EDITORIAL BOARD Editor-in-Chief J. A. Hudson Imperial College of Science, Technology & Medicine, London, UK
E. T. Brown
University of Queensland Brisbane, Australia
Senior Editors C. Fairhurst
University of Minnesota Minneapolis, MN, USA
E. Hoek
University of Toronto Canada
INTERNATIONAL ADVISORY BOARD G. Barla
Politecnico di Torino, Italy
Y. D. Diadkin
St. Petersburg Mining Institute, Russia
P. Londe
Pierre Londe & Associates, Puteaux, France
Y. Nishimatsu
University of Tokyo, Japan
Y. Ohnishi
University of Kyoto, Japan
T. Ramamurthy
Indian Institute of Technology, New Delhi, India
J.-C. Roegiers
University of Oklahoma, Norman, OK, USA
M. Romana
Universidad Politecnica de Valencia, Spain
O. Stephansson
Royal Institute of Technology, Stockholm, Sweden
Tan Tjong Kie
Chinese Academy of Sciences, Beijing, China
H. Wagner
Chamber of Mines, Johannesburg, South Africa
W. A. Wittke
Technische Hochschule Aachen, Germany
EDITORIAL BOARD Editor-in-Chief J. A. Hudson Imperial College of Science, Technology & Medicine, London, UK
E. T. Brown
University of Queensland Brisbane, Australia
Senior Editors C. Fairhurst
University of Minnesota Minneapolis, MN, USA
E. Hoek
University of Toronto Canada
INTERNATIONAL ADVISORY BOARD G. Barla
Politecnico di Torino, Italy
Y. D. Diadkin
St. Petersburg Mining Institute, Russia
P. Londe
Pierre Londe & Associates, Puteaux, France
Y. Nishimatsu
University of Tokyo, Japan
Y. Ohnishi
University of Kyoto, Japan
T. Ramamurthy
Indian Institute of Technology, New Delhi, India
J.-C. Roegiers
University of Oklahoma, Norman, OK, USA
M. Romana
Universidad Politecnica de Valencia, Spain
O. Stephansson
Royal Institute of Technology, Stockholm, Sweden
Tan Tjong Kie
Chinese Academy of Sciences, Beijing, China
H. Wagner
Chamber of Mines, Johannesburg, South Africa
W. A. Wittke
Technische Hochschule Aachen, Germany
COMPREHENSIVE ROCK ENGINEERING Principles, Practice & Projects
Editor-in-Chief JOHN A. HUDSON Imperial College of Science, Technology & Medicine, London, UK
Volume 4 EXCAVATION, SUPPORT AND MONITORING
Volume Editor JOHN A. HUDSON Imperial College of Science, Technology & Medicine, London, UK
PERGAMON PRESS
OXFORD · NEW YORK · SEOUL · TOKYO
U.S.A.
Pergamon Press Ltd, Headington Hill Hall, Oxford, OX3 OBW, England Pergamon Press Inc., 660 White Plains Road, Tarrytown, New York 10591-5153, USA
KOREA
Pergamon Press Korea, KPO Box 315, Seoul 110-603, Korea
JAPAN
Pergamon Press Japan, Tsunashima Building Annex, 3-20-12 Yushima, Bunkyo-ku, Tokyo 113, Japan
U.K.
Copyright © 1993 Pergamon Press Ltd All rights reserved. No part of this publication may be reproduced, stored in any retrieval system or transmitted in any form or by any means: electronic, electrostatic, magnetic tape, mechanical, photocopying, recording or otherwise, without permission in writing from the publishers. First edition 1993 Library of Congress Cataloging in Publication Data Comprehensive rock engineering: principles, practice, and projects/ editor-in-chief, John A. Hudson.— 1st ed. p. cm. Includes indexes. ISBN 0-08-035931-0 (HC) 1. Rock mechanics. I. Hudson, J. A. (John A.) TA706.C642 1993 624.1'5132—dc20 92-18616 British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library. ISBN 0-08-042067-2 (Vol. 4) ISBN 0-08-035931-0 (Set)
® The paper used in this publication meets the minimum requirements of the American National Standard for Information Sciences-Permanence of Paper for Printed Library Materials, ANSI 239.48-1984.
Printed in Great Britain by BPCC Wheatons Ltd, Exeter
Preface It is 30 years since the International Society for Rock Mechanics (ISRM) was formed. Since then, an enormous amount of rock mechanics research has been conducted and a huge number of structures have been built on or in the world's rocks, sometimes with the benefit of rock mechanics, sometimes without. From all these experiences, a great deal has been learnt - but before now there has been no single source providing the quintessence of our knowledge base. We have many textbooks and journal issues on rock mechanics and rock engineering, we have electronic access to databases of abstracts, there are many practitioners, there are many relevant teaching institutions; but there has been no attempt at unification before - to look deep into the very soul of rock engineering. The task of editing this compilation was rewarding, frustrating, exhilarating and exhausting! Our objective was to produce a benchmark knowledge statement for rock mechanics and rock engineering that represents what has been learnt since rock mechanics emerged as a discipline in its own right about 30 years ago. We have succeeded - and I believe that there is more to these volumes than just their component parts. As one looks through the chapters and absorbs the distilled experience of all our contributing authors, a synergistic phenomenon is definitely experienced: the sum of knowledge in these volumes goes beyond just the additive information of the chapters. Here we have the very essence of our subject: a heady mixture of the purity of mechanics, the idiosyncrasies of nature and the determination of mankind. My main thanks go to the contributing authors. They were each asked to write down in condensed form 'what they had learnt in life'. They took this to heart - and that is why the chapters are in a different genre to the usual scientific paper: the impact of the chapters is far greater and more significant than a 'normal' scientific paper. Cumulatively, the information is what we now know about rock mechanics and rock engineering. The three Senior Editors supporting me on this project all worked above and beyond the call of duty, demonstrating the truth of the old adage, 'If you want a job done, ask a busy man.' Professor Ted Brown was the President of the ISRM from 1983-1987 and is now the Deputy Vice-Chancellor of the University of Queensland. He has demonstrated a professionalism and work ethic that is unparalleled. Professor Charles Fairhurst is the current President of the ISRM, for the period 1991-1995, and is one of the world's wisest and most experienced rock mechanics academics. Professor Evert Hoek is one of the world's best, if not the best, rock engineering practitioners, and has certainly travelled further than anyone else in pursuance of his rock engineering goals. These were my three colleagues in this Comprehensive Rock Engineering venture. I thank them from the bottom of my heart. The project would not have been possible without our publisher, Pergamon Press. Jim GilgunnJones was the prime mover. Dr Colin Drayton, Dr Helen McPherson, Tim Jackson, Tracey Wells, Johanna Reilly and Peter Frank organized and carried out the copy-editing and production work. I should like, therefore, to thank Pergamon Press as an organization and to thank separately all of the people who were involved with the production of Comprehensive Rock Engineering. It has been a pleasure to work with them too. Readers should know that credit for the high presentational quality of the volumes lies directly with them. In terms of the coordination of contributors and manuscripts, the person who has done the most work is my wife, Carol. All authors, whether knowingly or unknowingly, have received letters written by her, and had their manuscripts scrupulously scanned to ensure that they had complied with 'instructions'. Without her help, the whole process would have taken much longer, perhaps for ever. Lastly, to anyone who is outraged that this reference work is not truly comprehensive, the Oxford English Dictionary includes in its definition of the word 'comprehensive' the phrase 'including much'. As Editor-in-Chief of Comprehensive Rock Engineering, I guarantee that this is true! JOHN A. HUDSON Welwyn Garden City, UK vu
Contributors to Volume 4 Mr Ch Amstad Rock Engineering Department, Swiss Federal Institute of Technology, Zurich, ETH-Hönggerberg, CH-8093 Zurich, Switzerland Dr D. A. Anderson Tensor Technologies, PO Box 92, Hazleton, PA 18201-0092, USA Professor L. Baochen Vice Director of Changsha Research Institute of Mining & Metallurgy, PO Box 67, Changsha 410012, Hunan Province, People's Republic of China Professor P. Choquet Department of Mines & Metallurgy, Université Laval, Ste Foy, Quebec G1K 7P4, Canada Dr E. P. Deliac Elf Aquitaine Production, 26 avenue des Lilas, F-64018 Pau Cedex, France Professor C. H. Dowding Department of Civil Engineering, Northwestern University, Evanston, IL 60208-3109, USA Mr D. F. Fawcett Babite Shaw & Morton, Consulting Engineers, 64 London Road, Maidstone, Kent ME16 8QW, UK Professor W. L. Fourney Department of Mechanical Engineering, University of Maryland, College Park, MD 20742, USA Dr R. J. Fowell 12 Otters Holt, Durkar, Wakefield, West Yorkshire WF4 3QE, UK Dr P. Fritz Rock Engineering Department, Swiss Federal Institute of Technology, ETH-Hönggerberg, CH-8093 Zürich, Switzerland Dr J. Hadjigeorgiou Department of Mines & Metallurgy, Université Laval, Ste Foy, Quebec G1K 7P4, Canada Dr S. Hibino CRIEPI, 1646 Abiko, Abiko-Shi, Chiba-ken 270-11, Japan Dr M. Hood Queensland Centre for Advanced Technologies, 2643 Moggill Road, Pinjarra Hills, Kenmore, Queensland 4069, Australia Professor J. A. Hudson Department of Mineral Resources Engineering, Royal School of Mines, Imperial College of Science and Technology, London SW7 2BP, UK Professor P. K. Kaiser Geomechanics Research Centre, Fraser Building F217, Laurentian University, Ramsey Lake Road, Sudbury, Ontario P3E 2C6, Canada Professor K. Kovari Rock Engineering Department, Swiss Federal Institute of Technology, ETH-Hönggerberg, CH-8093 Zürich, Switzerland Professor S. Littlejohn Department of Civil Engineering, University of Bradford, Bradford, West Yorksire BD7 1DP, UK Professor B. Lundberg Department of Technology, School of Engineering, Uppsala University, PO Box 534, S-751 21, Sweden Dr M. P. Luong CNRS-Laboratoire de Mécanique des Solides, Ecole Polytechnique, F-91128 Palaiseau Cedex, France ix
x
Contributors to Volume 4
Dr V. Maury DREA, Société Nationale Elf Aquitaine, Spécialité Mécanique des Roches, Centre Nicoulau, Avenue P Angot, F-64018 Pau, France Dr C. K. McKenzie Australian Blasting Consultants Pty. Ltd, PO Box 818, Toowong, Queensland 4066, Australia Dr K. W. Mills Strata Control Technology Pty. Ltd, PO Box 824, WoUongong East, New South Wales 2520, Australia Professor Y. Mizuta Department of Mining & Mineral Engineering, Yamaguchi University, Ube 755, Japan Sir A. M. Muir Wood Sir Wm Halcrow & Partners, Vineyard House, 44 Brook Green, London W6 7BY, UK Dr M. Motojima CRIEPI, 1646 Abiko, Abiko-Shi, Chiba-ken 270-11, Japan Dr P. P. Nelson Department of Civil Engineering, University of Texas at Austin, Austin, TX 78712-1076, USA Professor M. J. Pender Department of Civil Engineering, The University of Auckland, Private Bag, Auckland, New Zealand Dr D. J. Reddish Mining Department, University of Nottingham, Nottingham NG7 2RD, UK Professor S. Sakurai Department of Civil Engineering, Kobe University, Rokkodai, Nada, Kobe 657, Japan Dr B. L. Stillborg JA A AB, Aurorum 30, S-951 75 Luleâ, Sweden Mr A. G. Thompson Perth Laboratories, Division of Geomechanics, CSIRO, PO Box 437, Nedlands 6009, Western Australia, Australia Dr T. Vladut Hydro Environmental Research Group, Suite 1, 1715-27 Avenue, NE, Calgary, Alberta T2E 7E1, Canada Professor B. N. Whittaker Deceased, formerly Department of Mining & Mineral Engineering, The University of Leeds, Leeds LS2 9JT Mr C. R. Windsor Perth Laboratories, Division of Geomechanics, CSIRO, PO Box 437, Nedlands 6009, Western Australia, Australia
Contents of All Volumes Volume 1
Fundamentals
Overview 1
The Nature and Fundamentals of Rock Engineering
Geological Setting 2 3 4
The Significance of Structural Geology in Rock Mechanics The Mechanics of Natural Rock Deformation Rheology of Rocks and Plate Tectonics
Rock Mass and Site Characterization 5 6 7 8 9
The Role of Engineering Geology in the Design of Surface and Underground Structures Engineering Properties and Characterization of Rock Discontinuities Modern Developments in Rock Structure Characterization Groundwater in Rock Engineering Mechanisms and Consequences of Creep in Crystalline Rock
Strength and Deformation Properties 10 11 12 13 14 15
The Influence of Microstructure on Rock Deformation Rock Strength Criteria: The Theories and the Evidence Mechanical Behavior of Anisotropie Rocks Strength and Modulus Responses of Anisotropie Rocks Effect of Joints on Rock Mass Strength and Deformability Soft Rock Engineering
Constitutive Models and Numerical Modeling 16 17 18 19 20 21 22 23
Constitutive Behavior and Numerical Modeling Constitutive Models for Intact Rock, Rock Joints and Jointed Rock Masses Modeling Discontinuities in Numerical Analysis An Overview of the Boundary Element Methods Applications of Finite Element Analysis to Mining Engineering Rock Rheology Computer Simulation of Fracture Processes Application of Bifurcation Theory to Rock Mechanics Problems
Dynamics/Rock Excavation 24 Dynamic Behavior of Rock 25 The Boundary Element Method for Elastodynamics 26 Theories of Rock Cutting Integration/Application 27 28
Understanding Deformations in Tunnels Case Examples of Rock Mechanics Principles Used in Rock Engineering
Subject Index
Volume 2
Analysis and Design Methods
Overview of Design 1
Analysis and Design in Rock Mechanics - The General Context xi
Contents of All Volumes
XU
Rock Mechanics Continuum Modeling 2 Constitutive Modeling for Rocks and Joints with Comments on Numerical Implementation 3 Numerical Modeling of Yield Zones in Weak Rock 4 Time-dependent Response of Rock Around Tunnels 5 Fundamentals of Poroelasticity 6 Computational Methods in Fluid Flow 7 Thermal-Hydraulic-Mechanical Coupling Analysis of Rock Mass 8 Continuum Models for Layered and Blocky Rock Rock Mechanics Discontinuum Modeling 9 Numerical Modeling of Discontinua 10 An Introduction to Distinct Element Modeling for Rock Engineering 11 Determination of the 'Design Block' for Tunnel Supports in Highly Jointed Rock 12 Some Modern Developments in Block Theory for Rock Engineering 13 Rock Block Modeling with Interactive Graphics 14 Design of Pressure Tunnels and Shafts Applications to Rock Engineering - Civil Engineering 15 Interactive Computer Graphics Analysis of Rock Block Movement 16 Analysis of Explosions in Hard Rocks: The Power of Discrete Element Modeling 17 The Analysis of Fractures, Stress and Water Flow for Rock Engineering Projects 18 Stability of Underground Openings in the Storage of Low and High Temperature Materials 19 Radioactive Waste Repository Design 20 The Design and Construction of Underground Hydraulic Structures in Permafrost Soil Applications to Rock Engineering - Mining Engineering 21 Stress Analysis in Mine Design 22 Planning Mass Mining Operations 23 Soft Rock Properties and Strata Control 24 Design of Coal Pillar Arrays and Chain Pillars 25 Fundamentals of Mine Roadway Support Design: Rock-Support Interaction Analysis 26 Computer Aided Design and Rock Mechanics for Coal Mine Layouts and Operation 27 Design of Pillars with Backfill Interaction - A Case Study 28 The Use of Numerical Modeling for Underground Coal Mine Design Overview Aspects of Rock Engineering Design 29 Safety Concepts Applied to Rock Masses 30 Risk Analysis of Old Mine Workings 31 Design Methodology for Rock Engineering: Principles and Practice 32 Empirical Design and Rock Mass Characterization Subject Index
Volume 3 Rock Testing and Site Characterization Overview 1
Rock Properties, Testing Methods and Site Characterization
Basic Rock Properties 2 The Measurement and Estimation of Basic Rock Strength 3 Uniaxial Strength Testing 4 Triaxial Testing for Rock Strength 5 Hardness Tests for Rock Characterization 6 Time-dependent Behavior of Rocks 7 Characterizing Clay Shales
Contents of All Volumes
Xlll
Discontinuities 8
The Collection and Analysis of Discontinuity Orientation Data for Engineering Design, with Examples 9 Modern Developments in Discontinuity Analysis - The Persistance-Connectivity Problem 10 Pattern Analysis and Simulation of Joints for Rock Engineering 11 Construction of Equivalent Discontinuum Models for Fracture Hydrology
Stress and Stress Measurement Methods 12 13 14 15 16
Stresses in Rock and Rock Masses CSIRO Triaxial Stress Measurement Cell The Hydraulic Fracturing Method of Stress Measurement: Theory and Practice The HTPF and the Integrated Stress Determination Methods Measuring In Situ Rock Stress by Borehole Slotting
In Situ Stress 17 18 19 20 21
Rock Stress in the Fennoscandian Shield Rock Stress and Rock Stress Problems in Norway Rock Stresses and Rock Stress Monitoring in Canada Case Studies of Hydraulic Fracture Stress Measurement in Australia Measuring Rock Stress: Case Examples of Rock Engineering in Japan
Rock Mass Classification 22 23
Classification of Rock Masses for Engineering: The RMR System and Future Trends A Geomechanical Classification for Slopes: Slope Mass Rating
Geophysics 24 Dynamic Elastic Tests for Rock Engineering 25 Seismic Investigation for Rock Engineering 26 Geophysical Testing for Rock Engineering 27 The Use of Cross Well Seismology to Characterize and Monitor a Steamed Oil Reservoir Case Examples of Testing 28 29 30 31
Borehole Dilatometer Testing for Rock Engineering How Do Some Field Tests Really Work? The Case of the NX-Borehole Jack The Phenomenon and Examples of Rock Creep The Importance of Creep and Time-dependent Dilatancy, as Revealed from Case Records in China 32 Laboratory Experiments: Their Role in the Problem of Rock Burst Prediction Site Characterization 33 Modern Surveying Techniques for Mining and Civil Engineering 34 Case Study of Hydraulic Fracture Experiments at the Multiwell Experiment Site, Piceance Basin, Colorado, USA 35 Rock Mass Investigations in Hydroengineering 36 Rock Mass Response to Thermal Loading and Unloading at the Spent Fuel Test 37 Design, Execution and Analysis of a Large-scale In Situ Thermomechanical Test for Siting High-level Nuclear Waste Repository 38 The Atomic Energy of Canada Limited Underground Research Laboratory: An Overview of Geomechanics Characterization Subject Index
Volume 4
Excavation, Support and Monitoring
Overview 1
The Construction Process
Contents of All Volumes
XIV
Blasting 2 3 4 5
Mechanisms of Rock Fragmentation by Blasting Methods of Improving Blasting Operations Blast Monitoring: Regulations, Methods and Control Techniques Blast Vibration Monitoring for Rock Engineering
Mechanized Excavation 6 7 8 9 10 11
Computer Modeling and Simulation of Percussive Drilling of Rock The Mechanics of Rock Cutting Theoretical and Practical Rules for Mechanical Rock Excavation The Use of Water Jets for Rock Excavation TBM Performance Analysis with Reference to Rock Properties The Effects of Rock Properties on the Economics of Full Face TBMs
Support 12 13 14 15 16 17 18
The Design of Support for Underground Excavations Development of Tunnel Support Philosophy An Overview of Tunnel, Underground Excavation and Borehole Collapse Mechanisms Overview of Rock Anchorages Rock Reinforcement - Technology, Testing, Design and Evaluation Rock Mass Response to Large Blast Hole Open Stoping Coal Mine Support Systems
Back Analysis Monitoring 19 20 21 22 23 24 25 26 27 28 29
Back Analysis in Rock Engineering Decision Making in Tunneling Based on Field Measurements Deformation Monitoring for Stability Assessment of Underground Openings Rock Mass Behavior During Large-scale Cavern Excavation Predictive Calculation and Monitoring of Rock Stress and Displacement Induced by Ore Extraction A Method for Monitoring Rib and Lining Pressure Dynamic Indications of Rock Mass Failure Infrared Thermographie Observations of Rock Failure In Situ Testing and Monitoring of a Test Drive in an Underground Coal Mine Subsidence Behavior of Rock Structures Ground Surface Movements Due to Underground Excavation in the People's Republic of China
Subject Index
Volume 5
Surface and Underground Project Case Histories
Overview of Underground Space and Developments 1 2 3
The Expanding Role of Rock Engineering in Developing National and Local Infrastructures Subsurface Space - An Important Dimension in Swedish Construction Recent Developments in Rock Engineering in Norway: Gas-tight Rock Caverns, Subsea Road Tunnels, Steel-fiber Reinforced Shotcrete
Developments and Case Studies: Civil Engineering 4 5 6 7
Design of Large Powerhouse Caverns in Weak Rock Power Caverns of Mingtan Pumped Storage Project, Taiwan The Agua del Toro Dam, Mendoza, Argentina - A Case of Insufficient Surface Geology Studies Affecting Underground Excavations The Rio Grande Pumped Storage Complex, Cordoba Province - A Case Study of Excavations in Contrasting Rock Anisotropy
Contents of All Volumes
xv
8
A Case History in Argentina - Rock Mechanics for the Underground Works in the Pumping Storage Development of Rio Grande No. 1 9 Rock Instrumentation - Developments and Case Studies from Australia 10 Lessons from Two Large-scale Underground Rock Mechanics Projects: Crestmore and Climax/NTS Developments and Case Studies: Mining Engineering 11 The Use of Rock Mechanics Principles in Canadian Underground Hard Rock Mine Design 12 Case Study of Rock Mechanics in the Masua Mine, Italy 13 Calculation Methods and Experience of Using Energy Saving Systems for Controlling Local Climate in Mines, Tunnels and Underground Constructions 14 Caving Geomechanics 15 The Role of Geological Discontinuities and Tectonic Stresses in Mine Seismicity 16 Experiences with the Application of Modern Rock Classifications in Coal Mine Roadways 17 An Overview of Application of Coal Mine Ground Control Techniques in the USA 18 Residual Subsidence Over Abandoned Coal Mines 19 Case Studies in Coal Mines in India Developments and Case Studies: Geothermal Energy and Radioactive Waste Disposal 20 21 22 23
Analytical and Numerical Modeling of High Pressure Fluid-Rock Mechanical Interaction in HDR Geothermal Energy Reservoirs Rock Mechanics for Underground Nuclear Waste Disposal in France Rock-Backfill Interaction in Radwaste Repositories Man-made Induced Seismicity
Developments and Case Studies: Petroleum Engineering 24 The Use of Rock Mechanics in Petroleum Engineering: General Overview 25 Hydraulic Fracturing - The Significance of In Situ Stresses and Rock Mechanics 26 Advances in Shale Mechanics - The Key to Wellbore Stability Predictions 27 Perforation and Stimulation Design for Deviated Wells at the Kuparuk River Field, Alaska Further Developments and Case Studies 28 Réévaluation of the Stability of Large Concrete Structures on Rock 29 The Use of Rock Engineering to Overcome Adverse Geology at Revelstoke Dam 30 Large Piles in Weak Rock - West Gate Freeway Project 31 Flexural Toppling of Siltstones During a Temporary Excavation for a Bridge Foundation in North Devon 32 Preliminary Analysis of Quarry Slopes in a Weathered Rock Mass Profile 33 Examples of Rock Engineering in Chile Cumulative Subject Index
1 The Construction Process JOHN A. HUDSON
Imperial College of Science, Technology and Medicine, University of London, UK
1.1
INTRODUCTION
1
1.1.1 Rock Engineering and the Systems 1.1.2 The Engineering 'Perturbation 1.1.3 Analyses of the System Response 1.1.3.1 Cellular automata 1.1.3.2 Primary state variable evolution 1.1.3.3 Interaction matrix energy flux 1.2
1.1
The Basic Concept: Quality Types of Monitoring
The Construction Process The Way Ahead
17 19 22 22
The Basic Concept: Satisfying 'Natural' Support Ground Response Curve
the Engineering
Objective
22 23 27 29
Control
29 32
CONCLUSIONS
1.5.1 1.5.2 1.6
Distribution
MONITORING
1.4.1 1.4.2 1.5
17
The Basic Concept: Alteration of the Size Types of Excavation The Interface with the Support Objective
SUPPORT
1.3.1 1.3.2 1.3.3 1.4
3 7 9 9 11 12
EXCAVATION
1.2.1 1.2.2 1.2.3 1.3
Background
32 Summarized
32 34
REFERENCES
35
INTRODUCTION
The subjects of the volumes in Comprehensive Rock Engineering have been arranged in the following order: Volume 1, Fundamentals, edited by Professor Brown; Volume 2, Analysis and Design Methods, edited by Professor Fairhurst; Volume 3, Rock Testing and Characterization, edited by myself; this is Volume 4, Excavation, Support and Monitoring, again edited by myself; the last volume. Volume 5, Surface and Underground Project Case Histories, is edited by Professor Hoek. The logic in the presentational order of this rock engineering knowledge base is to provide the fundamentals of the supporting rock mechanics subject first in Volume 1, then in Volume 2 to discuss how to model the rock mass and develop design methods for rock engineering projects. In Volume 3, rock characterization is presented, with all the complications of dealing with the natural rock material. Now, in this volume, the subject is construction - to be followed in the next volume by case studies, i.e. descriptions of 'what happened' during a wide variety of construction operations. I mentioned in the main Preface that these volumes contain a heady mixture of the purity of mechanics, the idiosyncrasies of nature and the determination of mankind. In Volumes 1 and 2, there is much discussion of pure mechanics and how to deal with the idiosyncrasies of nature as manifested in natural rock masses and their response to engineering activities. This is reinforced by the chapters in Volume 3, where there is an emphasis on how to test the rock for its properties and 1
Overview
2
Effect I: Displacements and rock failure
ntact rock squeezed out Displacements occur because rock resistance removed
Block slides
Excavation
Discontinuities
Rock mass
Effect 2: Stress rotation
rincipal ^ ^ \ ess ^ y^ \
Normal and shear stresses become zero at excavation - which becomes a principal stress plane
Minor principal stress
,ln the rock, the ' principal stress ' magnitudes and ' orientations are / a l t e r e d - one ' principal stress ' being perpendicular / t o the excavation ''boundary
Principal stresses rotated to become parallel and perpendicular to an unsupported excavation boundary
Effect 3; Water flow
-Water flow induced Hydraulic head reduced to zero, excavation becomes a sink
'/
Excavation acts as a sink
Discontinuities
Figure 1 The three primary effects of excavation (after Hudson [27])
how to characterize a rock mass destined to host an engineering structure. There, we come face to face with the idiosyncrasies of nature. Now, here in Volume 4, Excavation, Support and Monitoring (i.e. the construction process), we deal with the determination of mankind. Thus, it is not only the principles and design ideas: the excavation, support and monitoring of rock are being considered with a specific engineering objective in mind. In Chapter 1 of Volume 3,1 finished the overview of rock properties, testing methods and site characterization with a figure illustrating the three primary effects of excavation, and I promised to maintain continuity of the discussion by presenting the same figure as Figure 1 of this Chapter-which indeed it is. When the rock is excavated, displacements occur because rock resistance is removed, the normal and shear stresses become zero at the unsupported excavation boundary, the hydraulic head is reduced to zero in the excavation, and the excavation is a sink. In this chapter, I will start by continuing the systems interpretation of rock engineering that was introduced in Chapter 1 of Volume 3. The systems interpretation considers the engineering
The Construction Process
3
'perturbation' to the rock mass; then there is discussion and presentation of three ways currently being developed of generically analyzing the effects of construction (the three primary effects being illustrated in Figure 1). This is followed by discussion of excavation (interpreted as altering the rock block size distribution), support (interpreted as satisfying the project objective), and monitoring (interpreted as quality control). Finally, in the conclusions section, there is a summary of the construction process and discussion of 'the way ahead'. 1.1.1 Rock Engineering and the Systems Background In the first chapters of Volumes 1 and 2, Professors Brown and Fairhurst have both drawn attention to the antiquity of rock engineering: its history is lost in the mist of time, but many ancient structures still exist. The largest rock structure ever built is the Great Pyramid of Giza in Egypt. It is at least twice the volume and 30 times the mass of New York's Empire State Building [1,2]. The masonry volume, as originally constructed, is more than 2 500000 m3. It was the largest building in the world from ca. 2500 BC to this century; it must have taken thousands of people to build; and no one is quite sure how it was constructed or even exactly what its purpose was. It seems ironic that in the future we will be building radioactive waste repositories, and their purpose, although clear to us, is unique: once the waste is emplaced, as little disturbance as possible should occur [3, 4] -just like the pyramids. As Professor Fairhurst mentioned in Chapter 1 of Volume 2, the later Gothic cathedrals of the twelfth to sixteenth centuries represent a zenith in the art of masonry construction, with concentration of forces in long slender columns. I have also drawn attention to these [5] in the context of the structures being built without the advantage of the present rock engineering knowledge base. Even relatively simple constructions made from rock blocks can be elegant. Consider the old stone bridge near Cayeli in Turkey, which is illustrated in Figure 2. The central portion of the bridge is made from a single arch of stone blocks without any mortar and is one of the most extraordinary rock structures I have seen. There are many stone bridges of structural interest [6, 7], but the one in Figure 2 really fires the imagination. How could one have the confidence to build this without modern technical knowledge?-probably by many years of trial and error, as indeed was the case with the cathedrals, many of which collapsed during and after construction. The flying buttress
Figure 2 Masonry bridge with an arch consisting of a single set of rock blocks (northern Turkey)
4
Overview
method of supporting cathedral walls was a practical solution to wall instability and was gradually refined until the flying buttress itself became an integral part of the aesthetic appeal of cathedral architecture. A source book for all the major stone monuments built by mankind is [8]. Now, whether we are considering civil, mining or petroleum engineering projects, the design of rock engineering systems is becoming increasingly complex (e.g. underground space in general [9] and hot water storage in caverns [10]). This is because of the larger scale of projects, newer types of engineering, the increasing availability of technical information, enhanced computational capabilities, and the need to interface with other parts of the total system. Not only is there the basic need to cope with this increasing complexity, there is also the requirement to improve our 'auditing' procedures in the context of validating proposed technical solutions and associated computer programs. Also, we must ensure that quality assurance procedures are both sensible and implementable. The construction of dams [11] and hydroengineering in general [12] are just two further examples of this complexity. As a result of this complexity, the author has proposed that a 'top-down' analytic approach be developed to supplement the 'bottom-up' synthetic approach that has been primarily used to date in modeling procedures. The distinction between the two approaches is shown schematically in Figure 3. This is explained in Chapter 1 of Volume 3 and explained further elsewhere [5]. An abbreviated explanation is also presented here in order to allow this volume to stand alone. Basically, the conventional synthetic model, shown at the lower left of Figure 3, is an 'exact representation' model in which the components are assumed and the model constructed, or synthesized. Thus, given the assumption of the model's components, the synthetic model is a high fidelity model with exact mathematical characteristics. In the illustration, the parallel linkage of a spring and a dashpot (the viscoelastic Kelvin model) can be completely described via the relevant equations. If the values of the material constants in the model are in some doubt, there are methods of accommodating uncertainty by probabilistic and fuzzy methods. There can be confusion between the words 'analytic' and 'synthetic'. In this text, the two words have been used according to their dictionary definition for normal usage: 'analytic' to mean breaking down to find the components, and 'synthetic' to mean building up from the components. Also, it is conventional practice in rock mechanics to use the term 'parameter' as opposed to 'variable', e.g. the parameters in a rock mass classification scheme. This has become rather awkward as the development of the rock engineering systems approach becomes more mathematically based, because the use of the term 'variable' would be more appropriate. In the mathematical sense, a variable is defined [13] as 'a symbol, such as x, y or z, representing an unspecified member of a class of objects, numbers, etc". The word 'parameter' has several meanings, but in this context is defined [13] as 'an arbitrary constant whose value affects the specific nature but not the formal properties of a The analylic and synthetic approaches to rock engineering modeling
Analytic
model
ToP-down
approach
• Known extent of application • Inexact representation of whole system • Interfacing with other systems easy • Convergent to correct model
• Contains all aspects of the problem • Modeling has sufficient fidelity • Necessary auditing procedures automatically generated
} Synthetic model
• Not necessarily convergent to correct model • Interfacing with other systems difficult • Exact representation of part of system • Unknown extent of application Bottom-up approach
Figure 3 The synthetic and analytic approaches to rock mass modeling
The Construction Process
5
mathematical expression, such as the arbitrary constants a and b in ax2 + bx + c = 0.' The term 'parameter' has been used in the text here following rock mechanics convention and the use of the term in [5] but, with continued development of the subject, it is anticipated that the variables and parameters per se will have to be unambiguously distinguished. In this increasingly complex world, the problems with the synthetic model for modern rock mechanics and rock engineering design are that it has an unknown extent of application and is not easily interfaced with other systems, e.g. the environmental system [6]. Perhaps its greatest drawback is that there is no guarantee that, by continual modification and extension, it will be convergent to the correct model, or at least to a model which approximates reality sufficiently well for the project objectives to be achieved. On the other hand, the analytic model, shown schematically at the top left of Figure 3, is an 'inexact representation' model in which the components are established by breaking down the system into its component parts, i.e. by analysis. This involves establishing which are the main parameters and which interactions may occur: in other words, solving the fully coupled problem with all the elements present. Characterizing the components has to be inexact initially because the behavior of many components will be unknown or unmeasurable directly. The boundary of applicability is defined beforehand and is therefore known. Interfacing with other systems is easier (e.g. [14]). Most importantly, the model naturally converges to the correct model - because that is the very essence of the analytic method. Everything necessary is defined to be within the solid border line of the top left sketch in Figure 3. It is expected that, in due course, the two approaches will be combined and the associated hybrid analytic - synthetic models will benefit from the advantages of each: the benefits of systematically establishing the complete model from the analytic approach, and the ability to have high fidelity simulation via the synthetic approach (e.g. [15]). This is illustrated by the central box at the left of Figure 3. Another benefit accruing from the analytic approach is that validation and auditing procedures are automatically generated. How much information is required to solve the problem? Do we have enough information? The idea of the information and other audits is discussed later in Section 1.5.2. Much work has already been conducted 'in synthetic mode' to solve the basic and coupled rock mechanics problems, as has been described in detail in Volumes 1 and 2. The analytic method is fundamentally different in principle from the synthetic method and requires a systems approach. The subject of systems has a long history and is well developed (e.g. [16], but it is necessary to develop a systems approach specifically for rock engineering owing to the idiosyncrasies of the rock material and the nature of rock engineering. The essence of this special character is that all rock masses that will host engineering projects are already in existence, and have been so for a long time. The engineering is therefore a perturbation to an existing system for a specific rock engineering objective. The interaction matrix is used as a device for listing all the primary state variables and their interactions. The procedure has been summarized in Chapter 1 of Volume 3 and presented in detail elsewhere, together with an atlas of rock engineering mechanisms [5]. However, a brief description is included here for completeness. The basic concept of the interaction matrix is shown in Figures 4(a) and 4(b). The main subjects or variables in the rock engineering problem are listed along the leading diagonal of a square matrix. This is the diagonal from the top left to the bottom right of the matrix. The influence of one subject or variable on another is then included in the appropriate off-diagonal box, as shown in Figure 4(a). This is an analytic method, because we are not just establishing a pedagogic device for locating existing knowledge: the matrix is created by including any variables that could be involved, and then establishing what interactive mechanisms are in the off-diagonal boxes. There may be sufficient existing information tofillthe boxes; or there may not be. Note that a clockwise convention has been used to locate the influence of A on B as opposed to the influence of B on A (see Figure 4a). An example 4 x 4 matrix is shown in Figure 4(b). This has the leading diagonal terms rock mass structure, in situ stress, water flow and construction. Examples of the interactions between the different leading diagonal terms are shown by the sketches in the off-diagonal boxes. Note that the information in the off-diagonal boxes is illustrative of the types of mechanisms that can occur in these locations; the information is not intended to be comprehensive. Within the context of this volume on construction, it can be seen that the boxes in the last column of the matrix (many of the subjects discussed in Volumes 1-3) are related to how the rock mechanics parameters potentially affect construction. The complementary boxes in the last row of the matrix (many of the subjects discussed in Volumes 4 and 5) are related to how construction potentially affects the rock mechanics parameters. The idea of increasing the size of the interaction matrix to accommodate any number of leading diagonal terms and to provide theflexibilityto analyze any problem at any resolution is shown in
6
Overview (α) Influence of A on B Box ij
Box // A
Γ\ f
Influence of Bon A Box ji
Subject B
Box jj
Figure 4 (a) The principle of the interaction matrix (after Hudson [5]). (b) Illustrative 4x4 interaction matrix with leading diagonal terms rock mass structure, in situ stress, water flow and construction (after Hudson [68])
Figure 5. With few leading diagonal terms the matrix is simple, but a complete characterization of the off-diagonal mechanisms is difficult because they are complex. With many leading diagonal terms, N, the matrix will have many off-diagonal mechanisms, N (N — 1) mechanisms, but there is a much better chance of being able to specify them and indeed to be able to establish the associated behavior. One interpretation [17] of the resolution of the matrices in Figure 5 is that the top level 3x3 matrix would be the one of interest to the client. What is the project? Where is the site? What is the rock? The intermediate M x M matrix would be the one of interest to the consultant. What aspects of this problem do I have to consider? For example, in [5], the 12x12 matrix presented for underground openings has the following 'parameters' for the 12 leading diagonal terms: excavation dimensions, rock support, depth of excavations, excavation methods, rock mass quality, discontinuity geometry, rock mass structure, in situ stress, intact rock quality, rock behavior, discontinuity
The Construction Process
7
Top level matrix Coarse resolution (Rock) (Site) (Project)
Intermediate level matrix Medium resolution
Lowest level matrix Finest resolution P P
Figure 5 Increasing the dimension of the interaction matrix increases the resolution
Monitoring
.. Construction stops
Time
Parallel system
Construction starts Site investigation
Figure 6 The systems interpretation of construction (after Hudson [5])
aperture, and hydraulic conditions. Finally, the lowest level matrix with the finest resolution will be of interest to anyone trying to establish the fundamental rock mechanics mechanisms. 1.1.2 The Engineering 'Perturbation' Thus, given the operation of a matrix such as that shown in Figure 4(b) and hence the interactions between all the parameters, we know that an NxN matrix is a representation of the process-response system for all rock masses that are currently in existence. The nature of engineering via the systems approach can then be interpreted as shown in Figure 6. (In this context, there are four basic types of system. The morphological system is concerned with the components. The cascading system is concerned with what happens, given the morphological system. The process-response system is concerned with how it happens, given certain morphological and cascading systems [18]. Engineering is the control system, relating to the specific project objective.) The naturally operating interaction matrix representing the rock mass is shown at the lower part of Figure 6 where, as in Figure 4(b), the operations introduced by man are considered in the lower right-hand box of the matrix. This is the sense in which engineering is regarded as a 'perturbation'. Firstly, we start with site investigation to find out the rock properties and the mechanisms that are
8
Overview
involved at the particular site. This is interrogation of the system to discover its morphological components and cascading behavior. Even this procedure in itself will cause a disturbance to the stability of the interaction matrix. In some cases, the disturbance will be insignificant; in other cases, as for example the extra permeability introduced by drilling site investigation boreholes, the disturbance could prejudice the integrity of the proposed engineered project. Referring to Figure 6, construction then starts-as indicated by the base of the darker shaded portion of the construction bar in the diagram. Although some components of construction could occur simultaneously, the essence of the whole construction operation is that it is a serial set of perturbations through time imposed on the parallel process-response system modeled by the interaction matrix. In other words, there will be continuing disturbance to the bottom right-hand box during construction operations. Eventually the project is complete and construction stops-as again shown in Figure 6. This may then be followed by a period of monitoring in order to ensure that the rock properties continue to be as expected, the mechanisms are proceeding as expected, and that the structure continues to fulfil its objective. Finally, when monitoring ceases, the host rock mass supporting or containing the project structure reverts back to a natural process-response system, which may or may not be operating in a similar way to that before site investigation. Thus, the total systems understanding of rock engineering is summarized in Figure 6. The significance of the engineering 'perturbation' via the interaction matrix analytic approach is shown in Figure 7. In (a), construction starts-as represented by the black box at the bottom right of the matrix. The leading diagonal terms representing the different state variables involved in the behavior of the rock mass are indicated along the leading diagonal, and the lighter shaded boxes in the off-diagonal parts of the matrix represent the mechanisms linking the leading diagonal parameters. When construction is switched on, as shown in (a), all the parameters are potentially affected via the interactions represented by the last row of the matrix, as shown in Figure 7(b). In (c), these initial changes in the parameters then affect construction through the interactions in the last column of the matrix. This is the first 'kick': in other words, the construction causes a disturbance to the natural process-response system, which then kicks back. Finally, as shown in Figure 7(d), all the parameters start affecting each other, i.e. the whole matrix is activated. This is potentially a much greater response, with continual changes reaching the construction box via the interactions in the last column of the matrix - the second 'kick'. In fact, the effects in the 'double kick', shown in Figure 7 and described above, will be smoothed out to a more or less continuous response because of the different times involved in the operation of
Figure 7 Construction and 'the double kick'
9
The Construction Process At time t, the parameters have values
p*
IP, I
p
I
2
X 1
f P, I
At t i m e / + Δ / , the parameters have values
Ρ?+Δ>
Either as a discrete process through repeated ùts or as a continuous process, the alteration in the P, values with time can be studied,e.g.
Time
Time
Time
Time
Figure 8 The evolution of the parameter values with time
each of the different mechanisms in the off-diagonal boxes, i.e. the different lag times. However, it is helpful to distinguish, via Figure 7, how the rock is affected by construction in these two ways, both directly (through affecting the parameters that then immediately affect construction) and indirectly (by activating all mechanisms). Together, these are the generic engineering 'perturbation'. This indicates the requirement to study the evolution of the values of the leading diagonal terms with engineering increments and with time-parameter value analysis. A variety of potential behavioral modes are shown in the lower part of Figure 8. If, after the parameters are disturbed, their values attenuate as shown in the first case the matrix will stabilize. If a parameter value increases asymptotically to a certain value, then again it is likely that the matrix will stabilize. If, on the other hand, as in the last two cases, there is a continuing increase in the value of a parameter, then it may reach some threshold representing an engineering interpretation of instability, e.g. the strength of the rock is reached or a rock block slides. It is advantageous, therefore, to have models that will allow consideration of the whole of the process-response system representing the rock mechanics and rock engineering circumstances. These are also of help in the 'thinking process'. It is crucial to be able to model the introduction of a perturbation associated with construction and then to be able to establish whether the total process-response system will stabilize, or whether there is any potential for instability. In fact, it is helpful to go further and establish, via this type of modeling, the optimal method of construction. The reader will find that the interpretations of excavation, support and monitoring as linked to the project objective and as described in the following chapters of this volume are much clearer within the systems context for construction. In the last subsection of this part of the chapter there is a brief presentation of the analyses of system responses that are currently underway.
1.1.3
Analyses of the System Response
There are several methods for studying the total systems behavior of a rock mass subjected to engineering. Current work on establishing the generic behavioral modes is advancing in three areas.
LI.3.1
Cellular automata
Cellular automata are dynamic models of assemblages of elements of the rock mass that are being governed by a set of rules (the canon), and which evolve patterns of behavior with time and engineering perturbations [19, 20]. If the cell linkages and behavior are considered as a priori information, the automaton compilation and operation will be in synthetic mode (cf. Figure 3). Conversely, if the components and mechanisms are established via a hierarchical top-down analysis
10
Overview
of rock mass properties and behavior, establishing the automaton's structure will be in analytic mode (cf. Figure 3). Indeed, the cellular automaton approach may be the ideal one for rock mechanics because it can be used in both modes. Also, cellular automata have the potential to provide a revolution in modeling capability since all existing methods, as described in Volumes 1 and 2, can be considered as particular cases of cellular automata. If, for example, each cell is operating internally and in conjunction with its neighbors according to the equations governing the basis of the theory of elasticity, then elastic problems can be solved. Similarly, distinct elements can be modeled. If quanta can move from cell to cell, mass transport can be modeled, e.g. water flow. Moreover, coupled problems are automatically solved via the basic canon applied to the cells and are in principle no more difficult to solve than any other problem. Inhomogeneity and anisotropy are potentially inherent aspects of the automaton's morphology. Establishing the macro- and microbehavior of the cellular automaton with time together with parameter value analysis are again inherent aspects of the output of the automaton as it changes with repeated time increments. Finally, the 'cognition' of the automaton's behavior can be characterized and enhanced via neural network analysis. In short, the automata model the rock directly as each element of rock reacts to its neighbors' (changing) conditions. These models, as applied to rock mechanics and rock engineering, are still in their infancy and have not been capable of implementation earlier because of the lack of the computer capacity required. However, all that is about to change and we can look forward to major developments in our subject area through the use of this approach. One example of a Rock Mass Automaton (RMA) (from [19]) is shown in Figure 9. This is a 20 x 20 2-D automaton. The boundary conditions are set,
Figure 9 (a) Initializing the structure of the cellular automaton, (b) Establishing the boundary conditions for the cellular automaton, (c) The quanta value distribution in the cellular automaton after 850 time steps, (d) Contour plot of the quanta values after 850 time steps (all after Millar and Hudson [19])
The Construction Process
11
the process-response units are assigned values for N (the number of matter units) and q (the number of quanta, or discrete energy particles) and then the rock mass automaton rules applied for each process-response unit. The events are implemented and the process repeated. The N values of the process-response subsystems are here arranged to model a rock fracture (see Figure 9a). The boundary q value conditions superimposed on initial q value conditions are shown in Figure 9(b). The q value distribution after 850 time steps is shown in Figure 9(c) with an associated contour plot shown in Figure 9(d). These diagrams give an indication of the method's promise and its utility in providing the understanding of complex coupled mechanisms and engineering guidance. Moreover, the capability of the automaton to adapt to spatial distributions given input conditions in a manner consistent with the training of neural networks is evident from this work [21].
1.1.3.2
Primary state variable evolution
This is an approach that formalizes the interaction matrix description of the system [17, 22] and uses graph theory [23] to study concurrent and consecutive mechanism concatenations to simulate system behavior and to be able to predict the consequences of any type of construction. Consider the concept of a pathway through the generic matrices in Figure 5. This would represent a particular sequence of events. For example, rock is removed by blasting, the excavation becomes a water sink, water flows from the rock discontinuities into the excavation, the water pressure in the discontinuities is reduced, slip occurs on the discontinuities, a block is released and falls into the excavation, this releases another block which falls in, then a larger discontinuity is exposed and more water flows into the excavation, the stress distribution around the excavation is altered, the strength of the intact rock is reached, failure occurs . . . Such a sequence might be represented by a path through the interaction matrix as in Figure 10. One of the methods of studying such paths has been developed by Yong [17], as shown in Figure 11. Although a 4 x 4 interaction matrix is illustrated, the method is suitable for a matrix with any dimension. Consider the variables xx to x 4 along the leading diagonal and known binary relations between each pair of variables in the off-diagonal boxes. Remembering the clockwise rotation convention, note that the x1 to x2 relation, with x1 as the independent variable, is not the same as the x2 to xx relation, with x2 as the independent variable. In Figure 11, and for the path χλ to x2 to x 3 to x 4 , the value of x1 gives the value of x2, which then gives the value of x 3 , which then gives the value of x 4 . (During the progression through the path, the variables switch from dependent to independent variables.) Since the relations can be approximated as piecewise linear relations to any required resolution, the off-diagonal relations can be of any kind. This type of interaction matrix pathway analysis enables rules to be established for the overall stability of the matrix for all paths and for the critical pathways to be established according to criteria set by the project objectives.
Figure 10 A concatenation of consecutive mechanisms initiated by construction can be represented as a pathway through the interaction matrix
12
Overview
Figure 11 A concatenation of consecutive mechanisms through the interaction matrix can be explicitly evaluated if the off-diagonal relations between the parameters are known (after Yong [17])
7.7.5.3
Interaction matrix energy flux
A related approach [24] is to consider energy flux in the interaction matrix, and hence enable analysis of the potential rock mass behavioral modes resulting from construction. Consider that each of the leading diagonal boxes in the interaction matrix is associated with an energy potential, Et. This can be considered as a concept in isolation, or the energy potential can be associated with specific primary state variables on the leading diagonal, say the x1 to x 4 variables in Figure 11. Via the mechanisms in the off-diagonal boxes, energy is transferred from the ith leading diagonal box, with energy Ei9 to the;th leading diagonal box, with energy E} (i refers to the matrix row number,; refers to the matrix column number). For a full matrix, this means that energy can be transferred from all leading diagonal boxes to all others-as occurs in Figure 7(d). However, energy is only transferred from a higher potential to a lower potential, with the effect that, for one matrix energy transfer increment and for a full matrix, only half the off-diagonal boxes will be transferring energy. Readers may find it helpful to think of an analogy where there are vertical pipes arising from the leading diagonal terms (with the height of water in them indicating the potentials) and horizontal pipes with one way clockwise flow connecting the bases of all vertical pipes to each other. These horizontal pipes have different diameters so that the energy transfer rates between the Ets are different. For a matrix with JV leading diagonal terms, or N vertical pipes, there are N(N - 1) offdiagonal terms, or N(N — 1) horizontal one-way flow pipes with different diameters. Given a set of initialized potentials along the leading diagonal of the interaction matrix, the transfers of energy from Et to Ej9 i.e. ΔΕΦ or Ej to Eh i.e. AEji9 in an increment of time are given by Δ£0· = «uM^i or
E
J)
(1)
ΑΕβ = qjikjiiEj - Et)
whichever is positive, where qu are energy transfer efficiency coefficients and fc0. are the energy transfer coefficients (or diameters of the pipes in the analogy). These equations represent the case where the increment of energy being transferred from one leading diagonal box to another depends on the difference in potentials, on damping through which 'usable' energy is lost, and on the ease with which the energy can be transferred. There are many modifications that could be made to this basic canon. Mechanisms could become inoperative if too much energy is transferred (representing failure in the rock mass); conversely, mechanisms could be inoperative until a sufficient rate of energy transfer rate is possible (crack propagation). It is particularly useful to consider two separate overall effects: the effect of a leading diagonal term on the system; and the effect of the system on the leading diagonal term. Considering the row
13
The Construction Process Main parameters
Interactions /,y in
Pj along leading
off-diagonal boxes
diagonal
Sum of energy in row boxes /'. e. energy from P. Influence of parameter on system: Ordinate C, (for 'Cause')
Sum of energy j in column boxes i.e. effects?
to P,
Influence of system on parameter: Ordinate E, (for 'Effect')
Thus,incremental energy transfer coordinates for P. are (£,£". )
Figure 12 Generation of the incremental energy transfer coordinates for parameters along the leading diagonal of the interaction matrix
<; \
\
Pi
'ij
\ 1 — 1 \ t * ' \ \ Pj 1 **-" 1
Ci
Assume that P; values represent the energy states associated with each parameter and that the Ijj allow energy flux from Pt to Pi under steaay state conditions
.^
C-, represents all the energy flux from P-, via the Ijj Ej represents all the energy flux into P, via the Ijj If P, is in this region, Pi is an energy sink Cause -If Pj is in this region, Pj is an energy source
Energetic |/dorninance of/^
(C-E)/*J2
(source or sink)
Energy flux from P} C
Figure 13 Identification of the energetic intensity and dominance (as sources or sinks) of the parameters from their incremental energy transfer coordinates (C, E)
and column through a leading diagonal parameter P, in the interaction matrix (Figure 12), it can be seen that, for a given time increment of energy transfer, the sum of all the energy increments in the row through Pf represents all the energy leaving P f , whereas the sum of all the energy increments in the column through Pf represents all the energy being transferred to P f . Using the convention developed in [5] the row sum, Ch is termed the 'cause' and the column sum, Ei9 is termed the 'effect'. These provide (Cf, Et) energy transfer coordinates that allow the system changes to be plotted. The value of Cf + Et is a measure of the total energy transfer occurring via a particular P f . Ct - E( is a measure of the parameter's energy effect, i.e. whether it is dominant or subordinate. When Cf - E{ is positive, Pf is acting as a source; when Q - Et is negative, P, is acting as a sink. The concept of plotting the (Ch £,) coordinates is shown in Figure 13. The purpose of the
14
Overview
computer program written for this analysis was not only to calculate the energy changes that could occur within the interaction matrix but also to generate displays that would help in the 'cognition' of the resulting behavioral modes. The components of the screen display are shown in Figure 14. Component A of the screen display (the matrix at the top left) shows the basic interaction matrix. For the purposes of the simulation shown in Figure 15, the values of the parameters have been initialized at values between 0 and 100 units. The values of the parameters at any time are indicated by the color scheme-in equal increments from blue, representing a value between 0 and 10, through various shades of purple, to red for parameter values between 90 and 100. The off-diagonal terms represent the energy transmission coefficients, the ktj in equation (1), and are color coded with 10 green shades: dark shades of green represent low energy transfer coefficients, light shades of green represent high energy transfer coefficients. Thus, the leading diagonal colors will change as the simulation proceeds, but the off-diagonal terms will not change because they represent thefixedfc^·energy transfer values (the diameters of the pipes). Component B of the screen display (the vertical bar graph at the bottom left of Figure 14) also shows the values of the 12 leading diagonal terms. The colour coding of these bars is the same as those in the diagonal of the matrix above. This vertical bar chart display has been included to provide an easier interpretation of the dynamic distribution of parameter energy values. Many of the simulations involve complex patterns and 'eddy currents' in the matrix. These are much clearer to interpret via the vertical bar graph display. Component C of the screen display (the matrix at the top right of the diagrams in Figures 14 and 15) represents the energy increments being transferred from one leading diagonal parameter to another during each time step. It represents all the energy transmission increments that are about to take place in the next step and can therefore be directly correlated with the values of the leading diagonal parameters in the matrix to the left. Because of equation (1) and the fact that energy is only transmitted from a high value parameter to a low value parameter, half of the off-diagonal terms will be colored to represent energy flow: if energy is flowing from parameter P( to Pj9 it will not then be flowing from Pj to P f . Thus, only one of the complementary pairs of the off-diagonal boxes, Ptj and Pji will be active. The off-diagonal boxes are colored from deep brown to bright yellow on a selfscaling equal class interval basis (between the highest and lowest energy transfer values in each increment) as the simulation proceeds, i.e. separately scaled for each increment of energy transfer. Thus, the colors in components A and B are fixed for the energy ranges but the colors in component C are self-scaling for each step from the highest to the lowest values of energy transfer, except when the energy transfer value drops below a cut-off value when no energy transfer color is shown. Component D of the screen display (the graph in the bottom right of the displays shown in Figure 15) is a graph of the 'energy out' versus 'energy in' for each parameter. As described earlier when referring to Figure 13, from the matrix C above the graph, the total energy leaving a parameter is found by the sum of the increments in the matrix row through the parameter. The total energy arriving at a parameter is found from the sum of the increments in the matrix column through the parameter. Each of the parameter points on the graph has coordinates (Ch £ f ). These points have
Figure 14 Components of the computer screen displays shown in Figure 15
The Construction Process
15
also been color coded on an absolute system according to the value of C + £, i.e. the parameter's total energy interaction intensity for each increment. The color coding is from bright red for the highest value of C + E to blue for the lowest values of C + E. Also, the diagonal line representing C = E has been shown. The brown spot on this line is the mean of all the Cf values and the mean of all the Et values-which are equal through the conservation of energy (theorem 1 in [5]). Thus, the position of the brown spot along the C = E line is a direct indication of how much energy in total is being transferred throughout the matrix during a particular time increment. Finally, component E of the screen display (the blue vertical bar originating at the center of the base of the screen display) is colloquially termed the entropy bar. The qu are energy transfer efficiency coefficients: energy is lost in what could be stress waves, hysteresis, heat losses, etc. What was previously usable energy has now been converted into a form that can no longer be used directly: this process is termed entropy and occurs with all real mechanisms. The lost' energy is put into the entropy bar, component E. With the inclusion of energy transfer efficiency coefficients, the more energy that is being moved around the matrix, the higher the entropy bar will rise. The numerical values of the total energy transfer for each matrix increment and the entropy level are also given on the screen displays. Thus, with these five main components, together with the two numerical values, the behavior of the matrix-given its fundamental off-diagonal structure and the initialization of the leading diagonal terms - can be followed clearly and is of major help in interpreting the modes of engineering behavior that are possible in any engineering system and as a result of any construction activities. Depending on the conditions assumed, this model has manifested attenuatory, oscillatory and chaotic modes. The output screens for the simulation shown in Figure 15 represent chaotic behavior. Note the initial conditions representing equilibrium among the first 11 leading diagonal parameters (in the top left output screen), except for the last leading diagonal box into which is put a sudden energy spike - representing construction. In this simulation, high energy transfer rates, the kij9 have been assumed with a mean k of 0.364. The transfer efficiency coefficients, qij9 were all 95%. This simulation was part of an overall study [22, 24] to establish potential behavioral modes of rock engineering systems by computer simulation of interaction matrix energy flux; the leading diagonal terms are generic, i.e. this simulation is not explicitly linked to specific parameters. The next step is to link this model with the previous approach, shown in Figure 11, in order to study specific coupled behavior and engineering perturbation. With high energy transfer rates, chaotic behavior is often observed with this model. Note that in Figure 15, the brown spot (representing the mean energy transfer for that time step) on the C versus E graph does not move monotonically towards the origin. This type of behavior can be modified to relatively stable attenuation if lower energy transfer efficiency and lower energy transfer coefficients are used. Within the context of the water analogy, one is reminded of the behavior of the Lorenz waterwheel [25], which has a steady rotational behavior when the water flow is low but becomes chaotic, oscillating backwards and forwards, when the water flow is at a higher level. The model is also similar to chemical reaction rate models discussed in [26], where steady state reactions proceed until certain reaction rates are exceeded. Chaotic behavior arises from a simple set of rules, which then lead to an unpredictable outcome and the possibility of major system changes. (Anti-chaos is where a similar set of simple rules can lead to clear structural arrangements, as is the case in the genetic and structural information in the development of organic structures.) It is instructive, therefore, to study all these behavioral modes through computer simulation to know the range of activity that one can expect and also to know how to control the rock mass system during construction in order to achieve the engineering objectives. With this background of the systems interpretation of construction, let us now consider the separate subjects of excavation, support and monitoring.
1.2 1.2.1
EXCAVATION The Basic Concept: Alteration of the Size Distribution
At its most basic level, the excavation process is an alteration of the in situ rock block size distribution to the fragment size distribution after excavation. This is illustrated in Figure 16 (from [27]). The block edge length is plotted on the x-axis and the percentage passing on the y-axis, so that the curves in this diagram are similar to the particle size distributions in soil mechanics. For a particular block edge length, the percentage of material that would pass through a sieve with that
16
Overview
(α)
(b)
(c)
Figure 15a, b, c
The Construction Process
17
(d)
(e)
(f )
Figure 15 Potential behavioral modes of rock engineering systems by computer simulation of interaction matrix energy flux. This example, with a high energy transfer coefficient, is illustrative of oscillatory and chaotic modes (after Hudson and Hudson [24]). (a) Initialization of primary state variables-note large energy input, i.e. construction box; (b) step 1; (c) step 2; (d) step 3; (e) step 4; (f ) step 5
18
Overview
0.1
I
10
100
Block edge length (m)
Figure 16 The process of excavation interpreted as changing the preexisting natural rock block size distribution to the debris fragment size distribution
mesh spacing is given on the y-axis. Obviously, with block edge lengths of 1 m, 10 m and 100 m, we cannot actually sieve the material, but there is a calculable in situ block size distribution before excavation because of the preexisting rock discontinuities. When the rock is excavated, either by hand, by blasting or by machine, the requirement is to alter the in situ block size distribution to a fragment size distribution. Thus, the excavation process is shown by the large arrow in Figure 16. In theory, it would be possible to calculate the amount of energy that is required to alter the preexisting rock block size distribution to the fragment size distribution. This could be done by considering the extra energy required to produce the new free surfaces needed to create the smaller blocks. However, because of the large losses that occur during the excavation process (it is estimated that less than 1 % of the energy supplied to a tunnel boring machine is actually used for the creation of new rock surfaces), it is not useful for practical purposes to conduct the calculation. The important point is that all excavation can be considered in terms of the diagram in Figure 16, with the direct implication that the engineering objective must be taken first. What is the purpose in moving from the preexisting rock block size distribution to the fragment size distribution? In civil engineering, the purpose is to remove rock in order to create an excavation which will perform some function. Therefore, the characteristics of the fragment size distribution will be more related to transportation and disposal issues. On the other hand, if the rock is being excavated to obtain minerals, then the excavation process shown in Figure 16 is the first stage in the overall comminution process, and there will be further stages of crushing and grinding in order to release the ore particles. In the mining context, it is necessary to remove the rock fragments from the mine and also to produce a particular fragment size for mineral processing. This may well affect the way in which the excavation process is conducted. There is much more latitude in mining excavation methods than in civil engineering construction. Mining methods are classed into 'naturally supported', 'artificially supported' and 'caving' methods [28-30]. Interesting early and recent texts on gold mining are [31-33]. In the naturally supported methods, such as the room and pillar method, the rock is removed leaving sufficient supporting pillars to sustain the extra load transferred from the areas where rock has been removed. Knowing the stresses in the pillars and the strength of the pillars, the mining can be designed in the same way as a civil engineering project with the associated factors of safety for the pillars. In the artificially supported methods, broken ore may be used to support the stope walls, as in shrinkage stoping, or artificial support in the form of cable bolting and similar measures may be introduced. Finally, in the caving method of mining, direct use is made of induced instability of the rock and the fact that it will break up when it is undercut and extracted from draw points. This latter method is far removed from civil engineering procedures because the very design of the whole operation is that it should collapse-with the rock mass moving into the failure portion of the complete stress-strain curve as an integral part of the mining design. There will be elements of the mine which have to remain as civil engineering structures in the sense of being permanent for the life of the mine, e.g. the main shaft [34], haulage ways and underground crusher chambers, but the basic concept of a caving mining method is very different in concept to civil engineering design. Also, the dimensions of a block caving operation can be very large, e.g. 300 m by 300 m by 300 m in the El Teniente mine in Chile. Needless to say, subsidence [35] can be a significant problem with the caving method of mining.
19
The Construction Process
The main consideration is therefore the fundamental objective associated with the excavation purpose. The next consideration is how to excavate the rock-which is outlined below. 1.2.2 Types of Excavation There has been much innovation in tunneling [36, 37] and excavation, but there have been very few major revolutions in excavation technology. The use of explosives is the main one, i.e. being able to use an energy inuput rate to a small part of the rock mass greater than can be supplied by one person. The other revolutions were the ability to pump water out of excavations, the use of machines, and now the use of computers. It is perhaps surprising that there are only three main methods of excavation: by hand, by blasting and by machine. Theflexibilityand robustness of the blasting method of excavation should not be underestimated. One example is illustrated in Figure 17 (from [27], from the source [38]). This shows the boreholes along the line of a presplit plane. The presplit method of blasting is used where there is concern about the final excavated surface, usually the final slope of a surface excavation. To avoid damage caused by stress waves and the gas pressure during explosive breakage of the rock, the final excavation plane is presplit first so that the initial fracture forms the final plane. Potential damage caused by subsequent bulk blasting up to this final plane is minimal because stress waves are reflected from and gas pressure vented along the presplit plane. Figure 17 is included to show that the presplit method of blasting can be robust and successful in the face of adverse factors (and despite the fact that the mechanics are not totally understood). At the left of Figure 17, there are low angle discontinuities-the traces of which can be seen in the diagram. Even though the discontinuities can be adversely orientated and consequently the presplit plane rather ragged, as shown by the heavy line in the diagram, this planar irregularity is insignificant compared to the damage that can be caused if the presplit plane is not already formed. Similarly, if, as in the central part of Figure 17, there are high angle discontinuities traversing the proposed presplit plane, these will again cause a micro-textural effect on the presplit plane between the boreholes-but again this is of minor significance compared to the damage that would be caused if the presplit plane were absent. If presplit blasting is well managed on site so that the basic criteria necessary for its successful operation (e.g. parallel boreholes, simultaneous detonation of low density explosive, etc.) are satisfied, then there are very few factors that can inhibit the successful formation of a presplit plane. One of these is shown in the right-hand side of Figure 17: a major principal stress
Low angle discontinuities
High angle discontinuities
In situ stress
Figure 17 The effects of discontinuities and in situ stress on the creation of the presplit plane
20
Overview
is adversely oriented to the intended presplit blast direction. Rock tends to break perpendicular to the least principal stress, and hence the direction of the maximum principal stress shown in Figure 17 would be awkward. In fact, presplit blasting is usually used for surface slopes where the stress is rarely high enough to cause the type of effect shown in Figure 17. Over the years, blasting technology has been refined a great deal and now is an effective and flexible method of excavating rock. Although, as mentioned, not all the dynamic effects are totally understood (e.g. the exact roles of the stress wave and the gas pressure during the formation of a presplit plane as shown in Figure 17), the methods have been optimized empirically. Professor Fourney discusses the mechanisms of blasting in Chapter 2 in this volume. With recent emphasis on enhanced quality control, both in terms of the explosives, detonators and site management, blasting is now effective, as is explained by Dr McKenzie in Chapter 3 in the volume. It can also be elegant, as with presplit blasting and postsplit or smooth wall blasting. If an attempt is made to use presplit blasting at any significant depth, where the stresses have significant values, then the result will be as shown in the right-hand side of Figure 17. This is overcome via postsplit or smooth wall blasting. Consider the excavation in Figure 18. Firstly, a rough excavation is created as shown in the figure. We noted in Chapter 1 of Volume 3 that excavation has a dramatic effect on the in situ stress state: whatever the stress state was before the excavation shown in Figure 18, after excavation the principal stresses are perpendicular and parallel to the excavation surface. As shown in Figure 18, the principal stress normal to the excavation surface is zero. Hence, if an initial 'rough' excavation is made and then a ring of blastholes made afterwards for secondary blasting, as shown in Figure 18, then a postsplit plane can easily be formed between the boreholes, because the least principal stress is the zero stress prependicular to the excavation surface. During this secondary blasting, the rock breaks perpendicular to the zero stress-forming a smooth wall. Blasting always produces vibrations. The methods of recording these, together with the regulations and control, are presented by Dr Anderson in Chapter 4. The interactions between blast vibrations and structures is explained by Professor Dowding in Chapter 5 (both in this volume). The other major type of excavation method is machine excavation. These machines come in various types, such as coal cutting machines used on longwall coal faces, the road header machines which have a rotating head on the end of an arm, and the full face tunneling machines often used in civil engineering projects where long lengths of relatively straight tunnel are required. Optimizing machine excavation is more difficult than optimizing blasting, because of the many components that are used in machines. A logical place to begin is at the cutter fragmenting the rock. The basic types of cutter are drag picks, disc cutters and button cutters, respectively dealing with harder rocks. In Chapter 6, this volume, Dr Lundberg explains the benefits of computer modeling and simulation of the percusive drilling of rock. The mechanism by which a drag pick cuts the rock has proved to be intractable. In Figure 19, there is a plot of tangential force against displacement, recorded during a 250 mm cutting traverse using a drag pick [39]. The force is oscillating: it increases as the pick moves forward, a chip is formed, the force reduces, and then increases to form another chip. The wide variety of localized peak forces is a function of the nature of the total cutting system, including the stiffness of the pick
Rough opening created first
Rock mass
Blasthole
Ring of 'postsplit' or smooth wall blastholes
Figure 18 Once an opening has been created underground, the magnitudes and directions of the stress field have been changed, with one of the principal stresses being perpendicular to the excavation surface and having zero value. 'Postsplitting' or smooth wall blasting then works very well because rock tends to break perpendicular to the least principal stress
The Construction Process
21
Figure 19 Record of tangential force during a 250 mm cutting traverse using a drag pick with a depth of cut of 0.25 mm on Villette limestone (after Almenara [39])
/
>
«
Large magnitude short duration pulses associated with blasting
Blasting cycle time
«
Machine servicing
Machine being used
I
I
Time Lower magnitude, essentially continuous input associated with mechanized excavation
Figure 20 Input energy versus time for blasting and mechanized excavation
and its holder. For the design of such picks, one wishes to know the maximum force that would be applied to the pick. For the total cutting operation, one would be interested in the area under the curve, representing the energy required. Dr Fowell covers the subject of the mechanisms of rock cutting in Chapter 7. This is extended by Dr Deliac in Chapter 8 in a discussion of the theoretical and practical rules for mechanical rock excavation. Developments in the use of water jets for rock cutting are described by Dr Hood in Chapter 9 (all in this volume). These chapters relate to the 'front end' of the system: there are many other factors to consider, especially in the case of a full face tunnel boring machine, which is a large machinefillingmost of the tunnel. Rock cutting occurs at the rotating head, the material has to be excavated from the tunnel, support may have to be erected. In Chapter 10, Dr Nelson describes tunnel boring machine performance analysis with reference to rock properties. This is extended in Chapter 11 by Mr Fawcett to include the effect of rock properties on the economics of full face tunnel boring machines (both in this volume). With only two main types of excavation available to us, it is interesting to compare the respective energy input rates and durations, as shown in Figure 20. In this figure, blasting and mechanized excavation are shown on the same graph. The blasting energy is input over a few seconds every, say, 8 hours or so: this is represented by a series of widely spaced vertical lines on the diagram. For machine excavation and assuming a machine were operating continuously except for servicing periods, the input energy per unit time is much lower, shown by the low level horizontal line just above the x-axis in the diagram. It is interesting to consider whether it might be possible to combine blasting and mechanized excavation to combine the advantages of each-the high energy input rates of blasting and the continuous input rates of machine excavation. Perhaps a tunnel boring machine that incorporated blasting in some way is the route ahead. The blasting would not necessarily have
22
Overview
to be by explosives: it could be via compressed air. Such a combined or hybrid method of excavation would be represented by rectangles on the diagram in Figure 20 and would certainly enable the base level energy input rate of machine excavation to be raised. There is no reason why machine tunneling rates should not be increased by an order of magnitude. Energy input to the excavation is not a problem; energy input focused at the excavation face is the problem. Once this problem is overcome, the back-up systems can be developed to accommodate the much higher rate of excavation. 1.2.3 The Interface with the Support Objective After excavation, whether by blasting or by machine, the rock may or may not need to be supported. The excavation and support objectives are quite different, as is indicated by reference to the complete stress-strain curve in Figure 21. For excavation, the rock must be made to reach the failure part of the complete stress-strain curve. Therefore the objective of excavation is to go beyond the compressive, shear or tensile strengths. Once this is achieved and the excavation space has been established through the removal of rock, then the proximate rock should be stable, i.e. be in the prefailure portion of the complete stress-strain curve in Figure 21. Thus, the physical boundary of the excavation between the rock and the air in the excavation represents a difference between two objectives: the removal of rock to form the excavation; and the stability of rock to retain the excavation. It is not surprising therefore, that problems occur at this interface-because it is the interface between two objectives [40]. As was noted with reference to presplit blasting, if precautions are not taken, the excavation objective can damage the proximate rock, prejudicing the support objective. Thus, there is an interface with the support objective because the support can be significantly influenced by the excavation construction. The excavation and support processes should be considered together as two parts of the same system, in the context of the overall project objectives and the subobjectives connected with the separate subjects of excavation and support. 1.3 SUPPORT 1.3.1 The Basic Concept: Satisfying the Engineering Objective In the previous section on excavation, we saw that the basic concept relating to excavation is alteration of the size distribution of the rock blocks. This applied whether the project objective was related to mining engineering (the excavated rock being required) or civil engineering (the excavated space being required). In the former case, the alteration of the rock block size distribution is the first stage of comminution in the mineral processing cycle; in the latter case, the alteration of the size distribution is to allow removal of the rock in order to create the excavated space. With support, the basic concept is also related to satisfying the engineering objective. In mining engineering, the ore extracted from the ground is the objective; hence, it is not directly obvious what the support philosophy will be. Some parts of the mine, such as the shaft and the access tunnels, will
Strain, E
excavation and support objectives
Figure 21 The boundary of an underground opening represents the interface between two engineering objectives: excavation and support, which are linked to the postfailure and prefailure portions of the complete stress-strain curve respectively
The Construction Process
23
require support so that their specific engineering function can be achieved, i.e. the transport of workers and materials. Other parts of the mine can be allowed to collapse on a controlled basis. In civil engineering, there are many functions that an excavation opening might be required to fulfil. For example, for the Channel Tunnel between England and France, there is the strict requirement to maintain the opening for an anticipated period of 125 years so that the trains can travel safely. Very little lining movement in the running tunnels is to be tolerated. Alternatively, the objective might be an underground excavation to host the generators of a hydroelectric scheme. In this case, more movement of the roof and walls, but not the floor, could be allowed, providing that the engineering objective is not prejudiced. The level of support will also depend on the consequences of failure. For example, the Channel Tunnel is a high prestige, high profile construction in which failure could have severe consequences and would be communicated worldwide via the media. Conversely, a main sewer tunnel beneath a road might fail in the same way and also directly prejudice the engineering objective, the transport of materials. Although, in the second case, the failure might be similar, it would not have such potentially severe consequences, nor would it be of such media interest. Moreover, the financial consequences of the failure of a sewer tunnel would not be as great as failure of the Channel Tunnel-and therefore the factor of safety for the sewer tunnel support will be less [41]. In Chapter 12, Drs Choquet and Hadjigeorgiou provide a review of the design of support for underground excavations and in Chapter 13 Sir Alan Muir Wood discusses the development of tunnel support philosophy. Dr Maury explains tunnel, underground excavation and borehole collapse mechanisms in Chapter 14 (all in this volume). All these factors indicate the critical requirement to identify the engineering objective beforehand. It may be that there is more than one engineering objective. For example, in the case of emplacement tunnels for radioactive waste disposal, there is an initial objective to maintain the tunnels open so that the waste can be emplaced; but, after the tunnels have been backfilled and the repository closed, slow creep and failure of the tunnels might be part of the design objective. It is essential to consider the engineering objective and the local site circumstances and then to design support according to these fundamental criteria and circumstances.
1.3.2
'Natural' Support
It will always be helpful to utilize natural support for sustaining the objective of the engineering structure: this will always have advantageous financial ramifications. Consider the modes of slope failure that are illustrated in Figure 22 [27]. It is usually not practical to anticipate a failure of the kind illustrated in these diagrams and then artificially to support a slope to avoid failure. Grouting the ground and possibly cable bolting through the anticipated failure zones would be expensive. It is far better to adjust the site excavation, the dip of the slope and the orientation of the slope such that failure will not occur. Where failure in hard rocks will occur along preexisting discontinuities, there are effective overlay techniques for analysing the potential for slip on discontinuities. These are well explained in [42]. In Figure 23 [27], there is a diagram of such an overlay on a stereographic projection of the discontinuities. There are six discontinuities, A to F, labeled on the projection. For a slope that is being considered with a dip of 75° and a dip direction of 295°, the overlay shows where wedge instability is possible, i.e. on discontinuity intersection lines IBE and IAB. This type of analysis indicates that by reducing the slope to, say, 55°, then wedge failure will not occur. Thus, if the dip direction has to be 295°, because it is the slope of a highway cutting and the highway orientation cannot be changed, then the possibility of wedge failure can be eliminated by reducing the slope from 75° to 55°. This type of flexible approach, maximizing the use of 'natural' support, is better than a rigid approach deciding on the slope characteristics and then constructing the slope come what may, possibly using expensive support measures. It is also possible to study uncertainty in this analysis in a similar way [43]. Many parameters will affect support. Some of these for slope stability have been indicated in Figures 22 and 23. Others have been reported elsewhere in connection with the systems approach [5, 44, 45]. The types of interaction matrix that were presented earlier have been studied for underground excavations utilizing the 12 parameters. These are listed in the lower part of Figure 24 and all influence failure. Parameters 6 and 7, discontinuity geometry and the related parameter rock mass structure, are the most dominant. Parameter 10, rock behavior, is the most subordinate, because it is dictated by the behavior of the other parameters. Number 3, excavation depth, has the highest parameter interactive intensity. Parameters 4 and 6, excavation method and discontinuity geometry, have the lowest interactive intensity. The underground circumstances are more interactive
24
Overview
Figure 22 Development of curvilinear slips in different geotechnical circumstances (from lecture notes for the Engineering Rock Mechanics MSc Course at Imperial College, London, UK, and after Hudson and Harrison [27])
than those for surface slopes with many mechanisms interacting, as illustrated by the underground excavations constellation in the cause-effect space shown in Figure 24. There are many sources of information on underground excavations, of which [29, 46, 47] are recommended. As shown in Figure 25 (from [27]), the overlay stereographic projection can also be effectively utilized, especially with inclined hemispherical projection, for underground excavations. This is further explained by Priest [48]. In the same way that the kinematic feasibility of blocks sliding or toppling can be established from overlays on the stereographic projection for surfaces slopes, so the possibility of blocks sliding or falling out from any rock surface underground can be studied. This can lead directly to the design of rock reinforcement measures for enhancing natural support. Block theory is also well advanced [16, 49]. Another factor in connection with underground excavation support is the in situ stress, remembering that the in situ stress components are always significantly affected by excavation. The principal stresses are all altered in magnitude and direction. In Figure 26, there is an illustration of a two-dimensional case showing that the load previously taken by the vertical component of the stress
25
The Construction Process N
Figure 23 Example of assessment of potential wedge instability using an overlay on a stereographic projection of the rock mass discontinuities (after Hudson and Harrison [27])
Φ
71 // \\\ >^
—7—ST
ς—"7
V ' Parameter N 40 f \dominance
\ i a
N
/ \
\
. \
V^
| \ 2 ' 3 Parameter\J intensity 1 • • · •ö \ / 2.
4
\ .
^
10
/
\
• v
\
>i. V
/v
\
/ \ 30
\
\X 40
Cause (influence of parameter on system) 1 2 3 4 5 6
Excavation dimensions RocK support Depth of excavations Excavation methods Rock mass quality Discontinuity geometry
7 8 9 10 I I 12
Rock mass structure In situ stress Intact rock quality Rock behavior Discontinuity aperture Hydraulic conditions
Figure 24 Parameter constellation in cause-effect space for the generic underground excavations interaction matrix (after Hudson [5])
tensor has to be taken up at the side of the excavation, resulting in an increased stress at the axis level. This applies to all underground excavations. It is necessary to establish whether the concentrated rock stress is likely to reach the intact rock strength, either at the time of excavation or subsequently as the intact rock degrades at the excavation surface. Again, through block analysis and stress analysis results, it is prudent to utilize natural support if at all possible. It is much better to arrange the shape, size and orientation of the excavation to minimize the possibility of block fallout and to minimize stress concentrations. All other factors being equal, tunnels are more stable when driven subperpendicular to major joint sets rather than subparallel to major joint sets. Similarly, it is sensible to orientate a tunnel parallel to the maximum principal stress so that the stress component
26
Overview
In practice, the inclination as shown here, should be conducted such that the point representing N the outward-directed normal from ' * the rock face moves to become the center of the projection
Upper hemispherical projection
N
f
Inclination angle
Horizontal inclination angle
Figure 25 The stability of underground blocks can be assessed in the same way as for surface blocks (see Figure 23) and by inclining the hemispherical projection to be coincident with the excavation surface (after Hudson and Harrison [27], and Priest [48])
Concentrated stress Load'gained' Preexisting rockst
Figure 26
When an underground opening is created in a rock mass, the load previously supported by the removed rock must be taken up elsewhere
that is subjected to the maximum concentration (cf. Figure 26) is not the maximum principal stress but the intermediate or minimum principal stress. If the vertical and horizontal stress components are different, elliptical excavations can be devised to minimize the maximum stress concentration, at least according to the theory of elasticity, and again this maximizes the use of natural support, in the sense that one is avoiding failure. These are very simple guidelines, but they can have very effective results where they can be applied. In Figure 27 there is a sketch of the stress concentrations around a longwall coal mining panel. (Coal mine ground control is discussed in [50].) The stresses are concentrated on each side of the panel as a result of the load redistribution. They are also concentrated ahead of the longwall face, again due to the load redistribution. In the same way that the stress concentration around a tunnel can be reduced by orientating the tunnel parallel to the maximum principal stress, so the stresses around longwall faces can also be reduced by suitable orientation of the longwall face. In the case of mining a 'bulk' material such as coal in which the variation in the excavation direction is possible, again the use of natural support in this way is possible. At the ends of the longwall face, there are peaks in stress concentration and indeed this is where most of the problems occur. In soft rock such as the strata within which the coal is often found, the 'sharp' stress concentrations can become attenuated with failure and creep, as is also indicated in Figure 27. Yet again, if these effects are understood and can be controlled, they can be used in the design to minimize the introduction of artificial support.
The Construction Process
27
Figure 27 The stress distributions in the rock around an advancing longwall face
Professor Littlejohn provides an overview of rock anchorages in Chapter 15, and Messrs Windsor and Thompson present a comprehensive explanation of the principles and hardware for rock reinforcement in Chapter 16 (both in this volume).
1.3.3
Ground Response Curve
There may be cases where, however lateral our thinking is in terms of utilizing natural support, some support must be introduced in order artificially to support the excavation and sustain the engineering objective. Useful references on support are [46, 51, 52]. When this occurs, one of the most useful concepts that covers many of the philosophical and mechanical aspects of support is the ground response curve, as illustrated in Figures 28(a) and 28(b). On the x-axis is the boundary displacement at the excavation wall. On the y-axis is the support pressure required to hold the excavation at that particular boundary displacement. The curve shown in the diagram in Figure 28(a) is a function of the rock type and the position on the excavation periphery, e.g. the roof or the walls, and is known as the ground response curve. It is a useful concept for considering the principles of artificial support and is described further in [46]. If the engineer is unwise and decides that no boundary displacement is allowed whatsoever, then the support pressure will have to be equal to the stress components that preexisted in the ground. An old mining maxim is that 'One cannot stop the roof coming down; one can only control its descent.' This also applies with slight modification to all support. There is no point in eliminating all boundary displacement-indeed, this will always be both unnecessary and impossible. The idea is to bend with the wind and only put in such support as is necessary. If the ground response curve intersects the x-axis (see one of the ground response curves in Figure 28b), then the excavation stabilizes at a small displacement with no artificial support. The associated boundary displacement is generally acceptable when it is in the order of millimeters. If, however, the displacement is unacceptable or the curve does not intersect the x-axis then some support is required. As shown in Figure 28(a), the installation of a stiff support is represented by a steep line in 'support pressure'-'boundary displacement' space; whereas a soft support will be represented by a gentler slope. (Note that the support cannot be installed straight away and the support lines should therefore start after some boundary displacement has occurred.) The kernel of understanding provided through this curve is that enough support should be installed to 'catch' the ground response curve, without putting in an unnecessary amount of support. One of the methods that has been used to achieve this when the ground response curve is unknown is to use a yielding support. This is also shown in Figure 28(a). In this case, the support load increases as the ground load on it increases until the support yields at a predetermined load, plastically deforming until the ground response curve is intersected. On a longwall coal face, this would involve hydraulic rams with a maximum allowable pressure; in simpler cases, it could be compressible inserts between lining segments. Consider the ground response curve which does not intersect the x-axis (the topmost curve in Figure 28b). Firstly, as indicated by all ground response curves, it is not necessary to install a
28
Overview
Ground support pressure required to hold excavation boundary at given displacement position
(α)
(b)
Figure 28 (a) The ground response curve and the principles of stiff, yielding, and soft support, (b) The type of excavation method can influence the form of the ground response curve
support pressure equal to the original ground 'pressure'. Indeed, this is not possible because some elastic displacement will occur instantaneously with excavation. Next, it is not necessary to put in a stiff support if a soft support will be sufficient. In physical terms, this means that the thickness of a continuous lining will be less or the separation between discrete steel arch supports will be greater. However, some support must be installed because the curve rises again (due to the rock losing its integrity after a certain amount of displacement). If the support is put in too late, even a stiff support will not catch the ground response curve, because the curve will then have started to rise. The author has found that the ground response curve, together with the available support lines, and the whole interpretation of this diagram provide a useful method of thinking about the type of artificial support that may need to be introduced into the excavation. Even if one were to consider filling the excavation with a backfill, as in a mine stope, the ground response curve indicates many things: the backfill must be in contact with the sidewalk, it is necessary to introduce the backfill at an early stage, and some type of yielding backfill could be advantageous. Finally, the link with the excavation method should not be forgotten. This is illustrated in Figure 28(b) where there are four ground response curves, corresponding to 'perfect' excavation (meaning the minimum disturbance possible), tunnel boring machine excavation, 'good' blasting and 'bad' blasting. These ground response curves are all for the same rock and excavation geometry: they are different because the type of excavation method will deleteriously alter the rock to different degrees. At the end of Section 1.2 on excavation, there is a discussion of the fact that the excavation method is linked with the support objective. The complement to this is that the support requirement can be intimately linked with the excavation method. There are many types of artificial support, varying from rock bolts that reinforce the rock by holding the rock blocks together to the type of precast concrete lining segments shown in Figure 29 Ground supports in weak rock are the subject of [53]. Point loads should not be applied to such a lining, and therefore some form of backfillingbetween the lining and the rock is advantageous. Also,
The Construction Process
29
Direction of installation
Knuckle' joint allows thrusts but not moments to develop
Figure 29 Precast concrete lining segments Adjacent living segments or rock blocks
Before rotation (load evenly distributed, low stress)
After rotation (load all transmitted through the corner, very high stress)
Figure 30 The dramatic effect of rotation occurring between adjacent lining segments or rock blocks
it is helpful if the precast segments have knuckle joints so that there is only a thrust in the lining and bending moments are minimized (see the inset sketch in Figure 29). The difficulty with either rock blocks or tunnel lining segments rotating is illustrated in Figure 30. Before rotation, the stress will be transmitted evenly across the discontinuity or the lining joint. Once the rock blocks or the concrete cast segments have rotated, as shown in Figure 30, all the load is transmitted through the small area remaining in contact. Very high stresses will be developed, because of the small area, and failure will result. This indicates the need, both from the point of view of supporting the rock blocks and for installing the lining, to inhibit such rotations. Indeed, this is one of the primary functions of shotcrete (cement-mortar sprayed onto the excavated rock surface). Dr Stillborg provides a case study of mine support in complex circumstances in Chapter 17 and Professor Whittaker reviews coal mine support systems in Chapter 18 (both in this volume). 1.4 1.4.1
MONITORING The Basic Concept: Quality Control
In excavation, the basic concept is the alteration of the rock block size distribution. For support, the basic concept is to satisfy the engineering objective. The basic concept of monitoring is quality control. It is necessary to check that all aspects of the excavation and support have been correctly implemented so that the function of the engineered structure is achieved. Thus, the monitoring can
30
Overview
be of many types. For example, it may be necessary to measure the size distribution of the excavated particles in order to check that optimal fragmentation has been achieved. Similarly, it may be necessary to monitor the movement of the excavation periphery to check that the support concept, whether natural or artificial, is indeed being successfully implemented [54]. Professor Sakurai reviews the subject of back analysis for rock engineering in Chapter 19, and Professor Kovari and Dr Amstad discuss decision making in tunneling based onfieldmeasurements in Chapter 20 (both in this volume). One of the most important aspects of monitoring is that it must be possible to decode the monitored results. This is indicated in relation to the interaction matrix in Figure 31. On the left is the coarsest resolution interaction matrix, containing just the three leading diagonal terms Rock, Site and Project. On the right is afinerresolution matrix of dimension JV. The lower the dimension of the matrix and hence the fewer the number of terms on the leading diagonal of the matrix, the more complicated the off-diagonal interactive mechanisms will be. It follows that results obtained from monitoring parameter values infinerresolution martices will be easier to decode. It will also depend on how interactive the mechanism is. If one considers the steps illustrated in the color pictures in Figure 15, representing a generic systems analysis of excavation, one can imagine how difficult it would be to decode what was happening if only one of the parameters in this system was being monitored. Without a knowledge of all the other parameters it is impossible to decode the total context. On the other hand, there may be a simple objective in the monitoring, e.g. to decide whether the displacement at the excavation boundary is excessive or not. Thus, monitoring that is directly aimed at establishing the value of a specific site parameter 'for its own sake' (e.g. blast vibrations [55]) will be easier than attempting back analysis. Conversely, successful back analysis closes the design loop and provides the greatest confidence, e.g. checking the validity of numerical models such as those in [56]. In Chapter 21, Professor Kaiser discusses monitoring the parameter of deformation for stability of underground openings. The review by Drs Hibino and Motojima in Chapter 22 illustrates the decoding problem. In Chapter 23, Professor Mizuta discusses the whole subject of prediction, calculation and monitoring with reference to rock stress and displacement induced by ore extraction. The specific subject of monitoring rib and lining pressure is discussed by Dr Fritz and Professor Kovari in Chapter 24 (all in this volume). One of the most dramatic illustrations of the advantages of monitoring that the author has encountered is in association with the geothermal project in Cornwall, UK [57, 58]. This is illustrated in Figure 32. Two boreholes were drilled during the Hot Dry Rock Project for the extraction of geothermal energy. Cold water was pumped down one borehole to a depth of 2 km. It was intended that the water should pass through the fractured rock and then be discharged at the surface from the other borehole. Because the site had been instrumented with seismic monitors, it was possible to observe the seismic events associated with the pumping of the water. In fact, in the initial tests, most of the water did not come back up the second borehole but migrated downwards from the 2 km level to about 3 km. This is evident from the microseismic events plotted in the diagram in Figure 32. A great deal of information was supplied by the monitoring of these seismic events leading to the design of practical geothermal systems.
Coarse resolution1 RSP
Finer resolution : P\,...,P[u
Extremely complicated interactive mechanisms,/^·
Less complicated interactive mechanisms, In
Ψ
'l2
R
RS
RP
SR
S
SP
PR
PS
P
'2. *l
KM >NZ
7iw]
fej
^j
Results obtained from monitoring parameter values in finer resolution matrices will be easier to decode. However, it may be necessary to monitor many P/ when working with a fine resolution matrix.
Figure 31 Interpreting monitored values during and after construction may be extremely difficult
The Construction Process
31
Figure 32 A dramatic illustration of the value of microseismic monitoring during water injection in a borehole at the Hot Dry Rock Geothermal Project in Cornwall, UK - viewed in the approximate direction of both the maximum principal stress and one of the joint sets (after Pine and Batchelor [57])
Figure 33 The foundation beneath a multistorey car park on the island of Jersey, UK
32
Overview
Dr Vladut reviews dynamic indications of rock mass failure in Chapter 25. Infrared thermographie observations of rock failure are explained by Dr Luong in Chapter 26 (both in this volume). 1.4.2 Types of Monitoring It cannot be emphasized enough that much thought should go into choosing the best parameters to observe and monitor and how to decode the results, especially if these are intended for back analysis using methods such as those in [59]. Consider the photograph in Figure 33, which is of the foundation beneath a car park on the island of Jersey, UK. It can be seen that this strong fractured rock has been rock bolted around the foundation support. If this support were to move, is the structure more stable or less stable? It could be either. If the foundation were to move by a sufficient amount, there could be cracking in the structure above with the structural integrity of the concrete in the car park prejudiced. On the other hand, if there were to be a small amount of movement, which did not damage the concrete above but which enhanced the stiffness of the bolted discontinuities, then the movement could effectively strengthen the foundation rock. There have been several conferences with papers on monitoring (e.g. [60-62]). The whole subject of monitoring is well illustrated in Chapter 27 by Professor Pender and Dr Mills with reference to in situ testing and monitoring of a test drive in an underground coal mine. The subsidence behavior of rock structures is explained by Professor Whittaker and Dr Reddish in Chapter 28, and Professor Liu Baochen in Chapter 29 (all in this volume). 1.5 CONCLUSIONS 1.5.1 The Construction Process Summarized It was seen in relation to Figure 6 how the evolution of the interaction matrix through time summarizes the parallel and serial nature of rock mechanics mechanisms and construction. This diagram representing the construction process is expanded in Figure 34 to include the near-field and the far-field. There is a natural process-response system before construction starts. The rock is investigated through the site investigation process, which slightly perturbs the rock. A particular construction scheme is decided upon and then implemented, effectively in a serial way compared to the parallel nature of the rock mechanics process-response system. Construction is implemented in the zone where the structure is required, and the near-field is affected. There is a far-field, which is defined as the zone not significantly disturbed by the engineering works. Construction then stops and monitoring occurs as necessary. Then the total system reverts to a natural process-response system with the structure contained on or in the rock mass.
Figure 34 The systems interpretation of construction, considering also the near field and the far field (after Hudson [5])
33
The Construction Process
This leads to the ideas illustrated in Figure 35. Engineering is an intelligent control system superimposed on the rock mass. Within the context of the project objective, full control is exercised in this region-interpreted as a cybernetic feedback control system implemented on the rock mass. Around the project there is a zone affected but not fully controlled by engineering; in this region there is partial control. Beyond this region, there is the natural process-response system in which there is no control. It is always helpful to think about these three regions when deciding on the design of construction. We can go further in Figure 36 and consider how the parameter interaction intensity indicates the choice of engineering control techniques. At the top of the diagram in Figure 36, there are two classes of possibility indicated. On the left, there is effective parameter independence with a low interactive intensity between the parameters, as indicated by the low mean parameter interactive intensity on the cause versus effect plot. On the right, there is a high parameter dependence with a high interactive intensity between the parameters, and the position of the mean parameter interactive intensity value is high on the C = E line. These two classes of interactive intensity indicate the possibility of direct control or indirect control, respectively. If the parameters are effectively independent, then one can safely alter one of the parameters without triggering (possibly uncontrollable) activity in the interaction matrix, e.g. a rockburst [63]. On the other hand, if the parameters are significantly
Process-response system (natural rock mechanisms) Zone affected but not full)1 controlled by engineering Intelligent control system (project engineering) Full control Partial control No control The term'full control'is used within the context of the project objective Figure 35
Construction will involve three zones in the host rock mass: a zone of full control (the term being used within the context of the project objective); a zone of partial control; and a zone of no control
(a)
(b) Parameter i ndependence E
/ I
/
/
/
/
'
Most Ijj have low values
(i)
Parameter dependence Λ
Low interactive itensity
1
E
/ /
/
Most Ijj have high values
/
/
/
p\
High interactive itensity
If the Ijj are close to zero, the parameters are almost independent and engineering control can be applied directly to each parameter (if physically possible)
(i i) If the Ijj have high values, the parameters are interactive and engineering control has to be indirect to some extent. Control must involve predicting how a specific disturbance will attenuate in the matrix to an equilibrium state Figure 36
The way in which control can be exercised via construction depends on the degree of parameter dependence
34
Overview
dependent, then control will have to be indirect in the sense that alteration of any one parameter will lead to major activation of the matrix mechanisms, and hence a wide variety of complications and interactions. Thus these are the two main methods of engineering control techniques, as indicated in Figure 36. If the off-diagonal terms are close to zero, the parameters are almost independent and engineering control can be applied to each parameter separately (if physically possible). If the off-diagonal terms have high values, the parameters are interactive and engineering control has to be indirect to some extent. Control must then involve predicting how a specific disturbance will attenuate in the matrix to an equilibrium state. This is the value of producing the types of constellation on the cause versus effect plot such as that shown in Figure 24. Naturally, there will be a wide spectrum of possibilities between these two extremes and it may well be that some parameters can be independently controlled, whereas others cannot be in the same system. Whichever is the case, it is prudent to study the interactions between all the parameters to avoid triggering an unstable event. 1.5.2 The Way Ahead The way ahead for construction is to enhance design methodologies through the many aspects that have already been mentioned in this chapter and through the implementation of improved site practices and contractual procedures. I will not forget one consulting experience when the two sides in dispute, the consulting engineers and the contractors, both separately explained to me with some passion that all the problems were caused by the rock: everything was 'the rock's fault'. We can be sure in engineering that the rock will always be innocent. Contractual arrangements that are inherently adversarial must be altered. Construction should be based on a total understanding of the whole design procedure and associated construction process, with everyone working together with the rock. Financial risk is discussed in [64]. The systems approach, which has been the perspective within which the text of Chapters 1 of both Volumes 3 and 4 has been presented, leads naturally to rock engineering audits. Even with the best arrangements, things can go wrong and constant monitoring of developments is required. Professor Bieniawski has suggested a set of principles [65] and I have proposed a suite of audits [5,22]. There should be an information audit to establish that enough information is available to solve the project engineering problem. There should be a technical audit to decide whether the design is indeed correct. There should be a financial audit to check that the design can be implemented in an acceptable financial manner. There should be an environmental audit. There could be an energetic audit to consider the movements of energy throughout the rock mass. I believe that there should also be an entropie audit. All mechanisms will lead to energy losses and an increase in entropy. Increases in entropy are generally associated with disorder. We therefore have to consider whether the local order created by the construction project and the benefit associated with it outweigh the inevitable greater increase in disorder with the associated disbenefit. These audits are illustrated in Figure 37.
Figure 37 Study of the alteration in the interaction matrix as a result of construction leads automatically to the consideration of rock engineering systems audits, of which seven are shown
The Construction Process
35
Thus, the way ahead is to utilize the principles that are explained in Volume 1, and which permeate all volumes, to implement the continuum and discontinuum analyses that have been presented in Volume 2, to improve the characterization of the rock properties and the sites as discussed in Volume 3, to utilize the construction philosophies that are described in the chapters in this volume, and to consider fully the precedent practice case studies in Volume 5. We are at a very interesting stage in rock engineering where, through Comprehensive Rock Engineering as a compilation of rock mechanics and rock engineering knowledge, a great deal of information is directly available in one source-basically equivalent to about 5000 direct personyears of experience (a mean of 25 years per author for 200 authors). There is increasing global communication and very rapid increase in computing capability and information technology [66, 67]. Thus, there is now the opportunity to enhance the methods by which we construct traditional structures. We should also be able to design and construct nonprecedent practice structures in an optimal way. Considering the amount of material in all these volumes, it is likely that the next step will be computer multi-media formats where the information can be presented in the form of text, photographs, videotape, sound, etc., and be interlinked via iconic interfaces. This has the potential also to enhance construction significantly. One or more persons, via a computer, should have an overview of the total construction process in order to ensure all the subsystem interfaces are correctly operating, and that person should have a very clear idea of the engineering objective. ACKNOWLEDGEMENTS I should like to express my personal thanks to all the authors in this volume for their contributions. Many authors have made considerable efforts to produce comprehensive articles describing their experiences in life, what they have learnt and what they consider to be most important aspects of rock mechanics and rock engineering. Some of the chapters are written in a style and contain information that is unobtainable elsewhere. Thus, special thanks go out here to all the authors of the chapters in this volume. The content of this first chapter of Volume 4 is based on 26 years of research, teaching and consultancy supported by the US and UK Governments and many other clients in many countries. I should like to acknowledge everyone who has helped me during this time-especially at the University of Minnesota, Minneapolis; the Transport and Road Research Laboratory; the Headquarters of the UK Department of the Environment; the University of Wisconsin, Madison; the Building Research Station; Imperial College and in many holes in the ground in many countries. I am grateful to the UK Construction Information Research and Information Association for permission to include Figure 4(b) (from [68]), to my colleague John Harrison for computer drawing Figures 1,16, 17, 22, 23 and 25 (from [27]), to Dean Millar for Figure 9, to Jiao Yong for Figure 11, to Raimundo Almenara for permission to use the plot in Figure 19 and to Ellis Horwood (the publishers of [5] from which several diagrams have been used). The latter were computer drawn by my son, Miles. My daughter, Jenifer, made sure that the references were supplied in the correct format. Over the last few years, I have had a major thrust on rock engineering systems. Relating to this subject area, I appreciate all the discussions which I have had with Peter Arnold, Christine Cooling, Lyn Flook, Kemal Gokay, John Harrison, Carol Hudson, Fin Jardine, Dean Millar, Max de Puy, Doug Spencer, Akio Tamai, Branko Vukadinovic and Jiao Yong. Dean Millar wrote the cellular automaton program and produced the diagram in Figure 9. Jiao Yong had the idea for the explicit evaluation of sequential mechanism analysis illustrated in Figure 11. My son, Jonathan, wrote the machine code for the computer screen color display in Figure 15. Without them, I would not have had so much to say and show. My wife, Carol, not only helped in the production of the whole of Comprehensive Rock Engineering but also in the editing of this chapter. Without her, I would not have enjoyed it so much.
1.6 REFERENCES 1. Edgar J. Great Pyramid Passages, p. 301. Bone and Hulley, Glasgow (1910). 2. Lemesurier P. The Great Pyramid Decoded, p. 350. Element Books, Tisbury, UK (1977). 3. HMSO. Assessment of Best Practicable Environmental Options (BPEOs)for Management of Low- and Intermediate-Level Solid Radioactive Wastes, p. 80. Her Majesty's Stationery Office, London (1986).
36 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54.
Overview NRPB. Boad Statement on Radiological Protection Objectives for the Land-Based Disposal of Solid Radioactive Wastes, p. 25. National Radiological Protection Board, Didcot (1992). Hudson J. A. Rock Engineering Systems: Theory and Practice, p. 185. Ellis Horwood, Chichester (1992). de Mare E. The Bridges of Britain, p. 226. Batsford, London (1954). Hopkins H. J. A Span of Bridges, p. 288. Praeger, New York (1970). Boudet J., Manceron C. and Ostier J. The Great Works of Mankind, p. 293. Bodley Head, London (1962). Inada Y. Underground, Underground, Underground, p. 157. Printed in Japan in Japanese, ISBN 4-627-48190-X (1992). Rehbinder G. (Ed.) Hot Water Storage in Rock Caverns, p. 223. Be-Fo and Swedish State Power Board, Stockholm (1988). Kollgaard E. B. and Chadwick W. L. (Eds.) Development of Dam Engineering in the United States, p. 1072. Pergamon Press, Oxford (1988). Thiel K. Rock Mechanics in Hydroengineering, p. 408. Elsevier, Amsterdam (1989). Borowski E. J. and Borwein J. M. Dictionary of Mathematics, p. 659. Harper-Collins, London (1989). Yegulalp T. M. and Kim K. (Eds.) Proceedings of the First International Conference on Environmental Issues and Waste Management in Energy and Minerals Production, p. 602. Batelle, Columbus (1992). Wei Lingli Numerical Studies of the Hydro-Mechanical Behavior of Jointed Rocks, p. 297. Ph.D. Thesis, Imperial College, University of London (1992). Goodman R. E. Introduction to Rock Mechanics, p. 562. Wiley, New York (1989). Yong J. Formalizing the Systems Approach to Rock Engineering, Ph.D. Thesis, Imperial College, University of London. In preparation (1993). Dury G. H. An Introduction to Environmental Systems, p. 366. Heinemann, London (1981). Millar D. L. and Hudson J. A. Rock engineering system performance monitoring using neural networks. In Artificial Intelligence in the Minerals Sector, Proceedings of an Institution of Mining and Metallurgy Conference, April (1993). Millar D. L. Neuro-Control of Rock Engineering Systems, Ph.D. Thesis, Imperial College, University of London. In preparation (1994). Aleksander I. and Morton H. An Introduction to Neural Computing, p. 240. Chapman and Hall, London (1990). Hudson J. A. Rock Engineering Systems 2: Implementation. Ellis Horwood, Chichester. In preparation (1995). Carre B. Graphs and Networks, p. 277. Clarendon Press, Oxford (1979). Hudson J. A. and Hudson J. L. Establishing potential behavioural modes of rock engineering systems by computer simulation of interaction matrix energy flux. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. In press (1993). Gleick J. Chaos, p. 352. Penguin Books, London (1987). Prigogine I. and Stengers I. Order out of Chaos, p. 349. Flamingo, Harper-Collins, London (1985). Hudson J. A. and Harrison J. P. Engineering Rock Mechanics. Pergamon Press, Oxford. In preparation (1994). Hustrulid W. A. (Ed.) Underground Mining Methods Handbook, p. 1754. American Institute of Mining, Metallurgical and Petroleum Engineers, New York (1982). Brady B. H. G. and Brown E. T. Rock Mechanics for Underground Mining, p. 527. Allen and Unwin, London (1985). Kidybinski A. and Kwasniewski M. (Eds.) Modelling of Mine Structures, p. 184. Balkema, Rotterdam (1988). Watermeyer G. A. and Hoffenberg S. N. Witwatersrand Mining Practice, p. 895. Transvaal Chamber of Mines, Johannesburg (1932). Jeppe C. B. Gold Mining on the Witwatersrand, p. 1821. Transvaal Chamber of Mines, Johannesburg (1946). Budavari S. (Ed.) Rock Mechanics in Mining Practice, p. 282. South African Institute of Mining and Metallurgy, Johannesburg (1983). Richards L. R. (Ed.) Shaft Engineering, p. 378. Institution of Mining and Metallurgy, London (1989). Whittaker B. N. and Reddish D. J. Subsidence Occurrence, Prediction and Control, p. 528. Elsevier, Amsterdam (1989). Harding Sir H. Tunnelling History and My Own Involvement, p. 258. Golder Associates, Toronto (1981). West G. Innovation and the Rise of the Tunnelling Industry, p. 355. Cambridge University Press, Cambridge (1988). Worsey P. Geotechnical Factors Affecting the Application of Pre-Split Blasting to Rock Slopes, p. 515. Ph.D. Thesis, University of Newcastle-upon-Tyne (1981). Almenara J. R. Investigation of the Cutting Process in Sandstones with Blunt PDC Cutters, p. 165. Ph.D. Thesis, Imperial College, University of London (1992). Whittaker B. N. and Frith R. C. Tunnelling: Design, Stability and Construction, p. 460. Institution of Mining and Metallurgy, London (1990). Megaw T. M. and Bartlett J. V. Tunnels: Planning, Design, Construction, p. 284. Ellis Horwood, Chichester (1981). Hoek E. and Bray J. W. Rock Slope Engineering, p. 402. Institution of Mining and Metallurgy, London (1977). Priest S. D. and Brown E. T. Probabilistic stability analysis of variable rock slopes. Trans. Inst. Min. Metall., Sect A 92, Al-12 (1983). Nathanail C. P., Earle D. A. and Hudson J. A. Stability hazard indicator for slope failure in heterogeneous strata. In Proc. EU ROCK '92 Conf. Rock Characterization, Chester (Edited by J. A. Hudson), pp. 111-116. British Geotechnical Society and Telford, London (1992). Arnold P. The Development of a Rock Engineering Methodology Using a Systems Perspective. Ph.D. Thesis, Imperial College, University of London. In preparation (1993). Hoek E. and Brown E. T. Underground Excavations in Rock, p. 527. Institution of Mining and Metallurgy, London (1980). Saari K. (Ed.) Large Rock Caverns, p. 1673. Pergamon Press, Oxford (1986). Priest S. D. Hemispherical Projection Methods in Rock Mechanics. Allen and Unwin, London (1985). Goodman R. E. and Shi G.-H. Block Theory and Its Application to Rock Engineering, p. 338. Prentice-Hall, London (1985). Peng S. S. Coal Mine Ground Control, p. 491. Wiley, New York (1986). O' Rourke T. D. (Ed.) Guidelines for Tunnel Lining Design, p. 82. American Society of Civil Engineers, New York (1984). Kaiser P. K. and McCreath D. R. Rock Support, p. 706. Balkema, Rotterdam (1992). Ward H. Ground supports for tunnels in weak rocks, Geotechnique IS, 133-171 (1978). Brown E. T. and Hudson J. A. (Eds.) Design and Performance of Underground Excavations, p. 518. British Geotechnical Society, London (1984).
The Construction Process
37
55. Dowding C. H. Blast Vibration Monitoring and Control, p. 297. Prentice-Hall, London (1985). 56. Crouch S. L. and Starfield A. M. Boundary Element Methods in Solid Mechanics, p. 322. Allen and Unwin, London (1983). 57. Pine R. J. and Batchelor A. S. Downward migration of shearing in jointed rock during hydraulic injection. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 21, 249-263 (1984). 58. Parker R. H. (Ed.) Hot Dry Rock Geothermal Energy, p. 139. Pergamon Press, Oxford (1989). 59. Pande G. N., Beer G. and Williams J. R. Numerical Methods in Rock Mechanics, p. 327. Wiley, Chichester (1990). 60. Hudson J. A. (Ed.) Rock Characterization, p. 486. (Proceedings of the EUROCK '92 Symposium, 1992, Chester). Telford, London (1992). 61. Kovari K. (Ed.) Field Measurements in Geomechanics, p. 1453. Balkema, Rotterdam (1984). 62. Sakurai S. (Ed.) Field Measurements in Geomechanics, p. 1271. (2nd International Symposium on Field Measurements in Geomechanics), vols 1 and 2. Balkema, Rotterdam (1987). 63. Fairhurst C. (Ed.) Rockbursts and Seismicity in Mines, p. 439. Balkema, Rotterdam (1990). 64. Thompson P. and Norris C. The perception, analysis and management of financial risk in engineering projects, Proc. Inst. Civ. Eng. 97, 42-47 (1993). 65. Bieniawski Z. T. Engineering Rock Mass Classifications, p. 251. Wiley, New York (1989). 66. Mutagwaba W. Design of an Intelligent Mining Decision Support System, p. 350. Ph.D. Thesis, Imperial College, Universtity of London (1991). 67. Gokay K. Developing Computer Methodologies for Rock Engineering Decisions, Ph.D. Thesis, Imperial College, University of London (1993). 68. Hudson J. A. Rock Mechanics Principles in Engineering Practice, p. 72. Butterworths, London (1989).
2 Mechanisms of Rock Fragmentation by Blasting W. L. FOURNEY University of Maryland, College Park, MD, USA
2.1
INTRODUCTION
39
2.2
HISTORY OF THE CONTROVERSY
40
2.3
STRESS WAVE MECHANISMS
41
2.4
GAS PRESSURIZATION MECHANISMS
47
2.5
CRATER BLASTING
51
2.6
CONTROLLED FRACTURING
59
2.6.1 2.6.2
59 62
Oil and Gas Well Stimulation Fracture-controlled Blasting
2.7
APPLICATIONS IN CONSTRUCTION AND QUARRY BLASTING
67
2.8
SUMMARY
68
2.9
REFERENCES
68
2.1
INTRODUCTION
The manner in which geological material is broken by explosive loading is not well understood. This lack of knowledge still exists even though explosives have been widely used for centuries for the purpose of resource removal and construction. It is important to understand the mechanism of rock breakage by explosives, since the production and sale of resources are becoming more and more competitive. It is necessary for nations that have developed a high standard of living to understand more completely the exact nature of rock breakage by explosives if they are to remain competitive in areas where labor rates are lower. In particular, the recent developments in computer technology have reached a stage where more use is being made of computer codes to predict the breakage of geological materials by explosive loading. Some of these codes do a very good job of predicting part of the breakage associated with rock fracture and fragmentation, but much more remains to be done. If these codes are to predict rock breakage accurately, more needs to be known about the breakage mechanisms, and the codes must address the inhomogeneous and anisotropic nature of rock. When explosives are used to break rock, the normal technique is to drill holes into the rock mass and then to place the explosive into the holes. When the explosive is detonated two things occur. The relatively small mass of chemical explosive is transformed into a very large volume of gas. This process is accompanied by the generation of very large gas pressures - in the hundreds of kilobars range - and large increases in temperature. The result of this detonation is pressurization of the borehole and fractures and a very strong shock wave which travels out into the rock. The long-raging debate is whether the resulting breakage is due to the large amplitude stress waves which travel through the medium, or due to the very large gas pressures, or both. Many research publications over the past 30 years have given evidence that either one or the other, gas pressures or stress waves, has been shown to be responsible for the breakage. It is a difficult problem 39
40
Blasting
to research for various reasons. The time required for detonation of the explosive is measured in the tens of microseconds range, while stress wave propagation also occurs in the tens or hundreds of microseconds range. Gas pressurization, however, persists for many milliseconds. Thus, all of these phenomena occur very quickly. The movement of the rock mass begins to occur in the millisecond range and takes seconds or tens of seconds to develop fully. The material under investigation ranges widely in material properties. In the worst case it can be classified as an inhomogeneous, nonisotropic, faulted and jointed material. These properties are not well defined and in many cases an exact representation would require the determination of more properties than it is possible to measure accurately in simple laboratory tests. Most of the evidence that has been presented in the literature, therefore, is not conclusive, nor can it be, due to the very complex nature of the material involved. Most of the testing that has been conducted to prove or disprove various theories, therefore, has utilized materials that are better behaved than the actual materials of concern; this has served further to confuse the issue in some cases. That is to say, in large-scale testing in thefield,the material is so complicated and so many parameters are inadvertently changed - even in tests that are conducted in sites that are adjacent to one another in supposedly the same material - that the scatter in the results is so great that the issues being examined become obscured. In the laboratory, where better control is possible, the important issues are compromised as a result of simplifying the materials being used. It is the intent of this chapter to review some of the previously conducted work aimed at identifying the mechanisms of rock breakage by explosive loading, in order to try and shed some light on the present state of knowledge. The views presented are therefore from the perspective of the author; they are intended to point out that stress waves and gas pressures both play an important role in the rock breakage process, and that the importance of one versus the other really depends upon the application at hand. Furthermore, it is felt that steps can be taken to ensure that both factors play an optimum role in the blasting process. Only by forcing this dual role can we optimize the results obtained from the utilization of explosive loading.
2.2 HISTORY OF THE CONTROVERSY The perceived value of the role of gas pressurization versus stress waves hasflippedback and forth over the years. In the late 1950s Hino [1] and Duvall and Atchison [2] emphasized the role of dynamic stress waves in the fragmentation process. This idea was fortified in 1966 by research results presented by Starfield [40]. Most of the later evidence seemed to point more towards the importance of the role of gas pressurization. Langefors and Kihlstrom [3] and Persson et al [4] in the 1960s, concluded that fragmentation blasting could be treated as a quasi-static problem. They felt that stress waves could be ignored and that fragmentation was only due to borehole pressurization. Kutter and Fairhurst [5] in the early 1970s hinted at the importance of stress waves in preconditioning the burden so as to make the borehole pressurization mechanisms more effective. In 1972 Persson [6] acknowledged the possible contribution of the reflected stress waves, but still felt that this contribution was only effective when large concentrations of high explosives were used. Hagen [7, 8] indicated that pressurization is the primary factor in fragmentation and advocated keeping the borehole pressures below a certain limit, so as to minimize the crushed zone around the borehole. This idea is also supported by Bligh [9], Persson et al [10], Melnikov [11] and Warpinski et al [12]. This control of pressure in the borehole has lately become known as Tailored Pulse Loading (TPL). Advocates of TPL recommend that not only the pressure amplitude but also the pressure rise rate be kept below critical values to enhance fracture propagation distances. This pressure control has mainly been introduced with the application in mind of oil and gas well explosive fracturing. In fact, the use of propellants rather than explosives in this technique is usually recommended. Bhandari [13], Barker and Fourney [14] and Winzer and Ritter [15] are the most recent researchers to acknowledge the importance of stress waves in the fragmentation process. Bhandari [13] utilized large block-models in an experimental program, and reported that by reducing burden it is possible to aid the fragmentation process because the reflected waves are made more dominant. Barker and Fourney [14] used models, which had controlled model-flaws introduced, and dynamic photoelasticity to demonstrate the importance of both small and largeflawsin increasing the role of stress waves in the fragmentation process. Winzer and Ritter [15] carried out field tests using high speed photography to assess the blasting results, and verified the findings of the small-scale model testing conducted by the University of Maryland.
Mechanisms of Rock Fragmentation by Blasting
41
More recently, Brinkmann [16] has used borehole liners to remove the gas pressurization factor from the blasting process. He feels that the resulting fragmentation is then seriously degraded. In the sections that follow, experimental evidence will be presented to attempt to show that in some situations gas pressurization is the most important factor, while in others stress wave fracture dominates the process.
2.3 STRESS WAVE MECHANISMS Several different reasonable mechanisms of failure have been proposed over the years to explain the fragmentation and fracture of rock by explosive loading. The first theory proposed falls into the stress wave category, and involves the reflection of outgoing stress waves from free surfaces. At the current time there is no agreement on how much of the energy that is released when an explosive detonates is converted into stress wave energy, how much is available in the high pressure gases, and how much is lost to other sources (such as temperature increase, air blast, fly rock, etc.). The percentage that is converted into stress waves is recognized by most to be quite small - no more than 20%. This of course will vary with the type of explosive being used, that is, TNT and like explosives are classed as high in stress wave energy and low in gas production, while ANFO and others are considered high in gas production and low in stress wave energy. In the original stress wave theory proposed to explain rock fracture and fragmentation Hino [1], in a series of papers, examined the interaction of the outgoing compressive stress wave with a free surface, and predicted how this reflected tensile stress would fracture the rock mass. Hino was looking mainly at crater blasting, and, assuming an outgoing pulse of triangular shape, predicted the number of 'slabs' and the thickness that would be broken in a given rock type. The thickness of the slab was given by t = (L/2)(SJPd)
(1)
where L is the length of the pulse, St is the tensile strength of the rock and P d is the peak stress in the impinging wave. The number of slabs was found from N = Pd/St
(2)
Duvall and Atchison [2], in a Bureau of Mines Report that was published shortly after Hino's papers, reported results from a comprehensive experimental study that among other things provided measurements of the strain pulses in four different rock types from various buried explosive sources. They then compared the fragmentation results with calculations using the theories proposed by Hino. Figure 1 shows the results for strain measurements made by Duvall and Atchison. A different explosive was used in each of the rock types, and the size of charge varied from about four pounds to a little over 20 pounds. The results represent a variety of explosive types, i.e. some were of the high stress wave generating type and some were of high gas-generating type. The results presented in Figure 1 have been altered to the extent that all values were normalized so as to give the same value of strain at a distance of five feet from the charge. The decay of strain with distance appears to be exponential - especially for the chalk and marlstone. A reasonable fit for the data appears to be strain (microstrain) = 31 250 exp( — 1.5d)
(3)
where d is the distance from the charge. The very high initial strains in the near vicinity of the borehole do, of course, decay at an even greater rate, due to the crushing of the rock in that area. After the strain levels become less than the compressive strength of the medium, equation (3) gives an estimate of how quickly the strain amplitude decays. This decay rate assumes that no fracture of the rock occurs due to compressive strains once borehole crushing is complete. In the theory presented by Hino and others the stress wave only begins to cause fracture and fragmentation after it travels as a compressive pulse to the nearest free face and is reflected as a tensile pulse, and then only insofar as it exceeds the tensile strength of the rock. This is the mechanism of fragmentation known as spall; it has been shown by some researchers to hold merit especially for very large charges detonated in close proximity to a free face (see, for example, the work of Bhandari). This theory of breakage, as presented, is based upon onedimensional wave theory, but in fact all blasting applications are three dimensional. The theory is also founded on the assumption of wave propagation in homogeneous isotropic materials.
42
Blasting 4.0 3.5 ~
r°
D + O Δ
l·-
3.0
=t 2:. 5 L
D
Granite Maristone Sandstone Chalk
fO
O 2.0 I .5
\-
è
6
â O 1 20 Distance from charge ( F T )
S
—±
Figure 1 Strain decay as a function of distance from charge in four rock types (data from Duvall and Atchison [2])
At the University of Maryland, an extensive series of two-dimensional model tests was conducted to investigate the effects of flaws on fragmentation. The objective was to determine the resulting damage in a brittle material due to the stress wave portion of the total energy released in an explosive detonation. In particular, the interest was in determining the fracture that would result in rock-like materials due to the outgoing and reflected stress waves. The resulting damage caused by all of the stress waves, not just the P (primary) wave, was studied. The technique used in the study was dynamic photoelasticity, which made it necessary to use model materials that were transparent and biréfringent. In the photoelastic technique the transparent model is viewed under special lighting conditions. The light source is monochromatic (a single wavelength), and by use of special optical elements the light is split into a fast and a slow component as it passes through the model. The difference in the speeds of the two components is a function of the stress state in the model. By recombining the two components after the light transits the model a series of dark fringes is obtained. These fringes represent lines along which the principal stress difference in the model is a constant; that is, the fringes represent lines of constant maximum shear stress expressed by radial stress — tangential stress = constant
(4)
By analyzing the fringes it is possible to determine at each and every point in the model the maximum shear stress. By placing the special optical arrangement in a high speed camera, dynamic stress patterns can be captured and stored on film for detailed analysis. For these tests a multiple spark-gap camera capable of taking pictures at framing rates up to one million frames per second was used. Each test produced 16 frames of the dynamic action during the event. This also provided the opportunity to view fracture initiation occurring at the same time as the stress pattern. Thus, it was possible not only to view the fracture process but also the mechanism responsible for fracture. There are two main disadvantages to this technique. First, the model is made of a transparent, biréfringent material which is brittle but which otherwise does not represent rock well. Second, the state of stress is one of plane stress rather than three dimensional. That is, the value of stress in the model in the direction perpendicular to the model is essentially zero. The results obtained therefore only give an indication as to what might be expected under similar circumstances in a threedimensional situation in a rock specimen. It should also be borne in mind that the models used were quite small, meaning that reflections from the boundaries occurred relatively quickly. Figure 2 presents a typical fringe pattern obtained in one of the dynamic photoelastic tests. The model size was 300 x 300 x 6.4 mm, and the size of charge only a few hundred milligrams of lead azide. The symmetric circular fringes represent the P-wave. Figure 3 presents the fringe order in the P-wave as a function of radial distance obtained in a typical test. The leading edge of the P-wave is to the right in Figure 3. Since the state of stress in the model is plane stress, it is possible to use equations from the theory of elasticity to solve for both stresses and strains from the photoelastic data taken in the test. Figure 4 presents the strains in the model as a function of radial position for the instant that the photoelastic data were taken (40 μ8 after the charge was detonated). Note that in the leading edge of the stress wave the radial strain is compressive and reaches a peak value of about 9000 microstrain while the tangential strain is always tensile and only reaches a peak value of about 2000 microstrain. The trailing part of the P-wave radial strain is tensile and reaches a peak
Mechanisms of Rock Fragmentation by Blasting
Figure 2
43
Dynamic photoelastic data from typical charge detonation in Homolite 100
P-wave character /=40.5/*s
_L
σ> o
40
Λ J_
\
60
I
.
80
\
100
_L
120
i Position, r (mm)
\
V
-5
Figure 3
/
Fringe order as a function of distance from the charge for a model test
o.009io.008 l·0.007 k 0..006 f0 005 H 0.004 h-
D Tangential + Radial
003 hf °002 H S 0.001 J— .£ 0. c .00!
I -o.002 oö -0003
-0..004 -0..005 -0..006 -0..007 -0 .008 -0 009 -0.
Figure 4
Radius (mm)
Model results for strains resulting from charge detonation
magnitude of more than 8000 microstrain. The stresses determined in the model are presented in Figure 5. Both the radial and tangential stresses in the leading edge of the P-wave are compressive, with the radial stress being five to six times larger than the tangential stress. In the trailing part of the P-wave both the tangential and radial stresses are tensile, with the tensile peak of the radial stress almost equal to the compressive peak value (about 6800 psi; 1 psi = 6895 Pa). These data were taken very early in the dynamic event. These stresses will change significantly as the fracture and fragmentation patterns continue to develop. It is likely that as these trailing tensile stress components continue to travel outward fracturing will occur (both radial and circumferential), so that stress wave energy will be consumed. The high tangential stress in the tail of the P-wave is the cause of the formation of the radial crack system, and the high compressive radial stress is responsible for the crushing that occurs around the borehole. This crushing will continue as long as the compressive stress exceeds the compressive strength of the rock being fragmented. As pointed out by Atchison and Duvall, the maximum value of the compressive stress that travels out into the medium past the crushed zone is fixed at the compressive strength value« This value varies with rock type and depends largely upon the condition of the joint and bedding sets in the rock mass.
Blasting
44 7
+
6
~
D
3
8. 2 O
D D
1
3 Λ
□ Tangential + Radial
+
+
5 4
+ D D
m
a
D
+
1-
+
OT - 2 -3 -4 -5 -6
50
D
D D
+ +
I
+
70
.1
90
Radius (mm) Figure 5
Model results for stresses resulting from charge detonation
Free surface
Charge
Figure 6
System of stress waves produced from P- and S-waves interacting with a free surface
The results of the model tests conducted at Maryland indicated that ifflaws(both large and small) are taken into consideration, then stress waves can account for far more fragmentation than would be predicted by the stress wave theory as presented in the early 1950s by Hino [1] and Duvall and Atchison [2]. Figure 6 presents a schematic representation of the system of stress waves generated from a point explosive source near a free surface. As shown in the figure, both a P-wave and an S-wave are generated. If in fact no fracturing occurred in the vicinity of the borehole, only a P-wave would form and travel out into the medium. But, as pointed out above, fractures do result at the borehole from a combination of the large compressive stresses in the leading edge of the P-wave and from the tensile tail of the outgoing P-wave. This fracturing destroys the symmetry at the borehole, and so a shear wave also travels out into the rock mass. The speed of the P-wave is about twice the speed of the S-wave. The deformation in the shear wave is primarily distortional, whereas in the Pwave the deformation consists of contraction and extension. When the P-wave interacts with a free
Mechanisms of Rock Fragmentation by Blasting
45
boundary at other than a normal incidence, both a dilatational and a distortional wave result. That is, a PP- and a PS-wave are generated in order to satisfy the requirement of a zero-force boundary condition at the free surface. These two waves travel back into the medium and interact with the outgoing S-wave. When the S-wave reaches the boundary it too forms two separate wave types, SSand SP-waves. Both of these travel back into the medium. As indicated by the early stress wave theory, there is also a sign change associated with the reflections from the boundary. That is, a compressive outgoing wave becomes a tensile ingoing wave. Figure 7 shows the result of the interaction of the PP-wave with two outwardly propagating radial cracks. In the first frame presented (taken at 101 μ8 after detonation), the radial cracks are traveling at a relatively high rate. The energy driving them is quite high, so that they are on the verge of branching. In the second frame, taken 18 μ8 later, the PP-wave has just passed over the crack tips. This changes the stress state dramatically, causing these cracks to turn and run in a circumferential direction. The energy contained within the PP-wave is quite high; as a result, as the cracks are driven in the circumferential direction they branch numerous times. As the PP-wave passes beyond the crack tips they once again turn and run in the radial direction, as dictated by the stress state arising from pressure in the borehole. The net result of this interaction of the PP-wave with the outwardly propagating radial cracks is therefore intense fragmentation. The final appearance of the interaction is as if the outward-traveling cracks had hit a 'barrier' (the PP-wave), spread out and branched. We therefore termed this mechanism 'barrier branching'. The outwardly propagating P-wave needs to reflect from a free surface in order to trigger this mode of fragmentation. The more remote from the charge that this surface is located, the less energy the PP-wave will contain, and hence the less intense will be the fragmentation. In a truly jointed and flawed rock mass the surface of reflection might be slightly open joints (or bedding planes) which might be located quite close to the explosive source. Hence the fragmentation caused by this mechanism can be very extensive. Many other mechanisms for fragmentation involving the interaction of stress waves in materials which contain smallflawswere observed in this series of tests. A summary of the results obtained is
Figure 7 Barrier branching caused by PP-wave pausing over outgoing radial cracks
46
Blasting
presented in Figure 8. As indicated in the figure, possibilities for nonradial crack initiation are uniformly distributed over the material mass between the borehole and the free surface. These include initiation offlawsof all orientations by the tensile stresses contained in the trailing portion of the outgoing P-wave as well as the stresses in the outgoing shear wave. In addition, very good opportunities for nonradial fracturing exist wherever the inwardly propagating waves interact with the outwardly traveling radial cracks. Also, the occurrence of spall-type fracturing near the free surface is one mechanism for fragmentation. The series of tests conducted also investigated to some extent what the effect of largeflawsmight be on fragmentation results. In this case strips of a transparent material (Homolite 100) were prepared by cutting a larger sheet with a band-saw and then routing the surfaces of the strips to make them relatively smooth. A typical model is shown in Figure 9. In the figure shown the model was built up by gluing together six of the strips described above. The glue used contained cyanoacrylate ester ('superglue') and the resulting joints were fairly weak. That is to say, if these models were handled carefully they would remain intact, but if dropped or bent the joints would separate. The model shown in the figure contained two boreholes, since the study conducted was investigating the effect of charge delay on fragmentation results. The results obtained by photographing the event in the multiple spark-gap camera are presented in Figure 10. The first frame presented (Figure 10a) which was taken 34 μ8 after detonation of thefirstcharge, shows the P- and Swaves traveling out from the borehole. An intense amount of fracturing is evident above and below Barrier branching in borehole cracks
Initiation zones Borehole
Assumptions. Cp/Cs =0.60 Radial crack = 0.2 CP P - wavelength = 0.58 S - peak lags front by 0 . 3 £
|PPand PS Shadow zone Free surface
Figure 8
Summary of locations where flaws can be initiated by stress waves (left) and locations where barrier branching occurs (right)
.i
1
(
,
b E m o ro c
*£ E
ç in
t i'
U-5 in (127 mm)·** ^
\**
Figure 9
IU in lÉ.wJt mm;
"""
)
15 in (381 mm
*.
Model geometry used in large flaw tests. Note that: (i) the boreholes are 0.25 in (6.35 mm) diameter; and (ii) the model consists of six 2 in (51 mm) wide, 0.25 in (6.35 mm) thick H-100 strips
Mechanisms of Rock Fragmentation by Blasting
47
the borehole at thefirstjoint. In the second frame (Figure 10b; taken at 62 μ8), this intense cracking pattern has occurred all along the joints just above and below the borehole. Two more frames taken later in the dynamic event are also presented in Figures 10(c) and 10(d). It appears from the observations conducted with the high speed photography and from analysis of the location of the stress waves that the resulting fracturing (which we call 'joint initiation fracturing') is caused by the shear stresses contained in the S-wave. The direction in which the fractures travel (very nearly perpendicular to the joint sets) indicates that they were initiated from shear stresses (mode II fracture loading). Similar fracture initiation was also observed in rock model tests by Hollo way et al. [17] and Winzer and Ritter [15], although not to the degree seen in the polymeric models. The results of the tests described above demonstrate that much more of the fragmentation can be attributed to the stress waves than anticipated by the classical theory of spall as proposed by early investigators. The tests also demonstrated that much of the experimental results of studies of fragmentation in rock, which utilize materials such as fine-grained granites or other rock types which do not contain joint sets and bedding planes, underestimate the amount of damage attributed to the stress wave as opposed to gas pressurization. This conclusion is not intended to indicate that the damage in rock blasting is due only to stress waves. As stated earlier, the author feels that both stress waves and gas pressurization contribute greatly to the fragmentation process and that truly efficient fragmentation results cannot be obtained unless both mechanisms are fully utilized. The natural mechanism which dominates depends upon the type of blasting being conducted. 2.4 GAS PRESSURIZATION MECHANISMS The argument for gas pressurization has also received considerable notice over the past few years. The main theory proposed notes that the pressurization that occurs is of considerable duration compared to the time of loading of the material by a passing stress wave, and thus holds that the
Figure 10 Fractures initiated at joints by outgoing stress waves, (a) Frame taken at 34 μβ. (b) Frame taken at 62 μβ. (c) and (d) are successively later frames.
48
Blasting
fragmentation process can be viewed as quasi-static. Among the major contributors to the gas pressurization theory are Ash [18], Porter [19] and Coursen [20]. Porter [19] analyzed the strain measurements made by Bureau of Mines researchers. He noted that for gauges located at some distance from the borehole, as the stress waves passed over the gauge location, the strains increased and then quickly returned to zero, as would be expected from a passing wave. At close-in gauges, however, he noted that the strains increased and did not return to zero over the several milliseconds of recording time of these gauges. He rejected the notion that this was due to plastic deformation of the rock surrounding the borehole since the same gauges were reported to be used again in later nearby tests without mention of recalibration. He therefore took this to imply that in the vicinity of the borehole the pressurization by the detonating explosive set up a load state that could be considered to be quasi-static, and he proceeded to investigate the fragmentation process from this standpoint. Porter presented both analytical and experimental evidence to show that fragmentation could be addressed in this fashion. Analytically, he showed that the direction of crack propagation under this quasi-static loading will always coincide with a principal stress trajectory. That is to say, cracking is caused when the tensile stress becomes large enough to overcome the strength characteristics of the rock, and the crack always propagates in a direction that is perpendicular to the maximum stress direction, i.e. along the principal stress trajectories. He used finite element methods to determine the stress trajectories for the geometry being considered, and compared those trajectories with the directions of crack propagation in model tests. Figure 11 presents some of the experimental results obtained by Porter. He used glass plates that were loaded by pressurizing the borehole with oil. The plates were taken to represent one slice perpendicular to the borehole. The pressure in the borehole was taken to a level near the critical value at which crack initiation and propagation would occur, and held there. A sudden increase in pressure was then achieved by using an exploding wire. This sudden increase was enough to cause two opposing cracks to initiate and propagate in the glass plates. In Figure 11 the directions of crack propagation are compared with the directions of the stress trajectories from the results of 14 tests. Porter also used high speed photography to view the cracks as they initiated and propagated along the stress trajectories. He found that the cracks initiated at the borehole wall and propagated outward towards the free surface. In the experiments as conducted by Porter only two opposing cracks were initiated at the borehole. In a truly dynamic event, more than two cracks would be initiated and the crater formed would be defined in shape by the cracks on the boundary side of the crater. The theory presented by Ash is similar to the one proposed in Porter's work, except that instead of looking at the stress trajectories he speculates on thefinalloading state. Ash'sflexuralrupture theory is based upon the idea that gas pressurization in the borehole loads the rock burden as a beam in bending. He views the borehole containing the long cylindrical charge to be a slab of rock which is pinned at its top and bottom but free to bend in the middle. Figure 12 shows the mode of bending which would occur with the flexural rupture theory. In this case the 'beam' is fixed at the top (the collar region) and at the bottom (the toe region) but is free to bend in the central region. This freedom to bend is afforded by the free surface and the freedom created by the borehole. This mechanism of fragmentation and that proposed by Porter are essentially the same. Porter was
Plate edge
Figure 11 Results obtained by Porter [19] showing crack paths in comparison to isostatics
Mechanisms of Rock Fragmentation by Blasting
49
looking at a two-dimensional view of the bending situation presented in the flexural rupture mechanism. In reality the situation as described by Ash should be enlarged to include more than one borehole, and the bending (or bulging) of plates rather than of beams should be considered. In any event, the gas pressurization mechanism as proposed by Ash and Porter is based upon a quasi-static loading of the borehole region by the high pressure gases that are generated as a result of the detonation of the high explosive. One serious difference between Porter's results and the flexural rupture mechanism is that Porter found that the fractures initiated at the borehole and traveled out towards the free face, while in the flexural rupture theory the fractures initiate at the free face (at points of high tensile stress) and travel inward towards the borehole. In his experimental studies, which included some testing with propellants, Porter also found that the explosive gases did not enter, or entered only for a small distance, the cracks that were initiated at the borehole. In the gas pressurization mechanism of fragmentation proposed by Coursen [20] the assumption is made that the gases produced during the detonation are able to enter and pressurize the radial cracks in the borehole region. The pie-shaped segments which are bounded by the borehole and two radial cracks, as shown in Figure 13, are therefore loaded by high pressures on the two sides formed by the radial cracks. According to the Coursen mechanism this pressure is sufficiently high to cause fragmentation of the rocks due to clamping pressure. The fragmentation therefore is the result of high compressive loading, just as is the crushed zone around the borehole. The fragmentation is not as severe in this outer zone since the compressive stresses are not as great as they are in the immediate vicinity of the borehole. Wilson [21], in a Ph.D. dissertation completed in 1987 at the University of Maryland, investigated the mechanisms responsible for fragmentation in a quarry-blasting situation. He conducted both an experimental and an analytical investigation. In his experimental program, tests were run in models made of small aggregate, low water-content concrete, in both single and multiple borehole configurations. His models were quite large, being about 2 m x 1 m x 1 m. His tests were highly instrumented with accelerometers, crack detection gauges, pressure gauges, velocity of detonation gauges and surface strain gauges. In addition he used a high speed camera to photograph the front and top faces of the quarry-blasting model. These surfaces were painted with grid markings so as to make it possible to follow more easily the deformations that were occurring after the explosive was detonated. Figure 14 shows a typical multiple borehole configuration tested by Wilson. For his analytical work he included a hybrid experimental-computational approach. A finite element program was used to compute the stresses and nodal displacements in the model, while crack velocities, gas penetration rates and other parameters needed in advancing the model configuration for the next quasi-static step were based upon values determined from the laboratory experiments. The mechanisms proposed by Wilson were comprehensive and included some of those recognized by others as described above, but also included some new ideas. According to Wilson the following events occur in a quarry-blasting situation. A tensile stress is produced at the free face due to the reflection of the P-wave. This tensile stress is not the same as the tensile stress referred to by Hino [1] and Du vail and Atchison [2]. This tensile stress is due to the £/Z_=l/3
Borehole center- line
Borehole center - line
Borehole center - line
Figure 12 Flexural rupture mode of failure as described by Ash [18] (B/L, slenderness ratio; L, bench height; B, burden distance; Γ, stemming length)
50
Blasting
Figure 13 Schematic of failure mechanism as proposed by Coursen [20]
Mortar joints, 2 cm thick, typical
Figure 14 Test set-up used by Wilson [21] in his fragmentation tests
curvature in the P-wave front, and results in vertical cracks forming at the free face and propagating back into the rock towards the borehole. This tensile stress, which was verified by Wilson in his experiments, is maximum for normal incidence. According to Wilson, a similar tensile stress can also result in fractures forming on the free surface above the borehole at the top of the bench. He observed the formation of these cracks in his high speed photographic results. These fractures occur before the spall fractures of the Hino theory and are orthogonal to them. They also would be
Mechanisms of Rock Fragmentation by Blasting
51
perpendicular to the fractures which would be formed according to Ash's flexural rupture theory. Wilson points out that increased burden would have two adverse effects on the formation of these fractures. First, the amplitude of the stress in the P-wave decreases with increased burden (as pointed out by Bhandari [13]); and second, as the burden increases the curvature of the P-wave front decreases, thereby decreasing the tensile stresses at the free face upon reflection. This is a new mechanism of fragmentation that had not been identified prior to Wilson's work. Wilson's work also showed that stresses imposed by the gas pressurization tended to open and propagate radial cracks in a preferential fashion depending upon their orientation. The cracks most forcefully driven by the gas pressurization were identified to be those that are also most effectively advanced by the reflected stress waves. These cracks are ones that lie to either side of the borehole, propagating to the free face at angles between 20 and 30°. That is, the breakout 'wedge' would have an included angle of between 120 and 140°.These preferred cracks are forced open even if they are not internally pressurized, which in turn subsequently increases the gasflowinto them. Wilson also points out that there is a strong tendency for other radial cracks to be closed, especially those in front of the borehole between the two dominant radial cracks. In this region the material experiences a biaxial compression during the pressure-loading period (similar to the mechanism proposed by Coursen). The state of stress at the free face in front of the borehole is tensile, while at the two sides of the borehole at the free face the stress is compressive. This is similar to the flexural rupture mechanism as proposed by Ash but is oriented at 90°. Wilson also indicates that it is unlikely that any new cracks would be initiated during the later period of gas pressurization (his analytical model was homogeneous and isotropic). Even though the gas pressurization phase acts over a much longer period of time than the stress wave stage, the tensile magnitudes in the stress waves were found to be higher than those created by the later gas pressure. This is because the initial detonation pulse that is the source of the stress wave amplitude has a much higher pressure than the subsequent explosion gas pressure in the borehole, and also because the amplitude of a quasi-static stress state attenuates with distance more rapidly than does the amplitude of a stress wave. For a material with joints and bedding planes, the bending process itself should initiate and generate additional fractures. Many more authors than those mentioned above have contributed to the various mechanisms of rock fragmentation as outlined. Ash, for example, indicates that the concept offlexuralrupture is not newly discovered by him, attributing the original reference to Daw and Daw [22] in 1898. However, the main champion of the flexural rupture theory in recent years has been Ash. None of the mechanisms presented above are adequate in explaining fragmentation by explosive loading if it is considered to be the only mechanism that is active. Unfortunately, it is not easy to combine the mechanisms since there are discrepancies between the theories. In the following sections, examples of blasting results will be presented and discussed which tend to support one or other of the mechanisms outlined above. Examples will be presented to show particular instances in which either gas pressurization or stress wave mechanisms are dominant. In some cases, results will be presented which make it evident that steps can be taken to make both gas pressurization and stress wave mechanisms interact, and hence optimize the blasting results. 2.5 CRATER BLASTING Crater blasting is a situation in which there is only one free face in the vicinity of the charge; it is often studied both in the laboratory and in the field. A common crater-blasting use is as the first charge fired in a tunneling situation. It has also been extensively studied in research programs dealing with the formation of underground retorts for modified in situ removal of kerogen from oil shale. The classic description of cratering involves the use of spherical charges, but in practical applications the amount of explosive needed to create the desired size of crater has required the use of cylindrical charges. Crater blasting has also been used as an aid in the study of quarry blasting, where two free faces exist - one parallel and one perpendicular to the charge axis. It has also begun to receive renewed attention due to the possiblility of using explosives on the moon or on other planets for the rapid construction of permanent or temporary shelter for personnel during space exploration [23]. The author took part in a very large research program in the early 1980s to investigate the removal of kerogen from oil shale. The program was sponsored by a consortium of oil shale companies, and was conducted in the Anvil Points oil shale mine near Rifle, Colorado. This mine is owned by the US Navy, as a part of their oil reserve plans. In this program, which was supported by technical assistance from the Los Alamos National Laboratory (LANL) and by Sandia National
52
Blasting
Laboratory (SNL), the intent was to investigate the formation of large underground retorts by blasting to permit removal of the oil without taking most of the shale to the surface. It was hoped that the experimental program would be able to define the mechanisms of fragmentation to the extent that blasting results for the formation of future retorts could be predicted by computer codes. The first step in this test program involved an intensive study of crater blasting. In addition to tests conducted under this program, earlier tests had been conducted by LANL in the Colony mine, located a few miles from the Anvil Points facility. Both LANL and SNL then conducted cratering tests after the conclusion of that earlier program at the Anvil Points site. (Anvil Points was also one of the test sites used by Atchison and Duvall in their original experimental work to study the stress wave mechanism of fragmentation, described in Section 2.3.) In all, 27 single-charge cratering tests were conducted in the Anvil Points and Colony mines. The results of those single-charge cratering tests were disappointing. In most of the tests the crater produced was much smaller than would have been desired. The bottoms of the craters in most of the tests (which were completely excavated) were located at or near the top of the cylindrical charge. Prompted by these disappointing results, a series of model tests was conducted at the University of Maryland in the hope of shedding more light on the mechanism of fracturing that operates in crater blasting. The results of the test series have been published [24], and indicate that the mechanism of fragmentation is dominated by stress wave generated fracture. Three different materials were used in that study: PMMA (Plexiglas, Perspex), hydrostone and granite. Hydrostone is a fast-setting gypsum cement; other than some plasticity in the immediate borehole vicinity, it has been shown to respond to explosive loading in a fashion similar to finegrained rock. Figure 15 shows one frame of 16 pictures taken during a test on one of the PMMA models. This picture was taken in a multiple spark-gap camera; it shows the event about 100 μβ after the small cylindrical charge was detonated. At this time the radial fractures have already formed, and gases generated by the explosive are being ejected from the top of the borehole. The item of extreme interest is the deformation occurring on the top surface of the model. The wave shown is visible to the naked eye and is similar in appearance to that of a pebble dropped into still water. The wave is traveling outward from the top of the borehole and from the velocity of propagation appears to be either a shear wave or a Rayleigh wave. The fact that it is visible indicates that the displacement
Figure 15 High speed photograph showing surface wave system generated on top surface by detonating cylindrical charge
Mechanisms of Rock Fragmentation by Blasting
53
perpendicular to the surface is quite large. The location of the wave indicates that it might be the result of a stress wave which has traveled up along the inside of the borehole and then propagated along the free surface, or that it is the surface wave that has resulted from the P- or S-wave reflecting from the free surface. Figure 16 shows two scaled depth of burial versus scaled volume curves obtained in the test series one for PMMA and one for hydrostone. Such curves are frequently used to describe the results of crater blasting, and are used to determine the optimum depth of burial. The procedure is as follows. A given charge is buried at a given depth and the resulting crater volume determined. The same size charge is then buried at a different depth and the crater volume once again determined. The depth which produces the maximum volume is defined as the optimum depth of burial. The scaling factor with regard to the crater volume is the amount of explosive used (in this case in grams), while the scaling factor for the scaled depth of burial is the cube root of the amount of explosive used. In the particular series of tests described in Figure 16 all charges were the same size, so scaling is not necessary. The scaled volume (SVOL) is the amount of material removed divided by the amount of explosive used; the scaled depth of burial (SDOB) is the distance from the free surface to the center of the charge divided by the cube root of the amount of explosive used. If gas pressurization plays a major role in the cratering then the results obtained should be different for stemmed and unstemmed charges. The points in Figure 16 are labeled with either a 'u' (for unstemmed) or an V (for stemmed). It appears from Figure 16 that any difference between the stemmed and unstemmed tests is less than the natural scatter. This similarity between the stemmed and unstemmed results could simply mean that the stemming used was ineffective and had no effect on the gas pressurization within the borehole. In an effort to investigate this possibility, measurements of the pressures in the borehole were conducted in the models being used. Figure 17 shows the model used in the pressurization tests. PMMA models were instrumented with Kistler transducers capable of dynamic response in the cross hole intersecting the borehole (Figure 17). A 300 mg charge of PETN was used in both tests - one stemmed and one unstemmed. The pressures recorded in the two cases are shown in Figure 18. As is evident from the figure the pressure measured in the stemmed case was four times greater than the pressure measured in the unstemmed case. The duration of the pressures in both cases was about the same, 300 μβ. Another possibility for the similarity between the stemmed and unstemmed cases could be the creation of a standing shock wave at the borehole top in the unstemmed case, which would cause a pressure increase equal to the pressures that result when stemming is present. That is, as the high pressure explosive gases exit to the atmosphere, a shock wave is set up at the mouth of the borehole to give an effect similar to stemming. To investigate this possibility another test was conducted, this time using an aluminum model. Aluminum was chosen so as to eliminate any reduction in pressure that would occur as a result of volume increases due to fragmentation of the model. The model used was similar in geometry to the PMMA model used in the pressure tests for the stemmed and unstemmed charges. Figure 19 shows the geometry of the aluminum model used. In this case two transducers were used, one located 19 mm below the free surface and the other located 19 mm below the first. The results obtained are presented in Figure 20. The lower curve shows the pressure recorded nearest to the charge, and it agrees well with the pressure curve for the measurements taken at about the same location in the unstemmed case (Figure 18). As can be seen
40
60
SDOB (mmg J 6)
40 60 SDOB (mmgJ/3)
Figure 16 Scaled volume (SVOL) versus scaled depth of burial (SDOB) for model testing of (a) Plexiglas and (b) hydrostone (u = unstemmed, s = stemmed)
54
Blasting
*(mm) Y (mm) CT - 38 unstemmed 61 127 C T - 4 0 stemmed 52 95
Figure 17 Test geometry used to determine pressure in borehole (stemmed versus unstemmed)
^Plexiglas, stemmed
Figure 18 Comparison of pressures in Plexiglas models with and without stemming
from Figure 20, there is an increase in borehole pressure as the free surface is approached. This pressure increase is about a factor of two instead of the factor of four observed in the stemmed case. Hence, if gas pressurization plays a major role in fragmentation in the crater-blasting case, significant differences should be evident in the size and geometry of the craters obtained. Figure 21 presents a crater profile obtained from one of the model tests conducted in the cratering series described above. This was typical of the tests conducted in the hydrostone models. The plasticity effects mentioned earlier are evident in the vicinity of the charge. The 20 mm cavity shown is a result of this behavior, which is not so evident in the examination of craters in more brittle rocks. Other than this, however, the crater profiles obtained in the model tests compare well with the craters obtained from the tests conducted in oil shale - especially those that resulted in poor craters. Figure 22 shows the comparison between one of the model tests (CT-4) and one of the poorer craters obtained in oil shale, test SB-1. This was one of the poorer performing tests and appeared to result in
Mechanisms of Rock Fragmentation by Blasting o-
55
A
19 mm Upper £,< Transducers Lower
19 mm _f
,*-j
&'
6 mm5mmi
11 I mm I I I
73 mm
|
125 mm 300 mg PETN
50 mm - 50 mm ·
Figure 19 Aluminum model used to determine pressure variation in borehole (unstemmed) as free surface is approached
150
200
250
300
Time (/xs)
350
400
450
500
Figure 20 Pressures measured in an open borehole (aluminum model test). Top curve near the free surface
one of the smallest craters. Other than the very wide spall failure in the model test, CT-4 and SB-1 appear to have produced craters very similar in scale. CT-4 in fact performed better from the standpoint of depth of pull compared to location of the charge. The model tests conducted at Maryland are felt to represent to a reasonable degree the results obtained in the field. The model test series points out other features which pertain to the mechanism of crater blasting, in addition to the fact that it seems to be stress wave dominated. It appears that the radial fractures form well before the time that the fractures defining the actual crater form. This is evident after viewing Figure 23, which is a photograph of a crater formed in a hydrostone model. Note in particular the discontinuity marked in thefigure:this was formed by a radial fracture. The fractures making up the crater on the lower side of this radial fracture are deeper (farther from the surface) than the cone-type fracture above the radial crack. This strongly implies that the radial fracture
56
Blasting
5.3 mm
Figure 21 Crater profile obtained in hydrostone model
Surface
CT-4 SB-I
SB - I Full-scale tests in oil shale C T - 4 Small-scale test in hydrostone
S B - I Charge center
- Explosive column All charges bottom detonated
Figure 22 Comparison of crater slopes obtained in model tests (hydrostone) and field (oil shale)
formed before the crater fracture. It also appears from these tests that the crater starts from the borehole and propagates towards the free surface rather than the other way around, as is implied from a spall mechanism theory, and this is substantiated by the crater shown in Figure 23. If the crater started from the surface and worked its way back towards the borehole, the discontinuity evident in Figure 23 could not have occurred. Tests which used PMMA as a model material are even more convincing. In the case of PMMA models, whenever a stress wave passes over the tip of a propagating fracture the direction of propagation changes momentarily, and the outline of the location is 'marked'. These marks are called ripple marks in fracture mechanics, and are frequently used to determine crack velocities if the times of passage of the stress waves are known. In the case of crater blasting these markings are always concave towards the borehole. Since a fracture propagates faster in the interior of a material than it does on the surface (plane strain versus plane stress conditions), and since the radial fractures form first, this implies strongly that the fractures which form the crater start at the borehole and work their way to the free surface. Thus it appears from both model and field tests that: (i) crater blasting is strongly stress wave dominated; (ii) crater blasting is very inefficient; (iii) radial fractures form well ahead of the craterproducing fractures; and (iv) the crater starts at the borehole and works its way towards the free surface. Knowledge of these facts can be used to make crater blasting more efficient in ways that will be described below.
Mechanisms of Rock Fragmentation by Blasting
57
Figure 23 Top view of crater obtained in hydrostone model. Note discontinuity caused by radial fractures forming first
One way to increase the efficiency of the crater blasting process would be to force the crater to initiate at a greater depth from the free surface. This can be done by introducing a flaw of proper size at the location where it is desired to have the crater initiate. Flaws which normally are located on the borehole wall might be the result of imperfections along the wall due to the drilling process itself, or might occur naturally due to the structure of the rock, such as locations of bedding planes, etc. A series of model tests was conducted to investigate further the possibility of initiating the location of the crater (Wang et al [25]). In these tests PMMA models were used in conjunction with circumferential grooves located at various positions along the borehole. The geometry of the model is shown in Figure 24. The groove was placed in the borehole wall using a broaching tool in a milling machine, cutting the groove to a depth of 1.5 mm in a borehole of 6 mm diameter. The broaching tool had a 30° included angle. The groove location was varied from the bottom of the borehole to just above the top of the Charge (at the location where the crater was observed to form naturally). The charge size and geometry were kept constant (600 mg of PETN in cylindrical form - 19 mm long by 6 mm in diameter). In all cases the crater was forced to form at the groove location. In the situation where the groove was two-thirds of the distance along the charge (from the bottom), the volume of the crater created was six times larger than the crater that formed when no groove was used. In cases where the groove was located closer to the bottom of the charge, even though the crater was initiated at the desired location there was insufficient energy to drive the potential crater all the way to the upper surface. In some cases, where the location of the groove was very near the bottom of the charge, the 'crater' began to form but ran to the bottom of the model (even though the bottom was further away than
58
Blasting
102 mm
1.5 mm Groove details
Figure 24 Model geometry used in grooved crater model tests
the top surface) or exited the sides of the model. The results of these tests are encouraging, but for the explosive and stemming being used this technique was not successful in creating craters which initiated near the bottom of the borehole and were capable of traveling all the way to the upper surface. This was felt to be due to the explosive's gases not getting quickly enough into the fractures forming the crater. That is, the gas pressurization was relieved quickly through the borehole and did not pressurize and drive the fractures forming the crater to the surface. Young et al. [26] have overcome the inefficiencies associated with crater blasting by altering both the geometry of the borehole and the type of explosive. In an effort to develop new bedrock removal techniques for use in the tunneling and mining industries, they developed a method which more fully utilizes the gas pressures associated with explosive detonation. In this technique short, squat boreholes, which place the pressure source in close proximity to the free surface being blasted, are used. The technique is called 'penetrating cone fracturing', and the geometry used is shown in Figure 25. The technique was first explored by model testing in PMMA models, conducted by the Photomechanics Laboratory at the University of Maryland. Those tests showed that, for the proper borehole geometry and type of charge, fractures could be initiated at the bottom of the borehole which would initially propagate away from the free surface but then turn back towards and intersect the free surface - removing a substantial volume of material in the process. In the laboratory the technique was found to work best if the borehole was drilled such that a sharp borehole bottom resulted. It was also found that it was best to use a propellant as opposed to an explosive. The propellant was found to provide a pressure rise rate high enough to initiate the fracture and enough gas pressure such that the fracture would continue to propagate until it intersected the free surface. In later testing conducted in rock, best results were obtained when a heavy inertial bar that just fitted into the borehole was used to stem the flow of gases from the borehole. The geometry of the borehole that was found to produce optimum results was one which had an aspect ratio in the range of three to four, i.e., the borehole length was only three to four times the hole diameter. The technique has been refined in field tests and efficiencies that are typically three to four times greater than the best drill and blast results have been obtained. In rock it was found that normal percussion drilling resulted in boreholes with small microcracks located at the bottom of the hole. These microcracks proved to be more efficient than sharpened holes, the best of shots removing cones of rock with the diameters of 1.22 m and depths of about 125 mm, using only 20 g of gunpowder. It was also found to be more efficient to place the charge in a 'gun' located outside the borehole with its barrel placed within the borehole. The 'gun' was held in place with an inertial stemming bar, ensuring that the pressurization of the borehole lasted for a long enough period to permit full
Mechanisms of Rock Fragmentation by Blasting Hole stemming ( Inerîial mass )
59
Decoupled charge ( Propel lant )
at sharp hole bottom Rapidly pressurized Dore hoie
Figure 25 Cone fracturing geometry (Young et al. [26])
pressurization of the fractures forming the cone. An added advantage of the penetrating cone fracture technique is that the size of charge being used permits work to continue at the face as long as personnel are protected by a metal shielding plate. Further development of the technique calls for this type of fracturing to be incorporated into a small-charge robotic miner, wherein it will not be necessary to evacuate the face, as is necessary in the typical drill, load, shoot and muck cycle that is currently used for tunnel blasting in hard rock. This is a good example of making blasting more efficient by taking advantage of the more potent mechanisms of gas pressurization, rather than the less efficient mechanism of stress wave fracturing. In the cone-type fracturing observed in PMMA models in the laboratory there was little evidence of any contribution from stress waves. 2.6 CONTROLLED FRACTURING In some applications it is desirable to exercise precise control of the fractures that result from the use of explosive charges. Two examples are the stimulation of oil and gas wells, and in presplitting or postsplitting situations. In presplitting and postsplitting applications it is desired to be able very carefully to create a plane in the blasted area beyond which no fractures propagate. For oil and gas wells the object is to drive fractures long distances into the pay zone in a controlled fashion. This section examines in some detail the research into the means of propagating fractures in predictable directions. 2.6.1 Oil and Gas Well Stimulation In the extraction of oil and gas from drilled wells it is common for the recovery rates from a drilled hole to be low. This is especially true in low pressure reservoirs, due to the fact that only a relatively small number of natural fractures intersect the wellbore. In many cases the drilling process itself tends to introduce a skin on the wall, impeding theflowinto the wellbore. Two common techniques for increasingflowinto the well are hydraulic and explosive fracturing. In the early days of explosive fracturing, small charges of nitroglycerine were used to introduce new fractures to connect the borehole with the natural fractures in the reservoir rock. As these charges were made larger in an effort to drive the fractures further into the rock, mixed results were obtained. In fact, in many cases production was actually decreased by the use of explosive fracturing in wells. Hydraulic fracturing also has drawbacks: it tends to be complicated and expensive to use, and it requires a large amount of support equipment. It requires, as well, a 'proppant' to ensure that the fractures, once created, will remain open after the hydraulic pressure is removed. Considerable work has been conducted in an effort to make the explosive fracturing of wells more attractive. Of particular interest is the work conducted under a Department of Energy program within the United States through the Eastern Gas Shales Program. In experimental work conducted by Sandia National Labs [27, 28] and by the University of Maryland [29, 30], as well as in earlier work by Bligh [9], it was recognized that if the process was too dominated by the stress wave system then the technique would be unsuccessful. If a high energy explosive is used in the normal fashion,
60
Blasting
then the resulting fracture system is nearly all the result of stress waves. The normal procedure in using a high energy explosive would be to drill the well to the desired depth (the pay zone), implant the explosive in the bottom of the well and then detonate the explosive. This results in very large amplitude stress waves being sent out into the rock mass, and the gases generated are expelled through the top of the well. The stresses sent out into the rock crush and pulverize the rock mass in the immediate vicinity of the borehole. In this situation there is no free face involved since the wells are hundreds of feet deep. As the stress waves move away from the well, their amplitudes very quickly decay to point where the tensile tails become too small to initiate cracking. Due to the high compressive stresses in the leading edge of the stress wave, it was postulated that, after the borehole expands and then relaxes as the compressive wave passes out into the medium, a compressive stress cage is set up around the borehole. This residual stress cage preventsflowof oil and gas back into the borehole. Understanding this mechanism has led to a type of blasting which has been termed Tailored Pulse Loading' (TPL). With TPL it is found to be advantageous to keep the stress amplitude and the pressure rise rate within the borehole below certain limits. The pressure rise rate has been shown by researchers at Sandia to control the number of fractures that form at the wellbore, while control of the stress amplitude reduces the likelihood of the formation of a stress cage. The common method of achieving this control on both stress amplitude and pressure rise rate was to substitute the use of propellants for explosives. The propellant deflagrates rather than detonates, and thus both the pressure rise rate and stress amplitude are checked. However, propellants are both expensive and tricky to use, the latter because, for most propellants, the pressure control within the propellant must be correct or detonation instead of deflagration will occur. In the well stimulation research conducted at the University of Maryland under the DOE program, our approach was different from others. We chose to use explosives and control both the stress cage effects and the pressure rise rates by geometric changes in the blasting configuration. In a series of model tests using PMMA and the multiple spark-gap camera, it was found that when an open section of borehole was present above the charge then the gases generated by the detonation of the explosive could be contained within the wellbore. Figure 26 shows the model configuration used in the test. The explosive was placed in the bottom of the borehole and a stem was placed at the top of the borehole near the free surface. When the explosive is detonated a gas shock wave travels up the 12.7 mm diameter grooved borehole
i l^l.
Ww^v 25 25
"Charge location
-152 mm ■
Figure 26
Model geometry used in stem-induced fracturing tests
Mechanisms of Rock Fragmentation by Blasting
61
borehole and interacts with the stem. The pressures associated with the traveling shock wave are then reflected back into the borehole with the same sign as the pressures in the upward-traveling shock wave. The pressure at the stem is therefore at least twice the pressure for the case when the stem is not present. Pressure transducers were used to monitor the pressures in the vicinity of the stem. Due to the compressibility of the medium in the borehole, pressures at the stem were measured to be in excess of 2.5 times the pressures at the position in the borehole away from the stem. The rise time of the pressure at the stem was found to be longer than that in the immediate vicinity of the charge. This is due to the attenuation of the pressure in the shock front as the shock wave propagates upward from the charge. All of the crushing and the stress cage effects due to the detonation of the explosive occur in the immediate vicinity of the charge. Figure 27 shows the dynamic event as recorded by the multiple spark-gap camera after the detonation occurs. Immediately upon detonation, fractures begin to form and propagate at the charge location. These fractures slow and begin to arrest at about 140 μ8 after detonation. At about the same time (for the model geometry shown), fractures begin to initiate and propagate in the vicinity of the stem. For the test being described, the borehole wall was grooved longitudinally to ensure that the initiated fractures would be such that the fracture planes formed would be parallel to the front surface of the model. This enabled the fracture surface to be viewed as a plane. (The grooving also resulted in the fractures being initiated at a lower pressure, but other model tests as well asfieldtests indicated that grooves are not necessary for such fractures to form). These 'stem-induced' fractures continued to propagate and in this case, as in most of the model tests conducted, resulted in the model being cleaved in half along the central plane.
Figure 27
Four frames from a stem-induced fracturing test. Fracture at charge location quickly arrests, fracture at stem cleaved model in two
62
Blasting
Figure 28 Picture of fractures resulting from stem-induced fracturing of an oil well in western Pennsylvania
In a field situation the desired operation calls for the well to be drilled deeper than needed, so that the explosive charge can be placed well below the pay zone, and for the stem to be located at the top of the pay zone. Upon detonation of the explosive, the shock wave travels to the stem and initiates fractures which connect the natural fracture system to the wellbore. However, these fractures are not closed off by a stress cage effect after the pressure in the borehole decays. All of the unwanted damage is confined to the charge vicinity where it cannot harm the production capacity of the well. This technique has been field tested by Young et al. [31] in stripper wells in Western Pennsylvania and has proven to be successful. Figure 28 shows a picture taken by Young with a specially designed 35 mm downhole camera. As is evident from the photograph, which was taken at a location in the borehole below the stem, the fractures produced are quite wide and no other apparent damage to the wellbore has occurred. In this case a predominantly stress wave controlled blasting situation has been transferred into one that utilizes the large amount of energy present in the gas pressurization phase, and has resulted in much more efficient use of the explosive. Also, nearly all of the energy in the explosive is used to initiate and drive the fractures. After detonation everything is retained in the borehole, and it is necessary to drill back through the stem to relieve the residual gas pressures still present in the wellbore.
2.6.2
Fracture-controlled Blasting
The resulting crack pattern from an explosive detonation normally is a dense radial pattern around the borehole. This crack pattern is randomly oriented and very little control of the fracture plane is achieved. Where control is desired this normal blasting procedure is modified. Presplitting and postsplitting as well as smooth-blasting procedures have been developed that to some degree can control the fracture process. In presplitting a row of closely spaced and highly charged holes are detonated simultaneously. The resulting stress waves interact to produce more cracking in the region between the holes where the stress waves overlap and increase the dynamic stresses. These highly charged holes also produce extensive cracking at the borehole and weaken the resulting excavations. The simultaneous detonation also results in excessively high ground shocks which are undesirable for excavations in urban areas. In presplitting, the row of closely spaced holes is detonated before the other rounds being used in the excavation. Postsplitting is almost identical to presplitting, except that the control holes are fired not before but after the other rounds in the excavation. In presplitting the round is more tightly confined than in postsplitting but more
Mechanisms of Rock Fragmentation by Blasting
63
protection is provided by shielding the remaining rock from the propagation of unwanted fractures into it. In smooth blasting the holes are drilled on very close centers and cushioned charges are used. Smooth blasting gives satisfactory results when enough holes are drilled and when the charge is properly cushioned; however, the large number of holes required increases the cost of excavation due to the increased drilling and loading costs. Work conducted at the University of Maryland by Ladegaard-Pedersen et al [32] using dynamic photoelasticity showed that truly to achieve fracture control the loading applied to the borehole had to be altered significantly. Any guiding of the fracture control crack is accomplished by interaction of either the stress waves and/or the gas pressurization, and it was shown that the randomly oriented cracks produced immediately after detonation of the explosive cause damage in the area beyond the desired limits of the excavation before this interaction occurs. The solution to the problem was a scheme which controlled not only the initiation sites of the fractures from the borehole but also the direction of the subsequent propagation of these selected fractures. The initiation sites can be controlled by a careful selection of the size of the charge in the borehole, coupled with notching the borehole. This notch was also found to control the direction of crack propagation during the period of interaction of gas pressurization and stress waves between boreholes. From the principles of linear elastic fracture mechanics it is evident that the pressure amplitude required to initiate fractures is a direct function of the character and number of flaws present in the vicinity of the borehole. Figure 29 presents results obtained by Ouchterlony [33] for the stress intensity factor which exists at the tips of cracks emanating from a pressurized hole of radius R. As can be seen from thefigure,Ouchterlony has developed solutions for cases where various numbers of cracks have initiated at the borehole. These are static solutions but are assumed to yield valid results when used in the case of explosive loading. This curve can be used in two different fashions; (i) it can be used to give a maximum value of pressure based upon the natural flaws present at the borehole (the grain size in this case); and (ii) it can also be used to determine the size of the groove necessary on the wall of the borehole to ensure that fractures are initiated. Notice that for very short cracks the solution is independent of the number of cracks. If the initiation fracture toughness for the rock being blasted is known then the stress intensity at the tips of the cracks can be compared with this fracture toughness and a range of pressures for successful crack initiation can be determined. The results of Ouchterlony can be used to compute the pressure required to initiate cracks at the tips of sharp notches on the borehole wall, as shown in Figure 30. Two facts are evident from the figure. First, the borehole pressures required to initiate cracks are quite low even for very shallow notches. This implies that high explosives with detonation pressures above 200 kbar give overpressures which are too large by a factor of 30. The second result is that cracks can be initiated at notches with low pressures for all rock types. The rock property that is important is the fracture toughness of the material Klc. Limited data that exist indicate that this value ranges between 0.18 and 15 MPam 1 / 2 for most types of commonly excavated rock. Using limestone as an example, Figure 30 shows that a crack can be initiated from a notch 0.5 mm (0.02 in) deep if the pressure exceeds 10.3 MPa. The same notch in granite would require 24.8 MPa.
Figure 29 Stress intensity at the tips of cracks from a pressurized borehole (Ouchterlony [33])
64
Blasting 80 70 60 50
1 Q.
gT 40 30 20
jg fc. È
S
10 (x10 3 psiin 1 / 2 ) (MN m3/2) Fracture toughness, K,
Figure 30 Minimum pressures required to initiate fractures in rock
Figure 31 Maximum permissible pressures for fracture-controlled blasting
Another factor that must be considered is the possibility of overpressure. The pressures in Figure 30 are minimum pressures that should be exceeded to ensure initiation. The amount of overpressure that can be tolerated can also be determined from Ouchterlony's results. If the natural flaw length, af, is equated to natural flaw size, then the same expression can be used to determine maximum pressure. For natural flaw sizes assumed to range from 0.025 to 0.25 mm, the maximum pressures that can be tolerated are presented in Figure 31. This figure shows that very fine grain rock materials with flaw sizes of 0.025 mm support much higher pressures prior to random crack generation than do coarse-grained material. The pressure range in which crack initiation can be controlled will depend upon three factors: the fracture toughness X lc , the natural flaw size af and the depth of the side notches a. The allowable pressure ranges for several sets of operating conditions are presented in Table 1. This theory was investigated both in the laboratory, using models made of either a biréfringent polymer or rock-plates and in the field. Figure 32 shows a typical set of dynamic pictures taken during a fracture control experiment and the model used in the tests. Eight of the 16 frames taken are shown, covering a period of 677 μβ after detonation of a small lead azide charge. Unsymmetrical loading, due to unequal packing of explosive around the through-bolt (used to retain the gas
65
Mechanisms of Rock Fragmentation by Blasting
pressurization of the hole), is evident in the first picture, taken 47 μ8 after detonation. The P-wave at this time has propagated far out into the model, while the S-wave is visible near the discs on the pressure containment device. The horizontal line along the diagonal indicates the specified fracture control plane (the direction of the groove). In the second frame presented, cracks are evident on the diagonal, about 10 mm beyond the pressure cap. The cracks propagating along the fracture control plane are more easily observed in the frame taken at 102 μβ. The reflected PS-wave, which is generated by the reflection of the P-wave from the model boundary, is propagating back towards the borehole. It is noteworthy that the cracks are propagating in the region behind the stress waves and are being driven by gas pressurization. The fact that the cracks propagate at high velocity in the low stress region behind the stress waves is even more evident in subsequent frames where the higher amplitude stress waves are all located near the boundary. Several fringe loops at the crack tips indicate a significant amount of energy available to drive the cracks. A reduction in the amount of explosive used also tends to create less (or no) borehole crushing, and this permits the cracks to be more easily pressurized. The effect of gas flow into the cracks is to produce an increase in the stress intensity factor K along with increases in the length of the propagating crack [34]. This possibility for free gas flow into the cracks will permit the cracks to extend over longer distances before crack arrest occurs. This then requires that fewer boreholes need to be drilled and the excavation job becomes less expensive. The technique of fracture-controlled blasting as outlined here has been successfully implemented in the field in various applications, including the driving of tunnels [35]. The main disadvantage is of course that the boreholes must be notched, which is time consuming and expensive. Mechanical
Table 1 Pressure Ranges for Controlling Crack Initiation with Side Grooving Rock grain size
Notch size
Very fine Fine Medium Coarse
Deep Medium Medium Shallow
a/
a
Pmax
Pmin
(mm)
(in)
(mm)
(in)
(MPa)
(psi)
(MPa)
(psi)
0.025 0.050 0.125 0.250
0.001 0.002 0.005 0.010
5.00 2.50 2.50 1.25
0.20 0.10 0.10 0.05
110 76 48 34
16000 11000 7000 5000
7.6 11.0 11.0 15.9
1100 1600 1600 2300
For granite with Klc = 1.8 MPa m1/2 (2 x 103 psi in1/2).
(α)
Figure 32a
Pmax/ Pmin
14.5 6.9 4.4 2.2
66
Blasting
Figure 32 Model geometry and results from fracture-controlled testing, (a) Geometry, (b) Early time, (c) Late time
Mechanisms of Rock Fragmentation by Blasting
67
notching, water jet cutting and linear-shaped charges have been used to accomplish the notching [36, 37]. Unless some of the problems associated with notching can be overcome, it is unlikely that wide acceptance of the technique will occur. The important factor to note, however, is that the presplitting/postsplitting technique, which is predominantly a stress wave controlled event, can be more efficiently accomplished by changing the mechanism to one in which gas pressurization dominates. This is a good example (as is well stimulation) of how a technique that has been developed and shown to work ceases to work with increased charge sizes. This occurs as a result of not taking advantage of the gas pressurization effects in the borehole. 2.7 APPLICATIONS IN CONSTRUCTION AND QUARRY BLASTING Chiappetta and Mammele [38] used high-speed photography to evaluate air decks, stemming retention and gas confinement in several commercial applications. Their research was aimed at providing a better understanding of the mechanisms of blasting. They in effect were evaluating in the field the mechanisms defined in the model testing conducted at Maryland, both from the standpoint of stem-induced fracturing and in fracture-controlled blasting as described above. They used high speed photography to diagnose tests with and without open spaces in the borehole above the charges, in conjunction with stemming and open hole (no stemming) blasting. They evaluated the use of these 'air decks' when used in cratering, presplitting, bench blasting and reclamation projects. In effect, they extended the use of the stem-induced fracturing as investigated at Maryland to include applications in fragmentation-blasting situations based on the results obtained from Russian literature (Mel'nikov [39]). Chiappetta and Mammele, in their testing of presplitting, conducted experiments in highwall control. In one test they used nine boreholes with diameters of 23 cm drilled on 5.2 m centers to depths of 14.3 m. ANFO was bulk loaded in the bottom of the holes for a length of 2.4 m - about 178 kg of explosive. The length of the air deck was 9.2 m with the remaining borehole being filled with stemming. The explosive charge by volume was about 17% of the drill hole and about 27% of the air deck. Figure 33 shows the results of the presplit tests at the collar regions of the holes, the presplit crack and the final wall after excavation. They felt that there may have been a little more breakage at the collar regions of the boreholes than desired but felt that the final presplit line and the integrity of the highwall were quite acceptable. Figure 34 shows a direct comparison of the results obtained with the air deck technique and conventional techniques being used in the mining operations and shows quite good fracture control. Chiappetta and Mammele have called this technique ADP (Air Deck Presplitting). In their paper they point out 'the Air Deck Presplitting technique has been tried [by them] in a wide variety of formations with fair to excellent results. Where conventional presplitting techniques succeeded, the ADP techniques achieved equivalent or better results in borehole diameters ranging from 13 to 30 cm. On an economic basis.
Figure 33
Results of Air Deck Presplitting showing (a) the collar region, (b) the presplit crack and (c) the final wall (after Chiappetta and Mammele [38])
68
Blasting
Figure 34
Direct comparison of results from Air Deck Presplitting and no presplitting: ADP on the right, no presplit on the left (after Chiappetta and Mammele [38])
the ADP technique has reduced costs from 10 to 46% compared to conventional techniques'. Their full-scale testing has shown that, based on successful presplit blasts, the explosive load with respect to borehole volume should be 8-11% and 14-18% with respect to the air deck volume. They found that the loading density per unit area of presplit surface ranged from 0.24 to 0.98 kgm~2. They found that significant cost savings were a direct result of lowered explosive costs, larger hole spacing and lower labor costs. This is yet another example of how proper use of both stress waves and gas pressurization can achieve as good or better results. If proper use of explosives is to be achieved, both aspects must be used to advantage. 2.8 SUMMARY The information presented here is intended to show that both stress waves and gas pressurization are important in the fracture and fragmentation of rock. It is felt that neither of the two mechanisms alone can effectively fragment rock. Of the two, the energy contained in the gas pressurization phase is much greater than the energy contained in the stress wave component. Nonetheless, the stress waves in most situations are quite useful in preconditioning the rock mass for later action by gas pressurization. If the blasting event is controlled by the stress wave phase alone, the results will be very disappointing and very inefficient. On the other hand, in most situations, gas pressurization alone cannot effectively fragment the rock. Most efficient results can be obtained if proper use can be made of both components. In this chapter the attempt has been to review the current state of understanding of the mechanisms of fragmentation which result after the detonation of an explosive charge in rock and rock-like materials. The state of understanding at this time is far from complete, but much progress has been made over the past 20 years. The literature is vast and many of the pertinent publications are not easily obtainable. In some cases major contributions have not appeared in the open literature. Much of the information presented reflects the work conducted by the author and various colleagues at the University of Maryland. 2.9 REFERENCES 1. 2. 3. 4.
Hino K. Fragmentation of rock through blasting. Colo. Sch. Mines Q. Rep. 51, 191-207 (1956). Duvall W. I. and Atchison T. C. Rock breakage by explosives. Rep. Invest. - U.S. Bur. Mines RI-5356 (1957). Langefors U. and Kihlstrom B. The Modern Technique of Rock Blasting, pp. 18-28. Wiley, New York (1963). Persson P. A., Lundborg N. and Johansson C. H. The basic mechanism in rock blasting. In Proc. 2nd Congr. Int. Soc. Rock Mech., Belgrade, vol. 3, pp. 19-33. (1970).
Mechanisms of Rock Fragmentation by Blasting 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38.
39. 40.
69
Kutter H. K. and Fairhurst C. On the fracture process in blasting. Int. J. Rock Mech. Min. Sei. 8, 181-202 (1971). Persson P. A. Fragmentation systems. In Proc. 3rd Congr. Int. Soc. Rock Mech., Denver, pp. 153-156. National Academy of Sciences, Washington, DC (1974). Hagen T. N. Rock breakage by explosives. In Australian Geomechanics Nat. Symp. Rock Fragmentation, Adelaide, pp. 1-17. (1974). Hagen T. N. and Just G. D. Rock breakage by explosives - theory, practice and optimization. In Proc. 3rd Congr. Int. Soc. Rock Mech., Denver, p. 1349. National Academy of Sciences, Washington, DC (1974). Bligh T. P. Principles of breaking rock with high pressure gases. In Proc. 3rd Congr. Int. Soc. Rock Mech., Denver, vol. 2B, pp. 1421-1427. (1974). Persson P. A., Ladegaard-Pedersen A. and Kihlstrom B. The influence of borehole diameter on the rock blasting capacity of an extended explosive charge. Int. J. Rock Mech. Min. Sei. 6, 277-284 (1969). Mel'nikov N. V. A method for improved explosive fragmentation of rock. In 6th Int. Colloq. Gas Dynamics of Explosives and Reactive Systems, Stockholm, pp. 1113-1127. (1977). Warpinski N. R., Schmidt R. A., Cooper P. W., Walling H. C. and Northrop D. A. High energy gas fracture - multiple fracturing in a wellbore. In Proc. 20th U.S. Symp. Rock Mech., Austin, TX, pp. 143-152. (1979). Bhandari S. On the role of stress waves and quasi-static gas pressure in rock fragmentation by blasting. Acta Astron. 6, 365-383 (1979). Barker D. B. and Fourney W. L. Photoelastic investigation of fragmentation mechanisms - Part II. Report to the National Science Foundation from the University of Maryland (1978). Winzer S. R. and Ritter A. W. The role of stress waves and discontinuities in rock fragmentation. In Proc. 21st U.S. Symp. Rock Mech., Rolla, MO (Edited by D. A. Summers), pp. 362-370. University of Missouri (1980). Brinkmann J. R. Separating shock wave and gas expansion breakage mechanisms. In Proc. 2nd Int. Symp. Rock Fragmentation by Blasting, Keystone, CO (Edited by W. L. Fourney and R. D. Dick), pp. 6-15. Society of Experimental Mechanics, Bethel, CT (1987). Holloway D. C , Fourney W. L. and Barker D. B. Dynamic crack propagation in rock plates. In Proc. 21st U.S. Symp. Rock Mech., Rolla, MO (Edited by D. A. Summers), pp. 371-378. University of Missouri (1980). Ash R. L. Flexural rupture as a rock breakage mechanism in blasting. In Fragmentation by Blasting (Edited by W. L. Fourney, R. Boade and L. Costin), pp. 24-29. Society of Experimental Mechanics, Bethel, CT (1985). Porter D. D. A role of the borehole pressure in blasting: the formation of cracks. Ph.D. Thesis, University of Minnesota, Minneapolis (1970). Coursen D. L. Cavities and gas penetrations from blasts in stressed rock with flooded joints. Acta Astron. 6, 341-363 (1979). Wilson W. H. An experimental and theoretical analysis of stress wave and gas pressure effects in bench blasting. Ph.D. Thesis, University of Maryland (1987). Daw A. W. and Daw Z. W. The Blasting of Rock: in Mines, Quarries, etc., p. 8. Spon, London (1898). Dick R. D., Fourney W. L., Goodings D. J., Lin C. P. and Bernold L. Use of explosives on the Moon. J. Aerospace Eng. 5, 59-69 (1992). Fourney W. L, Dick R. D. and Simha K. R. Y. Model study of crater blasting. Rock Mech. Rock Eng. 21,183-205 (1988). Wang X. J., Fourney W. L. and Dick R. D. Model studies of optimized crater blasting. In Proc. 3rd Int. Symp. Rock Fragmentation by Blasting, Brisbane, pp. 137-142. Society of Experimental Mechanics, Bethel, CT (1990). Young C , Chapman, Dick R. D. and Fourney W. L. Small charge cone-fracturing technique for rapid excavation. In Proc. 3rd Int. Symp. Rock Fragmentation by Blasting, Brisbane, pp. 129-136. Society of Experimental Mechanics, Bethel, CT (1990). Cuderman J. F., Cooper P. W., Chen E. P. and Northrop D. A. A multiple fracturing technique for enhanced gas recovery. In Proc. Int. Gas Research Conf, Los Angeles, CA, pp. 657-667. (1981). Cuderman J. F. Multiple fracturing experiments - propellant and borehole considerations. In Sandia National Laboratory Report, SAND 81-2224C (1981). Fourney W L., Barker D. B. and Holloway D. C. Model studies of explosive well stimulation techniques. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 18, 113-127 (1981). Fourney W. L., Barker D. B. and Holloway D. C. Model studies of well stimulation techniques using propellant charges. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 20, 91-101 (1983). Young C , Barker D. B. and Clark H. C , Jr. Field tests of the stem-induced explosive fracturing techniques. SPE J. Prod. Eng. 266-274 (1986). Ladegaard-Pedersen A., Fourney W. L. and Dally J. W. Investigation of presplitting and smooth blasting techniques in construction blasting. Report to National Science Foundation NSF-RA-T-75-015 (1974). Ouchterlony F. Fracture mechanics applied to rock blasting. In Proc. 3rd Congr. Int. Soc. Rock Mech., Denver, vol. 2B, pp. 1377-1383. (1974). Fourney W. L. Fracture control blasting. In Rock Fracture Mechanics (Edited by H. P. Rossmanith), pp. 301-319. Springer-Verlag New York (1983). Sperry P. E., Thompson D. E., McKown A. F. and Fourney W. L. Controlled blasting experiments at Porter Square pilot tunnel. In Proc. Rapid Excavation Tunneling Conf, Atlanta, GA, vol. 2, pp. 1130-1157. (1979). Holloway D. C, Bjarnholt B. G. and Wilson W. H. A. field study of fracture control techniques for smooth wall blasting. In Proc. 27th U.S. Symp. Rock Mech., Tuscazoosa, AL (Edited by H. L. Hartman), pp. 456-463. (1986). Holloway D. C , Bjarnholt B. G. and Wilson W. H. A. field study of fracture control techniques for smooth wall blasting: part 2. In Proc. 2nd Int. Symp. Rock Fragmentation by Blasting, Keystone, CO (Edited by W. L. Fourney and R. D. Dick), pp. 646-656. Society of Experimental Mechanics, Bethel, CT (1987). Chiappetta R. F. and Mammele M. E. Analytical high-speed photography to evaluate air decks, stemming retention and gas confinement in presplitting, reclamation and gross motion applications. In Proc. 2nd Int. Symp. Rock Fragmentation by Blasting, Keystone, CO (Edited by W. L. Fourney and R. D. Dick), pp. 257-301. Society of Experimental Mechanics, Bethel, CT (1987). Mel'nikov, N. V. Utilization of energy of explosives and fragment size of rock in blasting operations. Gornyi Zhurnal 5, 61-64 (1940). Starfield A. M. Strain wave theory in rock blasting. In Proc. 8th U.S. Symp. Rock Mech., University of Minnesota, 1966 (Edited by C. Fairhurst), pp. 538-548. Port City Press, Baltimore, MD (1967).
3 Methods of Improving Blasting Operations C A M E R O N K. M C K E N Z I E
Australian Blasting Consultants Pty. Ltd, Toowong, Qld., Australia
3.1
INTRODUCTION
71
3.2 FRAGMENTATION 3.2.1 Blast Event Monitoring 3.2.1.1 Limitations of event monitoring 3.2.1.2 Common blasting problems 3.2.2 Blast Performance Monitoring 3.2.2.1 Fragment size distribution 3.2.2.2 Blasthole velocity of detonation 3.2.2.3 Burden movement velocity 3.2.2.4 Induced shock energy
72 72 73 74 75 75 75 11 78
3.3 DAMAGE 3.3.1 Vibration Influences 3.3.2 Displacement Influences 3.3.3 Blast Design for Smooth Blasting 3.3.3.1 Loading density 3.3.3.2 Minimum standoff distance
79 79 81 82 82 84
3.4
ENVIRONMENTAL ASPECTS
84
3.4.1 Overpressure 3.4.1.1 Sources of overpressure 3.4.2 Ground Vibration 3.4.2.1 The seed waveform model 3.4.2.2 Superposition of waveforms 3.4.2.3 The coupling factor 3.4.2.4 The influence of blast size 3.4.2.5 The influence of delay timing
85 85 88 88 88 90 91 93
3.5
3.1
REFERENCES
93
INTRODUCTION
With the appearance of modern instrumentation for monitoring, increasingly sophisticated models for blast design and blast prediction, and more versatile explosives and initiation systems, modern blasting is moving more and more towards a science. Blasts should be designed with a high degree of confidence of achieving specific targets. In general, rock blasting is undertaken to facilitate the removal of rock from an excavation, and critical design factors associated with blasting are: (i) cost effective fragmentation and excavation of the rock; (ii) minimized damage to the surrounding rock mass to control stability and dilution; and (iii) minimized environmental impact to protect nearby residences and sensitive structures. 71
72
Blasting
Steps initiated to improve blasting operations will address one or more of the above issues. The following sections present some insight into methods of investigating each of the above factors. The object of this chapter is to present a methodology for the evaluation of blasting and blast designs to permit their fine tuning or optimization.
3.2 FRAGMENTATION Critical design parameters such as blasthole diameter and the separation of blastholes can be estimated using a suite of models and simple equations [1-3], but these should be regarded only as initial estimates. Fine tuning and optimization of designs require a more intimate knowledge of the complex interaction between the local rock mass and the explosive being used. This more intimate knowledge can only come from quantitative measurement and monitoring of blasting performance [4]. With modern instrumentation now readily available to blasters, it is frequently possible to locate gauges around a blast so that the detonation of individual charges can be sensed. The procedure is generally loosely described as 'blast monitoring', and refers to any form of recording which takes place during the period in which a blast pattern is initiated and detonated. The time period over which the whole event occurs is generally less than 2 s, but may be extended in conventional tunnel blasting (drifting) to around 10 s. In some South African underground mines, single-panel blasts may last for up to 15 min. The primary monitoring techniques for blasting are: (i) event monitoring, designed to detect the initiation or detonation of each charge, or selected charges, in the blast pattern; and (ii) performance monitoring, designed to provide information concerning the efficiency with which each charge detonates, and the effectiveness of the explosive/rock interaction.
3.2.1 Blast Event Monitoring Event monitoring is an essential component of any program of blast design optimization. Before a design can be assessed or compared with another, it must first be established that the detonation of charges proceeded according to the designed sequence, and that each charge performed the required amount of work on the surrounding rock. Blast event monitoring uses a variety of sensors to detect the detonation or initiation of separate explosive charges. The detonation is sensed using a range of sensors including vibration sensors, electromagnetic sensors, radio frequency detectors, infra-red sensors, microphones, pressure gauges and impulse detectors. Of the techniques listed, the most advanced and commonly employed technique is vibration monitoring. Vibration gauges are attached to the rock in close proximity to the blast pattern, and detect the intense shock impulses produced by individual charges as they detonate. The commonly used gauges are either geophones (velocity proportional gauges) or accelerometers (acceleration proportional gauges). In most applications, either gauge can be used, though the higher 'shock resistance' and frequency response of accelerometers make them preferable when monitoring very close to (within meters of) the explosive charges. Examples of the vibration impulses collected using geophones in single blastholes are shown in Figure 1. Note the bipolar nature of the signals, indicating times when the rock around the gauge is undergoing stages of alternative compression and dilation. From a geophone gauge, the amplitude of the signal is directly proportional to the particle velocity, and the units are therefore shown as m s _ 1 o r more commonly as mms" 1 . For an accelerometer gauge, the amplitude would be directly proportional to acceleration, and the units would be m s"2, or V (g = acceleration due to gravity = 9.8 m s"2). Note also that the time of the event can be measured with great precision using digital recorders with sampling speeds up to 1 MHz (1 x 106 samples per second). Extending the monitoring procedure to a multihole blast should yield a sequence of pulses similar to those shown in Figure 1, with the time interval between successive pulses representing the actual delay interval between successive charge initiations. Figure 2 shows such a record from a trench blast in hard rock, from which the detonation of each hole can be confirmed and the precise timing of initiation determined for each hole.
Methods of Improving Blasting Operations
73
(a) .rω
200
F E
100
0
ö -100 o
>ω
-200
(b)
^
400
F E
200
V)
300 100
>»
0 'ϋ -100 o -200 ω -300 -400
>
0.04 Time (s)
Figure 1 Single-hole blast vibration responses for two different rock types: (a) hard rock and (b) soft rock
300—I 200 — E
E
100 —
ÜM~
ΰ -200 H -300 H
V-4T~^—l· 0.3 Time (s)
Figure 2 Blast vibrations from a multihole blast showing individual hole detonations
3.2.1.1 Limitations of event monitoring Blast monitoring using vibration sensors will not always identify the initiation of every explosive charge. From the above presentation, vibration monitoring is oriented towards the identification of separately delayed charges. If many charges share the same delay, then vibration-based monitoring may not be able to detect the detonation of all charges, especially if the charges are all detonated simultaneously or near-simultaneously. Figure 3 presents two cases where charges have initiated near-simultaneously. In each case two charges are discernible, although some degree of operator interpretation is required. Where more charges are involved, or where the time between individual initiations is less, it may not be possible to say how many charges have detonated, although the complex vibration interference patterns will still indicate enhancement of vibration levels and multiple charge initiation. A second complication is caused by the properties of the rock being blasted. The two single-hole vibration pulses in Figure 1 were obtained from vastly different rock types. The high frequency wavelet was produced by a small, single-hole charge of ANFO in hard, brittle rock in an underground environment. The low frequency wavelet was produced by a long, single-hole charge of ANFO in a softer rock in a surface environment. The duration of the wavelet is influenced by the length of the charge, as described by Grant et al. [5], but is probably even more heavily influenced by the modulus of the rock surrounding the blasthole. Where the delay interval between successive charge detonations is less than the individual wavelet duration, interaction and enhancement of vibrations will occur and the resulting vibration waveform can become too complex to deconvolve, like the waveform shown in Figure 4. This commonly happens in soft rock formations, large open-cut blasts, and in large underground mass blasts where the average delay interval between charges is very small and the number of charges can be in excess of 1000. Better discrimination can often be obtained by moving the vibration sensor closer to the blastholes.
74
Blasting (α) ~
300-
< ' />
200—I
I
loo-
>> o o -100 "
1
0.212
0.216
#1
0.228
0.224
0.232
100 —
E E
50 —
υ _o
-50 —
Φ >
0.220
Time (s)
(b)
(/)
♦'
#2
ΛΛ J\l\r\
y"~^~s^\s**<.
-100 —
#1
#2 3.064 Time (s)
Figure 3 Near-simultaneous detonations of blastholes. Charge detonations: (a) 3.1 ms apart; (b) 3.8 ms apart
Time (s)
Figure 4 Vibration response from many holes initiated near-simultaneously (mass blast)
3.2.1.2
Common blasting problems
The types of commonly observed problems in blasting include misfires (undetonated charges), instantaneous initiation (charges detonated by the initiation system) and sympathetic initiations (charges initiated by the impact of adjacent charges). Some examples are shown in Figure 5. In the first instance, charges in the critical burn cut section of a tunnel blast failed to detonate. After repeated failure, the holes were omitted from the design without affecting the performance of the blast. In the second instance in Figure 5, instantaneous initiations occurred within several rings of blastholes in a mass blast. The initiation times evident from the record reflect surface delay times for the rings which detonated well before the designed initiation time of the 'first' blasthole. When charges in back rows initiate before a free face has been created, little useful work can be performed, and the charge has a high probability of disrupting surrounding charges, so that this type of defect can be particularly harmful in terms of controlling damage or fragmentation. When interpreting blast vibration records, it may be difficult to differentiate between 'misfire' charges and 'sympathetic initiations'. Charges located too close together may be initiated sympathetically by the earliest firing charge, and appear as 'misfires' by the absence of a vibration response at their nominal initiation times. Where vibration responses are consistently absent, blasthole spacings should be increased, or blastholes removed from the pattern, to see if the problem is resolved. Blast monitoring therefore represents a pragmatic approach to blast design optimization, providing real information about the performance and interaction of individual charges, and permitting decisions regarding blasthole spacings, explosive strength, charge size and delay timing to be made based on measured and observed responses. Importantly, blast monitoring provides a positive means of assessing the effects of changes in blast designs.
Methods of Improving Blasting Operations
t:
υ o >
-100
75
^r^^^
- 2 0 0 —| -300 -400
| Nominal start time 0.3
Time (s)
Figure 5 Some initiation malfunctions in production blasting: (a) misfire (drifting round); (b) instantaneous initiation (mass blast)
3.2.2
Blast Performance Monitoring
Blast performance monitoring involves the collection of data that describe the performance of the explosive or the response of the rock to the explosives. These data can be used to quantify or compare explosive performances or blast patterns. This type of monitoring has a primary application to modeling, providing the basic information on explosive performance and explosive-rock interaction which is required to calibrate and verify model predictions. To a large extent, performance monitoring can be used whenever the explosive user feels that the explosive product, initiation system or blast design may be faulty in one regard or another. Common performance indices for explosives derive from measurements of fragment size distributions, blasthole velocity of detonation, burden movement velocities and displacements and levels of induced shock energy. 3.2.2.1
Fragment size distribution
In many respects, fragment size distribution should be one of the primary indices of explosive performance since it directly assesses one of the primary goals of blasting - the need to fragment the rock to facilitate rapid excavation and removal. However, the measurement of fragmentation from blasting is one of the most vexing issues facing the blasting technologist. Modern technology is investigating automatic photographic and video image scanning methods, but to date there is no cheap or simple method, and few technical groups are prepared to mechanically screen the entire muckpile from a blast of between 10000 and 1000 0001. This mechanical sizing is particularly difficult in underground excavations where primary crushing must be conducted prior to haulage to the surface. To be complete, fragmentation data should be viewed in the perspective of the in situ block size distribution, particularly where one blast pattern or explosive type is being compared with another. In situ block size is determined by fractures, joints, foliations and bedding planes. A jointed rock mass can be considered to be a system of weakly cemented blocks of varying size, with surfaces defined by structural discontinuities. By conducting face mapping of the exposed faces to determine the spacings between fractures in three dimensions and the orientations of the primary fracture sets, structural models can be developed to approximate the distribution of block sizes. From this information, a weight-size distribution of the in situ blocks can be determined and compared with a measured, estimated, or modeled size distribution obtained from blasting, to determine the amount of breakage performed by the explosive and the blast pattern. 3.2.2.2
Blasthole velocity of detonation
Explosive velocity of detonation has long been used to check the 'state' of an explosive, and has generally been conducted in a small sample tested on the surface. This testing is sufficient to draw
76
Blasting
conclusions regarding the extent to which a sample of explosive has aged or deteriorated due to storage or regarding the evaluation of the degree of quality control in manufacturing. Such unconfined tests, however, give no indication about how an explosive will react inside a blasthole under varying degrees of confinement and mixing with water, mud or drill cuttings. It is well documented that for nonideal explosives, such as the ANFO-based explosives commonly used in commercial blasting applications, the performance varies according to blasthole diameter and degree of confinement. In particular, the partitioning of total energy between shock energy (primarily responsible for fragmentation) and heave energy (primarily responsible for muckpile displacement) can be strongly influenced by confinement and charge diameter. Justification for the measurement of the blasthole velocity of detonation comes largely from the assumption that the fragmentation potential of an explosive is directly related to the detonation pressure generated in the blasthole, and that this pressure is related to the blasthole velocity of detonation by the equation Λΐ °C PexpVOD2 ctual
(1)
3
where pexp is the explosive density (kgm~ ), VODactual is the actual blasthole velocity of detonation (ms _1 ) and Pd is the detonation pressure produced by the explosive in the blasthole (Pa). Further, it was shown that the proportion of explosive contributing to the detonation reaction at the detonation front was determined from the equation ( VODactual Y
(2)
V VOD,,
where VODmax is the maximum steady state velocity of detonation for the explosive (ins -1 ) and n is the volume fraction of explosive contributing to the detonation reaction. On the premise that the fraction of explosive contributing to the detonation reaction produces shock energy, and that the remainder of the explosive reacts behind the detonation front to produce heave energy, it has been considered desirable by most practitioners to utilize explosives with the highest detonation velocities when blasting high strength, massive rock types. Common techniques for measurement of blasthole velocity of detonation include continuous measurement systems and multiple point systems. For the continuous measurement systems, a
(a)
12 m ANF0 column
1.0 0.5 ω
o
>
°-0.5 -
mAnnWr 0.033
-1.0 -
0.034
Time (s) (b)
ANFO (100 mm diameter column) VOD= 3985 m s"1 Standard deviation = 25m s"1
r2 ' 0.99915
500
1000
1500
2000
2500
3000
Time (/is) Figure 6
Blasthole velocity of detonation: (a) measurement and (b) regression analysis (raw data ■ ; best fit —)
Methods of Improving Blasting Operations
77
length of cable is inserted into the blasthole prior to charging. As the explosive detonates (starting from the bottom of the column), the length of cable decreases continuously and surface-monitoring equipment produces a signal enabling the length of cable to be calculated at any instant. The advantage of continuous measurement systems is that they enable assessment of detonation behavior over small intervals, such as around the primer(s) or at interfaces between different explosive types. Two types of continuous velocity of detonation measurement systems are available for use in long column charges: the SLIFER (Shorted Location Indication by Frequency of Electrical Resonance) system and the TDR (Time Domain Reflectometry) system. Multiple point systems for the measurement of blasthole velocity of detonation are cheaper and more simple. Multicore cable is loaded into the blasthole at the time of charging, with electrode pairs located at known intervals up the explosive column. As the detonation front reaches each electrode pair, surface electronics produce timing pulses which can either be recorded directly for waveform processing or which are used to trigger crystal clocks that measure the time interval between successive pulses to an accuracy of ± 1 μ8 or better. By inserting an array of these electrode pairs over the full column length, the average velocity of detonation can usually be calculated to better than + 1%. In some cases fiber optic probes are preferred to electrodes, though there appears to be no improvement in accuracy relative to their electrical counterparts. An example of signals from a multiple point system inserted into an ANFO explosive is shown in Figure 6, together with the linear regression technique and an estimate of the error in the calculation of blasthole VOD. Blasthole velocity of detonation measurement has been used to evaluate the consistency of performance of bulk emulsion and water gel explosives, the effect of primer size on explosive performance and toe conditions, and the susceptibility of emulsion and slurry explosives to dynamic desensitization (the desensitization of an explosive in a blasthole by the shock action of earlier-firing blastholes in close proximity).
3.2.2.3
Burden movement velocity
The strength or performance of an explosive is frequently judged by the displacement of the muckpile. This is particularly relevant for cast blasting, for example. Providing that the volumes and degree of fragmentation of two muckpiles are equal, it is quite reasonable to state that the blast which produced the greatest displacement of rock, per unit charge weight, used a more energetic explosive. Alternatively, comparison of displacement profiles for the same explosive with different blasthole patterns, or different delay timing and sequencing, can be used to optimize cast blast design. The monitoring technique commonly used to gauge the effectiveness of a blast pattern is that of burden motion recording through high speed photography or video analysis. Burden motion studies provide an indication of the suitability of selected blasthole burdens, and are probably the only means of assessment of blasthole-stemming performance. Both of these factors are assessed through the measurement of velocities of motion and the time after detonation before motion commences. Motion studies from high speed films or video recordings require computer digitization. Successive frames, at known time intervals, are digitized to determine the displacements of markers at known locations, as shown in Figure 7. If the viewing angle is known, then displacements on the two-dimensional screen image can be transformed to actual displacements in any desired plane. If it is desired to control burden movement, such as in cast blasting (maximize displacement) or 'paddock' blasting (minimize displacement), blast patterns can be adjusted according to feedback provided from motion studies. Similarly, stemming lengths can be adjusted to permit the loading of maximum charge lengths to achieve maximum fragmentation over the full bench height. The heave or displacement potential of an explosive is its ability to displace the rock burden. This is expected to have a strong influence on the diggability of the muckpile after blasting, and therefore a strong influence on excavator productivity. Displacement of the rock burden is produced as a result of the peak blasthole pressure produced in the blasthole in the region behind the primary reaction zone. The effect of this pressure on the rate and timing of burden movement will be dependent upon the size of burden and the density of rock, or the total mass of rock displaced. A reasonable indicator of the available heave energy can therefore be expected from calculation of the kinetic energy Eh of the burden from the equation mV2 £h = ^ ~
(3)
78
Blasting
Marker I ö
°
o°
o o o o
oo
Marker 2
Figure 7 Digitization of face markers for measurement of face velocities
where m is the mass of rock moved (kg) and Fis the average velocity of movement for the entire rock mass (ms _ 1 ). Equation (3) suggests that for a constant explosive type and quantity the burden velocity is inversely proportional to the square root of rock mass, so that for a bench of fixed height and fixed blasthole spacing the burden velocity is inversely proportional to the square root of burden. This, however, is too simplistic, and assumes no energy loss due to the breakage of rock or the venting of gases to the atmosphere. In practice, burden velocity is frequently seen to be inversely proportional to burden to the power 1-1.2. The kinetic energy imparted to a rock burden is calculated by first measuring or estimating the velocity of motion of the burden. Face velocities can be measured using high speed photography or video recorders, the latter in conjunction with computer digitization, and these techniques can provide information about differential face movement and stemming ejection. However, face velocities are not necessarily representative of average burden velocities, and equation (3) requires the average velocity for the full burden to be calculated. Average burden velocity can frequently be calculated by considering the motion of the center of gravity of the material in the rock burden. This technique requires accurate surveying or photogrammetry before and after the blast to determine profiles, and only provides information about the average burden velocity. After locating the centers of gravity of the preblast and postblast burden material, the average velocity of movement F av for the entire muckpile is determined, using simple equations of motion, from the horizontal and vertical displacements of the center of gravity 9 Y/2 where Xd and Yd represent the displacements (in meters) of the center of gravity in the horizontal and vertical directions, respectively, and g is the acceleration due to gravity (ms~ 2 ). The simple technique above assumes that the original direction of motion was horizontal only, and therefore may not apply to heavily inclined blastholes or benches with large stemming columns.
3.2.2.4
Induced shock energy
When an explosive detonates, the available energy is commonly categorized as either shock energy or heave energy. Shock energy is said to accompany the rapid change of state from a solid (or liquid) to a gas which occurs at the detonation front of the explosive. As the proportion of explosive reacting in the detonation zone increases, the explosive is considered to have a greater brisance, or fracturing potential, and its relative shock energy increases. The partitioning between shock and heave energy is frequently evaluated in underwater tests, with the shock component £ s , derived from the pressure-time history P(t\ measured using a pressure gauge located in the pond at a distance R from a sample of explosive. The following equation can be employed
:«.}
EsocR2
[P(i)] 2 dt
(5)
Methods of Improving Blasting Operations
79
In the same manner, investigators have compared explosive performances in rock using measurements of induced ground vibration in close proximity to the blasthole. The peak level of borehole strain eb, developed by an explosive in a blasthole, can be calculated from equation (5), based on the final equilibrium pressure P e q of the explosion gases and the Poisson's ratio v and Young's modulus E for the rock
Using equation (6), the explosive generating the higher borehole strain has been identified as the explosive displaying the greater fragmentation potential. The comparison has been effected using measurements of peak particle velocity (PPV), under the assumption that PPV, the rock compressional wave velocity Vp and the borehole strain ε are related according to the equation PPV Experimentally, measurements of the level of vibration obtained at a fixed distance behind blastholes charged with equal amounts of different explosives, in the same rock type with the same burden and spacing, have been used to identify explosives showing a higher generation of borehole strain. The higher levels of peak particle velocity have therefore been used as an indicator of greater fragmentation potential, thus allowing a more appropriate choice of explosive for application in rock types requiring a high degree of fracturing. A comparison of vibration or shock levels from blastholes must be undertaken with considerable care and caution. Variability in shock levels can be very high due to factors other than explosive properties and performance. Factors such as degree of fixation (incorporating burden dimension, rock strength, bulk modulus, rock density, degree of water saturation and others) will influence the effective charge weight in the blasthole, and therefore the level of induced strain and vibration. Tests may need to be repeated several times to gain an acceptable level of statistical significance. It must also be stressed that evaluation of explosive performance in this manner does not provide a reliable indication of total explosive energy, since it provides little information about the heave or displacement potential of the explosive.
3.3
DAMAGE
Successful blasting produces material which is easily excavated, leading to higher excavator productivities, and which is easily handled at the primary crusher, permitting higher crusher throughput. If the blasting is achieving its objectives, then production targets can most easily be achieved, and the operation is well on the road to minimizing total production costs. If blasting does not fully achieve its objectives, production costs increase in areas including secondary breakage, loading, maintenance (excavators, trucks, hoppers, grizzlies, etc.) and crushing. Blasting therefore has the potential to influence the operating economics of many downstream processes, and to promote operating efficiency operators will often err on the side of overblasting. A complication to production blasting philosophies is the sometimes major issue of rock stability - with respect to either the long term stability of permanent openings or ultimate slopes, or the short to medium term stability of interim openings or slopes. Blasting adjacent to these structures must still achieve economic extraction of rock, but the destructive impact of the explosives must be reduced or tempered to maintain the structural integrity of the final structure. To control the impact of explosives on adjacent rock structures, it is necessary to recognize the principal mechanisms whereby stability can be reduced by the effects of nearby blasting. The primary mechanisms are related to the shock or vibration influences of the explosive detonation and to the dilation of fractures by either explosion gases or vibration acting on the rock mass. 3.3.1
Vibration Influences
High vibration levels can damage solid rock masses by initiating fresh fractures or extending and dilating existing fractures. Vibration in this context can be considered as indicative of strain or stress in the rock mass. At low levels of vibration, such as at relatively large distances from the blast, the
Blasting
80
levels of induced strain are too low to induce fracturing of the rock mass. However, very close to the blastholes vibration levels will increase sufficiently to fracture even large blocks of surrounding rock. Equation (7) presents the relationship between peak vibration levels PPV and induced strain ε for a rock mass of compressional wave velocity Vp. From Hooke's law, and assuming a brittle failure mode of rock, the maximum particle velocity PPVmax which can be withstood by the rock before tensile failure occurs can be computed from equation (8) if we know the tensile strength στ, the Young's modulus E and the P-wave velocity of propagation Vp, i.e. PPVmax = ^
(8)
E
Holmberg and Persson [6] used the above approach to arrive at a maximum particle velocity of between 700 and 1000 mm s~λ for hard igneous rocks. Although these levels of vibration were shown to be reliable indicators of incipient damage, readily observable damage is frequently taken to be approximately four times the level for incipient damage. To estimate the level of vibration PPV at any distance X from a blasthole containing weight Woï explosive, charge weight scaling equations similar to equation (9) are used FPV = KXaWß
(9)
where X, a and ß are site specific constants. However, these equations can only be applied in the far field, where the underlying assumption of a point source is valid. In the near field (very close to the blasthole where fracturing is occurring), equation (9) must be modified to account for the long, cylindrical shape of the charge. Equation (10) was developed for near-field vibration prediction by Holmberg and Persson [6] (the terms are explained in Figure 8)
PPV
,10)
H L ; + (*..a^-»)y·}'
where K, a and ß are the same constants as for equation (9) and Z is the linear charge concentration in the blasthole (kgm - 1 ). For the common situation where square-root charge scaling is applied (i.e. β = α/2), the equation of Holmberg and Persson reduces to / / Υ/2Γ
/K 0 tantf>-H\
|«/2
(11)
From knowledge of the vibration propagation characteristics of the rock mass and the relationship between vibration and strain it is possible to establish contours of fracturing around a blasthole. These contours represent the zone around a blasthole in which fracturing will occur as a direct result of vibration from the detonating explosive in the blasthole. Such a contour diagram provides a good indication of the amount of backbreak behind a blast, and the minimum required standoff between the back row of blastholes and the toe of the final wall. In many operations, both mining and civil construction, smooth blasting is undertaken to produce faces which are not only stable but which are also so smooth that loose, superficial spalling is eliminated. Where smooth blasting has been undertaken, it is imperative that subsequent blasting does not produce overbreak behind the designed exposure, and explosive loading in the back rows of
Stemming 11
—
. Explosive column
R
*
°
■—Geophone
H
^ ^
Φ
Figure 8 Integration over charge length to calculate near-field particle velocity (after Holmberg and Persson [6]) (x,, distance from the base of charge to the elemental charge; dx, height of elemental charge)
Methods of Improving Blasting Operations Presplit hole
Damage contour, (presplit hole)
v
Damage contours meet
/
81
Production hole
\ Stemming
Damage contour \ ^/(production hole)
-Explosive
i Figure 9 Placement of blastholes to control damage inside a presplit line
blastholes (buffer holes) must be adjusted to control strictly the extent of damage. Fracture contouring can be used here to examine the extent of damage around blastholes in determining the optimum diameter and standoff distances between blasthole rows. Figure 9 illustrates how blasthole spacing can be selected in perimeter blasting to ensure that rock damage is confined inside a presplit line for any configuration of blasthole diameter and explosive loading. Fracture contouring is the first important step in effective smooth-blasting design, and can be used to determine charge loads in perimeter holes and standoff distances for other charges. It can also be applied to tunneling and cavern excavation, open cut slopes and underground blasts against pillars or hangingwalls/footwalls. 3.3.2 Displacement Influences As solid explosive is converted to gas, extreme pressures (in excess of 1 GPa, or 10000 atm) are developed, acting in all directions around the blasthole. Behind the blasthole, the pressure is resisted by the retaining rock mass. In front of the blasthole, where a vertical free face exists just one burden dimension away, the retaining forces are less, and bulk movement occurs in that direction. Where no free bench face exists, the horizontal bench surface becomes the site of pressure relief, and pronounced surface swell occurs. In the presence of horizontal and vertical bench faces, a combination of forward and vertical movement occurs. Importantly, vertical displacement always occurs, starting from the base of the blasthole and continuing until the blasthole pressure is relieved by forward movement of the burden. The displacements are largest in the vicinity of the free surface, since the vertical retaining stresses are least in this part of the rock mass. The amount of vertical movement can be minimized by reducing as fully as possible the time before forward movement occurs. As the rock above the base of the blasthole is pressurized and pushed upwards, nonvertically oriented fractures and joints will dilate, permitting the high pressure gas to penetrate into the rock mass. This penetration of the gas into the rock mass, and the dilation of fractures, produce vertical displacement of the rock around the blasthole. Even after the explosion gases have been vented and the blasthole pressure removed, there is a permanent vertical displacement. Dilated fractures do not return to their original state of closure, and the number of intact rock bridges across the fracture planes can be greatly reduced, lessening considerably the shear strength of the weakness planes and possibly the peak friction angle of the rock mass. Both gas penetration and fracture dilation have been measured in field studies of rock blasting. Figure 10 shows how the pressure inside a sealed borehole behind a blasthole first registers a
82
Blasting Distance = Im O
0.6 0.4
Ü ° α>
2
ο
w -0.2
Time (s)
Figure 10 Gas pressure response inside a sealed blasthole behind a blast pattern
negative pressure, caused by vertical swell as horizontal fractures dilate, followed by an increase in pressure above ambient levels as explosion gases permeate through the system of dilated fractures. The monitoring establishes that dilation occurs prior to gas penetration, so that gas penetration is probably a symptom, more than a cause, of fracture dilation behind blastholes. The swell is confirmed by extensometer measurements, and is easily measured tens of meters behind blastholes, beyond the range of permeating gases. Even where the blasthole charge densities are very low, as in presplit holes, significant vertical displacements can still be measured behind the blastholes. After a relatively long period with respect to the duration of the blast itself, the pressure in the sealed borehole returns to zero. Extensometer measurements, however, indicate that even after the pressure has returned to ambient levels there is a permanent displacement, indicating a permanent dilation of joints or fractures. The effect of fracture dilation is to reduce the joint friction, or friction angle. As the aperture between the fracture surfaces increases, the frictional force acting to lock the surfaces together also decreases, so that the blocks are more likely to slip. The primary design feature of limits blasts is therefore to minimize the time over which the blasthole pressure can act against the surrounding rock. For presplit design, for example, this means that blastholes should not be stemmed. Unstemmed presplit holes will produce much higher noise and overpressure levels than stemmed holes, but will have less impact on the final wall condition. The immediate release, or venting, of gases from the presplit blastholes has no adverse effects on presplit performance since it is the interaction between shock waves from adjacent blastholes which produces the smooth wall, and gas pressure plays no significant part in the preshearing process. Blast pattens adjacent to final limits should also feature small burdens in order to minimize the confinement effect and subsequent heaving of the bench. Small burdens promote high burden velocities, and increased muckpile displacement, and may lead to an increase in powder factor. Limits blast design does not have to be a compromise between fragmentation and stability, but does require specialized blast designs commonly featuring smaller diameter blastholes, reduced burdens and low explosive-loading densities in the back rows of blastholes.
3.3.3
Blast Design for Smooth Blasting
Common methods adopted in smooth blasting include presplitting, postsplitting and trim blasting. All methods aim to produce a surface which is smooth, stable, and free from loose material. Perhaps the two most important aspects of smooth blast design are: (i) the determination of the most appropriate loading density of explosive in the blasthole; and (ii) the minimum standoff distance between the final face and the nearest blasthole.
3.3.3.1
Loading density
Blasthole loading densities are generally reduced in perimeter blastholes of sensitive structures in order to reduce the peak blasthole pressure. Upon detonation, fully coupled explosives exert a peak blasthole pressure P b , dependent upon the explosive density p exp and the velocity of detonation VOD of the explosive according to P b *0.25p exp VOD 2
(12)
Methods of Improving Blasting Operations
83
For ANFO at a density of 8 5 0 k g m " 3 and a velocity of detonation of 3 5 0 0 m s " 1 a peak blasthole pressure of 2.6 GPa will be generated. A fully coupled emulsion explosive of density 1200 k g m " 3 and velocity of detonation of 5500 m s " 1 will generate a peak blasthole pressure of around 10 GPa. These pressures are well in excess of the compressive strengths for rocks which are generally less than 250 MPa, or a maximum of one-tenth of the peak blasthole pressures. Tensile strengths of rocks will generally be less than one-hundredth of the peak blasthole pressures for fully coupled blastholes. In smooth blasting, the peak blasthole pressures are reduced to be only slightly more than the compressive strength of the rocks being blasted. This reduction is generally achieved by a reduction in the effective explosive density, either by diluting the explosive with an inert material or by decoupling the explosive from the rock. When an explosive is decoupled from the rock, the blasthole is only partially filled with explosive, so that a large reduction in peak pressure is achieved as the explosion gases expand to the full blasthole volume. Lateral decoupling of an explosive is achieved when the diameter of the explosive is less than the diameter of the blasthole. Air decking, on the other hand, involves the use of a fully coupled explosive for only a fraction of the length of the blasthole, with a column of air or other inert material between the explosive charge and the stemming column. The extent to which peak blasthole pressure is reduced therefore depends on the degree of charge decoupling. For a laterally decoupled charge, if the explosive diameter is reduced to one-third of the blasthole diameter, then the peak pressure will be reduced to approximately one-ninth (assuming ideal gas behavior) of that for a fully coupled charge. Generally, the reduction is greater than this because most explosives exhibit a decrease in velocity of detonation as charge diameter and degree of confinement decrease. To a reasonable approximation, the peak blasthole pressure P% for a decoupled charge can be determined from a knowledge of the coupling ratio/ c (defined as the ratio of charge volume to borehole volume), the explosive density and the velocity of detonation of the explosive, as follows P* = 0.25fl2PtxpWOO2 (13) Although presplitting and postsplitting have been successfully conducted using very large degrees of decoupling to reduce the peak blasthole pressure to levels equal to the tensile strength of the rock, most presplitting is performed with a charge diameter between a quarter and a half of the blasthole diameter, reducing the pressure by a factor of between 5 and 30. Perimeter charging generally features a charge distribution of between 0.5 and 1.0kgm" 2 . Selection of the combination of blasthole loading density and blasthole spacing is important in achieving a high quality final face. Drilling requirements can be reduced by using higher charge densities, but at the expense of increasing damage to the rock behind the blastholes. Figure 11 presents the (readily observable) damage contours around blastholes for various blasthole loading densities. The most common methods of reducing the loading density of explosives in blastholes include laterally decoupling the charges (charge diameter less than blasthole diameter), air decking, and
Loading densities (kg m"1)
Figure 11 Damage contours around a blasthole with various loading densities
84
Blasting Damage contour (shoulder hole) Back holes
Damage contour (back hole)
Designed excavation boundary
Figure 12 Damage from inner blastholes extending beyond perimeter blastholes
forming low density explosives by mixing the charge with an inert material such as polystyrene, salt, sodium nitrate, etc. Of equal importance in perimeter blasting as the correct selection of loading density in the perimeter holes is the loading density in the holes adjacent to the perimeter blastholes. Fully coupled explosive in blastholes placed too close to the lightly charged perimeter blastholes will produce damage extending beyond the perimeter blastholes, as indicated in Figure 12. The appearance of the face after blasting will indicate that the smooth blasting was unsuccessful, but the real fault may lie with the adjacent blastholes. It will frequently be necessary to adjust also the loading density in the holes adjacent to the perimeter holes.
333.2
Minimum standoff distance
When estimating the minimum standoff distance between the perimeter blastholes and the next row of blastholes the effect of fracture dilation due to vertical heave must also be considered, especially in large-scale bench-blasting operations. Under a high degree of fixation, large diameter charges in the back rows of blast patterns can produce extensive cratering, creating a zone of vertical heave which can extend backwards a distance greater than the height of the bench. It becomes very important when designing perimeter blasts to ensure that the degree of fixation of the charges does not increase as a result of reducing the charge density in the blastholes. A small, over-confined, air-decked charge, for example, can create more damage through fracture dilation than a large, fully coupled charge with a low degree of fixation. The primary requirements in perimeter blasting are a reduced charge density and an even distribution of charge, and this does not necessarily require a reduction in powder factor. For high benches, such as a 45 m highwall in a strip coal mine, fracture dilation will be minimal except in the upper 10 m of the bench. In relatively low benches, such as a 15 m bench in a large open-cut mine where large diameter blastholes are used, significant fracture dilation may occur over the full depth of the bench. In this latter case, it may be reasonable to assume that cratering due to the back row of blastholes will extend backwards for a distance at least equal to the depth of burial of the charge. The effect of fracture dilation on the stability of the exposed faces will be dependent on the orientation of the fractures relative to the exposure and on the roughness of the fracture surfaces. Certainly, as the apertures of the fractures increase, the shear strength decreases.
3.4
ENVIRONMENTAL ASPECTS
With increasing environmental constraints on the levels of disturbance induced by blasting operations upon nearby residents, there is an increasing need to be able to design blasting operations with greater precision. Environmental constraints on ground vibration from blasting vary around the world from 2 to 25 m m s " 1 , and over a similarly wide range for airblast overpressure. Environmental constraints on the levels of induced ground vibration and overpressure are becoming so demanding that many operations are incurring significant cost penalties in order to
Methods of Improving Blasting Operations
85
comply with required levels. The cost penalties are incurred as the sizes of blasts are decreased, operating bench heights are decreased and blasthole diameters are decreased. All of these factors are tending to reduce the number of tonnes which can be produced or excavated per manshift, and therefore tend to increase the cost of extraction. As a result, it is necessary for many operations to minimize these cost penalties by designing blasts to achieve levels of ground vibration and overpressure disturbance as close as possible to the permissible levels.
3.4.1
Overpressure
Equations (14) and (15) are commonly used equations for overpressure prediction and indicate that, like ground vibration, the peak level is controlled by the charge weight of explosive per delay and the distance from the blasthole dBL = 164.4 - 24 log
D W 1/3
(14)
or alternatively Pover = 3 3 0 0 ( ^
(15)
where dBL is the overpressure decibel level (linear weighting), D is the distance from the blasthole (m); W is the weight of explosive detonating per delay (kg) and P over is the overpressure level (Pa). Equations (14) and (15), although representing the best-fit expressions for describing a large database of overpressures, do not providevan indication of the degree of scatter in the data or the confidence in the prediction of levels. Figure 13 presents some of the data from the literature, plotted in comparison with equations (14) and (15). The comparison shows the inadequacy of the equation to predict accurately levels of overpressure from blasting, with the total scatter in level exceeding 20 dB at any value of scaled distance.
3.4.1.1
Sources of overpressure
The inadequacy of the overpressure equations is partly the result of the variability in the basic mechanisms producing the overpressure. The data in Figure 13 include cases of stemming ejection, face blowouts, exposed initiation systems, unconfined blasting and normal, well-controlled blasting. There may be several sources of overpressure from the one blast event, including the initiation system itself (particularly where surface-detonating systems are used), the venting of explosion gases either through the blasthole collar or through the free face, the vibration of the rock mass, and the movement of rock at the bench face. It is commonly accepted that the largest overpressure peaks will be produced by the venting explosion gases, and that after elimination of these the next major contribution occurs as a result of dBL= 165-24 log (scaled distance)
CD
Q.
120 \-
10
100
Scaled distance (m kgH/3)
Figure 13 Collection of overpressure data from a wide range of quarrying operations
86
Blasting
the rock movement at the bench face. Literature further suggests that the minimum possible level of overpressure at a location will be that level produced by the ground motion at the monitoring point. The results of recent detailed studies, however, suggest that after venting has been eliminated the peak levels of overpressure are caused by ground vibration at the face. Simple experiments, involving simultaneous measurement of vibration levels and overpressure, reveal a linear relationship between ground vibration and overpressure (Figure 14). This relationship shows how vibration produces its own source of overpressure. Vibration at the bench face is converted to overpressure which then propagates at the characteristic sound wave velocity through air, arriving at the monitoring location significantly after the vibrations propagating through the ground. Ground vibration at the monitoring location also produces an overpressure signal, explaining why there is always a low amplitude precursor overpressure signal, arriving at the same time as the ground vibration wave, before the onset of the main overpressure signal. Further evidence that the vibration produces the overpressure pulse is seen in Figure 15, which shows the similarity in waveform shapes for the vibration at the face of a brick wall and the overpressure measured very close to the brick wall when the wall is struck with a large hammer. The data of Figure 14 clearly indicate a linear relationship between the level of vibration and the measured level of overpressure using a linear weighting. Vibration-induced overpressure levels (measured in Pascals) are directly proportional to the level of vibration, according to the equation (16)
Pover = 0.38PPV where PPV is the peak particle velocity measured in mm s *.
200
^s^m m
150
m m
^M
m
-^m
100
50
Λ ^ ^
1
1
1
1
1
1
-1
Vibration (mm s )
Figure 14 Correlation between peak overpressure and vibration levels at a rock face
Overpressure
0.002
0.004 Time (s)
Figure 15 Agreement in shape between vibration and overpressure waveforms indicates a common source
Methods of Improving Blasting Operations
87
The above equation predicts that a vibration level of 2650 mm s "* will produce an overpressure level of approximately 1000 Pa (154 dBL re 20 x 10 " 6 Pa), and that a vibration level of 10 mm s " 1 in the ground will produce a level of overpressure of 3.8 Pa (106 dBL). The results of prediction using equation (16) are in agreement with other literature [7], and it must therefore be considered reasonable to expect high levels of overpressure to be generated at the face of a bench blast or the surface of a fully confined blast, due solely to the levels of induced vibration which occur before any gas-induced rock displacement. At many blasting sites, high levels of overpressure are being experienced despite the complete elimination of venting from either the blasthole collars or the bench faces. Furthermore, monitoring using synchronized video recording and full waveform monitoring indicates that the peak level of overpressure is often achieved prior to any detectable movement at the bench face. The conclusion from these observations is that the peak level of overpressure, at least in cases where venting and stemming ejection have already been eliminated, can only be produced by the vibration or shock levels generated at the bench face by the detonating explosive, and that rock movement plays a secondary role in overpressure generation. If face vibration is a major source of overpressure, then the peak level can be reduced by reducing the peak level of vibration. Using the near-field form of the scaled distance equation to calculate the levels of induced vibration very close to blasthole, the level of vibration at the bench face can be calculated for any diameter and length of blasthole, with any type and strength of explosive. Applied to the case of a 12 m bench and a blasthole of 75 mm diameter and 13 m length (1 m of subdrill), pour-loaded with ANFO, the calculated level of vibration at the face in front of a blasthole is approximately 1750 m m s ' 1 . From equation (16), the calculated overpressure level at the bench face is around 150 dBL, and this level will reduce at the rate of approximately 7-9 dBL per doubling of distance. In the absence of venting, peak overpressure levels have always been observed to be generated by face holes, and frequently by the first hole to detonate. This immediately suggests that blasts with three or four rows of blastholes will have less environmental impact than blasts with the same number of holes and only one or two rows of blastholes. Furthermore, large blasts fired less frequently will produce less environmental impact than small blastsfiredmore frequently. To reduce the environmental impact of overpressure it is necessary to reduce the number of face holesfiredand the frequency of blasting. In order to reduce further the level of overpressure, the level of vibration at the bench face can be reduced in several ways. (i) Introducing an air deck in each face hole approximately halves the amount of explosive in those holes relative to other holes. Using equation (10), this will reduce the level of vibration at the face from 1750 mm s " x to 1100 mm s ~ *, producing a reduction in overpressure of around 5 dBL. Note that air decking is frequently introduced without changing the burden and spacing on the front row blastholes. (ii) Reducing the blasthole diameter in the front row holes only will have the following effect. Using equation (10), a reduction in hole diameter of 20%, while maintaining a constant burden, will reduce the vibration levels at the bench face by around 30%, producing a reduction of around 3 dBL. (iii) Increasing the front row burden relative to the burden on all other rows will also be effective. Using equation (10), an increase in front row burden of 20% will decrease vibration levels at the bench face by around 20%, producing a reduction of around 2 dBL. Full-scale blasts have been fired, utilizing 50% air decks in all face holes of the pattern. Overpressure levels at 180 m were reduced from 132 dBL to 127 dBL, in good agreement with the expected reductions indicated by the calculated reductions in vibration level. These reductions have been achieved in conjunction with an increase in the average size of the blast, accomplished by the firing of more rows of blastholes. Success, however, also requires that stemming ejection must be totally eliminated. All of the above forms of overpressure reduction can be expected to cause an increase in the percentage of oversize material generated in the front row of blastholes, particularly if the rock mass is hard and blocky. The methods also considerably reduce the burden movement velocity, so that the resulting muckpile is considerably higher and less scattered. This may present some problems for some types of excavators. An alternative which has been successfully implemented is to apply one of the above design modifications to only the first few front row blastholes to initiate. This has been conducted in situations where the peak levels of overpressure are consistently generated by the first hole to detonate, and reductions in level of around 3 dBL have been consistently achieved.
88
Blasting
Changes in fragmentation due to reduced charging of front row blastholes has not been observed. This is possible because the front row generally produces relatively coarse fragmentation anyway, as a result of preconditioning by the previous blast. Generally, fragmentation has been improved by an increase in the number of blasthole rows, since most oversize appears to be from the first and last rows of bench blasts. 3.4.2 Ground Vibration Ground vibration levels are generally predicted using expressions such as the USBM equation ppv
=
1143
(^i) "
(17>
where the symbols are as defined earlier (equations 7, 14 and 15). In general, the USBM equation gives reasonable estimates of the level of vibration, but users must again realize that the standard deviation for this equation is high, so that for a predicted mean level of 5 m m s " 1 the actual level can be expected to lie somewhere in the range 2.5-10 m m s - 1 . Expressed differently, in order to ensure that the level of vibration is less than 5 m m s " 1 on 95% of occasions, the blasting operator must design for an average level of around 2.5 m m s " 1 . Further shortcomings associated with the normal vibration equations relate to their inability to predict the effects of various important blast design aspects such as delay sequencing, delay intervals and the number of blastholes. An alternative model is suggested, for very site specific applications, which will permit accurate evaluation of the effects of varying all of the above blast design parameters, and which will also permit estimation of the statistics of vibration scatter, from which 90 or 95 percentile limits of vibration can be determined. The model is based on the measured vibration response from a single blasthole in alliance with the principle of superposition. 3.4.2.1
The seed waveform model
The principle of superposition states that providing the separate ground vibration responses can be described as linear elastic, then the resulting vibration from two or more sources can be obtained by simple addition of the separate responses, taking their phases into account. The procedure is demonstrated in Figure 16 for two blastholes separated by 25 ms. The approach was used by Blair [8] and Hulmes et al [9] to model the surface ground vibration response due to approximately 1000 delayed charges detonated in a large underground mass blast. Implicit in the application of this model is the assumption that the vibration pulse shapes and amplitudes from identically charged holes will be the same. The validity of the superposition principle and the reproducibility of vibration waveform shape can readily be confirmed experimentally. Figure 17 presents the recorded vibration waveforms from four separate, single blastholes, measured at the same location approximately 500 m away. The upper two waveforms were recorded on one day, without moving the triaxial gauge between firings. The lower two were fired approximately four months later, using a different explosive type, and measured at the same location as for the previous firings, although the vibration sonde had been removed and relocated. The figure also indicates the charges in each blasthole. In Figure 17 the similarity in shape, not only between signals recorded on the same day, but also between the two sets of signals, confirms the assumption of reproducibility in waveshape for these studies. 3.4.2.2
Superposition of waveforms
Since the exact initiation timing of every hole in a blast is rarely known, Monte Carlo simulation is performed, based on the experimentally determined scatter of the delays used in the blast. The singlehole 'seed' waveform is added to itself, after appropriate delaying for each blasthole. The firing times for blastholes are varied and normally distributed about the nominal firing time, so that a different waveform is produced for each simulation.
Methods of Improving Blasting
"> E E
Operations
89
400 H 200 -
0-200 "55 - 4 0 0 —\
>
'
-600 H
Figure 16 Superposition of waveforms
Figure 17 Reproducibility in vibration waveform shape for single-hole blasts, (a) 55 kg ANFO; (b) 75 kg emulsion; (c) 70 kg emulsion; (d) 60 kg emulsion
90
Blasting
Because the degree of interaction between successive 'seed' waveforms also varies with the individual blasthole initiation timing for each simulation, the peak vibration amplitudes for each simulation vary. By recording the peak levels for each simulation and repeating the simulation many times, the model is able to give estimates of the scatter such that 95 percentile vibration levels can be determined. It will be immediately noted that the level of vibration from a multihole blast is considerably higher than the level from a single hole. The simple USBM type of equation, however, indicates that the levels should be the same, since the factor controlling the peak level in these equations is the charge weight per blasthole. To predict the levels of vibration from production blasts at a specific location it isfirstnecessary to establish the single hole response at that location, and this is done for blastholes detonated at various locations around the site. Although the reproducibility of waveform shape from single holes in close proximity to each other is easily verified experimentally, it will also be observed that there is quite major variation in waveform amplitudes. The variability may be as high as a factor of two or more and cannot be explained by varying charge weights, suggesting that there are other factors exerting a strong influence over the peak levels of induced vibration. The primary factors expected to influence the levels of induced vibration include: (i) the degree of confinement, or the amount of burden on the blastholes, and the competence of the rock around the charges (degree of fixation); (ii) the degree of water saturation, affecting the degree of coupling of shock energy to the rock; and (iii) the detonation efficiency of the explosive, affecting the partitioning of shock and heave energy of the explosive. Most of these factors are beyond the control of the blaster and act to increase the variability in vibration levels which can be expected from 'identically' charged blast patterns. Variable coupling at the monitor is another influence which can be eliminated in the simple experiment by not disturbing the vibration sensors between successive single-hole firings. Under normal conditions, where the sensors are reinstalled for each monitoring, the efficiency of coupling will contribute to the data scatter. Permanent monitoring stations are one way to avoid this problem. 3.4.2.3 The coupling factor It has been observed during field studies of single-hole vibration monitoring that where the vibration levels from similarly charged blastholes differ markedly, the vibration attenuation curves display a vertical displacement. That is, the slopes of the attenuation curves remain constant but the vertical intercepts vary. Figure 18 presents the scaled distance curves for two single holes measured at five locations simultaneously. Clearly, each scaled distance curve is indicating very similar behavior, but, equally clearly, there is considerable vertical offset between the different curves. Despite the similarity in waveform shape, each single-holefiringhas produced a scaled distance curve with the same average
\ \ \ \ J
I
l+i » I
l/2
Scaled distance (m kg" ) Figure 18 Variation in vibration scaling for two single-hole blasts (46 kg emulsion ■ ; 49 kg emulsion + )
Methods of Improving Blasting Operations
91
V)
E E
υ o
0)
> \>
o a.
J
I
I
Scaled distance (m kg
I
I l/2
I" I I I
)
Figure 19 Vibration scaling using the coupling factor to adjust effective charge weight (CF adjusted ■ ; best fit —)
slope but with a quite different intercept, indicating a variability in the initial amplitude at the blasthole. This is considered to reflect a difference in energy coupling. The variability of the 'seed' amplitudes is therefore interpreted as a variability in the 'apparent' weight of explosive in the blasthole. In some cases, 10 kg of explosive produces the equivalent vibration to a 20 kg charge, and on other occasions produces the equivalent of 5 kg of explosive. A coupling factor term has been used to define the 'apparent charge weight' according to Wapp = CFW
(18)
where Wapp is the 'apparent charge weight', CF is the coupling factor and W is the actual weight of explosive in the blasthole. The 'apparent charge weight', when applied to equation (18), alters the intercept of the vibration attenuation curve and places all data on the same regression line, as shown in Figure 19 for the data of Figure 18. The range of coupling factors required to 'normalize' the vibration data in this manner may vary over different intervals for different explosive types. In practice, the coupling factor is commonly observed to vary over the interval 0.5-2.0, and may vary over different intervals for different explosive types. Bulk emulsion explosives, for example, appear more variable than ANFO. Field studies indicate that the coupling factor is influenced by factors such as degree of fixation and velocity of detonation, and is a major source of scatter in local vibration versus scaled distance graphs. The single-hole blasting trials indicate that the coupling factor for each hole in a blast may vary, and that the range of variability may be related to the type of explosive used. These variabilities can readily be incorporated into the seed waveform model using Monte Carlo techniques. The effect of including a variable coupling factor for each blasthole, with an average coupling factor of unity, is to increase the maximum variability in peak amplitude with little or no influence on the mean amplitude. Where the coupling factor does not have an average value of unity, then both the mean amplitude value and the variability in peak amplitude will be affected. The total scatter in amplitude can therefore be attributed to two separate mechanisms for monitoring stations that feature a permanently located monitor: (i) the scatter in delay firing times for detonators; and (ii) the variability in vibration amplitude at the blasthole caused by variable confinement, coupling or explosive performance. Where the monitoring is performed using a vibration gauge which is constantly moved and relocated, an additional degree of variability will be produced by virtue of the varying effectiveness of the coupling at the gauge.
3.4.2.4
The influence of blast size
Trial blasting is frequently undertaken to establish site specific vibration attenuation equations. Trial blasting is generally conducted using single blastholes, and the vibration parameters are used to predict levels from large blast patterns. This approach is considered to be the most accurate and
92
Blasting
reliable method, but the user may frequently find that the equations describing single-hole behavior underestimate the levels of vibration from a large blast pattern. Although the commonly used vibration equations indicate that vibration levels are dependent on charge weight per delay, and independent of the number of blastholes in the blast pattern, in practice it is observed that there is a vibration enhancement effect when multiple holes arefired.The extent to which levels are enhanced is very site specific, and influenced by delay timing. The enhancement effect is equivalent to an increase in the apparent charge weight per delay. Using the seed waveform model to simulate the range of vibrations at a particular location, it is possible to develop simple linear approximations to predict the apparent charge weights as a function of blast size. Typically, for a blast containing around 30 blastholes the apparent charge weight will increase above the single-hole charge weight by around 5% per additional blasthole, so that for a blasthole charge weight of 10 kg the effective charge weight for a blast containing 11 holes would be around 15 kg. The effect of increasing the blast size for two particularly sensitive locations is shown in Figure 20, using the characteristic single-hole vibration waveforms from the two sites shown in Figure 21. In both instances, the peak level of vibration increases approximately linearly over the range studied, though the rate of increase is quite different for each site. The variability in apparent charge weight according to blast size is one more factor tending to increase the degree of scatter in the blast vibration data collected from sites.
.'■■
E ^
12
/" ■
o 10 If-
>
Γ / 8
o
6 I ■-■
a>
a) o
Case 1
/
^Λ"
/
Γ
. - + - —+
-+ Case II
+
- ""
+ 1
1
*¥'
1
1
10 15 20 Number of blastholes
1
. ...J 30
Figure 20 Effect of blast size on peak vibration levels for two different operations
(a)
Figure 21 Variability in characteristic vibration response for different operations, (a) Case I: 30 kg at 145 m. (b) Case II: 4.25 kg at 150 m
93
Methods of Improving Blasting Operations 1 1
E
E
t: Ü o
9 8
s
7
Û.
°
l +
Case 1 / \
/
, Case II
S 4
Q_
C)
\
1
1
1
1
1
1
1
IC
20
30
40
50
60
70
8<
Delay interval (ms)
Figure 22
3.4.2.5
Effect of delay interval on peak vibration levels for two different operations
The influence of delay timing
Delay timing will invariably influence the peak amplitude of ground vibration by affecting the degree to which vibrations from successive blastholes will enhance the overall level. Some degree of enhancement is inevitable, since the duration of vibration from a single hole is invariably considerably longer than the effective delay interval between charges. The variability in duration and shape for single holes is illustrated in Figure 21 for vibration responses from single holes measured at approximately the same distances in two different mine sites and rock types. In general, the duration of the waveform increases with increasing distance from the blast site. The logical conclusion from this is that the site vibration attenuation equation determined from single-hole blasts will underestimate the levels of vibration induced by multihole blasts with similar blasthole charge weights, even where holes are individually delayed. In practice, the extent to which peak vibration levels are enhanced by the firing of multiple holes is determined by the number of blastholes, the effective delay intervals between successive hole detonations and the features of the single-hole, characteristic vibration waveform. Figure 22 presents the variation in peak vibration levels for several effective delay intervals using the two characteristic waveforms of Figure 21. The simulations for this figure were conducted using the seed waveform model described in Section 3.4.2.2, assuming a blast size of 30 holes and a fixed hole geometry and charge loading. In the first case, there is a distinct minimum peak vibration level for an effective delay interval of around 20 ms, equivalent to the electric 'L-series'. In the second instance, the minimum peak vibration level occurs for an effective delay interval of around 5 ms, equivalent to the use of a typical nonelectric initiation and delaying system. In situations where environmental compliance is marginal, adjustment of delay intervals in conjunction with seed waveform modeling can provide sufficient relief to avoid litigation and complaint. A second major influence of delay timing is on the effective vibration frequency, and on the ability of the induced vibrations to excite resonance in structures such as nearby housing. This aspect of vibration control has been covered well by Djordjevic et al [10] and Crenwelge [11]; it is also covered in Chapters 4 and 5 of this volume. 3.5
REFERENCES
1. Langefors U. and Kihlstrom B. The Modern Technique of Rock Blasting, 3rd edn., pp. 28-64. Wiley, New York (1978). 2. Clark G. B. Principles of Rock Fragmentation. Wiley, New York (1987). 3. Afrouz A., Hassani F. P. and Ucar R. An investigation into blasting design for mining excavations. Min. Sei. Technol. 7, 45-62 (1988). 4. McKenzie C. K. Blasting in hard rock: techniques for diagnosis and modeling for fragmentation and damage. In Proc. 6th Int. Congr. Rock Mech. Montreal (Edited by G. Herget and S. Vangpaisal), pp. 1425-1431. Balkema, Rotterdam (1987). 5. Grant J. R., Spathis A. T. and Blair D. P. An investigation of the influence of charge length upon blast vibrations. In Proc. 6th Int. Congr. Rock Mech. Montreal (Edited by G. Herget and S. Vangpaisal), pp. 637-641. Balkema, Rotterdam (1987). 6. Holmberg R. and Persson P. A. Design of tunnel perimeter blasthole patterns to prevent rock damage. In Proc. I MM Tunneling '79 Conference, London, pp. 280-283. (1979).
94
Blasting
7. Siskind D. E., Stachura V. J., Stagg M. S. and Kopp J. W. Structure response and damage produced by airblast from surface mining. Rep. Invest. - U.S., Bur. Mines, RI-8485 (1980). 8. Blair D. P. The measurement, modeling and control of ground vibrations due to blasting. In Proc. 3rd Int. Symp. Rock Fragmentation by Blasting, Brisbane, pp. 88-101. Society of Experimental Mechanics, Bethel, CT (1990). 9. Hulmes M., LeJuge G., Ellison C. and McKenzie C. K. Improvements in blasting practices at Mount Charlotte through vibration monitoring and analysis. In Proc. 2nd Int. Symp. Rock Fragmentation by Blasting, Keystone, CO (Edited by W. L. Fowney and R. D. Dick), pp. 530-540. Society of Experimental Mechanics, Bethel, CT (1987). 10. Djordjevic N., Kavetsky A. and Scott A. Blast design optimization to minimize induced vibrations of structures. In Proc. 3rd Int. Symp. Rock Fragmentation by Blasting, Brisbane, pp. 373-380. Society of Experimental Mechanics, Bethel, CT (1990). 11. Crenwelge O. E., Jr. A frequency domain approach for predicting and minimizing blast-induced ground vibration. In Proc. 2nd Int. Symp. Rock Fragmentation by Blasting, Keystone, CO (Edited by W. L. Fowney and R. D. Dick), pp. 114-119. Society of Experimental Mechanics, Bethel, CT (1987).
4 Blast Monitoring: Regulations, Methods and Control Techniques DOUGLAS A. ANDERSON Tensor Technologies, Hazleton, PA, USA
4.1
INTRODUCTION
95
4.2
BLAST VIBRATION - GENERAL
96
4.2.1 4.2.2
Ground Vibration Airblast
96 97
4.3 METHODS 4.3.1 Instrumentation 4.3.1.1 Components 4.3.1.2 Placement 4.3.1.3 Procedure 4.3.2 Analysis 4.3.2.1 Peak levels 4.3.2.2 Waveforms 4.3.2.3 Fourier spectra 4.3.2.4 Response spectra 4.3.2.5 Velocity exposure level 4.3.2.6 Airblast analysis 4.4 REGULATIONS
97 97 97 98 99 99 100 100 100 101 102 102 102
4.4.1 US Bureau of Mines/ OSM 4.4.1.1 Ground vibration 4.4.1.2 Airblast 4.4.2 ISO/European Standards 4.4.3 Complaints versus Compliance 4.5
102 102 103 104 104
CONTROL
105
4.5.1 Ground Vibration 4.5.1.1 Scaled distance 4.5.1.2 Vibration control using delays 4.5.2 Airblast 4.6
105 105 105 107
PRODUCTIVITY
4.6.1 4.6.2
108 108 108
Determination of Firing Times Relative Fragmentation and Displacement
4.7
CONCLUSIONS
109
4.8
APPENDIX - RESPONSE SPECTRUM CALCULATIONS
109
4.9
REFERENCES
110
4.1 INTRODUCTION Much of Comprehensive Rock Engineering deals with what I will call direct engineering: how to get a job done safely, efficiently, economically and, if possible, elegantly. In these times, nonengineers (lawyers, government regulators, town councils, etc) may become part of the engineering process 95
96
Blasting
because of potential adverse environmental impacts. Though their input is typically nontechnical, it must evoke a technical response. This response, which I will call indirect engineering, may be at odds with the rest of the engineering process. An organized approach which has not considered the environmental impact from the outset may be compromised. Furthermore, a response based solely upon attempts to comply with regulations may result in poor engineering practices, and may not even satisfy the regulations. Direct blast engineering has been discussed in previous chapters by Fourney and Mackenzie (Chapters 2 and 3, this volume). This chapter deals with indirect engineering, specifically considering environmental effects of blasting. The effects discussed here are ground vibration and air overpressure or airblast, which I will refer to collectively as blast vibration (other effects, such as flyrock, dust and production of noxious or flammable gases in underground workings are not directly treated here). Blast vibrations may disturb surface structures such as houses, below ground structures such as pipelines, or the workings of underground mines. In this chapter blast monitoring in its simplest form means measuring blast vibration and comparing the measurements with regulations. Most countries have regulations which specify acceptable levels of vibration generated by blasting. The regulations may be implemented at national or local levels, or both. They are typically based upon research which relates vibration levels to structural damage. If there is an indication that the measurements exceed the regulations, blast designs must be changed. Though regulations are the main reason for blast monitoring, compliance with regulations does not guarantee that there will not be complaints. It must be remembered that regulations are to protect those who are disturbed by the vibration, and their concern that the vibrations are damaging a structure. Though blast vibration levels are oftçn kept well below the established criteria for avoiding damage, they may still annoy neighbors if they are low frequency and resonate the structures. Such annoyance will lead to public relations problems and may lead to litigation. This point will be stressed repeatedly. Mine operators may randomly try one blast design after another in the search for a solution. Recent research offers hope for this situation. Rather than the usual trial and error methods to control vibration problems, an operator may use an integrated, intelligent approach. Actions taken to comply with regulations may adversely affect productivity. I will discuss how the blaster may comply with regulations without unduly affecting productivity. The issues I will discuss are as follows: (i) discussion of blast vibration generation; (ii) monitoring methods; (iii) analysis to determine causes of existing or potential problems; (iv) control techniques to solve problems; and (v) productivity concerns. Since this volume is meant as an engineering guide, emphasis will be placed upon understanding problems and finding appropriate solutions. Though regulations must be complied with, existing control techniques to achieve compliance do not always properly address the problem. Much of the literature on instrumentation and regulations is in government publications, and on control and productivity is in the technical literature. A sampling of these will be discussed later. For general reference, two available books on blast monitoring are Dowding [1] and the classic work of Bollinger [2]. Other information is available in handbooks on blasting given by explosives manufacturers. 4.2 BLAST VIBRATION - GENERAL Detonation converts a solid explosive to a very hot gas in a very short period of time. The rapid expansion of this gas exerts a pressure pulse on the borehole wall which propagates away from the borehole. In the neighborhood of the borehole the stress induced by this pulse exceeds the elastic limit of the rock, and fracture occurs. The processes which produce this fracture are complex, and are discussed in detail by Fourney (Chapter 2, this volume). However, to complete this rudimentary picture of the rock fragmentation process, the gases, after giving the initial punch to the rock, expand more slowly,fillingthe new and existing cracks in the rock, pushing and moving the rock mass. The punch gives rise to ground vibration, and the push to normal airblast. 4.2.1 Ground Vibration The pulse produced by the explosive decays as it propagates from the borehole, due to the work done by fragmentation, heat generated by anelastic processes and geometric spreading. Eventually,
Blast Monitoring: Regulations, Methods and Control Techniques
97
the stress is below the elastic limit of the rock and passes through the rock as an elastic wave. Like echoes in a canyon, the waves spread out in many directions. Waves propagating through a rock differ from echoes in a canyon because the rocks are layered, inhomogeneous and have different velocities. The waves are reflected and refracted as they pass through the rock, and the pulse, which initially had a duration approximately equal to the detonation time of the explosive, is lengthened into a wave train which may be several seconds long for a single borehole at a distance of a few hundred meters. Depending on the geology and the delay sequence of a blast, various frequencies may be accentuated. Waves propagated from the blast may become trapped in layers near the surface. These trapped waves will resonate (much as the structures resonate) at frequencies determined by the thickness and type of material at the surface. When these waves arrive at structures near the blasting operation, problems may arise from two sources: high amplitudes, which may force a structure to move, and resonances, which transform a low-level vibration into one which may be annoying or damaging. If the frequency of the ground vibration matches that of a structure, the vibration may be amplified. Most of this effect occurs near the receiver site (the house or structure). The mine operator has no control over local geology; however, he can minimize the effects of unfavorable geology by altering the blast design.
4.2.2
Airblast
As discussed above, the movement of the rock mass generates primary airblast. If a rock face is to move, there must be airblast. In open pit operations the movement of the face may approach 50 m s"*. This rapidly compresses the air, which then propagates as a wave away from the blast site. This pressure pulse is infrasonic ( < 20 Hz), and may be within the frequency range of residential structures, which may amplify this pulse. The audible 'crack' heard from a blast is not airblast, and, while it may be annoying, it will not be a cause of structural damage. Other pulses of air may be generated by a blast, but are not necessary like the primary airblast. Several factors may cause such an air pressure pulse to be generated: (i) venting of material through the collar of a borehole due to inadequate stemming; (ii) ejection of material from the face due to inadequate burden or clay seam; and (iii) heave of back rows of a shot due to excess confinement. These types of air pressure pulses are usually higher frequency than the primary airblast pulse and may be accompanied by flyrock or excessive dust. Proper blast practice (discussed later) should alleviate them. Finally, large, low-frequency motion of the ground surface away from the blast site may generate an air pressure pulse, but this is usually of secondary importance. The airblast generated by any of the above causes may be modified by atmospheric conditions, primarily wind and temperature inversions. Wind may add to an airblast pulse if it is moving in the same direction. A temperature inversion will reflect the airblast pulse, much as a rock layer reflects a ground vibration wave. Since these effects are unpredictable, it is best to avoid blasting if possible if it is very windy or there is a temperature inversion (sometimes indicated by a low cloud ceiling). 4.3
METHODS
4.3.1
Instrumentation
The instrumentation chosen for blast monitoring must be appropriate for the purpose. Conformance to regulations may dictate a particular type of instrumentation. However, other considerations, such as understanding why certain blast designs annoy neighbors or why blasts are performing poorly, may indicate that another type of instrumentation should be used. Rather than discuss elements of instrument design and performance (these can be obtained from brochures and spec sheets), I will emphasize the components of instrumentation which will provide the most useable information from a rock engineering standpoint. Then I will discuss instrument placement and procedures for obtaining useful data.
4.3.1.1
Components
Every instrument must have three components: (i) a transducer or transducers to measure the phenomenon; (ii) signal conditioning circuitry; and (iii) recording medium. The transducer to measure ground vibration should measure the appropriate regulatory criterion directly. Most criteria are based upon particle velocity, and a velocity transducer is appropriate. In
98
Blasting
some cases acceleration must be measured, and in this case an accelerometer is appropriate. Displacement, velocity and acceleration as a function of time may each be obtained from the other measurements by differentiation or integration; however, there are potential problems with these procedures, and it is generally best to record the desired measurement directly. Some regulations specify measurement of only the vertical component of vibration. Structures are more sensitive to horizontal vibration, and it is recommended that three components of vibration be measured, even if only one is used for compliance with regulations. Annoyance from ground vibration within compliance may be indicated by vibration on horizontal components. The critical performance characteristic of a transducer is a linear response across the frequency range in consideration, or if the response is nonlinear, appropriate signal conditioning circuitry that produces output which is linear. The frequency range is generally specified in regulations. The frequency range for ground vibration is typically 2-200 Hz. This is generally not a problem with geophones with signal conditioning. There is a fundamental problem with practical microphones for measuring airblast. The microphones generally do not measure the low frequency pulse directly, and must be compensated either through signal conditioning or by using a regulatory criterion which takes this lack of complete measurement into account. This will be discussed in more detail in Section 4.4 (Regulations). In general, the signal conditioning for both ground vibration and airblast should not be the concern of the user. The appropriate specifications should indicate if the instrument is appropriate for the intended use. The recording medium, though, is critical for the type of engineering purpose. For compliance with regulations, a particular medium may be designated. The minimum is a printed record of peak values or a meter indicating peak values which may be transcribed. Rarely should a practicing engineer be satisfied with such a record. It is impossible to determine if the readings are actually due to the blast, or may be due to some extraneous source. Even if the record appears to be due to the blast, no understanding of the vibration generated can come from such a record, and therefore no adequate means of control can be adopted. The complete waveform should be recorded. This is usually practical, since blast vibrations are transient events. While a printed record only may be useful in some instances, it is difficult to do any further analysis on the data, since they must be digitized. It is far more effective to record the data on magnetic tape or disks directly. 43.1.2 Placement The instrument should be placed properly at an appropriate location. Proper placement involves both the geophone and the microphone. (Parts of these recommendations are taken from ISO 4468 and US Bureau of Mines RI 8969 [3].) The appropriate location may be dictated by regulations, occurrence of complaints or both. In all cases, be as consistent as possible from blast to blast, and note clearly when any modification of either recording technique or location is made. The accuracy of any ground vibration recording is only as good as the coupling of the geophone to the ground surface. Any decoupling of the geophone from the ground may result in transducer movement different from the ground movement. When decoupling occurs, the seismic reading will almost always be greater than the ground movement. The problem of decoupling is most often seen at short distances, where the highest frequencies are present, as well as the highest particle velocities. Burying the geophone may seem to afford the best coupling. However, this method is often impossible due to the disturbance to the site. In addition, at sites where the transducer is left for a long period of time, coupling with the soil may change with time (due to freeze-thaw cycles and water saturation). Regardless of the vibration level, recordings should be made with the geophone mounted on solid earth, whenever possible. It is important to make certain that no part of the geophone is supported by grass or grass roots. Spiking the geophone will inhibit excess horizontal motion relative to the ground. Such a spike should be only a few centimeters long and about a centimeter in diameter. If used properly, a sandbag can prevent decoupling in most situations. The geophone mustfirstbe firmly secured and a sandbag placed over it. The sandbag should be loosely packed and large enough so that, when placed on the geophone, all sides of the bag are resting firmly on the ground surface. In some cases, regulations require measurement on a structure or slab. Solid surfaces should have a firm attachment, preferably by studs or high-modulus resin. Double-faced tape should be avoided. A slab may amplify certain frequencies of the ground motion, which can result in a higher reading.
Blast Monitoring: Regulations, Methods and Control Techniques
99
This does not mean that the resulting seismic data are invalid. However, make sure that you provide detailed information as to how and where the geophone was mounted so that the data can be correctly interpreted. The microphone should be either hand-held or mounted on an appropriate permanent mount supplied with the instrument. In either case, it should be above the ground, at a distance from a reflecting surface such as a wall at least half the height of the surface. Different types of microphones require that it point either to the blast or straight up - follow manufacturer's recommendations. A windscreen may be used to reduce extraneous measurements from wind. Unattended operation on a mount may result in movement of the microphone due to the ground vibration, known as microphonics. While this can often be distinguished from airblast because the airblast will arrive later than the ground motion, it is best to avoid it in the first place. The human body is a pretty good filter for microphonics, and a hand-held microphone appears to be best. Monitoring for surface mines is generally done at noncompany owned structures, usually for both airblast and ground vibration; however, this is location dependent. Monitoring for underground workings may be done at surface locations (for ground vibration only), and in the workings for protection of the working, facilities and personnel. The best place for a seismograph to be placed is at the corner of the structure being monitored closest to the blast, on the ground. It should not be inside the structure or on a concrete or macadam surface outside unless required by regulations. Sometimes an additional seismograph may be placed inside the structure, but this instrument should not be used either to determine compliance with regulations or for vibration control techniques. The 'structure being monitored' is important. Some regulations stipulate that the nearest noncompany dwelling be monitored; if so, comply. Other regulations do not have such a stipulation. In this case, it is best to monitor at a location at or near where there have been complaints. Do not put it in an empty field somewhere. This does not give any information of use.
4.3.1.3
Procedure
The actual recording procedure should be kept as simple as possible. (i) Place the instrument and turn it on. For permanent locations it should be waterproofed and secured. (ii) Check battery and trigger levels (if appropriate). (iii) Make sure the recording medium (disk, tape, paper) is in place and sufficient to record the expected events. (iv) Calibrate. (v) Put instrument in monitor mode. (vi) Record information after shot or shots. Internal calibration of the instrument is generally done by applying a voltage pulse to a geophone and determining the response. A microphone calibration is generally only of the internal circuitry. It is recommended (and some regulations require) that the seismograph's internal calibration be verified by an external calibration by the manufacturer. This is normally done annually. The external calibration of the geophone is done on a shake table, ensuring that the geophone functions properly over the range of frequencies in the specifications. The calibration of the microphone is done with a pistonphone, which applies a calibrated pressure to the microphone itself. As important as the actual vibration recording is the back-up paperwork about the blast. This includes, as a minimum: (i) time; (ii) date; (iii) blast location; (iv) instrument number, location and distance from blast; (v) total pounds; (vi) pounds per delay; (vii) meter or visual readings; and (viii) weather conditions. Other information may be useful, either in defending damage claims or to determine effects on productivity. Blast design (such as delay times and number of holes and rows) can aid in determining why some shots perform either well or poorly. When in doubt, put it down. It is rare indeed when someone has a problem because they have recorded too much information.
4.3.2
Analysis
Blast vibrations are waves, which all instrumentation will record and/or analyze. Some form of analysis is necessary to reduce the information in the waves to numbers which can be related to regulations or other uses. The simplest form of analysis is peak level determination. Since drastically
100
Blasting
different waves can have the same peak levels, use of peak levels only can be deceiving. The digitized record for even a simple wave may comprise several thousand points, so if analysis is to be done on the entire waveform, some form of summary is necessary. This summary must adequately represent the wave, and provide numbers which are meaningful for the purpose at hand. In most cases a type of spectral analysis is appropriate. Analysis of vibration and airblast data may be left to a blast consultant. If the blast monitoring is being done for regulatory purposes, it is important to have third party analysis of the data. First, there is then no question of the operator manipulating the data to suit his liking. Second, problems with instrumentation, calibration and improper recording procedures can be flagged before there is a bigger problem. Third, potential problems can be detected by the blast consultant. However, it is important that the end user understand the analysis and the terms used to describe the analysis. Furthermore, the blast monitoring may aid in understanding problems with the blasting process.
4.3.2.1
Peak levels
Peak particle velocity, the greatest velocity of a particle about its rest state as a blast vibration wave passes, is the basis for most regulations. It is easy to determine, and provides a single number to be compared with the regulations. However, because waveforms from blasts are complex, the peak particle velocity is only a gauge of the level of vibration. In a sense, peak particle velocity criteria assume that the spectral content and shape of all waveforms are the same. To overcome this limitation of peak particle velocity criteria, regulations now often incorporate a frequency along with a peak level (see Section 4.4). There are some deficiencies to this method, though. First, the regulations usually specify only a single dominant frequency, and vibration at another frequency which may excite structural response may be overlooked. Second, the association of a peak particle velocity with a particular frequency is not strictly meaningful except for pure sine waves. The peak velocity is determined by amplitude and phase relations for the frequencies present in the wave. The peak velocity can be altered significantly while keeping amplitudes for all frequencies the same by changing the phase relationships. Amplitude-frequency criteria are an improvement over straight particle velocity criteria; however, they still are not the best way to indicate probable damage potential. These are the current regulations, though. The proper way to determine compliance with these regulations is to determine peak particle velocity and a dominant frequency by spectral analysis.
4.3.2.2
Waveforms
Direct examination of waveforms may be useful in determining if an event is 'real', i.e. that it is a result of a blast, monitored properly. Some simple waveforms are obviously due to nonblast events, such as a kick to the transducer. Others take a trained eye to discriminate. The waveform may also indicate whether there are coupling problems. When in doubt, it is best to ask a vibration consultant to determine if an event is well recorded and from a blast. Waveforms are also useful in determining whether a shot has fired properly. In many cases a sharp spike in the vibration indicates a problem in the shot, such as with excess confinement. The waveform of the airblast record can indicate if there are blowouts. Repeated spikes may indicate insufficient stemming. Because waveforms are complex, resulting from explosive performance and efficiency, detonation times and the travel path, direct examination is of limited usefulness. Spectral analysis is usually the preferred analysis technique.
4.3.2.3
Fourier spectra
The classical analysis technique to determine the frequency content of a signal is the Fourier transform. The appropriate Fourier transform for sampled data is the discrete Fourier transform as follows N-l
Ηω = Σ hkexp(2nikœ/N) k=0
Blast Monitoring: Regulations, Methods and Control Techniques
101
where Ηω is the transform (complex) at frequency ω, hk are the data points, and N is the number of data points (excellent detailed discussions of the procedure, sampling considerations, and computer code are found in Numerical Recipes in C: The Art of Scientific Computing [4]). Fourier analysis effectively splits a signal or waveform into sinusoidal waves for each of the frequencies. Each of these sinusoidal waves is for the entire length of the sample. For a transient signal such as a blast wave, there are non-zero amplitudes at frequencies that are not 'present' from a structural response viewpoint, but which must be present to correctly produce the actual waveform. Fourier analysis is particularly useful if there is a single-charge waveform which gives the effect of the travel path (known as the 'Green's function') on the source impulse. By deconvolution using Fourier analysis, the seismic efficiency and firing times may be determined for a given shot. However, since both of these are not necessarily known a priori, the solution is indeterminate. Work on this particular area of blast monitoring is currently in progress. A critical shortcoming of Fourier analysis to determine the effect of vibration on structures is the effect of duration. A high-amplitude pulse at a given frequency will, for damped structures, produce less of a response than several cycles at lower amplitude. The information about duration is effectively transferred to other frequencies, and is not readily apparent from Fourier analysis. Response spectra are more suited for these needs. As Dowding [1] has noted: 'Because of the similarity of [Fourier and pseudovelocity response] spectra, either can be employed to determine the dominant frequency in the ground motion. However, only the pseudovelocity response spectrum can be employed to predict structural response.'
4.3.2.4
Response spectra
Response spectral analysis, described by Hudson [5] and adapted to blasting seismology by Medearis [6, 7] determines, by using a single degree of freedom model, the effect of a given vibration on structures. This is commonly used by civil engineers to determine response of structures to vibrations such as those caused by earthquakes. Medearis has shown [6] that response spectra can be used to determine damage probability. Residential structures typically resonate at frequencies in the range from 3 to 18 Hz, and damping near 5% of critical for horizontal vibration, and near 20% of critical for vertical vibration [6], Any structure will respond to vibration if the applied vibration has a frequency matching one of the resonances of the structure. Increasing the number of cycles will, depending on damping, increase the response. This means that if the ground vibration has frequencies within this range, a structure will amplify the vibration. Three spectra are typically calculated, one for each component of ground motion. Each of them shows relative velocity of the structure with respect to the ground versus frequency (in Hz) for each of the three components. The response spectra are displayed on log-log scales, so that relative displacement and acceleration may be determined directly (see Dowding [1] for details). The details of the calculation are described in an appendix (see Section 4.8). Briefly, though, response spectra are calculated by a series of steps as follows. (i) An appropriate damping is chosen for the component of vibration being analyzed. (ii) A structure frequency is chosen (3 Hz to start out with). (iii) The vibration is passed (in the computer) through a structure of that frequency. The maximum movement (displacement) of the structure relative to the ground is determined for that frequency. (iv) This is then converted to a velocity of the structure relative to the ground. (v) This is then done for another structure with the next frequency, up to 50 Hz. Engineered structures, such as high-rise buildings and coal tipples, often have well-defined resonant frequencies. Put the seismometer in the building if necessary. Levels of vibration are not used (unless required by law) to determine damage probability. Evidence of nondamage, based upon peak level determination, is often insufficient to assure homeowners that no damage is being done to their homes. Certain types of vibration, which are clearly indicated by response spectra, may cause substantial annoyance, even though they are far below the threshold for damage. This annoyance, coupled with the knowledge of the blaster that no damage is being caused, creates a situation where both parties in the dispute are right: the homeowner is annoyed, and the blaster is not causing damage. A response spectrum plot which looks like 'Mt Fuji' is a clear indicator of potential annoyance. This type of response spectrum indicates that vibration is tuned to a particular frequency, due to geology and delay times. Fortunately, this type of vibration problem is usually amenable to correction by vibration control using delays.
102
Blasting
4.3.2.5
Velocity exposure level
Another technique is the velocity exposure level (VEL) defined as [8] VEL = 10 log -
L^O J 0
t;2(r)df
This is analogous to sound exposure level, and measures the energy of a signal within specified frequency limits. This is cited primarily for completeness, since current studies [8] indicate that VEL is not an accurate predictor of damage.
4.3.2.6
Airblast analysis
Airblast has traditionally been analyzed strictly on the basis of peak levels. Dowding et al. [9] have shown that response spectral analysis can be applied to airblast, but modifications are necessary. For typical low-frequency airblast waves and typical residential structures, the pressure envelops the whole structure and the push on one side cancels the push on the other side. For substantially higher frequency airblast this will not occur. However, this type of airblast analysis has not yet come into general use, and peak levels remain the primary consideration.
4.4
REGULATIONS
The purpose of discussing regulations is to determine the types of regulations which exist or are being considered. It is impossible to give an accurate summary, because standards (such as US Bureau of Mines or ISO) are often superseded by local or regional regulations. These differences may not be merely changes to a more restrictive allowable limit, but may in fact be major reinterpretation of relevant research. For example, in the US, where frequency of ground vibration is part of regulations in many areas, there is considerable discrepancy in how to determine the appropriate frequency or frequencies for regulation. Is one type of spectral analysis technique preferred? Should only the predominant frequency be considered? If other frequencies are considered, how are they judged relative to each other? How does one associate a peak particle velocity (time domain) with a frequency (frequency domain)? These questions are as yet unanswered. The key is to associate damage and annoyance data with current analysis techniques. This has not yet been done, and until it is done, the regulations are likely to inadequately protect both neighbors and operators.
4.4.1
US Bureau of Mines/OSM
Currently most of the newer regulations in the US are based upon research done by the US Bureau of Mines on structure response and damage due to blast vibrations [8] and airblast [10]. These works summarized and extended earlier work by the Bureau. While they do not have regulatory weight on their own, they are used as the basis of national, regional and local regulations.
4.4.1.1
Ground vibration
For ground vibration, the criterion is contained in Appendix B of RI 8507. A combination of peak particle velocity and frequency gives a ramped function (Figure 1 shows the actual regulations of OSM). Peak particle velocity is that measured in any of the three orthogonal planes. A vector sum is not used. Unfortunately, as discussed above, while peak particle velocity is easy to determine, the appropriate frequency is not. No clear directions as to how to unambiguously associate a frequency with peak particle velocity were given in RI 8507. An alternative interpretation to the peak particle velocity/frequency quandary is to use the figure itself, where both peak particle velocity and peak displacement are used to draw the graph. The caption of Appendix B states 'Safe levels of blasting vibration for houses using a combination of velocity and displacement'; there is no mention of frequency. Vibrations exceed the Appendix B criterion if one of the following occurs: the peak particle velocity is greater than 2.0 inches per
Blast Monitoring: Regulations, Methods and Control Techniques
103
T3
0>
2.0 in s
O
c
Ö
o
>
υ
5
""I
10
50
100
Frequency (Hz) Figure 1 U.S. Office of Surface Mining regulations for ground vibration
second; or the peak (integrated) displacement is greater than 0.030 inches ( 1 inch = 25.4 mm); or the peak particle velocity is greater than 0.75 inches per second for drywall construction or 0.50 inches per second for plaster/lath construction; and the peak (integrated) displacement is greater than 0.008 inches. This simple measure appears to fit the Appendix B curve, and is consistent with the text of RI 8507. A low displacement and high particle velocity imply high frequency, but do not involve the determination of frequency in itself. OSM regulations do not have a constant displacement segment in the range from 10 to 30 Hz; however, a variable displacement criterion could easily be determined from the regulations. Even though the displacement is variable, it does not change much. The displacement (sinusoidal approximation) at 2 inches per second and 30 Hz is 0.0106 inches, and the displacement at 0.75 inches per second and 11 Hz is 0.01085 inches. An approximate average value of 0.0107 inches may be appropriate.
4.4.1.2
Airblast
The recommendations by the US Bureau of Mines for airblast are contained in RI 8485 [10]. As noted by the authors, air pressure pulses from blasting have a significant component in the 0.5 to 2 Hz range. There are two general types of airblast, called Type I and Type II. Type II, in addition to the low frequency energy, has substantial energy above 6 Hz. This energy is most likely to cause structural response and/or damage. However, for consistency, all of the energy is to be measured. Because of difficulties in producing low-cost airblast instrumentation at the very low frequencies (0.1 Hz), many manufacturers produce instruments with higher low frequency response, either 2-200 Hz, or 5-200 Hz, or what is called 'C-slow' response, a standard for sound-level meters. The recommendations (which are then incorporated into regulations) are then a function of the type of instrumentation as shown in Table 1, where the dB levels are related to peak sound pressure by the following relationship dB = 201og10— and P0 is a reference pressure of 20 x 10~ 6 N m~ 2 (2.9 x 10 ~ 9 psi).
Table 1 Types of Instrumentation Instrument type 0.1 Hz 2 5 or 6 C-slow
Allowable peak levels (dB) 134 133 129 105
104
Blasting
4.4.2 ISO/European Standards European standards do not include airblast. Regulations are currently national, but will likely be unified under ISO standards in the future. At the time of writing, ISO 4866, 'Mechanical vibration and shock - vibration of buildings - guidelines for the measurement of vibrations and evaluation of the effects on buildings' is about to be approved. (There is an equivalent US ANSI draft standard.) This does not set actual recommended vibration levels, nor does it indicate probability of damage. Instead, it addresses the methods for measurement and how analysis should be done. The European standards (which are given as guidelines) are peak particle velocity levels as a function of two factors: (i) the frequency of vibration; and (ii) the type of structure. In some cases these dependences are explicit; in others they are implicit. It should be noted that the determination of frequency and structure type is not objective, and the cautions mentioned in the previous section should be recalled. The German standards, from DIN 4150, are summarized in Table 2. The French standards (87/70558) are similar (see Table 3). The Swedish standards (SS 460 48 66) (see Table 4) are based upon: (i) peak levels based upon ground conditions; (ii) type of structure; (iii) distance; and (iv) type of blasting (long or short duration). The frequency is implicitly contained in the dependence of the allowable peak levels on ground conditions and the distance. Allowable vibration levels are lower at greater distance, apparently because lower frequencies predominate at greater distance. The ground conditions influence the frequency, with unconsolidated materials giving rise to lower frequency vibrations. These values are then multiplied by factors considering the type of structure, distance and type of blasting. 4.4.3 Complaints versus Compliance Neighbors often insist that 'the blast is shaking my house to the ground', when vibration criteria indicate that this could not be the case. Response spectra may indicate the cause of the problem. Table 2 German Standards Structure type Commercial Residential Sensitive
<10Hz 20 5 3
Table 3\ Structure type Resistant Sensitive Very sensitive
4-8 Hz 8 6 4
Peak particle velocity (mm s *) 10-50 Hz 50-100 Hz 20^0 5-15 3-8
40-50 15-20 8-10
French Standards Peak particle velocity (mms - 1 ) 8-30 Hz 30-100 Hz 12 9 6
15 12 9
Table 4 Swedish Standards Subsoil Unconsolidated strata of moraine sand, gravel, clay Consolidated strata of moraine slate, soft limestone Granite, gneiss, hard limestone, quartzitic sandstone, diabase
Vibration (mm s *) 18 35 70
Blast Monitoring: Regulations, Methods and Control Techniques
105
Vibration may have a strong component at a frequency which could possibly be amplified by the structure itself, or rattle windows, shake dishes on shelves, etc. A neighbor will often assume that such rattling and shaking is evidence that the dwelling is being damaged; it is difficult to counter this kind of 'evidence' with proof of compliance with regulations. The neighbor will likely question the credibility of the regulations, the vibration recordings or both. The fundamental problem, then, is with complaints and not compliance. Compliance does not become an issue unless there are complaints; on the other hand, if there are complaints, compliance may not be sufficient.
4.5 CONTROL Control requires understanding the nature of the problem, and then using the appropriate tools to correct the problem. Analysis techniques discussed earlier provide data for determining the problem. We will now discuss methods to control the problem.
4.5.1 Ground Vibration Until recently, since regulations have been based upon peak particle velocity (a scalar), methods to control vibration have also been based upon a scalar, 'scaled distance'. Understanding of the problems of vibration have indicated that more sophistication is necessary in controlling vibration. Effective reduction of vibration is in fact not the reduction of a single peak, but modification of the entire waveform so as to reduce structural response potential. We will review the traditional approach, indicate why it is lacking and follow it with a more appropriate approach.
4.5.1.1 Scaled distance The simplest method used to predict peak ground vibration levels from blasting is scaled distance - the distance from a blast divided by the square root of the pounds per delay. Pounds per delay is defined as the number of pounds of explosive detonated within an 8 ms interval. Based upon this definition, regression plots of peak particle velocity versus scaled distance have been used to determine likely values of peak particle velocity. There are several problems with this approach. Peak particle velocity versus scaled distance plots show an enormous amount of scatter. Typical plots show a variation in peak particle velocities at a given scaled distance of a factor often or more. This is not surprising, since these plots are compiled from different sites with varying geology and blast designs. The scaled distance approach does not deal with the problem of the frequency content at all. Lowering the charge level may reduce peak levels, but will not alter either the delay times or geology, which contribute to the frequency. Unfortunately, it is also often forgotten that too little explosive can increase vibration if fragmentation is ineffective. If a blast has too much confinement, the rock does not break efficiently. This inefficiency is translated into wasted energy transmitted as ground vibration. Inaccuracy in firing times may also contribute to excess confinement.
4.5.1.2 Vibration control using delays In Bureau of Mines Bulletin 442, Thoenen and Windes [11] showed that delay times could be chosen to cancel vibration waves from a test production shot. However, they concluded that: 'It is difficult to decrease vibration by delay blasting because often the vibration has no regular frequency, elimination of one component frequency does not appreciably affect the resultant amplitude, and the timing method must be quiteflexible,yet accurate. It is concluded, therefore, that delay shooting of the type described in this paper is not practical for reducing vibration in commercial practice.' At the time of this study (1942) this was not practical because neither the appropriate data recording nor processing techniques were available. In addition, inaccuracies in delays precluded the control necessary.
106
Blasting
In 1983, Anderson, et al [12] described a method for choosing delays to decrease ground vibration. This method is site-specific, and applicable to complex vibration. It relies on three critical assumptions as follows. (i) At a given location, the vibration from each explosive charge in a blast will be very similar to that from other charges of approximately equal size. This characteristic vibration we call the signature. (ii) The signature is determined primarily by the geological characteristics of the path between the blast and the seismometer location. (iii) The signature from a single-charge blast contains all of the information needed for predicting vibration: one does not need to know how the geology influences the vibration signature; all one needs to know is what that signature is. These assumptions were tested and verified by Anderson et al. [13] and Reil et al [14]. Chironis [15] used an analogy to describe the method: 'Everybody knows how to give someone a nice ride on a swing. The pushes (pulses) to the swing must be timed precisely with the natural frequency and position of the swing so that each push adds to the swing's momentum. If the pushes are not synchronized properly, or if the period between pushes is too short or too long, the swing begins to lose its momentum and, in fact, can be brought almost to a stop.' What we wish to do is to 'stop the swing'. The key to the method is single-charge blast. A seismogram generated at a particular site by this blast is called the signature. The signature is determined by the properties of the rock and soil in the path between the blast and the receiver site. Detailed knowledge of the geological causes of the vibration is not necessary. Each hole in a blast in that general area will produce the same signature. The vibration from a production shot is then a summation of signatures. Returning to the swing analogy, the signatures must be added with delay times so that the pushes to the swing teqd to stop it rather than make it go higher. To choose appropriate delays, then, a series of synthetic (computer-generated) seismograms is generated from the signature. Each synthetic seismogram is created by adding a number of signatures (corresponding, for example, to the number of holes in a row) delayed by a given time interval. This creates a seismogram which is that expected from a shot with that number of holes in a row and that delay interval. This is typically done for delays from 0 (all charges firing at once), in increments of 4 ms up to 200 ms. Frequency spectra are determined for each of these seismograms. A plot of these spectra is generated, a 'map' in frequency-delay space, with amplitude proportional to the darkness on the plot (see Figure 2). This map is somewhat like a topographic map, with the high elevations the darkest. The task is to find the best east-west route (constant delay) with the flattest and lowest elevation (or vibration level). This method was tested withfiveproduction shots at each of two sites, using times chosen to give both good and bad vibration. (So that firing time scatter was not a problem in this study, the delays were achieved using seismic (instant) detonators and a programmable sequential blasting machine. This is obviously not a recommended standard operating procedure.) A shot with bad vibration would be one that produces a large amount of energy at frequencies which match the natural frequency of residential structures (i.e. 3-18 Hz). When this occurs, maximum energy is transmitted into the structure and the potential for complaints and possible damage is increased. In the test studies, the measured and predicted spectra were virtually identical. The good delay sequence (out-of-phase) drastically reduced the low frequency vibration peak. Such agreement and effectiveness were found at all the locations. Other workers (Crenwelge and Peterson [16], Hinzen et al. [17] and Blair [18]) have confirmed that delays can be used to reduce vibration. It must be noted that the technique is site specific. Each operation needs to have the seismic signatures determined for the neighbor locations of concern. In addition, delay times which are appropriate at one neighbor location may be bad for another. In such a case, the delays are chosen such that they are the best possible for all the locations. Also, the delay times are strongly dependent upon blast design. As an example, for a simple sine wave, optimum cancellation would be at 1/2 the period of the wave for two holes, while it would be 1/3 or 2/3 the period for three holes. Finally, the delays should be accurate for the vibration to be reproducible and for the best control. Anderson and Reil [19] have shown that predictability and control of ground vibration is substantially enhanced by accuracy in the initiator firing times. A 2 ms standard deviation in firing times produces acceptable results, comparable to those obtained with exact firing times. Vibration control is still effective with a standard deviation of 5 ms, although resonant peaks will sometimes appear. Standard deviations of 10 or 15 ms render both prediction and control almost useless. The
Blast Monitoring: Regulations, Methods and Control Techniques
107
Figure 2 Delay versus frequency plot for choosing best delays to control vibration (see text for discussion)
best results should be obtained with electronic initiators. If accurate initiators are used, they should be used along with the appropriate tools to avoid constructive interference at the resonant frequencies. Measurement of the accuracy of initiators will be covered later. 4.5.2 Airblast Control of airblast normally amounts to observing proper blasting practice. As of now there are no reliable methods for controlling airblast like those described for ground vibration. Normal face movement, which contributes to primary airblast, is in general unavoidable. Casting (used in some coal operations) is accomplished by short delays between holes in a row and long delays between rows. This may create strong airblast. If primary airblast is a problem the amount of cast should be reduced by modifying the delays. Secondary airblast is diagnostic of a problem with the blast. Venting (which shows up as a series of peaks on an airblast record) can be avoided by assuring that there is adequate stemming of the holes, and that there is adequate relief (delay time) between rows. Inaccuracy infiringtimes can also cause venting, and is only cured by obtaining a more accurate initiator system. Blowouts from the face or bench surface may be caused by incompetent material which is inadequately confined (such as a clay seam), inadequate burden of rock in front of or above the explosive, or excessive explosive in a part of the borehole due to caving. In either case laser profiling equipment is now available to determine burdens on a face. The best way to determine if there is incompetent material or caving is from driller information. Caved portions of the hole should be cased, and incompetent material should be decked.
108
Blasting
If there is a recurring airblast problem, weather should be checked to determine wind and if there is. a temperature inversion. One final precaution is that if holes across a face detonate at a rate greater than the velocity of sound, the face movement from these holes adds up, and a greater than normal airblast will result. In this case, the delays should be increased so that there is relief greater than 0.33 m ms ~λ (1.1 ft ms "* ) between holes in a row. 4.6 PRODUCTIVITY 4.6.1 Determination of Firing Times Proper performance of a blast requires that holes fire as designed. Vibration from a blast can indicate when holesfire.Examination of a simple waveform can indicatefiringtimes from the peaks. More complex waveforms require a more sophisticated approach. In 1978, Winzer [20] showed that commercially available initiators had substantial scatter, and that such scatter was detrimental to blasting performance. The release of that paper and an accompanying film documenting the scatter caused quite a stir. Subsequently, the US Bureau of Mines released a Report of Investigations [21] confirming the general conclusions of Winzer. Since the Winzer paper, most manufacturers have introduced initiators which are advertised as being substantially more accurate than the standard initiators available in 1978. The best pyrotechnic initiators currently available are said to have a scatter of about 2 ms one standard deviation. Electronic initiators have been announced [17], and are said to have a scatter of typically 0.4 ms, which is a factor of 10 better than existing pyrotechnic initiators. At the 99% confidence level, these initiators would then fire within 1.2 ms of the mean. Other manufacturers have indicated that electronic initiators are in the offing. One way of determining firing times is to use high-speed cameras, with tell tales of shock tubing extending from the boreholes. Normally this will give (for 500 frame s "* cameras) an accuracy of ± 2 ms, and this is sufficient for most purposes. However, it is expensive to film every shot. An alternative is to use a seismograph to determine the firing times. With sufficiently high sampling rate (on the order of 5000 samples per second), thefiringtimes can be routinely determined quite accurately. Several instrumentation manufacturers offer such equipment. Either acceleration or velocity measurements may be used to determine the firing times. The simplest records will be obtained close to the shot (of the order of 10 m). Adequate coupling of the transducer is essential. In addition, explosive efficiency may be gathered using this method, as covered in the next section.
4.6.2 Relative Fragmentation and Displacement Conventional wisdom has it that the worst fragmentation is observed for the opening and final holes on a blast. Work on the delay effect on fragmentation (Fourney, Chapter 2 and Mackenzie, Chapter 3, this volume) show that there is some interaction between the action of individual boreholes in a multiple-hole blast which improves fragmentation, though the basic mechanisms for this interaction remain uncertain. In some way the precracking of the rock by a borehole allows the strain wave and gas pressure for succeeding holes to fracture the rock more effectively. A single-hole shot has an unfavorable partition between the energy used for fragmentation and that wasted in ground vibration. Holes detonated subsequent to the first hole should then generate a ground vibration with lower amplitude than the single-hole shot. A synthetic seismogram generated by the linear superposition of single-hole waveforms will then predict a vibration somewhat higher than that actually observed for a multiple-hole blast. The efficiency of a blast cannot be determined, for different blast designs, simply by comparing the ground vibration peaks or spectra. Changing the delay times will change the observed vibration due to constructive or destructive interference, irrespective of the efficiency of the blast. Therefore, the appropriate measures need to be made for the vibration predicted by summing single-charge waveforms, compared with the observed. Anderson and Reil [19] have shown that there is a relationship between the efficiency of a blast, in terms of a screen size determined by mathematically analyzing face fragmentation, and the resultant far-field ground vibration as compared to the predicted. The data are limited, but the approach appears to have promise. An effective blast should waste less energy in ground vibration than an ineffective one.
Blast Monitoring: Regulations, Methods and Control Techniques
109
Close in to the shot the seismic records are simpler. Rise times and integrated energy of the vibration pulses will give information about the explosive efficiency. This is complicated, however, by the attenuation of the rock mass, which needs to be taken into account. Work is still progressing on this front. 4.7 CONCLUSIONS Blast vibration monitoring was originally developed to protect structures near blasting from damage. Early monitoring focused on peak levels, which gave a general indication of the levels of vibration. Research on structure response has indicated that peak levels are insufficient for adequately determining damage potential. Ground vibration is a function of charge weight in a hole, delay timing and the travel path. The waves from a production shot are approximately determined by linear superposition of the wave from a single-hole shot. This fact can be used to determine appropriate firing times to reduce vibration by cancellation of waves. Airblast is due to normal rock movement and to abnormal movement due to blowouts from the face and stemming ejection. Wind and temperature inversions also affect airblast potential. Proper blasting practice is the most effective cure for airblast problems. Analysis techniques for airblast and ground vibration have been mostly confined to measurement of peak levels, with the addition of a dominant frequency for ground vibration. Regulations have been based upon these analysis techniques. More sophisticated spectral analysis, with emphasis on response spectral analysis, promises to achieve both a better understanding of problems and more meaningful regulations. Additional techniques for monitoring blasting for determination of firing times and productivity are currently under development. These techniques should allow blasters to be more productive while reducing the potential for damage or complaints from neighbors. 4.8 APPENDIX - RESPONSE SPECTRUM CALCULATIONS We take the input to be the particle velocity as a function of time, û(t). To determine the relative displacement y(t) as a function of time, at constant undamped natural frequency cou and fraction of critical damping /?, we use the following equation [1] y(t) = -
Jo
ù(T)exp[-0û> u (t-T)] cos[a>d(f - τ)]
L
= =
2
Jl-ß
sin[û>d(t - τ)] \άτ
J
(1)
2
where cod, the damped natural frequency is equal to (ouy/l — ß . The maximum value of the displacement occurs at fm, and \y(tm,(o9ß)\ is the displacement response spectrum Sd. For typical residential structures, as shown by Medearis [6], the appropriate damping is 5% for horizontal components and 20% for vertical components. The pseudo-velocity spectrum, 5 pv is defined as 5pv = couSd
(2)
The significance of Spv is as follows: the maximum displacement corresponds to a condition of zero kinetic energy and maximum strain energy 0.5kSj. If this energy were in the form of kinetic energy, 0.5m(y)2 = 0.5kSj (where k is stiffness and m is mass), then the maximum relative velocity would be y(to) =
-Sd \] m
= couSd = Spy
(3)
This function, y(co), is the response spectrum. ACKNOWLEDGEMENTS I thank the late G. Alan Foster of Vibra-Tech Engineers for providing me with information on European regulations and current ISO and ANSI standards and I dedicate this chapter to his memory.
Blasting
110 4.9
REFERENCES
1. Dowding C. H. Blast Vibration Monitoring and Control, p. 297. Prentice-Hall, Englewood Cliffs, NJ (1985). 2. Bollinger G. A. Blast Vibration Analysis, p. 132. Southern Illinois University Press, Carbondale, IL (1971). 3. Siskind D. E. and Stagg M. S. Blast vibration measurements near and on structure foundations. U.S. Bur. Mines RI 8969, Pittsburgh, PA, p. 20. (1985). 4. Press W. H., Flannery B. P., Teukolsky S. A. and Vetterling W. T. p. 735. Numerical Recipes in C: The Art of Scientific Computing. Cambridge University Press, Cambridge (1985). 5. Hudson D. E. Response spectrum techniques in engineering seismology. In Proc. World Conf. Earthquake Engineering, Berkeley, CA, pp. [4]1-[4]12 (1956). 6. Medearis K. The development of rational damage criteria for low-rise structures subjected to blasting vibrations. In Proc. 18th U.S. Symp Rock Mech., Keystone, CO (Edited by F.-D. Wang and G. B. Clark), pp. 1-16. (1977). 7. Medearis K. Dynamic characteristics of ground motions due to blasting. Bull. Seismol. Soc. Am. 69, 627-639 (1979). 8. Siskind D. E., Stagg M. S., Kopp J. W. and Dowding C. H. Structure response and damage produced by ground vibration from surface mine blasting. U.S. Bur. Mines RI 8507, p. 74. (1980). 9. Dowding C. H., Fulthorpe C. and Langan R. T. Simultaneous airblast and ground motion response. J. Struct. Div. Am. Soc. Civ. Eng. 108, 2363-2378 (1982). 10. Siskind D. E., Stachura V. J., Stagg M. S. and Kopp J. W. Structure response and damage produced by airblast from surface mining. U.S. Bur. Mines RI 8485, p. 74. (1980). 11. Thoenen J. R. and Windes S. L. Seismic effects of quarry blasting. Bull. - U.S. Bur. Mines 442, 83 (1942). 12. Anderson D. A., Winzer S. R. and Ritter A. P. Synthetic delay versus frequency plots for predicting ground vibration from blasting. In Proc. 3rd Int. Symp. Computer Aided Seismic Analysis and Discrimination, Washington, DC, pp. 70-74. (1983). 13. Anderson D. A., Ritter A. P., Winzer S. R. and Reil J. W. A method for site specific prediction and control of ground vibration from blasting. In Proc. 1st Mini-Symp. Explosives and Blasting Research, San Diego, CA, pp. 28-43. (1985). 14. Reil J. W., Anderson D. A., Ritter A. P., Clark D. A., Winzer S. R. and Petro A. J. Geologic factors affecting vibration from surface mine blasting (Contract H0222009) BuMines OFR 33-86; NTIS PB86-175858, p. 191. (1985). 15. Chironis N. P. Accurate timed delays improve vibration control and breakage. Coal Age, January, 62-64 (1986). 16. Crenwelge O. E., Jr. and Peterson T. A. Overburden blasting vibrations: analysis, prediction, and control. In Proc. 12th Conf. Explosives and Blasting Technique, Atlanta, GA, pp. 269-283. (1986). 17. Hinzen K.-G., Lüdeling R., Heinemeyer F., Röh P. and Steiner U. A new approach to predict and reduce blast vibration by modeling of seismograms and using a new electronic initiation system. In Proc. 13th Conf. Explosives and Blasting Technique, Miami, FL, pp. 144-161. (1987). 18. Blair D. P. The measurement, modeling and control of ground vibrations due to blasting. In Proc. 2nd Int. Symp. Rock Fragmentation by Blasting, Keystone, CO (Edited by W. L. Fourney and R. D. Dick), pp. 88-101. Society for Experimental Mechanics, Bethel, CT (1987). 19. Anderson D. A. and Reil J. W. Effect of cap scatter on blast-induced ground motion. J. Explosives Eng. 5, 22-28 (1987). 20. Winzer S. R. The firing times of ms delay blasting caps and their effect on blasting performance. Report to the National Science Foundation, Contract DAR-77-05171. (1978). 21. Bajpayee T. S., Mainiero R. J. and Hay J. E. Overlap probability for short period delay detonators used in underground coal mining. U.S. Bur. Mines RI 8888, p. 22. (1985).
5 Blast Vibration Monitoring for Rock Engineering CHARLES H. DOWDING Northwestern University, Evanston, IL, USA 5.1
INTRODUCTION
5.2
RANGE OF BLAST EFFECTS
112 112
5.2.1 Permanent Degradation and Displacement of Adjacent Rock 5.2.1.1 Degradation 5.2.1.2 Displacement 5.2.1.3 Fly rock 5.2.1.4 Soil densification 5.2.2 Structural Response to Transient Displacement 5.2.3 Blast-induced Air Overpressures 5.2.4 Human Response 5.3 CHARACTER OF BLAST EXCITATION AND STRUCTURAL RESPONSE 5.3.1 Ground Motion 5.3.2 Sinusoidal Approximation 5.3.3 Kinematic Relationships of Ground Motion 5.3.4 Transient Nature of Blast Motions 5.3.5 Estimation of Dominant Frequency 5.3.6 Propagation Effects 5.3.7 Blast-induced Air Overpressures 5.3.7.1 Propagation effects 5.4
MEASUREMENT INSTRUMENTS AND THEIR DEPLOYMENT
5.4.1 5.4.2 5.4.3 5.4.4 5.4.5 5.4.6 5.4.7 5.4.8 5.5
Structural Strains versus Particle Velocity Appropriate Measurement of Particle Velocity Transducers Transducer Attachment Digital, Tape and Hard-copy Recorders Calibration Number of Instruments Instrument Deployment during Test Blasts
113 113 113 113 113 113 114 114 114 115 116 116 117 117 118 119 119 120 121 121 121 122 122 123 123 124
STRUCTURAL RESPONSE
124
5.5.1 Structural Response and Frequency Effects 5.5.2 Origin of the SDF Model 5.5.3 Estimation of Dynamic Response Properties 5.5.4 Response Spectrum 5.5.5 Fourier Spectra do not Directly Predict Response 5.5.6 Case Histories Demonstrate Importance of Response Spectrum Analysis 5.5.6.1 Surface mine blast 5.5.6.2 Urban construction blast 5.6 CONTROLLING BLAST EFFECTS 5.6.1 Regulation to Prevent Cosmetic Cracking of Residential Structures 5.6.2 Statistical Analysis of Data with Pre- and Post-blast Inspection 5.6.3 Distinction of Blast-induced Cracking from Natural Cracking 5.6.4 Multiple Origins of Cracks 5.6.5 Response of Structures to Everyday Activities 5.6.6 Comparison of Blast and Environmental Effects 5.6.7 Special Considerations
in
124 125 125 126 111 127 111 128 129 129 129 130 131 131 132 132
112
Blasting
5.6.7.1 Engineered structures 5.6.7.2 Restrained structures 5.6.8 Fatigue or Repeated Events 5.6.9 Rock Mass Cracking and Displacement 5.6.10 Frequency-based Control with Dominant Frequency 5.6.11 Regulatory Compliance for Air Overpressures 5.7
REFERENCES
132 132 133 133 133 133 134
5.1 INTRODUCTION Monitoring and control of blast effects near critical rock masses or constructed facilities depend upon two main considerations. First, shot designs must reduce the amount of explosives detonated at any instant and adjust the initiation sequence to reduce resulting ground and airborne disturbances. Secondly, the amount of explosives detonated per volume of rock and the shot pattern must be adjusted to ensure adequate fragmentation. Therefore, the initiation sequence must be separated in time but not in space. There is an optimum design which achieves both objectives of control of disturbances and production of adequate fragmentation. This optimum can be reached only through an understanding of the physics of rock mass and structural response to blast disturbance, and the interaction between rock fragmentation and shot design. This chapter summarizes the state-of-the-art in vibration measurement and structural response, to facilitate such an optimum blast design. Much of this chapter can also be found in Monitoring and Control of Blast Effects [1]. Chapter 3 in this volume summarizes the state-of-the-art of blast design. Advances in earthquake engineering and nuclear blast protective design are transferred to blast vibration monitoring and control, while recent experimental observations of mining-induced ground motions and structural response are summarized. It is hoped that such a transfer and summary of the state-of-the-art will mitigate several recent trends. For one, there has been a general downward trend in regulatory limits on allowable blast-induced vibrations. In addition to new observations, this drift, in part, can be attributed to the tendency to take the limit of the last study and divide it in half 'to be safe'. Unfortunately, too many studies whose limits were divided were themselves only summaries of past work that had also divided past limits. The discussion here presents the background for original experiments conducted to determine safe blasting controls and therefore will allow the reader to set appropriate limits based upon the original past work within the framework of existing regulations. Another trend is the misapplication of peak particle velocity limits that were determined for cosmetic cracking of residential structures. These limits have been applied to tunnel liners, radio towers, slabs-on-grade, and curing concrete. This chapter draws attention to studies made to determine limits specifically for these and other cases. Where no studies exist, it presents methods based upon response spectra or ground strains that allow setting of appropriate criteria or limits. Frequency of vibration and ground strain form the foundation for this presentation; The importance of frequency cannot be over-estimated, as it is as critical as peak particle velocity in determining the response of above-ground structures. For below-ground structures, frequency, in combination with propagation velocity, controls response. In both cases, cracking results from induced strains, where particle velocity is employed as an index of the strain level. In addition, computerized and self-triggering monitoring instrumentation is described. Such computerization simultaneously increases monitoring efficiency and decreases costs, both original capital costs as well as those associated with record keeping. The latter labor-saving efficiency associated with automated record keeping continues to be undervalued in many mining and construction operations.
5.2 RANGE OF BLAST EFFECTS Blast effects on surrounding earth materials and structures can be divided into permanent and transient displacements. While the focus of this chapter is the transient displacements, effects of permanent displacements are presented as they are associated with significant transient effects at relatively short distances.
Blast Vibration Monitoring for Rock Engineering 5.2.1
113
Permanent Degradation and Displacement of Adjacent Rock
Permanent effects, with the exception of fly rock, are only found within a few hundred meters of the explosion, and can be divided into degradation and displacement.
5.2.1.1
Degradation
Degradation is normally described by cracking intensity. Such blast-induced cracking has been observed experimentally to vary with hole diameter and rock type [2, 3]. Small-hole-diameter construction blasting has induced cracking at distances of 1 to 2 m, and larger-hole-diameter mining blasts are capable of producing cracks at distances of 10 to 20 m. Careful blast design can dramatically reduce these maximum distances.
5.2.1.2
Displacement
Displacement can be produced by either delayed gas pressures (those that accumulate during detonation) or by vibration-induced shaking. Delayed gas pressures have dislocated blocks as large as 1000 m 3 during construction blasting [4]. Such movement is unusual but is associated with isolated blocks, leakage of gas pressures along open joints, and poor shot design with large burdens. Vibratory or shaking-induced displacement is normally associated with unstable blocks in rock slopes, and can occur wherever static factors of safety are low and ground motions produce permanent displacements that are greater than the first-order asperity wavelength of rock joints [5]. Gas pressure related displacement can occur out to tens of meters. 5.2.1.3
Fly rock
Fly rock is a special case of permanent displacement of rock by explosive expulsion from the top of the blast hole. Rock has been observed to have been propelled as far as 100 to 1000 m [6]. Statistical studies have shown that the probability of these extreme events is quite low under normal circumstances, 1 in 10000000 at 600 m [7]. Since the probability increases with decreasing distance, blasting mats to prevent all fly rock are required in any construction blasting in an urban environment. 5.2.1.4
Soil densification
Another special case of permanent displacement is the vibratory densification of a nearby mass of loose, clean sand. The propensity for such densification is a function of the soil's density, mineralogy, and grain-size distribution. Soils that can be densified are loose sands, with less than 5% silt-size particles. These clean sands were densified out to distances of 20 m [8] after detonation of single, 5 kg charges within the loose sand mass itself. Soils that are either slightly cemented or contain more than 5% fines are a great deal less subject to vibratory densification from typical ground motions.
5.2.2
Structural Response to Transient Displacement
Transient effects result from the vibratory nature of the ground and airborne disturbances that propagate outward from a blast. In this discussion, it is assumed that no permanent displacements are produced. Thus the only effects are those associated with the vibratory response of facilities in or on the rock or soil mass surrounding the blast. Transient means that the peak displacement is only temporary, lasts less than one-hundredth of a second, and the structure returns to its original position afterwards. Transient structural effects can be arranged to reflect the expected distance from a blast. Beginning with the closest, transient effects are structural distortion, faulted or displaced cracks, falling objects, cosmetic cracking of wall coverings, excessive instrument and machinery response, human response and micro-disturbance. The first four effects, those that relate to structural response, are normally grouped together for experimental observation, and do not normally occur when vibration levels are regulated to prevent cosmetic cracking.
114
Blasting
Excessive structural response has been separated into three categories arranged below in the order of declining severity and increasing distance from the occurrence [9,10]. Beginning with effects that occur closest to the blast the categories are listed here. (i) Major (Permanent distortion) Resulting in serious weakening of the structure {e.g. large cracks or shifting of foundations or bearing walls, major settlement resulting in distortion or weakening of the superstructure, walls out of plumb). (ii) Minor (Displaced cracks) Surficial, not affecting the strength of the structure (e.g. broken windows, loosened or fallen plaster), hairline cracks in masonry. (Hi) Threshold (Cosmetic cracking) Opening of old cracks and formation of new plaster cracks, dislodging of loose objects (e.g. loose bricks in chimneys). These specific definitions of response should not be described collectively as 'damage'. To do so blurs the distinction between threshold or cosmetic cracking and major response or structural distress. 5.2.3 Blast-induced Air Overpressures Blast-induced air overpressures are the air pressure waves generated by explosions. The higherfrequency portion of the pressure wave is audible and is the sound that accompanies a blast; the lower-frequency portion is not audible but excites structures and in turn causes a secondary and audible rattle within a structure. Overpressure waves are of interest for three reasons. First, the audible portion produces direct noise. Secondly, the inaudible portion by itself, or in combination with ground motion, can produce structural motions that in turn produce noise. Thirdly, they may crack windows; however, airblast pressures alone would have to be unusually high for such cracking. Previous researchers [11, 12] have found that response noise within a structure (from blasting and sonic booms, respectively) is the source of many complaints. It appears that structure and wall motions, which are vibrationally induced by air blasts and sonic booms, rattle loose objects within the structure, which then startle the occupants. 5.2.4 Human Response Humans are quite sensitive to motion and noise that accompany blast-induced ground and airborne disturbances. Therefore human response is significant in the reporting of blast-induced cracking. Motion and noise from blasting can be startling and lead to a search for some physical manifestation of the startling phenomena. Many times a previously unnoticed crack provides such confirmation of the event. Furthermore, if a person is worried and observes a crack that was not noticed before, the crack's perceived significance increases over one noticed in the absence of any startling activity. In typical mining situations, significant blast-induced inaudible air overpressure and audible noise immediately follows the ground motion and intensifies human response. Both the ground and airborne disturbances excite walls, rattle dishes, and together tend to produce more noise inside a structure than outside. Thus both the audible noise as well as the wall rattle produced by inaudible pressures contribute to human response. To complicate matters even more, inaudible air overpressures can vibrate walls to produce audible noise at long distances, which are inaccurately reported by occupants as ground motions. 5.3 CHARACTER OF BLAST EXCITATION AND STRUCTURAL RESPONSE As shown in Figure 1, both the ground and air-borne disturbances (Figure la) produce structure response (Figure lb). Because of the importance of frequency, the full waveform or time history
115
Blast Vibration Monitoring for Rock Engineering Maximum ground motion
(a)
L
«Wyy[/vvA/V\/\A
T
—"^AA/^VSAA^
Air (b)
Maximum superstructure response
',
tf2(T)
p/S P/S
^ R
1" '"
A ÂR
*»■ ■ * î ^ v v v » ^ ^ * v
IWv
*^
0.12e iPs
oi63
^*>/^VN/Wn
^ { l ^ ^ l ^ ^
109 dB 0.322 ips
'^l^yUVV^^
* w r λ* ν \ / ^ Λ / \ / \
Ti ps )
velocity
0.106
°·
/ V / V ^ ^ V V /
Mid waM
286
0.361
Ex t t lon
0 364
(15-25 Hz) Response velocity
0
Super structure
7
(6-7Hz)
^=Ι/Λ Time
Figure 1 Comparison of (a) blast excitation by ground and air-borne disturbances and (b) residential structure response of walls and superstructure. Measurements were made some 2000 ft (600 m) from a typical surface coal-mining blast (after Dowding [27])
should be recorded. When a critical location is known, blast response is best described by the strain at that location. Alternatively, particle velocity (that shown in Figure 1) can be measured outside the structure of concern, as many recent cracking studies have correlated cracking with excitation particle velocity measured in the ground. 5.3.1 Ground Motion Ground motion can be described by three mutually perpendicular components labeled L (longitudinal), T (transverse) and V (vertical) (Figure 1). The L and T directions are oriented in the horizontal plane with L directed along the line between the blast and recording transducer. When a study focuses upon structural response, axes can be labeled HI, H2 and V, with HI and H2 oriented parallel to the structure's principal axes. Variation of peak motions in each component (L, V and T in Figure 1) has led to difficulty in determining which is more important. Horizontal motions seem to control the horizontal response of walls and superstructures, and vertical motions seem to control the vertical response offloors.In an absolute sense, the peak ground motion is actually the maximum vector sum of the three components, which usually occurs at the largest peak of the three components, the dashed line in Figure 1. This TRUE maximum vector sum is not the FALSE maximum vector sum calculated with the maxima for each component (dots in Figure 1) no matter their time of occurrence. The FALSE maximum vector sum may be as much as 40% greater than the TRUE maximum vector sum, which is normally 5-10% greater than the maximum, single component peak. In general, experimental observations of threshold or cosmetic cracking, which form the basis of blasting controls in North America, have been correlated with the maximum single component regardless of direction. Therefore, use of the FALSE maximum vector sum for control provides a large, unaccounted for, factor of safety. Two wave types are produced by blasting, body (P/S) and surface (R) and are illustrated by the ground motion in Figure 1 measured some 600 m from a typical surface coal-mining blast. Body waves travel through earth materials, whereas surface waves travel along surfaces and interfaces of earth materials. The most important surface wave is the Rayleigh, denoted R on the vertical trace in Figure 1. Body waves can be further subdivided into compressive (compression/tension) or soundlike waves, and distortional or shear waves, denoted as P/S on the vertical trace in Figure 1. Explosions produce predominantly body waves at short distances. These body waves propagate outward in a spherical manner until they intersect a boundary such as another rock layer, soil or the
116
Blasting
ground surface. At this intersection, shear and surface waves are produced. Rayleigh surface waves become important at longer transmission distances as illustrated in the vertical trace by the relatively larger 'R' amplitude compared to the T/S' amplitude. 5.3.2 Sinusoidal Approximation Typical blast vibrations, no matter the wave type, can be approximated as sinusoidally varying in either time or distance along the radial or longitudinal line as shown by the time variations in Figures 2(a) and (b). This approximation is useful because it makes calculations for strain and acceleration from particle velocity much simpler than that for an irregular pulse. Ground motion from a blast is similar to the motion of a cork caused by a passing water wave. Displacement of the cork from its at-rest position is similar to the displacement, u, of a particle in the ground from its atrest position. Similarly, the cork's velocity, as it bobs up and down, ti, is analogous to that of a particle in the ground, hence the term particle velocity. The water wave that excites the cork can be described by its wavelength, λ, the distance between wave crests; the wave speed or propagation velocity, c, at which the wave travels past the cork; and the frequency,/ or the number of times the cork bobs up and down in one second. Frequency,/ is equal to I/ΓΟΓ the reciprocal of the period or time it takes the cork to complete one cycle of motion. Frequency is measured in cycles per second or Hertz, Hz. Propagation velocity, c, should not be confused with particle velocity, û; as c is the speed with which the water wave passes by the cork, and ù is the speed at which the cork moves up and down while the wave passes. Blast vibration waves also can be described by their wavelength, propagation velocity and frequency in the same fashion as the water wave. 5.3.3 Kinematic Relationships of Ground Motion The general form for the sinusoidal approximation is best understood by beginning with the equation for sinusoidal displacement, u u = Usin(2nft)
(1)
where U is maximum displacement,/is frequency, and t is time. The relationship between the maximum particle displacement, wmax, particle velocity, timax, and acceleration, timax, is also greatly simplified by the sinusoidal approximation and is found through differentiation of equation (1) with respect to time, as shown below, whenever the sin/cos function maximizes at 1 "max = U (max. displacement) "max =
U2nf
«max = U4n f
(max. particle velocity)
271/U max
= 2 2
(2)
(max. acceleration)
= 2nfùn
Usually, acceleration is normalized (divided) by gravitational acceleration, 9814 mm s 2. Therefore, an acceleration of 2000 mm s 2 is 2000 9814
or two-tenths that of gravity.
= 0.20
(a)
Ai\
(3)
(b)
\
A
^
1
Distance, x
u
s\f\
—
T
-|
\ Time, /
Figure 2 Sinusoidal approximations: (a) sinusoidal displacement at afixedpoint (x = constant); (b) sinusoidal displacement at one instant (t = constant) (Dowding [4])
Blast Vibration Monitoring for Rock Engineering
117
Kinematic relations between particle displacement, velocity and acceleration for complex waveforms are exactly related through integration or differentiation of any of the waveforms. For instance, an acceleration time history can be integrated once for a velocity time history, which in turn can be integrated for a displacement time history. Even though a particle velocity record.can be differentiated to find acceleration, it is not recommended, as the procedure is sensitive to small changes in the slope of the velocity time history. Further discussion of the inaccuracies of differentiation and integration can be found in Dowding [4] and in texts devoted to interpretation of time histories [e.g. ref. 13]. 5.3.4 Transient Nature of Blast Motions Great care should be taken not to confuse the effects of steady-state, single-frequency, harmonic motions with those of transient, irregular blast motions. Most vibration studies conducted by personnel trained in mechanical and electrical engineering and geophysics implicitly assume that the motions are continuous (last many cycles), and steady state (have constant frequency and amplitude). As can be seen in Figure 1, blast-induced motions last only one or two cycles at a relatively constant amplitude and frequency. Such conditions are not similar enough to steady-state motions to allow specific application of steady-state approximations such as resonance. 5.3.5 Estimation of Dominant Frequency Adoption of frequency-based vibration criteria has made the estimation and calculation of the dominant frequency an important concern. Dominant frequency can be estimated through visual 0.4
Surface coal mine blasting 0.3 Ùi
L £
n
03
hi f- principal frequency
Quarry blasting
L£L
/
Construction blasting
0 2 \—
r
1 40
" ^ 60
Τ^ Π 80
100
120
Principal frequency (Hz)
Figure 3 Dominant frequency histograms at nearest structures categorized by industry. Dominant frequency is defined in the inset (after Siskind et al [10])
118
Blasting
inspection of the time history or calculated with Fourier frequency spectra or, alternatively, response spectra. The accuracy or difficulty of visually estimating the dominant or principal frequency depends upon the complexity of the time history. The easiest type of time history record for frequency estimation is one with a single dominant pulse like that shown in the inset in Figure 3. This dominant frequency can be determined through the manual or visual measurement of the time of the two zero crossings on either side of the peak. The difference between these times is one-half of the period, which is the inverse of twice the frequency of the dominant peak as shown in the inset. As shown in Figure 3, the relatively large explosions produced by surface coal mining, when measured at typically distant structures, tend to produce vibrations with lower principal frequencies than those of construction blasts. Construction blasts involve smaller explosions, but the typically short distances between a structure and a blast, as well as rock-to-rock transmission paths, tend to produce the highest frequencies. Such high-frequency motions associated with construction blasts have less potential for cracking adjacent structures [4]. The most difficult type of record to interpret is that which contains nearly equal peaks at two dominant frequencies, such as that in Figure 1. The two dominant frequencies are the initial 15 to 20 Hz portion (peak A) and the later 5 to 10 Hz portion (peak B). As can be seen in the figure, the initial portion produces the highest wall response while the second produces the greatest superstructure response. For the best frequency correlation of both types of response, both frequencies should be calculated. The best computational approach to determining the dominant frequency involves the response spectrum. The response spectrum is preferred over the Fourier frequency spectrum because it can be related to structural displacement and thus strains [4]. A compromise approach is to calculate the dominant frequency associated with each peak by the zero-crossing approach described above. Since not many time histories contain as broad a range of dominant frequencies as that in Figure 1, most approaches require only the calculation of the frequency associated with the maximum particle velocity for blasts that produce low particle velocities. The more complex frequency analyses are employed only when peak particle velocities approach control limits. 5.3.6 Propagation Effects Ground motions always decrease with increasing distance. Effects of constructive and destructive interference and geology are included within the scatter of data about the mean trend of the decay in amplitude with distance. While this scatter is large, the associated decay with distance is observed in
Square root scaled distance (ft lb" 05 )
Figure 4 Attenuation relationships showing scatter from geological and blast design effects as well as high expected velocities from confined shots, such as presplitting (after Siskind et al. [10])
Blast Vibration Monitoring for Rock Engineering
119
all blast-vibration studies. Typical examples of this decay are shown in Figure 4 where maximum particle velocity is plotted as a function of square-root scaled distance from the blast. Square-root scaling, or plotting peak particle velocity as a function of the distance, R, divided by the square root of the charge weight, R/W°'5, is more traditional than the cube-root scaling, which incorporates energy considerations [14]. Both square- and cube-root scaling can be employed to compare field data and to predict the attenuation or decay of peak particle velocity; however, square-root scaling is more popular. Several square-root attenuation relationships employed in the United States are shown in Figure 4. They are banded to reflect scatter, which is typical of blasting operations. Curve P should be used for presplitting, cratering and beginning new bench levels. It is also the basis for the United States Office of Surface Mining (OSM) regulations for conservative shot design when monitoring instruments are not employed. Dominant frequencies also tend to decline with increasing distance and with increasing importance of surface waves. At longer distances typical for mining, higher-frequency body waves begin to have relatively lower-peak amplitudes than the lower-frequency surface waves, as shown in Figure 1. Since lower frequencies can elicit greater structural response [15] (as shown in Figure 10), OSM scaled-distance limits decline with increasing absolute distance.
5.3.7
Blast-induced Air Overpressures
Just as with ground motions, blast-induced air overpressure waves can be described with time histories as shown in Figure 1. The higher-frequency portion of the pressure wave is audible sound. While the lower-frequency portion is not audible, it excites structures, which in turn causes a secondary and audible rattle within the structure. The air-blast excitation of the walls is shown by comparing the last 1/4 of the time histories of air blast and wall response in Figure 1. Unlike ground motions, air overpressures can be described completely with only one transducer, since at any one point air pressure is equal in all three orthogonal directions.
5.3.7.1
Propagation effects
Propagation of blast-induced air overpressures has been studied by numerous investigators and is generally reported with cube-root rather than square-root scaled distances. Peak pressures are reported in terms of decibels, which are defined as dB = 2 0 1 o g 1 0 ( ^ )
(4)
where P is the measured peak sound pressure and P0 is a reference pressure of 2.9 x 10 " 9 psi ( 2 0 x l 0 - 6 Pa). Figure 5 summarizes the effect of two important instrumentation and shot variables. First, the effect of the weighting scales is dramatically evident. C-weighting greatly reduces the recorded peak pressure at any scaled distance. This does not mean the peak is reduced by changing instruments, but rather that the C-weighting system does not respond to the low-frequency pressure pulses. These low-frequency pressure peaks excite structures and occupants whether or not they are sensed by the measurement instruments. The other (5 and 0.1 Hz) labels denote the lower-frequency bounds of the recording capabilities of these so called 'linear' systems. Secondly, the effect of gas venting caused by inadequate stemming in shot holes can be observed in Figure 5 from the higher average pressures produced by the parting shots at any scaled distance. Parting shots are detonated in thin rock layers between coal strata in surface mines. Consequently, there is less hole height available for stemming, and these shots eject the stemming many times and thereby produce abnormally high air overpressures. The unconfined relationship should be used for demolition of structures after modification for effects of weather and ground reflection. Various effects of the wind have been reported and should be added to the average relations presented in Figure 5. Wiss and Linehan's study [16] of air overpressures produced by surface coal mining showed that in moderate winds the typical 7.7 dB reduction for each doubling of distance is reduced by 7.7 - 1.6Fmohcos0dB
(5)
Blasting
120 ιον
10"'
131
α> V.
o 101
10"
91
— — . . . . . . —·— ·— 10«
'
10
'
I
Highwall parting Unconfined
l l i il I
_l
100
I
81
1 I I I I II
_l
1 I I 1 I I I
1000
10000
Cube root scaled distance (ft lb~ l / 3 )
Figure 5 Attenuation relationships for air overpressures produced by confined (highwall) and partially confined (parting) surface coal-mining blasts as well as unconfined blasts (Siskind et al. [34])
where Vmph is wind velocity in miles per hour and Θ is the angle between the line connecting the blast and transducer and the wind direction. An air temperature inversion causes the sound pressure wave to be refracted back to the ground and at times to be amplified at small, 16-acre-sized locations. Such an inversion occurs when the normal decrease in temperature with altitude is reversed because of the presence of a warmer upper layer. Schomer [17] has shown that for propagation distances of 3-60 km, inversions produce zones of intensification of up to three times the average, attenuated or low air overpressures at those distances, with an average of 1.8 times (5.1 dB) increase. At distances less than 3 km, where high air overpressures are likely to occur, his measurements show no inversion effects.
5.4
MEASUREMENT INSTRUMENTS AND THEIR DEPLOYMENT
This section describes characteristics of instruments that measure the ground motions (acceleration, velocity, displacement) and air blast (air overpressure). Since there are many excellent sources for information on instruments, the principal characteristics of available systems will be summarized
4
fitifl
(TO oooo 3
j
1 2 3 4 5
/Ύ^-· Y y-—)2
1
Velocity (3 orthogonal) and sound pressure transducers Cables Amplifier Recorder (tape, disk,or memory) Light beam oscilloscope or dot matrix printer
Figure 6 Idealized, field-portable, blast-monitoring system that shows the schematic relationship of the five principal components (Dowding [4])
Blast Vibration Monitoring for Rock Engineering
121
rather than exhaustively reviewed. The most complete single reference for detailed instrumentation information that is updated periodically is the Shock and Vibration Handbook [18]. Specific information on blast vibration monitors is contained in various government publications [i.e. ref. 19 in the United States]. An idealized, field-portable blast-monitoring system operating on a 12 V battery is illustrated in Figure 6. It consists of transducers (1) that convert physical motion or pressure to an electrical current, which is transmitted through cables (2) to an amplifying system (3) and (4), a magnetic tape, paper or computerized digital recorder, which preserves the relative time variation of the original signal for eventual permanent, hard-copy reproduction by a pen or light-beam galvanometric recorder or computer printer (5). As one can imagine, there is an almost endless variety of configurations of these five basic components. However, the best involve microprocessors (computers) for data acquisition, storage and reproduction. 5.4.1 Structural Strains versus Particle Velocity While particle velocity is the traditional measurement of choice, structural strains control cracking. They should be measured directly from relative displacements on structures or within rock masses when critical locations are known, and can be obtained with a variety of strain and relativedisplacement gauges [20]. Unfortunately, these critical locations may be either unknown or too many in number to economically measure. Therefore, some means of estimation is necessary. Ground motion and air overpressure time histories can be employed to calculate the relative displacement of structural components with a knowledge of the responding structure's dynamic response characteristics [4]. These relative displacements can in turn be employed to calculate strains. The accuracy of these estimates is limited by the degree to which the structure behaves as a single degree of freedom system, and the accuracy of the estimate of the dynamic response characteristics. 5.4.2 Appropriate Measurement of Particle Velocity While any of the three kinemetric descriptors (displacement, velocity or acceleration) could be employed to describe ground motion, particle velocity is the most preferable. It has the best correlation with scientific observation of blast-induced cracking, which forms the basis of vibration control. Furthermore, it can be integrated to calculate displacement. If acceleration is desired, it should be measured directly to avoid differentiation of the particle velocity time history. The location for measurement varies throughout the world. In North America, the excitation or ground motion is measured on the ground adjacent to the structure of interest. In Europe, the excitation motion is measured on the structure's foundation. The difference stems from historical precedent and location of transducers during scientific observation of cracking rather than differences in philosophy. In North America, frequently it is impossible to place transducers on adjacent property owned by a party not involved in the project. Furthermore, if it is desired to describe the excitation motions, then those motions should be measured outside of, and not on, the structure. If it is desired to measure structural response motions then they should be measured on the most responsive structural members, which are not the basement or foundation walls because of the restraint provided by the ground. Time histories of the three components of motion should be measured because of the importance of excitation frequency. Recording of peak motions will not yield information about the dominant frequency and time history details that control structural response. Peak motions and dominant frequency can be employed to describe low-level, noncritical motions. Therefore machines employed to monitor critical motions (Type I, Section 5.4.7) should be capable of recording time histories of selected critical motions. Machines that record only peak motions (Type II, Section 5.4.7) can be employed with those that record time histories to provide redundant measurement where frequency content does not vary widely, and where particle motion is low. 5.4.3 Transducers Transducers are one of the weaker links in the measurement system because they must translate kinematic motions or pressures to electrical signals. The remaining components transform electrical
122
Blasting 40 30 20 15 £
I
i D
8
-
k
**
6
4
3— f
—«^,^.^.
~/~X^1 /
1 1^1.1,1 0.2 0.3 0.4 0.60.8 I
1 2
.1.1.1.1 3 4 6 8 10
π - c\ ° n β - 0 7
1 20
,
1 . 40
Frequency (Hz)
Figure 7 Example response spectra of a velocity transducer with differing percentages of damping. With 70% of critical damping this system is x 3 dB ( x 30%) down 1 Hz (Dowding [4])
signals or light beams and are not restricted by mechanical displacement. The main characteristics of transducers that affect their performance are sensitivity and frequency response. Sensitivity of an instrument is the ratio of its electrical output to its kinemetric displacement, velocity, and acceleration or overpressure for energy-converting transducers (i.e. do not require an energy source). Since allowable limits are specified in terms of ground particle velocity, all blast monitors come equipped with velocity gauges. Frequency response is the frequency range over which the electrical output is constant with a constant mechanical motion. This constancy is normally expressed in terms of decibels (dB). For instance, linear within 3 dB between 5 and 200 Hz means that the transducer produces a voltage output that is constant within 30% between 5 and 200 Hz. Generally, it is better to look at the transducer's response spectrum (such as those shown in Figure 7) to determine the frequencies where this difference occurs. For example, the difference occurs at low frequencies for the velocity transducers in thefigure.The importance of the frequency response of air overpressure transducers was discussed in Section 5.3. 5.4.4 Transducer Attachment One of the most critical aspects of vibration monitoring is the mounting of the transducers in the field. The importance of mounting is a function of the particle acceleration of the wave train being monitored. The type of mounting on a horizontal surface is the least critical when the vertical maximum particle accelerations are less than 0.3g. In this range, the possibilities of rocking the transducer or the transducer package are small, and the transducer may be placed upon a horizontal measurement surface without a device to supply a holding force. When the maximum particle accelerations fall between 0.3# and 1.00, the transducer or transducer package should be buried completely when the measurement surface consists of soil [21]. When the measurement surface consists of rock, asphalt or concrete, the transducers should be fastened to the measurement surface with either double-sided tape, epoxy or quick-setting cement (Hydrocal or other gypsum-based cements set within 15 to 30 min). If the above methods are unsatisfactory or accelerations exceed l.Og, only cement or bolts are sufficient to hold the transducer to a hard surface. All transducers mounted on vertical surfaces should be bolted in place. Air overpressure transducers should be placed at least 3 ft (1 m) above-ground, pointed downward (to prevent rain damage) and fitted with a wind screen to reduce wind excitation-induced false events. 5.4.5 Digital, Tape and Hard-copy Recorders Microprocessor (computer) or digital recording systems now dominate technical recording because of the ease of computer linkage. The signal is sampled at a certain rate, say, 1000 times per second, and each sample is converted to a single magnitude. This magnitude and its associated time are then stored in computer memory. Digital recording has several advantages. It is very accurate, as
Blast Vibration Monitoring for Rock Engineering
123
variation in the speed of the tape has no effect, and records can be directly accessed by a computer. Details of the digitization process are discussed elsewhere [4]. Of those blast-monitoring systems with tape recorders, most employ compact FM cassettes. Many of the tape systems involve separate record and reproduction modules to reduce the complexity of recording. Care should be exercised to determine the exact details of the system before purchasing, as tape recorder performance varies at low temperature. A permanent record or 'hard copy' of the vibration time history is usually made on photographic film, floppy disk, battery-powered memory chips or paper. Almost all the current film-based recorders employ special field-developable, ultraviolet light sensitive paper in combination with light-beam galvanometers to record high-frequency motions. The newest generation recorders employ dot matrix printers and/or floppy disks with microcomputers. Unfortunately, those monitors that automatically print after a vibration event may not be recording another event while printing. If multiple shots are likely, this reset time should be determined. Furthermore, printer behavior in cold weather is variable and should also be investigated. Most recorders can be bought as either single- or multi-channel units. A four-channel unit is necessary in blast monitoring to record simultaneously the three components of the ground motion (L, V and T) and the air blast. The present trend in vibration equipment is to include a signalconditioning amplifier in the recorder to allow flexible amplification of the signals. Frequency analysis of records requires a time history and thus some form of permanent record. Instruments recording only peak particle velocities will not allow a frequency analysis. Sending permanent records through the mail for interpretation results in a delay offivedays, and sometimes up to a month. Systems with light-sensitive paper or dot matrix printers allow immediate interpretation of frequency without additional costly equipment.
5.4.6 Calibration It is obvious that the entire vibration measurement system should be calibrated, as it is futile to record data if they cannot be exploited because of a lack of reference. Manufacturers supply calibration curves with their instruments that are similar to the response spectra for transducers shown in Figure 7. Recalibration, or checking, requires special vibrating platforms where frequency and displacement are controlled, and, in thefield,a calibrating circuit to pulse the magnetic core of the geophone [22].
5.4.7 Number of Instruments While the obvious irreducible number of instruments for each blast is one, two would provide a more thorough documentation of the spatial distribution of effects. If only one instrument is employed, then it should be located at the nearest or most critical receiver. This single, Type I instrument should record time histories of the three axes of particle velocity as well as air overpressure. Since it must monitor continuously, it must trigger (begin recording) automatically, and be capable of monitoring even while printing or communicating results. When blasting will occur at more than one general location (i.e. involve different nearest structures separated by hundreds of feet or meters), then two and four are the irreducible and optimum number of instruments respectively. A third should be available, but not deployed, to insure continuous coverage in case of failure. The second and fourth instruments in the situations described above may provide a lower level of information and will be termed Type II. They must at least continuously record the peak particle velocity in one axis and may or may not measure air overpressure. The best axis is the vertical, since no horizontal direction decision is required and surface waves usually involve a significant vertical component regardless of the direction of the maximum horizontal component. These instruments should be located at a greater distance than the nearest structure to monitor a large area. The third or spare instrument can be either Type I or II. Where air overpressures will be problematic or frequencies critical, the spare should be Type I. This spare instrument can also be employed to monitor sites where complaints develop. This public relations work is essential in North America where law suits arise even when all blast effects comply with regulatory guidelines. The above approach describes the least number of instruments. Applicable regulations and mining schedules may require a larger number.
124
Blasting
5.4.8 Instrument Deployment during Test Blasts When blasting projects begin, when geological conditions change radically, or when new initiation systems are introduced, test blasts should be conducted to minimize the number of instruments necessary to monitor production blasts. These tests are conducted to produce project-specific attenuation relations for both air overpressures and ground motion. Such relations vary from project to project because of changes in geology and blasting practices. Additionally, the test blasts allow the determination of the frequency content of motions at different scaled and absolute distances. Frequency is important in estimating structural response through response spectrum analyses. The attenuation relation is not solely a site property. Although it is dependent upon geology, it is also heavily dependent upon the blast geometry and timing. For instance, with the same charge per delay, a blast with a larger burden will produce an attenuation relation having a similar slope or decay with distance but with a larger intercept. Furthermore, differing initiation timing will produce changes in the time history, both length and frequency content. During test blasts, a minimum of four instruments should be deployed to measure peak particle velocity at widely differing scaled distances for the same blast. Therefore, for any one blast design, parameters and initiation sequence are constant, and the resulting attenuation relationship shows only the effect of distance, direction, and/or geology. Seismographs and/or transducers should be placed along a single line with constant geology to determine the best attenuation relationship, or at all critical structures to determine the effects of direction and variable geology. Ideally, the linear orientation should be along a path with constant thickness of soil and not cross any large geological discontinuities such as faults. If geology changes radically, then two such attenuation lines are necessary, but not necessarily with each blast. A number of approaches to blast design for vibration control are now available that employ a single-delay, single-hole test blast and a number of instruments to record the attenuation and frequency change around the site. These single-time histories are then synthesized to reproduce the additive time history effects of multiple-delay, multiple-hole blasts at the differing instrument sites. Such synthesis of time histories to guide blast design has met with variable success, but does not replace monitoring of blast effects at critical structures during production blasting.
5.5 STRUCTURAL RESPONSE Documentation of blast effects involves two radically differing endeavors: measurement of ground and air disturbances as well as observation of cosmetic cracking. Measurement can now be accomplished remotely with computers to eliminate completely human interaction, whereas scientific observation of cracking must involve meticulous human inspection immediately before and after a blast. While the focus of this section is on instrumental monitoring, the alleged appearance of cracking by neighboring property owners is nonetheless a very serious consideration. Principal problems in the evaluation of measured effects involve (i) accounting for geological and weather effects on the overall attenuation with a small number of instruments; and (ii) incorporating structural response and frequency effects. Principal problems with the observation of blast-induced cracking involve (i) separating blast-induced from environmentally and human-induced cracking; and (ii) reducing the enormous amounts of time necessary for direct observation. Observational problems are normally overcome by employing measurable blasting controls at particle velocities low enough to prevent the threshold of cosmetic cracking even to old, degraded structures and eliminating observation altogether. Otherwise blast-induced cracking can be observed only with immediate before and after blast inspection. The remainder of this section will concentrate on the instrumentation approach and calculation of structural response. 5.5.1 Structural Response and Frequency Effects Structures respond to both ground- and air-borne disturbance, as shown by the bottom four time histories in Figure 1. Walls respond more to the higher frequency (15 to 20 Hz) waves in the early portion of the ground motion, while the superstructure, or overall skeleton of the structure, responds more to the last or lower frequency (5 to 10 Hz) portion. Walls are again excited by the late arriving air-pressure wave. Structural response can be calculated from the ground motions and air pressure excitation if the natural frequency and damping of structural components are known or estimated.
Blast Vibration Monitoring for Rock Engineering
125
Langan [23] has shown that measured structural response has a higher correlation coefficient with calculated single degree of freedom (SDF) response than with peak ground motion. Therefore structural motions can be estimated more accurately by assuming that they are proportional to response spectrum values at the particular structure's natural frequency than by assuming that they are proportional to the peak ground motion. This improved correlation is largely a result of the consideration of frequency in the response spectrum, which is calculated from SDF response. 5.5.2 Origin of the SDF Model One of the critical structural response factors is the amount of differential displacement (<5 in Figure 8) that occurs between or along structural members because it is proportional to strain which, in turn, causes cracking. Such displacements can be computed with a mathematical idealization of the SDF model shown in Figure 8. It is necessary to simplify a structure so that computations are practical. The fundamental characteristics of a structure that govern its behavior under vibratory or dynamic loading are: (i) the masses of the main components (analogous to floor and roof masses); (ii) the spring stiffnesses of the main components (analogous to wall stiffness); and (iii) the amount of damping or energy dissipation (analogous to differential movement in cracks, joints and connections). Behavior of one- or twostory buildings is directly analogous to the behavior of an SDF system when movement in only one direction is considered. When multi-storied structures are considered, it is necessary to model the structure as multi degree of freedom systems. However, even multi degree of freedom systems may be idealized as a single degree of freedom system to calculate the fundamental mode of response. If a structure's dampened natural frequency,/d, and its fraction of critical damping, /?, are not known, values of dynamic properties,/d and /?, can be accurately measured from a free vibration time history of the building response. These measured parameters automatically account for the factors that are difficult to quantify, such as the degree offixityof the columns and the damping coefficient. As shown in Figure 1, these parameters can be measured from the structure's free response. Time between peaks is the period, T= l//d, and the decay of free oscillation is proportional to the damping, ß. 5.5.3 Estimation of Dynamic Response Properties The fundamental natural frequency of the superstructure of any tall building can be estimated from compilations of work in earthquake engineering (Newmark and Hall, 1982) (6)
'- - 5^v
where N is the number of stories. Substitution of 1 and 2 for residential structures for JV yields / values of 10 and 5 Hz for one- and two-story structures, which compares favorably with the results of actual measurements. Damping, /}, is a function of building construction and to some extent the intensity of vibration. Thus is cannot be simplified as easily as the natural frequency. Measurement reveals a wide range of damping for residential structures with an average of 5% [24]. This value is also appropriate for
Spring
i
|
i
|
k
1 j
Γ
V
//////////////. Figure 8
1vlass
Dashpot ~~ damping Base
Single degree of freedom model of house that shows relative displacements of the walls, <5, and the analogy between model mass and roof mass, model stiffness, and wall stiffness (Dowding [4] )
126
Blasting Table 1 Natural Frequencies for Unusual Structures Type
Height (m)
/(Hz)
30 21 60 27
3.8 1.2 0.6 3
Radio towera Petroleum distillation tower3 Coal silo Bryce Canyon rock pinnacle0 a
Medearis [37]. b Dowding and Kendorski [36].
initial estimates involving taller engineered structures. Further details for engineered structures can be found in Newmark and Hall (1982). Walls and floors vibrate independently of the superstructure and have their own, but similar, fundamental frequencies of vibration that range between 12 and 20 Hz with an average value of 15 Hz [24]. Floors tend to have lower natural frequencies in office buildings with large floor spans but are similar to wall natural frequencies in residential structures. Dynamic response properties of some tall, unique structures cannot be estimated with the 1/0.IN equation. Field-measured natural frequencies for these types of structures are given in Table 1. 5.5.4 Response Spectrum The pseudo velocity response spectrum of a single ground motion, such as that of a seven-delay quarry blast in Figure 9, is generated from the relative displacement, (5max, values of a number of different SDF systems when excited by that motion. Consider two different components of the same structure, the 10 Hz superstructure and the 20 Hz wall. If the ground motions, ù(t) in Figure 9(b), of the seven-delay quarry blast are processed twice by the SDF response equation with/= 10 and 20 Hz and ß held constant at 3%, two <5max values will result.
IOO
20
IOO
Frequency (Hz) (b)
Ü m a x =27.9mm s"1 ( I I I in s"1) I
-^·
Figure 9 Construction of a pseudovelocity response spectrum: (a) response spectra; (b) associated excitation time history. The time history is operated upon by the single degree of freedom (SDF) equation to produce a computed relative displacement (<5), which is then multiplied by the circular natural frequency [2π/; or 2π10 for point (1)] to produce the pseudo response velocity (Dowding [4] )
Blast Vibration Monitoring for Rock Engineering
127
The first computation is made with the 10 Hz system, which has a circular natural frequency of p = 2π(/) = 2π(10)
(7)
and results in an SDF equation computed <5max = 0.25 mm (0.01 in)
(8)
This <5max is then converted to pseudovelocity, PV, as PV10 = pc5max = 2π(10)(0.25) = 15.7 mm s"1 (0.62 ins"1)
(9)
and is plotted as point 1 in Figure 9(a). The same computation is then repeated for the 20 Hz system p = 2π(20) <5max = 0.5 mm (0.02 in) PV20 = 2π(20)(0.5) = 63.5 mm s" ^
ins- 1 )
and PV 20 is plotted as point 2 in Figure 9(a). If the same ground motion time history in Figure 9(b) is processed a number of times for a variety of/s with β constant, the resulting pseudo velocities will form the solid line in Figure 9(a).
5.5.5 Fourier Spectra do not Directly Predict Response With increasing use of computers, calculation of various spectra from time histories has become commonplace. The two most common are the Fourier frequency and pseudo velocity response spectra. Although they are essentially different in meaning and typical use, they are similar for undamped response where the maximum motion occurs near the end of the time history [4, 13]. Since response spectra are calculated for damped response and peaks normally occur in the middle as well as the beginning of the time history, the two spectra are not usually the same. Only the pseudo velocity response spectrum can be employed to calculate directly the structural response. Because of the similarity of Fourier and response spectra, either can be employed to determine the dominant frequency in the ground motion. 5.5.6 Case Histories Demonstrate Importance of Response Spectrum Analysis Figure 10 compares time histories and response spectra from the longitudinal components of an urban construction blast and a surface coal-mine blast. Although the peak particle velocities are similar (3.8 mms" 1 for the construction blast, A, and 3.3 mms" 1 for the surface mining blast, B), the response spectra differ radically. This difference is greatest in the range of natural frequencies of residential structures and their components, 5 to 20 Hz. In this range the surface mining motions produce response velocities that are 10 times greater than the construction blast.
5.5.6.1 Surface mine blast Surface mining-induced ground motion was produced by a multiple-row blast. Some 60, 25 m deep, 380 mm diameter, holes were arranged in a 4-row pattern. The burden between rows was 6.1 m and the hole spacing was 7.6 m. Each hole contained four decks (or charges that are detonated at intervals separated by at least 17 ms). The ammonium nitrate-fuel oil (ANFÖ) charge weight per deck ranged from 45 to 60 kg. Therefore the largest charge per delay was 60 kg, and the total charge was 12600 kg. Geology between the blast and the transducer, located at the nearest residence, consisted of sedimentary rock with 3-10 m of overlying silty glacial till. A small, 10 m deep, gully was located some 400 m north of and between the blast and transducer. Soil depth at the transducer was 3-4 m. Some 825 m separated the shot from the transducer, where the longitudinal velocity time history in Figure 10 with a peak of 3.3mms" 1 was recorded. The accompanying transverse and vertical peak particle velocities were 4.6 and 2.3 mm s" 1 with dominant frequencies between 11 and 13 Hz. Other
Blasting
128 I.Oin
O.I in
4
6
8 10
20
40
60 80 100
Frequency (Hz)
Figure 10 Comparison of time histories spectra from construction and surface mining blasts respectively lasting 0.15 and 2.0 seconds. Even though the particle velocities are approximately equal, responses in the 5-20 Hz frequency range differ greatly
similarly designed shots with distances between 580 and 825 m produced peak particle velocities between 4.3 and 5.8 mms" 1 with dominant frequencies between 13 and 17 Hz. These particle velocities are high at 825 m according to scaled distance relationships. These greater than normal particle velocities may be a result of unusual confinement (too large a burden) or delay overlap. For instance, if two delays had overlapped additively, then the maximum charge per delay would have doubled and the square-root scaled distance would have declined by 30%. Particle velocities measured at smaller scaled distances were closer to expected levels.
5.5.6.2
Urban construction blast
Construction blast-induced motion was produced by a much smaller shot than the surface mining example. Some five, 3.6 m deep, 38 mm diameter holes were arranged in a single row. Each hole was charged with a stick of gelatin dynamite and initiated separately with a constant 25 ms between each delay. The burdens and hole spacings were small, approximately 0.6-0.9 m. The total charge was 9 kg and the maximum charge per delay was 2.3 kg. The structure of concern, a historic theater, and the recording transducers were located some 15 m away. Rock being fragmented consisted of granitized biotite schist. This shot produced peak particle velocity in the L, T and V axes of 3.8, 4.1 and 7.1 mms" 1 respectively with dominant frequencies between 75 and 125 Hz. Because these dominant frequencies were so high compared to the natural frequencies of the theater's structural components, their response was less than the peak excitation particle velocity. The transducer recording the excitation motions was located in the basement because there was no stable location outside as rock was being removed immediately adjacent to the theater's wall.
Blast Vibration Monitoring for Rock Engineering
129
Another vertical axis transducer was placed at the mid span of the theater's balcony, whose leftmost support was immediately above the transducer measuring the excitation motions. The peak vertical particle velocity of the theater's balcony was only 3.6 mm s - 1 , approximately half of the peak excitation motion in that axis. 5.6 CONTROLLING BLAST EFFECTS Direct regulation or specification of effects, rather than blast design, is the most effective control from a regulatory viewpoint because effects are so dependent upon details of the shot geometry and initiation sequence. Such dependency renders control impossible by simple regulatory specification of two- or three-shot design parameters. For instance, consider control by specification of the maximum charge weight detonated per instant at given distances from the nearest structure. Even with such detailed specification, intended vibration levels at the structure may be exceeded because of poor choice in the location of holes and/or their relative time of initiation. Present regulatory control limits in many countries are below those levels at which cosmetic cracking may appear. There are two principal reasons for such tight restrictions. First, regulatory limits are influenced heavily by human response to blast-induced vibration and noise. Since humans are approximately 10 times more sensitive than structures to vibration, low regulatory limits are understandable. Secondly, many regulations appear to have been adopted without documented, scientific experimentation necessary to determine the vibration levels that cause cracking. 5.6.1 Regulation to Prevent Cosmetic Cracking of Residential Structures Regulatory controls in North America are based on the occurrence of threshold cracking of plaster and gypsum wallboard in wood frame, residential structures [4, 10, 20, 35]. Observed cracking is cosmetic in nature and does not affect structural stability. These cosmetic cracks are hairsized and are similar to cracks that occur during the natural aging of structures. In fact they are indistinguishable from those that result from natural aging. Control limits are based upon direct observations of test homes immediately before and immediately after blast events to avoid confusion with the similar cracks that might occur from natural processes. These controls do not apply to engineered structures that are constructed of steel and concrete, buried structures, or adjacent rock. 5.6.2 Statistical Analysis of Data with Pre- and Post-blast Inspection Factors that cannot be measured can be taken into account indirectly by considering the appearance of cosmetic cracks as a probabilistic event. In order to investigate the effects of certain data sets on the overall conclusions, the probability computations of threshold or cosmetic cracking at given particle velocity levels have been made several times [10,25]. All of the observations studied by Siskind involve both immediate pre- and post-blast inspection of walls in residential structures, many of which were old or distorted and whose walls were covered with plaster. Definitions of the observed cracking in each study are described in Section 5.2. Data from various sets of observations were analyzed with cracking observations and the assumption that every cracking observation excludes the possibility of non-cracking at a higher particle velocity [ref. 10, page 55]. If the probability of cracking is calculated as the percentage of observations at lower levels of velocity, the result is the log-normal scaled plot of the probability of cracking versus particle velocity in Figure 11. This approach seems conservative as low particle velocity observations do not count noncracking at higher levels. According to Figure 11, there appears to be a lower limit of particle velocity of 0.5ins" 1 (12mms _1 ) below which no cosmetic or threshold cracking (extension of hairline cracks) has been observed from blasting anywhere in the world. This observation includes the data with unusually low frequencies that were collected by Dvorak [26]. His data are those that tend to populate the lower region of Figure 11. High-frequency data (>40 Hz) show that a 5% probability of minor cracking does not occur until particle velocities reach 3ins" 1 (75mms _1 ) [10]. Admissibility of Dvorak's data has been questioned because of the absence of time histories; some of the other studies, such as that by Langefors et al [27], are also plagued by the same lack of time histories. To resolve this difficulty, only the new US Bureau of Mines observations have been
Blasting
130
Particle velocity (mm s"1) 10
-i
1000
100 i i 11
m—i
• Threshold damage Δ Minor damage ▲ Major damage
95 90 80
70 60 50 40 30 20 10 5
-I
I 1 I I I I 1
Particle velocity (in s"1)
Figure 11 Probability analysis of worldwide blast cracking data (Siskind et al. [10] ). Threshold damage is the occurrence of hair-sized, cosmetic cracks similar to those caused by natural, environmentally induced expansion and contraction Particle velocity (mm s"1) 99 F
=
T
Γ
60
E 50 o fc. 40 & σ | Û
30 20 10
0.2
0.5
I
2
5
10
20
50
1
Particle velocity (in s" )
Figure 12 Probability analysis of blast-induced threshold cracks observed by US Bureau of Mines (Siskind [25] )
included in a recomputation of probabilities in Figure 12. The observations include low-frequency motions associated with surface mining. Again there is a particle velocity, 0.79ins" 1 (20mms_1), below which no blast-induced cracking was observed. 5.6.3 Distinction of Blast-induced Cracking from Natural Cracking Control of blast-induced transient effects to prevent threshold or cosmetic cracking reduces blastinduced displacement or strains in structures to or below that caused by everyday human activities
Blast Vibration Monitoring for Rock Engineering
131
and changes in the weather [20, 28]. In many cases these cosmetic cracks are smaller than cracks caused by other natural or occupant-initiated processes that are active in all constructed facilities. Thus blast-induced threshold cracks can be scientifically observed only with visual inspection immediately before and after each blast. Observations made under less stringently controlled conditions have little scientific merit because of the high probability of environmentally produced cracks occurring between or before visual inspections. 5.6.4 Multiple Origins of Cracks Several institutional references [29-31] present excellent summaries of the multiple origins of cracks. Basically, cracks are found to be caused by the following. (i) Differential thermal expansion. (ii) Structural overloading. (iii) Chemical changes in mortar, bricks, plaster and stucco. (iv) Shrinkage and swelling of wood. (v) Fatigue and aging of wall coverings. (vi) Differential foundation settlement. Over time, all of the causes listed above are likely to crack walls, whether or not blasting occurs. There are three important implications associated with the list above. Structures expand and contract preferentially along existing weaknesses (cracks). Seasonal expansion and contraction along these cracks will return patching and repainting to the original cracked state within several years. This persistent cracking is annoying to those owners who are unaware of the difficulty of patching existing cracks of any kind. Secondly, the distortion that caused the cracking also creates stress concentrations which may lower a wall-covering's resistance to vibration cracking; however, current regulatory limits already implicitly include these distortion effects as explained in Section 5.6.2. Thirdly, these natural cracks continue to occur over time. Therefore, any post-blast inspection at low vibration levels is likely to find new cracks from natural aging unless a pre-blast inspection is conducted immediately before the blast. 5.6.5 Response of Structures to Everyday Activities A comparison of strains produced by blast vibrations and everyday events with those needed to fail wall-covering materials gives perspective to the observation of cracking at low-particle velocities. Table 2 compares strains from daily environmental changes (temperature and humidity) and household activities measured in the US Bureau of Mines test house [20]. The door was slammed adjacent to the wall on which the strains were measured. It appears that in the course of daily life an active family will produce strains in walls similar to those produced by blasting vibrations of 2.5-12 mm s" *. Most astonishing are the measurements in a wood-framed home of relatively enormous strains from daily changes in temperature and humidity. These alone are large enough to crack plaster.
Table 2 Comparison of Strain Levels Induced by Household Activities, Daily Environmental Changes and Blasting
Loading phenomena
Site'
Daily environmental changes
Ki
Household activities Walking Heel drops Jumping Door slams Pounding nails
K2
s7 s7 s, Si S,2
Microstrain induced by phenomena (juinin -1 )
Corresponding blast level* (mm s l) (ins" 1 )
149 385
1.2 3.0
30.0 76.0
9.1 16.0 37.3 48.8 88.7
0.03 0.03 0.28 0.50 0.88
0.8 0.8 7.1 12.7 22.4
" K-! and K2 were placed across a taped joint between two sheets of gypsum wallboard.b Blast equivalent based on envelope line of strain versus ground vibration. Source: Stagg et al. [20].
132 5.6.6
Blasting Comparison of Blast and Environmental Effects
Crack-width changes from ground motions less than 25rams" 1 are less than those caused by the weekly passage of weather fronts [28]. This conclusion was reached after measuring the displacement response of a poorly built, non-engineered, wood-frame house to surface coal-mining vibrations for some 8 months. Displacements were measured at 10 different wall positions that included cracked and uncracked wall covering. Weather and blast-induced crack displacements across the most dynamically responsive wall-covering crack are compared in Figure 13. The continuous and highly cyclical curve is that of displacements produced by environmental change. The small circles are the maximum, zero-to-peak, dynamic displacements recorded by the same gauge. Even though the maximum recorded particle velocity was as high as 2 4 m m s _ 1 , the maximum weather-induced displacements were 3 times that produced by blasting. On other gauges, weather changes produced displacements that were 10 times greater than those produced by blasting.
5.6.7
Special Considerations
The statistically determined control limits are too low for cracking in concrete basement walls, rock masses and engineered structures. They were based on response of residential structures and the lower-limit cases involved cracking of above-ground plaster or gypsum wallboard wall coverings in older, distorted structures.
5.6.7.1
Engineered structures
Concrete is a good deal stronger than plaster. Therefore, engineered structures constructed of concrete can withstand maximum particle velocities of at least 100mm s" 1 without cracking [32]. Furthermore, buried structures such as pipelines and tunnel linings are not free to respond as were the above ground residential structures whose response provides the data from which most limits are chosen. Specific engineered structures should be analyzed in terms of the strain that can be withstood by critical elements and the strain should be measured. This approach is particularly appropriate for individual structures with isolated portions nearer to the blast source.
5.6.7.2
Restrained structures
Buried or restrained structures such as pipelines and rock masses cannot respond as freely as above ground structures and therefore have much larger allowable particle velocities. Whereas
Year
Figure 13 Comparison of crack opening displacements produced by weather-induced changes in humidity and temperature (continuous line) with those produced by surface coal-mine-induced ground motions ( o )
Blast Vibration Monitoring for Rock Engineering
133
strains in a freely responding structure are proportional to the relative displacement between the ground and the superstructure as shown in Figure 8, strains in a restrained structure such as a pipeline will usually be those of the surrounding ground, can be approximated as those produced by plane wave propagation, and are ù û ε = — and y = — cs cc
(10)
where ε and y are axial and shear strains, cc and cs are compressive and shear wave propagation velocities, and ù are maximum compressive and shear wave particle velocities [4]. For cases involving one critical location along a pipeline, the pipe strains should be measured directly on the metal [33]. For cases involving tunnel and/or cavern liners critical strains can be estimated through calculation of the relative flexibility of the rock and liner [34].
5.6.8
Fatigue or Repeated Events
Since current regulatory limits are so low as to restrain blast-induced displacements below those caused by the weekly passage of weather fronts, the question of repeated events becomes moot. Weather by itself over the years produces greater repeated event effects than does blasting. A repeated event experiment conducted at the US Bureau of Mines test house [20], confirms the low level of current regulatory controls with respect to fatigue cracking. The test house was framed in wood with paper-backed gypsum-board interior walls. When continuously vibrated at an equivalent ground particle velocity of 12 m m s " 1 , no response was observed until 52 000 cycles, when a taped joint between sheets of gypsum board cracked. These taped joints are the weakest and most compliant zones in a house with paper-backed gypsum-board walls.
5.6.9
Rock Mass Cracking and Displacement
Cracking in rock immediately adjacent to a blast can be controlled by limiting the particle velocities to 700 mm s - 1 in the volume of rock to be protected [3]. Rock displacement by forces produced by delayed gas pressure cannot be controlled by specifying an allowable particle velocity. Fortunately, these displacements occur only very close to a blast (within 30-50 m) and are associated with blocks that are unconstrained by other surrounding rock. Sliding instability of individual rock blocks must be evaluated on a case by case basis. Each block must have an adequate factor of safety to prevent static failure [5].
5.6.10
Frequency-based Control with Dominant Frequency
Figure 14 shows the limit adopted by the US Office of Surface Mining, that is based on a suggested, but not rigorously validated, proposal by the US Bureau of Mines [35]. Point (1) represents the lowest particle velocity at which USBM personnel have observed cosmetic cracking. Neither of the points (2) and (3) have been confirmed by direct USBM observation. The dominant frequency that is consistent with Figure 14 is that associated with the peaks in the time history with amplitudes greater than 50% of the peak or maximum particle velocity. The frequency of these peaks was calculated from the zero-crossing method as shown in the inset for Figure 3. Determining frequency from that associated with the peak particle velocity is a good first approximation and eliminates the need for sophisticated Fourier or, alternatively, response spectra analysis. Response spectrum analyses are the most precise approach to account for the frequency effects of structural response and should be employed in singular cases where an exacting analysis is required.
5.6.11
Regulatory Compliance for Air Overpressures
Although broken glass is normally associated with excessive air-blast overpressures, limits in the United States are based upon wall response necessary to produce wall strains equivalent to those produced by surface coal mining-induced ground motions with peak particle velocity of 19mms" 1 .
Blasting
134
F
{
k
c-1
,_%
(3)/
p σ
E 3
0.2
s
0.1
E
m
y
r Γ(ΐ)
λ
/ _/
/ //
/
1 1
1
K i
1 1 1 1 1 1 i i 11 1
J L 1 4 10 20 30 Blast vibration frequency (Hz) i l l
ILiJ
Figure 14 Frequency-based blast vibration control limit to protect residential structures modified by US Office of Surface Mining from US Bureau of Mines suggestion (Federal Register, vol. 48, no. 46,1983). Point (1) is verified. Dotted line has been employed safely for close construction blasting near engineered structures. Dashed line has been employed safely for construction blasting in urban areas near older homes and historic buildings
Table 3
Air Overpressure Control Limits as a Function of Instrument Frequency Weighting
Lower frequency limit of measuring system (Hz, — 3 dB) 0.1 or lower - flat response8 2 or lower - flat response 6 or lower - flat response C-weighted - slow response1 1
Maximum level (dB) 134 peak 133 peak 129 peak 105 peak dBC
Only when approved by the regulatory authority.
These limits are presented in Table 3. If a wall-strain level equivalent to that produced by 25 mms" 1 particle velocity (measured in the ground) were chosen, the allowable overpressure would increase by 3 dB. Most cases of broken glass are reported to have been observed at air overpressures of 136-140 dB (as measured with a linear transducer). Because of the different sound-weighting scales that might be employed by monitoring instruments, the recommended levels in Table 3 differ by instrument system. Since structures are most sensitive to low-frequency motions and the greatest air pressures occur at these inaudible frequencies, A-weighted scales cannot be employed at all. Since C-weighted scales are the least sensitive at low frequencies, their use requires the most restrictive limits.
5.7
REFERENCES
1. Dowding C. H. Monitoring and control of blast effects. In SME Mining Engineering Handbook, 2nd edn. (Edited by H.Hartman) (1992). 2. Siskind D. E. and Fumanti R. Blast-produced Fractures in Lithonia Granite. US Bureau of Mines, Report of Investigations 7901 (1974). 3. Holmberg R. and Persson P. A. The Swedish approach to contour blasting. In Proc. 4th Conf Explosives and Blasting Techniques, Society of Explosives Engineers, Montville, OH, pp. 113-127. (1978). 4. Dowding C. H. Blast Vibration Monitoring and Control Prentice-Hall, Englewood Cliffs, NJ (1985). 5. Dowding C. H. and Gilbert C. Dynamic stability of rock slopes and high frequency traveling waves. J. Geotech. Eng. Div. Am. Soc. Civ. Eng. 114, GT10, 1069-1088 (1988). 6. Roth J. A Model for the Determination ofFlyrock Range as a Function of Shot Conditions. Report prepared for the US Bureau of Mines by Management Services Association, Los Altos, CA NTIS, PB81-222358 (1979). 7. Lundborg N. The Probability ofFlyrock, Report DS 1981:5. Swedish Detonic Research Foundation, Stockholm (1981). 8. Ivanov P. L. Compaction of Noncohesive Soils by Explosions (Translated from Russian by the National Science Foundation). Available from the library of the US Water and Power Resources Services, Denver, CO (1967).
Blast Vibration Monitoring for Rock Engineering 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37.
135
Northwood T. D., Crawford R. and Edwards A. T. Blasting vibrations and building damage. The Engineer 215, 5601 (1963). Siskind D. E., Stagg M. S., Kopp J. W. and Dowding C. H. Structure Response and Damage Produced by Ground Vibrations from Surface Blasting. US Bureau of Mines, Report of Investigations 8507 (1980). Kamperman G. and Nicholson M. A. The Transfer Function of Quarry Blast Noise and Vibration into Typical Residential Structures. US Environmental Protection Agency Report, EPA 55/9-77-351 (1970). Borsky P. N. Community Reactions to Sonic Booms in the Oklahoma Area. Final Report for Aerospace Medical Research Laboratory, Wright-Patterson Air Force Base, AMRL-TR-65-37, vol. II (1965). Hudson D. E. Reading and Interpreting Strong Motion Accelograms. Earthquake Engineering Research Institute, Berkeley, CA ( 1979). Hendron A. J. Engineering of rock blasting on civil projects. In Structural and Geotechnical Mechanics (Edited by W. J. Hall). Prentice-Hall, Englewood Cliffs, NJ (1977). Medearis K. The Development of a Rational Damage Criteria for Low Rise Structures Subjected to Blasting Vibrations. Report to National Crushed Stone Association, Washington, DC (1976). Wiss J. F. and Linehan P. W. Control of Vibration and Blast Noise from Surface Coal Mining. Research Report for the US Bureau of Mines, Contract 10255022 (1978). Schomer P. D., Goff R. J. and Little M. The Statistics of Amplitude and Spectrum of Blasts Propagated in the Atmosphere. US Army Construction Engineering Research Laboratory, Technical Report N-13 (1976). Harris C. M. and Crede C. E. (eds.) Shock and Vibration Handbook. McGraw-Hill, New York (1976). Rosenthal M. F. and Morelock G. L. Blasting Guidance Manual. Office of Surface Mining Reclamation and Enforcement, US Department of the Interior, Washington, DC (1987). Stagg M. S., Siskind D. E., Stevens M. G. and Dowding C. H. Effects of Repeated Blasting on a Wood-Frame House. US Bureau of Mines, Report of Investigations 8896 (1984). Johnson C. F. Coupling Small Vibration Gauges to Soil. Earthquake Notes 33 (3), 40-47. Eastern Section, Seismological Society of America (1962). Stagg M. S. and Engler A. J. Measurement of Blast Induced Ground Vibrations and Seismograph Calibrations. US Bureau of Mines, Report of Investigations 8506 (1980). Langan R. T. Adequacy of Single-Degree-of-Freedom System Modeling of Structural Response to Blasting Vibrations. M.S. thesis. Department of Civil Engineering, Northwestern University, Evanston, IL (1980). Dowding C. H., Murray P. D. and Atmatzidis D. K. Dynamic response properties of residential structures subjected to blasting vibrations. J. Struct. Eng. Am. Soc. Civ. Eng. 107, ST2, 1233-1249 (1981). Siskind D. E. Open File Report of Responses to Questions Raised by RI 8507. Available for inspection, US Bureau of Mines, Minneapolis, MN (1981). Dvorak A. Seismic Effects of Blasting on Brick Houses, Prace Geofyrikenina Ustance, Ceskoslovenski Akademie, Ved., No. 159, Geogysikalni, Sbornik (1962). Langefors U., Westerberg H. and Kihlström B. Ground vibrations in blasting. Water Power, Sep. (1958). Dowding C. H. Comparison of environmental and blast induced effects through computerized surveillance. In The Art and Science of Geotechnical Engineering at the Dawn of the 21st Century, R. B. Peck Honorary Volume (Edited by W. J. Hall), pp. 143-160. Prentice-Hall, Englewood Cliffs, NJ (1988). Blasting Claims: A Guide to Adjusters, Association of Casualty and Surety Companies (available by writing to the Association, 110 William St., New York, NY 10038) (1956). Cracking in Buildings. In Building Research Establishment Digest 75. Building Research Station, Garston, UK (1977). Thonen J. R. and Windes S. L. Seismic Effects of Quarry Blasting. US Bureau of Mines Bulletin 441 (1942). Crawford R. and Ward H. S. Dynamic Strains in Concrete and Masonry Walls. Building Research Note 54. National Research Council of Canada, Ottawa (1965). Dowding C. H. Construction Vibration Monitoring and Control. Prentice Hall, Englewood Cliffs, NJ (in press). Hendron A. J. and Fernandez G. Dynamic and static design considerations for underground chambers. In Seismic Design of Embankments and Caverns (Edited by T. Howard), American Society of Civil Engineers, Special Technical Publication (1983). Siskind D. E., Stachura V. J., Stagg M. S. and Kopp J. W. Structures Response and Damage Produced by Airblast from Surface Mining. US Bureau of Mines, Report of Investigations 8485 (1980). Dowding C. H. and Kendorski F. S. Response of rock pinnacles to blasting vibrations and airblasts. Bull. Int. Assoc. Eng. Geol. 20, 271-281 (1983). Medearis K. Structural Response to Explosion Induced Ground Motions. American Society of Civil Engineers, Special Publication, New York (1975).
6 Computer Modeling and Simulation of Percussive Drilling of Rock BENGT LUNDBERG Uppsala University, Sweden 6.1
INTRODUCTION
137
6.2 ELASTIC WAVES IN RODS 6.2.1 Elastic Waves in a Uniform Rod 6.2.2 Normal Force and Particle Velocity 6.2.3 Momentum and Energy 6.2.4 Nonuniform Rods and Boundary Conditions 6.2.5 Initial Conditions and Impact
138
6.3
141
138 139 139 140 141
MODELING PERCUSSIVE DRILLING
6.3.1 6.3.2 6.3.3 6.3.4 6.3.5
141 142 144 146 147
Components and Members Segment Members Interface Members End Members Dimensionless Representation
6.4 SIMULATION AND VALIDATION 6.4.1 Features of Simulation Programs 6.4.2 Efficiency of a Percussive Process 6.4.3 Down-the-hole Drilling 6.4.4 Energy Dissipation at a Joint 6.4.5 Extension Drilling
148 148 148 148 150 151
6.5
SUMMARY AND CONCLUSIONS
153
6.6
REFERENCES
153
6.1
INTRODUCTION
The main common feature of different methods of percussive drilling is the use of impact as a means of generating the large forces needed to crush rock. In churn drilling the hammer and the bit form a unit which impacts on the rock. In down-the-hole drilling, drifter drilling and extension drilling, there are intermediary members which transmit force and energy from the hammer to the rock. In down-the-hole drilling this member is the drill bit. In drifter drilling it may be an integral drill steel or a drill rod with an adapter and a drill bit attached to its ends. In extension drilling it is usually a drill string composed of an adapter, a number of joined extension rods and a drill bit. As a result of impact elastic waves [1, 2] are generated. They are associated with elastic strain energy and kinetic energy and play important roles in the conversion of the kinetic impact energy into work. In drifter drilling and extension drilling, for example, the waves should carry the impact energy to the drill bit with a minimum of losses. These may be due for instance, to reflections which occur at changes in the cross-sectional area or in other properties of the drill string. Losses may also be due to frictional work in the threads of joints. Additionally, the waves have the detrimental effect of forming high stresses which may limit the life of the drill.
137
138
Mechanized Excavation
The hammer is commonly driven hydraulically or pneumatically. Typically, the impact frequency is 50 to 100 Hz, and the impact velocity is 10 m s" *. As a result of the reaction forces from the rock, the drill receives impulse or momentum directed out of the hole. In order to keep the drill bit in contact with the rock, it is therefore necessary to apply a feed force. It is also indispensable to rotate the bit a certain angle during each impact cycle in order to avoid the bit getting stuck. Furthermore, the hole must beflushed.This is normally done by letting air or water through a central axial hole in the drill string with outlets at the end of the drill bit. In order to understand, predict or control the operation of percussive drills, it is helpful to establish models which simulate important aspects of their behavior with sufficient accuracy. The models can be defined in terms of algebraic equations, such as Hooke's law; ordinary differential equations, such as Newton's second law; partial differential equations, such as the wave equation; and other mathematical-logical relations. Knowledge about the behavior of the real percussive drills is achieved from a study of such model relations. Early modeling and analyses relevant to percussive drilling were carried out by Donnell [3] and Dahl [4] in the early 1930s. They applied the one-dimensional theory of elastic waves in rods to the problem of percussive drilling, an approach which has remained valid and useful. The theory was applied in the 1950s by Takaoka and Hayamizu [5], by Arndt [6, 7], who used analysis methods for electrical transmission lines, and by Fischer [8], who applied and refined a graphical method of analysis introduced by De Juhasz [9]. Further modeling and analysis work was carried out by Fairhurst [10], Simon [11, 12] and Dutta [13] in the 1960s, by Fu and Paul [14], Hustrulid and Fairhurst [15] and Lundberg [16] in the 1970s, and by Carlvik [17], Lundberg et al [18-25], Nordlund [26], and Pang and Goldsmith [27] in the 1980s. Computer modeling signifies the establishment of models which are adapted to the use of computers. For an engineering application like percussive drilling it is essential to design the models for maximum simplicity consistent with the desired accuracy. Such models are more easily used and interpreted than more complex ones. Early computer modeling work with relevance to percussive drilling was carried out by Simon [11] and Dutta [13], who modeled the impact between the hammer and the drill rod of a percussive drill, and by Fu and Paul [14], who modeled the energy transfer through chains of impacting rods. After the advent of personal computers and easily available computer graphics at the end of the 1970s there has been a significant increase in the computer modeling of percussive drilling, [e.g. 18, 19, 23-27]. Work in this category has also been published in doctoral dissertations. Computer simulation studies of percussive drilling have been published concerning, for instance, the efficiency of down-the-hole drilling [20] and of extension drilling [21, 25], and the effect of thrust [26]. An important ingredient in the process of computer modeling and simulation [28] is model validation, i.e. the establishment that the model adequately represents the real system. This is done by comparing the results of simulations with those of carefully performed laboratory andfieldtests. If the model is found to be unsatisfactory it should be modified and tested again. Thus, computer modeling and simulation is an iterative procedure which normally involves experimental work. Examples of studies with emphasis on model validation are [22], which concerns a percussive process similar to down-the-hole drilling, and [23, 24], which pertain to transmission of elastic waves through drill rod joints. This chapter deals with computer modeling and simulation of percussive drilling of rock. Emphasis is put on wave-related phenomena during an impact cycle and on the efficiency of percussive drilling. Section 6.2 deals with elastic waves in rods, Section 6.3 with modeling of percussive drilling and Section 6.4 with simulation and validation. The last two sections are based on work carried out by the author and his coworkers at Luleâ University of Technology. 6.2 ELASTIC WAVES IN RODS 6.2.1 Elastic Waves in a Uniform Rod Percussive rock drills are largely built up of straight rods with constant properties along their lengths. Consider now such a uniform, straight and linearly elastic rod with cross-sectional area A, Young's modulus E and density p which are independent of the axial position x. The elastic waves generated in the rod as a result of impact are not free from three-dimensional (3-D) effects. Strictly, therefore, they have a complex behavior. In a percussive drill, however, the waves generated are generally much longer than the transverse dimensions of the drill string. Under these conditions the 3-D effects have little importance, and it is justified to consider the wave motion as 1-D [1, 2].
Computer Modeling and Simulation of Percussive Drilling of Rock
139
In what follows, it is assumed that initially plane cross sections remain plane so that the axial displacement u depends only on the axial position x and the time t. Also, the inertia associated with lateral contraction and expansion of the rod is neglected. Under these conditions, the equation of motion of an infinitesimal rod element results in the 1-D wave equation where
w« = c2uxx
(1)
c = (E/p)1'2
(2)
and where the subscripts indicate partial differentiations. The general solution of equation (1) is u(x, t) = up(x - ct) + un(x + ct)
(3)
where up and un are arbitrary functions. As these functions assume constant values where their arguments x — ct and x + ct are constant, they represent waves which propagate with speed c in the positive (increasing x) and negative (decreasing x) directions, respectively. These directions will also be referred to as right and left, respectively. For a drill rod made of steel, with E = 210 GPa and p = 7.8 x 103 kgm~ 3 , the wave speed is c = 5.2 kms" 1 . As the waves do not change their shapes when they propagate, the wave motion is qualified as nondispersive. 6.2.2 Normal Force and Particle Velocity The normal force N(x, t\ positive in tension, and the particle velocity v(x, t\ positive in the positive direction, are expressed in terms of the displacement u(x9 t) by the relations N(x,t) = AEux(x,t),
v(x,t) = ut(x9t)
(4)
Substitution of equation (3) into equation (4) yields JV(x,r) = Np(x - ct) + Nn(x + ct\
v(x,t) = (1/Z)[-N p (x - ct) + Nn(x + ci)]
(5)
where iVp = AEu'p and Nn = AEu'n are the normal forces associated with the waves traveling in the positive and negative directions, respectively, while Z = AE/c
(6)
is the characteristic impedance. The latter characterizes the cross section of the rod with regard to geometry and material. If the waves are travelling in one direction only, so that either Nn = 0 or Np = 0, equation (5) gives N/v = ±Z (7) where the negative and positive signs are valid for waves traveling in the positive and negative directions, respectively. For a drill rod made of steel with cross-sectional area A = 1.0 x 10 ~3 m2 the characteristic impedance is Z = 40kNsm _ 1 . Thus, a particle velocity of v = 5 ms" 1 , which is common in percussive drilling, corresponds to a normal force with magnitude | JV| = 200 kN, and to a normal stress with magnitude |σ| = 200 M Pa. This illustrates that common particle velocities correspond to large normal forces and stresses. 6.2.3 Momentum and Energy The momentum carried by elastic waves through the cross section x in the positive direction during a given interval of time can be expressed as the impulse of the force —N(x, t\ i.e. P(x) - I (-JV)dr
(8)
Substitution of the first part of equation (5) into equation (8) gives P(x) = - \(NP + Nn)dt
(9)
140
Mechanized Excavation
Correspondingly, the momentum at a time t carried by the elastic waves within a given portion of the rod is given by P(t) = \vpAdx
(10)
in terms of the particle velocity v(x,t). Substitution of the second part of equation (5) into equation (10), and using equations (2) and (6), gives Nnn)dx )d P(t) = (1/c)I \(-Np + N
(11)
The energy carried by the elastic waves through the cross section x in the positive direction during a given interval of time can be expressed as the work of the force —N(x,t), i.e. W(x) =
(-N)t;di
(12)
Substitution of equation (5) into equation (12) gives W(x) = (1/Z) i(N2p - N2)dt
(13)
Correspondingly, the energy at the time t carried by the elastic waves within a given portion of the rod is given by
■i
W(t) = (1/2) (N2/AE + v2pA)dx
(14)
In this expression, the first term represents the elastic strain energy associated with the normal force, while the second term represents the kinetic energy associated with the particle velocity. If there is no overlap between the waves travelling in different directions these terms have equal magnitude. Thus, in this case, the waves are associated with equal amounts of elastic strain energy and kinetic energy. Substitution of equation (5) into equation (14), and using equations (2) and (6), gives
•f
W(t) = (1/Zc) (N2 + iVn2)dx
6.2.4
(15)
Nonuniform Rods and Boundary Conditions
So far we have considered wave propagation in a rod with constant characteristic impedance Z. We shall now study the influence of a discontinuous change in characteristic impedance and of the conditions at an end of a rod. Consider first a rod with constant characteristic impedances Zx and Z 2 to the left and to the right, respectively, of the discontinuity. Then the force and the particle velocity can be represented by equation (5) with subscript T to the left and '2' to the right of the discontinuity. The requirement that the normal force N and the particle velocity v be continuous at the discontinuity gives the results Np2 = TpNpl + RnNn2,
Nni = RpNpl + TnNn2
(16)
for the waves leaving the discontinuity in terms of those arriving. The reflection and transmission coefficients in equation (16) are given by Rp = (Z2 - Zl)/(Zl and
+ Z2), Rn = (Z, - Z2)/(Z1 + Z 2 )
Tp = 2Z2/(Zi + Z2), Tn = 7Z1I(Z1 + Z 2 )
(17) (18)
respectively. If there is no discontinuity, i.e. Zt = Z 2 , it is noticed that Rp = Rn = 0, and Tp = Tn = 1. Thus, there are no reflections. Consider next the discontinuity as the right end of the rod on the left side, and assume therefore that there are no waves arriving from the right. Then, the second part of equation (16) gives the
Computer Modeling and Simulation of Percussive Drilling of Rock
141
boundary condition Nm = RpNPi
(19)
which expresses the reflected wave N n l in terms of the incident wave N p l . Free and fixed end conditions are represented by the limits Rp -► - 1 as Z2/Zt -► 0, Rp -► 1 as Z2/Z1 -► oo
(20)
which correspond to N = 0 and v = 0 at the rod end, respectively. More general boundary conditions can be formulated by prescribing relations between the normal force and the particle velocity at the rod end, and using equation (5). Generally, there is such a nonlinear relation at the end of the drill bit which corresponds to the force versus penetration relationship for the combination of cutter(s) and rock.
6.2.5 Initial Conditions and Impact Previously we have assumed the préexistence of waves Np and Nn in the rods considered. We shall now see that these waves can be determined from initial conditions. Consider a system of rods like that of a percussive drill. The rods may be separate, or they may be connected in different ways. The geometry and material of the rods are represented by the characteristic impedance function Z(x) which is assumed to be piece-wise constant. The initial distributions of normal force and particle velocity JV(x,0) = N0{x),
ν(χ,0) = v0(x)
(21)
are also presumed to be piece-wise constant. Substitution of equation (5) into equation (21), and solution of the resulting system of equations for Np(x) and Nn(x), gives Np(x) = (l/2)[JV0(x) - Z(x)v0(x)l
Nn(x) = (1/2)[N 0 W + Ζ(χ)υ0{χ)1
(22)
These relations determine the initial distribution of waves in the system. The subsequent evolution of waves can be determined using equations (16), (19), and similar relations. As an illustration we consider the impact between a hammer with length LH and impact velocity KH, and a long rod which is initially at rest. The two bodies have the same characteristic impedance Z and wave speed c, and they are initially free from normal force. This situation resembles that which may prevail in a hydraulic drifter. In this case equation (22) gives Np = — (l/2)ZVH and Nn = (1/2)ZKH in the hammer, and Np = Nn = 0 in the rod. After reflection of the wave Nn with reflection coefficient Rn= — 1 at the free left end of the hammer, a compressive wave JV = Np with normal force N = — (1/2)ZFH, particle velocity v = (1/2) VH and length 2LH propagates into the rod. By the use of equations (11) and (15) it can be verified that, in this case, all the initial momentum and energy of the hammer are carried into the rod by the wave generated through impact. If the characteristic impedance is Z = 4 0 k N s m - 1 and the impact velocity is VH= 10 m s - 1 , as in a typical percussive drill, the normal force and particle velocity associated with the wave are N = —200 kN and t? = 5ms' 1 , respectively. If the length of the hammer is LH = 0.5 m, then the length of the wave is 1.0 m. This is normally much more than the transverse dimensions of the drill string and justifies, as stated above, the use of 1-D theory. If, finally, the wave speed is c = 5.2 kms" 1 , as for steel, the wave passes a given cross section of the rod in 190 μβ. This illustrates that in percussive drilling, the large forces generated commonly act during short intervals of time.
6.3 MODELING PERCUSSIVE DRILLING 6.3.1 Components and Members Real percussive drills are constituted by components like hammers, adapters, rods, coupling sleeves, bits, etc. as shown in Figure 1(a). Two threaded rod ends which are kept together by a coupling sleeve form a joint. Similarly, model percussive drills can be considered to be made up of members which represent components, or parts or combinations of these. This is illustrated in Figure 1(b). Three different types of members are shown, viz. segment members, interface members and end members.
142
Mechanized Excavation (a)
Joint
00= *=tHËI3 Rod
Hammer
CO=M
Adapter
K
Bit
Rock
Coupling sleeve
(b)
Cutter/rock
Feed
Joint
Figure 1 (a) Real percussive drill; (b) model percussive drill
The segment members depicted are of the types hammer, swell and rod. They are built up of 1-D rod segments, each with constant characteristic impedance Z and all with the same transit time h for elastic waves. In simulations this transit time is taken as the time step. If the material is the same everywhere, all segments have the same length D = ch. This is commonly the case and will be assumed in what follows, although this is not a necessary restriction. The interface members shown are of the types joint and feed. Another interface member is labelled gap. They are located at the interfaces between different combinations of rod and swell members. Finally, the only end member illustrated is of the type cutter/rock. Other end members are designated free, fix and extended. They are located at the right end of the drill and represent the conditions below the drill bit. Next the properties of the different members will be defined, and rules will be given on how to combine the members in order to build up a model of an existing or a conceivable percussive drill. 6.3.2 Segment Members A hammer member (Figure 2a) consists of LH 1-D segments with characteristic impedances ZX,Z2, . . . , Z LH . The hammer is initially free from normal force and has impact velocity VH. Therefore, the initial state in the segments of the hammer is given by equation (22) with N0(x) = 0, i?o(x) = VH, and with Z(x) equal to the characteristic impedance of the appropriate segment. A hammer is necessary at the left end of a percussive rock drill. To its right there should be a swell, a rod, a cutter/rock or a fix member. A swell member (Figure 2b) consists of L$ 1-D segments with characteristic impedances Z 1? Z 2 , . . ., ZLs. Initially the swell is subjected to a constant normal force Ns, and it has constant velocity Vs. Therefore, the initial state in the segments of the swell is given by equation (22) with N0(x) = Ns,v0(x) = Vs, and with Z(x) equal to the characteristic impedance of the appropriate
Computer Modeling and Simulation of Percussive Drilling of Rock
143
(a) -►1/H
z
'
Zz
1
2
(b) ^"s
/Ve
Γ1
Zts
^2
I
1
■/V s
Ls
2
(0
Ή
/VR
I
Figure 2
2
I I ΙΚΓΤΤ
/VR
^.R
Segment members, (a) Hammer; (b) swell; (c) rod
segment. The initial normal force Ns originates from the feed force (s), and it is determined from equilibrium. The initial velocity Vs originates from previous blows. To the left of a swell there should be a hammer, a swell, a rod, a joint, a feed or a gap member. To the right there should be a swell, a rod, a joint, a feed, a gap, a cutter/rock, a free, a fix or an extended member. A rod member (Figure 2c) is a special case of a swell. It consists of LR 1-D segments, all with the same characteristic impedance ZR = Z. Initially the rod is subjected to a constant normal force JVR, and it has constant velocity VK. The rules for adjacent members are the same as for a swell. The total number of segments used to represent a percussive drill is L = LH + £ L s + Σ
^R
(23)
Let these segments have characteristic impedances Z (1) , Z ( 2 ) ,. . . , Z(L). Their initial state is given by the constant-amplitude waves JVP(1), iV p(2) ,. . . , NP(L) and iVn(1), iV n(2) ,. . . , Nn(L), determined from the initial conditions as described. From this state at t = 0 the states at t = h, 2Λ,. . . are to be determined. Consider now two adjacent segments 4(i)' a n d V + 1)' w ^h the interface '(i)'· These segments may belong to one or two members. Let the states at the previous time t — h and the present time t be labelled 'old' and 'new', respectively. Then, equations (16) to (18) give Npnew(i+1) — 7p(i)JVpold(0 + ^n(i)^nold(i!+ 1) Nnnew(i) = Kp(i)./Vpold(i) + T^iO^noldii + 1)
(24)
with ßp(i) = [Ζ(,·+ΐ) — Ζ ( ί ) ] / [ Ζ ( ί ) H- Z ( i + 1 ) ] ,
and
^p(i) — 2 Z ( i + 1 ) / [ Z ( i ) + Z ( j + 1 ) ] ,
Ä n(i) = [ Ζ ( ί ) — Z ( i + 1 ) ] / [ Z ( | ) + Z ( i + 1 ) ] r n ( i ) — 2 Z ( i ) / [ Z ( i ) -h Z ( i + 1 ) ]
(25)
(26)
For a rod member equation (24) yields the simple relations Npnew(i+1) — N po i d(l ·),
N n n e w ( f ) — Nnoid(i + i)
(27)
144
Mechanized Excavation
At the left end of the hammer member equations (24) should be replaced by the boundary condition Npnew(l) =
— Nnold(l)
(28)
for a free end. At the right end the conditions are the same as at a gap. Normal force and particle velocity in the segment '(i)' at time t are obtained from equation (5) which gives Nnew(i) =
iVpnew(l) + Nnnew(l·),
l\, e w(i) =
[ 1 / Z ( , ) ] [ ~ ^ p n e w ( i ) + ^nnew(i)]
(29)
The momentum and energy of a number of segments at time t can be obtained from equations (11) and (15) as P
=
hZl-Npn*Mi)
+ WnnewO·)]
(30)
I
and
W = /iZCV^olCiVp^wo·) + Nn2new(0]
(31)
i
respectively. Thus, for example, the initial momentum PH and the kinetic impact energy WH of the hammer can be determined from these relations by summing the contributions from the segments of the hammer at time t = 0.
6.3.3 Interface Members An interface member is located at an interface '(*')' between segments on the left and right sides with characteristic impedances Z (0 and Z(i +1}, respectively. These segments should belong to swells or rods. A joint member (Figure 3a) is constituted by a mass and an aggregate of two springs and a friction element. This aggregate provides the coupling of the mass to the interface (i). The segments on each side commonly have the same characteristic impedances Z (i) = Ζ(ί + υ = Z. The mass M represents the coupling sleeve, while the stiffnesses fcE and fcF and the force QF represent the coupling of the sleeve to the drill rods via the threads. The stiffness kE is associated with local elastic deformations of the threads, while the stiffness kF is also connected with slip. The force QF represents the friction between the threads. These three parameters depend on the prestrain in the sleeve and can be determined from laboratory tests. The parameter s represents the initial state of the joint, which depends on the loading history [24]. The waves leaving the joint can be expressed in terms of the incident waves by the approximate relations [25] Np„ew(«+1) =
Npoid(l) -
(l/2)ßav,
^nnew(i) =
N«old(i + 1) + ( l / 2 ) ß a v
(32)
where ß av is the average value of the coupling force Q(t) in the time interval (t — h,t). This force is obtained in terms of the incident waves by solving the nonlinear equation of motion of the mass M. It should be noticed that the relations (32) preserve the piece-wise constancy of the waves. The energy absorbed by the joint from the waves can be determined as the sum of net energies carried into the interface (i) by the waves on each side. Thus, equation (13) gives WD = (Ä/ZJECAftidci, + JV20ld(/ + 1) - iV n 2 new(0 - N 2 n e w ( / + 1)]
(33)
t
where the positive and negative terms represent energy transported to and away from the interface, respectively. Generally, the energy WO represents kinetic energy and elastic strain energy as well as dissipated energy. If the joint becomes quiescent after a certain time, however, this energy corresponds to the dissipated energy as illustrated by the shaded area in Figure 3a. A feed member (Figure 3b) consists of a constant force F{ which acts in the positive direction on the interface \i)\ If there is no gap between the feed member and the rock there results an initial normal force N0(x) = - F f in this part of the drill. Otherwise, there is no initial normal force in the drill due to the feed, which means that the force F{ is considered to be applied at time t = 0. If a drill contains several feed members, the initial normal force distributions due to all feed forces F{ are superimposed. Consider now an infinitely thin element which includes a feed member. This element is acted on by the feed force and by normal forces on each side. As the element has no inertia, the forces acting on it
Computer Modeling and Simulation of Percussive Drilling of Rock
145
(a) Q
Ä/pnew(/+l)
Λ/pold ( / )
I (/)ΐ/+D! (S+DÖF '•nnewi/')
/»noldi/'+D
LB-° Γ A/VH
o -*-
(b)
Λ ^
M
(c)
Wpold(/)
Λ/pnew {/ + D
Wpold(/')
Wpnew(/+I)
Π F
<
I-ZÎ/T
«2(/+l
&
1 I ^/nold(/+l)
Wnnewi/)
/V nc
/Vnn
Figure 3 Interface members, (a) Joint; (b) feed; (c) gap
must be in equilibrium. Furthermore, the particle velocity must be continuous across the element. These two requirements, and the use of equation (5), yield the relations Npnew(i+1) — ^ρθ') ^pold(i ) + ^n(i)^nold(i + 1) ~~ (1/2) T^p^f Nnnew(i) =
-Rp(i)^pold(i) +
^n(i)^nold(i + 1) + (1/2) 7^Ff
(34)
which differ from equations (24) by the last term of each relation. A gap member (Figure 3c) consists of a space of initial width d between the segments '(i)' and '(/ + 1)\ If d = 0 the gap is closed and if d > 0 it is open. As a result of the wave motion the gap may close and open repeatedly. If the gap is closed at time t — h and it is assumed that the gap remains closed during the time interval (t — h,t), the reflection and transmission of waves obey equations (24). Then, the normal force acting at the interface '(i)' during this interval must be non tensile, that is Npold(i) + -/Vnnew(i) ^
0
(35)
Otherwise the assumption that the gap remains closed must be wrong. Instead, the gap opens at time t — h and equation (24) should be replaced by the boundary conditions at the two free ends formed. These boundary conditions, which correspond to equation (19) and the first part of equation (20), can also be established from equation (24) by putting the reflection coefficients Rpii) and Rn{i) equal to —1 and the transmission coefficients Γρ(ί) and Tn(i) equal to zero. If the gap is open, or if it opens, at the time t — h, and it is assumed that the gap remains open during the interval (t — h,t\ the ends of the gap undergo displacements during this interval which can be established using the second part of equation (5). By adding the change in gap width for each time step to the previous width of the gap, it can be detected when the gap closes. If it is found that the width is positive (gap open) at time t — h but negative (overlap) at time i, then the gap must close
146
Mechanized Excavation
between these two instants of time. As only the times t = 0, h, 2ft,. . . can be handled, however, the gap width is put equal to zero at time t — h if the gap width at this time is smaller than the overlap at time t. Otherwise the gap width is put equal to zero at time t. 6.3.4 End Members An end member must be located at the right end of segment '(L)' with characteristic impedance Z(L). This segment should belong to a swell or a rod. A cutter/rock member (Figure 4a) consists of a force F which depends on the history of penetration p. This force versus penetration relationship is characteristic for a given combination of cutter(s) and rock, and it can be determined from laboratory tests under conditions similar to those which prevail in percussive drilling [29]. Commonly, this relationship is taken to be linear during loading as well as unloading. Then it is characterized by the initial gap pg between cutter(s) and rock, the threshold force F th , below which the rock undergoes elastic deformations only, and by the penetration resistances (force per unit of penetration) kx = k and k2. Thus, in this case, the interaction of cutter(s) and rock is determined by the four parameters pg, F th , k and fc2. An algorithm for determining the reflected wave Nnnew(L) in terms of the incident wave Npold(L), and other quantities, should preserve the piecewise constancy of the waves. Such an algorithm is presented in [19]. With minor modifications this algorithm can be used also for more general force versus penetration relationships which are piecewise linear during loading and unloading. The work performed on the rock is WK = J Fdp, Le. the area enclosed by the force versus penetration curve, as illustrated in Figure 4(a). Commonly, the efficiency of the percussive drilling process during one impact cycle is defined as the ratio of this work to the kinetic impact energy WH supplied by the hammer, i.e. (36) η = WR/WH If the kinetic impact energy of the hammer is the only energy supplied, this quantity is less than unity. This is the case if there are no feed forces and no initial velocities other than that of the hammer. Otherwise, it may be suitable to use a more general definition which includes other supplied energies. For the applications to be considered here, however, equation (36) is appropriate. The end members free, fix, and extended (Figures 4b-d) correspond to the boundary condition Nnnew(L)
—
(37)
Rp(L)^pold(L)
(a) /v,pold(Z.)
F ·1
jf^M
A,
C
/V„n —1
*—
«"
„ ,,,,.*,
, ,../
»
*3=0
(d)
(c)
(b)
NDl
^Î>old (L)
Λ/oc
1
V
\ZlL
Nnr
/Vn,
.
i
y y 1/
Λ/ηηβν»(Ζ.) βΟ
Figure 4 End members, (a) Cutter/rock; (b) free; (c)fix;(d) extended
Computer Modeling and Simulation of Percussive Drilling of Rock
147
with Rp{L) =—1,1 and 0, respectively. The first case represents the total absence of rock in front of the bit and corresponds to vanishing normal force, N = 0. The second case represents a bit which has got stuck and corresponds to vanishing particle velocity, v = 0. Finally, the third case represents matched conditions at the end of the bit. 6.3.5 Dimensionless Representation It is convenient to represent parameters and variables by dimensionless numbers. Thus, for example, the length of a segment, i.e. L* = D, is a useful unit for length. Another unit for length is the double length of the hammer, i.e. L** = 2LH, where prime (') denotes a dimensional quantity. As shown in Section 6.2.5 this is the length of the wave generated in a uniform rod through impact by a cylindrical hammer if the characteristic impedances and the wave speeds are the same for the hammer and the rod. The unit L*, and units based on it, are convenient to use in the numerical analysis. With regard to interpretation of results, however, the unit L**, and associated units, have the advantage of being independent of the segment length which is a characteristic of the numerical procedure rather than a property of the percussive drill. Two sets of units, associated with L* and L**, are given in Table 1. Where confusion may occur a prime (') has been added, as above, to indicate that a quantity is dimensional. When thefirstset of units (*) is used, k =fc'/fc*is the dimensionless and k! is the dimensional penetration resistance. For other quantities there are corresponding relations. With this convention, the equations in Sections 6.3.2-4 can generally be expressed in dimensionless form just by putting the time step h equal to unity. When the second set of units (**) is used, Greek letters are used for the dimensionless quantities. Thus, for example, ß = k'/k** is a dimensionless penetration resistance. Hammer, swell and rod members are characterized by their dimensionless lengths L, characteristic impedances Z and initial velocities V defined L = L'/L*,
Z = Z'/Z*,
V =
V'/V*
(38)
respectively, with indices omitted. The unit Z* is the characteristic impedance of a segment with cross-sectional area A* and can be chosen arbitrarily. Generally, it is convenient to take it as a 'typical' characteristic impedance such as that of the cross sections of the uniform part of a drill rod. The unit V* for velocity can also be chosen arbitrarily. Often it is convenient to take it as the impact velocity, which means that the dimensionless impact velocity is unity. A joint member is represented by the dimensionless mass μ and the coupling parameters κΕ and KF and χ defined (39) μ = M'/M**, κΕ = k'E/k**, KF = k'F/k**, χ = Q'F/F* The initial state of the joint is given by the parameter s, which is dimensionless by its definition. A feed member is represented by the dimensionless feed force Φτ = F{/F*
(40)
A gap member is defined by its dimensionless initial width (index omitted) δ = d'/u**
(41)
Table 1 Units for Dimensionless Quantities Quantity
Unit*
Unit**
Length Time Characteristic impedance Mass Stiffness, penetration resistance Velocity Penetration, gap Force Stress Energy
L* = D t* = L*/c = h Z * = A*E/c M* = A*L*p k* = A*E/L* V* (arbitrary) u* = V*t* F* = Z*V* σ* = F*/A* = EV*/c W* = F *u*
L** = 2L'H t** = L**/c M** = A*L**p k** = A*E/L** u** = V*t**
Mechanized Excavation
148
Finally, the cutter/rock member is characterized by the dimensionless initial gap <5, threshold force φ, penetration resistance ß and unloading parameter y defined by δ = p'Ju**, φ = FUF*, β = /c'//c**, y = k'lk'2
(42)
6.4 SIMULATION AND VALIDATION 6.4.1 Features of Simulation Programs A series of programs for computer simulation of percussive drilling of rock have been developed, tested and used [18-22, 24, 25] during the period 1980-1990 by the author and his coworkers at Luleâ University of Technology. The programs have been written in different programming languages for use with different personal computers, and they utilize, to different extents, the concepts introduced in Section 6.3. The latest simulation program is written in Modula-2 for IBM personal computers. It makes full use of the concepts presented. Thus, for example, percussive drills are built up in aflexibleway from segment members, interface members and end members. In this context, use is made of an editor which permits operations like insert, delete, copy, paste, etc. on members or segments. Strain gauges can be attached to segments where it is desirable to record the history of normal force or stress. Data for drills, or parts of drills, can be stored for subsequent simulations. Computer graphics are used extensively. Thus, for example, the distribution of normal force or stress can be displayed for each time step of simulation. After simulation the envelopes of normal force or stress, and the history of normal force or stress at the positions of the strain gauges, can be displayed. Furthermore, the efficiency of the drilling process can be presented as a function of desired parameters. Next, some results of computer simulations and of comparisons of these with experimental results will be presented. 6.4.2 Efficiency of a Percussive Process Figure 5(a) shows the experimental set-up used for the study of a percussive process [22] which resembles down-the-hole drilling. A cylindrical steel hammer with length L'H = 1000 mm was accelerated to a velocity V^ of approximately 2 m s"* by means of an air gun. The hammer impacted on a cylindrical steel bit which was terminated with a small wedge of hardened steel at the other end. The length L'B of the bit, excluding the wedge, was varied fro 1400 to 500 mm. The diameter of the hammer and the bit was 32 mm. In front of the bit there was a heavy block of concrete. The initial gap /?g between the wedge and a plane undamaged part of the concrete surface was varied from 0 to 1.6 mm in 0.1 mm steps. Two pairs of diametrically opposite strain gauges attached to the bit served to determine force F' versus penetration p' in each test. From this relationship the work IVi performed on the rock was determined. In addition, the dimensionless cutter/rock parameters β and y, defined in equation (42), were identified. The parameter φ was taken to be zero. The efficiency η of the percussive process was determined for each individual test according to equation (36). The system was modeled by a hammer member (LH = 40, ZH = 1, KH = 1), a rod member (LR variable, ZR = 1, VR = 0) and a cutter/rock member (φ = 0; δ, β, y variable). The efficiency η, based on simulation, was determined for each individual test by using the measured value for the initial gap δ, the values of the cutter/rock parameters β and y obtained through identification, and the actual value of the dimensionless bit length λ = L%/LH. Figure 5(b) shows results from the experimental tests and from the corresponding simulations for the efficiency η versus the bit length λ for the initial gap δ = 0. The points in the diagram represent average values. It can be seen that there is good agreement between results of simulations and experimental tests. The agreement is generally better for longer bits, λ > 1, than for shorter ones, λ < 1. In the latter case the percussive process is more complex due to the increased predisposition to repeated separations and contacts between the hammer and the bit. More extensive results are given in [22]. 6.4.3 Down-the-hole Drilling A significant difference between the percussive system studied in Section 6.4.2 and a downthe-hole drill is that the bit of the latter consists of a neck and a head. The head has larger characteristic impedance than the neck and determines the diameter of the hole.
Computer Modeling and Simulation of Percussive Drilling of Rock
149
(α)
(b) o Experiment • Simulation
1.2
1.3
1.4
Figure 5 (a) Experimental set-up for study of a percussive process; (b) comparison of results from experimental tests and simulations; average efficiency η versus bit length λ for initial gap δ = 0 (after Karlsson et al. [22])
We now consider the idealized down-the-hole percussive drill studied in [20]. It is characterized by the lengths L'H of the hammer, LB of the bit and L'h of the head, and by the characteristic impedances Z'H = Z'of the hammer, Z„ = Z' of the neck and Zi of the head. The drill was modeled by a hammer member (LH = 20, ZH = 1, VH = 1), a swell member (L s = LB = Ln 4- L h ; Ln variable, Zn = 1; Lh variable, Z h > 1 variable; Vs = 0) and a cutter/rock member (φ = 0; <5, jS, 7 variable). The efficiency η, defined by equation (36), was determined by simulation as a function of the dimensionless quantities λ = LB/LH9 Xh = L h /L H , Z h , <5, /? and y. For a given material the first three of these parameters represent the geometrical design of the drill. Figure 6 shows results for the efficiency η versus the bit length λ for different values of the penetration resistance β and the bit head impedance Z h . Figure 6(a) corresponds to the system with a uniform bit studied in Section 6.4.2. Some effects of nonuniformity of the bit can be observed by comparing Figures 6(a) and 6(b): for soft (β = 0.1) and medium hard (β = 1) rock, the peak in efficiency at λ = 1 for the uniform bit is moved to the left and its height is slightly decreased. Also, the width of the peak in efficiency increases significantly due to the nonuniformity of the bit. For hard rock (/? = 10), the efficiency is practically independent of λ both for uniform and nonuniform bits.
150
Mechanized Excavation (α)
(b)
Figure 6 Efficiency η versus bit length λ for soft (β = 0.1), medium hard (β = 1) and hard (β = 10) rock. Initial gap δ = 0 and unloading parameter y = 0.1. (a) Uniform; (b) nonuniform bit (after Lundberg and Karlsson [20])
The effect of the nonuniformity, in this case, is to raise the efficiency level significantly. Thus, the efficiency of the nonuniform bit is 0.388, whereas that of the uniform bit is 0.180. More extensive results are given in [20]. 6.4.4 Energy Dissipation at a Joint Figure 7(a) shows the experimental set-up for the study of dissipation in drill rod joints [23]. A cylindrical steel hammer with length LH = 300, 600 and 900 mm was accelerated to a velocity Vli by means of an air gun, and it impacted on a steel rod-joint-rod aggregate. The hammers and rods were made from standard Sandvik drill rods with external diameter 32 mm and thread R32. Therefore, as expected from the theory presented in Section 6.2.5, an approximately rectangular incident wave with length 2L'n and compressive strain amplitude (1/2) V^/c was generated in the input rod. The length of the hammer and the impact velocity were varied in order to modify the length and the compressive strain amplitude, respectively, of the incident wave. The joint to be considered here used a Sandvik Rock Tools' Coupling Sleeve 7993-3644 with length 150 mm and external diameter 44 mm. A pair of strain gauges at the mid-section of the coupling sleeve was used to measure the prestrain ερ due to the joint preload. One pair of strain gauges on the input rod and one on the output rod were used to measure the strains BU eR and ετ associated with the incident, reflected and transmitted waves, respectively. The dissipated energy WD was determined as the deficit of energy in the reflected and transmitted waves relative to the energy W[ of the incident wave. The formulae for determining energies from the recorded strains, given in [23], can be derived from equation (13). The accuracy in the evaluated relative energy dissipation W'O/W[ was much improved by requiring that the total momentum of the reflected and transmitted waves be equal to that of the incident wave. The relation which expresses this condition, given in [23], can be derived from equation (9).
Computer Modeling and Simulation of Percussive Drilling of Rock
151
(α)
Damper
Slide bearing
(b) 0.20
r-
l^(ms"')
Figure 7 (a) Experimental set-up for study of dissipation in drill rod joints; (b) comparison of results from experimental tests and simulations: relative energy dissipation W'OjW[ versus impact velocity V'u and hammer length L^ for joint prestrain ερ = 120 μ (after Beccu and Lundberg [23] and Lundberg et ai [24])
The system was modeled by two semi-infinite rod members (ZR = 1) with an intermediate joint member. In the input rod member there was initially a rectangular incident wave corresponding to a uniform hammer member (LH = 250, ZH = 1, VH = 1). The parameters μ, κΕ, KF and χ for the joint, defined in equation (39), were determined and the impact tests were simulated [24]. Figure 7(b) shows results from the impact tests and from the corresponding simulations for the relative energy dissipation W^/W[ versus the impact velocity V^ for different hammer lengths L'H. It can be seen that there is good agreement between results of simulation and impact tests, and that the relative energy dissipation decreases with increasing hammer length. More extensive results are given in [24]. 6.4.5 Extension Drilling Wefinallyconsider extension drilling with one or several joints as illustrated in Figure 8(a) [25]. The system consists of a hammer, an adapter, Nj joints and rods, and the rock. The hammers have lengths L'H = 300, 600 and 900 mm as in the experimental tests of Section 6.4.4. The impact velocity
152
Mechanized Excavation (a)
LU
VH
Z'
Z'
M
(b) 10
0.9
r - ^ 0 . 5 (medium rock) «p * 120/*(medium) a o v \- «p * 3 6 0 / * (high) ■·▼
0.8 0.7
4 (medium)
0.6 V
0.5 0.4 0.3 0.2 0.1 0
Figure 8 (a) Extension drilling; (b) efficiency η versus number of joints JVj, hammer-to-joint length ratio λΗ and joint prestrain ερ for medium hard rock (β = 0.5) (after Beccu and Lundberg [25])
is VH = 10 m s" 1 . The adapter is uniform with length L'x = 450 mm. The coupling sleeves are the same as in the experimental tests of Section 6.4.4 with length LJ = 150 mm and mass M ' = 0.940 kg. The extension rods are Sandvik Rock Tools' model 7853-3312-20 with length LR = 1220 mm. The bit is Sandvik Rock Tools' button bit 7733-6657-40 with length L'B = 120 mm. The characteristic impedance Z' is the same for the hammer, the adapter and the extension rods. The drill was modeled by a hammer member (LH = 20,40 and 60, ZH = 1, KH = 1), a rod member for the adapter (LR = 30, ZR = 1, VK = 0), a rod member for each extension rod (LR = 81, ZR = 1, VR = 0), a joint member for each joint (χ = 0; μ, κΕ, KF variable; s = 0), a swell member for the bit (Ls = 8, Ζχ - Z 6 = 2.47, ΖΊ-Ζ8 = 3.87, Vs = 0) and a cutter/rock member (δ = 0, φ = 0, y = 0.1; β variable). The efficiency η, defined by equation (36), was determined by simulation as a function of the number of joints Nj, the hammer-to-joint length ratio λΗ = L'H/L'j, the joint prestrain ερ, and the dimensionless quantity β° = 2Ljk'/AE9 which represents the hardness of the rock. The dimensionless penetration resistance /?, defined in equation (42), can be expressed in terms of β° and λΗ as ß = β°λγΐ' Similarly, the joint parameters μ, κΕ and KF, defined in equation (39), can be expressed in terms of AH. Thus, μ is inversely proportional to λΗ, while κΕ and KF are proportional to λΗ. Figure 8(b) shows the efficiency η versus the number of joints N} for different values of the hammer-to-joint length ratio λΗ, for different joint prestrains ερ, and for medium hard rock, β° = 0.5. It can be seen that the efficiency decreases slowly with the number of joints, and that it is the highest for the medium hammer. The joint prestrain has little influence on efficiency in the range studied. More extensive results are given in [25].
Computer Modeling and Simulation of Percussive Drilling of Rock 6.5
153
SUMMARY AND CONCLUSIONS
In this chapter the subject of computer modeling and simulation of percussive drilling of rock was introduced. The subject has its origin in research and development work performed over a long period of time. However, the conditions for work in the area improved significantly at the end of the 1970s with the advent of personal computers and easily available computer graphics. As percussive drills are largely built up of rods which are subjected to axial impact at moderate velocities, elastic wave propagation in such rods, and associated phenomena, are fundamental for the understanding of the mechanics of percussive drilling processes. Therefore, relevant parts of the theory of elastic waves in rods were presented in Section 6.2. As the designs of percussive drills are generally such that 1-D theory is adequate for engineering purposes, the presentation was limited to such a theory. Concepts like wave speed, characteristic impedance, normal force, particle velocity, momentum and energy, and phenomena like propagation, dispersion, reflection, transmission and impact, and their interrelations, were introduced. Numerical examples were given, and it was demonstrated that in percussive drilling large forces commonly act during short intervals of time. These large forces partially explain the success of percussive drilling methods. In Section 6.3 it was pointed out that models of existing or conceivable percussive drills can be considered to be built up of members which may or may not coincide with components of the real drills. Different types of such members are segment members, interface members and end members. The first type comprises hammer, swell and rod members. The second embraces feed, joint and gap members. The third, finally, includes cutter/rock, free, fix and extended. Detailed mathematical models were presented for several members of each type. For the joint and cutter/rock members, which are more complex than the others, references were given to the literature for complete descriptions. At the end of the section, it was shown how parameters and variables associated with the different members can be made dimensionless in different ways. In Section 6.4, finally, some features of simulation programs developed, tested and used by the author and his coworkers at Luleâ University of Technology were presented. With the latest version of these simulation programs, which is written in Modula-2 for IBM personal computers, percussive rock drills are built up in a flexible way from the three different types of members described in Section 6.3. Examples of results from simulation studies related to down-the-hole drilling and extension drilling were given. In two of these, comparisons were made with corresponding results of experimental tests. These comparisons validate the models used. It was seen from the examples that the effeciency of percussive drilling methods is relatively high. This provides another partial explanation for the successful use of percussive drilling methods.
ACKNOWLEDGEMENTS The author is in great debt to coworkers within Luleâ University of Technology, AB Sandvik Rock Tools and Tamrock Oy for their support and contributions. Mr Arne Nilsson, in parallel with his studies of computer science, wrote several versions of simulation programs with great skill and inventiveness. Sincere thanks are also due to Sandvik, Tamrock and the Swedish Board of Technical Development for their financial support, and to the University of Bordeaux 1 for providing an inspiring environment where this chapter was written. 6.6 REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.
Kolsky H. Stress Waves in Solids. Dover, New York (1963). Achenbach J. D. Wave Propagation in Elastic Solids. North-Holland, Amsterdam (1973). Donnell L. H. Longitudinal wave transmission and impact. Trans. Am. Soc. Mech. Eng. 52, 153-167 (1930). Dahl H. O. Hammare och borr. Jernkontorets Annaler 5, 205-219 (1932). Takaoka S. and Hayamizu H. Studies on the percussive deep hole drilling of rocks. J. Min. Metall. Inst. Jpn. 72,497-502 (1956). Arndt F.-K. Untersuchung über die Energieübertragung beim Schlagforgang im Blickfeld des schlagenden Bohrens. Ford. Forschg. Geb. Bohr-u. Schiesstechn 7, 1-132 (1959). Arndt F.-K. Der Schlagablauf in Kolben und Stange beim Schlagenden Bohren. Glückauf 96, 1516-1524 (1960). Fischer H. C. On longitudinal impact MIL Appl. Sei. Res. A8, 105-139, 278-308; A9, 9-42 (1959). De Juhasz K. J. Graphical analysis of impact of elastic bars. J. Appl. Mech. 9, A122-A128 (1942). Fairhurst C. Wave mechanics of percussive drilling. Mine Quarry Eng. 27, 122-130, 169-178, 327-328 (1961). Simon R. Digital machine computations of the stress waves produced by striker impacts in percussion drilling machines. In Rock Mechanics (Edited by C. Fairhurst), pp. 137-154. Pergamon Press, Oxford (1962).
154
Mechanized Excavation
12. Simon R. Transfer of the stress wave energy in the drill steel of a percussive drill to the rock. Int. J. Rock Mech. Min. Sei. 1,397-411(1964). 13. Dutta P. K. The determination of stress wave forms produced by percussive drill pistons of various geometrical designs. Int. J. Rock Mech. Min. Sei. 5, 501-518 (1968). 14. Fu C. C. and Paul B. Energy transfer through chains of impacting rods. Int. J. Numer. Methods Eng. 2,363-385 (1970). 15. Hustrulid W. A. and Fairhurst C. A Theoretical and experimental study of percussive drilling of rock. Int. J. Rock Mech. Min. Sei. 8, 311-333, 335-356 (1971); 9, 417-429, 431-449 (1972). 16. Lundberg B. Energy transfer in percussive rock destruction I—III. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 10, 381-399, 401-419, 421-435 (1973). 17. Carlvik I. The generation of bending vibrations in drill rods. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 18,167-172 (1981). 18. Lundberg B. Microcomputer simulation of stress wave energy transfer to rock in percussive drilling. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 19, 229-239 (1982). 19. Lundberg B. Microcomputer simulation of percussive drilling. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 22, 237-249 (1985). 20. Lundberg B. and Karisson L. G. Influence of geometrical design on the efficiency of a simple down-the-hole percussive drill. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 23, 281-287 (1986). 21. Lundberg B. Efficiency of percussive drilling with extension rods. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 24, 213-222 (1987). 22. Karisson L. G., Lundberg B. and Sundin K. G. Experimental study of a percussive process for rock fragmentation. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 26, 45-50 (1989). 23. Beccu R. and Lundberg B. Transmission and dissipation of stress wave energy at a percussive drill rod joint. Int. J. Impact Eng. 6, 157-173 (1987). 24. Lundberg B., Beccu R. and Nilsson A. Nonlinear dissipative spring mass model for a percussive drill rod joint of the coupling sleeve type. Int. J. Impact Eng. 8, 303-313 (1989). 25. Beccu R. and Lundberg B. Efficiency of percussive drilling of rock with dissipative joints. Int. J. Impact Eng. 9,277-287 (1990). 26. Nordlund E. The effect of thrust on the performance of percussive rock drills. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 26, 51-59 (1989). 27. Pang S. S. and Goldsmith W. Momentum and energy processes during jackhammer operation. Rock Mech. Rock Eng. 22, 205-229 (1989). 28. Neelamkavil F. Computer Simulation and Modelling. Wiley, Chichester (1987). 29. Carisson J., Sundin K.G. and Lundberg B. A method for determination of in-hole dynamic force-penetration data from two-point strain measurement on a percussive drill rod. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 27, 553-558 (1990).
7 The Mechanics of Rock Cutting ROBERT JOHN FOWELL University of Leeds, UK 7.1
INTRODUCTION
155
7.2
EXCAVATION MACHINES
156
7.3 DRAG PICK CUTTING 7.3.1 Drag Tool Relationships 7.3.1.1 Cutting speed 7.3.1.2 Front rake angle 7.3.1.3 Back clearance angle 7.3.1.4 Spacing 7.3.2 Cutting Head and Drum Design 7.5.3 Water Jet Assisted Drag Tool Cutting
162 163 164 164 165 165 165 166
7.4
166 167 167
CUTTING TOOL MATERIALS AND WEAR
7.4.1 Tungsten Carbide 7.4.2 Wear Mechanisms
169
7.5 DISK CUTTERS 7.5.1 Disk Cutting Relationships 7.5.1.1 Penetration 7.5.1.2 Tire edge wedge angle 7.5.1.3 Disk diameter 7.5.1.4 Speed 7.5.1.5 Spacing 7.6
169 169 169 170 170 171
CUTTABILITY AND MACHINE PERFORMANCE
7.6.1 Abrasivity 7.6.2 Rock Mass Properties
172 172 175
7.7
CONCLUSIONS AND THE FUTURE
175
7.8
REFERENCES
176
7.1
INTRODUCTION
This chapter introduces some practical and theoretical aspects of mechanical rock excavation. There is an established need for the extension of applications for mechanical excavation of rock materials, whether it is for the creation of nuclear waste repositories with no blasting induced fractures, excavation of rock trenches, tunnels in congested urban environments or stope development in the hard rock sector of the mining industry. Cutting tools are used to transfer energy from the excavation machine to the rock. The two principal tools used are the drag pick and the rotary disk cutter. The drag tool, though more efficient on energy grounds, is limited in terms of the strength and abrasivity of the rock it can economically excavate (Figure 1). The disk cutter is an effective mechanical tool principally for rock outside the range considered suitable for the drag tool (Figure 2). To work efficiently it requires high thrust forces to gain and maintain adequate penetration and is limited to tunnel boring machines (TBMs), raise boring and
155
Mechanized Excavation
156
Cutting force Normal force d
a » Rake angle β β Clearance angle d * Depth of cut
Θ» Breakout angle t*= Pick width
Figure 1 Drag pick
Rolling force
P = Penetration
Sideways force
Figure 2 Disk cutter
shaft sinking operations. For very hard rock tungsten carbide buttons are inserted into the periphery of the disk. Though blasting is the natural choice for hard rock mineral production, the advantages of mechanical excavation are becoming increasingly attractive for many projects for the following reasons. (i) It is economically advantageous (ii) It results in improved safety (iii) Its ease of automation (iv) Its accuracy of finished excavation dimensions (v) Excavation walls remain undamaged (vi) The product size can be handled by conveyors (vii) It has applications where there are limits on the level of vibrations allowed. In this chapter the types of machine employing drag picks and disk cutters are briefly reviewed, followed by a more detailed look at the mechanics of drag tool and disk cutter operations. Cutting tool materials and the influence of wear on performance are covered, along with a review of cuttability assessment and future developments. 7.2
EXCAVATION MACHINES
Historically, mining coal has provided the proving ground for the development of drag tool equipped excavation machines. This extended from coal winning to strata associated with the coal measures and moved into weaker sedimentary formations. Full face TBMs employing disk cutters for tunnels gained acceptance in the 1960s, running parallel with the developments in raise boring techniques.
157
The Mechanics of Rock Cutting
Stack, in her Handbook of Mining and Tunnelling Machinery [1], has comprehensively reviewed the development of many forms of rock cutting machine. Table 1 provides an overview of the main types of excavation machine. Drag tools are used exclusively for mechanized coal production. The two principal methods used for working coal seams are longwall mining and room and pillar mining. In longwall mining, the excavation machine travels along the coalface guided by an armoured flexible chain conveyor. The coal plough is the simplest form of mechanization and is basically an array of large drag tools which peel slices of coal oflf the face as it is pulled along by a chain (Figure 3). This machine requires a strong floor, weak coal and a good roof parting, and thus has limited application. The trepanner is another longwall machine which cuts a cylindrical core of coal parallel to the face line, this is broken up by the trepan head (Figure 4). Turrets laced with picks produce a parallel flat Table 1 Main Types of Excavation Machines Rock excavation machines
Drag tool
Disk cutter Soft materials only
Soft rock mineral production (coal)
Tunneling and development
Longwall machines
Coal plough
Continuous miner
I
Trepanner
Circular fullface tunnel boring machines (TBM)
Axial boring machines
I
Robbins mobile miner Roadheader
Shearer Axial Pickmat
Raise boring machines
Transverse
Hardhead
Figure 3 Coal plough
Shaft sinking machines
158
Mechanized Excavation
Figure 4
Trepanner
floor and roof conditions. This machine is being phased out in preference to the Anderton shearer, which is now the universally used machine for longwall mining (Figure 5). The shearer may take many configurations with one or two horizontal axis cylindrical drums which are capable of being raised relative to the main body of the machine. These machines are capable of being automatically steered within the coal seam. The placing of tools or lacing of the shearer drum will be considered later. Room and pillar mining consists of driving two sets of tunnels, usually at right angles, leaving pillars of coal or mineral to support the roof, which may be removed at a later stage. Continuous Miner machines are used for this method of mining. They are equipped with either a horizontal cylindrical cutting drum at right angles to the center line of the track mounted body or a number of parallel chains equipped with drag picks mounted on the links to produce a pick mat, which cuts and conveys the mineral onto a conveyor running through the body of the machine. The former type of Continuous Miner is known as a hardhead machine and is capable of cutting harder mineral than the pick mat machine (Figure 6).
Figure 5
Ranging drum shearer
The Mechanics of Rock Cutting
159
Figure 6 Continuous Miner
For mechanized mining of salt, potash and coal, a number of machines have been developed that have several rotary cutting heads mounted with their axis parallel with the center line of the machine. A drag tool equipped chain produces a flat roof and floor. The Marietta Borer is a typical example of this type of machine (Figure 7). For development tunnels associated with mining operations and for civil engineering tunnels in predominantly sedimentary formations, the roadheader boom tunneling machine has found acceptance. These machines may be track mounted or within a shield. The cutting head may be mounted axially to the boom (Figure 8) or two heads may be mounted transversely to the boom (Figure 9). The advantage of this type of tunneling machine is that it is not confined to circular cross sections and will excavate tight radius curves and junctions with selective mining capability. Increased production can be obtained by mounting two booms on the machine, or for larger cross sections with shield drivages, multiple booms. As with other machines in this section, most manufacturers
Figure 7 Twin rotor Marrietta Borer
160
Mechanized Excavation
Torque
Slewing force Sumping force Elevating force
Figure 8 Axial boom tunneling machine
Figure 9 Transverse boom tunneling machine
produce different sizes, powers and weights of machine to suit a range of rock characteristics, production rates and excavation dimensions. Disk cutter equipped machines are predominantly circular full-face tunnel boring machines (TBMs) (Figure 10) and raise boring machines (Figure 11) though the Mobile Miner, a new development by the Robbins Company which uses a large horizontal axis wheel with disk cutters mounted on the circumference to traverse the face of the drivage, which produces a rectangular cross section, has created much interest for future hard rock projects (Figure 12). „ Gauge cutter
Thrust cylinders Support erection platform
Gripper pad
Figure 10 Schematic drawing of full face tunnel boring machine (TBM)
The Mechanics of Rock Cutting
161
^//KN/AV/^-A
Figure 11 Raise boring Cutter head
Boom Grippers Cutter motor \ Roof support
Muck apron
Dust vents Conveyor Operators cab \ Dust extractors
\ Crawler Floor jacks
Figure 12 Robbins Mobile Miner (after Boyd [25])
The basic TBM consists of a rotating circular head with arrays of disk cutters arranged to cover the whole face. The head is thrust into the face and the circular grooves produced by the cutters interact to slab off the intervening rock, allowing the head to advance. After a head has advanced one cycle, the gripper pads that bear against the tunnel wall (or a thrust ring acting against the lining) are released and the machine body is advanced to start the next cycle of cutting. TBMs are employed for long tunnel drivages and are capable of consistently high rates provided unforeseen unstable ground or water problems are not encountered. As the machine has evolved features have been developed that have increased the range of applicable conditions. Pressurised chambers behind the head to control the inflow of water, avoiding the need for compressed air working, are being used to advantage. Provision for early support, ease of maintenance and replacement of components and efficient debris disposal have improved greatly since the first
162
Mechanized Excavation
machines were used. Integration and automation of support setting has greatly increased drivage rates. Nelson (Chapter 10 in this volume) reviews TBM performance with Fawcett (Chapter 11 in this volume) discussing the cost of tunneling. Raise boring is the reaming of a predrilled pilot hole to thefinisheddiameter. This is a commonly used technique for ventilation and ore passes in mining and isfindingincreased application in civil projects where there is access to the bottom of the pilot hole. Shafts have also been mechanically sunk using machines similar to TBMs operated vertically, though the major problems associated with this technique are in the area of debris clearance. Mellor has produced a highly recommended review of the 'Mechanics of Cutting and Boring' [2] which deals with the different types of excavation machines' cutting action, whether linear, transverse or axial rotation, or continuous belt action. The kinematics of each type of machine is analyzed with separate consideration of the dynamics and energetics of each machine type.
7.3 DRAG PICK CUTTING These tools are used on all coal and soft rock excavating machines and may take several forms, as illustrated in Figure 13. They comprise an alloy steel body with a tungsten carbide insert cutting tip, though alloy steels are used in weak, nonabrasive applications. The conical tool is the most popular on a worldwide basis, having the advantage that the cutting tool rotates in use, forming a stabilized geometry as the rock being excavated wears away the carbide and parent steel. The radial tool presents the same area of carbide to the rock with each pass of the tool, generating a wear flat that increases the forces several fold as wear progresses. The cutting action of drag tools has been studied in a number of laboratories and Clark [3] provides a review of these investigations. The practical requirements of a drag tool are the forces required to push it through rock at a given depth, the amount of rock excavated and the life of the cutting tool in terms of the number of cubic meters excavated before replacement. The force acting on a cutting tool is constantly changing in magnitude due to the brittle nature of most rock materials. The tool penetrates into the rock until a major chip is formed. Secondary chipping and crushing are also present in the process (Figure 14). Figure 14 illustrates a typical force signal obtained during drag pick cutting at constant depth. For analysis of cutting forces it is usual to resolve the resultant force acting on the tool into three orthogonal components (Figure 15): cutting force, normal or thrust force, and sideways or lateral component forces. Due to the transient nature of the forces, these components can be expressed in a number of ways. For constant depth of cut applications, the mean force component is the average force acting during the cut and is obtained by integrating the force level signal and dividing by time. The mean forces are related to the power of forces required from the machine for rock excavation. Another representation is mean peak force, which is the average of the peak forces for a given cutting condition. Mean peak forces are important for drag tool selection. High peak forces may result in gross fracture damage to the tungsten carbide cutting tip, or damage to machine components.
Retaining clip Radial
Forward attack
Point attack
Figure 13 Forms of drag tool
The Mechanics of Rock Cutting
163
Distance cut
Figure 14 Cutting force diagram
Figure 15 Pick force components on a simple chisel pick
A valuable parameter for mechanical rock cutting is specific energy, which is the energy required to excavate a unit volume of material, usually expressed in MJ m" 3 . Specific energy is an inverse measure of cutting efficiency; high specific energy values indicate low cutting efficiency. When using drag tools, the principal rock failure mechanism is considered to be predominantly tensile for brittle rock material, with shear failure applying to plastic rocks. In practise, both mechanisms are present. Evans [4], Nishimatsu, Roxborough [5] and Deliac have proposed models to describe the mechanisms of rock failure under attack by a drag tool. Evans has presented several modifications to his basic theory to account for blunt and conical tools, as well as spacing between tools. Nishimatsu has reviewed the theoretical aspects of rock cutting in Volume 1, Chapter 26 of this work. Deliac reviews theoretical and practical rules for mechanical excavation in Chapter 8 of this volume. The influence of the common variables in drag tool cutting will now be considered. 7.3.1 Drag Tool Relationships Though there are many cutting tool geometries in use, they all have similar characteristics in terms of their relationships to depth of cut, rake angle, cutting speed and spacing. Figure 16 shows the trends that have been established through laboratory linear testing on flat rock surfaces. Once a cutting tool has penetrated about 5 mm, the forces rise in an approximately linear manner with depth for both the mean and mean peak force components. The mean force is typically a third to a half of the mean peak force components for pristine tools. With the onset of wear, the normal force increases much faster than the cutting force components and may attainfivetimes or higher the pristine normal force level which should be allowed for in the specification of the machine. The product or debris produced for unrelieved cuts rises as a quadratic function, hence the specific energy drops with increasing depth. It is, therefore, most important that drag tools take a reasonable
164
Mechanized Excavation ^
10
E
3
6
9
12
3
Depth of cut (mm)
6
9
12
15
Depth of cut (mm)
Cutting force
H i.o
o
Ξ 3
5 2
H 0.5
10
0
-20
Rake angle0
0
^15
20
Spacing (mm)
0
10
20
30
Rake angle0
Mean cutting force 2 r
-10
Depth
UJ
(0
• 6 9 : 12 15
8
12
16
s/d
Figure 16 Typical drag tool relationships. Effect of (a) cutting depth, (b) rake angle and (c) spacing (after Roxborough [5])
depth of cut for a major part of the cutting cycle, to take advantage of the 'Deep Cut Principle'. Failure to arrange this results in inefficient rubbing cuts which rapidly wear the tools. 7.3.1.1 Cutting speed Cutting speed has been found to have no influence on the magnitude of the force levels recorded provided the wear effect has been discounted. The speed of fracturing through rock is higher than practical cutting speeds used on excavation machines. Wear rate increases with speed and this is an important consideration when cutting abrasive materials. 7.3.1.2 Front rake angle Force components increase with decrease in front rake angle on the tool with highest forces recorded for negative rake angles. Negative rake angles are used in practise to present a strong carbide geometry to the rock. High positive rake angles, though requiring less force to excavate rock,
The Mechanics of Rock Cutting
165
are prone to gross fracture of the carbide insert. High positive rake angle geometries are therefore only used for weak rocks and coal cutting. 73.13
Back clearance angle
The influence of back clearance angle has been found to have no effect on cutting force components providing it has a value of at least 10° and the influence between 5° and 10° is low. 73.1.4 Spacing For practical rock excavation spacing is a most important aspect. The cutting tools on a machine are arranged to cut in arrays. The previous tools in a sequence prepare the rock and reduce the forces compared with a tool cutting in the unrelieved condition. There are three spacing situations for a given depth of cut (Figure 17). Wide spacing, where the tools cut unrelieved, creating a groove deepening situation, resulting in the formation of a rib of rock between adjacent tool paths. Groove deepening produces very high force and specific energy levels with the tool eventually just removing the volume swept by the tool [6]. The second condition is where the tool spacing is too close and the tool releases only a small volume of rock. Specific energy values are high and the tool is operating inefficiently. For a given depth of cut, as the spacing is widened the forces rise until they reach a plateau value corresponding to that of unrelieved cutting. The amount of debris produced increases as spacing widens, producing a maximum as interaction between grooves is at its greatest, then drops to a constant value consistent with unrelieved cutting. The specific energy, being the quotient of mean force divided by amount of debris, reaches a minimum value. This minimum value has been found to be around a spacing to depth ratio of 1:2 for all depths tested. Tools on a rotary head, cutting in the traverse cutting mode, will cut in shallow and then attain maximum depth, then the cutting depth will reduce to zero as the tool leaves the cut. Interaction should be designed to take place over two-thirds of the cut. 7.3.2 Cutting Head and Drum Design Cutting heads are equipped with the cutting tool arranged in spirals to present a logical sequence of tools to the rock and also move the cut debris away from the face.
(b)
(α)
(c)
7T7^
Figure 17 Geometry of cutting situations: (a) coring due to too wide spacing, (b) tools too close and (c) optimum spacing
166
Mechanized Excavation (a)
(b)
Figure 18 Breakout patterns from CAD cutting head designs: (a) well balanced head design and (b) poorly balanced head design (after Morris [9])
Brooker [7] has produced a comprehensive practical review of shearer cutting drum design and manufacture, with many of the principles applying to cutting heads. It has been found to be important that the cutting sequence is arranged to avoid fluctuations in torque and horizontal and vertical forces acting on the cutting head or drum [8]. This may be achieved manually by summing all the force components for tools in cut for each degree of rotation of the cutting head. In practise, this is too time-consuming and computer programs have been written which undertake this task as well as providing breakout patterns for a given head design. The breakout pattern gives an indication of the work done by each tool (Figure 18). A well designed head or drum will have equal areas associated with each tool and the torque and force components on the head should be free from undue variation. Failure to observe this simple requirement will result in damaged gear boxes due to unwanted vibrations and the unequal duty will result in poor production rates due to rapid wearing of heavily loaded tools and inefficient rubbing cuts for the lightly loaded tools. This matter of dynamic balance is covered in depth by Deliac in Chapter 8 of this volume. For a given set of cutting conditions the cutting head designer has to set up the head for the worst cutting conditions. In weak rock with deep penetration of the tools, wide spacing can be adopted. For hard rock cutting conditions, narrower line spacing is required with usually a second start arranged so that the tools cut between the grooves of the first start; this is preferable to duplicating the cutting pattern of the first start. 7.3.3 Water Jet Assisted Drag Tool Cutting The addition of high pressure water jets to assist with the excavation of rock has been investigated and is available on boom tunneling machines and coalface shearers. The advantages claimed for water jet assistance include improved tool life and debris clearance, reduced vibrations and respirable dust generation and, under certain cutting conditions, reduced tool forces. Major benefits in terms of improved tool life are gained at pressures of 20-30 MPa, being a tradeoff between cost of providing the high pressure water and improved cutting performance. Deterioration of floor conditions due to excess water for track mounted machines is a factor that must be considered [10]. Jets may be placed in front of the tool or behind. Jets or sprays behind the tool are used to cool hot material in the cutter track where there is a danger of an ignition hazard in a potential methane contaminated environment. Jets in front of the tool are preferred for normal cutting operations. Hood will cover in greater detail the mechanics of water jet rock excavation in Chapter 9 of this volume. Before reviewing the cutting mechanics of the disk cutter, it is appropriate to consider the requirements of cutting tool materials and the influence of wear on drag tool cutting performance. 7.4 CUTTING TOOL MATERIALS AND WEAR Almost all drag tools are equipped with tungsten carbide inserts, though many other materials have been tried as alternatives. Synthetic diamond inserts are being developed but, due to their high cost, they will find difficulty in competing with tungsten carbide.
167
The Mechanics of Rock Cutting
There is a need for new cutting tool materials that are capable of transmitting greater power from the machine into the rock. The desirable properties are hardness to resist abrasive wear, toughness to resist impact failure and a good hot hardness characteristic which requires that material retains its hardness at elevated temperatures. Many materials have good hardness characteristics, but the many alternatives to tungsten carbide fail due to poor toughness or fracture resistance properties. 7.4.1 Tungsten Carbide Tungsten carbide is a composite of grains of tungsten carbide held within a matrix of cobalt. Other binder phases have been considered, but cobalt has remained the only binder in use to date. Other carbide additions have been used such as tantalum and titanium carbides, but their use is mainly restricted to metal cutting grades of carbide. For rock cutting applications the percentage of cobalt by weight is a major variable. Typical percentages range from 7% to 15%, with general purpose grades around 10%. Reduction in percentage of cobalt increases the hardness of the carbide and hence improves its abrasion resistance. Increase in cobalt content increases toughness at the expense of hardness (Figure 19). In practise, there are a number of other factors to take into consideration, including nominal grain size, grain size distribution, freedom from impurities, carbon content and other factors under the control of the carbide manufacturers. Tungsten carbide is obtained from scheelite or wolframite and is processed to form tungsten carbide powder. Many manufacturers buy in their supplies of tungsten carbide and cobalt powders; others have greater control over the powder production. Freedom from impurities and quality control checks are very important to provide a reliable product. The tungsten carbide tool manufacturer either ball or vibratory mills the tungsten carbide and cobalt powders together to smear the cobalt over the carbide grains. The powders, after quality control checks, are pressed to form the cutting insert geometry allowing for around 20% linear contraction during the sintering stages. During sintering, the tips are subjected to high temperatures in an inert environment which, on cooling, provides the finished carbide tip ready for brazing onto the parent steel body to form the finished drag tool. Variations in the manufacturing process can produce layered carbide which gives additional properties: a hard outer layer high in tungsten carbide to resist abrasion, a tough supporting layer and a hard core. This carbide has found application for button insert manufacture and is currently undergoing evaluation for radial tool applications. 7.4.2 Wear Mechanisms All cutting tools excavating rock suffer some form of deterioration. The most obvious forms are abrasion, gross fracture or thermal cracking, though many other forms of wear are present during rock excavation. These have been reviewed by Clark [3], Osburn [11] and Larsen-Basse [12].
600
\
\\
N . ^^-Compressive strength
-
1800
-
1000
|
a.
400 > v ^ - Hardness
" *—■
200 Transverse ^ \ ^ ^ rupture strength ^*s^. 1
1
Cobalt content (weight %)
Figure 19 Effect of cobalt content on mechanical properties of tungsten carbide-cobalt alloy (after Exner and Gurland [10])
168
Mechanized Excavation
Abrasive wear may be considered to be micromachining of the tool by particles of rock. This form of wear results in the generation of a wear flat, which increases the force components for a given depth of cut. To resist abrasion, the tool material must be hard and retain its hardness at elevated temperatures. Tungsten carbide, though considered to have a good hot hardness characteristic, exhibits a drop in hardness with rise in temperature to the extent that the hardness of quartz at ambient temperature is similar to tungsten carbide at 400 °C (Figure 20). Hence rapid wearing of localized hot spots is to be expected. Water cooling of cutting tools is, therefore, beneficial provided thermal shock can be avoided. Cutting tool speed is an important factor as the heat build up in the tool is a function of power transmitted which is the product of force on the tool and velocity. The heat generated is not removed in the chips as with metal cutting, but builds up in the tool. In drilling operations aflushingmedium removes heat as well as the debris, which allows diamond products to be used as a cutting material. Natural diamond graphitizes at 600 °C and is therfore unsuitable for tipping drag tools. 2000
.Quartz 800
400
400 800 1200 Temperature (°C)
Figure 20 Hardness versus temperature (after Osburn [11])
Ξ υ
0
100 200 300 Distance cut (m)
3 4 0 I Wear flat width (mm)
Figure 21 Rise in forces with wear flat development (after Kenny [3])
The Mechanics of Rock Cutting
169
For abrasive rock cutting, slow cutting speeds consistent with economic cutting rates and water cooling are required. In low abrasivity rocks, thermal fatigue can be observed as a crazed surface to the cutting tool. This is due to the heating of the carbide when in contact with the rock and cooling when in free space. Gross fracture usually results from the tool impacting against hard rock or a tunnel support, though it may also be due to extension of cracks formed from thermal fatigue failure. Carbide geometry and composition have to be modified when gross failure is a common occurrence. Increasing carbide grain size and cobalt content are remedial measures that may be adopted. From a practical point of view the carbide grade selected is an engineering compromise between hardness and toughness. The choice of tool geometry and whether radial or point attack are also important considerations. As the drag pick wears, the forces will rise, the normal force component rising up to four times the pristine value for the tool (Figure 21). The machine has to have hold-in forces to maintain an efficient depth of cut as wearing progresses. Tools should be inspected and replaced at regular intervals to maintain cutting performance in hard or abrasive ground. Kenny and Johnson [13] have undertaken a comprehensive program of cutting tool material testing, looking at tool material properties, drag tool geometry and operational conditions.
7.5
DISK CUTTERS
The application of disk cutters is confined to full face tunnel boring machines and raise boring machines, though recent developments have looked at their application to specialist machines such as the Robbins Mobile Miner. The primary requirement of a disk cutter is to have adequate thrust in order to gain penetration which, in turn, reduces the number of cutters required on the machine. The relationships for the disk cutter are similar in many respects to the drag tool. Terminology for component forces is different. The rolling force replaces the cutting force. Unlike the drag tool, significant damage is done to the rock being cut below the disk's path due to the compressive loading inducing tensile cracking in the rock. These cracks are exploited on subsequent passes of the tool. In laboratory testing, the experimental conditions need to be replicated for at least five passes to simulate normal field working conditions as there is a cyclic action to the mode of operation of the disk cutter.
7.5.1
Disk Cutting Relationships
The primary variables connected with disk cutting are: penetration depth, wedge edge angle, disk diameter, speed and spacing. The field and laboratory relationships derived for a disk cutter have been reviewed by Clark in his book on the Principles of Rock Fragmentation [3].
7.5.1.1
Penetration
For unrelieved cutting, the relationships for mean and mean peak rolling force are very similar in magnitude, unlike the equivalent relationship for a drag tool. This is due to the crushing action of the disk maintaining a high level of thrust between large chip generation events. The ratio of the thrust force to rolling force is about 10:1, unlike the drag tool, where normal and cutting force components are of similar magnitude for a pristine tool. A typical force penetration characteristic is given in Figure 22, along with the specific energy penetration characteristics. Specific energy drops with increase in penetration, which is of a much higher magnitude than that for equivalent depths taken with drag tools.
7.5.1.2 Tire edge wedge angle As the wedge angle is increased, the forces rise. Many operators use constant profile disk tires, where the area of contact is maintained constant for the majority of the life of the tire. The cutting
170
Mechanized Excavation
2
4
6
8
2
10
4
6
8
10
Penetration (mm)
Penetration (mm) 60 i -
-
40
50
60
70
80
90
100
50
60
40«-
70
80
90
100
Disk edge angle0
Disk edge angle0
Measured
2 a>
100
150
200
Disk diameter (mm)
50
100
150
200
Disk diameter (mm)
Figure 22 Influence of disk variables: (a) penetration, (b) disk edge angle and (c) disk diameter (after Roxborough [14])
edge in practise is worn to a flat contact surface and by having a parallel-sided section, the thrust forces do not have an increased area of contact, as with a triangular section tire (Figure 23). 7.5.1.3 Disk diameter The influence of disk diameter on rolling force is small and the specific energy is constant, though the thrust requirement increases with increase in diameter. In practise, in abrasive and hard ground, where space on the head allows, large diameter disks are used to reduce the wearing of the disk; also larger diameter disks allow larger, higher capacity bearings to be used. Disk diameters range from 250 mm to 500 mm. The thrust force for a given penetration is related to the area of contact between the disk and the rock [14]. 7.5.1.4 Speed Disk forces are considered to be independent of speed, provided wear is disregarded.
171
The Mechanics of Rock Cutting
Conventional tire profile
Constant tire profile
Figure 23 Disk cutter tire profiles
7.5.1.5 Spacing The spacing of disks on a cutting head is of primary importance. Where possible, single tired disks should be used, as it is the thrust that can be sustained by the bearing, typically 20-401, that is the governing factor. Two tires on the same bearing reduce the individual disk thrusts available and hence limit the penetration achieved. Optimum spacing is where the grooves created by the disk interact, though this may take several passes of the disk to achieve. Laboratory trials have indicated that optimum spacing is at penetration to spacing ratios of 10-17:1 [17] (Figure 24). Failure of the rock is by crushing of rock immediately below the disk cutter and a combination of shear and tensile failure between the adjacent cuts. Roxborough and Phillips [14], Farmer and Glossop [15], and Özdemir [16] proposed models to predict thrust and rolling force requirements giving results with adequate accuracy for the cutting head designer. As with drag picks, cutting on the periphery of the tunnel requires higher forces and closer spacings are adopted in this area to reduce the disk loading and to ensure the correct finished dimensions of the tunnel. (a) ~
250 40
200
a.
(b)
/
30
/
-■ 100
Penetration • 2 mm o 4 mm ■ 6 mm D 8 mm
50
20
40
60
80
J_
_L
20
U/s*
10
*··Ϊ*·-Τ
_L
40
100 120 140 160 180
120
(c)
E ~3
^
UJ
100
—
80
~
60
-
40
-
>N
2»
a> c UJ o
<> /
/
\
IV
^
«♦-
o a» a.
i
60
i
80
1 100
1 1 1 1 120 140 160 180
Spacing, 5(mm)
Spacing, 5 (mm) in
■ Penetration • 2 mm o 4 mm ■ 6 mm a 8 mm
2 - — 75 JC\
/ / ·' / 1/*/ 1 ■ 1 '/ Penetration • o ■ D
-oΆ)
2 4 6 8
mm mm mm mm
1
1
1
1
1
1
5
10
15
20
25
30
Spacing/penetration ratio, S/p
Figure 24 Effect of spacing and penetration on (a) normal force, (b) rolling force and (c) specific energy in Merrivale granite (after Snowden et al. [17])
172
Mechanized Excavation
7.6 CUTTABILITY AND MACHINE PERFORMANCE Before an excavation machine is put to work it is imperative that the task is fully defined. An assessment of the rock materials to be excavated is an important part of the appraisal. The rock material to be excavated is only one of the factors to be considered when assessing machine performance. The specification of the machine, size of excavation, presence of water and use of excavation are all important. Table 2 gives a number of factors that have been found to influence the rate of excavation and Table 3 lists the factors influencing the degree of utilization of an excavation machine, where utilization is the percentage of the available time the machine is working. Though there are numerous test procedures used worldwide, account must be taken of three basic quantities: a measure of the strength of the rock or how easily it may be excavated, the abrasivity or tool degradation property and the rock mass fracture density. Compressive strength is the most common method of assessing cuttability, not because it provides the whole picture, but because it is specified for most site investigation programs for support purposes as well as excavation assessment. Compressive strength does not model the fracture process and is subject to the influence of changes in test specimen size, accuracy, moisture content and loading rate. Within one lithology or group of associated lithologies, compressive strength has been found to provide a good indicator of machine performance. Evaporite rocks do not relate to the established scales of performance due to their inherent toughness. Howarth [18] has reviewed the methods used worldwide for cuttability assessment. Mosrmodel the cutting action of a drag tool to some extent. The author has found that the most effective way of predicting drag tool equipped excavation machines is by taking instrumented cuts with a standardized full-size drag tool in block or core samples of the rock to be excavated (Figure 25). The specific energy derived from this test has been successfully related to roadheader performance (Table 4 [19]). Specific energy in this context is taken as an index property of the rock as the only variable is the material being cut, unlike the previously quoted relationships for drag tools and disk cutters. 7.6.1 Abrasivity This is a term often used to express the rate of tool replacements required to maintain the production rate required. In practise, as stated previously, abrasivity is only one of the forms of deterioration experienced by cutting tools. Careful examination of failed tools can often lead to beneficial changes to operational practise, tungsten carbide grade, or tool geometry. Table 2 Factors Influencing Machine Performance Main factors Rock parameters
Variables (i) Intact properties (ii) Mass properties
(iii) Environment
Machine parameters
(i) Cutting head
(ii) Weight
(iii) Operational characteristics
(a) (b) (c) (a)
Strength cuttability Cutting wear (1) Abrasivity Slurry make (1) Slake Durability Discontinuities (1) Volumetric intensity (2) Orientation (3) Shear strength (b) Mixed face conditions (c) Degree of variation in strata (along line of tunnel) (a) Water (1) From within rock mass (2) From dust suppression (b) Tunnel geometry (1) Size (2) Shape (3) Gradient (c) In situ stresses (a) Number of tools (b) Tool type (1) Radial/conical (2) Tip geometry (3) Carbide composition of tip (a) Slewing + lifting forces (b) Head speed (c) Head power (d) Rigidity of machine construction (a) Profiling (b) Guidance (c) Degree of automation
The Mechanics of Rock Cutting Table 3 Factors Influencing Machine Utilization Main factors Downtime
Variables (a) Maintenance (b) Spares availability (a) Staff availability (b) Spares availability (c) Conditions in the tunnel
(i) Planned (ii) Unplanned
Support
Debris disposal
(i) (ii) (iii) (iv)
Type and amount required Erection system Degree of mechanization Ancillary operations
(a) Grouting (b) Lagging boards (a) Cleaning up (b) General mucking (c) Secondary breakage (a) Conveyors (b) Mine cars (c) Presence of water
(i) At the face (ii) Behind the face
Ancillaries
(i) Ventilation and dust extraction (ii) Extensions
Labour
(i) Availability (ii) Experience/skill (iii) Transport to face
Organization
(a) Water/pumping equipment (b) Air ducts for extraction (c) Track (d) Power cables (e) Telephone (f) Conveyor
(a) Distance/time (b) Method (a) Bonus schemes (i) Management (b) Communication (a) Total payable time (ii) Shift times (b) Production time Final use Engineering tolerance (a) Grade (b) Alignment Intergration Is tunneling the only on-site activity, or is it competing for resources with other operations (i.e. mining) A problem of disposal and drainage Water
Direction of cutting
Cut 3
Cut I
Cut 4
Figure 25 Arrangement for core cutting test
173
Mechanized Excavation
174
Table 4 Application of Roadheaders in Massive Widely Jointed Beds in Relation to Specific Energy [19]. Laboratory specific energy related to machine weight Heavy Medium (MJnT 3 ) (MJm" 3 ) 25-32
15-20
20-25
12-15
17-20
8-12
8-17
5-8
<8
<5
Generalized cutting performance
Machines can only cut these rocks at economic rates if they occur in thin bands (less than 0.3 m). Specialist advice should be obtained and modification for cutting hard rock may help. Poor cutting performance. Regular changes of slightly worn picks will improve energy requirements and will reduce component wear. Point attack tools may be more beneficial and low speed cutting motors and side stalls will improve stability. Moderate to poor cutting performance. For abrasive rocks picks must be inspected frequently as sharp picks will improve performance. Moderate to good cutting performance with low wear of machine components. Picks must be changed regularly when excavating abrasive rocks. Machines well suited to these rocks. High advance rates. Mudstones in lower end of category may be ripped rather than cut and very high cutting rates can be achieved.
Methods of assessing abrasivity include measuring quartz content, grain size, angularity and cementation in a sample, measuring the rate of material loss from a work piece rubbed against a rock sample, and rate of wear flat development. West [20] has reviewed some of the available methods. Two methods that have stood the test of time are the Cerchar Abrasivity Number and the Schimazek F value [21]. The Cerchar Abrasivity Number is obtained by dragging across the surface of the rock specimen a stylus with a 90 ° cone for 10 mm under a load of 7 kg. The cone is produced from a specified grade of steel with a Vickers pyramid hardness number of 660. The diameter of wear flat generated expressed in tenths of a millimeter is the Cerchar Abrasivity Number. This has been found to correlate well with roadheader tool consumption rates (Figure 26). The Schimazek F value is given by the following formula. _ Equivalent Quartz % x Grain Size x Tensile Strength FiNmrrT1) = 100
These two scales are linearly related and give an initial indication of the tool consumption per cubic meter of rock excavated, though where high pressure water jet assistance is employed, much lower tool consumption rates may be expected.
2.5
Γ
2
3
4
Abrasivity coeff.
5
6
(AB)
Figure 26 Abrasivity coefficient against tool consumption (after Johnston and Fowell [22])
The Mechanics of Rock Cutting
175
7.6.2 Rock Mass Properties The fracture spacing is a most important parameter when describing the rock mass for mechanical excavation purposes (Figure 27) [23]. With few fractures or discontinuities, the material is considered to be massive, and all material excavated will have to be cut by the machine. Depending on orientation, one favorably placed discontinuity plane can double the excavation rate in very hard strata for a roadheader. The influence of discontinuities on excavation rate rises as the strength of the rock increases for boom-type tunneling machines. Measurements of the discontinuity frequency are normally obtained from logging exposed faces or from borehole core samples. Care must be taken when assessing fracturing to ensure that the fractures assessed are natural and not drilling or handling induced. Rock mass rating classification systems modified for excavation purposes have shown that they are capable of taking into account the influence of r.ock mass properties [24]. It is important to bear in mind that the assessment of machine performance depends on a combination of all factors but that some will be dominant. For example, an extremely abrasive rock will govern the method of excavation no matter how weak the material. No definitive upper limit for the compressive strength of rock for drag tool application can be given as it depends on the combination of strength, abrasivity and fracture frequency. For heavyweight high powered road headers, an upper limit of 100 M Pa is quoted by a number of machine manufacturers. Nelson will review the performance of TBMs in Chapter 10. 7.7 CONCLUSIONS AND THE FUTURE Drag picks tipped with tungsten carbide are the principal excavation tools for the weaker rocks. For hard rocks, the disk cutter is the obvious choice, though at present limited to the creation of circular tunnels and raise-bored shafts. The initial application of the Robbins Mobile Miner [25] has shown the way forward in creating noncircular underground structures in hard rock and other manufacturers will also be exploiting the hard rock capabilities of the disk cutter. Tunneling machines for small diameter bores with totally automated operations will be more widely applied. Profile guidance, comprehensive health monitoring and integrated support systems will become standard for boom tunneling machines. New cutting tool materials will be developed that have both increased hardness and toughness compared with current grades of tungsten carbide, along with improved hot hardness characteristics.
80 |-
60
h
x + • A o
/
4
Upper mudstone undercutting Middle mudstone undercutting Upper mudstone sumping Middle mudstone sumping Lower mudstone overcutting
6
8
10
Break index (m-1)
Figure 27 Influence of fractures on cutting rate for a boom tunneling machine (after Fowell and McFeat-Smith [23])
176
Mechanized Excavation
TBMs will become more versatile in terms of the variety of ground they can economically tackle. Such features as slurry and earth pressure balanced head configurations for running and waterbearing conditions will become more common. For hard rock applications, the use of high pressure water jet assistance to reduce tool costs and the development of machines employing percussive action will see a revival. 7.8 REFERENCES 1. Stack B. Handbook of Mining and Tunnelling Machinery, p. 742. Wiley, Chichester (1982). 2. Mellor M. Mechanics of Cutting and Boring. Series of 8 Reports prepared by US Army Corps of Engineers, Cold Regions Research and Engineering Laboratory, Hanover, New Hampshire, USA (1975-1981). 3. Clark G. B. Principles of Rock Fragmentation, p. 610. Wiley, New York (1987). 4. Evans I. A theory of the basic mechanics of coal ploughing. In Proc. Int. Symp. Mining Research, University of Missouri vol. 2, pp. 761-796. Pergamon Press, Oxford (1961). 5. Roxborough F. F. Cutting rock with picks. Min. Eng. {London) 132, 445-455 (1973). 6. Roxborough F. F. Multiple pass sub-interactive rock cutting with picks and discs. In Proc. Conf Applied Rock Engineering, Newcastle upon Tyne (Edited by M. Jones), pp. 183-191. IMM, London (1988). 7. Brooker C. M. Theoretical and practical aspects of cutting and loading by shearer drums. Colliery Guardian Coal International, January and April (1979). 8. Hurt K. G., MacAndrew K. M. and Morris C. J. Boom roadheader cutting vibration: measurement and prediction. In Proc. Conf. Applied Rock Engineering, Newcastle upon Tyne (Edited by M. Jones), pp. 89-97. IMM, London (1988). 9. Morris C. J. and MacAndrew K. M. A laboratory study of high pressure water jet assisted cutting. In Proc. 8th Int. Symp. Jet Cutting Technology, Durham, UK pp. 1-8. BHRA, Cranfield, UK (1986). 10. Exner H. E. and Gurland J. A review of parameters influencing some mechanical properties of tungsten carbide-cobalt alloys. Powder Metall. 13, 13-21 (1970). 11. Osburn H. J. Wear of rock cutting tools. Powder Metall. 12, 471-502 (1969). 12. Larsen-Basse J. Wear of hard metals in rock drilling: a survey of the literature. Powder Metall 16, 1-32 (1973). 13. Kenny P. and Johnson S. N. The effect of wear on the performance of mineral-cutting tools. Colliery Guardian, June (1976). 14. Roxborough F. F. and Phillips H. R. Rock excavation by disc cutter. Int. J. Rock Mech. Min. Sei & Geomech. Abstr. 12, 361-366 (1975). 15. Farmer I. W. and Glossop N. H. Mechanics of disc cutter penetration. Tunnels and Tunnelling 12, 22-25 July (1980). 16. Özdemir L. A laboratory and field investigation of tunnel boreability. M.Sc. Thesis T1755, Colorado School of Mines, Golden, CO (1975). 17. Snowden R. A., Ryley M. D. and Temporal J. A study of disc cutting in selected British rocks. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 19, 107-121 (1982). 18. Howarth D. F. Mechanical rock excavation: assessment of cuttability and boreability. In Proc. Rapid Excavation and Tunneling Conf, New Orleans (Edited by Jacobs and Hendricks), chap. 11, pp. 145-164. AIME, New York (1987). 19. McFeat-Smith I. and Fowell R. J. The selection and application of roadheaders for rock tunnelling. In Proc. 4th Rapid Excavation and Tunneling Conf, Atlanta, pp. 261-279. AIME, New York (1979). 20. West G. A review of rock abrasiveness testing for tunnelling. In Proc. Int. Symp. Weak Rocks, Tokyo, pp. 585-594. Balkema, Rotterdam (1981). 21. Schimazek J. and Knatz H. Der Einfluss des Gesteinsaufbaus anf die Schnittgeschwindigkeit und den Meisselverschleib von Streckenvortreibmaschinen. Gluckauf 106, 274-278 (1970). 22. Johnson S. T. and Fowell R. J. Compressive strength is not enough: assessing pick wear rates for drag tool equipped machines. In Proc. 27th US Rock Mechanics Symp., Tuscazoosa, AL (Edited by H. L. Hartman), pp. 840-845. AIME, New York (1986). 23. Fowell R. J. and McFeat-Smith I. Factors influencing the cutting performance of a selective tunnelling machine. In Proc. 1st Int. Symp. Tunnelling Ί6, London, pp. 301-309. IMM, London (1976). 24. Johnson S. T. and Fowell R. J. A rational approach to practical performance assessment for rapid excavation. In Proc. 25th U.S. Symp. Rock Mech., Northwestern University, Evanston, IL, pp. 759-766. AIME, New York (1984). 25. Boyd R. J. Performance and experimental development of the mobile miner at Mount Isa. In Proc. Rapid Excavation and Tunneling Conf, New Orleans (Edited by Jacobs and Hendricks), chap. 48, pp. 747-768. AIME, New York (1987).
8 Theoretical and Practical Rules for Mechanical Rock Excavation ERIC P. DELIAC Elf Aquitaine Production, Pau, France 8.1
HISTORY AND ASSESSMENT OF EXCAVATING MACHINES
8.1.1 8.1.2 8.2
A Brief Historical Review Principles of Quantitative Evaluation of Performance
MODELING OF TOOL-ROCK INTERACTION
180 180 181 184 185
8.2.1 Picks 8.2.1.1 Single unrelieved cutting 8.2.1.2 Interactive deepened and inclined cutting 8.2.2 Disks 8.3
MODELING OF ROCK CUTTING HEADS
186 187 187 188 190 191 191 193 199 200 200 202 202
8.3.1 Kinematic Analysis of Excavation Heads 8.3.1.1 Cylindrical drums 8.3.1.2 Roadheaders in axial phase 8.3.1.3 Full face tunnel boring machines 8.3.2 Theoretical Model of Drum-type Heads 8.3.2.1 General method of calculation 8.3.2.2 Application to a simplified example 8.3.2.3 Computer simulation of cylindrical drums 8.3.3 Theoretical Model of Roadheader Heads 8.3.3.1 Axial cutting 8.3.3.2 Transversal cutting 8.3.4 Theoretical Model of TBM Shields 8.4 VALIDATION OF THEORETICAL MODELS AND MACHINE SIMULATION 8.4.1 Experimental Validation of Theoretical Models 8.4.1.1 Laboratory scale 8.4.1.2 Fullfield 8.4.2 A Method for Determining Machine Performance 8.4.2.1 General principle of the method 8.4.2.2 Application to drum shearers 8.4.2.3 Application to continuous miners and roadheaders 8.4.3 Computer Simulation of Pick Cutting Machines 8.4.3.1 The P.CD RUM program 8.4.3.2 The P.C.MAP program 8.4.4 Practical Examples 8.4.4.1 Evaluation of a drum shearer 8.4.4.2 Optimizing the design of a continuous miner 8.4.4.3 Selection of a well-suited roadheader 8.5
178 178 178
ADAPTATION OF MECHANICAL EXCAVATION TO A HARSH ENVIRONMENT
8.5.1 Adaptation of Machine Specification and Head Design 8.5.2 Adaptation of Cutting Tools 8.5.2.1 Diamond-tipped picks 8.5.2.2 Water jet assistance 8.5.3 A Simple Economic Model for the Selection of Equipment 8.5.3.1 When should cutters be changed? 8.5.3.2 How many tools should be changed? 8.5.3.3 Comparison between diamond and tungsten carbide picks 8.5.3.4 Comparison between dry and water jet assisted cutting
177
204 204 204 206 208 208 209 210 211 211 213 214 214 217 218 219 219 220 220 221 223 223 223 224 225
178
Mechanized Excavation
8.6
CONCLUSIONS AND FUTURE PROSPECTS OF MECHANICAL ROCK EXCAVATION
225
8.7
NOMENCLATURE
225
8.8
REFERENCES
226
8.1 HISTORY AND ASSESSMENT OF EXCAVATING MACHINES 8.1.1 A Brief Historical Review Rock excavating machines were first introduced during the 19th century. This was for tunneling purposes (Channel Tunnel, London Underground, etc.). The rational and industrial development of these machines is, however, more recent and can be traced back abou1<50 years. The first continuous miners were introduced in US mines during the Second World War. Soon afterwards, the drum shearers were invented in the UK during the early 1950s, followed by the coal ploughs (developed in Germany a few years later). The concept of roadheaders was born in Eastern Europe around 1960, but they became increasingly successful after use in the UK. All these machines were using picks or plough cutters of a similar design. Full face tunnel boring machines (TBMS, invented by the Robbins US company at the end of the 1960s) marked the starting point of new ideas: disk cutters fixed to huge rotating heads, giving perfect tunnel profiles well adapted to underground civil works, etc. Over the past 25 years, there has not been any other technical breakthrough in mechanical rock excavation, as far as the type of machine is concerned, but a great deal of good work has been carried out to understand the mechanics of rock cutting and thus optimize both the design of machine specifications and that of the cutting tools. It is thus possible today to provide a set of practical rules as to the design and selection of excavating equipment, based on theoretical and rational work well validated from thorough experiments. The purpose of this chapter is therefore to give an outline of these methods, starting with a logical approach to evaluate the performance of a given machine in a given rock environment. 8.1.2 Principles of Quantitative Evaluation of Performance The performance of any excavating machine is normally related to its advance rate into the rock formation and to the stability of the cutting head. A fast penetrating machine would soon face difficulties if it was vibrating badly at the same time. Fortunately, as observed underground, a high rate of penetration is usually bound to a low level of vibrations, so that 'performance' will hereafter refer to the speed at which the cutting tools are advancing into the rock. For low speeds it must, however, be borne in mind that vibrations must be evaluated. As illustrated in Figure 1, 'assessment of performance' will hereafter refer to the correlation between the machine specifications, on the one hand, and its production rate (or advance speed), together with the level of vibrations, on the other hand. Machine specifications consist of available torque, thrust force (to penetrate into the rock face) and maximum normal reactions (perpendicular to the thrust force). They are obviously related to the electrical power installed on the machine (together with the rotation speed of the head), its weight and the characteristics of the hydraulic devices which supply forces to the head. As shown in Figure 2, it is possible to correlate the pick penetration rate to the machine specifications, as soon as a proper 'model of the head' has been devised. Such a model will be dealt with in Section 8.2. Suffice here to say that it is a function of the cutting tools, the rock environment, and the head design (including cutter spacing). The first two lead to the 'tool-rock interaction', dealt with Section 8.2, which relates tool design (geometry of tip) and wear to the rock parameters (strength, stress, texture). The last involves a 'kinematic analysis', also detailed in Section 8.2, which in turn defines the type of tool interaction and results in the calculation of individual forces on each tool, hence the resultant efforts on the head. As a matter of fact, knowing the available mechanical power and the rotation speed of the head will yield the torque, e.g. cutting forces at the tips of the cutters, whereas the thrust force into the face will be equivalent to the normal forces on each tool. This is why, knowing the mechanics of cutting for the tools, with special reference to interactions between neighboring cutters, it is possible to relate
Theoretical and Practical Rules for Mechanical Rock Excavation
Specifications Forces Torque F, R, T
179
Performance Picks Rock Head design
Speed
Figure 1 Appraisal of an excavating machine SPECIFICATIONS
PERFORMANCE
Figure 2 Modeling of rock cutting machines
the depth of cut of each tool and the machine specifications. This depth of cut is a simple function of rotational speed and advance rate, hence the performance of the machine. Conversely, taking a desired minimum advance rate with a given type of machine in a given rock environment, it is the purpose of this chapter to explain how to calculate the required torque (hence the power) and thrust on the machine cutting head. The above preliminary considerations will therefore be followed by Section 8.2 on the mechanics of tool-rock interaction, with a brief summary of the main quantitative relations useful in determining the forces on the cutting tools (picks and disks). Section 8.3 deals with the modeling of cutting heads, starting from the kinematic analysis of the cutters to account for their interaction pattern, and eventually showing how to calculate the forces and torque on the head. After some validation of the above theoretical concepts, Section 8.4 applies this to the evaluation of machine performance, today based on computer simulations, and details practical examples of machine selection, together with design optimization criteria.
180
Mechanized Excavation
Afterwards, more practical rules will be dealt with (Section 8.5), by looking at the adaptation of excavation machinery to difficult ground conditions, as this is when design/selection criteria become of critical importance. This section will be ended by a simple economical model to show how to optimize the choice of the cutting tools. 8.2 MODELING OF TOOL-ROCK INTERACTION In this section, the behavior of mechanical rock breaking tools will be dealt with. Basically, excavation machines use two types of cutting tools, the pick and the disk. Other tools are well known, such as drilling bits or plough cutters, and, apart from impact rock breaking which is based on different rock mechanics, their breakage mechanisms are similar to those investigated here. It is now well known that the mechanics of the tool-rock interaction involves a variety of parameters [1, 2] such as rock brittleness, tool stiffness and sharpness, depth of cut, etc. In addition, the tool may encounter three different cutting conditions, which also interfere with the failure of the rock: single unrelieved cutting, interactive cutting and, finally, deepening an existing cut (Figure 3). Last, but not least, this interaction may also be influenced by lateral inclination of the tools (inclination with respect to the plane defined by the direction of advance and normal to the rock surface, vertical in Figure 3). In this section, these different situations are dealt with, starting with the pick and then moving on to the disk cutter. 8.2.1 Picks It is more and more widely accepted that the cutting action of the pick (or disk) implies creating a crushed zone of rock around the tool tip [3,4]. This has been experimentally evidenced [3] while cutting Berea sandstone (46 MPa uniaxial compressive strength) and Tennessee Marble (116 MPa). This crushed zone is necessary to transfer the tool forces as stress in the rock. Much research has been carried out since Evans' pioneering work [5], as explained in the previous chapter of this volume [1]. Several authors have attempted to establish theoretical models to account for the pick-rock interaction (e.g. calculate forces), mainly via geometrical simplifications and assuming a single failure criterion to be universally valid. Lebrun [6] was the first, to our knowledge, to address the three cutting situations previously mentioned, e.g. single unrelieved cutting, relieved cutting and deepened cuts, using a Mohr-Coulomb criterion and assuming a prismatic shape for the rock chips. Others have relied on empirical equations or numerical simulations [7, 8]. Single unrelieved cut
Direction of advance
/
Relieved cut
Deepening cut
Figure 3
The three cutting environments
Theoretical and Practical Rules for Mechanical Rock Excavation
181
However, the dual nature of the failure mode of rock under the compressive action of a cutter was only clearly analyzed recently [2] and later confirmed by detailed work on the cutting mechanism of picks in hard rocks [9, 10]. Altogether, as shown by Déliac [2], the rock may fail in shear compression, resulting in finely crushed particles together with chips of various sizes but of a typical shape, which can be approximated by a prism, or, alternatively, it may fail in tension (fracture propagation), often more efficient in terms of specific energy. The first failure mode is typical of soft rocks, wide or worn-out picks, deep cuts and deepening incremental cutting or very inclined cutting, whereas the second mode is usual with brittle rocks, sharp picks or disks, or also with interactive cutting. It is thus essential to account for these two types of rock-tool interaction in any modeling of the forces. Although the following is related to pick-rock interaction, the above considerations, as well as the rock failure, also apply to disk cutters. Single unrelieved cutting is generally addressed first, as it is the reference situation to model. After the modeling of forces resulting from pick-rock interaction for this type of cut has been dealt with, interactive cutting, deepened cutting and inclined cutting are described in this section, as compared to single unrelieved cutting. 8.2.1.1 Single unrelieved cutting Forces acting on a pick when cutting rock are called cutting (Fc), normal (Fn) and lateral (Fx) (nomenclature at the end of the chapter), as illustrated in Figure 4 (which also illustrates characteristic bit angles, such as rake and clearance). Only the first one delivers any mechanical work, as it is parallel to the direction of advance, and the resulting energy can be computed by integration of the mean value of Fc over the pick displacement. However, in order to fail the rock, the pick must periodically exert peak values of the three force components (Figure 4). Modeling the pick action therefore consists of finding mean and peak values of the three force components in various situations, such as unrelieved cutting (vertical or laterally inclined), interactive cutting (with neighboring tools) and deepening of a preexisting groove. The problem appears thus to be fairly complicated and, after the detailed explanations given in the previous chapter [1], only a brief summary of the major equations will be given here, largely based on the work by Déliac [2].
Figure 4
Forces induced during the cutting process
182
Mechanized Excavation
Déliac [2] called the two fundamental chipping modes mode A and mode B. The former is typical of shear fracture of the rock. Assuming a three-dimensional prismatic shape for the chip (Figure 5) and a Mohr-Coulomb criterion for the failure of the rock (after the crushing phase), he obtained a good agreement with experimental values for the maximum forces on radial drag bits. The latter is a fracture propagation mode that he put into equations by use of a simplified fracture mechanics approach. In that case the rock chip is assumed to be of a spherical shape (Figure 5). In most situations the tool-rock interaction involves a combination of the two modes. However, it is possible to state that: (i) mode A is predominant when the pick is wide, the rock is not very brittle (a first approximation of brittleness is given by the ratio between compressive strength and fracture toughness), or when the depth of cut is high; and (ii) mode B is predominant when the pick is sharp and rigid, and when the rock is brittle. When the rock is brittle and the pick is in good condition, chipping will be in mode B at a shallow depth of cut and will gradually shift to mode A, as illustrated in Figure 6. Table 1 gives some insight into the cutting mode, depending on the situation. In Table 1, the definition of shallow cuts is highly dependent upon the type of rock and the shape of the pick. For instance, for brittle rocks cut by a sharp tool, 'very shallow' means less than 1 or 2 mm, and 'shallow' would refer to cuts up to approximately 10-20 mm. On the other hand, for worn or wide picks, in granular/porous rocks, these definitions could be extended to 10 mm for 'very shallow' (shallow is meaningless there, since no mode B can be found). Table 2 summarizes this dual (b)
(α)
Tool width
Figure 5 The two chipping modes (after Déliac [2]): (a) mode A; (b) mode B
5
10
d (mm)
Figure 6 Mixed mode cutting of Serrouville iron ore - V-shaped pick (after Déliac [2])
Theoretical and Practical Rules for Mechanical Rock Excavation
183
Table 1 Description of Tool-Rock Interaction Modes Types of cut
Description
Cross section shape of cut
Characteristics
Very shallow Shallow Mode A
Grinding of rock
Exactly that of the bit
High energy consumption
Shearing of rock chips
Two symmetrical quasilinear sidewalls
Tensile fracture propagation
Circular
Shearing of rock chips
Two symmetrical quasilinear sidewalls
Soft rocks Worn tools Inclined picks Brittle rocks Sharp tools High rake angle Specific energy is minimized
Mode B Deep
Table 2 Chipping mode
Maximum Cutting Force for a Pick (after Déliac and Fairhurst [9]) Main rock parameters3
A
B
Ic'
Maximum cutting forceb A-d +
c
Cd312
Bd2
Comments
A and B proportional to ac B < A C proportional to Klc
a
(7c = unconfined compressive strength, φ = internal friction angle, Klc = fracture toughness (Mode I), ac = characteristic chip angle (see Figure 5). b d = depth of cut.
mode theoretical model, as detailed by Déliac and Fairhurst [2, 9] as far as the peak cutting force is concerned. This is only a short summary of the main findings, as the mechanics of tool-rock interaction are dealt with in the previous chapter of this volume [1]. A review of the calculations involved to get the equations in Table 2 can also be found in other publications [2, 9]. As can be seen in Table 2, the rock strength directly affects the forces, but its expression depends on the failure mode: compressive strength for mode A and fracture toughness for mode B. Also the A, B and C coefficients are affected by the tip shape and the state of wear of the pick. For each failure mode, a second rock parameter may influence the rock, but this is secondary when compared to the effect of rock strength. It is the internal friction angle for mode A (forces increase with φ) and the characteristic angle ac (forces increase when ac decreases, which indicates a higher degree of brittleness) for mode B. It is possible to define an 'effective depth of cut' above which mode A is generally predominant. For many rocks, this is the depth of cut at which the pick width is the full tip width [2]. By integration of the stress field at the pick-crushed zone contact [2] it is possible to calculate the Fn/Fc ratio between the components of the chipping forces, often called the normal/cutting coefficient, Kn. With mode A, this normal component is very small and the pick is bent (flexion), causing a rock-tool contact on the clearance side, hence an additional normal force to be added to the previous one (only due to the chipping process). Déliac [2] thus derived an expression for the maximum normal force, in good agreement with his experimental data, showing that the normal to cutting force ratio increases with rock strength and decreases when the depth of cut increases. Suffice here to say that Fn/Fc is usually close to 1 at shallow depths of cut for picks in good condition. A good approximation of the function Kn = Kn(d) has been found from a semiempirical analysis of tool flexion and contact action with the rock on the clearance side [2]. The resulting equation can be simplified to become the following approximate relation KM = K0 +
(l-K0)ll-(2ß0d)l'2l
where the coefficients are defined as follows: K0 is a residual normal/cutting force coefficient (when d increases Kn cannot become smaller than K0\ usually between 0.25 (V-shaped tools) and 0.4 (drag bits or pointed picks); and ß0 is an empirical coefficient close to 0.01 mm" 1 for most picks (except for drag bits, with ß0 » 0.02 mm - 1 ). When the pick becomes worn out, the above equation must be changed by adjusting K0 and modifying 1 — (2ß0d)1/2 into A — (2ß0d)1,29 with A > 1.
184
Mechanized Excavation
In order to calculate mean values of forces from maximum values, a mode of penetration of the pick into the rock must be assumed, which in term enables the integration of forces during a complete cycle (defined by the formation of two consecutive major chips). A wedge (mode A) or point-loading (mode B) type of penetration is often assumed. This results, for mode A, in a mean to maximum value of between 0.4 and 0.5 (respectively 0.35 to 0.45 for mode B) for a tool in good condition.
8.2.1.2
Interactive deepened and inclined cutting
At this stage, it is necessary to calculate how the forces are affected in the actual cutting environments encountered on a machine, e.g. relieved cutting and deepening of an existing cut (Figure 3). The interaction between two neighboring cuts is largely controlled by an 'interaction coefficient', k (depending on rock brittleness and on the failure mode - k usually ranges between 1.5 for soft porous rocks and 4 for hard or brittle rocks). For a given spacing, S (between cut sidewalls for mode A and cut centers for mode B), and for adjacent cuts of similar depths there is a critical depth dc = S/k such that the cut is relieved when the depth of cut exceeds dc. Cutting and normal forces are reduced, but this reduction is limited when the depth of cut is less than 2dc. A lateral force is generated, expressed by the F\/FG coefficient, K{ (increasing from zero to a maximum value usually close to 0.5 or slightly less). It should be pointed out that this situation is quite beneficial in terms of specific energy, as a high volume of rock is excavated with no increase in force (even a reduction sometimes). A simple approximation has been found as Kt = FJFC = A-d-dJd)
(ford>dc)
with the coefficient A between 0.33 and 0.5 [2]. When the adjacent cuts are not of the same depth (in most instances, the previously cut groove is less deep than the current one), the critical depth must be modified (increased) to account for a less favorable situation. A good approximation of dc is supplied by the following equation [2] where 6d is the depth difference between the cuts (usually positive, when the adjacent cut is shallower, but sometimes negative, e.g. with a deeper neighboring cut, like on boom-type roadheaders in axial penetration; this configuration is described in Section 8.3.1.2) de = (S/k) + (bd/2) Two adverse cutting situations arise when either cutting into an existing groove (deepening environment) or when cutting with a lateral inclination (inclined cutting). Both situations can be represented by equations where the forces are those computed for the standard unrelieved environment multiplied by an 'adjustment factor' AF. For the deepening environment, this factor increases sharply with the first deepening increments, then tends to stabilize around a maximum value AFmax depending on the depth of cut and on the failure mode of the rock [2] (usually between 3 and 10). Opposite to the previous cutting situation (relieved cutting), this one requires a high level of specific energy, as forces increase dramatically while excavated volumes tend to decrease (the cut sidewalls rapidly become vertical, so that the incremental rock volume is minimal). A multiplicative coefficient AF(p) can be used to express the forces during the pth deepening increment of the cut, so that
Fc(p,d) =
A^pyF^d)
In the above equation, Fc(0, d) refers to the single unrelieved cut (no deepening: p = 0). A simplified estimation of AF(p) is often taken as AF(p) = p + 1, until p = 2 or 3, e.g. Fc(d,p) = (p + l)Fc(d, 0), and then AF(p) gradually tends to the maximum value AFmax (for p over 2 or 3). Inclined cutting may be complex to model, as the failure mode of the rock can change from B to A when the inclination angle increases [2, 11]. There again, an adjustment can be calculated. The results show that, firstly, inclined cutting affects the normal force more than the cutting force, and, secondly, the increase with the inclination angle is initially moderate (of the order of tan a and (tan a)/2 for the normal and cutting forces respectively, a being the tilt angle in a relative value) and then tends to accelerate for inclinations above 30° to 45°. Obviously, in this situation a lateral force is generated, as compared to the negligible values recorded during noninclined unrelieved cuts. The Fi/Fc ratio increases up to a maximum of between 0.3 and 0.5 for a tilt angle of between 20° and 40°.
Theoretical and Practical Rules for Mechanical Rock Excavation
185
1.5
^—"* 1.0
0.5 '
-
'— ·
• ^^"*
1
0
1 5 Wear flat (mm)
10
Figure 7 Influence of the wear flat on the FJFc ratio (after Déliac [2])
To complete this brief description of the pick-rock interaction, mention must be made of the influence of wear on the forces. Much experimental work has been carried out on this topic (Roepke et al. [12], for instance), but modeling of the incidence of wear is recent [2, 13]. There again, the mode of tool-rock interaction may change, always from mode B to A, e.g. fracture propagation becomes more energy consuming than shear when a wear flat develops under the tip of the pick or when its sharpness is lost (not enough stress concentration). The change in cutting force may be computed with the above model, simply by modifying the geometry of the tool (width, angles). The increase in normal force, however, is more difficult to evaluate, as the tool-rock contact at the clearance side of the tip generates an increasing force. Some preliminary work by Sellami [13] gives a new insight into the calculation of the normal force as a function of the wear flat. As a matter of fact, this increase is much more substantial than that of the cutting force (Figure 7). 8.2.2 Disks The force components in this case differ from those of the pick (Figure 8). The rock breaking action is primarily due to the normal force, F n , which indents the rock during the rolling of the disk. This force is related to a lateral component on each side of the cutter and to a rolling force Fr parallel Fr
AxS / 2 Pi
"x^-
2
'<^s>'
I Figure 8 Geometrical decomposition of forces on the disc (after Roxborough and Phillips [14])
186
Mechanized Excavation
to the direction of advance. The resultant force must go through the disk axis to prevent friction on the tool. Some interesting modeling of the forces exerted on a disk cutter can be found in Roxborough and Phillips [14]. They assume the normal force to equal the product of the rock uniaxial compressive strength and the projected contact area. If δ is the wedge angle of the disk, d the depth of cut and D the disk diameter, an estimation of the contact area is A = 4d(Dd - d2)1/2tan(<5/2) hence the expression for Fn Fn = 4-<7c d'(Dd-d2)l/2-tm(ô/2) It is generally acceptable to consider d as very small with respect to D, so that the above equation can be simplified to Fn = 4-<7c-tan(<5/2)-D1/2-d3/2 Writing the no friction condition, the authors obtain the equation for the rolling force Fr = 4-
Α^η(δ/2)Όί/2·ά3'2
Fr = Β-Χ*η(δ/2)·ά2 It is interesting to note that the above expressions are quite similar to those derived by Roxborough and Phillips. Thereafter, Lebrun tries to extend this modeling to interactive and deepened cutting. A wealth of experimental evidence described by this author shows that forces are generally overestimated, particularly the rolling force. The agreement is, however, fairly good between measured and calculated values (for single unrelieved cutting) when the disks have a tip angle δ of over 90°. The success is limited when it comes to relieved or deepened cuts, the forces being systematically highly overestimated. A more sophisticated model has been published by Sanio [4], based on assumptions of rock failure in tension (crack propagation from a crushed zone). This seems to correctly represent the physics of brittle rock cutting, which corresponds to many TBM applications. The rock strength is here represented by the fracture toughness. Sanio's model has also been criticized [2] but, nevertheless, its application leads to quantitative expressions of the forces similar to those of Roxborough and Phillips or Lebrun. The above expressions can therefore be adopted as representative of the forces on a disk cutter during a single unrelieved cut, at least for δ higher than 90°, but more research is needed to explore their value for more complex situations. The value of the mode B relieving coefficient k determined for a pick (see Section 8.2.1.2) is likely to be a fair estimate of the corresponding disk-rock coefficient, since most rocks cut by disk cutters are brittle and the depth of cut is shallow. 8.3 MODELING OF ROCK CUTTING HEADS In this section, the following topics will be dealt with. (i) Principles of cutting head design - the arrangement of the cutting tools is of paramount importance to the efficiency of the machine head. Optimizing the head design requires a careful kinematic analysis of the tools during a complete revolution.
Theoretical and Practical Rules for Mechanical Rock Excavation
187
(ii) Computation of forces on a cutting head - as illustrated in Figure 2, the knowledge of the tool-rock interaction and the kinematic analysis from the head design lead to the calculation of the resultant forces and torque on the head during its revolution. The curves giving the forces and torque versus the advance speed of the machine are called cutting curves. 8.3.1 Kinematic Analysis of Excavation Heads It is possible to derive torque and force reactions on a cutting head from the forces on the cutters, which obviously depend on the cutting environment. This is why a kinematic analysis of the tools during each head revolution is necessary, in order to determine the depths being cut as well as the different conditions likely to be met: interactive cutting when the depth of cut is large enough, and deepening of cuts elsewhere. Only then will it be possible to determine the individual forces on each of the picks/disks. The kinematic analysis is obviously dependent upon the type of cutting head and the work phase (penetration/sweeping). Three possible configurations will therefore be presented in this section: the cylindrical drum, with a constant rotational speed and advance rate; the roadheader head in the axial phase (milling or penetration for boom-type heads, and sweeping or shearing for transversal heads); and the full face TBM head. Pioneering work in this area is due to Lebrun [6] and was further improved and developed by Déliac [2, 15] and Cordelier [16]. 8.3.1.1 Cylindrical drums These heads are found on drum shearers and continuous miners. As illustrated in Figure 9, their design generally consists of a vane section, where the picks are located on spirals, together with a clearance section on the side of the drum. The design of the vane section is fairly easy to describe. It includes the number of vanes (spirals), that of picks per line, the spacing of the 'cutting lines' (usually numbered 1 onwards from the first vane line adjacent to the clearance ring, see Figure 9), and the wrap angle. The latter is defined as the angle (measured around the axis of the drum) between the starting point of a vane and its final position on the free side of the drum (Figure 9a). Each line obviously corresponds to one cut in the rock face. If the spirals are not staggered, the number of picks per line equals that of the spirals (Figure 9a). Otherwise, it is a divider of this number (usually half the number of spirals, and sometimes a third or a quarter), as in Figure 9(b). The design of the 'clearance ring' may be quite complex, as there is usually little room to place the pick boxes. Clearance picks (also called gauge cutters) are often placed along the spirals (Figure 9a),
Key 5 Vane line B Clearance line 2 Vane number A Vane pick /Gauge pick
(b) Clearance sequence ABCDCD
Line spacing h—H
43
Lirîe spacing
m
42 4>-
Q
/
1
1 •
"S
1
vfc
/
8
7
6
5
4
Vane section
3
2
I
/
learance section Clearance
IO
/'
9
L_,_
/
8
7
'/1
Ί1Ϊ] I I M
k
/
/
Λk
f·
4ΊΙ
4
Δ L . _. 6 5 4 3 2
Vane section
Offset,
k
Ak
à Il
-"- —y
'''
& 9
/
/ 1/
//
/i k
v/
Ί
S
?
•
|AB<·'
Clearance section
Figure 9 Typical design of a drum: (a) simplified design (three nonstaggered spirals); (b) real design (variable spacing, four staggered spirals, complex clearance)
188
Mechanized Excavation
normally laced opposite to the vane spirals. This is to balance the lateral forces originating from interactions between cuts and from the lateral tilt of the gauge cutters. The picks are inclined outwards to clear the drum from the rock on its side. The tilt increases as the picks get further away from the vane. It is possible to describe a clearance design by firstly locating a reference pick (adjacent to the vane), then defining the line characteristics (spacing, offset and tilt; see Figure 9a), and finally placing the picks on the lines. The clearance lines are usually labeled A, B, C, . . . , starting from the vane section. The picks are generally grouped in sequences (normally a multiple of the number of vanes). Each sequence is defined by the order of penetration into the rock and the rotation angle of the head between the arrival of pick i in the rock and that of pick i + 1 (Figure 9). For instance, the three sequences in Figure 9(a) would be defined as ABCDCD. For continuous miners, the cutting head may consist of two such drums placed symmetrically on either side of a 'trim chain', or, alternatively, it may be made of one single drum with two clearance rings (on each side), with the vane picks placed on two opposite spiral patterns (Figure 10). On most continuous miners the design of the heads is quite simple, with one spiral only (one pick per line). Two situations may occur: penetration into the massif (forward drum of the shearer, or start of the cycle for the continuous miner, when it 'attacks' the top of the face); or shearing (rear drum of the shearer, or second step of the continuous miner cycle, when the head cuts downwards). The former is usually the most difficult one, limiting the performance of the machine, whereas the latter is 'easy' and productive. Penetration into the massif will therefore be emphasized here, with particular attention to the situation when half the drum is in contact with the rock. If Vr is the rotational speed of the drum and Va its advance rate into the rock, the advance per revolution of the head is simply VJVr. Therefore, if n picks are cutting in the same groove (number of picks per line), the corresponding depth of cut is dm = VJnVr This depth of cut is in fact the maximum depth cut by a single pick during a drum revolution. As shown in Figure 11, the depth of cut varies from zero to dm, then decreases back to zero when the pick loses contact with the rock. The problem is now to determine in which environment the pick is cutting, which will depend on its depth of cut and on the spacing between the lines. For clarity, a constant spacing S will be assumed here. In most situations, the maximum depth of cut dm is higher than the critical depth
Figure 10 Cutting head of a continuous miner
Theoretical and Practical Rules for Mechanical Rock Excavation
189
d«dmsin Θ
Figure 11 Depth cut during a drum revolution
dc mentioned in Section 8.2.1.2 (dc = S/k, where k is the interaction coefficient). The ridge of rock is then removed and the next pick cuts a fresh surface (single cut). The forces on the pick are given by Fc = Fc (mode A or B, relieved cut, rock, d) Fn = Kn(d)Fc Fx = Kl(d,S)Fc where the above expressions refer to results given in Section 8.2.1. Even when dm> dc, since the pick starts cutting from zero depth of cut, there will be a position where the depth of cut equals the critical depth (Figure 11), so that before this position the first cut is not relieved and the next pick will have to deepen the cut (hence a high energy consumption). When d is close enough to dC9 the second cut deepens the groove and breaks the rock between the neighboring cuts. The reduction in instantaneous forces due to this 'relieved effect' is negligible, but, most importantly, the next tool will cut a fresh rock surface. Therefore, in this part of the drum-rock contact there will be a breakout cycle consisting of a first cut (unrelieved) followed by a deepening one. The average forces are given by the following equations Fc = [^(unrelieved,
The normal forces are derived from the cutting forces and the lateral forces are negligible. This calculation shows that the cutting cycle of the tool (number of deepening increments before the tool is again cutting a new fresh unrelieved groove) depends on its position during the drum revolution, so that its average contribution to the forces and torque on the drum can only be computed by integrating the forces on the tip over the rotation angle from its starting point until it exits the rock. Let Θ be the angle between the vertical (start of cutting) and the current pick position (see Figure 11), e.g. the rotation angle of the pick in the rock. The depth of cut is then given by the
190
Mechanized Excavation
equation d = dmsin6 The various 'sectors' on the drum which define the breakout cycles can be determined as a function of the 'sector index', p, e.g. the number of deepening increments after the first cut. Sector p is given by the following expression dj(p +\) S/kdm The above equations demonstrate that the cutting sectors depend on a variety of parameters, such as the spacing S between adjacent cutting lines, the rotational speed and the rate of advance (the latter two affect dm). In addition, each sector consists of two areas, symmetrical about the horizontal. For complex designs such as staggered vane sections or clearance rings (Figure 9b, for instance), the above assumption that neighboring cuts are of similar depths is often not valid, because the drum has advanced significantly between the time a pick is cutting at a given angle and that when a neighboring tool comes to the same position. The calculation of dc is then more complex than that used above (dc is no longer equal to S/k). It is therefore often impossible to account for the kinematic analysis of the picks without the use of a computer which performs all calculations, cutting line by cutting line, looking for interacting neighbors of each pick at each position. It is the author's experience that any calculation which does not take into account the above detailed change of cutting conditions as the picks rotate about the drum axis would yield misleading results. However, for comparison purposes, it may be sufficient to only look at what happens at the position of maximum depth of cut. Such work has been published [17], to compare breakout patterns on drum shearers, in order to select optimized designs. Even so, it may be erroneous to decide on the optimized design at maximum depth of cut, since the breakout pattern itself may not be the same for other rotation angles of the drum (especially for complex designs). For instance, a cut deepened twice before breakout does not show the same shape (in three dimensions) as a single unrelieved cut of similar projected surface. In addition, the forces are quite different, the former cut resulting in efforts approximately three times higher than the latter. The correct method to accurately quantify forces on a cutting head thus impliesfinding,for each pick during its revolution (e.g. for a suite of positions along the drum-rock contact): (i) the elementary rock volumes removed by the cutter (geometrical part); (ii) the breakout cycle, together with the major interacting cut, hence the average forces required to complete the cycle, which should be accurately computed, bearing in mind the minimum and maximum possible values (mechanical part); and (iii) the contribution of the forces on the bit to the overall efforts supported by the drum. This method of calculation rests on the above detailed kinematic analysis, and it holds for all types of tunneling machines as well (roadheaders or TBMs).
8.3.1.2 Roadheaders in axial phase In the following section, an implicit assumption, for simplification purposes, is that the picks are placed on one or several spirals originating from the nose end of the head, with a fixed number of picks per cutting line. As mentioned above, roadheaders are faced with two situations during their cutting sequence: an axial phase, when the drum is advancing parallel to its rotation axis, and a lateral phase, when its direction of advance is perpendicular to the rotation axis. For boom-type machines, the axial phase
Theoretical and Practical Rules for Mechanical Rock Excavation
191
is the penetration into the rock face (also called sumping or milling), and the lateral phase corresponds to the lateral sweeping of the cutting head (also called shearing). A good analysis of the kinematics involved in these various situations has recently been delivered by Cordelier [16]. In most cases, the penetration phase is the critical one (during the total cycle of the machine) in evaluating the machine performance. If the machine is not of the boom type (Alpine AM 50 or AM 100, equipped with transversal heads for instance), this phase is kinematically similar to that of a continuous miner, except that the head is but a clearance ring (virtually no vane section). Section 8.3.1.1 therefore applies. Attention will thus be focused here on boom-type machines penetrating into the rock face. In this situation, the advance per revolution is still Ka/Kr, but the picks cut circular lines perpendicular to the direction of advance (hereafter numbered 1 onwards, from the 'nose' of the head to the body of the machine). As a result, if n is the number of picks per line, the 'apparent' depth of cut for each tool is constant during the head revolution and is equal to da = VJn-Vx In fact, as explained by Cordelier [16], this must be adapted because the picks do not work in the same way, depending on their position on the head, so that two groups must be distinguished (Figure 12). The first group consists of the picks laterally inclined by less than 45°. For these picks, the above depth is the actual depth of cut (d = da). Usually, due to design constraints, the end picks (placed on the nose of the head) can only cut deepened grooves. However, the next picks normally enjoy a good breakout pattern because the previous ones have created a free surface towards the center of the face (Figure 12). The depth of cut is therefore not a variable parameter here, and the forces on the picks depend basically on the inclination (tilt) angle and the breakout cycle. The main difficulty is to determine the critical depth dc for each pick, hence the number of deepening increments in the breakout cycle. If Zi is the offset of the picks on line i (e.g. the distance between the tips of the picks on line i and those of the tools on line i — 1, projected on the axis of the head), af their inclination angle, and rf is the radius of this line, then the following applies d = da = VJn-Vr S = dc =
ri-ri_l dc(S,Zi,0Li9k)
The second group of picks includes all the tools tilted at an inclination angle over 45°. In this situation, as shown in Figure 13, the picks actually cut in the direction given by their orientation, e.g. at an angle to the direction of advance. Ultimately, for the last picks on the head, the depth of cut may well be oriented perpendicular to the advance. The depth of cut is then best defined by comparison of the current groove with the Va< 1
F.rst spiral χ
yThird
1 X~ VV'*- _ I
'
s
P i r al
iC\
/xS~><\ Second spiral
»1 > * - - " v; />V< — ■-'s—^
Offset
-K. Line /'
Figure 12 Design of a boom-type head
192
Mechanized Excavation
Figure 13 The two groups of picks during the axial cutting phase of a roadheader
previous one in the neighborhood. Figure 13 shows that this only depends on the radius of the grooves being cut. More precisely, a good approximation of the depth of cut for each pick is the difference in radius between the radius of the previous groove and that of the groove being cut, which only depends on the head design. Since the picks cannot cut in a previously created groove (except for the very unlikely situation when the advance per pick per revolution coincides exactly with the spacing between lines), no deepening effect is observed and good breakout patterns are usual. As a matter of fact, the spacing between two interacting cuts is given by the advance per pick per revolution, since each pick interacts with its own previous groove (cut during the previous revolution) or with that cut by the previous pick of the same line, so that à =
ri-ri_l
S = VJn-Vr dc % S/k (negligible influence of α,) where r, is the radius of cutting line i. The above considerations show that the kinematic analysis of roadheaders in axial penetration is not very complicated, as long as proper attention has been given to separating the picks into two groups before computing the forces on each tool. For complex designs, such as staggered spirals, the only change is on the first group of picks, the calculation of dc being slightly modified (generally simplified as the offset zt is reduced). Finally, if the head is working on half its surface (transverse heads, during the sweeping phase), the only important change from the above is that the nose picks have a much better breakout cycle since there is a free surface nearby which reduces the forces substantially. 83.1.3 Full face tunnel boring machines A pioneering kinematic analysis has been published by Lebrun [6]. For this type of machine it is particularly simple, as the disks cut circular grooves of a constant depth. As a matter of fact, the advance per head revolution is again VJVX and, if n discs are cutting in the same circle, each one of them is then rolling at d = VJn-Vr For each line, the kinematic analysis is therefore reduced to determining the cutting cycle of each circular line, which obviously depends on the spacing with the next line (the next line is defined after the sequence of attack of the disks into the rock: from the center to the outside diameter of the
Theoretical and Practical Rules for Mechanical Rock Excavation
193
tunnel, or conversely). For instance, if the disks are cutting from the center outwards, each cut is relieved by the adjacent one towards the center of the tunnel. The head design is often such that there is only one disk per line, with one or two spirals. When there is one spiral only, the depth of the interacting adjacent cut is nearly the same as that being currently created. Consequently, if S is the spacing to the relieving line, the first cut has to be such that d > dc (dc = S/k) to interact with the neighboring cut and remove the ridge of rock between them. In most situations, as TBMs are used in hard rocks, this does not happen and d is well below dc. Since the cutting speed of the outer disks is much higher than that of the central ones (thereby exposing the former to faster wear), it is advisable to reduce the spacing as much as possible when going from the center to the periphery of the shield, in order to relieve the outermost cutters. Each cut has thus to be deepened a number of times p such that pd
<(p+
\)d
The forces (mostly rolling and normal) therefore increase for each disk, with constant depth increments, until the interaction between lines occurs, and then the cycle starts again. When the head design is such that the disk-bearing spirals are staggered (which implies that there are at least two of them), thereby reducing the line spacing, each disk is relieved by a disk of another spiral, but the depth of the corresponding line is approximately half that being cut. The critical depth dc must then be modified (see Section 8.2.1.2), but the same method of calculation as above applies.
8.3.2 Theoretical Model of Drum-type Heads Thefirstpublished works on the relationship between the forces on a machine head and the rate of advance were based on empirical correlations obtained experimentally. More recently, Lebrun [6] successfully managed to model the forces and torque on a drum analytically, with wide simplifications as to the pick-rock interactions and the head design (particularly the clearance ring). Encouraging results were, however, obtained from this preliminary investigation work [15, 18, 19] which led to an important research program dedicated to the simulation of pick cutting machines. Thanks to the above described research findings, it is now possible to quantify the forces and torque on a given cutting head, starting from the tool-rock interaction laws and the kinematic analysis of the investigated head. This type of calculation is the main object of this section, where it will be shown that the necessary simplifications for analytical calculations can be avoided (and the computation be carried out quickly) with the assistance of a microcomputer. The cutting curves can then be accurately quantified and used as the link between machine specifications and performance. A simple example is detailed, whenever possible, to illustrate the philosophy of the calculations.
8.3.2.1 General method of calculation In the section about kinematic analysis, it was explained that each pick goes through a series of different sectors where different equations apply to calculate the forces on its tip. For a given position, identified by the rotation angle 0, the depth of cut d = dm- sinö and the critical interaction depth dc (depending on the breakout pattern) define the cutting sector, hence the number of deepening increments p of the breakout cycle (see Section 8.3.1.1). Each force component can then be averaged as /=
ΓΤΤΣ/Μ.0
where/(rf, i) is the expression of the force component for the ith deepening increment of depth d. Besides, this average value off lies between a minimum and a maximum value defined as
mO)^f^f(d,p) These two extreme values will hereafter be referred to as/ min and/ max . The former occurs when the pick is cutting a fresh surface, e.g. at the beginning of the breakout cycle (that is to say, just after the interaction with a neighboring cutting line), whereas the latter corresponds to the last deepening increment, e.g. at the final step of the breakout cycle.
194
Mechanized Excavation
On drum shearers or continuous miners, the number of picks is usually high (typically above 50). It is therefore generally correct to use the average value of the work delivered by each tool (for each position) in order to compute its contribution to the overall forces and torque on the drum. As illustrated in Figure 11, this contribution can be expressed in terms of the horizontal, vertical and lateral forces (respectively called H, V and L) and torque (called τ) as follows H = Fc-cos0 + F n sin0 V = Fc-sin0 - Fn-cos0 L =
±F,
τ = F c D/2 The sign of L depends on the side where the relieving interaction occurs. Sign conventions are adopted here as follows: vertical force positive downwards (action exerted by the drum on the rock), thrust force positive forwards, lateral force positive from the clearance side to the other end of drum (this force balances out for continuous miners), and torque always positive. For each cutting line /, the average contribution to the forces on the drum can then be written as horizontal force (thrust F)
n ΓΩ Ff = — / 2 * Jo
vertical force (reaction Rv) lateral force (reaction RL) rotational torque (called T)
n f°
RL = — \ L
T, - "- Γ,
where n stands for the number of picks of the line and Ω is the arc of the drum in contact with the rock (Ω is 180° for the forward drum of a shearer or at the end of the penetration phase for a continuous miner). In the above equations, use has been made of the coefficient (ηΩ/2π), which expresses the average number of picks working at any time during the drum revolution (on each line). The total effort on a cutting line is obviously equal to the single force multiplied by this coefficient. Finally, to compute the total forces and torque on the drum it is necessary to sum the contributions of the cutting lines
F = Σ*Ί i
Ry = J > , ί
RH
= Z«i i
T = ΣΓ, I
This type of calculation can easily be performed by today's microcomputers, as will be shown later. Moreover, as explained for the single tools, the above efforts can be ranged between minimum and maximum values, derived from each minimum and maximum contribution to the drum. The resulting expressions are likely to be respectively well below and above the actual values. Much more interesting is the difference between them. When Fmax — F min is small with respect to F, this means that the efforts are stable during the drum revolutions. The vibrations are kept to a minimum and the machine is working properly. On the other hand, when this difference is relatively high, the efforts will fluctuate a great deal, due to picks oscillating between good interactions (minimum force) and deepened cuts (maximum force). The level of vibrations can then be expected to be important. This point will be discussed later, after the presentation of a simplified example.
Theoretical and Practical Rules for Mechanical Rock Excavation 8.3.2.2
195
Application to a simplified example
In order to practically illustrate the above calculations, a few simplifying assumptions are made for the coming subsection. They are mainly as follows. (i) Pick-rock interaction: the cutting force for the single unrelieved cut is expressed as Fc = Ad. The normal force is Fn = Kn- Fc, with Kn constant. The interaction effect of a relieved cut is assumed to be limited to the removal of the ridge of rock (reduction in forces negligible), and the lateral force is given by Fx/Fc = (1 — dc/d)/2 (for d> dc). The deepening cut results in the multiplicative coefficient AF(p) = p+ l,until p = 3, e.g. Fe{d,P) = AF(p) Fc(d,0) = (p + l)Ad(A¥(p) = 4 for p > 3). No pick is laterally inclined (otherwise, the effect of the inclination is neglected). (ii)Drum design: the example illustrated in Figure 9(a) is chosen here. The vane picks are placed on nonstaggered vane spirals, with a constant spacing S between lines. The breakout pattern is then simple, with each pick relieved by that of the same vane, towards the clearance ring. Clearance picks are few, regularly placed and only slightly inclined. They will therefore be considered as similar to vane picks. (Hi) Machine specifications: the machine is a singls-drum shearer. The available torque is T0 (related to the available power WQ by the equation W0 = π · Γ0 * Kr/30) and the available thrust is F0. These assumptions are basically those formulated by Lebrun [6] in his pioneering work. The cutting sectors are now easily defined by the constant critical depth dc = S/k and, applying the equations from Section 8.3.1.1, sector p is such that dj(p+ l)<
Fc =
- L £ F C ( < U ) = - ^ Γ £ ( / + 1 Μ · < ί + Χ4 P+h=o P+lli=o f=4 equivalent to Fc = M · A · d, with
Ad\ J
M(0) = (p + 2)/2 if p < 4 and M(0) = (4p-2)/(p+l)
ifp>3
On each sector, the contribution of each pick to the overall efforts is therefore H = Fc-cos0 + Fn-sin0 = Fc-(cos0 + Xn-sin0) V = Fc-sin0 - Fn-cos0 = F c (sin0 - Xn-cos0) L = 0 or (1 - d/dc) · FJ2
(if d > dc)
τ = FcD/2 Due to the symmetry of the problem about the horizontal line parallel to the advance, going through the drum axis, the factors with cos Θ will cancel out during the integration for Θ between 0 and π. Besides, replacing d by its value dm · sin 0, the contribution of each line to the total efforts can now be expressed as n Γ* n fπ F, = — Hd0 = — K n -M(0M·^· sin2 0d0 2π 2n Rv
./ο
Jo
n fπ
n fπ
= —
27I
Jo
νάθ
RLi = ^ i ^ d 0
= —
2n
Jo
= ^[
Μ(Θ)Ά •d m -sin 2 0d0
ε(θγ A'dm-(\
- djdm'smeysme
n fπ ΒΓ* T{ = — rd0 = — M(0) , 4 i i m D s i n 0 d 0 27t
Jo
47C
Jo
άθ
Mechanized Excavation
196
The above system can be simplified as follows Ft =
(n-A-Kn'dJ2n)r
RVi =
(n-A-dJ2n)r
RLi =
(n-A-dJ4n)n
Tt =
(n-ADdJ4n)
The coefficients Γ, n and Φ only depend on the ratio dm/dc (which in turn defines the cutting sectors). They are expressed as integrals which can be further detailed. All of them can be simplified by integrating on the interval [0, π/2] (and doubling the result) instead of [0, π]. The most simple one is n , as the function ε(0) is 1 when d > dc and 0 elsewhere. Therefore, if dm > dc, n is given by n = 22
Λπ/2
Je 0
(1 -(\-dJ djdm-sin Θ)· sin θάθ
where ö 0 is defined as the limit of the first sector (sector 0), e.g. by the equation d = dmsin90 = dc = S/k Solving the above integral leads to n = 2cos0o -
21(π/2)-eoydc/dm
The other coefficients involve the function M(0), which is constant on intervals [0 P , 0 p - i ] (defining sector p), where θρ is given by
For instance, the equation defining Γ is
which can be rewritten, replacing M(0) by its expression, as
Similarly, Φ can be expressed as the following series
Each of the above integrals can be explicitly calculated, so that the coefficients Γ and Φ can be approximated. As an example, consider a machine moving at a haulage speed of 5 m min" 1 , with a rotational speed of 40 rev min" 1 and a drum consisting of cutting lines spaced every 40 mm, working in a fairly soft rock with k = 2. It is simple to calculate dc = 20 mm and dm = 42 mm. The cutting sectors are then defined on a quarter drum as sector 0 angle (°) 29-90
1 2 14-29 9-14
3 7-9
4 6-7
5 5-6
6 4-5
This typical example shows that the error induced by computing forces, leaving out sectors with p > 5, is very small (only the portion of the drum for 0 from 0° to 5° is not accounted for). This is generally true, so that the coefficients Γ and Φ can be calculated for 5 or 10firstterms of the series of integrals. Figure 14 illustrates this example. Summing the force/torque contributions of the cutting lines leads to the calculation of the forces and torque on the drum. If N is the total number of picks on the drum (e.g. for n picks per line over
Theoretical and Practical Rules for Mechanical Rock Excavation
197
Sector 2 ^Sector I
Sector 0
dc = 2 0 mm d m = 42 mm
Figure 14 Cutting sectors for the simplified example
the entire drum), the above equations yield F = Rv =
(N-A-KJ2n)Fdm (Ν·Αβπ)Γάη
RL = (ΝΆΙAn) ndm T =
{NAD/4n)4>dm
This forms the basis of the cutting curves F = F(4n)> or F = F(Ka) by since Va is easily derived from dm by multiplying it by the rotational speed and by the number of picks per line. As mentioned earlier, it is possible to estimate the minimum and maximum values of the efforts on the drum. The former is quite simple to calculate, since the breakout cycle is simplified to the first single unrelieved cut (see above). The term Mmin(0)is uniformly equal to 1, which reduces the above integrals to Fmin =
^JV^A-sin 2 0d0 =
Rymin = ^A'dm-sin2ede RLmin = ^ j Φ)'A-dm-(l-
=
KnNAdJ4
N-A-dJA
djdm·sinΘ)·sinθάθ
7min = ^ j V ^ D - s i n Ö d Ö =
N'D'A'dJln
The computation of maximum forces and torque is slightly more complex, as M(0) is replaced by the maximum deepening coefficient for each sector, Mmax(0) = AF(p), e.g. for π/2 > Θ > θ0: for θ0>θ>θ1 for θί>θ>θ2 for 0 2 > 0:
AF(p) AF(P) AF(p) AF(p)
= = = =
1 (no deepening in sector 0) 2 3 4 (maximum force stabilized)
198
Mechanized Excavation
It can then easily be demonstrated that the coefficients Γ and Φ are replaced by Tmax and ^max such that Γ
* max
= Γ ■ + 2 *■ min
'
sin2 θ άθ + 2
Ή
sin2 0 d0 + j °sin2 0 d0
Tmax = (π/2) + (0O + 0t + 02) - (sin20o + sin201 + sin202)/2| ^max
=
^min
+
2
3
^
L Jo
^max = 2 +
θ
^
+
2
^
Je2
ö
^
+
^
J»!
ö
^
J
3 - (COS0O + COS θι + COS 0 2 )
The resulting forces and torque are obtained by applying the same set of equations as above (substituting Tmax and # max for Γ and Φ). The detailed calculation of these coefficients shows that the cutting curves are not smooth at points defined by dm = dc/m (m = 1, 2, 3). This is due to the assumed brutal change in breakout cycle (three deepening increments instead of two, for instance, for all picks at the same position) as the pick moves from one sector to the other. Obviously, in actual situations a more continuous behavior is observed. It is therefore necessary to smooth the cutting curves, as illustrated in Figure 15. In the figure, the minimum and maximum curves have been shown as dashed lines. Curves for the thrust and vertical forces are similar in shape (the lateral reaction is somewhat different and can be plotted from the above detailed equations which give RL as a function of n , and n as a function of
djdm).
The minimum curve is a straight line, which results from the simplified equations presented above (r min and # min are constants, independent of dm\ whereas the maximum and 'actual' curves have a similar shape, firstly convex, then gradually linear. Figure 15 shows that the three curves tend to join for high depths of cut per drum revolution (e.g. high haulage speeds). This can be easily understood in terms of cutting sectors. Let dm equal 2dc, for example. The sector with no groove deepening extends from Θ — 30° to Θ — 150°, e.g. the vast majority of the drum contact with the rock. Very good interactions exist between the picks and deepening increments are minimal (sector 1, with only one increment per cycle, covers half the remaining area). The above series of integrals tend to be close to the first term, which itself is similar to that calculated for the minimum curves. As a result, a very low vibration level is observed in such situations: if possible, given the design of the drum, it is desirable to set the VJVr ratio so that dm > 2dc. On the contrary, if dm < dc, then the picks are not relieved during their first cut, even at maximum depth of cut, and deepening is systematic. The average and maximum forces are then significantly
near domain
Maximum advance rate per pick and per revolution
Figure 15 Shape of the cutting curves
Theoretical and Practical Rules for Mechanical Rock Excavation
199
higher than the minimum ones. The resulting vibration level is high, resulting from the difference ^max - Fmin, relative to F. Apart from being a valuable quantitative tool to assess the instantaneous velocity of the drum (see Section 8.3.2.3), the cutting curves thus appear as an excellent way to estimate whether the working point of the machine is satisfactory or not, in order to predict short term changes in performance (high vibrations will lead to worn or broken picks, hence a rapidly worsening situation). They also confirm that, whenever it is practical to cut deep grooves in the rock, energy consumption is kept at a minimum. The volume of rock excavated grows faster than the power required from the machine, which results in an improved specific energy. Once the cutting curves have been established, it is easy to find the instantaneous performances of the machine if the available torque and thrust are known. As shown in Figure 15, two values of dm (or Va) are read off the curves, one for each machine specification. The smallest of the two corresponds to the actual advance rate and indicates which factor is limiting the machine: if the torque curve gives dm smaller than the thrust cutting curve, the machine is torque limited and the actual required thrust is that read from the thrust curve for dm derived from the torque T0. If the machine advances at a speed such that low vibrations occur, the curves can be approximated by the linear minimum curves, so that quantifying the effects of design parameters, such as the number of picks, rotation speed, diameter of the drum, etc. becomes quite easy, with the above equations relating F and T to
8,3.23
Computer simulation of cylindrical drums
The above example is obviously simplified far too much to accurately represent actual situations, although the qualitative conclusions expressed above remain correct. Firstly, the tool-rock interaction has been shown in Section 8.2.1 above to be fairly complex. Besides, forces on the picks normally increase more than linearly when the depth of cut increases. This is partly offset, however, by the lack of reduction assumed above in the forces when the picks interact with each other. The linear shape of the cutting curves at high depths of cut is therefore often observed in actual conditions. Drum design is also much more complex than that assumed above. This applies to the vane section (the spacing of the cutting lines is now generally increased from the clearance side to the arm of the machine; also, the vanes are often staggered, like in Figure 9(b), resulting in a very different breakout pattern), as well as to the clearance ring, where the picks are gradually more inclined, with varying cutting radii and complicated sequences of penetration into the rock. The depth differences in the interacting cuts influence the critical depth dc and hence the deepening sequence of the breakout cycle. On clearance rings, it may also happen that the picks are laterally inclined at angles well over 45°. In that situation, similar to that detailed in Section 8.3.1.2 for the axial cutting phase of roadheaders, the picks are not cutting in a plane orthogonal to the axis of rotation of the drum and the previous calculation of forces does not apply. The correct way to compute the forces is further detailed in Section 8.3.3. Finally, the drum is not always in contact with the rock over exactly half its surface, which removes the simplifications derived from the symmetry provided by a 180° contact arc. In practice, it is therefore necessary to use computer programs which accurately simulate the contributions of the cutting lines one by one for a given machine speed. Figure 16 illustrates an example of a 'good' simulation program. For each drum velocity, the program first calculates the depth of cut for each tool at a set of positions which reflects the drum revolutions in the rock massif (for instance, from 0° to 180°, by steps of 5° or 10°). In so doing, for each position, the program must look for neighboring picks and analyze the breakout sequence. On complex designs, it can often happen that the interacting neighbor is not located on the adjacent cutting line, but further away. Having determined dm and dc, the program can then move to the computation of the contribution of the said pick, keeping the minimum, mean and maximum values constantly in memory. To do so, it must check the pick-rock interaction (mode A or mode B), which itself may change from the first cut (single unrelieved) to the last deepening increment of the breakout cycle. Once the cutting, normal and lateral forces have been determined for each pick at each position, it is simple to compute the contribution of each pick to the effort on the drum during a complete revolution. The next step is to compare the resulting values with the machine specifications. This is dealt with later in this chapter (see Section 8.4 and later). A somewhat different approach has been published in the UK and in Germany [20-22]. The computer programs were aimed at optimizing head design by looking at the following problems: positioning of pick-boxes on the head (C.A.D. problem), assessment of the breakout pattern, and
200
Mechanized Excavation Machine Speed Maximum depth of cut
r_i
Rotation angled
♦ — -
Breakout pattern and cycle
I
Force contribution of pick at position Θ f Average force
NO YES NEW UNE
f I
Force contribution^ of cutting line J
Forces and torque on head(s)
J
Figure 16 Organization of a good simulation program for the modeling of cutting heads
evaluation of force fluctuations during head revolutions. At the time, this work was pioneering research in this area, particularly with the use of microcomputers at the National Coal Board of Great Britain [21] (now British Coal). It turned out to be successful in comparing different designs for a given environment (rock face). The approach is, however, purely geometrical, as the only computation of forces consists in evaluating volumes of rock cut by the pick, for different head positions, and then using very simplified empirical linear correlations to find the resulting values of the forces. As explained in Section 8.3.1.1 on the kinematic analysis of drums, this type of computation, which also leaves out minimum and maximum forces/torque (and hence is a useful tool to assess vibrations), is today felt to be much too simplified given the improvements achieved in the knowledge of pick-rock interaction. The engineer's way of thinking introduced into these programs to tackle the problems associated with mechanical rock breaking, especially when minimizing force fluctuations must, however, be kept in mind for the more sophisticated programs, such as those dealt with in Section 8.4.3. 8.3.3 Theoretical Model of Roadheader Heads The kinematic analysis in Section 8.3.1 has shown that two different situations must be distinguished: axial and transversal cutting of the head, with the advance speed parallel to the rotation axis of the head in thefirstinstance and perpendicular in the second. The two situations are therefore separately addressed here. 8.3.3.1 Axial cutting Two different behaviors are observed, as mentioned in the kinematic analysis: group I picks are laterally inclined at less than 45° and group II picks are inclined at more than 45° from the direction of advance (Figure 13). Figure 17 illustrates the force components on the picks for each group. Group I picks are considered to cut at a constant depth d = VJ(n-Vr) where n is the number of picks per cutting line (circle of radius rh as shown in Figure 13). If the axial
Theoretical and Practical Rules for Mechanical Rock Excavation
201
work is a penetration into the rock, the nose of the head is completely confined and thefirstpicks are continuously deepening their cuts, whereas some breakout can be observed for the next picks. The first cutting line is therefore contributing to the overall forces as follows Fl =
n(M+l)Fn(d)
7\ = nr^iM
+ l)-Fe(d)
In the above, M is a coefficient which expresses the increase in cutting and normal forces in the deepening situation (after stabilization, see Section 8.2.1.2). Vertical and lateral reactions can be neglected because of the symmetry of the problem (contact with the rock is over 360°, and the depth of cut is the same everywhere, so that forces perpendicular to the direction of advance will cancel out). The next cutting lines are evaluated from the breakout cycle. Since the depth of cut is constant during the head revolution, the interaction pattern is relatively simple to determine, hence the calculation of the number of deepening increments before breakout is obtained. This is normally a short cycle (one or two deepening increments), as the picks find a natural free face towards the line closer to the nose of the head. To detail an example, it is assumed here that forces increase proportionally to the number of increments. For the example where two deepening increments are necessary to break the ridge of rock towards the previous cutting line, the averaged cutting and normal forces during the breakout cycle are derived from Fc = lFc(ai9d)
+ 2F c (a„«0 + 3F c (a f , d)]/3 = 2F c (a„ d)
Fn = [F n (a |f
In the above expressions, the three summed terms stand for the forces during the cycle (the first one is the single unrelieved cut, the second one corresponds to the first deepening increment, and the third one to the second increment; for the example, the forces have been assumed to increase linearly with the number of deepening increments). at is the lateral tilt angle of the picks on line i. The resulting thrust and torque are easily derived from the above equations (same method as in Section 8.3.2.2).
y Advance
Group I (Fn is axial)
Groupll
Figure 17 Forces on the picks in axial work (roadheader)
202
Mechanized Excavation
It is thus relatively easy to calculate the contribution of each cutting line to the overall thrust and torque for group I picks. Like for drum shearers, the minimum and maximum values of the forces are derived from the assumption of systematic interactions between cuts (minimum forces) and simultaneous deepening at the final increment (maximum forces). Bearing in mind that the picks are laterally inclined, it is sometimes necessary to carefully study the interaction between adjacent cuts, as the critical depth dc (and hence the breakout cycle) may be different from that observed with picks cutting in the vertical plane of the groove. An implicit assumption in the above calculations is that the head is in contact with the rock over its full circumference. This is not always true (transversal heads in the sweeping phase) and some corrections must be brought about. It is in fact easy to show that the torque is reduced in proportion with the arc of contact, and that a reaction force is recorded, perpendicular to the direction of advance, with the major component being vertical (its direction depends on the rotation of the head). The computation is quite similar to that explained for cylindrical drums (see Section 8.3.2.2). Group II picks require a different calculation. As explained from the kinematic analysis, the depth of cut for picks on line i is now à = rt - *·,_! and the critical depth is constant at dc = VJ(n-k-Vr) The contribution to the torque is unchanged, depending only on the cutting force on the picks, but the contribution to the thrust now comes from the lateral force. Notwithstanding the interaction effect (the corresponding lateral force, directed opposite to the thrust, but often negligible), this lateral component primarily originates from the inclination of the picks. As explained in Section 8.2.1.2, it can reach a value of between 0.3FC and 0.5FC, and F c is itself increased because of the inclination. It must be pointed out that, in this situation, the inclination angle is no longer af but π/2 — a,·, since the plane of the cut is now at 90° from the previous one (see Figure 17). The contribution of group II picks can therefore be expressed as Ff = n-F,(a = π/2 - <χ„ d) = η·Κ,(π/2 - dt)'Fe(n/2 - a„ d) Tt = n-rf ·Fc(a = π/2 - ai5 d) This time, the computation is simplified by the absence of deepened cutting. 8.3.3.2 Transversal cutting In this situation, the kinematic analysis developed for a cylindrical drum applies, and the computation of forces is similar. As a matter of fact, a roadheader cutting transversally behaves just like the clearance ring of a continuous miner or a drum shearer. Group I picks, inclined at less than 45°, can be treated in exactly the same way as explained in Section 8.3.3.1. Group II picks, like those for axial cutting (see above), do not cut in the same direction. The reference plane of their grooves is not orthogonal to the rotation axis of the head any more, and the contribution to the total torque and thrust can be computed as in the previous case (group II picks, during axial cutting), with inclination angles changed to π/2 — a,·. 8.3.4 Theoretical Model of TBM Shields The computation of forces on the shield is quite simple, as there are only a torque and a thrust, straightforward to calculate from the individual forces on the disks: i
Τ = Σ^-r, i
where F m and F ri are the normal and rolling forces of disk i (see Section 8.2.2), and rt is the cutting radius (see Figure 18).
Theoretical and Practical Rules for Mechanical Rock Excavation
203
F r , Fn and FL are constant for all positions
Figure 18 Forces during a TBM shield revolution
For each cutting line, the depth of cut is constant during the shield revolutions (d = dm) and the breakout cycle has been shown to depend on the spacing with the relieving line (see Section 8.3.1.3). The number of deepening increments to complete the breakout cycle, p, is given by P < djdm < p + 1 with dm = VJn-VT (n is the number of disks on the same line, often equal to 1), and dc = SJk if the cuts are of similar depths (S,· is the spacing with the line relieving disk i). Each cutting line therefore has a constant breakout cycle during the shield revolutions, and no integration over the tool position is necessary, like it is for drums. Given p and the depth of cut, the mean forces during a complete breakout cycle are computed as
^ = 4 τ Σ F Ad, m) where F(d,m) is the force at the depth of cut d for the deepening increment m. As a simple approximation, it is common to write, as for picks F(d,m) = (m + 1)-F(d,0)
if m < M or m = M
F{d,m) = (M + 1)· F(d,0)
if m > M
The stabilizing threshold parameter M usually ranges between 5 and 10 for the shallow depths of cut found with this type of machine. The forces are then easy to compute, as are the values of torque and thrust, according to the above equations (for instance T — Frmean (d9m)·£/,·). Again, as for pick cutting heads, a minimum and a maximum value for the torque and the thrust can be calculated. With the above conventions, they are expressed as follows for each disk Fmin(d,p) = F(d,0) FuJid, P) = (P + 1) ' F(d, 0) (if p < M + 1) FBM(M) Usually, dm is much smaller than dc, hence p > M, so that the torque and thrust are given by Fmin = NFn(d,0) 7'»in = Z ^ ( * 0 ) T l = Fl(0)-Xrl i
i
Fm,x = N-(M + l)Fn(
i
204
Mechanized Excavation
If the spacing between lines is constant, it is simple to derive Ση = (N + n) · D/4 {D is the diameter of the shield). 8.4 VALIDATION OF THEORETICAL MODELS AND MACHINE SIMULATION The next step, after the cutting head has been modeled, is to compare the resulting values of the forces and torque for a given advance rate to the machine specifications. In this section, after a brief presentation of experimental validations of the above models, a method to correlate the machine performance to its specifications, based on the approach illustrated in Figure 2, is suggested, followed by a presentation of microcomputer programs that are based on this method. Practical examples then demonstrate the value of finding the machine best adapted to given conditions or, alternatively, estimating the maximum output of a given machine. Microcomputer simulation is shown to be of invaluable assistance to this problem, including the assessment of vibrations due to insufficient machine specifications or tool wear. 8.4.1 Experimental Validation of Theoretical Models Because of their complexity, it is essential, at this stage, to validate the above theoretical concepts. This can be done in the laboratory, but, ultimately, the validation must come in the field, on a full scale basis. It is the purpose of this section to gather the experimental evidence, published over the past 20 years, in order to check the validity of the above presented models. 8.4.1.1 Laboratory scale Full scale experimental tests in a laboratory are costly. As a result, only large organizations, such as the NCB/MRDE in Great Britain, the Bergbau Forschung in Germany or the CERCHAR in France could carry out such experiments in the 1970s or early 1980s. Reduced scale testing seemed to hold promise and was initiated in the UK in the early 1970s [23]. It was apparently abandoned shortly afterwards. Later, in 1980, a reduced scale cutting head was built in France [15, 18] after successful experiments on reduced scale single picks (Figure 19). Simultaneously, small scale tests of disks were started in Switzerland, followed by scale tests of a TBM shield [24]. In the USA, little experimental work was done (to our knowledge), apart from linear cutting tests with a single pick, until the introduction of the 'in-seam tester' by the U.S. Bureau of Mines [25]; however, this is not really a cutting head. It is difficult to find well-documented experimental results in the available publications. The work carried out in France [15, 18, 19, 26] is an exception to this finding. (i) Cylindrical drums Two sets of experiments were run from 1981 to 1983. The first one consisted of tests carried out simultaneously at 1/6 scale and at full scale [18], on two drum designs, with two different types of picks. No clearance ring was used, as the rock was free on both sides of the drum. The objective was to test the concept of cutting curves explained earlier and check the characteristic shape illustrated in Figure 15. As can be seen in Figure 20, this was well confirmed, and the simplified calculation detailed in Section 8.3.2.2 was successfully performed in agreement with the experimentally measured forces. The second set of experiments was carried out both in the field, (Lorraine iron mines, eastern France) with an instrumented continuous miner and in the laboratory with a 1/4 scale replica of the continuous miner half drum [19]. Again, with the three different designs tested, the cutting curves were correctly predicted by the theoretical model (Figure 21), both qualitatively and quantitatively. Another aim of this research work was to investigate the correspondence between full and reduced scales (force and torque ratios at the two scales). It was found that the only discrepancy from the expected ratios (derived from a dimensional analysis) was due to the well-known size effect of the rock strength (the smaller the scale, the harder the rock) or major discontinuities in the rock (rock specimens were from the same origin at the two scales). It was thus possible to correct the theoretical full/reduced scale ratios by a coefficient depending on the continuity of the rock. For instance, if s is
Theoretical and Practical Rules for Mechanical Rock Excavation
205
Figure 19 View of reduced scale testing rig at the Paris School of Mines (courtesy Paris School of Mines) (a)
(b)
Figure 20 Experimental results on cylindrical drums (after Déliac [15]): (a) full scale; (b) reduced scale. T= torque and F = thrust. The various curves refer to different rock samples and are only presented to show the agreement between calculated and experimental values
the scale ratio (4 or 6 here), the expected force ratio is s2 and the torque ratio can be shown to be s3. The correction coefficient x (between 0.1 for continuous rocks and 0.3 for porous or discontinuous rocks) is such that these ratios become s2~x and s 3 _ x respectively. This quite encouraging result boosted research in this area, and the reduced scale rig was adapted to simulate clearance rings and roadheader heads (Figure 22). (ii) Roadheader heads Quarter scale axial heads were manufactured with special features in order to be able to continuously move the pick-holders, thus simulating a variety of designs with one single head (Figure 22). Tests were run in different rocks [16, 26], and an example of the cutting curves is shown in Figure 23, together with the computer simulation based on the theoretical model. The agreement is very good for the torque, whereas the computation of the thrust force tends to slightly overestimate the actual values.
206
Mechanized Excavation (a)
2) Δ
ΔΙΙ ΔΙΟ Δ9 Δ8
_30°
'8,
r ^ k . 5 Δ17
4
7.5, / ' Δ
Δ4
|Δ3 12x7.5
Lîî
ΔΙ4
(b)
Δ7 Δ6
Δ5
Δ9
4
Δ2
-H be|Λΐ5
Δ3
^
Δ8
^
Δ3
'
.1
200 h
I4À
J3J_
Δ9
Δ1
1 3)
i
| 5
.
Δ2
Δ 10
Δ5
<
*
50
Ί
100 |"
_^ Δ3
Ό
Δβ
' 3Δ
3
4 5 d m (mm)
6
Figure 21 Cutting tests of a 1/4 scale continuous miner half-drum (after Déliac [19]): (a) the three designs; (b) the cutting curves. Dimensions are in mm. Effort = thrust/compressive strength of rock (mm2)
Figure 22 Adaptation of the Paris School of Mines rig to roadheader heads - view of the head during a cutting experiment (courtesy Paris School of Mines)
8.4.1.2 Full field As already mentioned, full scale tests have been carried out and documented on a fully instrumented continuous miner working in Lorraine (eastern France) iron mines [19]. These successful tests (see above, for the reduced scale simulations) confirmed the high level of vibrations observed when the maximum depth of cut per drum revolution dm is in the range of the critical interaction depth dc, and also the influence of operating parameters, such as rotational speed. The results will be used later in this chapter (Section 8.4.2.3), as a basis for the optimization of drum design and machine specifications.
Theoretical and Practical Rules for Mechanical Rock Excavation (b)
(a) 250
~
200
1-
150
Z
«A 3
JC
H
50 E
z ^ττιαχ
100
40
§ 30 CT
l·- 20 —
A— A —
A
Mnin
50 1
0
207
10
I 20
10
1
1
1
30
40
50
0
Vn(cm min"1)
Figure 23 Experimental and computed cutting curves for a two-spiral, one pick per line roadheader head (after Cordelier [16]). The three curves refer to the computed maximum, minimum and median curves. V = penetration speed
300
200
o
CD
0
50
100
150
Depth of cut, cfm (mm)
Figure 24 Measured and calculated values for the boom force (thrust F) of the instrumented continuous miner documented by Roxborough and Pedroncelli [27] (point-attack picks at 100 mm spacing, sumping phase)
An interesting study was carried out in South Africa in 1981-1982 on an instrumented continuous miner in coal mines [27]. The publication is well documented and gives enough information to compute forces with the model presented here. Figure 24 shows the experimental data obtained by Roxborough and Pedroncelli [27] for the boom force (thrust) during sumping, with point-attack picks cutting at a constant spacing of 100 mm and a rotational speed of 14.7 rev.min"*, together with the cutting curves estimated with the theoretical model from the available data on rock properties, pick shape, and a simplified clearance design. As can be seen in the figure, the agreement is fairly good between the measured and calculated values. As far as the torque is concerned, the agreement is not as good, the calculated values lying somewhat higher than the measured ones. The authors, however, argue that their values for torque at this spacing might be too low, particularly at shallow depth of cut, as the cutting curve could yield a negative intercept on the ordinate (torque axis), which is physically unacceptable. Torque values recorded at a 150 mm spacing, for instance, are well above those at 100 mm, whereas the two curves should cross in the vicinity of dm = 50 mm. In most instances, Roxborough and Pedroncelli only observed linear cutting curves, with positive intercepts, in line with the theoretical cutting curves approximated for dm> 1.5 dc (see Figure 24), e.g. when the mean actual cutting curve gets close to the minimum cutting curve. It is therefore reasonable to say that the theoretical modeling presented in this section is a good predictive tool to evaluate the forces and torque on a cutting head for a given advance rate of the machine. More experimental evidence would be interesting to further validate the model, particularly with roadheader heads and full face boring machines, but this is still scarce even today. A great deal of information is required to calculate forces: data for the tool-rock interaction (rock properties, bit design), design of the head (spacing of lines with number of tools per line, wrap angle, inclination of picks, radii of cutting lines at the tips of the picks) and rotational speed.
208
Mechanized Excavation
It is also essential to stress the importance of correctly assessing the behavior of the head at shallow depths of cut (low advance speed), as this is typical of hard rocks, where the penetration of the head is slow and vibrations are high. Most publications seem not to address this specific area of research, focusing on high depths of cut, where a high degree of linearity has been shown to exist for the force-advance rate correlations. This linearity vanishes at low advance rate and the calculation of forces and torque is much more complex. The computation of minimum and maximum forces is a powerful way to estimate the level of vibration. As a conclusion to this section, it is now possible to accurately simulate the behavior of cutting heads so as to determine the best design or selection of a machine in a given environment. This modeling highlights the importance of good tool interaction during cutting. 8.4.2 A Method for Determining Machine Performance 8.4.2.1 General principle of the method Determining machine performance can be done in two ways: by finding the advance rate of a given machine in a given environment, or by looking for required specifications which will achieve a desired production rate. In both cases, the most simple method is illustrated in Figure 25, which shows how to move from advance rate to power on the motors once a model of the cutting head(s) is available. The principle is basically to start from a given advance speed. The advance per revolution is then easily determined, and hence the forces on the head after the kinematic analysis and the integration of pick/disk force contributions over the head revolution can be calculated. The main problem is now to introduce the available thrust and torque, and possibly the maximum vertical or lateral reactions that the machine can tolerate. This is not always easy, as manufacturers' data seldom include the necessary information. This matter will be dealt with through practical examples in the next sections. When the available forces and torque have been calculated, they are compared to the efforts required from the head(s). If they are higher, the machine can go faster and the advance speed is incremented before the calculation is reset. If one of the available specifications falls below the head requirement, then the machine cannot achieve the rated advance speed, which must be reduced. The calculation stops when one of the available specifications equals the corresponding required effort on the head (or the two drums for a shearer) and when the other machine specifications are Machine Speed ? Maximum depth of cut
Force contribution of each cutting line T Forces and torque on head(s)
Figure 25
Principle of the performance-specifications calculation
Theoretical and Practical Rules for Mechanical Rock Excavation
209
Actual advance of machine Limitation is by torque
Advance rate
Figure 26 Selection of advance rate from the cutting curves and the machine specifications
higher than the forces/torque required. Usually, the limiting specification (hereafter referred to as the limiting factor) is either the thrust force or the torque (e.g. the power available from the motors), and the machines are designed to be able to withstand vertical and lateral reactions by appropriate hydraulic devices. In other words, as already mentioned in Section 8.3.2.2, if the cutting curves could be drawn with the available torque and thrust from machine specifications plotted on the drawing, the actual advance rate would be the smallest of the two values read from the graph (Figure 26). There is no point in increasing the power on the machine if it is thrust limited, and the vibration level only results from the limiting factor. 8.4.2.2
Application to drum shearers
As an example, the calculation is detailed here for a double-armed drum shearer. For the sake of clarity, the machine is supposed to be powered by a single motor. It is assumed that the total available electrical power We is known, as are the characteristic curves of the hydraulic power pack which delivers the thrust force (haulage force). For the sake of simplicity, it is also assumed that the vein is near horizontal. Let μ be the friction coefficient along the conveyor system. Figure 27 illustrates a simplified balance of the forces and torques, which results in the following set of equations Wl = 7\·ΚΓ·π/30 W2 =
T2Vrn/30
Wx + W2 + Wh = EeWe JI-(F W - Rwl + ÄV2) + F 1 + F2 = EhFh In the above equations, the subscripts 1 and 2 refer to forward and rear drums respectively, £ e and Eh are the mechanical efficiencies of the electrical motors and the hydraulic unit respectively (to be distinguished from the nominal efficiency, as explained below), Wh is the electrical power required from the hydraulic unit to supply the haulage force F h , and F w is the weight of the machine. Wh is calculated from Fh after the characteristic curve. The coefficients μ, Ec and Eh are quite important and cannot always be easily determined. Firstly, it must be borne in mind that the forces on the drums sometimes fluctuate greatly (particularly at low haulage speed, as shown in Section 8.3.2.2). Consequently, if We is the maximum rated power of the machine, it will have to overcome the peak values of the torques, resulting in an average delivered power significantly less than the rated power. In addition, the motor has an intrinsic efficiency (nominal efficiency, usually around 95%). Combining the two effects, the resulting efficiency Ee is then found to be between 0.6 and 0.8 in most situations. The hydraulic unit efficiency Eh is normally higher, in the vicinity of 0.8. Finally, the friction coefficient is highly dependent on the face condition. The theoretical value of a metal to metal friction coefficient should range between 0.2 and 0.4. In fact, because the haulage
210
Mechanized Excavation Seam
Figure 27 Main forces and torque on a double-armed drum shearer
conveyor is never perfectly straight and because some broken rock is virtually always found between the moving machine body and thefixedconveyor, μ is much higher than the theoretical value, lying between 0.6 and 0.9 (a value of 1 has even been recorded by the author in a phosphate mine). In the specialized module of a computer program, dealing with the evaluation of available specifications, the efficiencies, the friction coefficient and the Wh =/(F h ) curve must be input in addition to the power and weight of the machine. As explained above, this requires some expertise and the manufacturer's information is seldom sufficient to work the data out. From our experience, modern drum shearers are generally limited by the available mechanical power (or torque), rather than by the haulage force. This is a good point, as long as the available torque still ensures a fair working point (away from the high vibration zones), since the effect of wear will tend to increase the thrust requirement much more than the torque requirement. 8.4.2.3 Application to continuous miners and roadheaders The determination of machine specifications is somewhat different here. It often happens that two motors supply power, one for the hydraulic units (jacks) and the rock loading and conveying, and the other for the head revolution (torque). Let Wt be the electrical power used to provide torque on the cutting head. The thrust is supplied by a hydraulic system taking advantage of the weight of the machine (when this is insufficient it is sometimes necessary to anchor the body of the machine to the sidewalls or to the roof in order to increase stability and thrust). It is thus convenient to express the thrust force F as the product of the weight and an 'efficiency coefficient' EF. The vertical reaction Rw is oriented upwards for a continuous miner by a correct setting of the drum rotation (as in Figure 1), so that the weight of the drum is used to balance the reaction. This does not hold for roadheaders, since the sweeping phase (shearing) is done in two opposite directions for the same rotation. In one direction, the weight of the boom will balance the vertical reaction, and in the other direction, the two will add to generate a strong downwards force. This might be a limitation to machine performances, as it can only be balanced by hydraulic jacks, which supply a limited force. In any case, the critical working phase of a tunneling machine is the sumping phase, where such a problem does not occur. Forces and torque are substantially reduced in the shearing phase. During sumping lateral reactions cancel out for reasons of symmetry, so that the performance of the machine is basically dependent upon the available torque and thrust. This is expressed as follows £ e we =
TVrn/30
Wt + Wh + Wx = Wt EF'K
= F
where Wt is the total rated power of the machine, and Wh and Wx are the power required to activate the hydraulic devices and to load/haul/convey the broken rock respectively. The efficiency Ee is here in the range 0.9 to 0.95. EF must account for the hydraulic system efficiency. As a rule of thumb,
Theoretical and Practical Rules for Mechanical Rock Excavation
211
a value of 0.5 is often acceptable. However, most of this information is seldom easy to find in manufacturers' booklets or brochures. It is our experience that with tunneling machines the limiting factor is more often the available thrust than the torque (or mechanical power). This situation is worse in hard and abrasive rock, due to the dramatic influence of wear on thrust requirements (see Section 8.4.4). 8.4.3 Computer Simulation of Pick Cutting Machines In Section 8.3.2.3, it was explained that two generations of computer program exist today for pick cutting machines. The first one, initiated in Great Britain and Germany, is based on geometrical concepts and has been shown to be too oversimplified to predict forces accurately. Besides, the vibration analysis with these programs does not account for force fluctuations between a minimum and a maximum value for each pick position. The second generation of programs is more recent and has been intensively developed at the Paris School of Mines [2, 16, 26]. It starts from a kinematic analysis of the picks duty (variation of depth of cut, breakout pattern, frequency of deepened cuts, etc.), then calculates the force contributions of each cutting line according to the method described in Section 8.2.3 (illustrated in Figure 16), and eventually delivers a 'working point' for the machine, including production rate and vibration level, as well as a detailed analysis of the force/torque distribution across the cutting head (Figure 25). This set of programs is the basis of the coming section, and will also be used to present practical examples of machine selection, optimization and design. 8.4.2.1 The P.C.DR UM program The P.C.DRUM program was developed to assist the optimization or design of drum shearers [2] (P.C.DRUM and P.C.MAP are proprietary names, protected by copyright). It is organized in a modular way. As shown in Figure 28, the central part of a working session consists of a data files management system. Data files are independent and consist of the following. (i) Machine data: number of drums and motors, power from each motor, with the corresponding efficiency, power consumed by the hydraulic unit, if any (it is assumed here that Wh is roughly proportional to F h in the range of investigated values), power of the motor supplying the haulage force. (ii) Machine environment: weight of the body, friction coefficient along the conveyor, efficiency of haulage system (£h). (iii) Drum specifications: rotational speed, diameter. (iv) Vane section design: number of spirals (vanes), number of picks per line, wrap angle of vanes, spacing of lines (not necessarily constant). (v) Clearance design: angular position of the first gauge pick with respect to the neighboring reference vane pick, characteristics of clearance lines (spacing parallel to the axis of the drum,
Figure 28
Organization of P.C.DRUM
212
Mechanized Excavation
vertical offset, e.g. radius difference between the line and the previous one, inclination/tilt angle of the picks), number of sequences (groups of picks regularly placed on the ring, usually equal to the number of vanes: for instance, four sequences of nine picks, located at 90° from each other), with the number of picks per sequence, and sequencing of picks for one group (angular positions). (vi) Pick characteristics: the program asks for the type of pick (drag bit, V-shaped chisel, point-attack and heavy duty forward attack), then expects data from questions depending on the type of tool (for instance, the tip angle, its effective width at the junction with the body of the pick and the angle of attack for a pointed tool). A wear index is also input. This parameter varies from 0 (perfectly new pick) to 5 (when the tungsten carbide is worn to a stage where the steel body is in contact with the rock). Picks in good condition are usually entered with a wear index of 1 and average picks on a working drum shearer are usually considered at a wear index of 2. Higher values of the wear index are typical of poor pick condition. (vii) Rock characteristics: uniaxial compressive strength, internal friction angle (from shear tests), fracture toughness (mode I, e.g. Klc). If the latter is unknown, the program estimates a value from the uniaxial compressive strength and the friction angle. It also makes use of noncompulsory information (if available), the brittleness index. This ranges from 1 (soft granular/porous rocks, with very little chipping when the pick is cutting) to 5 (very brittle rock: mode B tool-rock interaction is dominant at shallow to medium depths of cut). P.C.DRUM estimates K]c using this index and the uniaxial compressive strength, or, alternatively, determines the brittleness index from the knowledge of <7C and KIc. Finally, the program suggests computed values of the rock interaction coefficient k (modes A and B), which may be changed by the user to any appropriate value. (viii) Seam parameters: dip angle, total thickness or thickness of up to four different strata (with different mechanical properties), relaxation factor (a coefficient which accounts for the reduction of rock strength, on the conveyor side, when the face is highly stressed). As can be seen from the above list, the amount of input data is large and, in order to develop user-friendly features, the program prompts for typical values whenever possible. Once a complete set of data files have been entered (for instance, machine + machine environment + two similar drums, e.g. design + vane + clearance + picks + seam (with three strata) + three rock types); they are built into a 'scenario'. It is easy to change the pick file (or set two different pick types for the vane and the clearance sections of a drum), by simply entering the number(s) of the new file(s) that substitute the previous one(s). Finally, when a scenario is ready, the calculations are started in three steps. The first one consists of 'preliminary calculations': pick-rock interaction laws (Figure 29), together with a preliminary 25 20
S 15 υ
£
l0 5
d (mm)
Figure 29 Mean force versus depth of cut, as computed by P.C.DRUM. Pick, chisel (V-shaped); rock, coal
Figure 30 Simplified view of the seam structure, displayed by P.C.DRUM. Pick, chisel (V-shaped); rock, coal. In this example, a different rock type is embedded within the coal seam
Theoretical and Practical Rules for Mechanical Rock Excavation 250
213
r-
100
200
300
400
500
VQ (cm min-1)
Figure 31 Approximate cutting curves for torque, with the available machine specification, as computed by P.C.DRUM. Maximum available torque = 109.8 kN m (total); maximum, rate (approximate) = 286 cm min -1 ; vibration index = 0.09
ιυυ 80
E S 60
-
Maximum
" Λ^ ^J^y^A >-5^^Λ
2" 40 β± 0)
Λ Λ A ^ Ä s ^ =^H= Minimum
20 0
j
I
1 180
1
1 1
(°)
Figure 32 Torque variation during a drum revolution, calculated by P.C.DRUM (Fa = 75cm min *; average torque = 50 kN m)
analysis of the relative positions of the picks for subsequent calculations of breakout cycles, with the decomposition of tool work in the different strata (Figure 30) and a first evaluation of the available torque and typical range of thrust (haulage force). The results of these calculations may be consulted (Figures 29 and 30, for instance), or they may be skipped to move to the next step. At this stage, the user may change some of the pick-rock interaction parameters, should he have experimental evidence from the laboratory (for instance, the FJFC ratio may be overestimated by the model). The second calculation step (which may also be skipped) is a fast and simplified computation of forces and torque, over a range of haulage speeds entered by the user. Its aim is to estimate an approximate value for the maximum advance speed of the machine, the limiting factor, and the resulting vibration level (Figure 31). The accuracy of the result in terms of advance speed is usually within 20%. It is the purpose of the third calculation step to compute accurately the forces and torque on the machine, then to check the compatibility with the machine specifications. A haulage speed (advance rate) is then entered (whereas it was calculated at the previous step), and the program calculates the distribution of forces on the drum(s) and on the machine, with mean, minimum and maximum components. Once the speed has been accurately determined, a variety of options allows the display such information as the thrust or torque variation during a drum revolution (Figure 32); the contribution of each cutting line to the overall efforts (Figure 33), which may be interesting to check whether the load on the drum is evenly distributed or not; or the load profile of a pick chosen in any line during its revolution (Figure 34 shows the torque profile as an example). Such simulations obviously provide a powerful way to assess the influence of design parameters (machine specifications, pick spacing, pick type), wear, quality of the rock, etc. This is why they will be used in the following section, dealing with practical examples of machine optimization. 8.4.3.2 The P.CM A P program The organization of P.C.MAP is similar to that of P.C.DRUM. The program can simulate pick tunneling machines and is mostly adapted to roadheaders [16]. Most modules are taken from
214
Mechanized Excavation
Vane section
Figure 33 Distribution of force contributions of the cutting lines, computed by P.C.DRUM (V% = 240 cm min *). This example shows the torque across the front drum of a shearer
Start
Stop
Figure 34 Torque contribution of a given pick (clearance line B) during its revolution, as computed by P.C.DRUM (Ka = 240cm min"1)
P.C.DRUM, except those dealing with the machine data files (haulage force is replaced by boom force), the kinematic analysis and the computation of forces in axial work (which has been shown to be less complex than the computation of forces in transversal work). As examples, Figure 35 illustrates a cutting head as displayed by P.C.MAP (the full 360° design is projected on a plane) and Figure 36 shows the cutting curves for axial penetration by a roadheader. 8.4.4 Practical Examples In this section, practical examples derived from the author's experience are described. They include a drum shearer, a continuous miner, and the introduction of a roadheader in a metal mine. 8.4.4.1 Evaluation of a drum shearer The example chosen here deals with a single motor (high performance) drum shearer recently introduced and tested in an underground colliery. The design of the drum was optimized for the loading and removing of cuttings. Its diameter was over 2 m, hence the choice of a low rotational speed in order to reduce the cutting speed of the picks as much as possible. The line spacing increases on the vane section from the clearance side to the machine side. The machine was intended to cut deeply into the coal and, due to the choice of four vanes on the drum, there was a choice to be made on the number of picks per line. The clearance design included four regular sequences of gradually tilted gauge picks. The seam is near horizontal, with a brittle (not very hard) coal, and the stresses on the face reduce the rock strength on the free face side. The simulation carried out with P.C.DRUM showed that, even though the machine was torque limited, it could still reach a haulage speed of over 5 m min" 1 when using the forward drum only, with a four picks per line drum design (Figure 37a). The vibration level was very low and each pick was cutting at a maximum depth of over 50 mm, which implies that any increase in haulage speed could result in pick-boxes being in contact with the broken rock (thereby reducing their lifetime). This is a typical example of a situation where the vanes should not be designed to give a staggered pick lacing, since the maximum depth dm would then have been doubled for the same haulage speed.
215
Theoretical and Practical Rules for Mechanical Rock Excavation Machine: Proto I
No I
200 mm 1000 mm
Figure 35 Design of a boom-type head, as displayed by P.C.MAP 250 r-
0
20
40
60
100
80
V0 (cm min"1)
Figure 36 Display of torque cutting curves by P.C.MAP (axial work). Available torque = 20.9 kN m; approximate advance rate = 14 cm min -1 (b)
(a)
250
500
200
400
«600
VQ (cm min"1)
1000
100
200
300
400
500
V0 (cm min"')
Figure 37 Simulation of a drum shearer in a coal seam: (a) monodirectional cutting with one drum (maximum available torque = 131.8 kN m, maximum rate (approximate) = 517 cm min"1, vibration index = 0.02); (b) bidirectional cutting with two drums (maximum available torque = 131.8 kN m, maximum rate (approximate) = 300 cm min"1, vibration index = 0.05)
When shearing with the two drums in a thick seam (3.5 to 4 m wide), the machine speed decreases to 3 m min" 1 , resulting in the same production rate as before and at a higher (although still acceptable) vibration level (Figure 37b). The machine is still torque limited, but an increase in wear would quickly result in all the available thrust being used. In order to increase machine performances, several directions have been investigated, as follows. (i) Available specifications: a single powerful motor is more efficient than two smaller ones, because the available power is distributed according to the various requirements (torque on the drums, haulage system, rock crusher, etc.), so that all the available power is actually used.
216
Mechanized Excavation
With two motors, the energy-consuming devices are related to one or the other, often resulting in one motor being saturated and the other one working at a small fraction of its rated capacity (a typical example is when each drum is separately powered: the forward drum then usually requires at least as much energy as the rear one, thus limiting the machine while a substantial amount of power is not used). The effect of increasing the available power of the motor is illustrated in Figure 38. The benefit of this modification is not obvious, as the shearer soon becomes thrust limited. In fact, this machine is fairly well balanced, in that some extra thrust force will be needed when the picks start deteriorating. When the power cannot be increased, the torque can still be enhanced by reducing the rotational speed Vr. As shown in Figure 39, the effect is complex and needs further explanation. Using the simplified example detailed in Section 8.3.2.2, and assuming the minimum curve to give a good representation of the efforts on the drums (which is valid, as shown in Figure 37), it is possible to use the equations giving torque and thrust F =
Kn'N-A-dJ4
T = N-D-A-dJ2n Bearing in mind that the maximum available torque and thrust are given, with the torque expressed as 30W0/(nVr\ and with dm = VJnVr, the advance rate is given by one of the following equations Va = 4n-Vr-F0/(Kn-N-A) (thrust limitation) Ka = 60nW0/(NDA)
(torque limitation)
(notation as in Section 8.3, W0 and F0 are the available power and thrust). The above relations show that, when the machine is torque limited, the rotational speed does not affect the instantaneous advance rate (as confirmed in Figure 39). It is then recommended to decrease Vr as much as technically possible, which increases the lifetime of the picks thanks to the reduction in their cutting speed. The major limitations in so doing consist of the difficulty in reducing Vx mechanically (speed reduction devices become more complex for low final rotational speeds) and the load transmitted to the picks by increasing their depth of cut (which also holds when increasing the motor power). Alternatively, when the machine is thrust limited, the advance rate Fa is directly proportional to VT, which is seen in Figure 39. It would, however, be dangerous to increase Vr too much, since this would firstly have a strong influence on the rate of wear of the tools, and secondly reduce the available torque. In general, the trend is definitely to reduce the rotational speed on excavation machines, together with a proper supply of mechanical power. (ii) Type of pick: four types of pick were simulated. In the brittle and soft coal occurring in the seam, V-shaped chisels or even forward-attack tools turned out to be the best-suited picks (they show good chipping characteristics and require less energy than pointed picks). (iii) Drum design: reducing the number of vane picks is also beneficial in this example. The problem is that the tools are cutting quite deep so that increasing the advance speed too much might result in the above-mentioned friction of the pick-boxes against the rock. The influence of the number of picks is often difficult to evaluate, as the optimum is a compromise between the breakout pattern (too few picks implies poor interaction between cuts), the depth of cut (e.g. the load per pick) and the pick lifetime. The industrial validation of these simulations came from in situ monitoring of machine performances, which confirmed the instantaneous haulage speeds calculated for both monodirectional and bidirectional shearing. The mining company involved in this investigation is currently trying to implement some of the above recommendations. — 300
'c
Thrust limitation
250
h F 200 υ
0)
co σ
■o
>
<
IbO 100
50 0
150
J 200
I 250
I 300 Power (kW)
L 350
400
Figure 38 Effect of available power on machine performance (simplified calculations)
Theoretical and Practical Rules for Mechanical Rock Excavation
217
Limit by torque Limit by thrust
Rotational speed (rev min ')
Figure 39 Effect of rotational speed on machine performance (detailed calculations)
8.4.4.2 Optimizing the design of a continuous miner As mentioned in Section 8.4.1.1 above, full scale in situ experiments have been carried out with a continuous miner in an iron mine, together with scaled tests in the laboratory. The full scale tests confirmed that the optimized design was design 3 (Figure 21), and force levels were recorded in good agreement with the scale model predictions. It is therefore possible to use some theoretical simulation to optimize the design or the specifications of the machine. It was found from laboratory testing, later confirmed by the manufacturer's technical information, that the machine was thrust limited. The vibration level was acceptable, but close to the danger level. After confirming the optimum drum design, several alternatives were considered to further improve machine performances. (i) Drum design: as illustrated in Figure 21, three designs were tested in situ. The main variables are pick spacing and wrap angle. For the first two designs, the spacing between consecutive picks on the vane is 30 mm (the angular rotation, which results in a difference in depths of cut, is then 30° and 70° respectively). With design 2, additional interaction may originate from picks located five lines away (spacing 150 mm, but no angular rotation), since the picks cut approximately at the same depth. Similarly, with design 3, the interaction may come from either the next pick on the vane (spacing 40 mm, angular rotation approximately 70°) or the pick at the same angular position (spacing 200 mm). The resulting critical depths have been evaluated in detail, using the relations given in Section 8.2.1.2. For design 1, dc is always in the vicinity of 5 to 6 mm, and the interaction occurs with the next pick on the vane. For design 2, dc is of the same order, plus an additional factor increasing with the advance rate Ka, for low speeds, and then switches to 40 to 45 mm ( Ka around 4 m min "*), so that the interaction is with the pick at the same angular position. For design 3, dc is around 10 mm and increases slowly with the depth of cut. As a result, in the range of practical speeds expected with the machine in fairly hard rock (
218
Mechanized Excavation
As a conclusion for this example, the best configuration would consist of a machine with more thrust (40 t instead of 301) and with a reduced number of tools in order to keep dc around 10 to 15 mm without increasing Vr.
8.4.4.3
Selection of a well-suited roadheader
The following example has been investigated in order to introduce mechanical tunneling in a uranium mine for production purposes. The main problem was related to the irregularity of the rock formations, ranging from very soft episyenites to fairly hard granite. Abrasivity was also expected to change quickly from one place to another. Scale testing was felt useful before any full scale field trial. Three different designs were tested in the laboratory, then simulated on a computer. In addition, some of the designs were tested in a suite of hard rocks, complementing the samples from the uranium mine. As shown in Figure 40, the familiar shapes of the cutting curves were observed during most tests, giving confidence in using the computer simulations to compare different scenarios. Some practical findings from this investigation can be listed as follows. (i) Type of head: the choice between transversal and boom-type head is often complex and depends on a variety of parameters. However, in hard or heterogeneous rocks, the latter has a distinct advantage in that it can penetrate relatively easily in a small part of the face (usually selected in softer areas), and from thereon widen its excavation to the full size of the tunnel. Boom heads also enjoy the advantage of a simpler transmission of the torque from the motor to the head. On the other hand, transversal machines are more productive, due to a larger cutting area (as long as the rock is not too hard), and they may be more stable thanks to a work load distributed on more picks. (ii) Lacing of the picks: in difficult conditions (rocks with a compressive strength above 50 MPa), the staggered one pick per line design always turned out to be more efficient, with less vibration. This was confirmed by computer simulations for one, two or three spirals. An interesting finding was the relative importance of the efforts on the nose section of the head. In some instances it could amount to 50% of the overall forces/torque, in spite of a reduced number of tools. There seems to be much benefit to be gained from an optimized nose design, but it must be remembered that locating pick-boxes in this area of the head is not easy. (iii) Available specifications: like for continuous miners, this investigation showed that most roadheaders are limited by their available thrust, rather than by the electrical power delivered by the motors. As a result, modifying the rotational speed will usually affect the instantaneous advance rate 3000
Z
2000
"to 3 f
1000
L / J
0 400
I
I
1
I
L
300
£ z
Φ
D cr
*%Μ^<
200
100
/" 0
J
1
2
I
I
3
I
4
I
5
I
6
I
7
dm(mm)
Figure 40 Experimental cutting curves at a 1/4 scale, with a two-spirals, one-pick per line design (after Déliac and Cordelier [26])
Theoretical and Practical Rules for Mechanical Rock Excavation
219
in the same direction, although an increase would not be beneficial in the long run (see Sections 8.4.4.1 and 8.4.4.2). Improvements to this situation are dealt with in Section 8.5. After this detailed analysis, it was possible to suggest a 'well suited' machine, e.g. a head design and desirable specifications. For financial reasons, the machine has not been introduced and tested yet, to confirm the laboratory work. Finally, these practical examples show that predictive calculations can be usefully carried out before introducing a machine in situ. Obviously, calibrating the calculated values with field experimental evidence is strongly advised, whenever possible. Yet the combined use of scaled heads, tested in the laboratory, with computer simulation proves to be of great assistance to the design engineer or the production manager. Systematic use of these possibilities should lead to a better preliminary evaluation of the possible performances of a given machine in given environments, thereby preventing major mistakes. 8.5 ADAPTATION OF MECHANICAL EXCAVATION TO A HARSH ENVIRONMENT When the rock becomes more and more difficult to excavate, the depth of cut decreases and wear increases. This in turn means a reduction in performance but, more importantly, a sharp increase in vibrations on the cutting head, transmitted to the machine itself. As soon as the vibrations reach a critical level (see Section 8.3), the machine limitation becomes severe and the deterioration is accelerated. Solutions to this problem can be to increase the specifications of the machine. Figure 41 shows that roadheaders have been continuously upgraded in weight and power over the years. However, thefigureindicates that there could be some kind of technological barrier in hard rock, at least due to limited available space in the developing tunnel, which in turn limits the size of the machine or the maximum load compatible with the tools' strength. Other solutions are found with 'activated' tools such as polycrystalline diamond (PDC) tipped picks or water jet assisted picks/disks. Although more expensive, these tools can be very profitable in difficult cutting conditions [28] and their beneficial effect can be quantified. This is the purpose of this section, which shows that it is now possible to outline which equipment should be recommended for a given harsh environment. The main problem here is thus to find the adequate machine (in terms of the available specifications), then to identify the proper enhanced tool, andfinallyto optimize the eventual choice from an economic point of view. 8.5.1 Adaptation of Machine Specification and Head Design Key factors for good machine behavior have been demonstrated to be the vibration level and wear rate of the cutting tools. This in turn is related to: (i) the individual depth of cut; (ii) the interactions between neighboring cuts; and (iii) the cutting speed. o 200
a. S
£
,75
|
150
σ»
Jet assisted roadheaders
(A
0)
0 x o i E
75 50
1 25 x o
J
I960
I
I
1970 1980 Year
_L
1990
Figure 41 Increase in roadheader specifications with time (after Cordelier [16])
220
Mechanized Excavation
From a machine specifications point of view, it is therefore essential to properly set the available torque and thrust, together with the rotational speed. As explained in Section 8.4.4, the latter should be as low as possible, compatible with the available thrust and the load on the cutters. As a matter of fact, the lower the rotational speed the higher the torque, hence the thrust requirement. As a result, should the machine be thrust limited, any reduction in Vr would result in a decrease in both the advance rate (with the same vibration level) and the wear rate. On the other hand, should the limitation be by the torque (or power), then the benefit would be manifold: an increase in depth of cut, a decrease in vibration (hence an improved advance rate) and a decrease in wear. However, in very hard rocks it may be impossible to cut deeply, as the required torque and thrust would be impractical due to the limitation in machine size and maximum admissible load on the cutters. The only solution is then to reduce the rotational speed to a practical minimum to adapt the thrust in order to match the available torque, and then to 'play' with the head design as well as with the type of cutter. The head design should itself be selected for an optimized interaction pattern, dependent upon the depth cut by each cutter. In hard rocks, it is now well established (and confirmed by computer simulations) that one pick per line staggered lacings behave better than regular designs with n picks per line and n spirals (when n > 1), because the reduced spacing between cuts, gives a much better interaction pattern. Selecting the best number of cutters is a complex matter, as a low number of tools results in deeper cuts, but with an important spacing, whereas many tools mean shallow cuts at a reduced spacing. However, because the specific energy is optimized for deep cuts, a reduced number of cutters is often preferable as long as the resulting load on each tool is compatible with its mechanical strength (see the example detailed in Section 8.4.4.2). In summary, although the final choice should be validated both by computer simulation and practical experiments, some basic rules can be drawn at this stage for machines in hard rock. (i) Thrust should not be a limiting specification; this is unfortunately not so with many pick tunneling machines. Like for TBMs, hydraulic jacks or other similar devices should then be used to enhance the available thrust. (ii) Rotational speed should be kept as low as practically possible (which depends on the type of electrical motors used: the reduction gears might not be capable of reducing the rotational speed as much as desired). (iii) Head design should be based on one pick per line when cutting in difficult conditions. 8.5.2 Adaptation of Cutting Tools As mentioned in the beginning of this section, two 'enhanced' types of cutters have been tested with some success in hard rocks: the diamond-tipped picks and the water jet assisted tools. Further investigations are underway to find new cutter designs (replacing diamond by ceramics, or trying picks made of special alloys from the body to the tip in a single piece, for instance), but these are still at the research stage and will not be dealt with in detail here. 8.5.2.1 Diamond-tipped picks A fairly comprehensive presentation of research undertaken on diamond-tipped picks has been recently published in French [29]. It outlines the development and trial of three generations of tools since the early 1980s. The first generation was reported in the USA, in South Africa and in Germany. The U.S. Bureau of Mines was looking for improvements in the safety of cutting in dangerous environments (gassy atmospheres), whereas the Sasol company investigated increased productivity in hard coal seams. German researchers (StBV) studied the behavior of diamond-tipped tools in the laboratory and in situ. The design of these first picks was based on the use of polycrystalline diamond compact (PDC) cylindrical bits fixed at the tips of radial picks. The cutting tests on laboratory rigs consistently showed that the diamond-tipped picks were requiring more specific energy than standard tools, mainly due to poor bit geometry (negative rake angle, for diamond-manufacturing reasons), and that the PDC bit was quite brittle (poor resistance to shocks). The safety of cutting in gassy atmosphere was, however, much better than with regular radial picks. In situ trials in South Africa demonstrated a better productivity of PDC bits in difficult seams, but diamond cutting was still not economical.
Theoretical and Practical Rules for Mechanical Rock Excavation
221
These preliminary results were felt encouraging enough to boost research, and a second generation of tools was tested later, mostly in South Africa [30]. A new tool design was developed by inserting a PDC bit into the body of a heavy duty forward attack pick. Much improvement was then observed, particularly concerning the strength of the bit-tool bonding and the wear resistance of the pick (the diamond gives better protection to the tungsten carbide on the clearance side). The in situ trials on a drum shearer showed that the machine was keeping its maximum advance rate for a much longer time than with standard carbide-tipped picks. In addition, less time is devoted to removing the worn-out tools and placing new ones in their boxes. Finally, safety conditions are improved and the size of dust particles is significantly increased. In spite of these successful results, the project was abandoned in South Africa, apparently because the PDC tools still did not prove economical. Further testing was then initiated in France [29] with the CERCHAR and the Paris School of Mines laboratories. After initial experiments on picks similar to the first generation, which confirmed the above-mentioned results, a new generation of diamond-tipped tools was developed and investigated. The design consists of a pointed pick body with a special brazed bit made of a tungsten carbide base with a dome shape on top, covered by a composite cobalt-tungsten carbide-diamond material, itself coated with sheets of polycrystalline diamond (Figure 42). Laboratory testing under severe conditions (cutting speed over 0.5 m s" 1 in a hard, very abrasive sandstone) confirmed the excellent behavior of the tool with respect to wear resistance. The temperature of the bit was always reasonably low, whereas an ordinary point-attack pick would inflame a sheet of paper after a few seconds of cutting. This tool was also highly resistant to repeated shocks. On the other hand, the specific energy was found to be systematically higher than for the tungsten carbide pick (up to twice as much at shallow depth of cut), which is obviously due to the peculiar smooth shape of the bit. However, experimental testing showed that the difference in forces between the two picks tended to decrease at depths of cut above 10 mm (unfortunately seldom to be found in hard rocks). This innovative tool was subsequently tried underground in a colliery. The overall good behavior was confirmed, but it was observed that the metal surrounding the diamond-coated bit would wear out fast in abrasive rocks, thereby allowing the bit to be removed accidently. Some research is currently being carried out to improve the bit-body fixing, which could hopefully result in a fourth and industrial generation of picks. Overall, it may be stated that diamond-based bits are now technically operational, but improvement is required concerning the wear resistance of the pick body and the strength of the bonding between the bit and the body. Should this happen, even though they are much more expensive than standard tools (price ratio ranges from 10 to 30), PDC bits may well prove quite economical in difficult cutting conditions. 8.5.2.2 Water jet assistance This will only be briefly dealt with here, as it is the main topic of the following chapter [31].
Polycrystalline diamond
Complete tool
Detailed view of bit
Figure 42 Design of the third generation of diamond-tipped picks
Mechanized Excavation
222
The pioneering work originated from South Africa in 1976, aimed at reducing the forces on bits in hard rock by placing powerful water jets along the tool or in front of it. Interesting experimental results were obtained, later confirmed in other laboratories, with force reductions reported of up to 50% or even more. However, such experimental evidence was obtained at very low cutting speeds and it was later demonstrated that the tool velocity is an essential parameter in assessing the efficiency of the water jet [32]. Recent research, carried out at actual cutting speeds (e.g. between 0.5 and 3 ms" 1 ), has shown that the major parameters affecting water jet assistance are the tool velocity, the water pressure, the nozzle diameter, the stand-off distance, and the distance from the bit to the impingement point of the jet. Under these cutting conditions, it has been experimentally shown that a 70 MPa water jet does not reduce the peak forces on a pick cutting rocks with a compressive strength above 50 to 70 MPa (single unrelieved cut) [32]. It is thought, however, that the water jet should have some beneficial influence on the specific energy during relieved cutting by lateral fracturing of the already fissured rock towards the neighboring cut (Figure 43). As a result, the use of efficient water jets on a tunneling or coal winning machine could mean a better interaction coefficient between the picks, hence a reduced critical depth dc, and an improved interaction pattern (together with a reduced vibration level, as explained in Section 8.3). Figure 44 shows a calculated example [3] illustrating the dramatic improvement in machine performance when the interaction coefficient increases. The above considerations could explain the 30% reduction in boom forces on a water jet assisted roadheader, published by British Coal engineers [28]. Yet other benefits are to be drawn from water jet assistance. The effect of a water jet on the wear rate of a forward attack pick at high speed (approximately 2.4 m s"*) has been investigated by Fairhurst [3]. The results show that, although wear (measured as the increase in normal force) is still important for a 70 MPa jet in a hard limestone (ac over 100 MPa), the jet has a significant influence, substantially increasing the bit lifetime. Finally, it is now well established that water jets modify the chipping cycle of cutting tools by continuously cleaning rock debris from around the tool tip, which usually means a reduced normal force and also a somewhat reduced mean cutting force (but not systematically). In summary, the benefits of water jet assistance are to be found in three directions: (i) hydraulic effect, by fracturing the rock laterally when the neighboring cut is not too far away (to be confirmed
Figure 43 Lateral fracturing of the rock during interactive cutting (after Déliac and Fairhurst [9]) 260
_r 220 c
e
E 180 o 140 Q.
0)
υ c σ > •o
<
100 60 20
2.0
2.5
3.0
3.5
4.0
4.5
Interaction coefficient,
5.0
5.5
6.0
k
Figure 44 Effect of the rock interaction coefficient on the performance of a drum shearer (after Fairhurst [3])
Theoretical and Practical Rules for Mechanical Rock Excavation
223
experimentally); (ii) thermal effect, by cooling the rock and the bit, thus reducing wear and improving the sharpness of the tool; and (iii) mechanical effect, by cleaning the cut and removing rock debris and particles, thus avoiding useless energy being spent in secondary crushing. This probably explains why the development of water jet assistance, both on pick and disk machines, is today at the industrial and commercial stage, although the physics of the water jet action are not yet fully understood.
8.5.3 A Simple Economic Model for the Selection of Equipment The economic comparison between a conventional drill and blast method and a continuous tunneling or mining method is highly dependent upon the stope cycle, e.g. the frequency and duration of machine stops for the latter. Let us assume that the machine is reliable when the tools are in good condition, but that their continuous use results in their gradual deterioration, with increasing vibrations on the cutting head and decreasing performance until it is stopped to change the worn-out picks/disks. A simple model can then help to answer the two following questions: (i) when should cutters be changed? and (ii) how many tools should be changed during one machine stop? The basis of this model, further detailed hereafter, has been published after some initial theoretical work on the influence of wear [29]. If Kam is the advance speed of the machine with new tools, and if Kac is the critical advance speed below which the vibration level would result in major damage, then the performance cycle can be illustrated as in Figure 45. Thefirstpart is a slow decline in speed due to wear gradually taking place. After some time, the thrust limitation becomes severe and the speed decreases rapidly, finally reaching Vac. The average speed, Kav, thus depends on Vam and on the duration of the first slow decline phase. Finally, let t be the production time during one cycle and let is be the stand-by time, so that the period of the cycle is tc = t + ts The average production rate of the machine, g, is given by the following equation, where X is a fixed coefficient (projected section of the machine head on the face)
If n picks are changed when Ka reaches Kac, and if c is the cost of one cutter, the cost incurred by the change of cutting tools (in monetary units per volume of rock) is c
=
ne ~Q7C
=
ne vavt-x
8.5.3.1 When should cutters be changed? For the sake of clarity, the slow decline shown in Figure 45 at the beginning of the cycle is assumed here to be negligible, so that Va is constant during a time t0i close to Vam. During this time ^av = ^ann and Q is proportional to the ratio t/(t + is) (see the above equation): the average rate increases with t. After a sufficient time, however, the decline becomes sharp and the ratio t/(t + is) tends to stabilize close to 1. As a result, Kav decreases faster than t/(t + is) increases and Q is a decreasing function oft. It is thus demonstrated that there exists an optimal time t09 between these two periods, where Q is maximized. This is the time when the cutters should be changed. This time is difficult to accurately quantify, as it depends on knowledge of the performance curve with time [illustrated in Figure 45 and related to the function wear = wear(time)]. 8.5.3.2 How many tools should be changed? Let nx and n2 be two numbers of tools being changed, with nx
224
Mechanized Excavation
" am
Vav
Vac
m
»I
I
Time
Idealized cycle of a tunneling or mining machine
Figure 45
machine, visually inspect the head, remove the worn-out tools and replace them, and finally restart the machine. The costs to be compared are =
Cl
n< -c ^avl
and
C2
=
'ti'X
"av2
'tl'X
It is therefore essential to determine Kav and t for each situation in order to find the smallest cost. A computer simulation can provide some answers to this question, by entering two different average wear conditions of the cutters on the head. The following example illustrates this further. 8.5.3.3 Comparison between diamond and tungsten carbide picks Simplified performance cycles for the two different types of picks are illustrated in Figure 46. The maximum speed with diamond-tipped picks has been shown to be somewhat lower than that with new tungsten carbide picks (see Section 8.5.2.1). Nevertheless, the cycle is then simple, since the speed is constant until some cutters have to be changed (which is when they start to lose their bits). Therefore, with diamond picks, if Vd is this constant speed, and td the duration of the cutting period, the cost of the cutters is given by Cd =
VytyX
The cost ratio with standard tungsten carbide picks is then C/Cd = (n/nd)'(c/cd)'(VJVJ'(td/t)
V0 Diamond-tipped pick
\
Tungsten carbide P i>ick c
._. /
Time Figure 46
Performance cycle with and without diamond picks
225
Theoretical and Practical Rules for Mechanical Rock Excavation
The South African experience seems to indicate that the individual cutter cost ratio c/cd can be of the same order of magnitude as the (n · td)/(nd · i) ratio. Should this be confirmed by further testing of diamond picks and reduction of their cost through the industrial production of large quantities, then the economic benefits of diamond picks would be illustrated by the ratio of the average speeds, which is obviously in favor of diamond tools in hard rocks.
8.5.3.4
Comparison between dry and water jet assisted cutting
Some data have been collected by Fairhurst [3] concerning the performance cycles of drum shearers with and without water jet assistance. The data can be summarized as follows. (i) Overall average speed: increased by 40% with water jets. (ii) Production cycle: t is doubled with water jets, and is is increased by 50%. (iii) Number of picks to change: lifetime of the water jet assisted picks is 5 times that of dry picks. The number of picks to change is thus 2.5 times lower than that of dry picks (production time is doubled). (iv) Direct cost of water jet assistance: picks are no more expensive, but the machine cost is 20% higher with water jets. Neglecting the change in t/(ts + t), it is easy to calculate the direct cost ratio, equal to 0.3, and hence the total water jet assisted cost Cw = 0.3-C + 1.2Cam where Cam is the amortized cost of the machine (over total production) to be compared with the 'dry pick' cost C + Cam. Therefore, if the cost related to the picks, C, is high (e.g. in difficult conditions), water jet assisted cutting should be more economical in spite of a more expensive machine. The very simple model outlined here can thus be useful in order to compare different situations, as long as some data are available as to the cost of the cutters and the performance cycle of the machine.
8.6
CONCLUSIONS AND FUTURE PROSPECTS OF MECHANICAL ROCK EXCAVATION
Mechanical rock excavation is still a young technique in the large scope of engineering. Yet it has achieved tremendous improvements since its early industrial introduction about 50 years ago. As often observed in science, the understanding of the physics behind rock cutting machines has progressed, but at a slower rate. It is now possible to quantitatively assess the performance and behavior of a given machine. The engineer can thus check the selection of his equipment, optimize the choice of the cutters or their location on the head, etc. It was our ambition to make this complex set of rules, equations and sometimes semiempirical reasoning more accessible to the reader. We hope that after reading this chapter he will feel more confident in dealing with rock cutting. If this is so, then the continuous improvement in rock excavation machines will continue at a steady pace. Today, it is difficult to cut rocks with a uniaxial compressive strength of over 100 MPa using picks. Tomorrow, however, thanks to the introduction of new materials to enhance the cutter characteristics, the increase in available specifications by use of reliable hydraulic devices to improve the thrust, and reduction gears which can produce very low rotational speeds, harder rocks will be won without having to increase the size or weight of the machine. Careful monitoring of parameters such as the forces on the head, or hydraulic pressures, should also lead to a better remote control of the machine behavior, by early diagnosis of tool failure, for instance. The scope for much improvement lies ahead, making mechanical excavation an exciting challenge to the drill and blast methods. 8.7 ac a ß0 δ
NOMENCLATURE characteristic chip angle inclination angle of picks (lateral tilt) empirical parameter to calculate Kn (0.01 to 0.02 mm" 1 ) wedge angle of tip or disk
(°) (°) (°)
226 Θ μ ac τ φ Ω d dc dm E F Fc Fh Fx Fn Fr H k Klc Kx Kn L N p Q rt RL Rv S T V Va Vr W Wh Zi
Mechanized
Excavation
position angle during cutting head revolution friction coefficient on haulage system unconfined compressive strength contribution to torque (pick) internal friction angle arc of contact between head and rock depth of cut critical depth of cut (for interaction effect) maximum depth οΐ cut/pick/head revolution efficiency of motor or power unit horizontal thrust force (head) cutting force (pick) haulage force (machine) lateral force normal force rolling force (disk) contribution to horizontal thrust (pick) interaction coefficient of rock fracture toughness (mode I) F{/Fc coefficient FJFC coefficient contribution to lateral effort (pick) total number of picks on head number of deepening increments in a breakout cycle machine production rate radius of cutting line i horizontal lateral reaction force (head) vertical reaction force (head) spacing between cutting lines rotational torque (head) contribution to vertical effort (pick) advance speed of the machine in the rock rotational speed of the cutting head power available/required from the machine power available for the hydraulic unit(s) offset between picks on lines i and i — 1
(°)
(MPa) (kNm) (°) (°) (mm) (mm) (mm) (kN) (kN) (kN) (kN) (kN) (kN) (kN) (MNm" 3/2 ) (kN) (m3 min * ) (mm) (kN) (kN) (mm) (kNm) (kN) (mmin *) (rev min" 1 ) (kW) (kW) (mm)
8.8 REFERENCES 1. Fowell R. J. The mechanics of rock cutting, Comprehensive Rock Engineering (Edited by J. A. Hudson), vol. 4, 2. 3.
4. 5. 6. 7. 8.
9.
10. 11.
pp. 155-189. Pergamon Press, Oxford (1993). Déliac E. P. Optimisation des machines d'abattage à pics. Doctoral dissertation, University of Paris VI, France (1986). Fairhurst C. E. Theory and practice of enhanced rock cutting picks: the water-jet assisted tool and the vibrating tool. Doctoral Dissertation (in French and English), Paris School of Mines, France (1987). Sanio H. P. Prediction of the performance of disc cutters in anisotropic rock. Int. J. Rock. Mech. Min. Sei. & Geomech. Abstr. 22, 153-161 (1985). Evans I. A theory of the basic mechanics of coal ploughing. In Proc. Symp. Mining Research, University of Missouri, Rolla (Edited by G. B. Clark), vol. 2, pp. 761-798. Pergamon Press, Oxford (1962). Lebrun M. Etude théorique et expérimentale de l'abattage mécanique; application à la conception de machines d'abattage et de creusement. Doctoral Dissertation, Paris School of Mines, France (1978). Roxborough F. F. Cutting rock with picks. Min. Eng. (London) 132, 445-455 (1973). Saouma V. E. and Kleinosky M. J. Finite element simulation of rock cutting: a fracture mechanics approach. In Proc. 25th U.S. Symp. Rock Mech., Evanston, IL (Edited by C. H. Dowding and M. M. Singh), pp. 792-799. Soc. Min. Eng. AIME, New York (1984). Déliac E. P. and Fairhurst C. E. Theoretical and practical investigations of improved hard rock cutting systems, In Proc. 29th U.S. Symp. Rock Mech., Minneapolis, MN (Edited by P. Cundall, R. L. Sterling and A. M. Starfield), pp. 553-562. Balkema, Rotterdam (1988). Sellami H., Cordelier P., Hefferman J. and Chaput E. The influence of rock properties on the efficiency of mechanised mining, with reference to hard rock cutting. Report No. 4 to the E.E.C., Edited by the Paris School of Mines, Paris and Imperial College of Science, Technology and Medicine, London (1990). Anon (coll. work) The influence of rock properties on the efficiency of mechanised mining, with reference to hard rock cutting. Report No. 2 to the E.E.C., Edited by the Paris School of Mines, Paris and Imperial College of Science, Technology and Medicine, London (1990).
Theoretical and Practical Rules for Mechanical Rock Excavation 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32.
227
Roepke W. W. and Hanson B. D. Effect of asymmetric wear of point-attack bits on coal-cutting parameters and primary dust generation. Rep. Invest.-U .S., Bur. Mines RI 8761 (1983). Sellami H. Etude de pics usés: applications aux machines d'abattage. Doctoral Dissertation, Paris School of Mines, France (1987). Roxborough F. F. and Phillips H. R. Rock excavation by disc cutter. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 12, 361-366 (1975). Déliac E. P. Recent developments in the design and optimization of drum-type cutting machines in France. In Proc. 7th Rapid Excavation and Tunnelling Conference, New York (Edited by C. D. Mann and M. N. Kelley), pp. 265-283. Soc. Min. Eng. AIME, New York (1985). Cordelier P. R. Modélisation du travail des machines à attaque ponctuelle. Doctoral Dissertation, Paris School of Mines, France (1989). Brooker C. M. Theoretical and practical aspects of cutting and loading by shearer drums. Colliery Guardian part I in 229, 9-16, and part II in 230, 41-50 (1979). Déliac E. P. and Gripp M. F. Etude quantitative de l'abattage mécanique par pics à partir d'essais en modèle réduit. Rev. Ind. Miner.-Mines 65, Les Techniques 5-83, 275-292 (1983). Déliac E. P. and Léonet O. Optimisation du matériel pour l'abattage en taille dans les mines de fer lorraines. Rev. Ind. Minér.-Mines 66, 331-340 (1984). Morris A. H. The design of shearer drums with the aid of a computer. Min. Eng. (London) 140, 289-295 (1980). Holt P. B , Morris C. J. and Owen R. J. Desk-top computers for design work, Min. Eng. (London) 143, 485-489 (1984). Knissel W., Mertens V., Kleinert H. W. and Mittmann M. Verfahren zur Auslegung and Optimierung der Schneidköpfe von Teilschnitt Vortriebmaschinen. Glueckauf 120, 1534-1539 (1984). Roxborough F. F. and Eskikaya S. Dimensional considerations in the design of a scale model for coal-face production system research. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 11, 129-137 (1974). Dubugnon O. and Barendsen P. Small scale model testing; a new approach in TBM development. In Proc. 7th Rapid Excavation and Tunnelling Conf, New York (Edited by C. D. Mann and M. N. Kelley), pp. 245-263. Soc. Min. Eng. AIME, New York (1985). Roepke W. W., Wingquist C. F., Olson R. C. and Hanson B. C. Bureau of Mines coal cutting technology facilities at the Twin Cities research center. Inf. Circ. - U.S., Bur. Mines 8951 (1983). Déliac E. P. and Cordelier P. R. Practical results from reduced-scale testing of pick cutting heads for tunnelling applications. In Tunnelling '88, London (Edited by M. P. Jones), pp. 79-86. Institution of Mining and Metallurgy, London (1988). Roxborough F. F. and Pedroncelli E. J. A practical evaluation of some coal cutting theories using a continuous miner. Min. Eng. (London) 142, 145-156 (1982). Morris A. H. and Harrison W. Significant advance in cutting ability - Roadheader. In Proc. 7th Rapid Excavation and Tunnelling Conf, New York (Edited by C. D. Mann and M. N. Kelley), pp. 317-340. Soc. Min. Eng. AIME, New York (1985). Déliac E. P., Sellami H. and Fairhurst C. E. Adaptation des outils d'abattage aux roches dures. In L'Abattage Mécanique, vol. 4. Paris School of Mines, Paris (1988). Collin W. D. and Kornecki J. A. The development and use of diamond picks for longwall shearers at Secunda Collieries. In Proc. Mining '85 Conference, Birmingham, pp. 153-163. Institution of Mining Engineers, London (1985). Hood M. The use of water jets for rock excavation. Comprehensive Rock Engineering (Edited by J. A. Hudson), vol. 4, pp. 243-274. Pergamon Press, Oxford (1990). Fairhurst C. E. and Déliac E. P. Water-jet assisted rock cutting - The effect of pick traverse speed. In Proc. 8th Int. Symp. Jet Cutting Technology, Durham, pp. 43-55. BHRA, Cranfield (1986).
9 The Use of Water Jets for Rock Excavation MICHAEL HOOD Queensland Centre of Advanced Technologies, Kenmore, Qld, Australia 9.1
INTRODUCTION
229
9.2 CONTINUOUS JETS 9.2.1 Kerf Cutting with High Pressure Jets 9.2.2 High Pressure Jet Drills 9.3
230 230 233
DISCONTINUOUS JETS
234
Theoretical Considerations interrupted Continuous Jets Water Cannons
234 237 237
9.4
CAVITATING WATER JETS
239
9.5
ABRASIVE WATER JETS
241
9.3.1 9.3.2 9.3.3
9.6 COMBINED USE O F HIGH PRESSURE JETS AND MECHANICAL CUTTING TOOLS 9.6.1 Mechanically Assisted Cutting 9.6.1.1 Principles of the cutting method 9.6.1.2 Drilling small holes (smaller than 50 mm diameter) 9.6.1.3 Drilling medium holes (larger than 50 mm but smaller than 300 mm diameter) 9.6.1.4 Drilling large holes (larger than 300 mm diameter) 9.6.2 Jet-assisted Cutting 9.6.2.1 Principles of the cutting method 9.6.2.2 Drag bit force reductions - general 9.6.2.3 Tool force reductions - the importance ofjet position 9.6.2.4 Tool force reductions - the effect of bit velocity 9.6.2.5 Reduction of dust, frictional sparking and machine vibration 9.6.2.6 Reduction in bit wear 9.6.2.7 Disc cutter force reductions
242 242 242 243 245 247 248 248 249 251 251 254 254 257
9.7
257 257 258 258
CONCLUSIONS
9.7.1 9.7.2 9.7.3 9.8
9.1
Continuous, Discontinuous, Cavitating and Abrasive Water Jets Mechanically Assisted Cutting Jet-assisted Cutting
REFERENCES
259
INTRODUCTION
The use of high pressure water jets for rock excavation systems has been the subject of considerable research for the past 25 years. The continuing interest in this technology is driven by a desire to overcome a fundamental limitation on the use of mechanical tools for rock breaking. Rock excavation today is conducted either with mechanical tools or with explosives. Rock cutting by machine offers many advantages over explosive breaking. However, with the notable exception of hard rock boring machines, the use of machines for rock excavation today is limited to weak rock materials, such as coal.
229
230
Mechanized Excavation
The difficulty in cutting rocks with mechanical tools lies in the strength of the tool materials. The rate of tool wear and breakage increases as the power applied to the rock through the tools increases. However, since the rate of excavation increases with the power applied to the rock, there exists an upper limit on the rate of rock excavation that is determined by the maximum power that can be applied to the tools without causing excessive rates of tool failure. Furthermore, since the power necessary to break rock at a given rate increases with the strength of the rock, the feasibility of using the cutting method to excavate rock, at rates of extraction that are of interest to the mining and civil engineer, decreases with increasing rock strength. Most rock cutting machines employ drag bits (or picks) as the cutting tools. These tools are more efficient, that is to say they require less power to break a unit volume of rock in a given time, than roller cutters. However, drag bits are more susceptible to wear and breakage than roller cutters. Hence, in the strong rocks that are cut with boring machines, roller cutters are employed. The machines that react to the very high cutting forces experienced by roller cutters necessarily are large, inflexible and difficult to steer. Consequently while these machines are ideally suited for drilling long, almost straight, circular holes, their usefulness in other applications is limited. One solution to this fundamental constraint on the use of mechanical tools is to apply the energy necessary to cut the rock by some means other than through mechanical tools. Many different methods have been investigated, including thermal, chemical and erosional (using high pressure liquid jets). It is widely recognized that the use of high pressure water jets is the most practical of these breaking methods. In this chapter we review the options for the uses of high pressure water jets both as an alternative to and as a supplement to the use of mechanical tools for rock cutting. 9.2 CONTINUOUS JETS 9.2.1 Kerf Cutting with High Pressure Jets When a high pressure water jet is traversed across a rock surface typically the jet erodes the rock grains and cuts a shallow kerf in the rock face. The effectiveness of this erosional process depends both on the rock type and on the jet characteristics: pressure, flow rate, etc. Many workers have investigated the erosive behavior of jets in rock kerfing operations. One of the best documented works is a paper by Harris and Mellor [1] and their results are used to illustrate the following general comments concerning trends in the cutting behavior of high pressure water jets. (i) The kerf depth cut by a jet, over a wide range of jet power, jet traverse speeds and rock types, typically is less than 25 mm and often is of the order of only a few millimeters. (ii) At low jet pressures the rock tends to spall or flake and a wide but shallow kerf is cut. At high jet pressures this spalling does not take place, instead a narrow kerf with sharp edges is formed. (iii) At very low jet pressures the jet may not erode the rock at all, in fact it is widely held that a minimum value of jet pressure is needed in order for any erosion of the rock to take place. This minimum pressure is termed the threshold pressure. Some workers have claimed that the threshold pressure has roughly the same numerical value as the uniaxial compressive strength of the rock, although there is no reason why this should be so. Other workers, including Harris and Mellor [1], have shown that in some rocks no threshold pressure exists. (iv) The depth of a kerf cut by the jet increases with increasing jet pressure (Figure 1). A small positive curvature is evident in these curves. (v) The depth of a kerf cut by the jet increases with increasing nozzle diameter (Figure 2). (vi) The depth of a kerf cut by the jet decreases with increasing jet traverse speed. When the traverse speed is increased from a low value this decrease in kerf depth is substantial, but at higher speeds the kerf depth is relatively insensitive to jet traverse speed (Figure 3). For a given jet power there exists an upper limit of jet traverse speed beyond which no rock erosion occurs. This point is illustrated by the points D and E in Figure 1(b). Harris and Mellor [1] found that in Barre granite with a jet diameter of 0.203 mm no damage was caused to the rock at jet pressures lower than 400 MPa when the jet velocity exceeded 177 mms" 1 . It is not surprising that increasing either jet pressure, p, or nozzle diameter, d, causes an increase in kerf depth. However, the power of a fluid jet, P Ρ~ρ15ά2
(1)
Hence the experimental approach followed by Harris and Mellor and by other workers, namely increasing p while holding d constant and then increasing d while holding p constant, means that the jet power changes in a nonlinear manner at each data point. Hood et al. [2] used a factorial
The Use of Water Jets for Rock Excavation
231
(a) Indiana limestone d =0008 in. (0.203 mm) Standoff:0.25in.(6.35mm)
.
E E
«c^ o
^
c 16 — +~* c o
■ 1-
o l_
+-
A B C 0 E F
12
Traverse speed f t sec"1 mm eec"1 6.7 0.022 0.051 15.5 57.9 0.19 177.0 0.58 975.0 3.2 1646.0 5.4
_ " o*-
Nozzle pressure (x I0 3 psi) J L 2
3
Nozzle pressure (xlO 3 bar) (b) Barre granite d =0.008 in. (0.203mm) Standoff: 0.25 in. (6.35mm) Line
|
o
0.2
A B C D E
Traverse speed f t sec- 1 mm sec -1 6.4 0.021 0.052 15.8 0.19 57.9 0.58 177.0 5.63 1716.0
J_ 20
J_
30
_L 40
50
Nozzle pressure (x I0 3 psi) J_ 2
3
Nozzle pressure (x I 0 3 bar)
Figure 1 Effect of jet pressure on kerf depth in (a) Indiana limestone and (b) Barre granite (Harris and Mellor [1]) (1 bar = 105 Pa, 1 psi = 6895 Pa)
5h E E w sz :
4
3
0.20
0.22
-L
0.24
0.26
0.28
0.30
0.32
Nozzle diameter (mm)
Figure 2
Effect of nozzle diameter on kerf depth (data from Harris and Mellor [1])
Mechanized Excavation
232 0Θ
20 U
Berea sandstone d = 0 . 0 0 8 i n . ( 0.203mm) Standoff :0.25in. (6.35 mm)
Γ
Nozzle pressure kbar psi 4.14 60 000 45 0 0 0 3.10 30 000 2.07 15 0 0 0 1.03
Curve
o £
12
A Θ C D
U .2 °
4Q>
|- c
0.
0.41 '
It >w_ 1
^^-^ά^ i
i
— r4
1
2
3
1
'
1
-1
Traverse speed (fts ) I l 0.8
1.2
Traverse speed (m s-1)
Figure 3 Effect of jet traverse velocity on kerf depth (from Harris and Mellor [1])
experimental design and surface response methodology to investigate the influence of the various parameters that affect rock erosion. A principal finding from this work is that in Indiana limestone the kerf depth is independent of both jet pressure and jet flow rate when the jet power is held constant (Figure 4). Figure 5 shows that the kerf depth increases monotonically with jet power.
(a)
Traverse velocity = I80mrn s"1
Traverse velocity =80mm s
■ 1.9kW O 4.3kW • 9.7kW
Ih 250
50
_L
100
Jet pressure (MPa)
6
(b) r~
Traverse velocity = 80mm s"
■ l.9kW 51- O 4 . 3 k W • 9.7kW
Traverse velocity = 180mm s ■ I .9kW f- O 4 . 3 k W • 9.7kW
H
K, Flow rate (ml s )
Figure 4 Insensitivity of kerf depth to (a) the jet pressure and (b) the jet flow rate when the jet power is held constant (from Hood et al [2])
The Use of Water Jets for Rock Excavation
233
20
E E
Ï
Q.
Ό
a> Q
Ï
*
5h
îi I 40
80
Power (kW)
Figure 5 Influence of jet power on kerf depth (from Hood et al. [2])
The finding that kerf depth is relatively insensitive to jet traverse speed at high traverse speeds is somewhat surprising and it is important. (Presumably the low efficiency of the rock removal process at low traverse velocities occurs because the frictional losses of the jet stream increase as the kerf depth increases.) The efficiency of the kerfing process is the inverse of the Specific Kerfing Energy, £ sk , where (Maurer [3]) E« =
Jet Energy Kerf Area on one side of kerf
Jet Power Kerf Depth x Traverse Speed
P ~dv
(2)
Now, at high traverse speeds d % constant as v increases and P = constant, hence Esk decreases. Thus, at high values of v, the efficiency of the kerfing operation increases as v increases. This says that rock excavation systems employing high pressure water jets should be operated in a manner that causes high traverse velocities of the jets across the rock surface. Obviously, this traverse velocity should not exceed the critical velocity beyond which no damage is caused to the rock. 9.2.2 High Pressure Jet Drills This method of rock destruction has been used successfully in a limited range of rock types for drilling small diameter holes. In this case an array of high pressure nozzles is arranged in a manner on the bit face both to cut kerfs in the rock and to erode the rock left between these kerfs. A typical hole produced by a water jet drill is shown in Figure 6. In some rocks very rapid penetration rates have been reported using jet drills of this type. For example, Summers and Bushnell [4] achieved rates of 7.12 m min" 1 in Berea sandstone using the drill shown in Figure 6. Figure 6 illustrates a number of features that are characteristic of holes drilled by water jets. One of these is the hole size. It can be seen that the diameter of the hole is considerably greater than the drill diameter. This is always the case. In a given rock type for a given jet pressure and flow rate, the hole diameter is controlled by the angular placement of the jet nozzles in the bit, the rate of drill penetration and the fluid pressure in the hole. Another feature of note is the rough sides to the hole wall. To some extent, this roughness can be controlled by adjusting the jet parameters and the drilling rate. In some circumstances a rough hole wall is advantageous. For example, in roofbolt holes the rough sides allow better retention of the resin and better grip of the bolt. A feature that is a principal disadvantage of water jet drills also is illustrated in Figure 6, that is the cone of rock material that remains intact in the bottom of the hole. This rock cone can develop to a size where further penetration in the hole by the drill becomes impossible. To a certain extent the size of this rock cone can be controlled by angular placement of the jet nozzles in the bit face. Drilling experiments conducted by Maurer and Heilhecker [5] in Berea sandstone and Indiana limestone indicated that, for a given nozzle arrangement in the bit, drilling rate increased roughly linearly with jet pressure. However, since jet power increases with jet pressure to the power 1.5, a jet pressure exists at which the specific energy, Es, of the drilling process is minimized. (In this case we
234
Mechanized Excavation
Figure 6 Hole drilled in Berea sandstone using a water jet drill (from Maurer [3]; published originally by Summers and Bushnell [4])
lOOOh-''
0
I
Nozzle-3mm=O.II7in.
I
10
I
20
I
30
_J 40
I
50
I
60
I
70
I
80
Pressure (lOOOpsi) Figure 7 The influence of jet pressure on specific energy (from Maurer [3]; originally published by Chadwick [6])
are interested in the excavation of the hole and not just the cutting of a kerf. Es # E sk . Es = Jet Energy Hole Volume = Jet Power/(Hole Area x Drilling Rate).) These authors found that this pressure was about 35 MPa for their bit in Berea sandstone. At this pressure Es = 2560 MJm" 3 and the drilling rate was 0.63 m min"1. This value for specific energy is probably an order of magnitude higher and the value for drilling rate is about a factor of two lower than could be achieved in this rock using a conventional mechanical bit. Thefindingof Maurer and Heilhecker [5] that increasing jet pressure caused the specific energy to decrease from infinity, because no rock is removed at low jet pressures, to a minimum value beyond which it increased, was verified by Chadwick [6] for different rock types in a series of kerfing experiments (Figure 7). 9.3 DISCONTINUOUS JETS 9.3.1 Theoretical Considerations These jets apply an impact, or impulsive, force on the target. The maximum pressure generated by this impacting process is higher, often considerably higher, than that generated by continuous jets. Consequently the destructive ability of discontinuous jets potentially is much greater.
The Use of Water Jets for Rock Excavation
235
When a body is fired normally at a semi-infinite target, compressional waves are set up both in the body and in the target. The magnitude of the peak pressure in this wave, pi9 is Pi =
(3)
peu
where: c = the compressional wave velocity in the impacting body ( = 5 km s~* for steel; 1.5 km s _ 1 for water) p = the density of the impacting body ( = 7.85 t m~ 3 for steel; = 1 t m " 3 for water) u = the velocity of the impacting body (m s" 1 ) When these units are used the pressure p{ is calculated in M Pa. The ratio of the peak pressures of an impacting and a continuous water jet is given bv 2pcu pu2
EL Po
1c u
(4)
The velocity of a continuous water jet = u = ^(Ip/p) The maximum pressure of these jets with current pump technology is about 345 MPa. Thus the maximum jet velocity of a continuous water jet is about "max = 830 ms"
2c
EL Po
= 3.6
(5)
Thus if a discontinuous water jet is directed at a target at the maximum velocity of a continuous water jet (about 830 m s " 1 ) the peak stress induced in the target is approximately 3.6 times the stress that would be induced by a continuous jet. If the impacting body is solid, then the compressional wave travels to the end of the body where it is reflected as a tensile wave. When this wave reaches the interface between the body and the target they separate; thus the time during which the compressional wave acts in the semi-infinite target, i s , is given by 2/
i. = — c
(6)
where / = the length of the impacting body. The situation is somewhat different when the impacting body is a liquid. When the liquid strikes the surface it spreads radially across the surface (Figure 8). If the slug of liquid is cylindrical, a compressional wave is initiated from the corner of the cylinder at the interface (Figure 8). When this wave reaches the center of the slug, it is cancelled by the wave traveling from the opposite side. Thus the time during which the wave acts on the target, £w, is tw
where r
(7)
=
the radius of the cylindrical slug of liquid
Thus
tw
1"
r
(8)
I" Z\
Figure 8
Impact of liquid cylinders (after Brunton [7]) (u, velocity of liquid cylinder; c, compressed wave velocity in the liquid)
Mechanized Excavation
236
Length of contact of droplet and plane surface Figure 9 Impact of a spherical droplet
The time during which this pressure acts is even lower if the liquid slug is not cylindrical but has a curved front. Consider, for example, a spherical droplet striking a plane target (Figure 9). Here the geometry allows the liquid to escape more easily. The same peak pressure p{ = peu is attained over an area defined by the radius x 0 , where x0 < r and r = radius of the sphere. In this case c
For small values of u, X 0 becomes vanishingly small. Table 1 gives peak pressures and the times during which these pressures are applied when two spheres are fired at a target. One sphere is a steel ball, the other is a slug of water. Three different cases are illustrated. The first example is a low velocity, large diameter sphere; the second is a high velocity, large diameter sphere; and the third is a high velocity, small diameter sphere. First, it is apparent that, for all three cases, the peak pressure for steel is greater than that for water by a factor of 26.67. This follows because rjrw = 8 and cjcw = 3.33, so that for a given slug velocity rscs/rwcw = 26.67. Another point to note is that the ratio of the time that the pulse is applied is 30 times greater for the steel sphere at the low impact velocity but this reduces to a factor of two at the high impact velocity. This can be seen from ii rw
=
JSzp. \cju
and since °- = 0.3 ± = 1.2^ c8 rw u
The energy imparted to the target by the sphere is computed by integrating the pressure over the area to which it is applied and then multiplying this value by the time. The calculations for peak pressures were made assuming the target to be a rigid solid. In practice of course the target will not be rigid and when impact occurs it will deform, probably elastically. In Table 1 Peak Pressures and Pressure Pulse Durations Velocity (ms" 1 )
Radius (mm)
Peak pressure (MPa)
Time
Steel Water
60
Steel Water Steel Water
900
37.5 37.5 37.5 37.5 0.5 0.5
2400 90 36000 1350 36000 1350
30 1 30 15 0.4 0.2
900
(μβ)
237
The Use of Water Jets for Rock Excavation this case p^pcu
but is given by Pi =
(10)
;
where pt = density of target material and ct = compressional wave velocity in target material. In summary, it is apparent that both the magnitude of the induced stress and the time during which this stress acts in the target are greater when the impacting body is a solid. The former parameter determines whether fractures are induced in the target. The latter parameter controls the length to which these fractures grow. On the other hand, when a liquid such as water is used as the projectile the liquid can penetrate and help propagate fractures induced in the rock by the impacting process. In this way liquid jets can be more effective than solid projectiles in effecting rock destruction. There are two main types of discontinuous water jets. One is simply an interrupted version of the small diameter (generally fractions of a millimeter to a few millimeters) continuous high pressure water jets. The frequency at which these small packets of water are directed at a rock face is variable but often it is of the order of several hertz. The other type of jet is a much larger slug of water that is fired at the rock face at a much lower frequency, and often at a much higher jet pressure. The jets in this latter category are fired from devices known as water cannons. 9.3.2 Interrupted Continuous Jets Because the stresses set up in the target by the impacting action of discrete packets of water can be much higher than the stagnation pressure beneath a continuous jet, there has long been interest in using discontinuous water jets for rock erosion. Early studies were conducted using a rotating plate to interrupt a continuous water jet. It is apparent that the efficiency of this process is not great because the jet energy is wasted whenever the jet strikes the plate. An ingenious solution to this problem was found by Nebeker and Rodriquez [8]. They employed a device termed a flow rate modulator (Figure 10) to modulate theflowof a continuous jet. In other words, this device cycles the jet discharge velocity with respect to the average jet velocity. The amplitude of this modulation is small compared with the average discharge velocity. The drag on the jet, which is still a continuous stream when the jet leaves the nozzle, causes the stream to break up into discrete packets (Figure 11). These jets, termed 'percussive water jets', have been found to be much more effective in causing rock damage than conventional continuous water jets. 9.3.3 Water Cannons In these devices a plastic or metallic piston is fired at high velocity into a long, slowly converging nozzlefilledwith water (Figure 12). The piston often is driven by a high pressure gas to velocities of several hundred meters per second. The nozzle geometry controls the jet velocity. Jet velocities in excess of 3500 ms" 1 (equivalent to jet stagnation pressures greater than 6.125 GPa) can be developed using these cannons. The damage caused to a rock surface by a water cannon was described by Young [10]. When the water slug isfirednormal to the face thefirstfew impacts produce large fragments and form a crater. Subsequent impacts in the same location produce smaller fragments and a hole starts to form Bearing
Nozzle
Stator cylinder Figure 10
Rotor cylinder
Shaft
Barrel wall
Flow rate modulator (from Maurer [3]; originally published by Nebeker and Rodriquez [8])
Mechanized Excavation
238
Discharge
Train of drops
Bunching stream
Figure 11 Percussive water jet (from Maurer [3]; originally published by Nebeker and Rodriquez [8])
Piston
Contact pins for piston speed measurement
Pressure Transducers
fcfe t Water
Water reservoir and nozzle assembly
Figure 12 Water cannon showing piston and nozzle assembly (from Edney [9]) (b)
Nozzle :H3 at 10mm for 1+2 at Omm for 3 at -10mm for 4+5
Fracture turns toward surface
Rock: bonus granite 5-developed large fracture II jet
Reference free surface (from previous bench advanced
New reference surface
Fracture II to jets 1st. impact \ crater 2nd impacf crater and h o l e \ 3 r d impact hole
Figure 13 Rock damage by water cannon producing (a) small fragments and (b) large fragments (from Maurer [3]; originally published by Young [10])
(Figure 13a). When the jet is directed at an angle (30° - 45°) to the face the first impact produces a crater at the toe of the bench. Subsequent impacts begin drilling a hole, and finally a large rock fragment is produced (Figure 13b). Although this water cannon method for rock breaking is effective and, from the viewpoint of the physics of rock breakage it is attractive, most of the research work in this area has been abandoned.
The Use of Water Jets for Rock Excavation
239
The reason for this, apparently, is the difficulty in engineering reliable units that can operate in a mine environment and that can fire water pulses at the face at a sufficiently rapid rate. 9.4 CAVITATING WATER JETS In these continuous water jets vapor bubbles are formed in the jet stream. When these bubbles collapse against a solid surface, such as a rock face, very high pressures are generated. These pressures can cause considerably greater damage to the surface than would result from conventional continuous jets. Thus, the advantage of a cavitating jet is that low jet pressure can be used to effectively erode a rock face. The disadvantage of this jet system is that in order for vapor bubbles to form and to travel from the nozzle to the rock face, a large diameter jet is required. Often jet diameters of several millimeters are employed. A common method used to form these vapor bubbles is to place an object in the jet stream (Figure 14). When these jets are used to cut kerfs in a rock face they generally result in wider and deeper kerfs than conventional continuous water jets operating at the same jet pressure and flow rate. Experiments have been conducted kerfing rock with the cavitating water jet nozzle and the rock sample mounted in a water-filled pressure cell (Conn and Radtke [12]). The ambient pressure of the water in this cell could be varied to simulate the high mud pressures that exist in oil and gas wells. It is to be anticipated that an increase in the ambient pressure downstream of the nozzle would serve to suppress the formation of cavitation bubbles, and hence to decrease the effectiveness of cavitating water jets. This expectation was realized in the experiments conducted by Conn and Radtke. They found that the kerf depth in Berea sandstone decreased from 54.8 mm to 19.1 mm when the fluid pressure in the cell was increased from atmospheric to 20.7 MPa. However, the decrease in jet effectiveness with ambient pressure was less clear when cutting tests were made in Indiana limestone. Additional work by these investigators using cavitating water jets to cut kerfs in Indiana limestone (Conn and Radtke [13]) indicated that the excavation rate actually increased, and increased sharply, with increasing back pressure at low values of this back pressure. Conn and Radtke claimed that a maximum excavation rate was observed beyond which the rate of rock removal decreased as back pressure increased. Inspection of their data (Figure 15) shows that some licence was used in reaching this conclusion (note the dashed portion of the curves in Figure 15). However, this finding by Conn and Radtke that a maximum rate of rock excavation is observed when the hydrostatic pressure increases is supported by an earlier study of rock erosion using cavitation (Figure 16b). Angona [14] did not use a jet in his work. Instead he induced cavitation in a waterfilled pressure chamber using a focused acoustic system (Figure 17) to produce negative pressures sufficient to induce cavitation in a particular small volume within the chamber. The rock sample was mounted at this focal point. The sound pressure and the hydrostatic pressure in the chamber could be varied independently over the range of 0.1 - 2.0 MPa. Thus, unlike the situation with the jet where cavitation is suppressed as the hydrostatic pressure increases, here cavitation could be maintained by setting the acoustic pressure always somewhat higher than the hydrostatic pressure. This was done and it was found that (i) Substantial damage was caused to the limestone samples tested.
Cylindrical center body Cavitation
Figure 14 The centerbody CAVIJET™ nozzle design (from Conn [11])
240
Mechanized Excavation 280i-
24.0
20.0 h
I I I I I I I I I I I '\
I
Λ
Tests at Drilling Research Lab., May 24-26,1978 Nozzle: plain l/4in.Cavijet TI Standoff distance: 5/8in. Drilling fluid :mud, 9.3ppg ° Rock:Indiana limestone
»Ap = 2950psi
'/ * 16.0
3
12.0
8.0
> < 4.0
1000
2000
3000
4000
Ambient (bore hole) pressure, Pa(psi) Figure 15 Cavitating jet data (from Maurer [3]; published originally by Conn and Radtke [13])
PA = 8.0atm w
2500 l·-
<
i 12
P L 0 (atm)
16
20
24
P L0 (atm)
Figure 16 (a) Weight loss Δ W as a function of hydrostatic pressure and (b) weight loss Δ W for constant values of applied acoustic pressure PA (from Angona [14])
(ii) The rate of damage increased as the hydrostatic (and acoustic) pressures increased (Figure 16a). (iii) When the acoustic pressure was held constant the rate of damage increased rapidly initially with increasing hydrostatic pressure. The maximum erosion rate was observed when the acoustic pressure and the hydrostatic pressure were about equal. When the hydrostatic pressure exceeded the acoustic pressure the erosion rate decreased (Figure 16b). Work continues to be conducted using cavitation systems for rock erosion. This approach holds promise for specialized applications, such as drilling.
The Use of Water Jets for Rock Excavation
241
Transducer
Figure 17 Cavitation pressure chamber with rock sample positioned at the focal point (from Angona [14])
9.5 ABRASIVE WATER JETS Two approaches for forming water jets with abrasive particles entrained in the jet stream have been developed. One of these approaches employs a high pressure (typically 200 - 340 M Pa) relatively low flow rate water jet system. The other approach employs a lower pressure (typically 70 M Pa maximum) higherflowrate water jet system. Figure 18 illustrates the nozzle arrangement of the higher pressure unit. The high pressure jet is formed using a sapphire nozzle, labeled as Instajet™ in this diagram. The velocity of this jet flowing through the chamber 3 pulls a vacuum that draws the abrasive particles into this chamber through tube 2. The abrasive particles and the jet enter the abrasive nozzle through the carbide sleeve and the particles are accelerated to a velocity approaching that of the water jet in this abrasive nozzle. The lower pressure system pumps a slurry of the abrasive particles and the water through the nozzle (Figure 19).
Instajet T Nozzle holder
Abrasive nozzle holder
Cable tip
Nozzle sleeve
Nozzle guard
Abrasive nozzle
Figure 18 High pressure abrasive jet nozzle assembly (from Hunt et al. [15])
Dry abrasive
I
Pressure vessel
100 bar pump
Pressurized abrasive/water mix
Low pressure " water
4
Jetting jig
Figure 19 Low pressure abrasive jet cutting arrangement (from Fairhurst et al [36])
Mechanized Excavation
242
Both of these jet systems offer an ability to make cuts in materials that are difficult to cut by other methods. For example, in cutting reinforced concrete they will cut both through the aggregate and the reinforcing steel. The advantage of the lower pressure system is the more simple operation that comes from low pressure units. However, the consumption of abrasive particles is the higher for this low pressure system, and since the cost of the abrasives is one of the most significant operating costs for these cutting units, this is a serious disadvantage of this system. Little has been published on the use of the low pressure system for mining and rock excavation application. A few investigations have been made using the high pressure system for this purpose. Fort et al [16] conducted experiments to investigate the parameters that had the greatest influence on the kerf depth. They concluded that the factor that most affected kerf depth was jet power. They showed that, over a broad range of jet pressures and flow rates, it was unimportant whether this power was provided as high pressure or higher flow rate. They showed also that, again over a broad range of values, the mass flow rate of the abrasive had only a secondary influence on kerf depth. This study and other work (Hashish, 1989, private communication) demonstrated that the type of the abrasive particles employed has a substantial influence on the cutting performance. The density, size, shape and hardness of the particles all are significant, but density is by far the most important of these properties. Apparently in this high pressure system the jet accelerates the particles to about 80% of the jet velocity. In this situation, obviously, the higher the density of the particles the higher their momentum and therefore the greater the rock damage. Field trials have been conducted in an underground gold mine using an abrasive jet system to cut kerfs in abrasive quartzite adjacent to the gold reef (Marlowe et al [17]). The concept in this case was to determine whether a method could be developed in which the narrow reef could be removed separately from the waste rock by cutting the reef from the face. Three different materials were used for the abrasive particles in these tests: garnet, chromite and quartzite. These tests showed that the 60 kW system employed could be used to cut 250 mm deep kerfs in the face at a jet traverse rate of 100 mm m i n - 1 using chromite as the abrasive particles. In this case chromite was selected as the ideal abrasive both because it was effective and because it was a waste product from nearby mines and therefore was inexpensive. The high pressure systems were found to operate satisfactorily in the very hostile environment of a deep-level gold mine. The conclusion of the test was that the results were encouraging but that deeper kerfs were needed if a technique was to be developed to remove the cut reef from the face. The studies described above all employed an abrasive jet nozzle that did not penetrate the kerf as it was being cut. Because the efficiency of the cutting operation decreases rapidly with increasing kerf depth this implies that high power systems will be needed to produce deep kerfs. An alternative approach to deep kerfing has been reported by Echert et al [18]. Echert, Hashish and Marvin used a high pressure but low power unit with a nozzle that penetrated to the bottom of the kerf to make 25 mm wide, 1.5 m deep kerfs in concrete. In another development a small diameter (35 mm) abrasive jet drill has been built and tested by the Bureau of Mines (Savanick [19]). This drill operates on the principle of the high pressure unit described above at a pressure of 70 MPa, with a water flow rate of 1.31 s" 1 , and with an abrasive flow rate of 10 kg min - 1 . Rocks with uniaxial compressive strengths as high as 500 MPa have been drilled, albeit at a fairly slow drilling rate (100 mm min - 1 ). However, weaker rocks such as Indiana limestone with a uniaxial compressive strength of 55 MPa have been drilled at a respectable 0.76 m min - 1 . In summary, abrasive water jets are unlikely to be a universal answer to the varied excavation problems that exist. However, this approach to rock cutting appears to offer significant potential in some applications. 9.6
COMBINED USE OF HIGH PRESSURE JETS AND MECHANICAL CUTTING TOOLS
9.6.1 9.6.1.1
Mechanically Assisted Cutting Principles of the cutting method
The most straightforward approach to rock cutting with high pressure water jets and mechanical tools is to employ the jets to weaken the rock. This is achieved by cutting a series of kerfs in the rock face with the water jets before the remaining rock is removed by the tools. This approach is termed 'mechanically assisted cutting' because the energy consumed in eroding the fraction of the rock removed by the jets generally accounts for 70-90% of the total energy required for the rock removal
The Use of Water Jets for Rock Excavation
243
process. This is the case despite the fact that usually only a small fraction of the rock (often less than 10%) is removed by the water jets. This demonstrates the low efficiency of high pressure water jets as a method for rock removal. However, since the strength of the tool materials provides a fundamental limitation on the rate of rock cutting or drilling with mechanical tools and drill bits in some circumstances the operator may be happy to pay the additional energy costs associated with the use of high pressure jets if this enables the rate of rock excavation to be increased. (The problem here is that the rate of rock cutting (or rock drilling) increases with the power applied to the cutting tools (or drill bit). Unfortunately the wear of the tool (bit) materials is proportional to the heat generated in the tool material during the cutting process. The amount of this heat increases with the power applied to the tool. Hence the upper limit on the rate of rock excavation using mechanical tools is given by the upper limit of the power that can be applied to the tool without causing excessive rates of tool wear.) Because this method relies on cutting a series of uniformly spaced kerfs in a rock face it has been applied mainly for drilling holes. For this purpose a drill bit is constructed with an array of nozzles mounted across the bit face. The mechanical tools, either drag bits or indentation type tools, are mounted on the bit to remove the ridges of rock left between the kerfs. The effect of these kerfs is to reduce substantially the forces necessary at the tool to cause rock failure because the presence of the kerfs promotes crack growth. This leads to chip formation with reduced bit forces. 9.6.1.2 Drilling small holes (smaller than 50 mm diameter) The Bergbau-Forschung (Feistkorn and Knickmeyer [20]) appears to have been one of the first organizations to investigate the benefits of a mechanically assisted rotary drill. Feistkorn and Knickmeyer conducted experiments both in the laboratory and underground using the drill bit shown in Figure 20. Their laboratory tests were performed in a sandstone with a uniaxial compressive strength of 80-100 MPa and containing up to 70% quartz. One of the interesting results reported by these authors is the approximately linear increase in the drilling rate with rotary speed at a constant thrust up to a critical rotary speed beyond which the penetration rate is independent of speed (Figure 21). Thisfindingcan be explained in terms of the results of erosion tests cutting kerfs in rock where it is found that over a wide range of traverse speeds the depth of the kerf cut by the jet is relatively independent of the traverse speed of the jet across the rock (Figure 3). Consequently, it might be anticipated that increasing the rotational speed of the drill will result in an increase in the drilling rate. It might be predicted that as the traverse speed of a jet across the rock surface increases, in the limit no erosion will take place because the dwell time of the jet on the rock is too short. This may explain why the curves in Figure 21 level out at rotational speeds in excess of 400 rpm (this corresponds to a linear velocity of 0.9 m s~ * for the jets that cut the hole gauge). Feistkorn and Knickmeyer reported other very interesting results. They showed that at a given thrust and rotary speed the drilling rate increases steeply with jet pressure (Figure 22). It should be
Figure 20 Rotary drill head with 10 nozzles (Feistkorn and Knickmeyer [20])
244
Mechanized Excavation Thrust (kN) 3.0h
2 5
I
|
2.0H σ c
o
l.5rSandstone Bit diameter 4 5 m m 10 Nozzles 0 . 4 m m Jet pressure 2 2 5 0 bar 0 . 3 % Nalcotrol B
I Or
0.5h
0
100 200
300
400
500
600 1
Rotational speed (min" ) Figure 21
Drilling rate as a function of rotational speed and thrust (Feistkorn and Knickmeyer [20])
Sanstone 10 Nozzle 0 . 4 m m 1 3.0 |_ Rotational speed 400min" Thrust 10 kN 2.5|
E E o c
2.0
1.5
1.01
Q 0.5
0
-j
i—
1500
2000
2500
Jet pressure (bar) Figure 22
Possible drilling rates as a function of jet pressure (Feistkorn and Knickmeyer [20])
noted that the water jet power needed at a pressure of 240 MPa (or 2400 bar) for ten 0.4 mm diameter nozzles (assuming a coefficient of discharge of 0.7 for these sharp orifice nozzles) is 146 kW. From Figure 22 it can be seen that this power input produced a penetration rate for this drill in sandstone of about 2.5 m min" l . Perhaps of most interest was their observation that when using water jets, the average drilling rate was reduced only from 2.8 m min" * when the bit was new to 2.3 m min" * when the bit had drilled 15 m of hole. This compared to the situation when the water jets were not used where the bit was worn down after only 0.3-0.4 m of hole drilled. Feistkorn and Knickmeyer concluded that, for this benefit of dramatically reduced bit wear alone, the use of water jets represented a significant advance in drilling technology. A considerable amount of work has been carried out with mechanically assisted small hole drills by Flow Industries, a Seattle based company. Almost all of this work remains unpublished although in concept the bits that they used were similar to those employed by Feistkorn and Knickmeyer. Most of the work conducted by Flow Industries used water jets at 380 MPa pressure. Their work was focused in two areas. One project was the development of a lightweight hand-held drill for very hard rock. Another was the development of a roofbolt drill for coal mines. The drill for this latter project was tested extensively in a mine. Although this unit suffered severe 'teething' problems,
The Use of Water Jets for Rock Excavation
245
reportedly by the end of the development program the drill worked reliably. It drilled 40 holes before it was necessary to resharpen the bit, compared with only 1/2 a hole drilled with a bit on a conventional roofbolt drill in that same section of the mine, and it outperformed the conventional drill by three to one in terms of drilling speed (McFarland, private communication, 1989).
9,6.13
Drilling medium holes (larger than 50 mm but smaller than 300 mm diameter)
Maurer and Heilhecker [21], working for Exxon, appear to have been the first workers to conduct experiments with mechanically assisted drill bits in oil wells. Maurer used three different types of drill bits in his experiments, a roller cone, a drag bit and a diamond bit (Figure 23). The high pressure fluid that he delivered to the rock face was drilling mud rather than water. The pressure delivered by the high pressure mud pump was 103 MPa, although inevitably pressure losses in the system reduced this pressure at the nozzles, particularly since some of the wells in which drilling was conducted were as deep as 3350 m. The field tests that Maurer conducted demonstrated that the use of jets to pre weaken the rock could increase drill penetration rates by two to three times. This is illustrated in Figure 24 which shows the results from two adjacent holes, one employing the high pressure water jet system and the other drilled conventionally. It is apparent that the time taken to deepen the mechanically assisted well from 2000 ft to 6000 ft was 24 hours whereas 67 hours were required to extend the conventionally drilled well over this interval. Despite promising results this work was eventually abandoned by Exxon. Several factors seem to have led to this decision, many of which were related to the reliability of the hardware. One persistent hardware problem was failure of the high pressure swivels and high pressure pumps. Another problem was excessive rates of erosion of the water jet nozzles. In some cases the pressure available at these nozzles was below the rock threshold pressure. Consequently, no erosion of the rock by the jets took place and no benefits in terms of increased drilling rates were observed. Substantial efforts were made to overcome these difficulties; however, although considerable progress was made, eventually the project was terminated. Other workers and other companies, including Shell Oil, continued to work in this research field but no systems were ever commercialized from this work. The most recent work in this area currently is still in progress and is being conducted by the Seattle company, FlowDril Inc. The approach adopted by this company is somewhat different from that employed during the Exxon and Shell tests. One of the difficulties, at least with the Exxon
Figure 23
Exxon jet bits (Maurer et al. [21])
246
Mechanized Excavation
Q.
Φ •Ό
( 2 0 0 0 psi)
High-pressure bits I0 0 0 0 - I 5 0 0 0 ( p s i )
10
J
20
I
L_
30
40
50
J_ 60
J 70
Rotating time (h) Figure 24 Oil well jet drilling data (Maurer et al. [21])
experiments, was that the power required for thefluidjets was very high. This is because the full mud flow in the hole was pressurized to 100 M Pa. Also erosion of the high pressure nozzles by the mud jets was problematic. FlowDril has attempted to overcome these difficulties by circulating the major part of the drilling mud (about 90% of the conventional mud flow) through two of the three nozzles in a rotary bit at conventional mud pressures. Only a fraction of the mud flow (the remaining 10%) is pumped at high pressure (about 170 MPa) through a concentrically mounted conduit in the center of the drill pipe. Before this smaller fraction of the drilling fluid is pressurized it is filtered in order to minimize wear both in the high pressure pumps and in the high pressure nozzle. This highly pressurized fluid is directed at the rock face in the vicinity of the hole gauge through the third mud nozzle. This third nozzle is extended to minimize the standoff distance to the rock. Because only a fraction of the mud is circulated at high pressure, the power requirements for the fluid system are not excessive. The jet pressure is higher than that used by most previous workers for Red fork sandstone 6 0 0 psi compressive strength
Flow drill,25OOOpsi Conventional
CL Q ÛC
632
636
640
644
<
Depth ( f t ) Arbuckle dolomite ISOOOpsi compressive strength
Q.
a cr
a. a
ÛC
1620
1640
Depth ( f t )
Figure 25
Field test results in (a) Red fork sandstone, (b) Mississippi limestone and (c) Arbuckle dolomite (Hood et al. [22])
The Use of Water Jets for Rock Excavation
247
this type of application and, because jet energy losses are minimized by minimizing the standoff distance, this pressure is greater than the threshold pressure of most sedimentary rocks. Consequently the jet is effective in cutting a kerf in the vicinity of the hole gauge in most of the rocks drilled during the drilling operation. This results in marked reductions in the force acting on the bit. Hood et al. [22] reported impressive improvements in drilling rates in three different rock types using this drilling system (Figure 25). 9.6.1.4 Drilling large holes (larger than 300 mm diameter) At least twofieldexperiments have been conducted using mechanically assisted cutting on tunnel boring machines (TBMs). In 1974 Wang et al [23] published results of laboratory tests where kerfs were cut with the water jets directed both in the path and between the paths of the disc cutters. Wang, Robbins and Olsen showed that, for a given disc thrust force, the depth of cut taken by a cutter was increased by a factor of about three, while the spacing between the cutters on the head also was increased. Wang, Robbins and Olsen concluded from these tests that a 300% improvement in the advance rates of TBMs could be achieved by using water jet kerfs to preweaken the rock. The first full-scale field test of these ideas was conducted by Wang et al [24]. A 2.1 m diameter TBM was equipped with 31 high pressure water jet nozzles, each of diameter 0.3 mm, and 21 disc cutters. The jet power employed was 750 kW and the jet pressure used was 393 MPa. The machine was used to drive a tunnel in granite with uniaxial compressive strength values ranging from 159-262 MPa. In contrast with the results from his earlier laboratory tests Wang found that no additional increases in TBM advance rates were realized when the water jets were directed in the line of the cutter paths. However, he reported an increase in these advance rates of 40-48% when the jets were directed between the cutter paths. One of the most serious operational problems with this field trial was the poor reliability of the high pressure pumping system. The state of the art pumps and other high pressure components were inadequate at that time to permit continuous cutting for more than a few minutes at a time. Another field test conducted a few years later by the Bergbau-Forschung used a 2.65 m diameter TBM equipped with disc cutters and with an installed mechanical power of 320 kW. The cutter head of this machine was fitted with about 100 high pressure water jet nozzles. The maximum pressure that could be generated across these nozzles was 400 MPa and the installed power for the water jet system was 1 MW (Henneke and Baumann [25]). This machine is shown in Figure 26. The results from these experiments, reported by Baumann and Henneke [26], showed that the advance rate of the TBM was doubled when the jets were used (Figure 27). Baumann and Henneke also noted that the concept of using water jets at pressures as high as 300-400 MPa was technically feasible in an underground environment. With this statement they indicated that the hardware for systems of this type had advanced substantially since Wang, Robbins and Olsen had carried out their studies.
Figure 26
TBM equipped with disc cutters (Henneke and Baumann [25])
248
Mechanized Excavation Nozzle arrangement 2 / Π α Type of rock sandstone Fv = 6 0 0 k N const. Waterpressure s 3 2 0 0 bar o Using water jets • Without water jets
Rotational speed,/? (mi rf1)
Figure 27 Influence of high pressure water jets and of rotational speed on the drilling rate
In the light of these various laboratory and field experiences, Hustrulid [27] examined the economic feasibility of mechanically assisted TBMs. He concluded that in hard rock in order to double the advance rate of a 6.1m diameter machine from 1.22 m h" 1 without water jets to 2.44 mh" 1 with jets, 192 water jets operating at a pressure of 345 MPa would be needed. This corresponds to an installed power for the water jet system of more than 4 MW, or about an order of magnitude more power than installed on a conventional machine. 9.6.2 Jet-assisted Cutting 9.6.2.1 Principles of the cutting method Rather than employing water jets to erode a rock face directly, this cutting method utilizes these jets to erode crushed rock debris formed by mechanical tools during the rock cutting process (Figure 28). The goal in any mechanical rock cutting operation is for the tool to induce fractures in the rock to form discrete rock chips. Unfortunately, in the process of initiating and propagating these fractures, all mechanical tools produce regions of crushed rock immediately beneath and adjacent to the tool. The stresses induced in the rock by the tool are reduced by the cushioning action of this crushed material. This raises the stress that it is necessary to induce in the tool in order to form a rock chip. A mechanism that has been proposed, and which is widely accepted, to explain why water jets are effective in reducing tool forces, is that the jets continuously flush this crushed material away from the tool during the cutting process (Dubugnon [28]). In order for this cutting method to be effective the jets must be directed accurately into the zone where the crushed rock is formed, as it is being formed. This means that the jets must be directed immediately adjacent to the cutting tools. Furthermore, the jets must be sufficiently energetic to erode this compacted crushed material. In addition to removing the crushed rock, Dubugnon [28] and Hood [29] suggested that the jets assist the rock breakage process by helping to propagate the fractures that create the major rock chips (Figure 28). These fractures are initiated by the tool in the region where the water jets are directed, consequently it is almost inevitable that water will enter
^
©
c
^
Figure 28 The point of impact necessary for a water jet to exploit the rock damage caused by a drag bit (after Dubugnon [28])
249
The Use of Water Jets for Rock Excavation
these cracks. Basic fracture mechanics principles tell us that the amount of external energy required to propagate a crack is reduced by the pressure energy applied internally within the crack.. Thus if crack propagation is assisted by the action of the water jets one would expect that the tool force necessary to drive the cracks would be reduced. The term 'jet-assisted cutting' comes from the fact that the principal source of energy for the rock breaking operation is provided by mechanical tools. The energy from the water jets supplements, or assists, this breakage process. It is apparent that this contrasts with mechanically assisted cutting in which the water jet energy is the principal energy source for the breaking operation. The benefits of jet-assisted cutting remain controversial. Several research workers claim the following substantial benefits for this cutting method: reductions in cutting tool forces, improvements in tool life, reductions in the dust generated during the cutting process, reductions in incidences of frictional sparking in gassy environments and reductions in vibration of the cutting machine. Other workers accept that all of these benefits can be obtained under laboratory conditions but they claim that in practical mining and tunneling situations many of these benefits cannot be reproduced. In large part this controversy seems to have developed because the benefits of jet-assisted cutting are influenced by a large number of factors. These include: the rock type, the depth of cut taken by the tool, the cutting velocity of the tool, the tool geometry, the number of water jets used to assist each tool, the position of these jet(s) with respect to the tool, the jet pressure and the jet flow rate. Studies conducted by the various research workers have examined some aspect of the problem but, to date, no unifying mechanistic theory of this cutting process has emerged to reconcile apparently contradictory findings by different research groups. In the sections below we have attempted to summarize these findings in a manner that conveys, hopefully, the current state of knowledge in this field.
9.6.2.2
Drag bit force reductions - general
Considerable work with jet-assisted cutting systems has been carried out over the past decade. Most of this work has employed drag bits as the cutting tools and a wide variety in the geometry of this type of tool have been examined. Also a wide variety of rock types have been cut in this experimental work. In general, these studies have confirmed Hood's findings [30] that an appropriately directed water jet system usually can reduce the bit cutting force by a factor of at least two and the bit penetrating, or thrust, force by a factor in excess of two. Typical results from laboratory work by Ropchan et al [31] and Dubugnon [28] are given in Figures 29 and 30. The work of Ropchan, Wang and Wolgamott is of interest because they used two different types of pick, chisel and point-attack and they conducted experiments directing the water jet at the rock-tool interface from a position behind the pick. Dubugnon conducted his work using a bit similar to that employed by Hood [30] and, like Hood, he used a two-jet arrangement for the water jets with these jets directed immediately in front of the bit, towards the bit corners. However, Dubugnon used three different rock types: a strong, finegrained sandstone (uniaxial compressive strength of 150 M Pa), Bohus granite with a uniaxial
(b) I500r-
(a ) •
Without water jet
o With water j e t , 10000 psi pressure •O
-
•
2000 h
3
1000 h
• a» σ
1000 h
•
•
o 1
0
o
O
z 1
1 0.6
1 0.8
0
0.4
0.6
Bit penetration (in.)
Figure 29 Bit cutting (drag) and bit penetrating (normal forces with and without jet assistance). The jet pressure used in these tests was 70 MPa (after Ropchan et al. [31])
250
Mechanized Excavation o • Δ A
100
o
T
-> "
Δ
Fc Fc FT FT
av. peak av. peak
'**-» •
"
^
^—
' Δ
1
1
L
— !..
1
Initial depth (mm)
Figure 30 The reduction of the bit forces (given as a percentage of the forces measured when cutting dry) using 85 MPa water jets to assist a drag bit cutting Bohus granite. Fc is the bit cutting force and FT is the bit penetrating force (after Dubugnon [28])
compressive strength of 200 MPa, and a tough, fine-grained, siliceous limestone (uniaxial compressive strength of 250 MPa). He found that the extent of bit force reductions caused by the jets was influenced strongly by the rock type. For example, the average bit thrust force was reduced by as much as 80% of the value measured when cutting dry in the granite but this same force component under the same cutting conditions was reduced by only 65% in the sandstone. Dubugnon suggested that these different responses in these different rock types may be attributed to the dilatational behavior of the rock. He argued that the ability of a water jet to penetrate rock zone adjacent to the tool is proportional to the rock dilatancy. Thus, he reasoned, rocks such as granite, that exhibit high dilatation, are more easily assisted by high pressure water jets. An explanation that sometimes is put forward to account for the lower values of the bit cutting force measured when using water jets to assist the cutting process is that the jets 'lubricate' the rockbit interface. From the results presented in Figure 30 it can be seen that this is not the case. Indeed the ratio of the bit cutting force to the bit normal (penetrating) force, sometimes termed the coefficient of cutting friction, actually is seen to increase in the work of Hood [30], Ropchan et al. [31] and Dubugnon [28]. In other words, the opposite of lubrication, increased friction, is observed. An explanation for this behavior suggested by Geier et al [32] is that the cushion of crushed rock on which the bit rides when water jets are not used acts as a lubricant for the bit. When this is removed by the jets the friction coefficient between the bit and the rock increases. The only commercial application of this technology to date has come about as a result of an extended development project undertaken collaboratively by the British Coal Corporation (formerly the National Coal Board) and the U.S. Bureau of Mines. The goal of this work was to learn whether the addition of a water jet assisted cutting system to a roadheading machine could extend the cutting capability of this machine from weak and medium strength rocks to strong rocks. Laboratory experiments conducted by British Coal showed that bit cutting force was reduced by about 30% and the bit penetrating force was reduced by about 50% when 5 mm deep cuts were made by a pick in a strong sandstone. The pick velocity during these tests was 1.14 m s~ * [33]. The initialfieldtrials with a 70 MPa water jet system mounted on a roadheading machine demonstrated that considerable benefits could be derived from the use of these jets. The rock cut at the trial site was, mostly, a limestone with a uniaxial compressive strength of 117 MPa and a uniaxial tensile strength of 21 MPa. An interesting result from these trials is illustrated in Figure 31. One of the traces in this figure is the pressure in the slew ram used to sweep the cutter head of the machine through the rock. In Figure 31a, which is the record of a cut in which water jets were not used, this trace is seen to increase steeply and then to fall suddenly. This indicates that the cutter head encountered higher and higher resistance as it progressed on its path through the rock. Eventually this resistance exceeded the force capacity of the slew ram and the head bounced back out of the cut, unable to complete its path. Figure 31b shows the equivalent trace in an adjacent cut made under similar conditions but this time the 70 MPa water jet system was employed. The reduced levels of the pick forces in this case enabled the cutter head to complete its sweep through the rock and consequently no increase in the slew ram pressure is observed.
The Use of Water Jets for Rock Excavation (a)
251
(b)
Figure 31 Measurements of parameters on a roadheading machine (a) without and (b) with water jets used to assist the cutting process (from Anon [33])
A later field trial using a different roadheading machine is described by Morris and Harrison [34]. This work was conducted in a British coal mine and the rocks cut were a variety of coals, fireclay, shales , siltstones and sandstones. One of the sandstones in this drivage had a uniaxial compressive strength of 170 M Pa. As with the previous field trial considerable improvements in cutting performance were reported when water jets were used to assist the cutting process. These benefits included a 50% improvement in cutting rate and a 30% reduction in the mechanical specific energy (that is, not accounting for the energy input of the water jets). Following these field experiences, two British manufacturers of roadheading machines, Dosco and Anderson Strathclyde, developed commercial versions of their machines with 70 M Pa water jets systems.
9.6.2.3
Tool force reductions - the importance ofjet position
One of the findings to emerge from the research work on this cutting method is the overriding importance of jet position. It is postulated that the main role of the water jets in assisting the cutting process is the removal of crushed rock adjacent to the cutting tool. Because this crushed material is concentrated in a small region, if the jets fail to strike that region, then the benefits to the cutting process are greatly reduced. Claims have been made by some workers that the optimal jet position is in front of a drag bit (Hood [30]; Dubugnon [28]; Hood [35]) while other investigators report that the best results are obtained when a jet is directed from behind the bit (Ropchan et al [31]). Whichever position is employed it is crucial that the jet(s) impinge at the tool-rock interface. If this target position is missed by even a millimeter or so then most of the benefits of jet assisted cutting are lost.
9.6.2.4
Tool force reductions - the effect of bit velocity
The effect of bit velocity on the bit force reductions caused by water jets remains a controversial issue. Some workers (Fairhurst and Deliac [36]; Fowell et al. [37]; Nienhaus et al. [38]) have conducted experiments using drag bits at tool velocities greater than 1ms" 1 . These workers conclude that at these high cutting speeds the benefit of reduced bit forces decreases to the point where it becomes negligible. However, a review of the experimental technique used by these various research groups reveals that their procedures called for the measurement of bit forces with increasing tool velocity while holding all of the other test parameters, including the water jet parameters of pressure and flow rate, constant. Obviously the potential problem with this experimental approach is that the energy deposited on the rock surface per unit cut length by the water jet decreases as the bit velocity increases. Under these circumstances it is not too surprising that the effectiveness of the water jets in reducing the bit forces also decreases because it might be anticipated that the jet energy is reduced to the point where it becomes incapable of eroding the crushed rock material from the region adjacent to the bit.
252
Mechanized Excavation
Perhaps the most detailed investigation of pick force reductions at various pick velocities was conducted by Morris and MacAndrew [39]. They used a chisel pick with the water jet arrangement shown in Figure 32. Their experiments were carried out cutting in two sandstones. The uniaxial compressive strengths of these rocks were 75 MPa and 150 MPa and their tensile strengths were 13.5 MPa and 23 MPa respectively. Morris and MacAndrew measured several parameters at three cutting speeds: 0.5,0.9 and 1.34 m s~ *. Unlike previous workers Morris and MacAndrew varied the jet power as they changed the cutting speed. This power was changed by varying jet pressure while holding the jet flow rate constant at 41 min"1. Five levels of jet pressure were examined: 0.2,10, 30, 60 and 138 MPa. These correspond to jet power levels of: 0.01, 0.67, 2, 4 and 9.2 kW, respectively. The depth of cut taken was 20 mm. These authors reported that, at least in the harder rock, almost no reductions in the pick forces were observed even at the highest jet pressure (Figure 33). The jet energy per unit length of cut for the experiment shown in this figure was 6.9 Jmm - 1 . Despite the potentially serious flaw in the methodology used by some of these research groups, their main conclusion that substantial force reductions are not achieved at high bit velocities seemingly was supported by the results of a field trial funded by the U.S. Bureau of Mines. These trials were carried out cutting in coal on a longwall face using an Eickhoff shearer equipped with a 70 MPa water jet system (Thimons et al [40]). This pick velocity for these tests was about 2.1 ms" 1 . The influence of using water jets was reported as negligible both on the cutting motor load and on the haulage force of the machine along the face.
Tool holder
/ /
Wimet hwpick
AJ
\ J e t block Single angle nozzle
2mm _
Figure 32 Pick and water jet arrangement used by Morris and MacAndrew [39]
1.0
1.5
2.0
2.5
Distance cut(m)
Figure 33 Bit cutting (light curve) and bit penetrating (heavy curve) forces with and without water jet assistance. The jet pressure used in these tests was 134 MPa. The cutting velocity was 1.34 m s - 1 (after Morris and MacAndrew [39])
The Use of Water Jets for Rock Excavation
253
The argument that bit velocity per se should not affect the reductions in bit forces was advanced by Hood et al [41]. Hood, Geier and Xu argued that provided that the jet power was increased proportionally with the tool velocity, so that the jet energy per unit of cut length remained the same, then the force reductions should be preserved. They conducted experiments cutting in Indiana limestone with a chisel pick and a single water jet directed 1 mm ahead of, and parallel to, the front face of the pick. They measured the reduction in the pick forces, with respect to cuts made without any water jets, usingfivedifferent jet pressures (maximum pressure was 70 MPa), three different jet nozzle diameters and two cutting speeds. Unfortunately the limitations of their apparatus limited the maximum cutting speed to 420 mms" 1 and therefore their findings, while interesting, are not conclusive. Hood, Geier and Xu found that, over the range of the parameters investigated, the reductions in bit forces were proportional to the jet energy per unit length of cut and were independent of the cutting speed. Over a range of jet energy values of nearly two orders of magnitude, from 1 -70 Jmm - 1 , the reductions in the average pick cutting force were observed to increase slowly from about 20 to 40%. On the other hand, the pick normal, or penetrating, force was observed to increase roughly linearly from only about 10% at low jet energy values to about 70% at high jet energy values (Figure 34). The importance of this result, if it can be verified at higher cutting velocities, is that it appears not to matter whether the jet energy is supplied as mostly pressure or mostlyflowrate. Provided that the energy content of the jet is adequate, the bit forces will be reduced. Support for Hood, Geier and Xu's thesis [41] that bit force reductions should continue to be realized even at high bit velocities provided that the jet power is increased proportionally with the bit velocity, came from another field trial using a water jet-assisted shearer. This trial was conducted by British Coal [42]. It employed water jets to assist only those 15 picks mounted on the face (or clearance) ring of the machine. Conventional water flushing was used for the 14 vane picks. Chisel picks were used for all of these cutting tools and the water jets were directed ahead of the front faces of the clearance picks. The high pressure pump fitted to the shearer was capable of delivering 55 1 min - 1 at a pressure of 69 MPa. The pick velocity was approximately 3 m s - 1 . Thus the total installed hydraulic power was 63.25 kW, or 4.2 kW pick -1 , and the jet energy per unit length of cut was 1.4 J mm" 1 . 100
(a) Nozzle diameter = 0.65mm
80
Nozzle diameter = 0.65mm
60 40 20 O
(b )
100 r-
Nozzle diameter =0.83mm
80 U 60 U
Nozzle diameter = 0.83mm
40 k 20 U θ'
100
(c ) Nozzle diameter= 1.05mm
80
601
Nozzle diameter = 1.05mm
40 20 0
J_
J
100
I
J
10 1
Jet energy per unit length (J mm" ) 16cm s~'
HH
4 2 c m s -1
H
'6cm e"'
□
42cm s"1
Figure 34 The reduction of the bit forces (given as a percentage of the forces measured when cutting dry) as a function of the jet energy per unit length of cut. The shaded areas are the 90% confidence bands for these results (after Hood et al. [41])
Mechanized Excavation
254
The maximum depth of cut on a shearer is controlled by the haulage speed. Three different haulage speeds were investigated: 47, 55 and 76 mms" 1 . These speeds correspond to cut depths of: 5.0, 5.9 and 8.1 mm pick -1 , respectively. It was observed that the water jets reduced substantially the power consumed by the drum. This is equivalent to stating that the pick cutting force was reduced. A comment is made in the report that this finding could be significant because it could be used as a method to extend the life of the gearhead. One of the key parameters measured during these experiments was the machine haulage force. This is equivalent to the pick thrust, or penetrating, force. Although considerable scatter was observed in the measurements of this haulage force, it was found that the average value of this force decreased markedly with increasing jet pressure. Perhaps the most convincing piece of evidence regarding the effectiveness of the water jets at this high pick velocity is the observation in the report by British Coal that when water jets were not used to assist the cutting process the shearer stalled continually at the highest haulage speed of 76 mms" 1 . However, using the water jet system, the machine cut smoothly along the face with no stalls. Thus, this report concludes 'for all practical purposes, the maximum cutting speed of the machine was substantially increased when high pressure water aided'. 9.6.2.5
Reduction of dust, frictional sparking and machine vibration
The results of all of the field trials reported in the literature are unanimous in proclaiming that dramatic reductions in dust production are realized when water jets are directed immediately adjacent to the cutting tools. The use of a high pressure (41-70 MPa) water jet to assist the cutting performance of a roadheader was reported to produce considerable (but unqualified) reductions in dust produced, compared to conventional dust suppression sprays [33]. It is noted also in this report that the incidence of frictional sparking was very high when cutting operations were carried out in sandstone using only conventional dust sprays. This problem was eliminated apparently completely by using the high pressure water jet-assisted system. Measurements made during later trials with roadheaders showed that dust levels were maintained consistently below 2 m g m ~ 3 provided that the jet-assisted cutting system was employed (Morris and Harrison [34]). Morris and Harrison note that substantial reductions in dust were observed at jet pressures well below the maximum 70 MPa available. Again, corroborating previous findings, frictional sparking was not observed during these trials when the water jet-assisted cutting system was used. A marked reduction in machine vibration was observed during these trials. Currently some 16 roadheading machines with high pressure water jet systems are operating in the Cape Breton coal mines in Canada (Haslett, private communication, 1989). The rocks cut by these machines are relatively soft, principally coal. Perhaps because the strata are not hard the principal benefit of using water jets on these machines has been found to be dust suppression. A study to examine the effectiveness of the water jets on these machines found that the dust levels decreased rapidly as the jet pressure increased, up to a jet pressure of 30 MPa. Beyond this value of jet pressure the dust level started to increase (Haslett et al [43]). In another coal cutting operation, a shearer trial with an Eickhoff machine, again the principal benefit of using water jets was found to be a substantial reduction in the dust generated during the cutting operation (Thimons et al [40]). Confirming the finding of previous workers, Thimons, Hauer and Neinhaus reported that substantial reductions in dust quantities were observed even at low water jet pressures (4 MPa). Increasing the jet pressures to 50 MPa did not produce any further substantial reductions in the measured dust loads. Similar findings were reported for the British shearer field trial (British Coal [42]) although in these tests some slight improvement in measurable dust concentration with increasing jet pressure was observed. The question of the influence of the water jets on the vibration of the shearer was addressed, but was not answered satisfactorily, during these trials. Apparently the machine operators were of the opinion that the water jets caused the shearer to function more smoothly. However, when accelerometers were fitted to the machine it was found that somewhat higher acceleration levels were measured when the water jets were in use. British Coal concluded that, at least in this situation, no reduction in machine vibration took place. 9.6.2.6
Reduction in bit wear
The rate of bit wear is related directly to excessive bit temperatures. Hood [29] measured the temperatures generated in the bit body during the cutting operation both using and not using water
The Use of Water Jets for Rock Excavation
255
jets. He showed that the bit temperatures were reduced substantially (by several hundred degrees Celsius) when moderate pressure water jets were used to assist the cutting operation. This result implies that the jet-assisted cutting method should result in greatly reduced bit wear rates. This expectation was realized in the underground tests where improvements in bit life, often by more than an order of magnitude, were achieved. Measurements of bit temperatures also were made by Nienhaus et al. [38]. Their results (Figure 35) show that bit temperatures increase with cut length. The two upper curves in Figure 35 show that much higher bit temperatures were recorded when the cutting velocity was doubled from 1 m s~ 1 to 2 ms" 1 . It can be seen from this figure that, at both of these bit velocities, the temperatures were reduced by a factor of about three when water jets, at a jet pressure of 45 MPa, were used to assist the cutting process. Perhaps of greatest interest is that the gradients of the lower curves, representing the bit temperatures using the water jets, are less throughout this domain than those of the upper curves representing dry cuts. This implies that the rate of bit wear was significantly less when the jets were employed. Furthermore, these lower curves appear to level out indicating that bit wear may not increase beyond the amount caused by cutting a distance of 1.5-2.1 km in this artificial rock. In other words, when using water jets the bit wears from its new condition to some point, but it does not wear, at least not substantially, beyond that point. This finding that the rate of bit wear levels out was confirmed by Morris and MacAndrew [39]. Morris and MacAndrew showed that even in situations when a jet, properly mounted with respect to the bit (see Figure 32), was inadequate to effect significant reductions to the bit forces, major longterm benefits were realized. Figure 36a shows the measured increases in the bit cutting and in the bit normal forces as a function of the distance cut. The steep slope of these curves obviously is caused by the rapid blunting of the bit when water jets are not used to assist the cutting process, Eventually these loads become so great that the tungsten carbide insert shatters. Figure 36b is a similar plot only in this case a relatively low pressure (10 MPa) water jet was used to assist the breakage operation. It is apparent that the rate of blunting of the bit, as indicated by the slopes of these curves, is much less than for the previous case. Morris and MacAndrew concluded that one of the most important effects of water jets that are directed adjacent to drag bits to assist the cutting operation is to provide effective cooling of the bits which, in turn, preserves the tool sharpness. This prediction that water jet-assisted cutting should improve bit life has been confirmed many times in the field. The initial roadheader field trial report (Anon. [33]) commented that significant improvements in pick life were achieved by using the water jet system. Morris and Harrison [34] reported that 'pick consumption (when using water jets to assist the cutting process) was very low, less than one pick per metre of advance'. They conducted a dramatic test to show how effective the
2ms"1 350
300
o
250
0 ω 200 +3σ α>
Q. E ω
2ms" 1 450bar
150
100
50
0
300
600
900
1200 1500 1800
2100
2400
Length of cut (m) Cutting speed Depth of c u t Compressive strength Flow rate V Flat bit
l m s"', 2 m s"1 I5mm 4 0 N mm"2 8Lmin"' HB 3 0 0 - 5 0 7 9
Figure 35 Measured temperatures in the pick body with and without water jets assisting the cutting process (after Nienhaus et al. [38])
256
Mechanized Excavation (a)
(b)
• Normal force O Cutting force
j _
I .0
0
0.2
Distance cut (km)
Figure 36 Pick cutting and penetrating (normal) forces as a function of the distance cut (a) without a water jet assisting the cutting process and (b) with a 10 MPa water jet assisting the cutting process (after Morris and MacAndrew [39])
RI.5
AT
I
Jet position
Disc cutter
Figure 37 The position of the jets used to assist a disc cutter (after Fenn et al [44])
The Use of Water Jets for Rock Excavation
257
water jets were in extending bit life. For this test they blocked the nozzle orifice to two of the 24 picks on the cutter head. The machine then was used to advance the tunnel by one meter. During this cutting period it was necessary to change one of the picks without a water jet 23 times and the other pick without a jet had to be changed twice. Only one of the remaining 22 picks, which were assisted by water jets, needed to be changed. 9.6.2.7 Disc cutter force reductions Fenn et al [44] conducted an experimental program in the laboratory to determine whether water jets could be reduce disc cutter forces. Their work was carried out cutting norite, a strong (254 MPa uniaxial compressive strength; 13.9 MPa tensile strength) but relatively nonabrasive rock. The work demonstrated that, with appropriately arranged jets, significant force reductions were observed. Four jets, two directed from each side of the cutter (Figure 37), were found to yield the best results. Jet pressures up to 40 MPa were employed. The diameter of the nozzles used was 1.2 mm. Typical results are shown in Figure 38. It can be seen from Figures 38a and 38b that substantial reductions in both thrust and rolling forces were measured when the water jets were employed. 9.7 CONCLUSIONS 9.7.1 Continuous, Discontinuous, Cavitating and Abrasive Water Jets Continuous high pressure water jets, by themselves, as a method of rock erosion are extremely energy intensive. Consequently this method is impractical unless the total quantity of rock to be broken is very small. Hence this method can find application for drilling small diameter holes. It might be used also for cutting deep (of the order of meters or tens of meters) kerfs in a quarry that produces dimension stone. These kerfs replace line drilling and serve to define blocks that later can be lifted using small explosive charges. The biggest problem that must be faced when using continuous water jets by themselves is the sensitivity of the jets, in terms of the efficiency of the erosion process, to changing rock types. In general, an excavation will penetrate many different rocks. Although the water jets may erode satisfactorily the majority of the rocks encountered during the excavation process, they may have difficulty, or be incapable of, eroding the remaining rocks. This behavior can be disastrous to an excavation system which, in general, must be capable of breaking any and all rock likely to be met. Discontinuous jets have much higher rock breakage efficiency than continuous jets. However, although work on these systems has been conducted over a considerable period of time no systems have been developed that have come close to commercial exploitation. This may reflect the considerable engineering difficulties that need to be overcome with these types of jet. Cavitating jets also have received considerable research attention. The erosive abilities of these jets are much greater than those of continuous jets. However, as with discontinuous jets, no cavitating water jet systems have found commercial application for rock breakage. JC
160
(α
Φ
o
«♦(0 3
*_ ·*Φ σ Φ
> σ Φ
>
σ
140
•/^
120 __
80
= 10 p
-
60
s • 'ο
40 20
υ
0
(b)
tt) I4l·-
100
3
Ε 3
Z 16
)
-
• Unassisted 0 Jet assisted 1 1 1 1
o
1 1 1 15 30 45 60 75 90 100
3
E O 0
• Unassisted o Jet assisted J15 30 I 45 I 60 I 75 I 90 L J 100
Cut spacing (mm)
Figure 38 (a) Thrust and (b) rolling forces measured on a disc cutter with and without water jets assisting the cutting process (after Fenn et al [44])
258
Mechanized Excavation
Abrasive water jets are the newest of these technologies. The use of these jets in rock excavation operations remains relatively unexplored. Because these jets are capable of cutting deep kerfs, even in very strong and erosive resistant rock, it seems likely that this technology will find application in rock excavation systems as a method of weakening a rock face. 9.7.2 Mechanically Assisted Cutting This approach to rock excavation is effective in the sense that there is no doubt that the force needed at the tools to break the rock is greatly reduced when kerfs are cut in the rock face by water jets before that face is attacked by mechanical tools. However, for this method to be effective the jet pressure must always exceed the threshold pressure of the rock. When a drill is developed specifically to work in some weak rock type, such as coal or uranium ore, relatively low jet pressures, of the order of 200 MPa, might satisfy this criterion.However, most drills need to be capable of operating in a wide variety of rocks. In practice it has been found that even in sediments, such as coal mine roof rock, jet pressures of the order of 350-400 MPa are needed for a drill to operate effectively. These high pressures pose two potential problems. First, 15 years ago whenfieldexperiments with this cutting method were initiated, pump technology was inadequate to deliver jets reliably at these high pressures. This is not a problem today. Pumping systems are available from several manufacturers that will operate without the need for major overhaul in an underground environment at pressures of the order of 380 MPa for hundreds of hours. The other potential problem is one of power consumption. In small holes the power consumed by the water jets is much greater than the mechanical power consumed by the bit, often by a factor of 9 to 1. The total power (water jet power plus mechanical power) consumed by this drilling system might be twice that of a conventional percussive or rotary drill. This is still not an excessive quantity of power, perhaps of the order of 50-60 kW, and the advantages offered by a mechanically assisted water jet drill (faster drilling rates, a lighter drilling machine, less noise and less dust, etc.) might make this type of drill attractive in spite of this higher power requirement. However, a difficulty arises as the hole size increases because the water jet power needed for this cutting method increases as the square of the hole diameter. Thus, with medium sized holes the power needed for the water jets to cut kerfs effectively in the rock face is of the order of several hundred kilowatts. Again, in many circumstances the benefits to be gained from this drilling method may justify the use of this level of power usage. However, it is difficult to see how the power levels of several megawatts needed for effective cutting of very large diameter holes, such as tunnels, can be economically viable. 9.7.3 Jet-assisted Cutting Substantial reductions in tool forces can be achieved when high pressure water jets are directed in the immediate vicinity of a mechanical rock cutting tool. In addition to reducing these forces these jets also are effective in cooling the tool during the cutting operation. These two factors of lower forces and cooler tools result in substantial improvements in tool life. Other advantages claimed for this cutting method include: significant reductions in the dust and in the incidences of frictional sparking and less machine vibration. Some investigators question whether some of these benefits, particularly the reductions in tool forces, can be achieved under the practical constraints of high tool velocities and deep cuts that are experienced with mining and tunneling machines. Other workers argue that if sufficient attention is paid to directing the jets into the region of crushed rock adjacent to the tool and if the jet energy is sufficient to erode this crushed material then the tool force reductions should be preserved, even at high tool velocities. More work is needed to resolve these questions. On the other hand, Morris and MacAndrew [39] have shown that, even if the cutting conditions are arranged so that tool forces are not reduced significantly by the jets when the tools are sharp, the cooling effect of the jets substantially retards the blunting process and therefore lower tool forces are experienced because the cutting operation is carried out with sharp tools. Other benefits, such as dust and spark suppression are widely accepted and it is found that these benefits can be achieved at low (of the order of 20-30 MPa) jet pressures. This cutting method has the potential for achieving a major advance in rock breaking. This potential is only now starting to be realized. The method reduces the mechanical tool loads, this is true over time and it may or may not be the case when the tools are new, and it reduces the thermal
The Use of Water Jets for Rock Excavation
259
tool loads. These benefits give excavation machines the capability of cutting in stronger rock than can be machined when water jets are not employed. Alternatively, the lower tool forces give machines the capability for higher rates of excavation than are possible without jets. Finally, because the water jet system is compact and because the use of this system results in lower forces at the cutter head, the equipment size needed for a given rate of rock excavation might be reduced by using a water jet-assisted cutting machine. In underground operations small, maneuverable machines with a capability of cutting hard rock could find widespread application. In other applications, for example coal mining, the benefits of dust and spark suppression might be adequate to justify the installation of a water jet system to the excavation equipment.
9.8
REFERENCES
1. Harris H. D. and Mellor M. Cutting rock with water jets. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 11, 343-358 (1974). 2. Hood M., Nordland R. and Thimons E. A study of rock erosion using high pressure water jets. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 27, 77-86 (1990). 3. Maurer W. C. Advanced Drilling Techniques, Vol. 1. The Petroleum Publishing Company, Tulsa, OK (1980). 4. Summers D. A. and Bushnell D. J. Preliminary experimentation of the design of a water jet drilling device. In Proc. 3rd Int. Symp. Jet Cutting Technology, Chicago, IL. pp. E2-21-E2-28. BHRA, Cranfield (1976). 5. Maurer W. C. and Heilhecker J. K. Hydraulic jet drilling. In Proc. 4th Conf. Drilling and Rock Mech., University of Texas, Austin, pp. 213-214 (1969). 6. Chadwick R. F. Continuous high velocity jet excavation - Phase 1. Bendix Research Laboratories, Southfield, Michigan, Final Report to the Bureau of Mines (1972). 7. Brunton J. H. The physics of impact and deformation: single impact. I. High speed liquid impact. Phil. Trans. R. Soc. London A 260, 79-85 (1966). 8. Nebecker E. B. and Rodriquez S. E. Percussive water jets for rapid excavation. Scientific Associates Inc., Santa Monica, CA. Final Report NTIS-772 931 (1973). 9. Edney B. E. Experimental studies of pulsed water jets. In Proc. 3rd Int. Symp. Jet Cutting Technology, Chicago, IL, pp. B2-11-B2-26. BHRA, Cranfield (1976). 10. Young C. Rock breakage with pulsed water jets. AS ME Energy Technology Conference and Exhibition, Houston, TX. ASME, New York (1977). 11. Conn A. F. Rapid cutting of pavement with cavitating water jets. In Proc. 8th Int. Symp. Jet Cutting Technology, Durham, UK, pp. 231-240. BHRA, Cranfield (1986). 12. Conn A. R. and Radtke R. P. CAVIJET augmented deep-hole drilling bits. Am. Soc. Mech. Eng. Pap. 77-PET-54 (1977). 13. Conn A. F. and Radtke R. P. Development of CAVIJET augmented deep-hole bits. Presented at DOE Geothermal Drilling and Contractor Review Meeting, Washington, D. C. (1978). 14. Angona F. A. Cavitation, a novel drilling concept. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 11, 115-119 (1974). 15. Hunt D. C , Kim T. J. and Sylvia J. G. A parametric study of abrasive waterjet processes by piercing experiment. In Proc 8th Int. Symp. Jet Cutting Technology, Durham, UK, pp. 287-296. BHRA, Cranfield (1986). 16. Fort J. A., Geier J. and Hood M. Deep-kerfing for selective mining in hard rock using abrasive water jets. In Proc. 2nd Int. Conf Innovative Mining Systems, Penn State University, PA (1986). 17. Marlowe A. C , Worsley S. L. and Price C. J. The use of abrasive entrained high pressure water jets as a tool for the nonexplosive winning of gold bearing quartzites. In Proc 8th Int. Symp. Jet Cutting Technology, Durham, UK, pp. 113-123. BHRA, Cranfield (1986). 18. Echert D. C , Hashish M. and Marvin M. Abrasive-waterjet and waterjet techniques for decontamination and decommissioning nuclear facilities. In Proc. 4th U.S. Water Jet Conf, Berkeley, CA, pp. 73-82. ASME, New York (1987). 19. Savanick G. A. and Krawza W. G. An abrasive water jet drill. In Proc. 4th U.S. Water Jet Conf, Berkeley, CA, pp. 129-132. ASME, New York (1987). 20. Feistkorn E. and Knickmeyer W. Tests on water jet assisted drilling of shot firing boreholes in abrasive rocks. In Proc. 7th Int. Symp Jet Cutting Technology, Ottawa, Canada. BHRA, Cranfield (1984). 21. Maurer W. C. and Heilhecker J. K. Hydraulic jet drilling. In Proc. 4th Conf Drilling and Rock Mech., University of Texas, Austin, pp. 213-214 (1973). 22. Hood M., Kolle J. and Reichman J. Water jet technology. In Proc 2nd Berkeley Symp. Topics in Petroleum Eng., Lawrence Berkeley Laboratory, CA, pp. 5-9 (1988). 23. Wang F.-D., Robbins R. and Olsen J. Feasibility study of hydraulic jet kerfing to improve the efficiency of mechanical disc cutting. Colorado School of Mines. Report for Dept. of Trans. DOT-TST-75-66. NTIS, Springfield, VA (1974). 24. Wang F.-D., Robbins R. and Olsen J. Water jet assisted tunnel boring. In Proc. 3rd Int. Symp Jet Cutting Technology, Chicago, IL, pp. X63. BHRA, Cranfield (1976). 25. Henneke J. and Baumann L. Jet assisted tunnel boring in coal measure strata. In Proc. 4th Int. Symp. Jet Cutting Technology, Canterbury, UK, pp. Jl-l-Jl-12. BHRA, Cranfield (1978). 26. Baumann L. and Henneke J. Attempt of technical-economical optimization of high-pressure jet assistance for tunneling machines. In Proc. 5th Int. Symp. Jet Cutting Technology, Hannover, FRG, pp. 119-140. BHRA, Cranfield (1980). 27. Hustrulid W. A technical and economic evaluation of water jet assisted tunnel boring. Report to the National Science Foundation, Final report (1976). 28. Dubugnon O. An experimental study of water assisted drag bit cutting of rocks. In Proc. 1st U.S. Water Jet Symp., Golden, CO, pp. II-4.l-II-4.il (1981). 29. Hood M., A study of methods to improve the performance of drag bits used to cut hard rock, Ph.D. thesis, University of the Witwatersrand, South Africa (1978).
260 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44.
Mechanized Excavation Hood M. Cutting strong rock with a drag bit assisted by high-pressure water jets. J. S. Afr. Inst. Min. Metall. 77, 79-90 (1976). Ropchan D., Wang F.-D. and Wolgamott J. Application of waterjet assisted drag bit and pick cutter for the cutting of coal measure rocks. Report to the U.S. Dept. of Energy. Final Technical Report ET-77-a-01-9082 (1980). Geier J. E., Hood M. and Thimons E. D. Waterjet-assisted drag bit cutting in medium strength rock: A fundamental investigation. In Proc. 28th U.S. Symp. Rock Mech., Tucson AZ (Edited by I. W. Farmer, J. J. K. Daeman, C. S. Desai, C. E. Glass and S. P. Neuman), pp. 953-961. Balkema, Rotterdam (1987). Anon. Field trials with a 10,000 psi prototype system. Seminar on Water Jet Assisted Roadheaders for Rock Excavation, Pittsburgh, PA (1982). Morris A. H. and Harrison W. Significant advance in cutting ability-roadheaders. In Proc. 7th Rapid Excavation and Tunneling Conf., New York, NY (Edited by C. D. Mann and M. N. Kelley), Vol. 1,2, pp. 317-340. Soc. Min. Eng. AIME, New York (1985). Hood M. Waterjet-assisted rock cutting systems - the present state of the art. Int. J. Min. Geol. Eng. 3, 91-113 (1985). Fairhurst C. E. and Deliac E. P. Water-jet assisted rock cutting - the effect of pick traverse speed. In Proc. 8th Int. Symp. Jet Cutting Technology, Durham, UK, pp. 43-55. BHRA, Cranfield (1986). Fowell R. J., Ip C. K. and Johnson S. T. Water jet assisted drag tool cutting: Parameters for success. In Proc. 8th Int. Symp. Jet Cutting Technology, Durham, UK, pp. 21-32. BHRA, Cranfield (1986). Nienhaus K., Weigelt H. and Thimons E. D. The development of a water-jet-assisted shearer loader. In Proc. 8th Int. Symp. Jet Cutting Technology, Durham, UK, pp. 79-92, BHRA, Cranfield (1986). Morris C. J. and MacAndrew K. M. A laboratory study of high pressure water jet assisted cutting. In Proc. 8th Int. Symp. Jet Cutting Technology, Durham, UK, pp. 1-8. BHRA, Cranfield (1986). Thimons E. D., Hauer K. F. and Neinhaus K. Water jet assisted longwall shearer: Development and underground test. In Proc. 4th U.S. Water Jet Conf., Berkeley, CA, pp. 113-120. ASME. New York (1987). Hood M., Geier J. E. and Xu J. The influence of water jets on the cutting behavior of drag bits. In Proc. 6th Int. Congress on Rock Mech., Montreal, Canada (Edited by G. Herget and S. Vangpaisal), pp. 649-654. Balkema, Rotterdam (1987). British Coal, HQTD. Water supply for shearer Venturi. Report to Commission of European Communities. Final Report on ECSC Research Project 7258-03/08/088 (1986). Haslett G. A., Corbett G. R. and Young D. A. An investigation into the effect of varying water pressure and flow rates upon the release of airborne respirable dust by a Dosco MkllB roadheader equipped with a water jet assisted cutting head. In Proc. 8th Int. Symp. Jet Cutting Technology, Durham, UK, pp. 103-112. BHRA, Cranfield (1986). Fenn O., Protheroe B. E. and Joughin N. C. Enhancement of roller cutting by means of water jets. In Proc. 7th Rapid Excavation and Tunneling Conf., New York (Edited by C. D. Mann and M. N. Kelley), Vol. 1, pp. 341-356. Soc. Min. Eng. AIME, New York (1985).
10 TBM Performance Analysis with Reference to Rock Properties PRISCILLA P. NELSON University of Texas, Austin, TX, USA
10.1
INTRODUCTION
261
10.2 TBM SYSTEM DESCRIPTION
262 262 266
JO.2.1 Component Systems and Operation 10.2.2 TBM System Performance Parameters 10.3
ROCK PROPERTIES
10.3.1 10.3.2 10.3.3
267 267 269 TJ\
Fragmentation by Cutting Tools Laboratory Rock Tests Rock Mass Properties
10.4 ROCK PROPERTY IMPACT ON TBM PENETRATION RATE 10.4.1 General Observations 10.4.2 Correlations with Intact Rock Characteristics 10.4.3 Rock Mass Characteristics Impact
273 273 275 279
10.5
280
ROCK PROPERTY IMPACT ON CUTTING TOOLS
10.5.1 10.5.2 10.6
280 281
Cutting Tool Failure Rock Property Impact on Failure Rates
ROCK PROPERTY IMPACT ON UTILIZATION
10.7
THE FUTURE
10.7.1 10.7.2 10.7.3 10.8
283 283 283 284 286
10.6.1 Management and Downtime 10.6.2 Impact of Cutting Tools 10.6.3 Impact of Geotechnical Factors 10.6.4 Other Impacts
287 287 288 289
TBM System Performance Prediction The Future for Equipment Developments Summary Comments
REFERENCES
289
10.1 INTRODUCTION In response to a question at the Tunnelling 76 conference, Innaurato et al [1] replied: 'Undoubtedly rock mass characteristics ... influence tunnel boring machine performance probably more than the laboratory properties of the rock matrix. Nevertheless, attempts to characterize the in situ properties of rocks have been directed principally towards the choice of tunnel supports rather than to the investigation of tunnelling machine performance. Further research into tunnelling machines will undoubtedly be directed towards the definition of a suitable scheme of rock mass properties in connexion with machine performance' ([1], page 226). Now, over 15 years later, such a 'suitable scheme' has yet to be identified and adopted in geotechnical engineering. There are many reasons for this, including: (i) limited opportunities for 261
262
Mechanized Excavation
geotechnical engineers to continue from design through construction; (ii) machine manufacturers, contractors and specialized consultants building their own proprietary data bases; (iii) hesitancy to release information because of the perceived potential for claims; and (iv) the very competitive industry rarely contributes funds to support data gathering by researchers. Whatever the reasons, it is clear that neither geology alone, laboratory and field testing alone, experience alone nor equipment design and operation expertise alone can get an engineer to the point where underground excavation is a clearly defined engineered process. Integration of all these knowledge bases is required to raise the level of engineering contribution to underground construction, and the entire excavation system must be understood before applying engineering expertise to the solution of expected or developing problems. A mechanized excavation system includes many component systems, broadly grouped into the tunnel-boring machine (TBM) and the back-up equipment. The TBM includes the cutterhead, with cutting tools and muck (broken rock) collection components; systems to supply power, cutterhead rotation and thrust; a bracing system for the TBM during mining; shielding to protect workers; and a steering system. Back-up equipment systems provide tunnel support, muck transport, personnel conveyance, ventilation, material supply and utilities. In this chapter, these systems will be briefly described so that the impact of geotechnical conditions on excavation system operation and performance can be assessed. Only full face hard rock rotary excavation tunnel equipment, operated by workers in the tunnel, will be discussed here. It is not the intention of this chapter to produce an exhaustive literature review of past efforts at performance prediction. Several such reviews of various aspects of performance evaluation have appeared in the recent literature. Neither is it the intention to exhaustively describe all rock index and property tests which have evolved to support the aim of performance prediction. In any event, many of the specialized tests require particular equipment not generally available except at a few laboratories. This chapter will concentrate on developing an understanding of TBM design and rock material characteristics which have an impact on the relative success or failure of a particular mechanical excavation system in a particular application. The aim here is to support a geotechnical engineer charged with site investigation, contract document preparation and resident engineer duties, so that he or she may be able to identify particular geological or geotechnical characteristics of concern for various alignment options, so that the potential impact of these characteristics on construction can be anticipated, and so that the impact of various design options for TBM systems can be appreciated. 10.2 TBM SYSTEM DESCRIPTION 10.2.1 Component Systems and Operation Figure 1 shows a typical open TBM designed for operation in hard rock. A TBM cuts rock with wheel-like disc tools attached to a full face circular cutterhead structure. The cutterhead is rotated and thrust into the rock surface at the heading of a tunnel, causing the cutting tools to penetrate and break the rock at the tunnel face. Thrust and torque reaction forces are transferred through a structural frame. For a typical hard rock unshielded (open) machine, the structural system is internal. For a shielded TBM, the external shield may serve as a major portion of the structural system. The reaction to the applied forces may be supplied by anchoring with braces (grippers) extended to the tunnel wall, bracing against support installed at the rear of the TBM, and friction at the shield/rock interface. Each component of an unshielded mechanized excavation system with interactive impact on geotechnical conditions is briefly described below. The cutterhead serves to support the ground at the tunnel face, to support the individual cutting tools, to transmit rotation and thrust to the tools, to gather muck and direct it into the transportation system, and to supply some dust suppression with water spray. The current machine design trend is to apply increasing cutter loads and rotate the cutterhead at faster rates, resulting in increased concern for cutterhead stiffness, vibrations and fatigue in the design of new and rehabilitation of old TBMs. In poor ground, blocks of rock may be loosened at the face and become wedged between cutters causing cutterhead and cutter damage. In anticipation of such a condition, a TBM may be fitted with recessed cutters, or a steel structural 'false face' may be built up. At the periphery of the unidirectionally rotating cutterhead there are openings through which muck can enter, called muck buckets. In poor ground, bars (grills) are often added across the openings to stop fall-out blocks from entering. Muck is dumped onto a conveyor which carries the muck to the rear of the TBM, where it is transferred to the back up transport system.
TBM Performance Analysis with Reference to Rock Properties
263
Drive train and bull gear motors installed through rear of cutterhead structure Figure 1
Unshielded TBM
As the cutterhead rotates, each cutter indents and traces a circle on the rock face. The array, or spatial arrangement, of cutters on the cutterhead is designed to facilitate rock chipping (kerf cutting) between concentric indentations. Discs cutting adjacent grooves are separated on the cutterhead and neighboring groove locations are not loaded at the same time. Peripheral cutters are called gauge cutters, cutters at the center of the cutterhead are called center cutters and those at other positions are called face cutters. Gauge cutters are usually positioned at an angle relative to the direction of thrust, have a high rolling velocity and must track through any muck accumulation at the tunnel invert. Center cutters roll in a tight radius, and are subjected to scuffing. Average spacing between adjacent cutter grooves is generally about 65-80 mm. Cutting tools are almost exclusively single disc rolling cutters, with replaceable disc rings of special hardened steels, selected to achieve a desired balance between hardness, toughness and abrasion resistance. In softer rock, where cutter loading is lower and less variable, disc material with reduced fatigue life but with increased material hardness may be chosen. In stronger rock, high thrust and impact loads are expected, and so fatigue, toughness and abrasion properties must be balanced. In extremely abrasive rock, tungsten carbide discs may be used if the wear rate of steel cutter rings becomes too high. Standard cutter diameters are 394 and 432 mm, but smaller diameters are available and 483 and 500 mm diameter cutters are under development. Cutter design has evolved over the past 35 years, and the earlier wedge-shaped section disc has given way to a 'constant section' disc which is more uniform in edge width when new but wears to a stable profile which is maintained over long periods of use. Typical constant section cutter tip widths are 12-19 mm, and recommended allowable cutter loading has increased from 220 kN up to about 270 kN. Higher loads and heavier cutters have also prompted redesign of the cutter holders, and improvements have been made to minimize the time required to replace worn cutters. In poor ground where safety is a major concern, rear-access cutters may be used which can be replaced without exposing workers to the unsupported face. The torque system comprises motors and a drive train including the bull gear and main bearing assembly. Motors are either hydraulic or electric. Hydraulic systems operate at about 65-70% efficiency with the possibility of having a full range of variable cutterhead rotation rates (rpm). Some TBMs include two-speed AC electric motors rated at about 112-190 kW power. Installed torque has often been limited by space and heat conditions at the heading. Newer designs incorporate watercooled AC drive motors rated at up to 335 kW, with variable frequency or stepped drive, and which operate at 85-90% efficiency. The high power motors increase the available torque without causing space constraints. The variable cutterhead rotation promises to help in ground control, since a very slow rpm which minimizes disturbance can be used in blocky ground with less disruption of the face while, in good ground, a high rpm will promote higher advance rates. TBM thrust systems are exclusively hydraulic with rated pressures about 35-50 MPa. These systems are very reliable. Main thrust cylinders have about a 1.2-1.6 m stroke, with smaller stroke increments used to negotiate curves or for steering corrections. Anchoring to provide thrust and
264
Mechanized Excavation
torque reaction is usually provided by pair(s) of diametrically opposed hydraulic grippers which are extended to engage the tunnel wall. For designs with one pair of grippers, the third balance point is a support at the base of the cutterhead structure. Such a system is liable to nose dive if soft rock is present in the tunnel invert. This is less of a problem with two opposed gripper pairs or with 'cruciform' bracing systems. TBM steering systems are also hydraulically actuated, and different designs permit steering correction while mining or only when mining has stopped at the end of the thrust cylinder stroke. The efficiency of TBM operation depends on the relative stiffness of the rock mass and the TBM cutting system, since relative stiffness affects cutting efficiency and peak load development on the cutters. Increased TBM stiffness (or reduced rock mass stiffness) leads to less demand on torque and a decrease in average cutter forces. TBM stiffness is dominated by the hydraulic system and stability of the cutterhead during mining, and machine stiffness is typically of the order of 200-400 kN mm - 1 , but varies inversely with thrust cylinder piston extension [2, 3]. In laboratory indentation testing of intact rock, typical intact rock stiffness may be about the same order of magnitude, perhaps 50-200 kN mm" 1 . In the field, TBM relative stiffness can be enhanced by discontinuities which greatly reduce rock mass stiffness. In massive rock, however, the TBM and rock mass may be of the same order of stiffness, a condition which can pose a limitation to efficient cutting. TBMs operated in a very stiff rock with high cutter loads are susceptible to fatigue problems with the cutters, cutterhead structure and bearings. In such cases, it is very important to provide bracing and support to improve cutterhead stability. Open TBMs may incorporate some capability for shielding and primary support installation. Drills for rock bolting are fixed to platforms at locations which are stationary during mining, and rock bolts can be installed through slots in roof shields, permitting bolt installation immediately behind cutterhead, perhaps 5-10 m behind the heading. Contractors have been hesitant to apply shotcrete near the TBM because of the clean up required. Shotcrete has been applied in TBM tunnels, but usually to the rear of the trailing floor perhaps 100-150 m behind the heading. For more difficult ground conditions, it is possible to install either steel sets (and lagging) or precast segmental lining at a location between the TBM and trailing floor, about 15-25 m behind the heading. Ring steel or segments can be installed in front of gripper locations, but the work is slow and the rings (and lagging) cannot be fully completed until grippers are past the installation point. For soft invert conditions, the contractor may install a single precast invert segment which can help to improve track stability and minimize mucking delays. If difficult ground conditions are expected, a shielded TBM may be selected. The engineer should be very sober in assessing ground conditions, however, since the decision to use a shielded TBM has significant impact on other aspects of construction, cost and schedule. Shielded operations are usually slower, and face access and ground control is more difficult. Shielded TBMs are more expensive than open TBMs, and steering is difficult if there are tight curves in the project alignment. A shield should only be considered if difficult ground cannot be handled with an open TBM, if estimated stand up time is very short, or if the poor conditions are expected over long tunnel stretches or at many locations. Authorizing conditions might include fractured/blocky/ravelling/ loosening ground, for which face instability is severe and stand up time is short, heavy ground water inflows, running or flowing ground, and sometimes in squeezing conditions. If the rate of squeeze is slow, it may be that an open machine will perform better and not run the danger of squeeze closure or pressure binding as may happen with a full shield. Shields in use are described in Table 1, including single shields (articulated and nonarticulated) and newer telescoping designs [4]. Single shields generate thrust by bearing on lining installed in the rear of the shield, and mining must stop for lining erection. This generally slows mining progress, perhaps giving squeezing pressures time to develop on the unmoving shield. A telescoping shield makes no such commitment to support installation, and includes a main sidewall gripping thrust system, with the lining thrust system considered auxiliary. Most types of segmental linings can be used with shields. However, some projects have experienced difficulty when overcutting (to minimize squeeze) or overbreak results in an enlarged tunnel diameter and expanded segments need to be installed. Extensive delays can occur when grouting/backpacking is required to stabilize the segment rings before mining can continue. For gassy ground, TBMs can utilize oil invert emulsions for hydraulic fluids, explosion-proof electrical systems, and intrinsically safe instrumentation and lighting. Natural gas detection systems are usually incorporated (and may be required by regulatory agencies). The ventilation/dust control systems usually incorporate unbalanced pressure systems, dust shields, and water spray for dust suppression. However, in weak rock of certain lithologies, introducing water into muck may cause more problems than it solves. Clay shales may become very sticky in handling, and some chalk and
Table 1 Shielding Options for TBMs Shield type
Thrust
Steering
Curve/clearance or ground squeeze
Roll control
Support
Other
Open
Propel cylinders anchored to sidewall grippers
Either at regrip or while mining, depending on design
Adjustable roof supports Little difficulty as minimal shielding
Cutterhead turns one direction, grippers anchor
Installation decoupled from boring. Bolt drilling during mining
Easier probing Easier pregrouting and face access possible
Single shield nonarticulating articulated
Propel jacks thrust on lining installed in rear of shield (or may use auxiliary gripper reaction ring)
Controlled hydraulic fluid flow to jacks Articulation via bevelled shield joint
Add gauge cutters Shim gauge cutter mounts Hydraulic 'copy cutters' for selective overcut Nonarticulated shields may nose-dive, especially with overcut at invert Shorter shield, less area for squeeze pressures to act, but single shield operations slower (support installation), so more time for squeeze to develop
Reversible cutterhead rotation
Boring stops for segment (or other as steel sets/ lagging) installation
Bevels for articulation difficult to manufacture, especially for large diameter TBMs
Main propel cylinders to sidewall grippers which extend through shield to tunnel wall Auxiliary system jacks act on lining
Main propel cylinders used for steering
Muck and ground squeeze can pack in telescopic joint May step or taper rear shield section to avoid pressure binding, but extra backpacking/grouting required behind segments Overcutting options as for single shield above
Main propel cylinders in an arrangement used to control roll
Double telescoping shield (variety double shield with extendable cutterhead, advanced independently of shield) Additional alternatives for squeezing or poor ground
May use gimballed cutterhead (independent attitude positioning)
Bentonite injection around shield (especially if progress shut-down occurs) Contracting (split) shields Walking gripper shields
If water inflows likely, need to waterproof seal the bevel joint
If good rock and sidewall grippers used, support installed on a no delay basis (if support is required) Auxiliary jacks thrust on lining in poorer ground conditions
If water inflow, seal required to waterproof telescoping joint (easier to seal against water pressure, but more complex and costly and cutterhead vulnerable when extended beyond shield)
Grill bars reduce muck bucket slot opening Cutterhead false face/recessed cutters stabilize face
Sticky muck - use teflon or other lining for conveyors
266
Mechanized Excavation
limestone materials may actually set up to form in situ concrete at the face and along the muck transport system. 10.2.2 TBM System Performance Parameters TBM system performance is evaluated using several parameters which require definition. In the literature, most of these parameters have been inconsistently used, and many literature citations of project performance are difficult to apply in comparisons because there is no certainty as to the definitions used in calculations. Contractors may use three 8-hour shifts per day, and maintain equipment as needed 'on the fly'. Some contractors schedule a special daily maintenance shift during which no mining is intended. As used here, the shift time on a project is all working hours, including time set aside solely for maintenance purposes. All time on a project is therefore either mining time when the TBM is operated, or downtime when repairs and maintenance occur. Therefore, Shift time = TBM operating (or utilized) time + downtime
(1)
When the TBM is operating, there is usually a clock on the TBM which records all operating time. The TBM clock is activated by some minimum level of propel pressure and/or by a minimum torque and the start of cutterhead rotation. This operating time is used to calculate the penetration rate (Pr) - how fast the cutterhead advanced per unit TBM time. Therefore distance mined
Pr =
(2) TBM operating time Penetration rate is calculated as an average hourly value over a shift, over a day, week, month, year, or the entire project. The basis for calculation should be clearly defined. When averaged over an hour or shift, Pr values can be of the order of 2-10 m h " 1 . The Pr can also be calculated as distance mined per cutterhead revolution, and expressed as an instantaneous penetration, averaged over each thrust cylinder cycle or other time period. The particular case of penetration per cutterhead revolution is useful for study of rock cutting mechanics, and is here given the notation P rev . Typical values of P rev can be 2-15 mm per revolution. The percentage of shift time during which mining occurs is the utilization, U. Utilization is calculated as U
%
( ) =
TBM operating time
u- f t ,.
shift time
( 1 0 °)
<3)
and is usually evaluated as an average over a specified time period. Utilization is typically lower in smaller diameter tunnels, with limited space for mucking and other activities. There is no clear evidence that projects utilizing a reconditioned machine will have a lower U than projects completed with a new machine; U depends more on contractor capabilities and maintenance plans. Values for U reported in the literature are particularly suspect. For example, sometimes U is reported which excludes the necessarily less productive start up time early in a project. It is important that U is reported together with the basis for calculation - whole project (including start up), after start up 'production' average, or U over some other subset of the job. Utilization depends on many things, including ground conditions and support, equipment condition/maintenance, contractor capabilities, project conditions (shaft/portal entry, alignment curves, surface space constraints on operations) and human factors (remoteness/access, underground temperature and environment). On a shift basis, U varies full range from nearly 100% to zero. When evaluated on a project-wide basis, U values of 40-50% are typical. Advance rate (AT) is defined on the basis of shift time as Ar =
distance mined shift time
(4)
If U and Pr are expressed on a common time basis, then the Ar can be equated to
A, -P.?™ r
100
(5)
TBM Performance Analysis with Reference to Rock Properties
267
The best linear Ax (rnh -1 ) is probably possible for 3.5-5 m diameter tunnels, but cost-effectiveness (as $m" 3 ) or efficiency (kWht - 1 ) (1 kWh = 3.6 x 106 J) is best on large-diameter machines. Advance rate variations can be caused by changes in either PT (as encountering very hard rock, or loosening some of the TBM drive motors), or in U changes (such as encountering very poor rock with changing support needs, unstable invert which causes many derailments, or highly abrasive rock which results in fast cutter wear). One additional parameter related to the Ar is the cutting rate (Cr), defined as the volume of intact rock excavated per unit time. Again, the averaging time unit must be defined clearly, and typical values of Cr range from 20 to 200 m 3 h _ 1 . Other performance parameters deal with cutter replacement rates, which depend on cutter position and type of cutter, rock properties (hardness, strength, abrasiveness) and also on thrust, cutterhead rotation rate and gauge cutter rolling speed. Parameters used to evaluate cutter change frequencies include: average TBM mining time before change, linear distance of tunnel excavated per cutter change, distance actually rolled by a cutter (the rolling life) and rates of material wear (disc weight loss or diameter decrease). Rolling life may be 100-300 km for abrasive rock, to more than 3000 km for nonabrasive rock. The disc mount (hub and cutter bearing) also suffer during mining, and statistics may be kept on hub life, the number of discs used on each hub, etc. 10.3 ROCK PROPERTIES 10.3.1 Fragmentation by Cutting Tools Figure 2 is a schematic drawing which illustrates many terms and concepts developed to understand disc cutting processes. Disc cutting tool action involves initial elastic indentation, inelastic crushing of rock material beneath the cutter disc, and chip breakout by fracture propagation from the crushed zone to the free surface at the adjacent groove. The muck created in this process includes fines from crushing and chips from fracture. The fines are active in disc wear, and rocks with high porosity (much crushing) and/or high content of abrasive minerals will promote abrasive wear. Rock chips have typical dimensions of 5-15 mm, thickness, widths of the order of the cutter kerf spacing, and lengths of the order of one to three times the chip width. The action of indentation causes cracking and general damage to the rock beneath and near the groove. This process is continuous and a steady state damage zone is extended into fresh rock with each cutterhead rotation. For nonporous brittle rock, which is dilative during crushing, disc load and penetration create a local zone of high triaxial stresses in the highly confined crushed zone. The intensity and stress direction within this zone are poorly defined for a moving cutter, but it is apparent that stress components develop normal to the direction of indentation. With adequate stress intensity in the crushed zone, crack propagation will occur from the crushed zone boundary into the unfailed rock. The development of brittle cracking is evidenced by the accompanying explosiveness, noise and dust ejected. Of interest to efficient cutting with discs are the cracks which propagate more or less parallel to the cut face laterally to the adjacent groove. Chipping is really a discontinuous process, such that chips do not form at all locations on every disc pass. However, in many cases chip thicknesses are of the order of the penetration per revolution, and it may be concluded that 'single pass' cutting is occurring.
Figure 2 Disc force and geometry definition for kerf cutting
268
Mechanized Excavation
From this brief discussion, it is clear that several things are important for efficient interactive or kerf cutting with disc cutters. (i) The cutter indenting (or normal) force and penetration must be sufficient to produce contact stresses adequate to form a crushed zone. (ii) The stresses in this crushed zone must be high enough to initiate crack propagation into the less damaged surrounding rock between grooves. (iii) The adjacent, unloaded, groove (a free surface) and its local zone of cracked rock must be near enough so that lateral cracks from the loaded groove can interact and extend to create a chip. Such lateral crack propagation may occur on loading or on unloading as a disc passes a particular location on the rock face. (iv) There must be a disc force component adequate to maintain cutter movement, in spite of the rolling resistance or drag associated with the indentation process. Rock properties and the disc geometry and penetration control the contact stress produced during indentation. The penetration is effected by the applied TBM thrust. The average thrust, or normal force (Fn), per cutter is calculated as Ncp'cnd2c Fn = — An
(6)
where Nc is the number of thrust cylinders; p'c is the net applied hydraulic pressure; dc is the diameter of each cylinder piston; and n is the number of cutters in the array. Disc rolling is effected by supplied machine power and cutterhead rotation. The average rolling force per cutter, Fr, is calculated as follows for electric systems NmPe 2nnrRc where Nm is the number of motors; P is the motor operating power level; e is the motor and drive train efficiency; r is the cutterhead rotation rate (rpm); and Rc is the weighted average cutter distance from the center of rotation, about 0.6 times the TBM radius for most cutter arrays. TBM operating conditions are not uniform. When chips form, rapid changes in cutter loads occur. Wear and staggered cutter replacements result in varying disc diameters, cutter spacing (s) and orientation are not uniform across the cutterhead, and some discs must roll through muck accumulations at the tunnel invert. Therefore, it is not likely that the forces calculated in equations (6) and (7) are actually developed for any particular cutter. In addition, rock is not homogeneous and varying intact and rock mass properties may produce uneven force distributions and impact loading. However, it is convenient to discuss disc/rock interaction in the context of these average forces, and also average disc cutter spacing (s) and disc indentation or penetration per revolution (p). The interaction of the normal and rolling forces and the resulting penetration is indicated in Figure 3. The force/penetration relationships are typically nonlinear, more so for the normal force relationship. This observation relates directly to the compound crushing/chipping process involved in disc indentation, with the initial low slope section associated with crushing and the steeper higher force section indicating the onset of groove to groove chipping. The transition between the two processes has been called the 'critical thrust': unless force of this magnitude can be applied, chipping between grooves will not occur. The critical thrust is directly related to rock strength or hardness, and also increases with cutter spacing and disc contact area which, in turn, is related to disc diameter, the disc section used and amount of wear on the tip. Several summaries of the important factors for rock/disc interaction are available [2, 5, 6]. Force/penetration relationships vary for different rocks, with either required forces increasing or resulting penetrations decreasing for stronger rocks. Although these relationships are known to be nonlinear, several parameters have been defined based on ratios or slopes derived from force/penetration plots. The ratio of rolling to normal force has been defined as the cutting coefficient (Cc), and the normal force to penetration ratio is defined as the penetration index (R{). Much information about rock/disc interaction has been gained from laboratory linear cutting tests, with the rotary cutting process modeled as linear paths of indexed cutter indentations. Testing indicates that neither disc rolling velocity nor linear versus curved travel paths affect force/penetration relationships. The specific energy of cutting (energy per volume of rock excavated) has been used to understand optimum cutting geometry for efficient cutting. The optimizing process involves the ratio of kerf spacing to penetration, the s/p ratio. For a wide variety of rock Hthologies it appears the Optimum' s/p is similar, indicating that the nature of kerf interaction is more a characteristic of
TBM Performance Analysis with Reference to Rock Properties
269
Low strength rock
High strength rock
Figure 3 General plot of disc cutter force variation with penetration for high and low strength rocks
geometry than rock properties - as long as chipping is occurring. For typical machine designs with set cutter spacings, an s/p ratio less than optimum occurs in weaker rock with high penetrations at lower cutter forces. Although such operation is less efficient than at optimum s/p ratios, resulting advance rates are usually adequate to satisfy any project schedule. However, for strong rock, reduced penetration and effectively increased s/p ratios can occur even at elevated normal force levels. For a given TBM, the critical thrust for hard rock may be too high to reach, and the penetration decrease can be so extreme as to obviate kerf chipping. In such a case, it is clear that single pass chipping cannot be achieved, and that multiple disc passes or several cutterhead rotations may be required for chip formation. Many theories have been invoked to explain the indentation process in brittle (nonductile and low porosity) rock, each of which makes certain assumptions as to material behavior or to simplify complex equations. There is no general consensus as to the appropriate model for indentation mechanics, although offerings have included force equilibrium, stress analysis and fracture mechanics approaches [e.g. 2, 6-11]. For ductile rock, the indentation process might be more correctly modeled by plasticity or with one of the general bearing capacity approaches developed for foundation design. For porous rock, indentation results in local rock fabric collapse, causing crushing with compaction of available pore space. This compaction occurs without volume increase or dilative effects so that high stresses do not develop in the confined crushed zone, and crack propagation and chip formation are inhibited. The volume of crushed material can be large and, if the powder is abrasive, discs may wear rapidly and fines can intrude into the bearings, a process exacerbated by the presence of water. 10.3.2 Laboratory Rock Tests The goal of a laboratory testing program for an underground construction project is different from that supporting other types of works, for which the focus is primarily to support the geotechnical design. For tunneling projects, the tests performed and recommendations made must also provide the contractor with a sensitivity to geotechnical conditions before construction, permitting estimation of cost and schedule, and supporting the selection of appropriate equipment. It is not surprising, therefore, that the tests used to characterize rock for excavation purposes are very different from tests utilized in other kinds of projects. Since the precise nature of chip formation involves several processes, many index tests have been created which either measure some correlatable characteristic of the rock, or which mimic the process in the laboratory. Several good reviews of these tests are available in current literature [2, 6, 12-16]. General characteristics of some tests will be discussed, and test procedures can be found in these references. Some of the earliest tests developed involve static indentation of confined rock specimens. A wide variety of indentors, testing procedures and equipment has been used. Characteristics of six developed tests are summarized in Table 2. These tests use different indentors, some tests use indexed penetration, and some do not load to chipping. Results can be reported as normal force versus penetration plots, and generally result in data trends which mimic those exhibited by TBMs in operation. However, the tests remain nonstandard, although machine manufacturers and others
Mechanized Excavation
270
Table 2 Static Indentation Index Tests Single or indexed penetration
Load to crack or chipping
Application
Button Morris Handewith Button Colorado Disc cutter School of Mines
Single Single Indexed
Yes Yes Yes
NCB O & K Wedge
Single Single
No Yes
Empirical Empirical Direct to penetration, cutter forces Empirical Empirical
Test
Indentor
Cone Wedge/tooth
have developed databases of test results which are valuable via empirical correlation with field performance. Unfortunately, these databases are largely considered proprietary. A second group of index tests can be called 'hardness' tests, including Shore hardness, Scleroscope hardness, Taber abrasion hardness, Schmidt hammer rebound hardness (HR), and a parameter called Total Hardness, which is calculated using Schmidt hammer and Taber abrasion hardness test results. Test procedures have been summarized and utilized in empirical performance prediction derivations [16,17]. Hardness testing requires specialized equipment, but the tests are relatively easy to perform on rock core specimens and therefore have sometimes been abused in performance and application by inexperienced professionals. Some dynamic impact tests have developed for application to TBM performance prediction. These include Rock Impact Hardness (RIH), Coefficient of Rock Strength (CRS), and the Swedish Brittleness Test (S2o)· Each test exposes a sample of rock fragments to a number of falling weight impacts, and uses the sample weight loss through a prescribed sieve as the metric. The S 20 value is incorporated in the performance prediction method developed by the Norwegian Institute of Technology, NTH [18]. A wide variety of'drillability' and 'abrasivity' index test procedures has also been developed [14, 19]. Some test characteristics are described in Table 3, and each test method requires specialized equipment. The CERCHAR (the Laboratoire du Centre d'Etudes et Recherches des Charbonnages de France) test has been utilized in assessing tool abrasiveness, and West [15] presents details of the test and correlation of results with mineralogical abrasiveness measures including quartz content and Moh's hardness scale. Abrasivity indices have also been developed which focus on mineralogical and petrographical characteristics, such as using Moh's or Rossival's hardness scale to calibrate an evaluated relative rock hardness. Some workers have simply used the percent quartz as a measure, and others have created compounded abrasion or drillability coefficients, as the DRI (Drilling Rate Index) used by the NTH. Application of these tests is exclusively through empirical correlations with TBM case study data. Empirically derived prediction equations have also incorporated results from conventional rock strength testing. The rock property most widely accessed in performance prediction has been the uniaxial compressive strength (UCS). Such usage is dependent primarily upon the availability of UCS test results, often the only test data available to contractors at time of bid. However, UCS is certainly not the ideal parameter to use in predicting the result of what is mostly a process of brittle fracture. In addition, UCS test results are sensitive to the core size, specimen preparation, specimen geometry, care given to maintain water content before testing and stress relief effects on core which can cause a test sample selection bias as well. Some methods have also used the Young's modulus determined in UCS testing as a prediction parameter, and other predictive models utilize information on the rock shear strength, as measured in a punch shear test or estimated from tensile and UCS test results. The rock tensile strength, often measured in a Brazil test, is underutilized for machine performance prediction. Brazil test results can be used to evaluate whether brittle behavior will occur on disc indentation, and the test can be used to evaluate rock strength anisotropy, more definitively than can the point load test which has also been used. Rock fracture toughness and the critical energy release rate (or critical crack driving force) are rock properties which have great potential application for machine performance prediction, but few rocks from tunneling projects have been tested so the demonstrated correlations to date must be considered preliminary. There are many methods for fracture property evaluation, as considered in the recent efforts by an ISRM Commission [20]. Most applications have considered Mode I (crack opening) toughness evaluation.
TBM Performance Analysis with Reference to Rock Properties Table 3
271
Drillability and Abrasivity Index Tests
Test
Drillability
Abrasivity
Goodrich Taber Voest-Alpine (rock cuttability index) CERCHAR Paddle Abrasiveness LCPC Abrasimeter Siever's J-value Norwegian Abrasion Value
Hole drilled depth
Tool wear width 1/wheel weight loss Tool weight loss
Groove depth 10 mm hole drill time Sieved weight loss Hole drilled depth
Tool wear flat diameter Weight loss of paddle Weight loss of paddle Tool weight loss
In general, other properties are also evaluated during site investigations, and empirical correlations have included properties such as density, water content and seismic velocities (compression and shear, field and laboratory) in linear regression equations. A measure of porosity is also of interest, as is information on nonlinear stress/strain response and development of inelastic deformation at low stress. For weak rock, Atterberg limits and clay mineralogy should be evaluated early in the site investigation, with more specialized testing for swell, squeeze and consolidation properties perhaps warranted on the basis of the results of index tests. Linear cutting (and rotary cutting) test equipment exists in the US, UK, Germany, Australia, Japan and no doubt elsewhere. Such testing has been performed primarily in research efforts to understand the interrelationships between rock properties, disc forces, disc geometry, cutter kerf spacing, machine stiffness and penetration. In some cases, linear cutter testing is performed for contractors who plan to make their own decisions about equipment purchase or reconditioning. Such testing is expensive, and not likely to be pursued for many tunnel projects. Linear cutter test results may be directly applicable to full-scale TBM penetration rate prediction when performed with identical cutter spacings and discs will be used on the project. Such an application should, however, consider the differences between the tested rock and the rock mass in situ, including likely differences in relative stiffness between the rock mass and TBM and between the laboratory rock specimen and laboratory equipment. Tests selected for a site investigation should depend on the application. For comparison of several alignments, a simple inexpensive test may be sensitive enough to detect differences in boreability, to identify where problems will be, and to give a general estimate on how much thrust and torque will be required. In such a case, several of these indices may be appropriate. At this time, there is no particular suite of rock property tests which is recommended for tunnel project investigations. Testing should, however, include both tensile and compression strength, an evaluation of porosity or other measure of dilative versus contractive response, and an evaluation of rock abrasivity. Care should be taken with core to minimize stress relief effects and moisture loss. Sampling biases in terms of very weak or very strong rock must be avoided as it is these extremes which often define success or failure for a TBM application. For use in specific predictive approaches, particular tests can be performed, such as the various hardness tests [16] or the suite of tests incorporated into the NTH methodology [18]. In all cases, specified equipment is mandatory and suggested procedures must be followed. In any event, test results require the interjection of judgement and experience for interpretation and before application. 10.3.3 Rock Mass Properties The primary impact of rock mass properties on TBM performance is on utilization, an impact which depends greatly on the chosen construction and support methods and equipment selection. To a certain extent, rock mass properties may also affect penetration rate and cutter wear. Site investigations should be geared to address certain basic questions on equipment selection. Some of these questions are summarized in Table 4. The actions and decisions associated with the answer to each geomechanics question are the responsibility of the contractor, but clear assessment of each geomechanics question is the responsibility of the investigating engineers and geologists.
272
Mechanized Excavation Table 4 Rock Mass Related Questions to be Answered During Site Investigations
Geomechanics questions
Contractor's questions
How hard is the hardest rock?
What thrust, torque, cutterhead rpm, cutters and cutter spacing should be used? Can sidewall grippers be used? Will the muck be hard to handle? Will the alignment be hard to maintain? Will track stability and derailments be problems? If a shield is used, can it get stuck?
How soft is the softest rock? Is there likely to be soft rock in the invert? For how much of the tunnel? Will the ground develop squeezing? If so, over how much of the alignment and how fast will squeeze occur? Are loosening or overstressed conditions likely to be developed? Is there very abrasive rock present, and over how much of the tunnel? How different can rock strengths be at the heading at one time? What is the shortest stand up time likely? How blocky is the rock? Is overbreak likely?
How likely is it that ground water inflow at high pressure or volume will occur? What is the ground water chemistry? Will the facility operation be impacted? Is the tunnel likely to be gassy? With what certainty are critical factors known?
Is a shield necessary? Should the cutters be recessed to provide ground control at the face? Is manned entry to the face required? What will cutter costs be and how will the schedule be affected? Is disc impact loading of severity likely to occur for long tunnel stretches? Will cutter costs be high? Will there be gripping problems when mining from" soft into hard material? How fast does support need to be installed? If faster than is possible, is a shield required? Are blocks likely to jam the muck system? Will there be problems with sidewall gripping? Can an expanded segmental lining be used? Will backpacking and grout costs be high and/or affect the schedule? Can the tunnel be driven upgrade for gravity drainage? Will there be muck handling problems? Should the equipment be waterproofed? What pumping capacity is needed? Will there be problems for water disposal? Will there be construction safety problems underground? Does equipment need to be specially designed? Is advance probing through the cutterhead required?
Evaluation of rock mass behavior must take advantage of available information from any previous underground works in comparable conditions, and field work should include geological mapping and outcropping rock mass description. For general geotechnical purposes, rock mass properties for performance prediction should be simple, cheap and standard and calibrated to field experience. Rock mass classification should be used as a basis for subdivision of the alignment into geotechnical and construction zones, so that the severity and extent of critical conditions as high strength or mixed face variable hardness conditions can be identified and planned for by the contractor. These zones or classes may be used as a basis for scheduling and payment [21]. There is much underutilized information available through geology and geological interpretation. For example, the cooling history of a massive granite body may be known well enough that the variation of stress and micro- and macro-structural characteristics across a site can be understood in this context. Geological information may also be used to deduce likely occasions for high stress conditions to be present, and where along an alignment these stresses may have been relieved. Experienced geologists and engineering geologists should always be utilized during the site investigation for any major project. For consideration of utilization impact, classification systems which yield an appreciation of rock mass variability and stand up time can be used, including the RMR, Q, RSR and RQD methods. For particular projects, some specialized ground classifications have been developed, some of which incorporate seismic velocities [22-24]. Although several attempts have been made to use various systems of rock mass classification to evaluate utilization impact, there is no clear recommendation to be made now. It is clear that varying ground conditions disrupt mining activities if ground support operations must be changed. The strength of the impact depends greatly on the equipment used on a specific project, and on how prepared the project management is to respond to changing geotechnical conditions. To minimize impact on TBM performance, the mining and support operations must be compatible, so that change in stand up time does not affect the ability to support
TBM Performance Analysis with Reference to Rock Properties
273
on a no delay basis. For best TBM operating conditions, the expected stand up time should be at least one hour. Discontinuities in a rock mass can certainly serve to increase penetration rate above that anticipated on the basis of laboratory testing on intact cores. The same rock mass classifications used for stand up time and support evaluation could also be applied to evaluate such an effect, but classifiers should note that the discontinuities selected for use in rock mass classification for ground stability applications may not be those with greatest impact on penetration. The most effective discontinuities for penetration rate are those oriented parallel to the excavated face and/or which have average spacing on the order of the disc cutter spacing. This means that indices, such as discontinuity frequency or RQD, evaluated on vertical core may not present information pertinent to penetration rate evaluation. For directed application in TBM penetration rate prediction, Bamford [12] suggests that two geotechnical factors based on rock mass classifications schemes warrant more study: GR based on the RMR system, using the sum of parameters 3 and 4 with Laubscher's adjustments; and GQ based on the Q system, using the product of the first two ratios RQD/Jn and Jr/Ja. However, neither of these suggested factors has yet been implemented in performance prediction. 10.4 ROCK PROPERTY IMPACT ON TBM PENETRATION RATE 10.4.1 General Observations The penetration rate of a TBM is the result of complex interaction between the intact rock mass, machine design and the conditions of machine operation. Any method which reliably predicts penetration rate must include each component. The 'simple' process of disc indentation and chip formation is actually quite complicated and, considering the variability of most rock masses, it is not likely that one rock property can describe the entire process. However, there is evidence that some description of rock does exist which could be applied to predicting performance. Figure 4 is a plot of TBM cutting rate versus the measured power consumed during cutting by single disc tools. Data are included from casefilesfor dolostone excavation by four TBMs at the TARP (Tunnel and Reservoir Plan) project in Chicago, IL, for two TBMs operated in the Seabrook, NH tunnel drives in quartzite and schist, for dolomitic limestone excavation by three TBMs for the Buffalo, NY subway tunnels [25], for one machine performing excavation of weak coal measures rock at four locations in the Selby mines in the UK [26], and for one TBM operated at two cutterhead rotation rates in basalt [27]. Straight lines are shown on the plot to indicate data trends. The cutting rate increases linearly with power delivered, and the rate of increase varies for different rock masses. This variation appears to be independent of the tunnel diameter (TARP tunnel diameters included vary from 4.3 to 10.8 m), an observation which is actually biased by the design of 200 175 150
~ 125
B
100 75 50 25
Power consumed (kW)
Figure 4 TB M data on cutting rate versus power consumed
274
Mechanized Excavation
torque capacity for these Robbins machines. It is clear that the sedimentary sequence at Selby is easier to cut than is the schist at Seabrook. The slope of these lines represents the specific energy of cutting, which is relatively consistent for each rock mass. In fact, even in relatively consistent rock, penetration rates can vary. The 21 TARP project tunnels completed from the mid 1970s to the present are in fairly consistent rock. These contracts range from 0.9 to 12.8 km in length, with tunnel diameters varying from 2.0 to 10.8 m. Both Jarva and Robbins TBMs have been used, and projects have involved both new and reconditioned TBMs. The range in project penetration rates is 1.6-5.1 m h" \ with an average rate of 2.3 mh" 1 [28]. In such consistent rock, it is clear that more than rock properties must be controlling the rate of progress. The biggest contribution to the variation comes from contractor management and equipment capabilities newer TBMs were more powerful and utilized larger, more highly loaded cutters. The influence of rock properties cannot be underestimated, however. Consider the information shown in Figure 5, case study information for a tunnel in Rochester, NY [25]. The variation in average Prev and average Fn are shown for each shift along the 4 km section of tunnel. Stationing in the figure is in feet, per US practice. The geological units encountered are shown at the top of the figure, and mining proceeded from left to right. In this tunnel, the motors were constantly operated above their maximum rating to deliver as much torque as possible. Changes in Prev clearly track the change in rock at the heading. Highest Prev was achieved for mining in shale, lowest for mining through sandstone and limestone. Thrust and penetration show an inverse relationship, so that the reduced Prev in the limestone was only achievable with significant increases in Fn. It is clear that Fn is a significant variable for any prediction model. TBM thrust delivered to the cutters is calculated from the cylinder pressure reduced by system efficiency and losses such as from back-up system towage during mining or friction losses from contact between the machine and the rock. For full shields this loss can be very high, and may ultimately stop forward progress if ground pressures on the shield are larger than can be overcome by available thrust. As a total, the net average cutter normal force can easily be 40% less than the average gross force calculated from hydraulic cylinder pressures [29]. For very hard rock, thrust Tunnel station (ft) J 3000
5000
Thorold sandstone
1
I
11000
Reynales limestone
\
/
^ ζ ζ ^
/ Grimsby sandstone
\
/
/ shale Maplewood
el»·»-'·. . ^/..vsS* - ..v.
3000
I
9000
7000
5000
9000
13000
I
15000
Irondequoit limestone
\
\
^ ^
/
Lower Sodus/and Williamson shale
«.]**?:· v
' <"**.
·. V .. A/** *: '~v>
15000
11000
150 125 100
* · * · Λ J * * ** ***
te' ·"
75 50 25 I 3000
5000
_L
7000
_L
9000
_L
11000
13000
Tunnel station (ft)
Figure 5 Geology, Prev and average cutter normal force data for TBM excavation of a tunnel in sedimentary rock [25]
TBM Performance Analysis with Reference to Rock Properties
275
losses may severely limit the penetration rate. The use of fewer, more highly loaded cutters in hard rock is an effective way to deal with thrust-limited mining. On a recent project in Norway, the Prev for 483 mm diameter cutters at 312 kN gross thrust per cutter was 25% higher than for 432 mm diameter cutters at 222 kN gross thrust per cutter in comparable hard rock ground [30]. TBM operation is softer rock is often limited by installed power on the machine. Torque losses arise with muck accumulation in the invert and in much buckets. Particularly at high cutter loads and therefore harder rock, cutter bearing friction also causes a loss to net torque. A 'frozen' or blocked cutter with a seized bearing will cause a significant torque loss, but the attendant vibrations make the operator aware of this problem almost immediately. Figure 5 can also be used to demonstrate the importance of machine torque to the cutting process. The low Prev for mining the first part of the Maplewood Shale developed when motor problems temporarily decreased the available torque, also sticky muck clogged the cutterhead and muck buckets resulting in torque losses from friction and drag against rotation. The net result was a very low P rev . When the muck jams were cleared and all motors were functioning, the Prev improved considerably. Before field performance can be understood, a sober assessment of system limitations and operating torque and thrust losses is important. Load capacity of a sidewall gripper system can also limit the level of thrust and torque that can be applied. With weak rock, the gripper may slide or develop local bearing capacity failure in the sidewall rock. In bad ground where overbreak develops, wood cribbing may be required if the overbreak is more extensive than the gripper cylinder stroke. These problems are particularly severe when mining from weak rock into hard rock, with high thrust desired for efficient cutting but with grippers bearing on low strength rock. For shielded TBMs, the strength of the lining may limit operating thrust and torque. All of this discussion has been rather pragmatic, and has not considered the inherent inefficiency of rock cutting processes. Of the supplied cutting power, most goes into cutter/rock friction, kinetic energy associated with explosive chip release, muck and cutter temperature increase, evaporation of water and strain energy in the damaged rock. Power utilization is low for the work of cutting and creating new surface area, and the surface energy component of work is dominated by the fines produced rather than the larger chips. Much engineering research has evaluated cutting optimization on the basis of specific energy of cutting (energy expended per volume material comminuted), using total torque delivered and not considering the energy apportionment to various dissipative processes. Precise accounting of energy is difficult, however, and TBM designers assign more value to optimizing penetration rate than to minimizing specific energy of cutting.
10.4.2 Correlations with Intact Rock Characteristics There is no widely accepted public domain method to predict P r from laboratory test results on intact rock specimens. Correlations between P r and various rock properties abound in the literature, but few empirically derived expressions are valid for rock masses and TBMs different from those in the original databases. Before considering some of these predictive methods, some initial comments are in order. (i) Empirical correlations require good laboratory test results, obtained using standard procedures and with due respect for stress relief and moisture content change effects. (ii) Most of the correlations involve only laboratory test results. Field operations certainly include effects of discontinuities which are not reflected in intact rock testing of the ordinary genre. This influence should be 'controlled' in the database. (iii) Clear descriptions of geotechnical conditions for database projects are required so that the pertinence of the predictive methods for each application can be assessed. (iv) Good correlations require good field performance information. In many cases, this information is poorly recorded or case history information is incomplete, and the more extended databases are considered proprietary/confidential. (v) Low thrust and low torque mining through alignment curves may result in locally reduced Pr which should be considered in developing a field case study database. (vi) Some databases include performance with single, double and triple disc cutters, a variation which greatly affects disc edge loading and average cutter kerf spacing. (vii) Disc cutter wear has an effect which is not often accounted for. However, a stable average disc wear level is probably reached early in a drive, so this is likely a secondary effect except in highly abrasive rocks.
276
Mechanized Excavation
As an additional cautionary note in developing database correlations, particular attention should be given to clearly defining the basis for evaluation of performance parameters. Consider the project data shown in Figure 6, here presented as shift averaged Prev information in histogram format [31]. Data have been divided corresponding to two geotechnical tunnel sections, a zone with several faults and a zone of nearly uninterrupted mining in good quality chalk. For each zone, Prev varied and was somewhat reduced in the fault zone. For empirical correlations, what value of Prev should be used? Some variation in each zone derives from rock mass variation along the alignment, and it may be that only a low estimate of Prev in good quality chalk should be used in correlations, since any rock mass effects would tend to increase penetration. In general, most databases use average field performance estimates for correlation purposes. Many efforts have been made to correlate laboratory index test results to TBM penetration rate. Recent reviews [12,19, 32] present equations for all variety of approaches, and an exhaustive listing of equations will not be made here. Prediction equations are either empirically derived or developed with a theoretical basis such as force equilibrium or energy balance. Simplified assumptions of disc indentation geometry and contact zone stress distribution are usually made. In some cases, an equation originated in theory contains coefficients derived from correlations with case history information. Most prediction methods agree on trends, but empiricalfitsare strongly case specific in terms of geology and machine characteristics. As an example of simple linear regression applied to case history data, Nelson [25] investigated relationships between carefully evaluated average penetration rates and rock properties including UCS, Brazil tensile strength, point load index and various hardness measures (rebound, abrasion and total). The statistical significance of the correlations was relatively poor until Fn was included. The correlation between R{ (the ratio of Fn to Prev) and total hardness, HT, was most significant. For demonstration purposes, rock properties and machine performance data for four tunnel projects in sedimentary rock are used to assess the predictive ability of four published correlations: Farmer and Glossop [8] using tensile strength; and Graham [33], Roxborough and Phillips [10], and Hughes [34] using compressive strength. Rock test results, TBM performance, and predicted penetration rates are shown in Table 5. Each project machine was fitted with 394 mm diameter single disc cutters, with average kerf spacings between 64 and 76 mm. Average disc forces vary widely but generally directly with compressive strength, and the cutter loads listed are well below the maximum load suggested (222 kN) for the cutters used. For each of these four TBM projects, PT and thrust were in fact limited by available torque or by the muck handling system capacity. The predicted penetrations in Table 5 are nearly always less than those achieved by TBMs in operation. Predictions are most different for the weakest rock, particularly for the chalk in Texas. This reflects the influence of the databases accessed in the original correlations (mostly stronger rocks), and perhaps a change in indentation mechanics which occurs in this weaker rock. The Farmer and Glossop equation yields consistently higher predicted penetrations and the Hughes predictions are consistently lowest. The influence of rock test material condition is indicated by the information for the Grimsby Sandstone. Much of the original testing on this project was performed
I
2
3
4
5
6
7
8
9
10
II
12
13
14
15
Penetration per revolution,Prev (mm)
Figure 6
Frequency distribution of Prev data in zones of good and poor quality rock, Govalle Tunnel, Austin, TX (author's files)
TBM Performance Analysis with Reference to Rock Properties
277
Table 5 Comparison of TBM Case Study and Predicted Penetration Rates Rock strength (MPa)
Project information
Location Buffalo (NY) Rochester (NY)
Chicago (IL) Austin (TX)
Rock unit Falkirk dolostone Oatka dolostone Williamson/ sodus shale Reynales limestone Maplewood shale Grimsby sandstone: wet dry Romeo dolostone Markgraf dolostone Austin chalk
TBMr performance
p
Brazil tensile
(kN)
188 139
13.3 13.0
134 108
80
—
128
UCS
Prediction method 1 - Farmer and Glossop [8], 2-Graham [33], 3-- Roxborough and Phillips [10], 4 - Hughes [34] 1
2
3
4
7.6 10.4
6.3 5.2
2.8 3.1
4.3 4.6
2.2 2.3
99
10.0
—
4.9
6.2
3.8
15.0
141
6.8
5.9
4.3
5.8
3.3
68
—
98
10.4
—
5.7
6.9
4.5
130 108 237
10.1 6.1 17.0
112
7.9
145
8.0
6.9 11.5 5.3
3.4 4.1 2.4
4.9 5.6 3.9
2.4 3.1 1.6
168
12.1
137
9.3
7.1
3.2
4.7
2.3
10
1.3
33
9.6
15.7
99.1
46.5
11.8
*■ rev
(mm)
Sources of case study information: NY and IL projects [25], TX project [31].
on air dry rock. When the rock was resaturated and tested, strength reduction was evident. This uncertainty as to intact strength can clearly affect the penetration rate predicted. The number of single parameter equations available leads to much confusion in Pr predictions, a difficulty addressed in a study supported by a mining research association in Australia [35], in which 24 predictive equations were considered. In applications, no single approach was highly recommended; rather the equations were considered useful to assist in design and selection of equipment, and for sensitivity studies of the relative importance of various factors. Overall, the interpretation and application of rock tests results requires experience, and more than one method of prediction should be used to develop insight into probable machine performance. For brittle rock, penetration and chipping are related closely to fracture mechanisms. Such an empirical relationship between R{ and Glc, the critical energy release rate, has been identified and the data are included in Figure 7 to correct a units error in the original publication [36]. The open symbols in this figure correspond to data for TBM performance in four fairly massive rock units. The closed symbols represent linear cutter testing of four brittle rocks [3, 37], with the cutter forces from the 200 mm diameter discs scaled to equivalent 394 mm diameter forces using the relationship derived in [10]. The indicated relationship between Rf and Glc is promising and deserves additional investigation. Some additional prediction equations utilize correlations with specific energy of cutting, Es (m3 h~x). Boyd [in 19] suggested such a relationship, which depends on proprietary rock test results with modifications for rock mass properties (a 'rock boreability index') to predict the gross Es as a rock mass property. This Es is used, together with installed power and tunnel dimensions, to evaluate penetration rate. This is a reasonable approach and, considering the linear trends displayed in Figure 4, additional work here is warranted. Slopes for the linear trends in Figure 4 are listed in Table 6, together with data given in the reference [19]. The Es values from Figure 4 have been increased by 25% (an estimate) to convert from net to gross energy applied. The Es values from the two sources are certainly of the same order of magnitude, and it is possible that a reliable laboratory test could be developed to evaluate Es. However, the influence of relative stiffness (rock/machine) is important, with high stiffness laboratory rock (rather than lower stiffness rock mass) yielding high laboratory values of Es. The Norwegian Institute of Technology (NTH) has developed the most thorough published predictive approach for TBM performance [18]. Most information has been derived from Norwegian projects, so that there may be a bias in rock masses included. However, it is certainly the most
278
Mechanized Excavation
E
30
E z 5 25 x
DReynales limestone olrondequoif limestone " Δ Markgraf do lostone o Romeo dolostone ■ Merrivale granite • Whinstone dolerite • Pias Gwilym limestone • Shap granite
•Ό
.£ 20
.i
I 15 ω c
CL
10
l·-
5 h
20
40
60
80
100
120
Critical energy release rate (N rrf')
Figure 7 TBM and linear cutter data for penetration index versus critical energy release rate Table 6 Specific Energy of Cutting Data Location
From Figure 4:
Specific energy of cutting ( k W h m - 3 ) Chicago Buffalo Seabrook Basalt
From Boyd [in 19]: Granite, gneiss, schist, quartzite Harder sedimentary rock Softer sedimentary rock
10.5-12.0 9.2 18.9 12.2 25.2 16.8 11.2
systematic method available in the public domain, and includes all desirable aspects of TBM design and operation, including thrust, torque, rotation rate, cutterhead profile (domed orflat),disc spacing and diameter, and disc bluntness. Intact rock tests required in the methodology include three specialized tests for abrasivity (AV or Abrasion Value), brittleness (S20 from the Swedish Brittleness test) and drillability (the Sievers J-value). Derived rock parameters include the Drilling Rate Index (DRI) and Cutter Life Index (CLI). The thrust versus 'basic' penetration relationship is nonlinear, and the concept of 'critical thrust' is incorporated as a normalizing parameter. Various factors are offered to modify the calculated penetration rate for differences in cutter diameter and kerf spacing, and similar factors are used in rolling force and 'torque demand' calculations. Correlations have also been developed between the results of NTH tests and other tests, including UCS. Blindheim et al. [38] reported on use of the NTH method to predict TBM performance for a tunnel to be excavated in basalt. The correlation based on DRI values indicated lower penetration rates than were expected, primarily because the crushed rock used in the brittleness test preferentially excluded the softer zones of the amygdaloidal basalt. This resulted in a low DRI and an underestimated boreability. This example illustrates the recommendation that tests not only be performed to generate numbers, but that the process of testing should be observed carefully to understand the meaning of the results. Linear cutter tests results have been used in machine performance prediction. For example, the Colorado School of Mines (CSM) has developed index indentation and linear cutter test correlations with full scale TBM performance. CSM used these equations to predict TBM performance, using field values of operating thrust, torque, cutter type and spacing [6]. The predictions held up well except for one project excavated in blocky and jointed rock. When applied to a TBM project in Chicago, using the same laboratory discs as were used in the field, the normal and rolling disc forces were very similar for linear cutting tests and field trials. The normal forces predicted by indentation
TBM Performance Analysis with Reference to Rock Properties
279
test equations were close, and rolling forces agreed in trend but predicted values were lower than experienced. Linear cutter testing has also been used for performance prediction in basalt [38], with the results compared to predictions from the NTH method. A match of disc cutter tip width and diameter between the field and linear cutter testing was considered important for accurate predictions of both forces and penetration. An additional complication for direct application to TBM performance, however, was noted to be the reduced cutting efficiency expected in the field because of inexact disc groove tracking (indexing) and cutter wear. Most of the theoretically derived equations expect single pass cutting, and multiple pass cutting is a complication on prediction that can occur in soft or hard rock. In soft rock, TBMs with typical kerf spacings (about 70 mm) can achieve very high penetration rates, but with a spacing/penetration (s/p) ratio much less than optimum. In chalk, Prev values of 10-15 mm are possible, and it might be expected that multiple chipping per pass might occur. By observation, however, the widths of many chips are about equal to the kerf spacing, but the chips are much thicker than the penetration, clearly indicating multiple pass cutting. Multiple pass cutting also may occur in hard rock where thrust is insufficient for adequate penetration and single pass cutting.
10.4.3 Rock Mass Characteristics Impact Rock mass characteristics impact penetration rate in several ways. As a generalization, if a mixed face of variable rock strength is present at the heading, the penetration rate is more typical of the stronger rock. If ground condition deterioration (as by geological structure or weathering) is severe, TBM thrust and torque may be reduced to promote face stability. For better ground, penetration rate will increase as more discontinuities are present at the face. In addition, the more parallelism between the discontinuity orientation and the rock face, the greater will be the penetration rate enhancement. These comments should be used to guide site investigation efforts. For the case of flat lying sedimentary rock and vertical exploratory core, a measure such as RQD clearly supplies little guidance as to the frequency of discontinuities important for penetration rate prediction, those which can be exploited in the process of chip formation. The same comments are generally true of rock fabric anisotropy. Sanio [39] demonstrated that, with schistosity oriented parallel to the tunnel face, Frev was six times greater than with schistosity oriented parallel to the tunnel axis. Wanner [40] found that penetration rates when fissility or foliation was parallel to the face were twice that for perpendicular to face. Banding in gneiss acted similarly to foliation in improving penetration rate, excluding those rocks toughened by hornblende crystal growth which cut across foliation. Fissility and schistosity effects may be included implicitly in intact rock prediction methods by controlling rock specimen orientation during laboratory testing. As far as larger scale rock mass effects, Aeberli and Wanner [11, 41] discussed the impact of rock mass effects in terms of 'efficiency' of discontinuity planes. Tensile joints were least efficient, particularly if healed by mineralization. Shear fractures, with rock damage zones extended into adjacent intact rock, were more efficient. Aeberli and Wanner also made a careful study of Rf variations during mining through several different rock masses. Table 7 includes data for one tunnel in sedimentary rock. Overall, Rf could be reduced by about 50%, but the precise variation depended upon lithology, strata sequence, stresses present and discontinuity orientation. Several investigators have noted that joint frequency can double or triple the penetration rate when joints approach cutter spacing. For TBM operation in dolerite, Korbin [2] reported that Pr increased from 1.3 to 3.9m h" 1 when the joint spacing decreased from greater than 1.0 m to less Table 7
Effect of Joint Spacing on Field Penetration Index [40] Field Penetration Index (kN mm * )
Rock type
<0.05 Sandy limestone Coarsegrained limestone Finegrained limestone
Joint spacing (m) 0.05-0.1 0.1-0.5
0.5-1.0
11
17
28 18
39 22
20
23
26
28
280
Mechanized Excavation
than 0.1 m. This effect may be understood on the basis of Rittinger's relation [2, 34], which may be expressed as Energy required = K{{DS)~' - (D{)~'}
(8)
where K is the 'effective surface energy', Ds is the product (muck) size, and D{ is the feed size or general spacing of discontinuities. Korbin [2] used the D50 size from a muck gradation as a measure of product size and noted that, to see a significant effect of reduced energy or increased P r , D{ needs to be less than aboutfivetimes Ds. In many cases, this means that Df should be roughly equivalent to the disc cutter spacing. Several investigators have developed indices to quantify the varying rock mass conditions. Aeberli and Wanner [41] used the ratio of the area of discontinuities per unit volume to track the influence of joints on P rev . Although no quantitative relationship could be suggested, a ratio increase resulted in a Prev increase at decreased F n , and that the effect was more significant for stronger rock. Eusebio et al [42] discussed the application of a 'Ground Difficulty Index' (GDI) [43] classification scheme to understand TBM performance during a long tunnel drive in highly variable rock. During mining, they recorded rock mass RQD and RMR classifications, and performed in situ Schmidt hammer tests to evaluate intact rock strength variability. From the starting point of a 'basic' penetration rate derived from case histories or from correlation with rock properties such as UCS, they suggest an empirical GDI correlation between RMR classes and PR, a correlation which includes the effect of Fn on penetration. A similar approach has been taken using the RSR classification scheme [44]. The NTH [18] method of performance prediction includes a 'fracturing factor' which is introduced in the Prev calculation to include rock mass effects. Discontinuities are described by type (continuous, noncontinuous and nonfractured rock), by filling (filling strength or healed by mineralization) and by spacing. This information is used to classify the rock intofissureor fracture classes, which are combined with information on orientation to produce the fracturing factor. This factor effectively modifies the thrust versus penetration relationship for a given intact rock, such that the more fractured a rock mass is, the higher the Prev achieved for a given Fn. This factor is also used in torque calculations since, in fractured rock, torque demand increases with increased penetration. As a final comment on rock mass effects, in situ stresses which are high relative to rock strength can promote stress slabbing at the face. At typical mining rates, this response may develop fast enough to result in an increased Prev if the rock is not too overstressed or susceptible to bursting. However, face deterioration and overbreak may develop which must be controlled with shielding, or cutterhead modifications as false facing in severe cases of unanticipated conditions. In fact, the TBM operator usually decreases F n and cutterhead rotation rate to improve face stability. To summarize, if ground support requirements are not changed significantly, a P r increase can be expected with increased jointing present in a rock mass. Such an effect is most important to consider in very strong rock for which modest increases in P r can significantly improve the economics of a project. In many rock masses, any potential P r improvement is usually conservatively ignored, in anticipation that the impact of any ground instability will dominate the performance response. The geotechnical engineering profession clearly does not have a recommended method for quantitative estimation of the effects of rock mass variations on P r . With more tunnel projects being supported in single pass lining operations, the time is closing on opportunities to access in situ rock mass exposures. The effort to gather additional rock mass information must be made in the immediate future, before the opportunity to describe the rock is lost. 10.5 ROCK PROPERTY IMPACT ON CUTTING TOOLS 10.5.1 Cutting Tool Failure Cutter failure may occur for many reasons, including excessive abrasive wear of the disc ring, brittle failure or chipping of the disc ring, splitting or loss offitbetween the replaceable disc ring and hub, and disc bearing failure or 'blocked' cutters. In blocky ground, rock blocks may fall out from the face and can become wedged between cutters, damaging the discs and hubs. The rate of abrasive wear depends on characteristics of the rock, and also on cutter forces, amount of muck accumulation in the invert and cutterhead design. If disc loads were uniform across a cutterhead, even rates of wear might be expected. However, loading is far from uniform as the rock is not homogeneous across the face, and cutters are not replaced simultaneously everywhere so that differences in disc diameter are present at any time. Recent projects most often have used the constant section disc edge, which promotes more even cutterhead loading and cutter loads. Linked
TBM Performance Analysis with Reference to Rock Properties
281
with abrasive wear, the high disc rolling velocities and loads produce heat at the disc edge which softens the ring, perhaps leading to plastic deformation or 'mushrooming' of the steel [45]. High temperatures can certainly exist at the disc edge, evidenced by the observation that muck transported on most projects is typically at higher than ambient temperature. Brittle failure or chipping of the ring steel is a problem when a cutter is highly loaded and, particularly, with impact loading from mixed face conditions or when chip formation results in rapid release of high disc loads. Recent attention to this problem by machine manufacturers has resulted in application of new high toughness harder steels for disc ring material. Where worn hubs have been used, new disc rings do not fit to the close tolerances intended. In relatively nonabrasive rock, 6-10 discs can be refitted on each hub before repair is necessary. However, in abrasive sandstone, one project reported a rate of only 3-4 discs per hub [46]. Because of abrasive conditions, carbide button cutters were tried on this project - but the steel matrix was abraded quickly and the buttons were plucked or chipped with overall disastrous consequences. Disc bearing failures or 'blocked' cutters tend to occur at high loads and high cutterhead rotation rates, with a resulting elevation in bearing temperature. High temperature ages and thins the oil, deteriorates the metal seals permitting abrasive material entry, and affects close dimension tolerances and preloading assumed in design and operation [45]. Cutter replacements rates are usually described in terms of cutters per linear tunnel distance, cutters per volume of rock excavated, machine operating hours per cutter or rolling disc life before replacement. The cutter rolling distance is calculated as the circumference of the cutter travel path multiplied by the number of times the cutterhead was rotated after installation of a new disc. The number of rotations can be estimated as the machine clock time since installation multiplied by the cutterhead rotation rate. Disc replacement rates vary across the cutterhead, with low rolling distance life associated with center cutter positions, where tight turning and scuffing reduce bearing life, and vibrations can cause particularly high rates of abrasive wear. For relatively nonabrasive rock, rolling distance life for cutters in gauge and face positions are comparable. However, gauge replacement rates are higher in terms of TBM operating time because the travel path is longer and the cutters 'wash' through muck accumulations. Gauge cutter rolling distance life is notably reduced in highly abrasive rock mining.
10.5.2 Rock Property Impact on Failure Rates Most project information on cutter failure rates and costs is proprietary, and few studies of cutter replacement rates have been reported. Information on the reason for disc cutter replacement is even more difficult to find in the literature. With such a limited database on failure occurrences, very little has been published on quantitative estimation of rock property impacts. Most attention has focused on rates of abrasive wear. In general, the more crushing and the more abrasive the rock powder, the higher the rate of abrasive wear. Heat makes the muck more abrasive, moisture decreases the abrasivity. Poor chipping means that much of the disc is in contact with the rock and crushed powder, resulting in higher rates of wear at the disc sides. For strong rock, resistant to penetration, abrasion of the disc point occurs. The relationship between rock properties and abrasive wear rates can be understood by considering data from one tunnel project [25]. High replacement rates of the 394 mm diameter cutters prompted the contractor to make regular measurements of changes in disc diameters. In highly abrasive porous sandstone, for which rock chips were small and much of the muck was fine material, about 25 mm of disc diameter decrease occurred after about 18 hours of operation, corresponding to a rolling distance of about 90 km. Cutters on this job were typically changed after the diameter had been decreased by about 65 mm. For less abrasive limestone, a comparable 25 mm diameter decrease occurred after about 300 km rolling distance, and in nonabrasive shales, after about 1000 km. With these measurements, it is possible to demonstrate the effect of disc wear and blunting on penetration rate. TBM performance was unaffected until the abrasive wear had caused about a 40 mm decrease in average disc diameter. For additional amounts of wear, similar penetration rates were only achieved with about 10% higher average cutter thrust loading. If thrust was not increased, the penetration rate achieved was reduced by 15 to 25%. Various empirical correlations between abrasive wear rates or cutter costs and rock properties have been developed. Bamford [12] suggested a multilinear regression equation to predict rates of wear (cubic meters excavated per cutter consumed) as a function of rock properties including Scleroscope hardness, shear strength, tensile strength, the Goodrich wear number and Taber
282
Mechanized Excavation Table 8 Cutter Failure Rate Data from TBMs in Sedimentary Rock [25] Rock type
Sandstone Bedded shale/limestone Weaker dolomitic limestone Stronger dolomitic limestone
Average cutter rolling life (km)
Average TBM operating time before cutter replacement at each position
Number of cutter replacements per linear 100 m mined tunnel
200-300 600
70-90 TBM hours 175 TBM hours
0.75 0.03
1500-2500
400-450 TBM hours
0.08
1000-1500
350-400 TBM hours
0.15
Abrasiveness. The r2 value associated with this correlation was 0.61. Tarkoy [16] has published a correlation between Total Hardness and cutter costs. The CERCHAR test has been used for wear prediction, but little information on relationship details is available [15]. Nelson et al [49] studied cutter replacement rates for six tunnel projects in sedimentary rock units, with 394 mm diameter cutters with 90° wedge cross section rather than the constant section discs in predominant current use. The rates of failure correlated closely with the Rock Abrasiveness value determined in the Taber Abrasion Hardness testing. The replacement frequency data are summarized in Table 8. The NTH methodology includes an approach to estimate cutter replacement rates [18]. The prediction is based on the Cutter Life Index (CLI), based on the Abrasion Value (determined for steel rings) and the Siever's J-value (a drillability test). The CLI is modified by various derived parameters to include the influence of disc diameter, disc spacing, cutterhead rotation rate, TBM diameter and cutterhead profile (domed or flat). In addition, a factor is suggested to correct for the presence of certain minerals in the rock which are difficult to crush and therefore are particularly effective at causing abrasive wear. The identified minerals are quartz, mica and amphiboles (as hornblende), and the Vickers Hardness number of each mineral is used together with volume percent to evaluate the mineral content correction. The CLI and other factors are used to determine the average cutter disc life, in terms of TBM hours per cutter. By the NTH data base, the reference 394 mm diameter rolling distance life varies from 200 to 1000 km for highly abrasive rock, and up to 5000-10000 km for nonabrasive rock. From this reference, cutter life is reduced by 30% for 356 mm diameter cutters, and increased by 50-65% for 432 mm diameter cutters. Cutters on domed cutterheads have 10% longer life than on flat cutterheads, and constant section cutters last 10-15% longer than do wedge section cutters with similar amounts of steel in the disc rings. Mining around tight curves reduces cutter life by about 75%. Published correlation have not, to date, included consideration of the different disc edge materials available (and evolving) for contractor selection. In order to develop an appreciation of the decision process in 'engineered' selection of disc cutters, the following case studies are noted. Deering [30] reported on a project in Norway where 483 mm discs were used on a 3.5 m diameter TBM. The project was started with 14.3 mm tip width constant section discs but, in the hardest rock with average gross normal force per cutter to over 260 kN, disc chipping became a problem. The contractor switched to 19 mm tip width cutter discs of harder steel, and the amount of chipping decreased significantly. An additional concern on this project was related to the high abrasive wear rates and numerous bearing failures noted for cutters in outer face (transition to gauge) positions. Transition cutter loads were found to be 28% higher than at other positions, and the contractor decided to compensate by adding cutters, reducing transition cutter spacing on future projects. Jordal and Hartwig [47] reported on a TBM tunnel excavated in hard quartzite, for which cutter wear was a particular concern at gauge positions. The cutter change frequency with normal disc material averaged 0.58 cutters per meter of tunnel (to a maximum of 1.89 cutters per meter). For discs made of a harder tool steel, the change rate decreased to 0.14 cutters per meter of tunnel. Davey [48] reported on TBM excavation in diorite with a relatively low quartz content, and with 0.2-0.5 m joint spacings. The penetration rate averaged 2.60mh" S with thrust and torque at 85% of installed maximum and 19 mm tip width constant section cutters. In more massive granodiorite (joint spacing to 1 m), the average penetration rate was 1.77 m h - 1 . In response to the reduced Pr, the contractor opted to keep the same discs but to increase thrust. It is possible that Pr could have been increased by using thinner tip widths to increase the contact stress, but only at a penalty of more replacements for abrasive wear since there would be less material on each disc.
TBM Performance Analysis with Reference to Rock Properties
283
In summary, the largest public domain database for abrasive wear rate prediction can be accessed through the NTH method, but specific rock tests must be performed which require special equipment. Other tests to be considered for application include the CERCHAR test and the Taber abrasion test, both of which also require specialized equipment, and which have been demonstrated to produce reasonable estimates of relative rates of abrasive wear. Little information ij available about rates for other types of cutter failure.
10.6 ROCK PROPERTY IMPACT ON UTILIZATION 10.6.1 Management and Downtime Improvement in utilization is the major task of an effective management system for underground construction projects. For most projects, the penetration rate achievable is fully adequate to meet desired advance rates, and TBM performance is not limited by the ability to cut rock. Instead, activities associated with muck transport and support installation systems often occupy increasing percentages of shift time. It is important to have on site management during construction, to develop concepts for alternative actions which may be required, to keep daily construction records and to perform routine scheduled maintenance [16]. In 1984, the US National Committee on Tunneling Technology (USNCTT) completed a study of the needs of geotechnical investigations for underground projects [50]. A key point in managing geotechnical impacts was for the contractor to have the right specialists on site. Most contractors have 'good' and 'bad' ground specialists, who develop different management styles and make different judgements on when to slow down or speed up, and when to change construction procedure. Clearly, if a contractor is prepared for the possibility for bad ground, he can handle it more efficiently. Therefore, the contractor and owner rely on the geotechnical engineers and their understanding of anticipated conditions. If the visualization of the ground response is wrong, it will be serendipitous if appropriate equipment were selected. Parkes [51] included 87 tunnel projects in a CIRIA (Construction Industry Research and Information Association) study. He concluded that successful projects included appropriate equipment and management which developed alternative or 'fall back' thinking. In variable ground, flexible equipment is required. For projects judged unsuccessful, reasons cited include: low advance rate, most often in variable ground; poor TBM function in the ground encountered, indicating poor visualization of ground conditions; inadequate mucking system capacity; and poor equipment choices and maintenance, especially on back-up system components. As an example of an effective management scheme, consider one project excavated in hard rock [30]. Site management initially was not concerned about operating the TBM with very worn cutters. However, with average cutter loads very high in the tough rock, the higher loads required for cutting with worn discs resulted in higher tool temperatures and bearing and seal failures. Newly replaced cutters were high profile, highly loaded and often failed shortly after replacement. A plan was developed which included regular inspection and establishing criteria for replacement which included a maximum wear limit, maximum wear difference between adjacent cutters, and a rule that new bearing hubs would be installed where load were highest, with older bearings relegated to other positions. With implementation of the plan, the results included reduced rate of abrasive wear, reduced frequency of catastrophic bearing failures (blocked cutters), reduced cutter change downtime because all such activities are scheduled and minimized cutter costs. 10.6.2 Impact of Cutting Tools If the expected number of cutter replacements can be established for a project, the downtime required for these changes can be estimated using case history information. Cutter change downtime data, from case histories in sedimentary rock, are summarized in Table 9. This information indicates that about 1.5 hours are required for a solitary cutter change [25]. This agrees with the experience of others [29] and, if several cutters are changed at one time, less time is required per cutter. Note that, for Project 4 in Table 9, the high downtime is closely correlated with large ground water inflows, which made cutter change activities time consuming. Cutter change downtime can also be expressed on the basis of available shift time. Nelson et al [52] report on cutter change downtime for six projects. For five of the projects in relatively nonabrasive rock, the downtime was fairly low, averaging about 2.9% (ranging from 1.7 to 5.5%).
284
Mechanized Excavation Table 9
Cutter Change Downtime on Four Projects [25]
Project
1 2 3 4
Number of cutters changed
Downtime hours per cutter
341 782 29 39
1.5 1.4 1.8 2.3
For one project in highly abrasive rock, however, cutter changes required 16.5% of all shift time clearly a significant economic impact. The CIRIA database [51] generally agrees with this information. Of the 10 projects with downtime distributions presented in the CIRIA study, the seven in nonabrasive rock had overall cutter change downtime averaging 3.5% (ranging from 2.3 to 7.0%). For the three in more abrasive rock, the average downtime was 16.5% (ranging from 15.8 to 17.1%). McFeat-Smith and Tarkoy [53] summarized project data in a different form, with downtime for cutter change recorded as hours required per meter of excavation. For nonabrasive rock, cutter change downtime was 0.02-0.05 h m" 1 . For more abrasive rock, downtime was increased to 0.1-0.2 h m" 1 . The recent trend towards larger disc diameters means heavier cutters requiring more access space in front of the cutterhead. TBMs which use larger cutters usually have equipment installed to facilitate cutter transport and installation. Such equipment will reduce the average change time, perhaps down to 30 minutes per cutter [30]. In addition, a new wedge lock housing has been developed which makes cutter changes much easier and which has proven to be very durable. Other improvements include the rear-access cutters which do not require access to the front of the cutterhead for replacement. In cases of face instability, these cutters greatly improve safety. However, rear-mount cutters are more expensive and take more time to replace. 10.6.3 Impact of Geotechnical Factors Good overviews of the importance of geotechnical factors in TBM utilization are available in recent literature [16, 50]. In weak rock, mucking and ground support are major downtime sources. In very strong rock, equipment wear at high loads and cutter wear are often the major downtime sources. In either case, correct appreciation of the problem or limitation before the equipment is ordered goes a long way to minimizing the geotechnical impacts. A brief summary of geotechnical impacts would include the following: (i) installing support under difficult conditions; (ii) cutters and cutterhead damage from face fallout; (iii) face fallout blocks get dragged around causing overbreak; (iv) muck jams along the conveyors and in muck buckets; (v) problems with removal of roof fallout muck from invert; (vi) blocky/seamy ground and sidewall fallout which affects the gripper bearing areas; (vii) ground loads on the TBM shield, with thrust and torque losses, steering difficulty, and delayed access for support installation; and (viii) unstable invert, giving track problems (derailments) and problems with steering. The biggest impact overall, however, is the change in support requirements - the break in routine and loss in efficiency. Case studies have been reported in the literature which provide guidance on estimation of ground condition related downtime. Table 10 includes time required to install steel sets as temporary support for sedimentary rock tunnels driven by unshielded TBMs [25]. These sets were installed for ground control, and a trend of less unit downtime for more sets at one location is notable. In another tunnel in sedimentary rock, mining delay to support blocky weathered ground amounted to an average of about 2.6 hours per linear meter of tunnel. This study included over 50 km of tunnel [25], and it was very typical that any potential penetration rate improvement possible in highly jointed rock was more than offset by delays associated with ground support, rock jams and grippers cribbing. During excavation at Kielder, the Robbins TBM in blocky mudstone developed support, gripping and mucking problems [2]. The ground support downtime information is summarized in Table 11. In faulted material, both Robbins and Demag TBMs encountered significant time loss in ground support. Water made the situation worse. In addition, steering in the bad ground zones was difficult
285
TBM Performance Analysis with Reference to Rock Properties Table 10 Downtime to Install Steel Set Temporary Support on Five Projects [25] Average number of steel sets installed per location 37 7 8 19 15
Average downtime hours required per installed set/lagging 1.0 3.8 3.9 0.9 1.5
Table 11 Ground Support Downtime at Kielder [2] Ground conditions
TBM and support
Blocky mudstone
Robbins-steel sets
Fault zone
Demag-bolts/mesh Robbins Demag
Ground water conditions
Time (per m) to support
Support time as %of shift time
Dry Water present Dry
1.08 h 2.02h 0.66 h 3.0 h 5.0 h
57 50 38 76 80
for the Robbins TBM, and 300 hours were required to reestablish grade. Associated delays occurred at interface transition mining between soft and harder material. When higher thrust was required to mine harder material, while the gripper locations were still in softer ground, downtime to crib the grippers was 3.7 to 4.1 h m" 1 . On a whole project basis, ground condition downtime for bolts or steel sets, gripper cribbing, crown scaling and removing muck jams on six TBM projects accounted for an average 12.6% of project shift time (range 6.3-17.8%) [25]. In the CIRIA study, similar averages for 10 TBM projects were 15.4% (range 8.8-19.9%) [51]. Utilization reductions have also been documented for other projects afflicted with severe and/or unanticipated poor ground conditions [e.g. 2, 22, 24]. It is clear that the impact of ground conditions depends greatly on the equipment employed. An interesting case in point is a project in Colombia [46]. No significant water problems were expected, and the sandstone encountered was expected to be quartz-rich (90%), but competent. Much concern was given to the presence of potentially squeezing ground in overstressed shales, and a telescoping double shield with many innovations was designed for the 9 km tunnel construction. Significant squeeze never developed. However, the sandstone was not well cemented, and was encountered as a friable abrasive ground mass which developed a flowing behavior in locally uncemented and saturated zones. Ground water inflow rates were typically 1200-2000 L min" *, and 20000 L min" l in one fault zone (with a 45 day delay). The porous sandstone did not chip well, and the resulting muck was more of an abrasive slurry which caused damage to all operating equipment. In clay/shale sections, the material mined as cohesive muck with abrasive sand, completely wearing out the TBM false face. The contractor attempted to take cores of the 'rock' after mining, but for 18 of 26 attempts only cohesionless material was retrieved. Shotcrete, bolts and arches were expected to be adequate ground support. In fact, the unstable face and walls made sidewall gripping impossible. The auxiliary thrust system, however, had to react against an expanded segmental lining. As overbreak developed, lining ring completion was complicated and much time was expended to grout/backpack the segments so the auxiliary thrust system could function. The USNCTT [50] studied geotechnical impacts on 87 projects in the US. Major impacts are summarized in Table 12, which also incorporates comments on ground impacts from other references [e.g. 54]. Included in this table is the 'Impact rating', based on how many of the problems resulted in claims. Note that the Impact rating does not gauge the duration, severity and costs in terms of construction and long-term operational impact. Major impacts most often related to: (i) stand up time not adequate for the chosen method of construction; (ii) overstress impacts - stress relative to available strength, causes face and wall instability, safety hazards and often occurs with a time-dependency difficult to establish ahead of time; and (iii) not anticipating ground water inflows.
286
Mechanized Excavation Table 12 Major Geotechnical Impacts on TBM Excavation [after 50, 54]
Major problem areas*
Impact rating*
Consequences/requirements
Loosening loads, blocky/slabby rock, overbreak, cave ins
4.2
Ground water inflow
1.8
Squeezing ground (shield stalled by pressures)
4.2 (10)
Ground gas/noxious or hazardous fluids/wastes Overstress, spalls, bursts
5.0 6.6
Hard, abrasive rock
4.0
Soft zones in rock mass, weathered soil-like zones (ifflowingground)
5.0
Soft, weak rock at invert
10.0
Time loss, immediate and additional support (perhaps grouting, hand mining), special equipment (perhaps machine modifications), gripper anchorage and steering difficulty, shut down in extreme cases of face and crown instability Lowflow/lowpressure - operating nuisance, slow down, pumping capability High flow and/or high pressure - progress slow or shut down, special procedures for support and water/wet muck handling Corrosive or high-salt water - treatment may be required before disposal, equipment damage, concrete reactivity, problems during facility operation (equipment modifications (as waterproofing) may be required if inflow unanticipated) Immediate support, equipment modifications, time loss, if invert heave and train mucking - track repair and derail downtime Safety concerns, time loss (perhaps project shut down), special equipment modifications, disposal problems Immediate support, perhaps progress shut down, safety, special procedures High cutter wear and cutterhead damage (esp. if jointed/fractured), slowed progress, may damage machine, cutterhead fatigue Slowed progress, immediate and additional support, potential for ground water inflow, muck transport problems, steering difficulty, weathering particularly important in argillaeous rock, if mixed-hardness rock - may have high cutter costs. Slowed progress, grade and alignment - steering problems
(8)
a Listed in ranked order according to frequency of occurrence. bImpact rating = [(incidence of claims)/(number of occurrences)] x 10.
The severity of each problem depended on the particular construction method and interactions between factors, and problem anticipation is important since equipment modification and repair is very costly and time-consuming underground. Although it is possible to envision direct application of rock mass classifications in utilization estimation, little has been published. Garrett [22] summarizes a rock mass descriptive classification used in one project (Koralpe), and reports on the advance rate achieved in each ground class. If more information were available, it might be possible to separate penetration rate and utilization reduction effects, and relate each to a rock mass classification. Such a goal is well worth additional efforts by the research and professional communities. 10.6.4 Other Impacts Ground water is a very common difficulty encountered in underground construction. A brief review of many case histories indicates that nuisance level inflows are associated with downtimes of the order of 3-5% of all shift time. Major inflows and high water pressures cause much more downtime. Direct water problems can be compounded by piped removal of jointfillingswhich can cause instability and loosening loads, problems with handling very wet muck and drainage into the tunnel which may promote consolidation of overlying soils. If a contractor is aware of a potential for significant volumes of ground water inflow, certain modifications can be made on a TBM to improve its water tolerance. The main bearing seal can be pressurized, and all electronics can be waterproofed. Cutterhead buckets can be designed to handle sloppy muck, and low angle or trough conveyors can be incorporated to retain and transport water. In addition, equipment for advance probing and consolidation grouting can be installed, and adequate pumping and disposal capacities can be provided [16]. If ground water is of unacceptable quality, a contractor may have difficulty in disposal and special treatments may be costly. In urban and coastal environments, ground water may well be contami-
TBM Performance Analysis with Reference to Rock Properties
287
nated by hazardous chemicals or gasoline. In addition to the problems of safe disposal, the construction environment becomes difficult, particularly if construction is contemplated under compressed air. Gassy ground nearly always has a significant impact on underground construction, particularly if it is unanticipated. The basic techniques to construct in gassy environments include sober attention to safety training and gas monitoring equipment with a reliable automatic shut off, using explosion proof equipment and safe hydraulicfluids,keeping turbulent airflowand adequate ventilation at the heading and elsewhere, and advance probing and predraining as required. If high ground temperatures are expected, the contractor must be apprised so that adequate ventilation, cooling and water heat exchange capacities can be provided. 10.7 THE FUTURE 10.7.1 TBM System Performance Prediction Prediction methods for TBM performance prediction have been under development for the past 25 years. Most of the methods are strongly database oriented, and resulting predictions tend to be biased towards the specific projects and contracting environments in place of database cases. Some methods are well used for actual performance prediction in absolute terms, and others are best used in comparative or relative application as feasibility evaluations. Some methods consider penetration rate prediction separately from utilization, and combine these two predictions to produce a predicted advance rate from which project schedule and costs can be evaluated. Other prediction methods are based on an input advance rate, determined by judgment and previous experience. Key to the use of most predictive models is the subdivision of a tunnel alignment into zones of geotechnical and/or construction significance. Criteria for subdivision can include intact rock property changes, rock mass changes, support requirement changes, variations in ground water inflow, project alignment changes or type of equipment to be used. Within each zone, the penetration rate and utilization may be more sensibly estimated, and the results combined to establish whole project expectations. Many prediction methods have been developed by individual contractors, machine manufacturers and consultants, and are considered proprietary. However, some models have been evolved in research efforts and are accessible in the public domain. Some of these models are briefly referred to below. Perhaps the first public domain model was developed by Wheby and Cikanek [55], based largely on tunneling experience in Chicago pre-1970. The program is called COSTUN and accepts rock RQD and compressive strength and ground water inflow rates as input parameters to estimate the project advance rate using a predictor equation. The advance rate can also be input directly if determined from other information sources. Tarkoy [17] has developed a microcomputer-based predictor model which has been generally described in the literature but is proprietary in use. The program is recommended for the following applications: (i) feasibility evaluations; (ii) to check on and update TBM performance in operation; (iii) to evaluate expected performance for different machines and cutters, assisting in equipment selection; and (iv) to provide guidance in comparing applications for new versus reconditioned equipment. Norwegian Institute of Technology researchers have developed their integrated case history based method into a computer program which is the most complete public domain model available to include all aspects of TBM design and performance considerations [18]. The Colorado School of Mines (Earth Mechanics Institute) has developed a microcomputer-based program called SIMTUN, which contains modules for cutterhead modeling (head shape, cutter arrays), mucking and back-up systems [56, 57]. SIMTUN also has an internal capability of providing estimation of penetration rate using one of three predictor equations. Output includes evaluation of TBM system capacities (calculations of power, torque and thrust), system component performance limit checks, material handling and back-up system impact on utilization, and overall project advance rate and cost estimation. SIMTUN is well suited for performance estimation and for cost and schedule sensitivity studies. Researchers at the Massachusetts Institute of Technology (MIT) [58, 59] have developed a computer code incorporating a probabilistic basis to consider uncertainty in geotechnical information. The model is geared towards optimizing design and exploration processes, and also includes a construction simulation module (SIMSUPER5). Geotechnical descriptive input includes rock type, RQD, degree of weathering and water inflow potential, and the model is used to create zones referred to as ground classes.
288
Mechanized Excavation
Additional work on this basic MIT model is currently ongoing as a partnership between MIT, the University of Texas, Austin and the University of California, Berkeley. The probabilistic basis for data input will be developed in conjunction with other modules designed to facilitate consideration of the impact of newly developed equipment components and opportunities for automation and real time system performance monitoring which should improve the reliability and flexibility of TBM excavation systems in the future. For application to scheduling, costing and machine selection and design, no public domain method is highly recommended for general application at this time. An engineer experienced in underground construction should always be involved in site characterization. The engineer should consider several alternative approaches, and access machine manufacturers and specialized consultants for testing appropriate to address particularly critical concerns.
10.7.2 The Future for Equipment Developments The evolution (or revolution) of mechanized excavation equipment developments will be directed to address improvements in both penetration rate and utilization. Penetration rate improvements are most likely aimed at extending TBM capabilities to be more effective in increasingly hard rock and in variable ground conditions. However, the biggest impact on overall system performance is likely to be attained with improvements in utilization, including system operating reliability and no delay ground support and single pass lining operations. The potential importance of utilization improvements is well documented by case studies in the literature. The CIRIA study [51] found that, for 51 projects with adequate information, the overall project utilization averaged 40% (range 22-71%). Projects with greater than 50% utilization almost all had consistent rock throughout the alignment, with little uncertainty and few support problems. The project utilizations for six sedimentary rock tunnels in another study [52] averaged 33% (range 29-36%). It is clear that the best way to improve TBM performance is to attack the typical low utilization. This means making equipment moreflexible,developing a management approach which is more responsive and investing in the ability to better anticipate what the ground conditions will be. Developments anticipated for the future include the following. The Cutterhead (i) Wider kerf spacing and excavation in stronger rock make higher cutter edge loads and faster cutterhead rotation rates desirable. Using water cooling on cutters [45] may keep tool temperatures down. Larger diameter cutters will permit larger bearings to accommodate the higher loads. (ii) Improved disc cutter array design on the cutterhead will reduce radial loads on the main bearing, increase penetration and reduce cutter wear. In particular, improvements in the lifespan of gauge and center cutters are expected. (iii) Metallurgical developments will permit longer cutter life by improving abrasion resistance of disc edge materials without sacrificing fatigue/impact loading resistance. (iv) Improved cutterhead stiffness and fatigue resistance will be required to maintain cutting efficiency under the load fluctuations and vibrations expected for high cutter loads and faster rotation rates. (v) Incorporating water jet assistance to extend cutter life and to improve TBM capabilities in harder rock without increasing cutter loads. Water jets can also reduce dust and sparking. (vi) Improved main bearing design (possibly hydrostatic bearings) to improve the quality of and ease of changing main bearings. This is especially important on long tunnel drives. Currently, main bearing replacement requires a minimum of four weeks shutdown, and involves over-excavation for access and pulling the machine from the face. (vii) Automated microprocessor control of TBM operation to vary thrust, torque and cutterhead rotation rate for optimization of tool forces and penetration rate in varying ground conditions. TBM Systems (i) Improved dust control systems, incorporating dust shield design which includes suction. (ii) Increased drive motor capacity will increase the power density and permit operation at less than 'red-line' current draw and may reduce maintenance requirements and improve reliability of operation [60].
TBM Performance Analysis with Reference to Rock Properties
289
(iii) Automated microprocessor control of TBM operation to include steering control, gripper reset, lining installation and equipment maintenance activities. (iv) Improved shield design will facilitate handling of highly stressed, swelling and squeezing ground conditions. Other System Components (i) Automated microprocessor control of mucking systems, including sensing to anticipate problems and minimize downtime for repairs. (ii) Developments in conveyor and pipeline systems for muck transport on a no delay basis. (iii) Ground support automation, including evaluation of support requirements with nonintrusive methods at the face during mining. Techniques for advance sensing of ground conditions will likely include radar and seismic methods. May also incorporate bolt and probe hole drilling performance data (thrust, torque, rpm) in support requirement evaluation and TBM operation optimization [19]. 10.7.3 Summary Comments In an ideal TBM application, the geotechnical exploration will be conducted to provide a rational basis for bid, avoiding overly conservative assessments. The design team includes those familiar with construction equipment and methods to obtain appropriate data and make good inferences, minimizing required contingency in the project bid. The contractor soberly assesses the information and his (or her) competitive bid includes selection of appropriately flexible equipment to maintain the advance rate required for project completion on schedule and in safety. The owner delegates a team of engineers to maintain and update the technical information database (including geotechnical information as encountered) during construction. The on site team maintains records and analyzes contractor operations during the project (including shift activity reports), compiling a final as built report which the owner appreciates as an unbiased record accessible for problem resolution. With project post mortems available for digestion, the perceived reliability of underground operations will be enhanced, improving the likelihood that underground space will be considered as a viable alternative for the needs of the future. The technology exists today for safe and efficient excavation of circular headings from 2 to 12 m in diameter, at an average advance rate between 30 and 50 m per 24 hour work day. TBM excavation systems have been used efficiently in rock as weak as 3 MPa uniaxial compressive strength, and in rock with strength in excess of 300 MPa. Technological developments will result in larger diameters possible, higher advance rates achievable and both very low and very high strength rocks mineable. Advance design concepts have been implemented to make TBM systems increasingly safe and versatile, with recent successful applications in previously considered difficult conditions such as extremely high strength abrasive rock and overstressed rock masses. Capital investment has been reduced by using reconditioned equipment, incorporating technological advances during the rebuild. The robustness of the basic design has permitted many TBMs to be reused on three to five projects, greatly improving the competitiveness of underground construction for many applications. The TBM of the future may well have evolved to the point that the rock under excavation is never seen but only remotely sensed, with workers never exposed to the hazards of the mined face, with the final lining installed during mining as extruded from the rear shield, with the muck processed and transported in containment. Machine/rock interaction is monitored continuously and mining rates are optimized for a maintained, if not maximized, advance rate.
10.8 REFERENCES 1. Innaurato N., Mancini R. and Pelizza S. Consideration of rock boring machines: analysis of Italian operations. In Proc. Tunnelling 76, pp. 219-223. (and discussions pp. 224-227.). Institution of Mining and Metallurgy, London (1976). 2. Korbin G. E. Factors influencing the performance of full face hard rock tunnel boring machines. US Dept. of Transportation, UMTA-CA-06-0122-79-1, available from National Technical Information Service, Number PB80112378(1979). 3. Snowdon R. A., Riley M. D., Temporal J. and Crabb G. E. The effect of hydraulic stiffness of tunnel boring machine performance. Int. J.Rock Mech. Min. Sei. & Geomech. Abstr. 20, 203-214 (1983). 4. Dowden P. B. and Cass D. T. Shielded TBMs - matching the machine to the job. In Proc. Rapid Excavation and Tunneling Conference, Seattle, WA, pp. 787-805. (1991).
290
Mechanized Excavation
5. Dollinger G. L. Hard rock tunnel boring: A summary of recent developments. In Proc. Rapid Excavation and Tunneling Conference, Chicago, IL, pp. 89-105. (1983). 6. Özdemir L. and Wang F. Mechanical tunnel boring, prediction, and machine design. Final report by CSM to NSF, NSF/RA-790161, available from National Technical Information Service under accession number PB80-101660, p. 204. (1979). 7. Dubugnon O. and Barendsen P. Small scale model testing - a new approach in TBM development. In Proc. Rapid Excavation and Tunneling Conference, New York, vol. 1, pp. 245-263. (1985). 8. Farmer I. W. and Glossop N. H. Mechanics of disc cutter penetration. Tunnels and Tunnelling 12 (6), 22-25 (1980). 9. Kutter H. K. and Sanio H. P. Comparative study of performance of new and worn disc cutters on a full face tunnelling machine. In Proc. Tunnelling "82, pp. 127-133. Institution of Mining and Metallurgy, London (1982). 10. Roxborough F. F. and Phillips H. R. Rock excavation by disc cutter. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 12, 361-366 (1975). 11. Wanner H. and Aeberli U. Tunnelling machine performance in jointed rock. In Proc. Int. Cong. ISRM, Montreaux, vol. 1, pp. 573-580. (1979). 12. Bamford W. E. Rock test indices are being successfully correlated with tunnel boring machine performance. In Proc. 5th Australian Tunnelling Conference, Melbourne, vol. 2, pp. 19-22. (1984). 13. Bamford W. E. Tests for assessing the drillability, cuttability and ripability of rocks are being internationally standardized. Excavation Characteristics Seminar, Melbourne, pp. 1-16. Australian Geomechanics Society (1984). 14. West G. A review of rock abrasiveness testing for tunnelling. In Proc. Int. Symp. Weak Rock, Tokyo, vol. 1, pp. 585-594. Balkema, Rotterdam (1981). 15. West G. Technical note: rock abrasiveness testing for tunnelling. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 26, 151-160 (1989). 16. Tarkoy P. J. Practical geotechnical and engineering properties for tunnel-boring machine performance analysis and prediction. Transportation Research Record 1087, Transportation Research Board, National Research Council, pp. 62-78. (1987). 17. Tarkoy P. J. Predicting raise and tunnel boring machine performance: state of the art. In Proc. Rapid Excavation and Tunneling Conference, Atlanta, GA, vol. 1, pp. 333-352. (1979). 18. Norwegian Institute of Technology. Hard rock tunnel boring. Project Report 1-88, Trondheim, Norway, p. 183 (1988). 19. Howarth D. F. Mechanical rock excavation - assessment of cuttability and boreability. In Proc. Rapid Excavation and Tunneling Conference, Los Angeles, CA, vol. 1, pp. 145-164. (1987). 20 Ouchterlony F. (coordinator, ISRM Commission on Testing Methods). Suggested methods for determining the fracture toughness of rock. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 25, 71-97 (1988). 21. Beckmann U. and Simons H. Tunnel-boring machine payment on basis of actual rock quality effect. In Proc. Tunnelling '82, pp. 261-264. Institution of Mining and Metallurgy, London (1982). 22. Garrett R. Koralpe Challenge. World Tunnelling 4, 215-218 (1991). 23. Ikeda K. and Nishimatsu Y. The effect of geotechnical properties on the productivity of tunnel boring machines. Report for Int. Tunnelling Assoc. Working Group Research, p. 14 (1980). 24. Mitani S., Iwai T. and Isahai H. Relations between conditions of rock mass and TBM's feasibility. In Proc. 6th Cong. ISRM, Montreal, vol. 1, pp. 701-704. (1987). 25. Nelson P. P. Tunnel boring machine performance in sedimentary rocks. Doctoral dissertation, Dept. of Civil Engineering, Cornell University, p. 448. Available from University Microfilms, Ann Arbor, Michigan, USA (1983). 26. Athorn M. L. Analysis of operational performance in the Selby Spine roadways. Department of Geotechnical Engineering Report, University of Newcastle upon Tyne (1985). 27. Samuel A. E. and Seow L. P. Disc force measurements on full face tunnel boring machines. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 21, 83-96 (1984). 28. Hingst H. T. TARP revisited. World Tunnelling 4, N13-N20 (1991). 29. Nord G., Persson P. and Prader D. European views on mechanical boring versus drill and blast tunneling. In Proc. Rapid Excavation and Tunneling Conference, Atlanta, GA, vol. 1, pp. 543-563. (1979). 30. Deering K., Dollinger G. L., Kranter D. and Roby J. A. Development and performance of large diameter cutters for use on high performance TBMs. In Proc. Rapid Excavation and Tunneling Conference, Seattle, WA, pp. 807-814. (1991). 31. Hemphill G. B. Mechanical excavation system: selection, design, and performance in weak rock. Ph.D. dissertation, University of Idaho, Moscow, p. 361 (1990). 32. Roxborough F. F. Research in mechanical rock excavation: progress and prospects. In Proc. Rapid Excavation and Tunneling Conference, New York, vol. 1, pp. 225-244. (1985). 33. Graham P. C. Rock exploration for machine manufacturers. In Proc. Symp. Exploration for Rock Engineering, Johannesburg, South Africa (Edited by Z. T. Bieniawski), vol. 1, pp. 173-180. Balkema, Rotterdam (1976). 34. Hughes H. M. The relative cuttability of coal measures rock. Mining Science and Technology 3, 95-109 (1986). 35. Howlett P. F. Practical applications of performance prediction algorithms for tunnelling projects. In Proc. Int. Congr. Progress and Innovation in Tunnelling, Toronto, vol. 1, pp. 107-110. (1989). 36. Nelson P. P., Ingraffea A. R. and O'Rourke T. D. Technical Note: TBM performance prediction using rock fracture parameters. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 11, 189-192 (1985). 37. Snowdon R. A., Riley M. D. and Temporal J. A study of disc cutting in selected British rocks. Int. J. Rock Mech. Min. Sei. ά Geomech. Abstr. 19, 107-121 (1982). 38. Blindheim O. T., Boniface A. and Richards L. A. Boreability assessments for the Lesotho highlands water project. Tunnels and Tunnelling 23, 55-58 (1991). 39. Sanio H. P. Prediction of the performances of disc cutters in anisotropic rock. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 11, 153-161 (1985). 40. Wanner H. On the influence of geologic conditions at the application of tunnel boring machines. Bull. Int. Assoc. Eng. Geol. 12, 21-28 (1975). 41. Aeberli U. and Wanner H. On the influence of geologic conditions at the application of tunnel boring machines. In Proc. 3rd Int. Congr. Int. Assoc. Eng. Geol, Madrid, section III, vol. 2, pp. 7-14. (1978). 42. Eusebio A., Grasso P., Mahtab A. and Innaurato N. Rock characterization for selection of a TBM for a railway tunnel near Genova, Italy. In Proc. Int. Symp. Mine Mechanization and Automation, Golden, CO (Edited by L. Özdemir, R. King and K. Hanna), vol. 1, pp. 4.25-4.35. CSM and U.S. Bur. Mines (1991).
TBM Performance Analysis with Reference to Rock Properties 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60.
291
Zaninetti A., Rondena E., Innaurato N. and Mancini R. Forecasting criteria in TBM work. In Proc. Int. Congr. Progress and Innovation in Tunnelling, Toronto, Canada, vol. II pp. 775-782. (1989). Cassinelli F., Cina S., Innaurato N., Mancini R. and Sampaolo A. Power consumption and metal wear in tunnel boring machines: analysis of tunnel-boring operations in hard rock. In Proc. Tunnelling '82 (Edited by M. J. Jones) pp. 73-81. Institution of Mining and Metallurgy, London (1982). Hartwig S. Internal water-cooling of cutters: the future ? In Proc. Int. Symp. on Mine Mechanization and Automation, Golden, CO, vol. 1, pp. 4.15-4.24. (1991). Grandori R., Doclini G. and Antonini F. The Rosales water tunnel in Bogota. In Proc. Rapid Excavation and Tunneling Conference, Seattle, WA, pp. 561-581. (1991). Jordal Y. and Hartwig S. The Ivar Project. In Proc. Rapid Excavation and Tunneling Conference, Seattle, WA, pp. 473-486. (1991). Davey G. M., Dickson K. R. and Gowring I. M. Kemano Power tunnel. In Proc. Rapid Excavation and Tunneling Conference, Seattle, WA, pp. 487-505. (1991). Nelson P. P., O'Rourke T. D and Kulhawy F. H. Cutter wear and its influence on tunnel boring machine performance. In Proc. Symp. Design and Performance of Underground Excavations, ISRM, Cambridge, pp. 239-246. (1984). US National Committee on Tunneling Technology, Geotechnical investigations for underground projects. National Academy Press, Washington, D. C. (2 volumes) (1984). Parkes D. B. The performance of tunnel-boring machines in rock. CI RIA Special Publication No 62, p. 56. (1988). Nelson P. P., O'Rourke T. D. and Glaser S. D. TBM system downtime - causes, frequency and duration on six tunnel projects. In Proc. Rapid Excavation and Tunneling Conference, New York, vol. 2, pp. 751-770. (1985). McFeat-Smith I. and Tarkoy P. J. Assessment of tunnel boring machine performance. Tunnels and Tunnelling 11,33-37 (1979). McFeat-Smith I. and Tarkoy P. J. Site investigations for machine tunneling contracts. Tunnels and Tunnelling 12, 36-39 (1980). Wheby F. T. and Cikanek E. M. A computer program for estimating costs of tunneling. In Proc. Rapid Excavation and Tunneling Conference, San Francisco, vol. 1, pp. 185-206. (1974). Sharp W. R., Kennedy E. R. and Little W. E. Estimating costs using an interactive computer model. In Proc. Rapid Excavation and Tunneling Conference, Chicago, vol. 2, pp. 1079-1094. (1983). Sharp W. and Özdemir L. Computer modelling for TBM performance prediction and optimization. In Proc. Int. Symp. Mine Mechanization and Automation, Golden, CO, vol. 1, pp. 4.57-4.66. CSM and U.S. Bur. Mines (1991). Ashley D. B., Veneziano D., Einstein H. H. and Chan M. H. Geological prediction and updating in tunneling - a probabilistic approach. In Proc. U.S. Symp. Rock Mech. pp. 361-366. (1981). Einstein H. H., Salazar G. F., Kim Y. W. and Ioannou P. G. Computer-based decision support systems for underground construction. In Proc. Rapid Excavation and Tunneling Conference, New Orleans, LA, vol. 2, pp. 1278-1308. (1987). Peach A. J. Remanufactured TBM's. World Tunnelling 1, 155-157 (1988).
11 The Effects of Rock Properties on the Economics of Full Face TBMs DAVID F. FAWCETT Babtie Group, Maidstone, UK 11.1
INTRODUCTION
293
11.2 THE ECONOMICS OF TBM TUNNEL DRIVING 11.2.1 General Approach 11.2.2 Plant Costs 11.2.2.1 Basic details 11.2.2.2 Depreciation allowances 11.2.2.3 Other plant cost headings 11.2.2.4 Utilization, availability and power consumption 11.2.2.5 Equipment cost implications 11.2.3 Labor Costs 11.2.3.1 Labor basis 11.2.3.2 Labor required 11.2.3.3 Labor numbers and duration requirements 11.2.3.4 Labor rates 11.2.3.5 Summary of labor costs 11.2.4 Materials Costs 11.2.4.1 Materials costs details 11.2.4.2 Implications of materials costs 11.2.5 Staff Costs 11.2.6 Cost Summary 11.2.6.1 Project basis 11.2.6.2 Implications of cost headings 11.3 COST/PERFORMANCE CRITERIA RELATING TO ROCK PROPERTIES
294 294 295 295 295 298 298 298 299 299 300 300 300 300 301 301 301 301 302 302 302 303 303 303 304 304 304 305 308 309 309 310
11.3.1 Objectives 11.3.2 Reasons for Poor Advance Rates 11.3.3 Current State of TBM Art 11.3.4 Relevant Rock Properties 11.3.4.1 Rock strength 11.3.4.2 Chemical composition 11.3.4.3 Discontinuities 11.3.4.4 Rock bedding 11.3.4.5 Hydrology 11.3.4.6 Depth of cover 11.4 CONCLUSIONS
310
11.5
311
11.1
REFERENCE
INTRODUCTION
The art of tunneling using full face tunnel boring machines has progressed significantly over the last 20 years. In particular, tunneling machines have become very much more mechanically reliable for performing the tasks for which they are designed. Two major advances have led to this improvement in performance. The first has been the experiences that have led to improvements in mechanical design and the second, very significant improvements in the performance of the cutting
293
294
Mechanized Excavation
tools used on the machines. These have come about due to the increased understanding of the cutting mechanisms of disc cutters and the application of modern metallurgy to achieve the required cutting performance. To date, however, the universal tunneling machine, i.e. one that will cope with all ground conditions, has not, in practice, been built. The cutting mechanisms of machines are designed for specific rock parameters with disc type and cutter spacing being matched to the rock types to be encountered in a particular tunnel drive. The configuration of the machine, in particular its ground support systems around and behind the head, are designed for the rock mass properties of the ground in order that the most effective methods of rock support for the particular ground being encountered may be installed most efficiently. One aspect of the tunneling process has not advanced at the same pace as the two mentioned above. This is the ability of the tunneler to predict the performance of his tunneling machine in the ground. Before a tunnel boring machine is put into the ground for a particular project, the key question is always 'How will it perform?' If the answer to this question is always known, then the excitement of the project would no longer exist for the tunneler. However, if tunnelers do not pursue the answer with some reasonable degree of vigor and success, then promoters will lose faith in the industry to supply the products required, i.e. to build tunnels within a budget and to a required timescale. Much work has been done to relate TBM performance to various rock properties. However, this work has tended to relate specific rock properties to specific performance characteristics. This fundamental work is necessary to provide a basic understanding of the problems that can be used to develop technology, such as cutter technology. However, as far as the consumer is concerned it is only the final cost of his project and the time to build it that matter. The promoter of a tunnel, i.e. the person paying for it, has little interest in the instantaneous penetration rate of a single 16 inch disc cutter under a load of 20 tons. In practice the instantaneous penetration rate of a tunnel boring machine is only one of a number of parameters that determine the weekly advance rate. It is this weekly advance rate that largely dictates the cost of the project. This chapter explains, in some detail, the parameters that affect the cost of TBM tunneling. An analysis is then made of how the cost parameters are affected by rock properties, with the ways in which the industry is currently attempting to overcome some of the main problems being discussed. It is hoped that once they have studied this chapter, readers will have a better understanding of the realfinancialeffects of various rock properties on tunneling projects. Once this understanding exists, it will be possible for tunnel engineers to concentrate their attention on rock properties that are financially significant to the project being considered, to the ultimate benefit of the industry.
11.2 THE ECONOMICS OF TBM TUNNEL DRIVING 11.2.1 General Approach The cost of driving a tunnel is neither fixed, time related, nor distance related, but is rather a combination of all three of these factors. Were the costs purely to be time related then it would be easy to see the effect of one day's lost driving due to rock support problems on the cost of the tunnel. Equally, were the cost entirely distance related, then this would mean that one day's lost driving due to supporting the ground had nofinancialeffect on the cost of the tunnel. Obviously, neither of these scenarios is correct and in practice a complex and varying relationship exists between parameters. One of the significant costs that can be treated in various ways is the cost of plant such as the tunnel boring machine itself. In order to assist in understanding the economics of TBM tunnel driving, there follows an example of a typical cost estimate for a project where the tunnel is to be excavated using full face TBMs. The cost estimate used as an example on the following pages is for a project where two parallel 16 km long tunnels are required. The tunnels are required to have an internal diameter of 7.2 m and the project also includes the necessary access shafts, cross passages, portal excavations and spoil disposal. Any fitting out of the tunnels for theirfinaluse has not been included in the example, as these items are not relevant to the driving costs. The cost of most types of construction work is estimated by multiplying the known price for a unit type of work by the quantity of that work required on that particular project, i.e. it may be known that concrete placed in a typical wall costs £300 per cubic metre, and that there are 10 cubic metres of this type of work required, giving a total of £3000. This type of cost make-up is totally inappropriate for tunnel projects and they are therefore normally costed using a resource method. In this method
The Effects of Rock Properties on the Economics of Full Face TBMs
295
the cost of carrying out the entire project is calculated from first principles and then, if necessary, such items as 'excavation cost' can be calculated by taking a proportion of this overall cost and dividing it by the number of cubic metres involved. There are three main cost headings which are used when calculating the resource required for a tunneling project, namely the plant (equipment), labor and materials. For the project we are considering here, it is assumed that one tunnel boring machine can drive a tunnel 16 km long in a total of 100 weeks. The effects of variations in this driving time are discussed later in the chapter. 11.2.2
Plant Costs
The plant make-up sheet for our project is shown in Table 1. The plant make-up shows the cost for one tunnel boring machine and all its associated equipment required for the duration of the driving of one of the two tunnels. All of the equipment needed for the driving of the tunnel is listed in items 1 to 52 down the lefthand side of the table. It has been assumed that the spoil will be brought out of the tunnel using rail-mounted locomotives and mine cars, it will be tipped near the tunnel portal and then taken away to a final tip in 20-tonne dump trucks. 11.2.2.1
Basic details
An indication of the typical quantity of plant requirements is given in Column 2 of Table 1 where the number of each item of plant required is shown. As some items such as rail track and electric cable are purchased or hired by the meter, it is the metric length of these items required for the project that is recorded in this column. The duration of plant requirement is shown in Column 3 where the number of weeks that the item of plant is required to be on the project is given. Care must be exercised in determining this number, as it may be used when calculating the hire cost, the depreciation, or the fuel used. Most of the plant is required for the duration of the tunnel driving only. However, some items that relate to the site setup such as staff transport, will be required for a number of weeks prior to the start of the tunnel driving and a number of weeks after it has been completed. When these costs are being included in the basic tunnel cost, these additional weeks must be taken into account. Column 4 gives an indication of the hire cost of an item on a weekly basis if the item is to be hired for the project. Hire generally applies to shorter duration projects when capital purchase is not viable, or where a joint venture is hiring plant from one of its constituent companies. Typical capital costs are shown in Column 5. This is the capital cost for the item of plant concerned were it to be purchased new. As well as being used for calculating depreciation charges this figure is also often used when calculating the amount that should be allowed for spares. 11.2.2.2
Depreciation allowances
Column 6 gives the depreciation percentage that is going to be allowed on this specific project for the particular piece of equipment concerned. It is in this column where the policy of the company making the estimate may vary the figures considerably. For example, in the case of the tunnel boring machine, the company may feel that it is unlikely to resell or reuse this particular machine after this project, in which case, most of its cost must be written off on the project. On the other hand, the company may feel that the machine could have a high resale value, or was of a diameter and type that was very likely to be used elsewhere and could be hired out to others at the end of the project. In this latter case, a much lower depreciation percentage could be allowed. The policy that has been used in the figures shown in Table 1 is that any highly specialized item of plant should be largely written off on the project at either 90 or 95% depreciation. Other specialized items of tunneling plant may have some resale value, but are likely to have been well used on the project and therefore 80% depreciation has been allowed. Linear items such as pipelines often have a reasonable resale value irrespective of the duration of the project and have therefore been written off at 50%. Other linear items such as electrical cabling suffer considerable damage when being removed from tunnels and have therefore been written off at 75%. Some of the other less specific plant such as tunnel fans and surface equipment are also considered to have high resale value and have been written off at 50%. Items such as the buildings, or the site infrastructure, are regarded as having very little resale or reuse potential and have been written off at 90%.
Table 1 1
1 TBM and backup 2 Locos 15 t 3 Locos 10 t 4 Mulhausser cars 5 Flat cars 6 Cement cars 7 Manriders 8 Grout pumps Schwing 9 Grout pumps Colmono 10 Grout mixers 11 Pumps 150 mm 12 Pumps 50 mm 13 Pump line 150 mm 14 Water line 75 mm 15 H.V. cable 16 Cable & lights 17 Communication cable 18 Fan duct 750 mm 19 Fans 60 kW 20 Lasers 21 Z E D equipment 22 Track 23 Probe drills 24 Breakers
2 Number
1 6 6 40 10 6 8 2 4 4 10 6 32000 16000 16000 16000 16000 16000 16 3 1 32000 2 6
3 Weeks
100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100
4 Hire (£)
Plant Cost Make-up Sheet (7.2 m Internal Diameter Tunnel, 16 km Long)
5 Cost (£) 8 500000 60000 35000 10000 4000 2700 1700 8000 10000 7500 10000 5000 10 5 22 25 12 20 12000 4000 22000 50 30000 750
6 7 Depreci- H. Charge ation (%) (£) 95 80 80 75 75 75 80 80 90 90 80 80 50 50 75 75 90 75 50 80 80 50 75 100
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
8 D. Charge (£)
9 Freight (£)
10 Erect (£)
8075000 288000 168000 300000 30000 12150 10880 12800 36000 27000 80000 24000 160000 40000 264000 300000 172800 240000 96000 9600 17600 800000 45000 4500
85000 3600 2100 4000 400 162 136 160 400 300 1000 300 3200 800 3 520 4000 1920 3200 1920 120 220 16000 600 45
85000 3600 2100 4000 400 162 136 160 400 300 1000 300 3200 800 3 520 4000 1920 3200 1920 120 220 16000 600 45
11 Spares
12 Spares
(%)
(£)
18 18 18 8 8 8 8 18 18 18 18 18 0 0 0 8 0 8 18 8 8 18 18 18
2942308 124 615 72692 61538 6154 2492 2092 5 538 13 846 10385 34615 10385 0 0 0 61538 0 49231 66462 1846 3 385 553 846 20769 1558
13 Hr/wk
14 UtiVn (%)
168 168 168 168 168 168 168 168 168 168 168 168 168 168 168 168 168 168 168 168 168 168 168 168
30 50 60
1500 80 60
6 6 6
20 20 50 100 100
10 10 10 20 5
6 6 6 6 6
50 100 100
60 1 5
6 6 6
5 5
10 1
6 6
15 Rating
16 Unit (pence)
17 Fuel (£) 453600 241920 181440 0 0 0 0 4032 8064 20160 201600 30240 0 0 0 0 0 0 483 840 3024 5040 0 1008 302
25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52
Torque spanners Burning equipment Air receivers Compressors 500 cfm Cap lamps Charger units Gas monitoring eqp. Cranes - wheeled Cranes - segment yrd Fuel tanks Cement/PFA silos Front end loaders D u m p trucks 20 t Flat trucks Crew buses Vans Cars Fitters workshops Elects, workshops General stores Changing rooms Toilets/showers Canteen Engineers' offices Heating/lighting Welders Elec. distribution Sewage
6 4 4 1 100 5 2 4 2 2 2 3 6 2 3 20 27 1 1 1 1 1 4 1
100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 150 150
100 100 100 100 100 100 100 100 100 100 100
50
60 60
5000 2500 150000 50000 5000 10000 200000 65000 25000 25000 8000 10000 100000 50000 50000 100000 100000 100000 250000 25000 8000 25000 25000
100 100 75 75 80 90 100 50 75 90 50 80 75 100 100 90 90 90 90 90 90 90 90 90 90 100
(£) 21 515492
300
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
4000 15000 26250 4000 22 500 5000 300000 75000 9000 10000 480000 292500 50000 75000
0 0 0 0 0 0 0 0 0 0 0
90000 45000 45000 90000 90000 90000 225000 22500 28 800 22 500 25000
180000 243000
423000
Totals (£) Total plant cost
50
1000 5000 35000
0 0
3 40 200 350 50 250 50
6000 1000
100 200
3 40 200 350 50 250 50
6000 1000
100 200
6000 3900
6000 3900
1600 2700 1000
1600 2700 1000
1000 1000 1000 2500
1000 1000 1000 2500
500 750
500 500
250 320 250 250
500 750
500 500
250 320 250 250
13 355680 165 366 165 366
18 18 8 18 18 18 18 18 18 0 8 18 18 18 18 18 18 18 18 8 8 8 8 8 18 18 8 18
104
1385 3077 12115 1731 8654 1731 207692 34615
0
3077 207692 135000 17 308 25962 83077 140192 34615 17308 7692 15 385 15 385 15 385 38462 8654 11077 3846 8654
5 105 169
168 168 168 168 168 168 168 168 168 168 168 168 168 168 168 168 168 168 168 168 168 168 168 168 168 168 168 168
0 0 0
60
100
6
60480
100
10
6
50400
50 30
5 10
35 6
58 800 6048
50 50 50 20 10 10 20 20 100 10 20 20 50 50 5
5 5 2 2 1 1 10 10 1 1 10 10 10 100 25
35 35 35 200 200 200 6 6 6 6 6 6 6 6 6
44100 88 200 11760 40320 100800 136080 2016 2016 1008
0 0
0 0
101
2016 2016 5040 $0400 5040
0 0
2300911
298
Mechanized Excavation
It should be noted that where an item like trackwork has been written off at 50%, this could then allow for either the purchase of new trackwork and its resale at the end of the project for half its original cost, or alternatively, the purchase of secondhand trackwork that can be obtained at half the new cost. None of these figures should be regarded as being firm, but rather figures that are in the correct order and give the correct feel for the cost of a particular item of plant on the overall project.
11.2.2.3
Other plant cost headings
The hire charges that are going to be put against this project for hired plant are a multiplication of the number of items by the number of weeks by the hire cost for the item concerned - shown in Column 7 of the table. The depreciation charge that is going to be charged to this project is shown in Column 8. It gives the number of plant items multiplied by the total cost multiplied by the percentage depreciation determined for the particular item. Freight costs for transporting the equipment to and from site, usually from the contractor's depot, are given in Column 9. Erection cost for the plant on site covers money to be allocated for craneage and specialist labour, where required, for the erection of equipment on the site and is shown in Column 10. An allowance must be made for spare parts (Column 11). This is normally a percentage of the new cost of the equipment that is to be allowed annually for the cost of spare parts and replacement items for the equipment concerned. Three figures have been used in the example: (i) 18% for items of plant which are working continuously at, or near, their design level; (ii) 8% for items of plant which are mechanically simple and therefore fairly reliable; and (iii) 0% for items such as pipes where no significant maintenance costs can be envisaged. These percentages for spare parts are comparatively high; however, this is because in the example the driving scenario that has been assumed is one where the tunnel is driven 24 hours a day, 7 days a week. Under these circumstances the plant gets very heavy use compared to the more conventional 40-50 hour week on a regular construction site. In addition, due to the lack of specific maintenance periods, items are often replaced rather than repaired, causing additional cost. In a scenario where driving takes place on two 10-hour shifts for 5 days a week it would be reasonable to halve the above spares percentage figures when looking at the annual cost of spare parts. Column 12 gives the actual allowance for spares for each item of plant for the example project.
11.2.2.4
Utilization, availability and power consumption
Table 1, Column 13 gives the hours per week for which the equipment is available for working on the project. For example, if a project worked 5 days a week, 24 hours a day, then the figure would be 120 hours, but for a project working 7 days a week, 24 hours a day, the figure is 168 hours. Utilization of the equipment is shown in Column 14. The utilization figure used in this column is the running utilization, i.e. the time when the equipment is actually working rather than standing waiting to work. In the case of pumps and fans, the utilization will be 100% of the total time, whereas in the case of such items as locomotives they are more likely to be pulling trains 50% of the time and waiting the other 50% of the time. Some items of equipment, such as grouting equipment, may be used less frequently and their actual working time could be 20%, or less. The figures are entered, using the experience of the cost estimator, and those shown in Table 1 are considered indicative of the sorts of figures that are reasonable at the current time. Column 15 gives the rating of the plant item in terms of power consumption. In terms of electrical equipment, this figure is installed kilowatts and in terms of diesel equipment it is the normal diesel consumption in terms of gallons/hour. Column 16 gives a unit cost of fuel, i.e. 6p per unit for electricity, 35p per gallon for diesel to be used on the site and 200p per gallon for road fuel for road vehicles (1990 UK prices). The final column, Column 17, then gives the fuel cost for the plant on the project by multiplying the number of plant items by the number of weeks by the hours per week by the utilization, the rating and then the unit cost of fuel.
11.2.2.5
Equipment cost implications
The summation of the costs for the items that are included in the overall cost make-up are given in the table at the foot of the columns: 7 for hire charges, 8 for depreciation charges, 9 for freight costs,
The Effects of Rock Properties on the Economics of Full Face TBM s
299
10 for erection costs, 12 for spares costs and 17 for fuel costs. In this way, the total cost for the equipment to drive a particular tunnel can be arrived at. Many of these items are a complex mixture offixedtime and distance-related costs. If one takes the example of the TBM itself, a large part of the cost is afixedcost and the depreciation of 95% would not be considered to change were the tunnel 15 or 17 km long rather than 16 km long. The freight and erection charges would remain as a fixed cost, independent of the tunnel length. The spares percentage is calculated, knowing the duration of the project and the rate at which the machine is going to work on the project, and is therefore effectively distance related. Likewise, the fuel cost is also basically distance related and will increase or decrease depending on the length of the tunnel. What we are seeking here, however, is the effect that rock properties may have on a cost scenario, such as the one shown in Table 1. Rock properties do not alter the length of the tunnel, but may, in fact, significantly alter the time taken to drive the tunnel. On a large tunnel project very few items of equipment do, in fact, have directly time-related costs. However, the capital cost of the tunnel boring machine and its back-up may vary quite significantly depending on the complexity of the rock conditions that it is anticipated the machine will have to cope with. What is shown, however, is that using this type of accounting, a delay of a period of time caused by encountering unexpected difficulties with rock conditions, would have little effect on the final equipment cost. Had the builder of the project used hire charges instead of capital costs and depreciations, then it would appear that the costs were very much time related. However, it is considered that in reality hire charges are an inappropriate way of considering costs for large tunnel projects. 11.2.3 Labor Costs 11.2.3.1 Labor basis Table 2 shows a cost make-up for the total cost of labor for one of the two tunnels on the project under consideration. It has been assumed for the project that in order to work 24 hours per day, 7 Table 2 1
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
TBM operators Fitters - tunnel Electricians - tunnel Miners Segment erectors Segment loaders Grouters General TBM backup Tracklayers Track maintenance Loco drivers Pit bottom men Fitters - surface Electricians - surface Segment yard men Front loader ops Dump truck drivers Canteen etc. Chainmen Crane drivers TBM shift boss Yard foremen Foremen - M&E Surface laborers Security staff
Totals Total labor cost
2 Number per gang
3 Gangs per day
4 Weeks
5 Shifts/ week
6 Shift rate (£)
2 3 2 6 1 2 3 10 4 9 8 6 4 2 8 4 6 6 6 5 1 1 1 10 6
3 3 3 3 3 3 3 3 3 3 3 2 2 2 2 2 2 2 3 2 3 2 2 2 2
100 100 100 100 100 100 100 100 100 100 100 100 125 125 125 125 100 125 125 125 100 125 125 150 150
7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7
210 110 120 210 210 110 110 110 110 90 90 80 90 100 70 95 95 50 50 95 210 120 130 70 70
116 £ 22 237 250
Labor Cost Make-up Sheet 7 Weekly cost/ man (£) 1470 770 840 1470 1470 770 770 770 770 630 630 560 630 700 490 665 665 350 350 665 1470 840 910 490 490
8 Total/ week (£)
9 Total cost (£)
8 820 6930 5040 26460 4410 4620 6930 23100 9240 17010 15120 6720 5040 2800 7 840 5 320 7980 4200 6300 6650 4410 1680 1820 9 800 5 880
882000 693000 504000 2646000 441000 462000 693000 2310000 924000 1701000 1512000 672000 630000 350000 980000 665000 798000 525000 787 500 831250 441000 210000 227500 1470000 882000
£204120
£22237250
300
Mechanized Excavation
days per week, then 3 gangs of tunnel workers will be required. Obviously, in order to operate continuous 7 day working, it will be necessary to employ more than 3 gangs working on a shift basis. However, as the cost per shift of a man is only incurred for the shifts he works, then it is possible to calculate a cost estimate, assuming that one of the gangs is always present in the tunnel. 11.2.3.2 Labor required The various disciplines and classifications of operatives required are given in Column 1, items 1-25 of Table 2. Operative classifications are needed for each type of operative where a different rate of pay may apply. The classifications and numbers of men shown are typical for the example project. Column 2 gives the number of operatives for each particular category in a gang of men who are working at any one time.
11.2.3.3 Labor numbers and duration requirements The number of gangs required per day to work the 24 hours is shown in Column 3. In general practice in the UK this is three gangs for 'face' work and two gangs working 12 hour shifts for surface work. The general principle being that production labor works continuous 8 hour shifts with changeover of work at the work place, while backup labor can work 12 hour shifts with a hiatus at shift changes. Column 4 gives the weeks for which one class of operative is required on the job and should, in general, coincide with the weeks for the plant item required if the operative is a plant operator. Some operatives, in particular fitters and surface labor gangs, are required for longer than the duration of the tunnel driving due to the need to establish the working site prior to tunnel driving and to dismantle it once it is completed. The number of shifts in the week that a particular operative type is required on the project is given in Column 5. 11.2.3.4 Labor rates Column 6 gives the shift rate, i.e. the cost to the project of the man for the shift. Thisfigureincludes work overheads, insurances, pension schemes, etc. and takes account of his skills and whether or not he is working an 8 or 12 hour shift. The rates quoted in this column are typical costs in pounds sterling applicable in the UK in 1993. (In practice local rates should be used.)
11.2.3.5 Summary of labor costs The weekly cost to the project of a particular type of operative is shown in Column 7. Column 8 gives the total cost to the project for the week for that type of operative and is the weekly cost per man multiplied by the number of gangs per day and the number in each gang. Column 9 gives the total cost to the project for the classification of operative in question. At the end of the columns the necessary totals are given, i.e. Column 2 - the total number of operatives required (excluding spare gangs); Column 8 - one can see the weekly cost of labor; Column 9 - the total cost of labor for the project is shown. It is the weekly labor cost which is of particular interest, as labor costs are wholly time related. Therefore, if particular rock properties caused difficulties that took one week to overcome, it can be seen that the cost of the tunnel in this case would rise by over £200 000. There is, therefore, in all tunnel work a very significant time-related cost in the labor element of the work that needs to be taken into account when considering the effects of rock properties on the economics of tunnel driving. It should be noted that the labor costs in Table 2 do not take account of production bonuses that are often paid to production workers on tunnel projects. This in no way invalidates the cost estimate, as the labor costs are given as the normal labor costs on achieving normal target production. However, if production increases beyond the norm and payments therefore also increase beyond the figures given, these bonus payments are generally related to the savings made by finishing the project early. The cost of bonuses can, therefore, be regarded as 'self cancelling' in that they are paid when they can be offset by the savings caused due to the shorter duration of the project.
301
The Effects of Rock Properties on the Economics of Full Face TBMs 11.2.4 Materials Costs 11.2.4.1 Materials costs details
Table 3 shows two scenarios from materials costs for our 16 km long tunnel project. It has been assumed that the tunnel is being driven through horizontally bedded sedimentary rocks, consisting of mud rocks, sandstones and limestones. Scenario 'Α' represents a situation where the tunnel is to be followed by a permanent in situ lining. Scenario Έ ' assumes what may be the 'norm' in the future, and one where the tunnel is lined throughout with a precast concrete segment lining as the tunnel is driven. In this latter method the final lining of the tunnel is achieved at the same time as driving and although the lining system appears expensive, the total project duration is considerably shortened. Typicalfiguresfor the use of rock bolts, steel arches and sprayed concrete are shown in Scenario ' A\ The total amount of these materials required has been calculated by considering the number used per metre (when they are used), the total length of the tunnel, the percentage of the tunnel drive of which it is considered the particular type of support will be needed, and the unit cost of the materials item. From this table it can be seen that the cost of rock support material is comparatively small when compared with the cost of the labor and equipment required for the project. 11.2.4.2 Implications of materials costs All of the costs are specifically linearly related, i.e. the costs increase in direct proportion to the meterage over which they are required. The other material costs included in this section are also directly linearly related, i.e. the cost of cutting tools for the tunnel boring machine. If the tunnel length is increased by 10%, and assuming uniform rock conditions, the cost of cutting tools for the TBM will also be increased by 10%. For this reason, the cost of TBM cutters is included in the materials section as opposed to the spares section of the tunnel equipment schedule. Scenario 'Β' shows a situation where precast concrete segments are used to line the tunnel as it is driven. In this situation the cost of materials appears high. However, it has to be remembered that in this case the cost of final lining of the tunnel, i.e. the segments, is included in this figure and that although the lining system is comparatively expensive, the cost of temporary support materials is also eliminated. The significant point to note is that for the 'normal' tunnel job the cost of temporary or permanent rock support materials is likely to be very small in comparison with the plant cost and the labor requirement cost for the project. 11.2.5 Staff Costs Labor costs that have not been taken into account in Section 11.2.3 are the costs of the management and engineering staff required for the project. A make-up for this cost is shown in Table 4 where one can see the typical site management costs for the project under consideration. These costs are comparatively small and are time related. Table 3 Materials Cost Elements No(m) 1 Rock bolts 2 Arches 3 Shotcrete 4 TBM cutters Total materials cost
15 1 1
Scenario 'A' Total (m) % Required 16000 16000 16000
25 5 50
Total
Unit cost (£)
Total cost (£)
60000 800 8000 745000
10.00 350.00 100.00 2.00
600000 280000 800000 1490000 £3170000
Number
Unit cost (I)
Total cost (I)
16000 30000 745000
2467.00 20.00 2.00
Scenario 'IT 1 Precast segments 2 Grout 3 TBM cutters Total materials cost
39472000 600000 1490000 £41562000
302
Mechanized Excavation Table 4 Staff Cost Make-up Sheet 1
1 2 3 4 5 6 7 8 9
2 Number
3 Weeks
4 Cost/ week (£)
1 1 5 3 6 3 2 6 10
150 150 150 150 150 150 150 150 150
750 650 600 550 500 600 650 375 250
Contracts Manager Agent Sub-agents Senior engineers Engineers (civil) Engineers (plant) General foremen Q.S. staff Clerks/typists etc.
37
Totals Total staff cost
5 Total/ week (£)
6 Total cost (£)
750 650 3000 1650 3000 1800 1300 2250 2 500
112 500 97 500 450000 247 500 450000 270000 195000 337 500 375000
£16 900
£2535000
£2535000
11.2.6 Cost Summary 11.2.6.1 Project basis In order to complete the overall cost estimate picture a summary of cost estimates for a complete project is shown in Table 5. In the project being considered two 16 km long tunnels were to be constructed and the necessary shafts, cross passages, portals, site establishment and spoil disposals facilities were also to be provided. The cost of driving one 16 km long tunnel is given by the addition of the summaries of the tables in Tables 1-4 inclusive. The summary, therefore, shows the cost of 2 such tunnels, a typical cost for access shafts associated with such tunnels, cross passages, portal excavations and the site establishment. In addition, an allowance has been made for spoil disposals facilities. It can be seen that these sundry items have added approximately 5% to the cost of this project. 11.2.6.2 Implications of cost headings In this summary we can see that we have a unit cost per meter for the tunnels of £6589. The detail of the make-up shows the complexity of the relationship betweenfixedcosts, time-related costs and distance-related costs, and also the high degree of interchangeability between these types of costs, Table 5 Cost Summaries
1. 2. 3. 4.
Plant Labor Staff Materials
Total 20% On-costs Total Unit cost per meter of tunnel
1. 2. 3. 4. 5. 6.
Tunnels 2 @ 16 km Shafts Cross-passages Portals Site establishment Tip establishment
Grand Total
Drive summary £ 21515492 22237250 2535000 41562000 £87849742 17 569948 £105419690 £ 6 589 Grand! summary £ 210839382 4000000 5000000 2000000 500000 500000
£222839 382
(including final lining)
The Effects of Rock Properties on the Economics of Full Face TBMs
303
£M 300 290 280 270 260 250 240 80
90
100
110
120
130
140
150
Number of weeks
Figure 1 Effect of variation in time on cost
depending on the accounting scenario being used for the project. Two very obvious examples are in the plant, where if the plant had been costed on a hire basis, the cost would have been time related. However, when estimated on the depreciation basis, costs are fixed and virtually independent of time. The other element is in the materials element where the materialsfiguresshown are inclusive of the lining for the tunnel with precast concrete segments. This makes the materials cost highly distance related. However, had it been assumed that the tunnel was to be driven and temporarily supported with a subsequent in situ lining, then this in situ lining would have had a high labor element, thus making a significant proportion of this material cost time related. The sensitivity of the cost of the project to time is shown in Figure 1. Figure 1 shows the effect on total cost for the example project of an increase or decrease in time from the assumed period. It can be seen that the cost increases by about £1 M per week of extra time taken to complete the work. Should, therefore, a rock condition be met which delays the project by one month, then a very high direct 'extra time' cost will be incurred. The actual materials cost for dealing with the problem is likely to be at least an order of magnitude lower than the time cost. It is an understanding of the concepts of cost make-up for TBM tunnel driving that is necessary prior to the reader being able to determine the effects of rock properties on the cost of tunnel driving. In simple terms it has often been assumed that tunnel driving is almost entirely time related as a cost. However, it can be seen from the above cost make-up calculations that there is a highly complex interrelated and interchangeable nature between the fixed time- and distance-related elements of tunnel costs. In general terms, however, time remains one of the most significant variable cost items. Materials used, particularly for temporary support, are generally an order of magnitude less in cost than the overall time cost for their installation into the tunnel. 11.3 COST/PERFORMANCE CRITERIA RELATING TO ROCK PROPERTIES 11.3.1 Objectives The objective of any tunnel drive is to achieve 'good' driving performance. This performance can be defined as either the economic performance generally accepted for that situation (with the current state of knowledge) or alternatively the most economic way of achieving a particular objective. Many tunnels are needed irrespective of cost and in this case it is the most economic solution which is being sought. In many other cases, however, tunneling is competing with other forms of construction and in these cases excessive costs will result in alternative forms of construction being utilized. 11.3.2 Reasons for Poor Advance Rates Poor advance rates in tunneling with tunnel boring machines are almost invariably associated with variable ground conditions. This poor advance rate is not due to the inability of the industry, with the current state-of-the-art, to build a tunneling system that will cope with a given range of
304
Mechanized Excavation
ground conditions. It arises because the tunneling system has been designed for optimum performance in a given set of ground conditions and insufficientflexibilityhas been built into the design to cope with ground conditions outside the perceived parameters. This largely arises from the methods of tendering in the construction industry. The contractor who foresees all possible eventualities, calculates his price for the project at a level so high that he is never awarded a project. However, it is prior to the project being designed and going out to tender that the tunnel engineer has the opportunity to define the ground conditions through which the tunnel is to be constructed. With modern site investigation techniques, defining the actual ground conditions through which a tunnel is to be driven is not particularly difficult. Site investigation has reached the stage where with the aid of both boreholes and geophysical methods, as well as the wealth of existing information, it is possible to predict ground conditions for a tunnel drive with a remarkable degree of accuracy. Even when specific geological phenomena have not been recorded in a site investigation it is usually possible for the geotechnical engineer to draw to the attention of the designer and the contractor other ground conditions which are likely to exist in the particular strata to be encountered in a specified area. Both site investigations techniques, and the interpretation of that site investigation in order to provide the tunnel engineer with an accurate assessment of the anticipated geological conditions along a tunnel, are now to a high standard. However, the state-of-the-art in the interpretation of rock and rock properties and ground conditions as they affect the performance of tunnel boring machines (and consequently the economics of these machines) is still very much an art rather than a science. This is due to the very small differences in ground conditions that can cause dramatically different performances in tunneling systems with modern TBMs. At some time in the future it may be that universal TBMs can be built economically. With the current state of knowledge in the industry it is, however, only practical to build a machine that will cope with a given predicted set of circumstances. One then builds into that machine all of the necessary equipment that will cope with any ground condition generally regarded as an economic proposal. 11.3.3 Current State of TBM Art The state-of-the-art in tunnel boring machines have reached the stage where they can be designed and manufactured economically to cope with almost any given set of circumstances that are likely to be encountered on a specific project. These circumstances can be defined by the various classifications of rock properties. 11.3.4 Relevant Rock Properties Rock properties that affect tunnel boring machine performance fall into six broad headings: (i) rock strength, (ii) chemical composition, (iii) discontinuities, (iv) rock bedding, (v) hydrology and (vi) depth of cover. Each of these rock property headings and the ways in which they can affect TBM performance are now discussed. 11.3.4.1 Rock Strength Rock strength is the parameter that has by far the most significant effect on the instantaneous penetration rate of tunnel boring machines. It is not unusual for a tunnel boring machine to achieve three times the penetration rate in a softer rock ofthat which machines are capable of in a hard rock. The rock strength affects the type, metallurgy and spacing of the cutters used on the head of the tunnel boring machine. All of these parameters can be designed for a specific rock type, and as such, inefficiencies are bound to occur when the machine has to cut more than one rock type on one tunnel drive. However, when a variable rock condition is to be encountered it is the hardest rock which dictates the cutter technology to be employed. Designing the cutter technology to achieve a good performance in harder rocks is of paramount importance. Figure 2 shows a typical modern cutter and Figure 3 shows the layout of these cutters on the cutter head. It has been shown in Section 11.2 that the economics of tunneling are highly dependent on overall advance rate. In turn the overall advance rate is closely related to instantaneous penetration rate and hence to rock strength. The relationship between rock strength and overall advance rate is, however, not linear. Circumstances have arisen where overall performance of a
The Effects of Rock Properties on the Economics of Full Face TBMs
305
Cutting disc Bearing
Housing
Figure 2 Typical TBM rock roller cutter
Figure 3 Typical layout of cutting discs on TBM cutter head
machine has been almost constant, irrespective of rock strength. On one particular drive a 3 m diameter hard rock TBM achieved an average progress rate of 95 m per week. The machine drove through 3 rock types; a tillite with a UCS (unconfined compressive strength) of about 150 M Pa; a shale with a UCS of 50 MPa and a sandstone with a UCS of 250 MPa. The weekly progress of this machine is shown in Figure 4 and it can be seen that although the penetration rate varied in accordance with the UCS the weekly progress rate remained almost unchanged throughout the tunnel driving. Various reasons exist for this constant performance, the main ones being the need for temporary support in the weaker rock. It was, however, very clear that it was the cutter technology that limited performance in the hard sandstone. Both the low penetration rate and the need for frequent cutter changes had significant effect on progress. Table 6 shows an analysis of where time was spent while driving in the various rock types. In the hard sandstone 33% utilization was achieved, demonstrating that even when penetration rate is the limiting factor only 1/3 of the time was spent cutting. Both cutter technology and machine utilization have come a long way over the last 20 years. Table 7 shows typical parameters for three similar-sized TBMs designed and built in 1972, 1979 and 1989. The very significant increases in power and thrust per cutter are clearly shown. These advances have resulted in very significant growth in the penetration rates possible in harder rocks. Improved penetration, combined with a general increase in utilization from 33% to nearer 50%, has led to very hard rock TBM driving becoming an economic option.
113.4.2
Chemical composition
The composition of the rocks encountered can have significant effects on tunneling performance in two ways. First, the quartz content is very significant in terms of cutter wear. Quartz is a highly
Mechanized Excavation
306
Table 6 Analysis of Activities in Three Rock Types. Machine Utilization (as % of total available shift time)
Drilling Awaiting spoil removal Cutter changes General maintenance Support Repairs to TBM Services Derailments Surveying Labor Other Av. penetration rate (m h *)
Tillite
Shale
Sandstone
All rocks
34.60 7.62 5.66 9.84 13.48 12.59 6.33 0.47 6.01 2.30 1.10
27.02 9.32 1.74 5.78 32.22 7.74 3.73 1.35 9.38 1.72 0.00
33.14 11.81 15.39 8.39 8.02 8.77 2.64 0.56 2.90 7.70 0.68
32.44 9.84 9.27 8.37 14.72 9.91 4.15 0.69 5.28 4.64 0.69
2.32
3.27
1.88
2.28
Table Mountain S/S
Dwyka Tillite
Dwyka Tillite
Ecca Shale
Av. prog.(m wk"1)
96
116
84
81
Penetration (m hr~')
2.32
3.27
1.88
2.32
Rock type
180 H
Λ^ u ir
160-1 140
« 120 H Progress ' m wk-1 l 0 ° 80
60 40 20
J
I
LJ
I
I
I
Overall av. 9 5 m wk~
il
L Weeks
Figure 4
Weekly progress of TBM in three rock types
Table 7 Year Model Diameter Geology Cutter types Thrust per cutter Total cutter head thrust Cutter drive Approx. weight
Development of TBMs
1972-73
1979-80
1989-90
142-145 4.27 m Limestone 32-305 mm 3-280 mm 89 kN 3115 kN 447 kW lOOt
147-210 4.32 m Dolomitic limestone 27-394 mm Center 178 kN 5425 kN 671 kW 113 t
1410-252 4.3 m Micaschist granite schist 29-483 mm 4-305 mm 314 kN 9100 kN 2345 kW 262 t
abrasive material, and when present in large quantities can cause extreme cutter wear. For a rock of a given strength, one with a high quartz content can require the use of 10 times the number of replacement cutters compared to one with a very low quartz content. This factor is significant not only because of the cutters, which are comparatively expensive disposable items, but more because of the time taken to change the cutters. Changing a set of cutters on a full face TBM may take, for example, a full working shift. Referring to the cost scenarios in Section 11.2, it can be seen that in the example used the time-related cost for labor only for a shift is 15 000 units, for a full set of cutters on a TBM the total replacement cost of the cutter may amount to 2000-3000 units. In terms of tunnel labor the time cost of changing the cutters is almost five times the material cost involved. The overall time cost of replacing cutters is, therefore, an order of magnitude higher than the material cost of the
The Effects of Rock Properties on the Economics of Full Face TBMs
307
cutters themselves. When looking at the quartz content of rock, the eventual cost of a high quartz content is very much higher than might be immediately apparent from the basic cost of the replacement cutters required. Table 6 shows that on one particular tunnel drive 15.4% of the time was spent changing cutters in a quartzitic sandstone compared to only 5.6% in a hard tillite. This difference of 10% of total driving time has a very high cost to the project when compared with the materials cost of the 202 cutters actually changed. The second significant chemical property that affects tunneling performance is that which relates to rocks which deteriorate on exposure to either water or air. Most rocks remain stable when exposed. However, some rocks, particularly weaker sedimentary and some volcanic rocks, can deteriorate in various ways when exposed to either air or water. Rocks which do so deteriorate need to be sealed in order to stop the deterioration causing physical damage and eventual failure of the tunnel. The exact timing of the need for this sealing is very critical to tunnel boring machine economics. Some rocks are best sprayed with a thin layer of concrete in order to avoid deterioration. If it is necessary to apply this thin layer of concrete within hours of exposure, then this spraying of concrete near the front of the machine can cause severe difficulties. Sprayed concrete and tunnel boring machines are not compatible. The machine itself becomes heavily laden with surplus sprayed concrete and the ventilation and electrical systems become inefficient due to the deposition of the chemicals in the sprayed material onfilters.The rebound material is extremely difficult to collect and remove underneath the tunnel boring machine, and hampers necessary operations in that area such as the laying of track work for spoil removal. If, however, the spraying can be carried out at the rear of the machine, then a different scenario exists, the problems outlined above do not occur and the sealing of the rock can become a simple efficient operation which has little or no effect on the driving of the tunnel. This latter scenario occurred in the shales of the tunnel outlined in Tables 6 and 7. In this case the shale did deteriorate, but is was found that sealing the surface within a week or so was sufficient to maintain structural stability. Had it been necessary to seal the rock within 24 hours then only half of the progress rate would have been possible. This would have had a very dramatic effect on the cost of the tunnel drive. This particular example serves to illustrate a very small change in property that could have seriousfinancialconsequences. Had the composition of the rock been only marginally different, then it would have been necessary to seal the surface within hours of excavation. As discussed earlier, it is now a simple matter to predict rock properties from site investigation. It is, however, still a much more difficult art to predict exactly how those known rock properties will behave in a given tunnel excavation. Machine designers are currently attempting to design systems that can help to overcome these problems. Figure 5 shows the layout of a proposed 'ΝΑΤΜ' ΤΒΜ. This TBM has a moveable forward canopy intended to allow access to spray shotcrete onto the rock immediately behind the cutters. Whereas this system will allow access and enable the rock to have immediate sealing/support if required, the problems of the incompatibility of the machine and shotcrete still exist. The particular machine layout shown in Figure 5 also allows steel arch rings to be fed in behind the head of the machine and erected there if necessary.
Movable canopy forward position
Ring beam erector
Movable conveyor aft position
Ring beam conveyor
Gripper shoes (slotted for 'steel rings)
A^fi=S
Cutter head
Shield thrust cylinders
Segmental lining if needed
Segment erector
Figure 5 'ΝΑΤΜ' TBM showing forward support features
308
Mechanized Excavation
11.3.4.3
Discontinuities
Rock jointing, and in particular joint spacing, has a significant effect on TBM performance. Very closely spaced joints in a very hard rock can, for example, make the rock comparatively easy for the cutters to excavate. Conversely it only needs these joints to be very slightly open and combined with weak bedding planes, to create a highly unstable roof that requires immediate and significant temporary support. Roof support itself only becomes a problem to the economics of a TBM once the amount or type of support required goes beyond that for which the system was designed to cope. The quality of discontinuities can also have a noticeable effect. Very tightly spaced joints that are also tightly closed create a rock mass which for support purposes acts as a single unjointed rock. On the other hand comparatively widely spaced joints at a spacing of anything up to 30% of the tunnel diameter can cause very significant problems if these joints are open (see Figure 6). In this latter circumstance the blocks are free to move and can either collapse into the tunnel, causing support problems, or onto the head of the tunnel boring machine, rendering further advance impossible. The spacing and type of jointing which is anticipated in a tunnel drive can have a noticeable effect on the machine configuration designed for the drive. In rock tunnels there is always a conflict between providing shielding that protects the machine and the workforce from failing roof rock and at the same time hides the rock, making it impossible to see conditions that may require early support. The installation of early support is also rendered difficult or impossible. The trend is to provide all machines with some degree of shielding (Figure 7) that does not extend far behind the Blocks fallen out
Figure 6 Effect of weak bedding planes and joints on profile of TBM driven tunnel Recessed
Short
Finger
Optional roof drill
Drive motors Cutter head
Propel cylinders
Main gripper
Cutter head support
Rear support
Figure 7 Typical rock TBM with short shield
The Effects of Rock Properties on the Economics of Full Face TBMs
309
head to a point at which it is possible to install temporary support if necessary. This allows the stand-up time for the ground to be comparatively short, but at the same time, the machine is far less likely to be stopped by falling blocks. Even when loose blocks of rock cannot cause significant support problems they can continue to cause difficulties with the mucking systems. Mucking systems are conveyor orientated and require the rock to be broken down to comparatively small sizes to be handled by the equipment. TBMs designed with a normal cutter head design are unable to break down the blocks once they have dislodged from the face or the roof of the tunnel. In these circumstances the rocks must be passed back through the muck collection system and along the conveyors passing hoppers, and leading to severe damage to conveyor systems, particularly at transfer points. In addition, large blocks are often unstable on the conveyor system, leading to a tendency for them to fall off, causing safety hazards to the workforce and delays to tunneling whilst they are collected from the invert of the tunnel and removed from the system. One of the noticeable inflexibilities of TBM systems is the single purpose of the excavation and mucking system. These systems are designed solely to excavate ground at the face and transport the spoil through the forward equipment on a conveyor. This conveyor is usually at, or above, axis and often enclosed. As a result, any material that either falls from the roof or off the conveyor is extremely difficult and expensive to deal with. All such material must of necessity be manhandled either back onto the conveyor or through the back-up equipment to the rear muck transfer point.
113.4.4 Rock bedding Bedding planes in rock, where there are discontinuities similar to the jointing, cause similar problems to vertical or sub-vertical rock jointing. However, weak bedding planes cause additional problems. Due to the nature of sedimentary rocks it is normal for bedding planes to be horizontal or sub-horizontal features. (Figure 6). It is also common for bedding planes to be weakly cemented, resulting in the rock mass having no tensile strength across the bedding plane. These two factors combine to make bedding planes the most common cause of minor roof failures in TBM-driven tunnels. It is a very common situation to find one or both shoulders of the circular profile missing in a tunnel, the cause being a weak bedding plane at or near the crown of a tunnel. Another common problem resulting from the bedding nature of sedimentary rocks in particular, is the variation in strength that can exist across a bedding plane. Situations often exist where two rocks with very different strengths meet at a particular bedding horizon causingspecial problems for full face TBMs. In a normal circumstance, the thrust of the TBM is applied equally to most of the cutters on the head of the TBM. However, in a situation where a significant proportion of the tunnel face is made up of a soft rock, the full load being applied to the TBM will be taken by the cutters attempting to penetrate the harder parts of the face. In this circumstance it is not uncommon for the bearings of the cutters to fail at the point where the moving cutter goes from the softer face in an unloaded condition to the harder part of the face, creating an overloaded condition. The real cost of time as well as materials to replace cutters has already been discussed and a similar concept applies to the cost of replacing complete cutters, bearings and housings. At first sight it may appear that this problem could be overcome by having the cutter bearings mounted on some form of 'sprung' base. However, as the efficiency of cutting with a disc cutter TBM is related to the stiffness of the system, any tendency to reduce this stiffness is counter-productive. 11.3.4.5 Hydrology The permeability of either a rock material or rock mass has little significance on tunnel driving where the tunnel is being driven above the water table. However, in circumstances where the tunnel is being driven below the water table, the amount of water ingress to a tunnel can have significant economic repercussions. The effects of permeability are invariably manifested in the effects of water on the tunneling progress. Water is rarely a benefit to tunnel driving and varying quantities can cause varying degrees of problems depending on the circumstances. The effects of water in the excavation area of a TBM tend to be magnified due to the difficulties of removing that water from the excavation area using conventional conveyor systems. This results in much of the water being 'recycled' causing problems out of proportion to the amount of water originally present. Amongst others, water causes the following problems: (i) mixture with fines in the rock spoil to form rock slurry; (ii) washing of material off spoil conveyors causing a build-up of material in the invert; (iii) damage to tunnel boring machine electrical equipment; (iv) difficulties with the application
310
Mechanized Excavation
of sprayed concrete support systems; (v) visual and physical difficulties for the tunnel labour; and (vi) severe wear to moving metal parts. 11.3.4.6 Depth of cover The significance of depth of cover is dependent on rock mass properties. In a circumstance where in situ stress does not cause failure of the opening when excavated, the depth has little relevance. However, where the in situ stress is such that some form of failure occurs, then severe problems can be caused. Failure occurs when the in situ stress exceeds the rock mass strength of the ground around the opening. Failure can vary from the squeezing of 'soft' ground in fault zones or weak beds to the rock bursts found in deep hard-rock metalliferous mines. In intermediate circumstances the in situ stress to competence ratio can affect the 'stand up' time of the ground. In any of these circumstances careful analysis is required to ensure that the correct type of equipment is used for the project concerned. The industry is currently seeking ways of overcoming the problems of the failure of softer ground. Figure 8 shows a recent fully shielded rock TBM that can fully support the ground until full circle support can be erected behind the machine. Whereas this approach appears to overcome some of the problems encountered with TBMs in 'bad' ground, it has a number of drawbacks. Not the least of these are the difficulty of steering the machine in hard rock and the problems of hard blocks of rock entering the anulus between the TBM shield and the driven bore. With all of the above potential problems, it is the marginal change in rock property causing a significant change in TBM performance that is of most interest to the tunnel engineer. Conditions such as fault zones where obvious major difficulties exist can be overcome as long as they do not affect significant proportions of the tunnel. However, there are comparatively minor changes in rock properties that can have significantly beneficial or detrimental effects on tunnel boring machine performance. Starting with a dry tunnel face, water will be added at the head using sprays both to suppress dust and cool the cutters. With minor water ingress the addition of water becomes unnecessary. As the amount of water ingress increases, the first problem to be noted is when restarting after a break. In this circumstance, water has collected in the head and is difficult for the rock removal system to cope with. Increasing amounts of water causes rock slurry to be formed that collects in the invert of the tunnel, requiring manual removal. Further increases in water usually dilute rock slurry allowing it to be pumped from the tunnel. However, once water has reached this level working conditions for both men and machines are usually becoming difficult. 11.4 CONCLUSIONS In a brief resume such as this chapter, space does not permit the detailed analysis that various rock properties may have on the multitude of potential circumstances. It is, however, hoped that the reader will gain an understanding of the economics of TBM driving and the potential effects on these economics of various rock properties. Armed with a basic knowledge of these facts, the rock Torque control system
"7
Thrust cylinders
Support erection area
2-Part telescopic shield
Figure 8 TBM with full shield
The Effects of Rock Properties on the Economics of Full Face TBM s
311
mechanics practitioner should be able to identify those rock properties that are of interest to the tunnel engineer. From this and a knowledge of the current state-of-the-art of TBM design, ways can be sought to quantify those rock properties that are significant to the circumstance so that uncertainties and 'surprises' are avoided when the project is constructed. Designs will continue to evolve and enable more varied conditions to be dealt with economically. However, for many years to come, TBM systems will continue to have limitations. These limitations must be identified and the detailed nature of rock properties that affect them investigated and understood.
ACKNOWLEDGEMENT The drawings that provide the basis for the figures were kindly supplied by Robbins UK Ltd. REFERENCE 1. Barnes A., Whitton T., Fawcett D. F. and Morraw T. M. Tunnelling in Rock Using Full Face Machine in Durban, Natal. In 'Methods & Machines in Underground Construction' - SANCOT SEMINAR (1983).
12 The Design of Support for Underground Excavations PIERRE CHOQUET and JOHN HADJIGEORGIOU Université Laval, Quebec City, Canada 12.1
INTRODUCTION
12.2
METHODS OF DESIGN
12.2.1
314 314 314
Civil and Mining Design Philosophies
12.3 DESIGN OF SUPPORT FOR STRUCTURALLY CONTROLLED ENVIRONMENTS 12.3.1 Reinforcement of Rock Blocks 12.3.2 Reinforcement of Stratified Deposits 12.3.3 The Rock Arch Concept 12.3.4 Rock-bolt Dimensioning Based on Stochastic Modeling
315 315 315 316 316
12.4
317
EMPIRICAL DESIGN BASED ON CLASSIFICATION SYSTEMS
320 321 321 322 322 324 326 326 326 326 328 328
12.4.1 Rock Mass Rating System 12.4.1.1 RMR mining applications 12.4.2 Mining Rock Mass Rating (MRMR) 12.4.3 Q-system 12.4.4 Rock Load 12.4.5 Rock Structure Rating (RSR) 12.4.6 Stand-up Time 12.4.7 RQD Method 12.4.8 Size-Strength System 12.4.9 Rock Mass (MR) 12.4.10 Minimum Rock Bolting Density 12.4.11 Assessment of Classification Systems 12.5
DESIGN BASED ON RULES OF EXPERIENCE
331
Rules of the US Corps of Engineers Rules of Farmer and Shelton Other Empirical Rules
331 332 332
12.6 RATIONAL METHODS OF DESIGN
333
12.6.1 Rock-Support Interaction Analysis 12.6.2 Convergence Control Method 12.6.3 Numerical Modeling 12.6.4 Lining Design
334 334 335 336
12.5.1 12.5.2 12.5.3
12.7 OBSERVATIONAL METHODS 12.7.1 New Austrian Tunneling Method (NATM)
336 336
12.8 SUPPORT SYSTEMS 12.8.1 Design of Concrete and Shotcrete Linings 12.8.1.1 Concrete linings 12.8.1.2 Design of concrete linings 12.8.1.3 Shotcrete 12.8.1.4 Design of shotcrete linings 12.8.2 Steel Arches 12.8.3 Mechanically Anchored Rock Bolts 12.8.4 Cable Bolts 12.8.5 Resin- and Cement-grouted Rock Bolts
337 337 338 338 338 339 339 340 341 341
313
314
Support
12.8.5.1 Resin-grouted rock bolts 12.8.5.2 Cement-grouted rock bolts 12.8.6 Friction Bolts (Swellex) 12.8.7 Friction Bolts (Split Set) 12.9
DESIGN OF SUPPORT FOR EXCAVATIONS IN SWELLING AND SQUEEZING ROCKS
341 342 343 343 344
12.10 SUMMARY
344
12.11
345
REFERENCES
12.1 INTRODUCTION The design of support for underground excavations has often been described as an art as well as a science. This can be construed as a compliment to the judgment necessary in design, as well as a criticism on the present state of knowledge. While the last 30 years have been witness to large developments in analytical and computerized techniques, including three-dimensional numerical modeling, empirical design techniques are still very much in use, particularly in mining. This chapter presents a design methodology for the support of underground excavations. The design process in rock engineering is often complicated due to the lack of control over the geological and stress environments. This is partially compensated by the use of a factor of safety, reflecting both data reliability and the impact of the consequences of failure. In developing an excavation design and support strategy it is traditionally accepted to distinguish between low- and high-stress environments. In low-stress environments, geological and structural discontinuities control excavation behavior and dictate support requirements. The design of support and the choice of excavation shape aim at maintaining the rock mass strength, while at the same time limiting gravity-induced movements across existing geological discontinuities. In increasingly higher field stress environments, stress rather than structural considerations dominate. Under these conditions support design is intended to mitigate the impact of distinct fracture zones (spalling) that tend to develop around the excavation. 12.2 METHODS OF DESIGN In recent years, despite the development of sophisticated analysis tools these have not yet replaced the traditional empirical methods. If anything, a resurgence of empirical design methods of support has been observed. This can be attributed to a great degree to the complexity of the operating geological domain. The rock mechanics engineer is often called to design in what is a divergent domain for which at times all that is available is incomplete information. This gap in knowledge was the driving force in the use of probabilistic [1] and statistical tool development, as well as fuzzy methodology [2, 3]. On the other hand there has been a great dissemination of information from documented case studies, allowing for the creation of databases and thefinetuning of existing design methodology. In general there are three approaches for the design of support in underground excavations. (i) Empirical, which quite often involves some form of ground characterization scheme, accompanied by a set of design recommendations, or employing sets of rules based on acceptable practice. (ii) Rational methods make use of analytical solutions, when available, and use numerical modeling to predict the influence of different support designs on the overall stability of an excavation. (iii) Observational methods call for the instrumentation of the excavation and the implementation of support as the design is developed. This is demonstrated in the New Austrian tunneling Method (NATM) and the Ground Response Curve (GRC). The three methodologies should be considered as essential parts of a unified approach to the design of underground excavations rather than independent techniques [4]. 12.2.1 Civil and Mining Design Philosophies The traditional distinction between civil and mining design of support assumes that civil engineering structures have a longer life span and are often used by the public, or can contain important and expensive facilities. Mining projects, in general, have a shorter excavation life and
The Design of Support for Underground Excavations
315
access is limited to trained personnel, but design considerations are often called to account for considerable fluctuations in stress during the life of a project. The distinction of two independent stages in the design of support, temporary and permanent, has been found to be expensive and inefficient [5]. This practice has stemmed from traditional civil engineering custom whereby two different crews were employed. The first, during the drill-muck cycle was responsible for installing temporary reinforcement, and the second group installed the permanent support. The contractor was generally accountable for the temporary support and for the crew safety. Responsibility for the permanent support design and implementation rested with the engineer who often did not integrate the capacity of the temporary support into the final design. The original distinction between civil and mining design of underground support has become more blurred in recent years. This is partially due to the use of traditional civil classification systems in mining applications, with or without modifications, and the transfer of technology from one domain to the other. In recent years an increasing tendency has been observed towards the use of systematic reinforcement patterns for both civil and mining applications. The ultimate goal of an efficient design is the utilization of the rock mass in such a way that it will render support to itself. It is clearly advantageous to integrate what has been regarded as temporary support into the final design, and for the design engineer to be prepared to use a range of tools and reevaluate design strategy when the quantity and quality of information, which may vary at the different design stages, so necessitates. 12.3 DESIGN OF SUPPORT FOR STRUCTURALLY CONTROLLED ENVIRONMENTS In low-stress environments, in underground excavations close to the surface, it is the structural regime that quite often dictates the possible ground failure mechanism and consequently the choice and design of support necessary to mitigate such failure. While most of the empirical tools discussed in Section 12.4 are applicable, design focuses on the reinforcement of structurally defined blocks of material susceptible to movement. 12.3.1 Reinforcement of Rock Blocks The determination of the influence of reinforcement on the movement of rigid blocks assuming slippage on cohesionless joints can be easily solved by statics [6]. A technique to define the number of bolts necessary to reinforce a rock wedge in an underground excavation under gravity loads, is given in Figure 1 (after [5, 7, 8]). The represented models, chosen partly due to their simplicity, include empirical recommendations regarding bolt spacing and applicable factors of safety. Rock wedge stability is a function of both the discontinuity planes but also of the in situ stresses. The in situ stressfieldcan have a stabilizing influence on such a wedge but its effectiveness is lessened by loosening [9]. A conservative estimate of the bolting force necessary to support a rock wedge has also been successfully employed in practice [10]. Based on observations in foliated rocks, the minimum conditions for wedge formation were found to be a function of the joint orientation as well as the roughness (waviness) of the discontinuities [11]. A powerful reinforcement design tool, for individual rock blocks, is provided by the use of block theory [15]. Key blocks that can initiate local instability are identified and the magnitude and direction of the necessary stabilizing bolting force are determined [16]. A different approach uses a block formulation, based on keystone response, in the design of rock-bolt support with the basic assumption being that the bolt will support the rock block in the same way as another block would [17]. 12.3.2 Reinforcement of Stratified Deposits Excavation reinforcement of stratified rock mass is represented in Figure 2. Design, under these conditions, assumes that the layers behave like slabs or beams [12]. A step-by-step design nomogram, allowing for the dimensioning of reinforcement in a horizontally bedded rock, has proven successful [13]. Other nomograms [14] are also available to assist in the design of bolting patterns and bolt tension of mechanical roof bolting. These are applicable to cases where support is by suspension and the bolt transfers the weight of the weaker strata to the stronger strata, as well as for cases where reinforcement is achieved by friction.
316
Support
Reinforcement of a wedge free to slide under gravity
Reinforcement of a wedge free to fall under gravity
W (f sin /3-cos β ran 4>)-cA Λ/ =
B (cos a tan φ + f sin a)
R = cA + Wcos£ tan (Stillborg [ 7 ] ) /" = 1.5 for grouted bolts f - 2.0 for non grouted bolts (Hoek and Brown [ 5 ] )
N-
= =
number of rock bolts weight of wedge safety factor bolt load bearing capacity dip of the sliding surface cohesive strength of the sliding surface A = base area of sliding surface R - resistance to sliding
(Stillborg C7])
s< Ze (Hoek and Brown [ 5 ] ) w> L + 1.0m (Schach état. C8]) 2<
N W f B ß c
w% f
f<5
a = angle between the plunge of the bolt and the normal to the sliding surface w - width of excavation s = bolt spacing e = joint spacing L - bolt length φ - joint surface angle of friction
Figure 1 Reinforcement of a rock wedge in an underground excavation (after Hoek and Brown [5], Stillborg [7] and Schach et al. [8])
12.3.3 The Rock Arch Concept The concept of a rock arch [12] assumes that rock reinforcement creates a load-carrying arch within the rock mass, stabilizing the tunnel roof (Figure 3). This allows the determination of the type and length of rock bolts as well as possible length reductions due to the arch effect [18]. Spacing recommendations and minimum tension requirements are readily available [7]. Based on earlier work [12], the conclusion was reached that the basic element of a rock bolted roof is the reinforced rock unit. This consists of an individual bolt and the immediately surrounding rock mass. A series of reinforced rock units constitute a reinforced rock structure and stability is ensured if the minimum rock-bolt tension is greater than a value T (Figure 4, after Lang and Bischoff [18]). 12.3.4 Rock-bolt Dimensioning Based on Stochastic Modeling Stochastic modeling techniques can provide a preliminary tool in selecting the design length and spacing of rock bolts to support kinematically isolated wedges of rock [1]. The method depends on determining the joint pattern in an underground excavation, as defined by the mean joint spacing x, the mean joint trace length / and the orientation of joint sets, and considering the span of the excavation. Parametric studies [1] have indicated that a critical bolt spacing (Scrit) and critical bolt
317
The Design of Support for Underground Excavations Reinforcement of a horizontally bedded weak rock
1 e
T h
H
Design procedure based on nomograph
(Panek [ 1 3 ] )
( a ) input bed thickness, e ; (b) choose bolt length, L\ (c) test bolt capacity; (d) determine number of bolts per set; (e) determine spacing of sets, 5 ( f ) input roof span , w ; (g) if Reinforcement Factor < 2 , reiterate design process by increasing number of bolts per set, or decreasing the spacing between sets.
Bed thickness (cm)
Reinforcement of an unstable layer of horizontal bedding planes overlayed by solid rock Solid rock Unstable layer of rock = h W* f-s-ch-p
(Stillborg C7D)
W* weight of rock to be supported by a single bolt
h = thickness of unstable layer of rock
f = safety factor ( l . 5 < f< 3 )
p = rock density
5 ■ bolt spacing, perpendicular to axis of excavation
e - bed thickness
c = bolt spacing, along the axis of excavation
w » width of excavation
Figure 2
Excavation reinforcement of stratified rock (after Stillborg [7] and Panek [13])
length (Ccrit) exist beyond which there is no significant increase in a wedge volume support. Furthermore, as Figure 5 reveals, the influence of joint length on both 5crit and Ccrit decreases. This is important in that it allows for preliminary design in the absence of / values, whose evaluation in the field is difficult. Application of stochastic modeling techniques can also lead to the use of the least number of rock bolts within a particular bolt pattern. Under these conditions the larger individual blocks of rock are supported by rock bolts, while employing wire mesh for the support of unravelling smaller blocks. 12.4 EMPIRICAL DESIGN BASED ON CLASSIFICATION SYSTEMS Empirical design is often based on rules of experience and/or some form of rock mass classification system. Rock mass classification systems and their merits and disadvantages have received
318
Support -0.75Z.
Assumed lower boundary for natural arch
.Loose L zone
Assumed lower r- boundary for | natural arch
0.4 L
The natural arch concept, which develops above the curved roof of an underground excavation in a moderately jointed rock.
The concept of a natural arch, developing above the curved roof of an underground excavation in heavily jointed rock. Use tensioned rock bolts
Use untensioned rock bolts /. = 1.40 + 0 . 1 8 4 * (Schach et al. C83) can use length reductions as in diagram.
Z_ = 1.60+ v/l.0 + 0.012 w2 (Schach et al. LQ1) L/w >2 to ensure development of a compressée zone 5 < Ze (Stillborg C7]) 0.5 5 < Γ < 0 . 8 5 (Stillborg [73 )
L- length of rock bolt (m) w- span of opening (m) e = joint spacing (m) s = bolt spacing (m) f = factor of safety T- applied tension to the bolt (kN) B- bearing capacity of the bolt (kN)
Shotcrete and wire mesh reinforcement necessary.
Figure 3 Reinforcement based on the rock arch concept (after Stillborg [7], Schach et al. [8] and Lang [12])
considerable attention in recent literature [19] and in Volume 3, Chapters 22 and 23 of Comprehensive Rock Engineering. This work will not duplicate existing information but rather will concentrate on the salient aspects of rock mass classifications specifically pertaining to the design of support. Classification systems and their recommendations for support design are not without some degree of controversy. The suitability of a system for a given task, its choice of constitutive parameters, and their assigned weight factors have all been the subject of argument. In favor of such design tools are the wide applications and the existing records of success, while retractors of these suggest that such systems invariably lead to overdesign, and in extreme cases to underdesign. Empirical systems are subjective, despite the use of quantifiable values, and the precise degree of inherent conservatism is not known. Another serious limitation is that such systems reflect 'current and past practice', which may have been influenced by legislation, local practices and particular geological peculiarities and do not necessarily constitute an optimum design methodology. Einstein et al. [20] discussed the necessity of developing a framework for determining the suitability of empirical methods in tunnel design and identified a series of conditions that the empirical methods should strive to satisfy: '(a) they should promote economical yet safe designs; (b) they must correctly be calibrated against test cases and those test cases must be representative of the field of application of future use; (c) they should be complete in that all relevant factors are included, yet they must be practical in that parameters can be determined and with acceptable certainty; and (d) they should have general applicability and robustness to the vagaries of use, yet they must be recognized as subjective.'
The Design of Support for Underground Excavations f
319
i
Reinforced rock unit
.
D
f·:·:·:·:·:·:·: x·:·:·:·:·:·:·!/
/ .
( · : · : ■ : :· : · : · : ·
:·:·:·:·:·:·:·:·| mm 1mm [xvWxl mm I:::·:·:·:·:11
1
j:·:·:·:·:·:·:· ■::·:·:·:·:·:·:Ι övXxXi
»:·:·:·:·:·:·:·
1^-5—H h— s
Span
Minimum rock bolt tension to ensure stability ayAR (ίαηφ)Α-
(■-*)
[£
f]
exp C - (ΐαηφ) * £ / / ? ] exp C - (tan) *Z.//?
Determination of the length of the rock bolts L = w2'* T P φ k
= minimum rock bolt tension - shear perimeter of reinforced rock unit (45) = angle of internal friction for the rock mass - ratio of average horizontal to average vertical stress R - shear radius of the reinforced rock unit (A/P= s/4) w = excavation span
A L s c D a
= = = = =
area of roof carrying one bolt isxs) bolt length bolt spacing apparent cohesion of rock mass height of destressed zone factor depending on time of installation of bolts γ = unit weight of the rock
Notes : 1. If the rock reinforcement is installed prior to the occurrence of significant deformation then it is considered to have an active contribution to the excavation stability (a = 0.5). 2. If passive (more conservative) reinforcement is assumed a= I.O. 3. Cohesion should be taken as zero for initial design.
Figure 4 Schematic of a reinforced rock unit (after Lang and Bischoff [18])
In recognition of the limitations of empirical systems it is suggested that the following procedure be adopted prior to application of these systems for design purposes: (i) review the original paper, and be aware that several systems have been modified by the authors as more data became available (for example the RMR classification [21, 25]); (ii) identify the nature of the constituent database; (iii) distinguish between civil and mining engineering databases, as they involve different degrees of overdesign and possibly different emphases; (iv) compare with more than one system; (v) attempt to verify the design using modeling and field monitoring; and (vi) update recommendations as more information becomes available. The methods that have received considerable attention and have found applications in underground excavation design are summarized in Table 1. The reader is directed to the original publications and Volume 3, Chapter 22 of Comprehensive Rock Engineering for the deriving
320
Support (a)
(b) 9 0 % Volume supported
>
9 0 % Volume supported
0.4
4 U W/l
l
o.i μ
15
30
w/x 5 c r if = critical spacing of rock bolts
w = span of excavation
Ccrir = critical length of rock bolts
7 = mean joint trace length 7 = mean joint spacing
Figure 5 Influence of joint length on critical bolt spacing and length for high volume of support (after Crawford et al. [1]) Table 1 Rock Mass Classification Systems for Underground Excavation Design Name of classification
Author
Country of origin
Applications
Rock Load
Terzaghi [37]
USA
Stand-up Time Rock Quality Designation (RQD) Rock Structure Rating (,RSR) Concept Rock Mass Rating (RMR)
Lauffer [42] Deere et al. [47]
Austria USA
Wickham et al. [40]
USA
Bieniawski [21]
S. Africa
Tunnels with steel supports Tunneling Core logging, tunneling Tunnels with steel supports Tunnels
Q-system
Barton et al. [28]
Norway
Size-Strength
Franklin [49], Louis [50]
Modified Basic RMR (MBR)
Kendorski et al. [23]
UK, France USA
Mining RMR Simplified RMR (SRMR) Rocha system
Laubscher and Taylor [22] Brook and Dharmante [34] Costa-Pereira and Rodrigues-Carvalho [51 ]
S. Africa Sri Lanka Portugal
Tunnels, large chambers Excavation, tunnels Metal mining Mining Mining Tunnels
methodology, evaluation tables and extensive lists of recommendations. Of particular interest to this work are the recommendations pertaining to the choice of support system and the appropriate dimensioning. 12.4.1 Rock Mass Rating System The RMR system, originally proposed by Bieniawski [21], has achieved considerable popularity. The method has evolved over a period of time in response to increased availability of information and at the same time incorporating ISRM recommendations for rock mass characterization. An important contribution of the RMR is that the system has stimulated the development of a plethora of more-specialized systems of ground evaluation, particularly in mining applications [22, 23]. The method is discussed extensively by Bieniawski in Volume 3, Chapter 22 of Comprehensive Rock
The Design of Support for Underground Excavations
321
Engineering. The RMR system considers six parameters: (i) uniaxial compressive strength of rock; (ii) rock quality designation (RQD); (iii) orientation of discontinuities with respect to opening; (iv) spacing of discontinuities; (v) condition of discontinuities; and (vi) groundwater conditions. It is important to note that, given the system's evolution, the latest version [24] should be used in any support estimation. The method does not lead to a quantitative prediction of the rock load, in hard rock conditions, but does lead to a prediction of the stand-up time. The system has been found somewhat conservative, when applied to the Scandinavian data base created by Cecil [26], while it accurately predicted the support of the Tauern and Arlberg tunnel sections in squeezing ground, [20]. In a study in sedimentary rocks in British Columbia, the system provided conservative estimations of the no support (NS) limits [27]. This led to the suggestion that better results could be obtained by accounting for opening size effects RMR(NS) = 22 In ED + 25
(1)
where ED is the equivalent dimension of the span [28]. 12.4.1.1
RMR mining applications
(i) Block cave mining The system has been modified for block cave mining, [23, 24]. It differs from the RMR in that it follows a different adjustment methodology and is capable of making support recommendations at different stages of design. The modified RMR asserts the competence of the rock mass without accounting for the type of opening and is only applicable for isolated single tunnels constructed in geologically similar environments to the production areas. (ii) Coal mining Predictive equations for determining the support pressure (P) in MPa, based on RMR values, the tunnel width (B) in meters and the rock unit weight y in M N m ~ 3 are available for coal mines in the United States (equation 2) [29] and India (equation 3) [30] P = (100 - RMR)/(100y£)
(2)
P = γΒ(Ι.Ί - 0.037RMR + 0.0002RMR2)
(3)
(iii) Semiempirical applications In the design of roof bolting, a semiempirical method has recently been introduced [31]. RMR values and in situ stress information are used in the determination of support. Based on documented field observations the method assumes an elliptically shaped zone of loosening above the excavation (Figure 6), whose geometry and magnitude are influenced by the existing ground conditions and reflected in the material RMR. The determination of the ellipse dimensions and the height of zone in the roof that requires reinforcement is derived by using Figure 6. These are obtained for when the maximum principal stress is horizontal as well as vertical to the excavation. 12.4.2
Mining Rock Mass Rating (MRMR)
The RMR system has led to the development of the Laubscher system, which has been calibrated for mine support design. The system has been well documented, including in Volume 2, Chapter 22 of Comprehensive Rock Engineering, and the different adjustment parameters have been refined [22, 32, 33]. Essentially, the system uses the RMR as its basis and makes allowances for adjustment based on the following: (i) potential for weathering; (ii) combination of joint orientation, mining direction and the potential of blocks to be formed in roof and sidewalls; (iii) mining-induced stresses; (iv) excavation technique and the depth of damage; and (v) intact rock strength. Consequently the Design Rock Mass Strength (DRMS) is derived which can be used in connection with the mining stress to characterize the material and make recommendations for support systems. A simplified rock mass rating system that draws from the work of Laubscher and Bieniawski has also been proposed [34], its most striking difference being the omission of RQD. The system was
322
Support -Maximum bounding ellipse _, \»L
3^1 — ,-.,4-^— ,4τ^7Ρ7/7^7/7Γ/77ν>ί^/
^
^Reduced bounding
'ectangular opening
Maximum principal stress is horizontal
Maximum principal stress is vertical
a b
= kb
a
= (H2+k2 £ 2 ) 0 5
σ'
= Bbx/{tS2-H2)0*
b a1
b'
=(1-0.01 RMR)Z?+/y(0.0l RMR)
b
= a/k =(1-0.01 RMR)a + H (0.01 RMR)
Ht Hl
= b-H =(1-0.01 RMR)/y L
HL Hi
=o-H = (1-0.01 R M R ) / / L
k 2.B 2H RMR
1
ratio of maximum to minimum stresses γ opening width HL height of opening Hi Rock Mass Rating a^a b,bx
= θο/(ο2-Η2)°>
unit weight of material maximum height requiring support reduced height requiring support semimajor axis of ellipse semiminor axis of ellipse
Figure 6 Semi-empirical design for roof bolting (after Stimpson [31])
employed in Sri Lanka and its results were close in range to those obtained by the DRMS. Consequently it has been argued that it can be employed in connection with Laubscher's support recommendations [33]. 12.4.3 Q-system A detailed description of the Q-system is given by Barton et al. [28]. This multiparameter index value Q is the product of three basic characteristics each defined by three parameters: (i) block size (RQD/joint set number); (ii) interblock shear strength (joint roughness number/joint alteration); and (iii) active stress (joint water reduction factor/stress reduction factor). Table 2, based on [28] and [35], provides an empirical estimate of support pressure as a function of the Q-value, allowing its application to the design of support and the dimensioning of reinforcement. The Q-system was originally based on case studies in Scandinavia [25], but it has received wide acceptance, particularly for civil engineering applications. Of interest is the assertion that it is not necessary to increase the support pressure as the cavern dimensions increase. The Q-rating has been found suitable for the selection of support for pressure tunnels with diameters ranging from 2 to 7 m [36]. Based on a number of reported case studies it is possible to select the necessary support using Figure 7 (after Benson [36]). In mining there has been a trend to use a modified version of the Q system, Q'. This involves the elimination of the stress reduction factor by assigning it a value of 1. Q' is used in relation with other parameters in the design of open stopes (see Section 12.8.4). 12.4.4 Rock Load This is a qualitative method by which the ground is divided into nine different classes, and the rock load is structurally estimated [37]. The method was derived and intended for steel sets and timber, based on the state-of-the-art information at the time. While the method is generally regarded as
The Design of Support for Underground Excavations Table 2 Support, Length and Spacing of Reinforcement Based on the Q-system [28, 35] Support pressure
Length and spacing of reinforcement
Proof = (0.2Q- 1/3 )/J r
if the number of discontinuity sets > 2 Proof = (0.2./n- Q- )/3./r if the number of discontinuity sets < 2 Pwall: calculated with the same formulae as Proof, by replacing Q by Q' with: Q' = 5 Q if Q > 10 Q' = 2.5Q if 0.1 < Q ^ 10 Q' = Q if Q <0.1 1/2
Bolts Lroof = (2 + 0.15£)/ESR Anchors £roof = (0.4ß)/ESR 5 = [(CxlO" 3 )/P roof ] 1 / 2
1/3
Bolts Lwa„ = (2 + 0.15tf)/ESR Anchors Lwal, = (0.35tf)/ESR S = [(CxlO- 3 )/P r o o f ]^
J r , joint roughness number; J n , joint set number; Q, NGI classification rating; P, support pressure (MPa); B, span of the excavation (m); H, height of the excavation (m); ESR, excavation support ratio; L, length of reinforcement (m); S, spacing of reinforcement (m); C, load exceeding yield strength of bolt. (kN)
90 Z 60
O
A
-
bU 40
£
^
^
^
"
• U/6 powerhouse or large tunnel A inverted U or horseshoe tunnel o Circular tunnel
1
A
30 20
10 Θ 6 5 4
/ i - / / / ♦ / A
-
- /
/ o
:/
/ °° /
' * °
3 2
-
A
*
·
X - ^
A
*
^ Ή*
* |A
•
A
1
..._l
._
15
10
1
K
** 1
25
30
Span (m)
Classification
Basic final lining
Additional support measures
No support
Un lined
Untensioned grouted anchors locally
Minor support
Unlined
Untensioned grouted rock anchors as necessary Untensioned grouted pattern rock anchors locally Local shotcrete
Moderate support
Shotcrete and pattern bolts
Additional reinforced shotcrete Tensioned grouted bolts or tendons Local concrete Consolidation grouting as necessary
Heavy support
Concrete horseshoe or inverted U
Steel ribs as necessary Reinforced as necessary Pattern consolidation grouting
Very heavy support
Circular concrete
Circular ribs and primary concrete lining Additional reinforcing as necessary Special grouting
Figure 7 Selection of support for pressure tunnels based on the Q-system (after Benson [36])
323
324
Support
conservative, it is of particular interest in that it has influenced subsequent design projects. The concept of rock-load prediction has been used in other subsequent systems and can be converted to support pressure by multiplying with the rock unit weight γ, in MN m " 3 . The immense popularity of the method, particularly in the United States, has ensured that a large number of case studies used in the development of empirical systems were probably designed based on the Terzaghi system (Table 3). The values proposed by Terzaghi have been found to be conservative, and it has been suggested that predicted loads could be reduced for design practices [38, 39]. 12.4.5 Rock Structure Rating (RSR) This method was primarily intended for steel-supported tunnels [40,41]. It is a three-step method that is relatively easy to apply. The three parameters assess the geology, the pattern of jointing and the water inflow and condition of discontinuities. The summation of these parameters is the Rock Structure Rating, which can also be modified to account for mechanically excavated tunnels (Figure 8, after Wickham et al [40]). The RSR method draws heavily from Terzaghi's work to arrive at empirical relationships capable of predicting the existing rock loads. For a given excavated tunnel width B (m) and height H (m) the rock load (kPa) can be obtained by [54] Wx = 0.26 (B + #){[8880/(RSR + 30)] - 80}
(4)
Although there is an absence of sufficient data to develop a high confidence correlation between rock load and bolt spacing, or shotcrete, the following tentative correlations were proposed [40]
Table 3
Rock Load Hp in Meters of Rock on Roof of Support in Tunnel with Width B (m) and Height Ht (m) at Depth of More Than 1.5C, where C = B + Ht (modified after [37])
Rock condition
RQD
Rock load, Ht
Remarks
1. Hard and intact
95-100
zero
2. Hard stratified or schistoseb 3. Massive, moderately jointed
90-99 85-95
0 to 0.5B 0 to 0.255
4. Moderately blocky and seamy
75-85
5. Very blocky and seamy
30-75
0.25ß to 0.35C 0.25B to 0.20CC 0.35 to 1.10C 0.20 to 0.60CC 1.10C 0.60 to 1.10CC 1.10 to 1.40CC
Light lining, required only if spalling or popping occurs. Light support. Load may change erratically from point to point. No side pressure.
6. Completely crushed but chemically intact
7. Squeezing rock, moderate depth 8. Squeezing rock, great depth 9. Swelling rock
a
3-30 0-3 —
1.10 to 2.10C
— —
2.10 to 4.50C Up to 75 meters irrespective of value of C
Little or no side pressure. Considerable side pressure. Softening effect of seepage towards bottom of tunnel requires either continuous support for lower ends of ribs or circular ribs. Heavy side pressure, invert struts required, Circular ribs are recommended. Circular ribs required. In extreme cases, use yielding support,
The roof of the tunnel is assumed to be located below the water table. If it is located permanently above the water table, the values given for types 4 to 6 can be reduced by 50%. Some of the most common rock formations contain layers of shale. In an unweathered state, real shales are no worse than other stratified rocks. However, the term shale is often applied to firmly compacted clay sediments which have not yet acquired the properties of rock. Such so-called shale may behave in the tunnel like squeezing or even swelling rock. If a rock formation consists of a sequence of horizontal layers of sandstone or limestone and of immature shale, the excavation of the tunnel is commonly associated with a gradual compression of the rock on both sides of the tunnel, involving a downward movement of the roof. Furthermore, the relatively low resistance against slippage at the boundaries between the so-called shale and rock is likely to reduce very considerably the capacity of the rock located above the roof to form a bridge. Hence, in such rock formations, the roof pressure may be as high as in a very blocky and seamy rock. c Modified by Deere et al. [38] and Rose [39]. b
The Design of Support for Underground Excavations Parameter A: rock type, strength index and geological structure
325
Maximum value 30
Basic rock type Hard
Medium
Soft
Decomp
Igneous
1
2
3
4
Metamorphic
1
2
3
4
Sedimentary
2
3
4
4
Geological structure
Massive Type I
30
22
15
Type 2
27
20
13
8
Type 3
24
18
12
7
Type 4
19
15
10
6
.ls_ v>
0.2
A
/
3 s ._., \L , 0.2 0.6 1.0 Thickness ( m )
Maximum value 4 5
Strike perpendicular to axis
Strike parallel to axis
Direction of drive
Direction of drive
5 4
9
Parameter B: joint pattern and direction of drive
J 1.0
I0·6
Slightly Moderately Intensely faulted or folded faulted or folded faulted or folded
With dip
Both
Against dip
Dip of prominent joints
Both Dip of prominent joints
II
13
10
12
9
9
7
2
Closely jointed
13
16
19
15
17
14
14
II
3
1 Very closely jointed
9
Moderately jointed
23
24
28
19
22
23
23
19
4 Moderate to blocky
33
32
36
25
28
30
28
24
5
Blocky to massive
36
38
40
33
35
36
34
28
6
Massive
40
43
45
37
40
40
38
34
Flat:0-20 e ; Dipping: 2 0 - 5 0 ° ; Vertical : 5 0 - 9 0 °
Parameter C : groundwater and joint condition Maximum value 2 5 Anticipated
| 3
3 i n f , - i ^ / ^ -i nrr mm i u u m Good None 22 Slight 9 < 0.75 m 3 min ' , y Moderate 0.7515 3.8 m 3 min Heavy Q >3.8 m 3 min
Sum of parameters A +B .44 | 4 5. 7 5
Fair 18
Joint condition Poor Good Fair 25 22 12
Poor 18
15
9
23
19
14
II
7
21
16
12
8
6
18
14
10
~ 10 4
I 15 φ » 20 σ ΰ25 2 30 c £ 3.5
Ê
6 % E σ 8 =Φ 5
.
10 g 1
1
1
1.05
1.10
1.15
1.2!0
^
RSR adjustment factor
Joint condition: good = tight or cemented; fair = slightly weathered or altered; poor = severely weathered, altered or open
Figure 8 Determination of the Rock Structure Rating (after Wickham et al. [40])
spacing of 25.40mm bolts in m = 0.3048(24/WT)05
(5)
spacing of 19.05 mm bolts in m = 0.3048(13.5/lVt)0·5
(6)
shotcrete thickness (mm) = 25.42(1 + WJ125)
(7)
The RSR system was meant to be a guide to the overall quantity of support required and not the exact support to be used at a specific location [40]. The method explicitly excludes squeezing tunnels from its range of applicability, but does not define what is the limiting squeezing condition.
326
Support
12.4.6 Stand-up Time Originally proposed by Lauffer [42, 43] and modified by Linder [44] this system has received considerable popularity in the past. It is a one-parameter system, stand-up time, which is defined as the time during which an underground excavation can remain unsupported without serious deterioration. Stand-up time is influenced by the orientation of the geological structure, the shape of tunnel cross section, the type of excavation and the type of support procedure. The system, however, is based on a single case study and the method can be influenced by the excavation procedures [20]. Bieniawski [45] reports how the stand-up time diagram has been revised to account for the influence of excavation by tunnel boring machine (TBM), and its relationship to RMR ratings [45,46]. 12.4.7 RQD Method Deere and his coworkers have related RQD to support requirements, distinguishing between TBM-driven and drilled and blasted tunnels [38, 47]. The method is applicable at the preliminary stages of design and is less conservative than the Terzaghi method. The limitations of RQD as a single parameter of ground characterization have been duly noted [38, 48]. A series of support recommendations for tunnels of diameter between 6 and 12 m is given in Table 4. The support recommendations were further revised (Figure 9, after [48]). The RQD method is of interest in that it can be used for the preliminary choice of support, as well as a constitutive parameter for more elaborate systems. 12.4.8 Size-Strength System This is a simplified size-strength classification for rocks, whereby the block size and point-load strength can be used to arrive at a single value of support [49]. This is represented in Figure 10, applicable to tunnels of radius 6-8 m. The method allows for a support-excavation strategy and specific support determination. The system proposed by Franklin is similar to that of Louis [50] since the two authors worked together at the initial development stages. The system can be used to assist in the selection of support at the early stages of design and its correct application is restricted to shallow tunnels, less than 300 m in depth. 12.4.9 Rock Mass (MR) The Rock Mass (MR) system, sometimes referred to as the Rocha classification, has proven popular in Portugal but not received much attention elsewhere [51]. The system quantifies information on the joint spacing, joint sets, shear strength and water pressure to determine an index
Excellent 100 quality rockhard, few joints
Tunnel width (m) 6.1 9.1 12.2
18.3
• o x Δ Poor quality rock-closely jointed and or weathered 0
20
30
No support Occasional bolts Pattern bolting Steel ribs
40
Tunnel width (ft)
Figure 9 Support recommendations based on RQD (after Merritt [48])
Table 4 Support Recommendations for Tunnels in Rock (6 to 12 m in diameter) [38] Rock Quality
Excellent3 RQD>90
Tunnelling method
A. Boring machine B. Conventional A. Boring machine B. Conventional
Fair 50 < RQD < 75
A. Boring machine B. Conventional
Poor* 25 < RQD < 50
A. Boring machine B. Conventional
Very poorb RQD < 25 (Excluding squeezing or swelling ground) Very poor (Squeezing or swelling)
A. Boring machine B. Conventional
A. Boring machine B. Conventional
Steel setsb
Rock bolts*
Shotcretec
None to occasional light set. Rock load 0.0 to 0.2Bd None to occasional light set. Rock load 0.0 to 0.3B
None to occasional
None to occasional local application None to occasional local application 50 to 75 mm
None to occasional
Occasional light sets to pattern on 1.5 to 1.8 m center. Rock load 0.0 to 0.4B Light sets, 1.5 to 1.8 m center. Rock load 0.3 to 0.6B
Occasional to pattern on 1.5 to 1.8 m center Pattern, 1.5 to 1.8 m center
Light to medium sets, 1.5 to 1.8 m center. Rock load 0.4 to \.0B Light to medium sets, 1.2 to 1.5 m center. Rock load 0.6 to 1.3B
Pattern, 1.2 to 1.8 m foot center Pattern, 0.9 to 1.8 m center
50 to 100 mm on crown
Medium circular sets on 0.9 to 1.2 m center. Rock load 1.0 to \.6B Medium to heavy sets on 0.6 to 1.2 m center. Rock load 1.3 to 2.0B
Pattern, 0.9 to 1.5 m center
100 to 150 mm on crown and sides. Combine with bolts
Pattern, 0.6 to 1.2 m center
150 mm or more on crown and sides. Combine with bolts
Medium to heavy circular sets on 0.6 m center. Rock load 1.6 to 2.2B Heavy circular sets on 0.6 m center. Rock load 2.0 to 2.8£
Pattern, 0.6 to 1.2 m center
150 mm or more on whole section. Combine with medium sets 150 mm or more on whole section. Combine with medium to heavy sets
Very heavy circular sets on 0.6 m center. Rock load up to 75 m Very heavy circular sets on 2 foot center. Rock load up to 75 m
Pattern, 0.6 to 0.9 m center Pattern, 0.6 to 0.9 m center
Pattern, 0.9 m center
None to occasional local application 50 to 75 mm Occasional local application 50 to 75 mm
100 mm or more on crown and sides
The Design of Support for Underground Excavations
Good* 75 < RQD < 90
Alternative support systems
150 mm or more on whole section. Combine with heavy sets 150 mm or more on whole section. Combine with heavy sets
327
a In good and excellent quality rock, the support requirement will in general be minimal but will be dependent upon joint geometry, tunnel diameter, and relative orientations of joints and tunnel. b Lagging requirements will usually be zero in excellent rock and will range from up to 25% in good rock to 100% in very poor rock.c Mesh requirements usually will be zero in excellent rock and will range from occasional mesh (or straps) in good rock to 100% mesh in very poor rock. d B = tunnel width in m.
328
Support Degree of support number
E 1000
2
II
3
4
5
6
7
E «υ a)
w
*-!
^.1
Bolt spacing (m)
c E 10
100
Compressive strength, ac (ΜΡα)
circumference bolted
(Assuming major principal stress σ, = 2ΜΡα)
Shotcrete thickness (cm) Contour numbers indicate the degree of support applicable to rock with given size-strength properties. Letters A-G indicate the 'degree of support class'. Degree of support
circumference 5 0 H shotcreted ιοο ->
A B C D E F G Heading area 1 —. ^ — Shield 2 b Full cross section Admissible advance before support 4 m - * - 3 m * 2 . 3 m - » l . 6 m * L 3 m * 0 . 8 m Stand-uptime
Years Months
Hours
(unsupported)
I 10
Ό
I 100
Minutes
Ribs per 100 m of tunnel
I 100 10 negligible
Figure 10 Support recommendations based on the Size-Strength classification (after Franklin [49])
rating (MR). This can be used in the determination of the rock load requiring support or can be used directly to assist in the choice of support strategy (Figure 11). 12.4.10 Minimum Rock Bolting Density Choquet and Charette [52] determined the rock bolting densities in 10 Quebec hard rock mines, in more than 57 drift portions, assessing ground conditions by means of the Q and the Mining RMR systems. Figure 12 indicates the minimum bolting density required (i.e. the minimum amount of support judged as necessary) as a function of the classification systems employed. It was recognized that overdesign existed in some of the surveyed drifts, as clearly demonstrated by the presence of data values well above the minimum density line. The equation of the line of minimum required support is as follows D = - 0.227 In Q + 0.839
(8)
D = - 0.0214 MRMR + 1.68
(9)
s = l/D{0.5
(10)
where D = number of bolts per square meter of roof and wall (if the latter is bolted), Q = rating of rock mass according to NGI classification, MRMR = mining RMR [33] and s = bolt spacing (m). The above equations were found applicable for conditions in the Canadian Shield and were based on spans of drifts between 2.8 m and 7.5 m, with the majority between 3.5 m and 5.5 m. Depths of drifts surveyed varied between 50 m and 1000 m, with the majority between 100 m and 500 m. 12.4.11 Assessment of Classification Systems The different empirical systems have been reviewed more extensively elsewhere [45, 53, 54]. The fact remains that the RMR and Q systems have received the most attention, particularly for civil
329
The Design of Support for Underground Excavations Joint spacing 25
e (cm) ? P.
50
100
200
H
I
—i—
15
0
25
20
Joint sets Four or more joint sets and random joints
Three joint sets and random joints
Three joint sets
Two joint sets and sparse joints
One joint set and sparse joints
Without joints or sparse joints
10
15
20
25
Shear strength Continuous stiff clayey filling or flat walls coated by low friction angle minerals
Continuous soft clayey filling
Φ
Flat walls and silty or siltysandy filling —
7.5
P-
I —
Smooth flat and fresh walls; or rough and slightly weathered walls 1
22.5
30
10
15
1
Rough flat and fresh walls , 375
Rough and irregular or undulating or discontinuous and fresh walls 1
65
H
—h-
25
20
Water pressure Erodible filling, water pressure >IOkg cm-2
Erodible filling, water pressure 2.5 kg cm-2
Erodible filling, Not erodible fillingJ Not erodible filling, Impervious rock water pressure water pressure water pressure mass or water table > 5kg cm-2 I kg cm-2 | > 10 kg cm-2 J[ 1 below tunnel floor
3 6 Rock Mass Quality { MR)
20
30
40
Obtaining
Class
III
50
60
70
80
90
15
12 ΡΛ+Ρ. +Pr
+Pn
100 MR
k from (MR) Support recommandations
MR
80-100
0-0.05
Sporadic supportiex··rockbolting, in accordance with the observed roof conditions
60-80
0.05-0.3
Systematic support in the roof
50 - 6 0
0.3-0.6
Systematic support in the roof. Sporadic support in the walls may also be necessary
IV
30-50
0.6-0.9
V
0-30
0.9-1
Determination of the rock load acting on the support for MR > 60
Systematic support in both roof and walls is necessary
hc- kw hn - 0
for 50 < MR < 6 0
0=
for M R < 50
hn~ 0.5h c
n<
0.5/?c
Figure 11 Rock mass classification system (after Costa-Pereira and Rodrigues-Carvalho [51])
engineering applications, hence it was inevitable that attempts would be made to formulate some form of correlation between the two systems (Table 5). It is evident that the two systems do not necessarily correlate in different environments, particularly at low and high values. This can be attributed to sampling variations and/or the influence of different parameters under different conditions.
Support
330
1.4 cvj 1
E
|
0 5-3
\ /
0 6-1
9 - m \
1.2
"ô £ i.o
Line of minimum rock bolting density
Si09-6 \o9-3
0 6-2 **
01-3
QD
S. D , "6133-5
03-β
a ° 10-7
X 2-210.3 4-4D V e O DO 03-2
0)
ting density P P
0 5-1
09-2 D* - 0 . 2 2 7 In 0 + 0.839 V /
1.6
o •ß 0.4 o
l - l .mine 1, drift I 1-2 :mine 1, drift 2
S 0.2
10-7: mine 10,drift 7
_, π D 77 '
7-47-3 0
J_
1 0.1
0.01
g5"2 4
] 1
' 2:β;ΐ&ηβ"5 » 10 40 100
Q 1 y , 1. f
1.6 1.5
<^
— a i.i >* 1.0 £ 0.9 — "ö 0.8 — ? 0.7 -
06-I D9-î\^
1-2
Ξ 06 o ■° 0.5 o
0.4
*
0.3 0.2
05-3
0 = - 0.0214 MRMR+1.68
1.4
Έ 1.3 75 o
Ö5H
D9-2
-
09-6/V /9-50
/
rock bolting density Line of minimum
06-2 09-3 v 03-5
01-3
NJC -6 03-8 53-5 l 0 - , l0-3 Ι0-7\03-7β^|0.4
αβ-l
I
N^H°Oe-3D|.p3-9D5.5 >Q2-2D3-2
-
OS C^fp/S
1 -1 : mine 1, drift 1 1-2: mine 1, drift 2
-
3
"3 0
0
1 20
.
L_
i
40
D 7 H
D5 -6
0 ? 4 Ü2-J4-I
10-7: mine 10, drift 7
0.1
010-5
60
\ 7 · 4 S-2 2-6 £4/7-3/ -2 6-5
ο-—α>οο-α-α 80
1
100
MRMR
Figure 12 Minimum rock bolting density based on the Q and Mining RMR classification systems (after Choquet and Charette [52]) Table 5 Correlation between RMR and Q Systems under Different Conditions Correlation
Source of case studies and reference
Comments
RMR = 13.5 log Q + 43 RMR = 9 In Q + 44 RMR = 12.5 log Q + 55.2 RMR = 5 In Q + 60.8 RMR = 43.89 - 9.19 In Q RMR = 10.5 In Q + 41.8 RMR= 12.11 log Q +50.81 RMR = 8.7 In Q + 38
New Zealand [55] Diverse origin [25] Spain [56] S. Africa [57] Spain [58] Spain [59] Canada [60] Canada [27]
RMR = 10 In Q + 39
Canada [52]
Civil engineering tunnels Civil engineering tunnels Civil engineering tunnels Civil engineering tunnels Mining tunnels, soft rock Mining tunnels, soft rock Mining tunnels, hard rock Civil engineering tunnels, sedimentary rocks (assumed normal distribution) Mining tunnels, hard rock
In another investigation [54], a comparison of Terzaghi's, RQD, RSR, RMR and Q methods at a 21m span, shallow excavation was found not to be greatly influenced by the subjectivity of the user. This, however, was attributed to the dampening of the rating differences when they were linked to the support recommendations.
The Design of Support for Underground Excavations
331
0.001
!o. •Ξ 0.2 "o £ 0.3 9 C
£ 0.4 V)
I 0.5 Q. 0.6 o "0.7
2 0.8
30
40
50
60
70
80
90
100
RMR
Figure 13 Preliminary support recommendations for large underground excavations at depth (after Hoek [61])
Both the RMR and Q systems, when used with information on the maximum compressive stresses of an opening and the intact strength of the rock, can be used to determine preliminary support recommendations for large underground excavations at depth (Figure 13, after [61]). Attention has also been drawn to the fact that the determination of support requirements directly from classification systems bypasses investigations into the excavation mechanics [62]. Consequently it is possible that unpredicted complications could arise during construction. In general, despite their limitations, empirical systems form an integral part in the design strategy of support for underground excavations. It is felt that when applied in full knowledge of their restrictions, and with reference to related case histories, they can be an invaluable aid in design.
12.5 DESIGN BASED ON RULES OF EXPERIENCE While classification systems incorporate design support recommendations of an empirical nature, there also exist several rules of thumb applicable to the selection of reinforcement type and dimensions. These can be used independently of classification systems and analytical design, but perhaps are best used as comparative tools to allow the choice of spacing and bolt length.
12.5.1 Rules of the US Corps of Engineers The rules were established after examination of more than 68 case histories of rock reinforcement in underground chambers, tunnels and shafts [6]. Widths of openings surveyed varied between 4.5 and 30 m and heights between 4 and 60 m. Depths were moderate, generally not exceeding 150 m. No mine openings were surveyed. These rules are presented in Tables 6 and 7. They allow for the estimation of length, spacing and support pressure. The rules, however, only give a preliminary configuration for rock reinforcement, which must be checked, analyzed and, as necessary, modified to meet the requirements of a specific rock reinforcement design. The use of Tables 6 and 7 requires, to start with, one assumed value for one of the variables, length or spacing. The next step is to verify that all specifications are met by going through the table as many times as is necessary. Table 7 provides values of support pressure which can be directly used for the calculation of bolt spacings. The working load of bolts to be used should be at yield point, as assumed by the Corps of Engineers. Taking a fraction of the yield limit load of the bolts would bring an additional factor of safety to the one already included in the projects surveyed.
332
Support Table 6 Minimum Length and Maximum Spacing for Rock Reinforcement [6] Parameter
Empirical rules
Minimum length
Greatest of: A. Two times the bolt spacing B. Three times the width of critical and potentially unstable rock blocks8 C. For elements above the springline: (i) Spans less than 6 m - 1/2 span (ii) Spans from 18 m to 30 m - 1/4 span (iii) Spans 6 m to 18 m - interpolate between 3 m and 4.5 m lengths, respectively D. For elements below the springline: (i) For openings less than 18 m high - use lengths as determined in C above (ii) For openings greater than 18 m high - 1/5 the height
Maximum spacing
Least of: A. 1/2 the bolt length B. 1-1/2 the width of critical and potentially unstable rock blocks C. 1.8 mb
Minimum spacing
0.9 to 1.2 m
a
Where the joint spacing is close and the span is relatively large, the superposition of two bolting patterns may be appropriate, e.g. long heavy bolts on wide centers to support the span and shorter and thinner bolts on closer centers to stabilize the surface against ravelling due to close jointing. b Greater spacing than 1.8 m would make attachment of surface treatment such as chain link fabric difficult. Table 7 Minimum Average Support Pressure for Rock Reinforcement [6] Parameter
Empirical rules
Minimum average support pressure at yield point of elements
Greatest of: Above springline: A. Pressure equal to a vertical rock load of 0.2 times the opening width8 B. 0.04 MPab Below springline: A. Pressure equal to a vertical rock load of 0.1 times the opening height0 B. 0.04 MPad
8
For example, if the unit weight of the rock is 0.023 M N m - 3 and the opening span is 2.5 m, the required support pressure is 0.2 x 25 x 0.023 = 0.115 MPa. bFor the maximum spacing of 1.8 m, this requires a yield strength of approximately 142 kN. cFor example, if the unit weight of the rock is 0.026 MNm" 3 and the cavity height is 45 m, the required support pressure is 0.1 x45 x 0.026 = 0.117 MPa. dThis reinforcement should be installed from the first opening excavated prior to forming the intersection. Stress concentrations are generally higher at intersections, and rock blocks are free to move toward both openings.
12.5.2 Rules of Farmer and Shelton These rules, presented in Figure 14 (after Farmer and Shelton [63]), are based on various authors' experience [64-66]. They provide design guidelines for the length and spacing of rock bolts for excavations in rock masses having clean, tight discontinuity interfaces and a maximum of three discontinuity sets. 12.5.3 Other Empirical Rules Table 8 provides a list of current reinforcement practices in different countries [33], while Table 9 provides empirical guidelines for temporary roof reinforcement [67] by means of rock bolts.
The Design of Support for Underground Excavations Number of Excavation span (m) discontinuity sets
Bolt design
<2 inclined at 0 - 4 5 ° îo horizontal
Comments
Z_ = 0.3£ S = 0.5L(depending on thickness and strength of stata) Install bolts perpendicular to lamination where possible with wire mesh to prevent flaking
< 2 inclined at 4 5 - 9 0 ° For side bolts: L> h sin ψ ( if installed perpendicular to discontinuity)'^ > h tan ψ {\1 to horizontal installed horizontally). See figure below for h and ψ ■ L = bolt length ; s = bolt spacing ; B =
excavation
Φ^,//////////////// L>h tan¥
>I5
<2
Roof bolting as above. Side bolts designed to prevent sliding along planar discontinuities. Spacing should be such that anchorage capacity is greater than sliding or toppling weight. Bolts should be tensioned sufficiently to prevent sliding
L=2s 5 = 3 - 4 x block dimension. Install bolts perpendicular to excavation periphery with wire mesh to prevent flaking
Bolts should be installed quickly after excavation to prevent loosening and retain tangential stresses. Prestresses should be applied to create a zone of radial confinement. Sidewall bolting where toe of wedge daylights in side wall
Lis0.'5Bl
primary bolting
L2=0.3S,
secondary bolting
Primary bolting conforms to smaller excavation design. Secondary (and tertiary) bolting supplements primary design (See figure below)
5| = 0.5 L| S2=0.5L2
Install wire mesh to prevent
spoiling
Z.i = 0.3£| primary bolting
< 3 with clean.tight interfaces
The purpose of bolting is to create a load-carrying beam over the span. Fully bonded bolts create greater discontinuity shear stiffness. Tensioned bolts should be used in weak rock, subhorizontal tensioned bolts where vertical discontinuities occur
. / //// //////
////-wU-Z_>/7sin* f
< 3 with clean,tight interfaces
333
5| = 0.5Δ|
5 2 =3/4 x block size: secondary bolting
Primary bolting should have sufficient capacity to restrain major blocks. Decisions on block size for secondary bolting should be left to the section engineer
L2=2S2
Figure 14 Empirical guidelines for the dimensioning of rock bolts (after Farmer and Shelton [63])
Table 8
Reinforcement Dimensioning, Rules of Thumb (after Laubscher [33])
Parameter
Empirical rules
Minimum length
A. Greater than half the width of excavation (S. Africa) B. Twice the bolt spacing (Australia) C. Three times the width of critical and potentially unstable rock blocks defined by average joint spacing in the rock mass (Australia) D. For spans of 18-30 m a length equal to 1/4 of the roof span; or, for excavations higher than 18 m, sidewall bolts equal in length to 1/5 of the wall height (Australia) A. 0.5 of bolt length (Australia), or B. 1.5 x width of critical and potentially unstable rock blocks defined by the average joint spacing in the rock mass (Australia) 2.5 m x 0.25 m x 0.004 m 250 mm x 250 mm x 10 mm
Spacing and orientation
Straps Bearing plates
12.6 RATIONAL METHODS OF DESIGN Rational design methods make use of strain and stress analyses for underground excavations with the solution methodology employing both analytical and numerical techniques. Hambley and Kendorski [68] have applied a simplified, somewhat conservative, analytical strategy to the reinforcement of circular underground openings up to 12 m in diameter. The method relies on estimating the Rock Mass Strength Determination (RMSD) [69], accounting for dilation
334
Support Table 9
Temporary Roof Bolting Design Recommendations (after Coates and Cochrane ^'^)
Parameter
Empirical rules
Load capacity of bolt
Use the smaller value of:
Length of bolt
Use the greater value of:
(a)ßa>es (a)L>lm
(b)ea>er (b)L>D
Maximum recommended length of bolt:
(a)ifr>0.5ß, (b) if Γ < 0.503 Bolt spacing
usQ
L
Maximum bolt spacing: s> 3e (a) when L > 4w (b) when L < 4w
use s < 0.9L use s < 0.5L
ßa, load capacity of bolt; , load capacity of the steel; , ultimate load of the bolt; 7, rock density; β, discontinuity spacing; w, span of the excavation; s, bolt spacing; T, tensile load on bolt; D, depth of the block containing collar of the hole.
effects, and determining the thickness of a loosened zone surrounding the excavation. Consequently the necessary reinforcement to stabilize the loosened zone is determined.
12.6.1
Rock-Support Interaction Analysis
The characteristic Hnes method of describing and analyzing the mechanical interaction between rock and tunnel support, for years employed as a quahtative tool, has, in the last 15 years, developed into a quantitative tunnel design support tool [ 7 0 - 7 2 ] . G r o u n d response is represented by a ground reaction curve a n d the lining by a support reaction curve. A simplified schematic version of the g r o u n d - s u p p o r t reaction curve is presented in Figure 15 (after [73]). O p t i m u m design is achieved when the pressure required to Hmit deformation is counteracted by the pressure available from the support. The available solutions to the ground reaction curve (GRC) have been summarized in [74]. The face remains, however, that the determination of the G R C for compHcated geotechnical material remains difficult. Solutions to the support stiffness and bearing capacity for different hnings are also available [5, 75, 76]. The selection of a support system necessitates the assessment of its ductility, strength a n d stiffness. In a combined support system it is the softest support component that dominates the resultant effective support. F o r axisymmetric problems it is possible to employ analytical and iterative solutions, while for nonaxisymmetric conditions it is necessary to resort to numerical methods. Hoek a n d Brown [ 5 ] provide a simpHfied methodology to determine the interaction of the g r o u n d - s u p p o r t system. While this method makes simplifying assumptions, it can be used to determine the required support Hne as well as the maximum support pressure for a concrete/shotcrete Hning, for blocked steel sets and for rock bolts and cables. The method has been appHed with particular success for creeping ground conditions [77].
12.6.2
Convergence Control Method
The design of grouted bolt reinforcement can also be undertaken, based on a control of convergence model [78]. The effectiveness of grouted bolts was assessed in terms of convergence reduction and the model was found valid for weak rock ( R M R < 40) for axisymmetric tunnels in homogeneous material behaving in an elastic, brittle plastic fashion. Displacement was restricted by using a bolt pattern, β greater than 0.15 and an L/a ratio greater t h a n 0.8
The Design of Support for Underground Excavations
335
In situ stress prior to excavation
Radial deformation Support reaction curves ( = load induced in support by deformation of excavation ) 1. Stiff support installed too early attracts excessive load. 2. Effective support at pressure A required to limit deformation = pressure available from support tunnel and support system in equilibrium. 3. Ineffective support not stiff enough and installed too late.
Figure 15 Simplified ground support reaction curve (after Douglas and Arthur [73])
where d = bolt diameter (m), λ = friction factor for bolt-ground interaction, a = tunnel radius (m), SL = longitudinal bolt spacing (m), ST = tangential bolt spacing (m), β = bolt density parameter and L = length of bolt (m). 12.6.3 Numerical Modeling Numerical modeling as a design tool in rock mechanics is covered extensively in Volume 2 of Comprehensive Rock Engineering. Recent years have seen a significant increase in the number as well as in the sophistication of numerical codes developed and applied to rock engineering. The International Journal of Rock Mechanics and Mining Sciences has in fact devoted an entire issue to listing and summarizing the characteristics and capabilities of currently available numerical codes applicable to rock mechanics [79]. In the design of support systems, numerical models are often used to arrive at quantitative solutions. In the absence of complete input data, they can be applied qualitatively to perform parametric studies, and in sensitivity trials, identifying the influence of specific parameters on different design options. A summary of and an introduction to the computational techniques used in rock mechanics for continua (finite element, finite difference and boundary element methods) and discontinua (discrete element) are given in [80]. The selection of the appropriate tool is a function of both the knowledge of existing geotechnical conditions as well as the ability of a particular code to accurately represent these in a way that can lead to a successful design. The limitations and capacity of different methods is covered in Volume 2 of Comprehensive Rock Engineering. It is of interest to note a trend, in the modeling of complex situations, towards hybrid computational codes coupling together a combination of meshes of the different methods. Lorig et al. [81] have used a distinct element model coupled with a boundary element model for the analysis of tunnels in jointed rock, while a hybrid boundary-finite element model has also been employed successfully [82, 83]. These methods take advantage of the reduced data requirements of the boundary element analysis while using the finite or distinct element to represent certain types of conditions. The subgrade reaction model, otherwise referred to as the Beam Element Method, can also be used to analyze a tunnel lining for simple ground and excavation conditions. The method assumes elastic support, whereby spring elements simulate normal and shear stresses [84]. In practice, numerical models are used to define the prevalent stresses and/or to explicitly determine the influence of rock reinforcement on the overall stability of an underground excavation. The support element representation, in the numerical model, and its limiting assumptions are of
336
Support
particular importance. The mechanical representation of rock reinforcement for explicit finite difference codes has been dealt with in detail elsewhere [85, 86]. Where the joint spacing in the reinforcement direction is greater than the development length it is possible to apply a local reinforcement representation. This utilizes force-displacement relations to describe both the shear and axial behavior of reinforcement across discontinuities. Alternatively, a global reinforcement model separates the entire reinforcement length into a series of lumped masses and springs. The suitability of three-dimensional over two-dimensional codes has recently received attention. The controlling elements in the selection of such codes still remain the complexity of the structure and the quality of input information. A trend has also been observed towards minicomputers. This is attributed to economic factors as well as to the advance of the state-of-the-art in microcomputer technology. A consideration that has always to be maintained is that there could be a difference in what constitutes a successful research code and a design tool which is called to assist in the design of support. 12.6.4 Lining Design Analytical methods applicable to the determination of thrusts and bending moments in linings are summarized by Szechy [87]. It has been argued, however, that the traditional methods for estimating tunnel-liner capacity predict nonconservative bending moments and could possibly lead to erroneous results [88]. The design process could be improved by considering the influence of variations in the ground pressure on the calculated bending moments. This allows the comparison between the resulting liner thrusts, bending moments and the reduced liner capacity [88]. A full evaluation of this method, however, necessitates further field data on the magnitude and extent of ground pressure variations and the subgrade modulus, which are difficult to determine in the field. 12.7 OBSERVATIONAL METHODS Observational methods utilize monitoring as an integral constituent of the design process [89]. Monitoring techniques are also employed to complement empirical, analytical and numerical methods. The aim is to determine ground response to reinforcement, allowing the early identification of possible problem areas and the verification of the implemented design. Instrumentation procedures have been reviewed by Dunnicliff [90]. For deep tunnels, constructed in rock, monitoring aims to measure excavation convergence, support performance and structure stability. The available technology for recording rock movements and the equipment used in measuring support pressures, applied loads and strains have recently been reviewed in [91] and [92], respectively. Table 10 summarizes reinforcement monitoring systems [93] while the deformation monitoring systems are listed in Table 11. Monitoring of any structure, either in the form of follow-up investigations or as an integral part of the design, as is the case in the New Austrian Tunneling Method (NATM), makes for good engineering practice. 12.7.1 New Austrian Tunneling Method (NATM) The different stages in the evolution of what has come to be known as the New Austrian Tunneling Method (NATM) have been traced elsewhere [94]. The method has suffered somewhat from different interpretations which has led to several works attempting to clarify and interpret the fundamentals of the NATM [95-97]. Current consensus is that the NATM is more of a design philosophy than a method or a classification system, even though it involves qualitative ground characterization. Any given NATM classification involves a level of detail directly dependent on the available site geological and geotechnical information, restricting its applicability to the given tunnel for which it was developed and modified. The NATM integrates fundamental rock behavior under load, monitors tunnel performance during construction and revises the design of support requirements as the encountered conditions dictate. The method has been mostly used in tunnels of diameters between 10-12 m. The implemented support systems, always including shotcrete, have been arrived at empirically. A simplified arrangement of NATM principles is given in Figure 16 (after Sauer and Gold [94]). The developed excavating classes, based on ground characterization, constitute an integral part of the contractual agreement between client and contractor [98]. Consequently, eificient application of
The Design of Support for Underground Excavations Table 10
337
Reinforcement Monitoring Systems (after Norris and Yearby [93])
Method
Principle of operation
Comments
Indicator washer Indicator: signal of telltale bolt
Deflects at predetermined load Point anchored above the normal bolt anchor horizon and protrudes freely from the hole Compression spring below bolt head raises a plastic indicator Torque required to turn nut Wire stretched between two adjacent roof bolt heads Direct measurement of compression between bolt head and bearing plate Measurement of bolt strain at any point along bolt
Inexpensive indication of bolt load Indicates vertical strata movement
Eaton 'flag' system Nondestructive bolt test Horizontal roof strain indicator Titanium load cell Vibrating-wire or mechanical strain gauges Acoustic integrity testing
Gives qualitative information on bolt load Qualitative information on bolt load Determination of horizontal strain and roof sag Accurate determination of bolt loads for point anchored systems Accurate determination of strain and hence load at a point for fully bonded bolts or along whole bolt for point anchored systems Indicates debonding of resin anchored systems
Base of the rock bolt is struck to induce acoustic vibrations. Monitors receive the signal and signal strength will depend upon integrity Mechanical device working on spring Indicates bolt movement compression and linkages Change in relative length of free wire Indicates strain and hence load in running to the end of hollow rock bolt point anchored rock bolt from free end Hydraulic jack measures load to pull Measures resin bond or mechanical out rock bolt anchorage strength
Mine roof movement monitor Hollow bolt indicator Pull testing
Table 11
Deformation Monitoring Systems
System
Device
Comments
Convergence meters
Dial gauge extensometer
Borehole extensometers
Vibrating-wire extensometer Tensioned tape Photoelastic disc Single point
Quick, economical, can be used in restrictive areas
Borehole inclinometers
Multiple point Strain gauged pendulum Servo accelerometers Pendulum with vibrating wire Pendulum with rheostat
Direct measurement of displacement magnitude, moderate or low angles Applicable for recording lateral movements
the NATM necessitates that the involved parties have the experience and willingness to engage in the necessary contractual agreements. 12.8 SUPPORT SYSTEMS An installed support system usually involves the use of reinforcement of the ground (pattern bolting, cables, etc.) and support provided by steel sets and shotcrete. The choice of support can be made based on any of the approaches - empirical, rational and observational - discussed in this chapter. The merits, applicability and disadvantages of the different support systems are summarized in [5] and [99]. 12.8.1 Design of Concrete and Shotcrete Linings The design of linings is influenced by the employed excavating technique and the resulting ground disturbance, the elapsed time between excavation and support installation, the geological structural
338
Support Avoidance of micro
Figure 16 Simplified arrangement of NATM principles (after Sauer and Gold [94])
conditions and the flexibility of the design support system. Concrete and shotcrete (pneumatically applied concrete) are commonly used to provide tunnel support for both civil and mining structures.
12.8.1.1
Concrete linings
Concrete tunnel linings are cast in situ or are segmentai. A comprehensive, qualitative review of concrete linings is given by Whittaker and Frith [100]. An inherent advantage of in situ placed concrete is that it can be designed to accommodate any desired shape of cross section. The lining is designed to function in compression in order to minimize the need for reinforcement. Circular segmental rings also provide an immediate permanent lining of great strength. This is so provided that when erected they can be brought into close contact with the excavated ground by grouting injection or otherwise. All the segmental types of concrete provide, immediately on erection, strong support and adequate flexibility. The timing of grouting operations is significant in developing the interaction between lining and ground.
12.8.1.2
Design of concrete linings
Szechy [87] has collected a series of analytical solutions for the structural design of concrete linings which account for uniform and nonuniform loadings of underground excavations of circular and other geometries. The design of the required concrete strength is covered clearly in a majority of concrete texts and manufacturers' literature. For the case of shaft lining design or other circular linings a series of simplified solutions is available to determine the required thickness [101].
12.8.1.3
Shotcrete
Shotcrete differs from concrete cast in place by its higher compaction and its lower water/cement ratio. Its application is often made on the recommendation of the different empirical systems, and it is used exclusively or in combination with other reinforcement methods. The use of shotcrete as a measure of support is an integral part of design philosophies such as the NATM.
The Design of Support for Underground Excavations
339
Good design of shotcrete takes into consideration its time-dependent effects, mix design, layer thickness and possible use of wire mesh. Two types of shotcrete are in use: dry-mix shotcrete, where the mix is dry and water is added at the nozzle; and wet-mix concrete, where the water is already added at the mixer (accelerator must be added at the nozzle). The typical mix by dry weight [7] consists of: cement (15-20%), coarse aggregate (30-40%), fine aggregate (40-50%) and accelerator (2-5%). The water/cement ratio for dry-mix shotcrete lies in the range 0.3:1 to 0.5:1 and is adjusted by the operator to suit local conditions. Accelerators allow the concrete to achieve early high strength, prevent sagging and sloughing of the shotcrete during application, reduce rebound and increase plasticity of the mix. The addition of 50 mm long and 0.4-0.8mm diameter steel fibers has been found to improve the toughness, durability and shear andflexuralstrengths of the shotcrete. and to reduce the formation of shrinkage cracks. More elaborate discussions on shotcrete characteristics and application procedures are given in [102-105]. 12.8.1.4 Design of shotcrete linings Shotcrete when first applied lacks strength. It is thus necessary to ensure that the stand-up time of the rock is longer than the setting time of shotcrete. The classification systems, Section 12.4, make support recommendations on the range of applications where shotcrete is suitable. The required shotcrete thickness differs for temporary and final support and is based on qualitative ground assessment and RQD ratings [106]. Rules of thumb regarding the thickness of the shotcrete, as applied in different countries, have been summarized as follows [53]: in Austria 1/40 to 1/50 of tunnel diameter; in Sweden 3 to 8 cm applied immediately behind the face in jointed rock; and in Germany 10 cm for tunnels up to 10 m in diameter. Alternatively the thickness of shotcrete is determined by using design loads, as predicted by the empirical systems [37,40]. If, however, shotcrete is applied at the early stage, the full predicted loads will not be allowed to develop and a layer of thickness from 50 to 150 mm, applied in 50 mm layers, will suffice [105]. A design procedure is also available to calculate the necessary thickness of shotcrete, i, to support a rock wedge liable to failure at the crown of a tunnel (see Figure 17, after Fernandez-Delgado et al. [107]) t = ^/[2r c (sine)/ c28 L]
(12)
where W = the weight of the rock block, fc28 = the 28-day unconfined compressive strength of the shotcrete, L = the length of rock block perpendicular to the plane of Figure 17, Tc = the thrust coefficient, given by the ratio of the axial layer load at failure to the maximum compressive strength times the cross-sectional area of the layer, and Θ = the abutment angle, defined in Figure 17. If shotcrete is used with rock bolts then the load is normally carried by the rock bolts and shotcrete is not considered as structural support. When shotcrete is employed as a structural reinforcement tool, it can be analyzed by numerical techniques. Modeling should account for the absence of a transfer of load from the rock mass to the shotcrete. In most cases the shotcrete, by locking together discontinuous blocks of material, develops a composite rock/shotcrete structure. This underscores the need for accurate determination of the in situ strength and stiffness of the composite rock/shotcrete structure to be used with any numerical model. Shotcrete has also successfully been employed as a permanent lining of a 15.4 km railway tunnel [108]. For this work, up to five shotcrete machines were employed simultaneously, achieving lining rates of 400 tunnel meters per month.
12.8.2 Steel Arches Steel arches are widely used to support roadways in coal mines, where they are often required to sustain quite large deformations, and in civil and mining environments that demand the support of high loads. Qualitative descriptions of the applicability of the different types of steel support are available [100]. In 1946 Proctor and White produced a design manual for the design of steel arches that is still pertinent today [37]. In the same reference Terzaghi presented his original rock classification system intended to be a tool in the design of steel arches. The RSR system [40] has been shown to be
Support
340 30°
Equivalent shotcrete geometry Joint set I °
Joint set 2
= 15°
ù=r, 0=30°, Tc=OA
b = \r, 0=15°, Tc =0.3
Figure 17 Determination of shotcrete thickness to support a rock wedge (after Fernandez-Delgado [107])
particularly applicable to the design of steel arches. This involves the use of a parameter called Rib Ratio (RR), which is 100 times the ratio of the theoretical spacing for a rib size, divided by the actual spacing for the same rib size as reported in the RSR database. The RSR can be linked to the RR by the following empirical equation (RSR + 30) (RR + 80) = 8800
(13)
The design of support using steel arches entails considerations of arch profile and steel characteristics (moment of inertia, allowable stresses) which are readily available from the manufacturers. The most popular types of arches are rigid and yielding. The design of rigid arches is undertaken using strength of material theory. Analytical solutions, dependent on the loading assumptions and blocked constraints, are available [109, 110]. Rigid arches are also readily analyzed using numerical [111] and physical modeling techniques [112]. Yielding arches are statically indeterminate. Consequently, empirical design often incorporating in situ measurements of deformation may be useful [110-112]. Steel arches behave as a passive support system; consequently, blocking assumes importance in ensuring that rock loads are transferred uniformly onto the steel sets. Choquet [113, 114] has analyzed the results of full-scale load testing trials for both rigid and yielding arches, and has shown that preliminary support value may be obtained for a horseshoe-shaped arch by F = 3.07 x 105(WX + 5
Wy)A
F = 5.82 x 10 (^ x + Wy)A
(yielding arches)
(14)
(rigid arches)
(15)
where F = maximum load carried by the arch at collapse (kN), A = interior area of steel arch (m2), Wx = modulus of inertia (m3), from manufacturer's data and Wy = modulus of inertia (m3), from manufacturer's data. The proper application of steel arches necessitates the use of blocking, usually with wood, to ensure proper distribution of load on the ribs and minimize undue bending. 12.8.3 Mechanically Anchored Rock Bolts These consist of a steel rod with one threaded end to which an expansion shell can be fitted and the other end is a forged head or threaded to fit a nut. Bolts are used with a head plate and tensioned by tightening mechanically. A more general description of mechanically anchored rock bolts can be founding, 7, 115].
The Design of Support for Underground Excavations
341
Design of reinforcement uses information on the pull-out strength of the bolts for the rock mass type for which the system is being designed. Procedures used to perform the pull-out tests are described in [116,117]. Pull-out test results [115] have confirmed the strong influence of drill-hole diameter tolerance on the pull-out strength. For instance, expansion shells designed for use in drill holes of 31.7 mm diameter may result in drastic reduction in pull-out strength if used in drill holes of dimensions exceeding a tolerance of + 1.6 mm or —0.8 mm. The system can apply a support pressure (P) up to (16)
pull-out /scS\
where P = support pressure in M Pa, Ppuii-out = pull-out load of bolt system in MN, sc = circumferential rock bolt spacing in m, and sx = longitudinal rock bolt spacing in m. 12.8.4 Cable Bolts The design of cable bolting reinforcement is given more extensively in Volume 4, Chapter 16 of Comprehensive Rock Engineering and by Xanthakes [118]. The design can assume the presence of isolated blocks (see Section 12.1) or make use of rules of thumb or empirical classification systems (see Section 12.4 and 12.5). Empirical approaches have been particularly successful in the design and dimensioning of cable bolts for open stopes in Canadian mines [119]. 12.8.5 Resin- and Cement-grouted Rock Bolts While other substances can also be used as grout, cement and resin are currently the most popular. Bolts have been designed to be grouted with polyester-resin cartridges previously introduced in the drill hole or with pumped cement grout. The selection of appropriate grout is assisted by considering the inherent advantages of each type (Table 12). 12.8.5.1 Resin-grouted rock bolts The grouted length with a resin cartridge can be determined through manufacturer's data, based on cartridge diameter and drill hole diameter (Table 13). Furthermore, it is necessary for the required grouted length to provide sufficient adherence of the bolt to the drill hole. This will ensure that in the event of failure during loading of the bolt this will occur in the bolt and not along the rock-grout or rod-grout interfaces. A simple empirical relationship can be used [73] provided that the bond length remains greater than 400 mm L = 2.5 P + 50
(17)
where L = required grouted length (mm), and P = working load of reinforcing element (kN). Table 12 Polyster resin
Advantages and Disadvantages of Polyester Resin and Cement Grout Cement grout
Advantages
Disadvantages
Advantages
Disadvantages
Quick installation (resins) Quick setting (1-30 minutes) Possibility of tensioning bolts using two resins with different setting times Very high holding power
Relatively high resin cost Average storage time (12 months) Resin vapors toxic to skin and eyes
Low cost of cement
Longer installation time
High holding power
Slow setting time
Good protection against bolt corrosion
More difficult installation in holes drilled upwards
Good protection against bolt corrosion Ease of installation
Decrease in mechanical properties with an increase in temperature Setting time varies with temperature Resins are flammable
Lack of control over grout quality (segregation) and anchor (when end portion of hole not full)
342
Support Table 13 Guidelines for the Selection of Grouted Length with a Resin Cartridge 30 cm Long
Bolt diameter (mm)
22
25
28
15 20 25 30 35 45
310 406
406
406 508
25
30
32 330 457
32
t\ L
1.8 L 1.6 L 1.4 L 1.2 1.0 l·0.8 L 0.6 L 0.4 0.2 r
35
Diameter of the cartridge 35
305 432
356
406 559
305 457
38
40
38
40
Diameter of the drill hole
^
\
5
\
305
45
48
51
330 57
—10.056
J
Drill hole: | 1/4' \ . (31.75 mm )
\ 4
^
J 0.042 A 0.028 A -i
V
*>*J
u I
381
■1
u
1.6
432 508
-A
Bolt: Γ (25.4 mm)
\
40
2 1
1 . .!__ 1 1 1 i l
3.2 4.8 Low
9.6
16
1
0.014
> ,JJJL
32 48
I Average | High
1
L_JL
96
160
| Very high |
Compressive strength of the enclosing rock (MPa)
1. Granite 2. Limestone
3. Sandstone 4. Coal
5. Chalk
Figure 18 Adherence values based on rock type and strength (after Franklin and Woodfield [120])
Alternatively, adherence factors are derived from pull-out tests and are usually given in grouting length per anchoring strength. Adherence values based on rock type and strength are also provided in Figure 18 (after Franklin and Woodfield [120]) and in [121]. Results based on pull-out tests, grouted over the whole length of the bolt or over 30 cm, are given in [122]. In other series of pull-out tests [123] it was observed that: (i) bars grouted in drill holes 6.35 mm greater than their diameter have a pull-out load equal to the bolt rupture load and display stiff pull-out curves; and (ii) a difference in diameter of 6.35 mm greatly facilitates the tearing of the cartridge envelope and proper mixing of the resin.
12.8.5.2
Cement-grouted rock bolts
Previous empirical work on the determination of the adhesion stress used in the design of cement grouted anchors is summarized in Table 14. While there exist sophisticated solutions to the determination of the required grouted length [124], for most purposes the values in Table 14 can be used in relation with the following equation [130] G = Ρ/τχ10 3 πΖ)
(18)
where P = failure load of steel rod (kN), D = diameter of drill-hole (m). G = required grouted length (m), and τ = smaller value of Trod_grout and Trock_grout. Trod_grout is the adhesion resistance along the rod-grout interfaces, Trock_grout is the adhesion resistance along the rock-grout interface and/ c is the uniaxial compressive strength of the grout.
343
The Design of Support for Underground Excavations
Table 14 Empirical Relationships for the Determination of the Adhesion Stress (τ) Used in the Design of Cement Grouted Bolts Steel-grout contact
Comments
Rock-grout contact
Comments
Smooth rod:
Values incorporate a factor of safety between 2.0 and 2.5 [124]
Deformed rod:
Use a factor of safety of 3
^ultimate = O - ^ c
[126]
τ = <7c/30 (MPa)
Values incorporate a factor of safety of 3 [128]
τ = 0.20/ c (max 1.1 MPa) Deformed rod: τ = 0.10 fc (max 2.4 MPa)
max 4.2 MPa for fc > 42 MPa
Smooth rod:
Minimum values [125]
τ = / c /30 (MPa) imax < 1.3 (MPa)
τ = 0.17/c (MPa) Deformed rod: τ = 0.96 (/ c ) 0 · 5 (MPa) Ribbed rod: τ = 0.17 ( / c )
05
Use of average value recommended [129] Use minimum of rock-grout or steel-grout
lower bound
τ = 0.50 (/ c ) 0 · 5 average
Ribbed rod: 0 5
τ = 0.17 (/ c ) · lower bound τ = 0.50 (/ c ) 0 · 5 average
Φ = 38 mm
1
1
1
1
40 41 38 33 Hole diameter (mm)
o r
IN)
o
'
—
t
/ /
\ \
o
^ // //
o
\ ^
Pull-out resistance I k N m " 1 )
//
o
3
ull-out resistance ( k N m - 1 ) — ro o o
P = 30 MPa
-
Use of average value recommended [129] Use minimum of rock-grout or steel-grout
20
25 Pump pressure (MPa)
30
Figure 19 Pull-out resistance of Swellex bolts as a function of the drill hole diameter and the inflation pressure (after manufacturer's data [131])
Alternatively, the grouting length (G), including a safety factor of 2.0 to 2.5, for a ribbed bar can be taken as a function of the rock quality and bar diameter [126] : sound rock, G = 30 bar diameters; fissured rock, G = 40 bar diameters; and weathered rock, G = 60 bar diameters. The choice of the diameter of the drill hole can vary from 12.7 mm to 25.4 mm larger than the bar diameter. 12.8.6 Friction Bolts (Swellex) The Swellex (registered trademark of Atlas Copco MCT AB, Sweden [131]) friction bolt is characterized by its variable holding power, influenced by both rock type and installation technique (diameter of drill hole and inflation pressure) [132]. This is demonstrated in Figure 19, after manufacturer's data [131]. The minimum anchoring length (L) can be obtained by L = (supported load at rupture)/(pull-out resistance)
(19)
Results of pull-out tests on Swellex bolts have been reported by the manufacturer. 12.8.7 Friction Bolts (Split Set) The Split Set (registered trademark of Ingersol Rand Co., USA [133]) friction bolt is inserted into the drill holes with a percussion drill. The holding power of Split Set bolts increases with time as
344
Support
Z
120 I 10 100 90 80
Stabilizers
Z ^ο % o °σ> S
τ?
ô I
60
50 Normal installation range
40 30
20
10 38
37
35
Hole diameter (mm)
Figure 20 Holding power at installation of SS-39 Split Set bolts as function of the drill hole diameter (after manufacturer's data [133])
a result of corrosion on the surface of the bolt barrels, improving friction characteristics, and as a result of ground displacements that may help wedge the bolts into the drill holes. For holding power immediately after installation, the main parameter that must be met is the diameter of the drill hole, as in Figure 20 from manufacturer's data for SS-39 bolts [133]. Holding power drops by a factor of 2.5 when the diameter of the drill hole exceeds the nominal diameter of 35 mm by more than 2.5 mm. Results on pull-out tests are readily available [134, 135]. 12.9 DESIGN OF SUPPORT FOR EXCAVATIONS IN SWELLING AND SQUEEZING ROCKS The particular problems associated with the design of support in swelling and squeezing rocks perhaps merit special attention. Swelling in rocks is caused by a combination of physicochemical reactions involving water and stress relief. Squeezing results from the influence of initial stress concentrations on shales and other argillaceous rocks, resulting in rock failure. Both processes are time dependent and can result to some degree in tunnel convergence and closure. Local geology, state of stresses and tunnel geometry have to be accounted for in the design of support in this environment. An overview has recently been presented [136]. Empirical design methods can explicitly allow for swelling and squeezing considerations [37], can provide limited information, as in the Q-system [28] where squeezing ground conditions are inadequately represented, or can specifically exclude such conditions from their range of applicability. Analytical and numerical solutions as applied to squeezing and swelling conditions can be limited by the absence of adequate empirical information and on the basis of the intrinsic assumptions [136]. Analytical solutions accounting for the influence of long-term rock deformation on lining pressure, using both the characteristic line and lining-rock interaction methods are available [76]. The actual design of support can be active aiming in the prevention of swelling and squeezing by employing chemical inhibitors or steel arches and rock bolts. Alternatively the effort can be applied to the containment of defofmation and can include the removal of ground. At times a combination of the two techniques is the most appropriate. The specifics of the problem often dictate the integration of design and construction. This explains the origins and the popularity of the NATM in squeezing and swelling ground. 12.10 SUMMARY This chapter has dealt with the design of underground support systems for both civil and mining applications. It is now apparent that the fulfillment of the main task, i.e. the successful design of a support system, can be accomplished by rational, empirical and observational methods. The presented techniques are but tools in this design process. Rather than argue the superiority of one method over another, it is felt that most techniques can be employed in a complementary fashion with their applicability influenced by the prevalent conditions, state of knowledge and presence of
The Design of Support for Underground Excavations
345
qualified personnel. The ultimate goal is the design of a safe and economical support system for the life of the project.
12.11
REFERENCES
1. Crawford A. M., Ng L. and Lau K. C. The spacing and length of rock bolts for underground openings in jointed rock. In Proc. 5th Int. Conf. Numerical Methods in Geomechanics, Nagoya, Japan (Edited by Z. Einsenstein) pp. 1293-1300 (1985). 2. Fairhurst C. and Lin D. Fuzzy methodology in tunnel support design. In Proc. 26th U.S. Symp. Rock Mech., Rapid City, SD (Edited by E. Ashworth), pp. 269-278. Balkema, Rotterdam (1985). 3. Nguyen V. U. Some fuzzy set applications in mining geomechanics. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 22, 369-379 (1985). 4. Duddeck H. Guidelines for tunnel design. Tunnels and Tunneling 21, 12, 42-44 (1989). 5. Hoek E. and Brown E. T. Underground Excavations in Rock. Institution of Mining and Metallurgy, London (1980). 6. U.S. Army Corps of Engineers. Engineering and Design: Rock Reinforcement, Engineering Manual EM 1110-1-20907. Available from the Office of the Chief of Engineers, Washington, DC 20314 (1980). 7. Stillborg B. Professional Users Handbook for Rock Bolting. Trans Tech Publications, Series on Rocks and Soil Mechanics, vol. 15 (1986). 8. Schach R., Garschol K. and Heltzen A. M. Rock Bolting: A Practical Handbook. Pergamon, Oxford (1979). 9. Crawford A. M. and Bray J. W. Influence of the in situ stress field and joint stiffness on rock wedge stability in underground openings. Can. Geotech. J. 20, 276-287 (1983). 10. Douglas T. H., Richards L. R. and Arthur L. J. Dinorwic power station: rock support of underground caverns. In Proc. 4th Int. Congr. Rock Mech., Montreux, Switzerland, vol. 1, pp. 361-369. Balkema, Rotterdam (1979). 11. Cording E. J. and Mahar J. W. Index properties and observations for design of chambers in rock. Eng. Geol. 12, 112-142(1978). 12. Lang T. Rock reinforcement. Bull. Assoc. Eng. Geol. 9, 215-239 (1972). 13. Panek L. A. Design for bolting stratified roof. Trans. Soc. Min. Eng. AIME 229 (1964). 14. Tang D. H. Y. and Peng S. S. Methods of designing mechanical roof bolting in horizontally bedded strata. In Proc. 25th U.S. Symp. Rock Mech., Evanston, IL, pp. 615-626. AIME, New York (1984). 15. Goodman R. E. and Shi G. Block Theory and its Application to Rock Engineering. Prentice-Hall, Englewood Cliffs, NJ (1985). 16. Shi G. and Goodman R. E. Keyblock bolting. In Proc. Int. Symp. Rock Bolting, Abisko, Sweden (Edited by O. Stephansson), pp. 143-167 (1983). 17. Warburton P. M. Implications of keystone action for rock bolt support and block theory. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 1A, 283-290 (1987). 18. Lang T. A. and Bischoff J. A. Stabilization of rock excavations using rock reinforcement. In Proc. 23rd U.S. Symp. Rock Mech., Berkeley, CA (Edited by R. E. Goodman and F. E. Heuze), pp. 935-944. AIME, New York (1982). 19. Kirkaldie L. (Ed.) Rock Classification Systems for Engineering Purposes, Proc. ASTM STP 984, p. 167 (1988). 20. Einstein H. H., Steiner W. and Baecher G. B. Assessment of empirical design methods for tunnels in rock. In Proc. Conf. Rapid Excavation and Tunneling, Atlanta, GA, pp. 683-706. AIME, New York (1979). 21. Bieniawski Z. T. Engineering classification of jointed rock masses. Trans. S. Afr. Inst. Civ. Eng. 15, 335-344 (1973). 22. Laubscher D. H. and Taylor H. W. The importance of geomechanics of jointed rock masses in mining operations. In Proc. Symp. Exploration for Rock Engineering, Johannesburg, South Africa (Edited by Z. T. Bieniawski), pp. 119-128. Balkema, Rotterdam (1976). 23. Kendorski F., Cummings R., Bieniawski Z. T. and Skinner E. Rock mass classification for block caving mine drift support. In Proc. 5th Int. Congr. Rock. Mech., Melbourne, pp. B51-B63. Balkema, Rotterdam (1983). 24. Kendorski F. S., Cummings R. A., Bieniawski Z. T. and Skinner E. A rock mass classification scheme for the planning of caving mine drift supports. In Proc. Conf. Rapid Excavation and Tunneling, Chicago, IL, pp. 193-223. AIME, New York (1983). 25. Bieniawski Z. T. The Geomechanics Classification in rock engineering applications. In Proc. 4th Int. Congr. Rock Mech., Montreux, vol. 2, pp. 41-48. Balkema, Rotterdam (1979). 26. Cecil O. S. Correlation of Rockbolts-Shotcrete Support and Rock Quality Parameters in Scandinavian Tunnels. Ph.D. Thesis, University of Illinois, Urbana, IL, pp. 414 (1970). 27. Kaiser P. K., MacKay C. and Gale A. D. Evaluation of rock classifications at B. C. Rail Tumbles Ridge tunnels. Rock Mech. Rock Eng. 19, 205-234 (1986). 28. Barton N., Lien R. and Lunde J. Engineering classification of jointed rock masses for the design of tunnel support. Rock Mech. 6, 189-236 (1974). 29. Unal E. Design Guidelines and Roof Control Standards for Coal Mine Roofs. Ph.D. Thesis, Pennsylvania State University, University Park, PA, p. 355 (1983). 30. Venkateswaelu V. Geomechanics Classification of Coal Measure Rocks vis-a-vis Roof Supports. Ph.D. Thesis, Indian School of Mines, Dhanbad, p. 251 (1986). 31. Stimpson B. A simplified conceptual model for estimating roof bolting requirements. Int. J. Min. Geol. Eng. 7,147-162 (1989). 32. Laubscher D. H. Geomechanics - classification of jointed rock masses - mining applications. Trans. Inst. Min. Metall. 86, A1-A8 (1977). 33. Laubscher D. H. Design aspects and effectiveness of support systems in different mining conditions. Trans. Inst. Min. Metall. 93, A70-A81 (1984). 34. Brook N. and Dharmante P. G. R. Simplified Rock Mass Rating system for mine tunnel support. Trans. Inst. Min. Metall. 94, A148-A154 (1985).
346
Support
35. Barton N. Cavern design for Hong Kong rocks. In Proc. Seminar Rock Cavern - Hong Kong, pp. 179-202. IMM (1988). 36. Benson R. P. Design of unlined and lined pressure tunnels. In Canadian Tunneling, pp. 37-65 (1987/1988). 37. Proctor R. V. and White T. L. Rock Tunneling with Steel Supports. With an Introduction to Tunnel Geology by K. Terzaghi. The Commercial Shearing and Stamping Company, Youngstown, OH (1946). 38. Deere D. U., Peck R. B., Monsees J. E. and Schmidt B. Design of tunnel liners and support systems. Dept. of Civil Engineering, Univ. of Illinois, for U.S. Dept. of Transportation, contract no. 3-0152, p. 287 (1969). 39. Rose D. Revising Terzaghi's tunnel rock load coefficients. In Proc. 23rd U.S. Symp. Rock Mech., Berkeley, CA (Edited by R. E. Goodman and F. E. Heuze), pp. 953-960. New York (1982). 40. Wickham G. E., Tiedemann H. R. and Skinner E. H. Support determination based on geologic predictions. In Proc. Conf. Rapid Excavation Tunneling, pp. 43-64. AIME, New York (1972). 41. Skinner E. H. A ground support prediction concept: the Rock Structure Rating (RSR) model. In Rock Classification Systems for Engineering Purposes, Proc. ASTM STP 984, 35-49 (1988). 42. Lauffer H. Gebirgsklassifizierung fur den Stollenbau. Geol. Bauwes 74, 46-51 (1958). 43. Lauffer H. Die neure Entwicklung der Stollenbautechnik. Oesterreichische Ingenieur Zeitschrift 3, 13-24 (1960). 44. Linder R. Spritzbeton im Felshohlraumbau. Bauternik 40, No. 10, 326-331, No. Il, 383-388 (1963). 45. Bieniawski Z. T. Engineering Rock Mass Classifications, p. 251. Wiley, Chichester (1990). 46. Lauffer H. Zur Gebirgklassifizierung bei Frasvortrieben, Felsbau 6 (3), 137-149 (1988). 47. Deere D. U., Hendron A. J., Patton F. D. and Cording E. J. Design of surface and near surface construction in rock. In Failure and Breakage in Rock (Edited by C. Fairhurst), pp. 237-302. AIME, New York (1967). 48. Merritt A. H. Geologic prediction for underground excavations. In Proc. Conf. Rapid Excavation Tunneling, pp. 601-622. AIME, New York (1972). 49. Franklin J. A. Safety and economy in tunneling. In Proc. 10th Can. Symp. Rock Mech., Kingston, Ontario, vol. 1, pp. 27-53 (1975). 50. Louis C. Reconnaissance des massifs rocheux par sondages et classifications geotechniques des roches. Ann. Inst. Tech. Paris 108, 97-122 (1974). 51. Costa-Pereira A. S. and Rodrigues-Carvalho J. A. Rock mass classifications for tunnel purposes - Correlation between the systems proposed by Wickham et ai, Bieniawski and Rocha. In Proc. 6th Int. Congr. Rock Mech., Montreal (Edited by G. Herget and S. Vangpaisal), pp. 841-844. Balkema, Rotterdam (1987). 52. Choquet P. and Charette F. Applicability of rock mass classifications in the design of rock support in mines. In Proc. 15th Can. Symp. Rock Mech., Toronto, pp. 39-48 (1988). 53. Tunneling Technology. Ontario Ministry of Transportation and Communications, p. 166 (1976). 54. Einstein H. H., Thompson D. E., Azzouz A. S., O'Reilly K. P., Schultz M. S. and Ordun S. Comparison offiveempirical tunnel classification methods - Accuracy, effect of subjectivity and available information. In Proc. 5th Int. Congr. Rock. Mech., Melbourne, pp. C303-C313. Balkema, Rotterdam (1983). 55. Rutledge J. C. and Preston R. L. New Zealand experience with engineering classifications of rock for the prediction of tunnel support. In Tunnelling under Difficult Conditions (Edited by I. Kitamura), pp. 23-29. Pergamon, Oxford (1978). 56. Moreno Talion E. Comparison and application of the geomechanics classification schemes in tunnel construction. In Tunneling '82 (Edited by M. J. Jones), pp. 241-246. Institution of Mining and Metallurgy, London (1982). 57. Cameron-Clarke I. S. and Budavari S. Correlation of rock mass classification parameters obtained from borecore and in-situ observations. Eng. Geol. 17, 19-53 (1981). 58. Tamames Celada B. Fourteen years of experience on rock bolting in Spain. In Proc. Int. Symp. Rock Bolting, Abisko (Edited by O. Stephansson), pp. 295-311 (1983). 59. Abad J., Celada B., Chacon E., Gutierrez V. and Hidalgo E. Application of geomechanical classification to predict the convergence of coal mine galleries and to design their supports. In Proc. 5th Int. Congr. Rock Mech., Melbourne, pp. E15-E19. Balkema, Rotterdam (1983). 60. Udd J. E. and Wang H. A comparison of some approaches to the classification of rock masses for geotechnical purposes. In Proc. 26th U.S. Symp Rock Mech., Rapid City, SD (Edited by E. Ashworth), pp. 69-78. Balkema, Rotterdam (1985). 61. Hoek E. Geotechnical design of large openings at depth. In Proc. Conf. Rapid Excavation and Tunneling, pp. 1167-1185. AIME, New York (1981). 62. Cording E. State of the art: rock tunneling. In Tunneling in Soil and Rock, pp. 77-106. ASCE, New York (1984). 63. Farmer I. W. and Shelton P. D. Factors that affect underground rockbolt reinforcement systems design. Trans. Inst. Min. Metall. 89, A68-A83 (1980). 64. Rabciewics L. Stability of tunnels under rock load. Water Power 21, 266-273 (1969). 65. Lang T. A. Theory and practice of rock bolting. Trans. Am. Inst. Min. Metall. Pet. Eng. 220, 333-348 (1961). 66. Alexander L. and Hosking A. D. Principles of rockbolting - Formation of a support medium. In Proc. Symp. Rockbolting, Australasian Institute of Mining and Metallurgy, Paper 1, Illawara (1971). 67. Coates D. F. and Cochrane T. S. Development of design specifications for rock boltingfromresearch in Canadian mines. Research Report R224, Mining Research Centre, Energy, Mines and Resources Canada, p. 30 (1970). 68. Hambley D. F. and Kendorski F. S. Design of rock reinforcement systems for underground openings based on critical assessment of geological conditions. In Proc. 14th Can. Symp. Rock Mech., Vancouver, CIM Special Volume 30, pp. 100-105 (1982). 69. Kendorski F. S. Field and laboratory assessment of rock mass strength for tunnel design with allowance for dilation. In Proc. 13th Can. Symp. Rock Meek, Toronto, CIM Special Volume 22, pp. 162-167 (1980). 70. Ward W. H. 18th Rankine lecture: Ground supports for tunnels in weak rocks. Geotechnique 28, 133-170 (1978). 71. Fairhurst C. and Daemen J. J. K. Practical inferences from research on the design of tunnel supports. Underground Space 4, 297-311(1980). 72. Brady B. H. G. and Brown E. T. Rock Mechanics for Underground Mining. Allen & Unwin, Boston (1985). 73. Douglas T. H. and Arthur L. J. A Guide to the Use of Rock Reinforcement in Underground Excavations, CIRIA Report 101. Available from CIRIA, 6 Storey's Gate, Westminister, London SW1P 3AU (1983). 74. Brown E. T., Bray J. W., Ladanyi B. and Hoek E. Ground response curves for rock tunnels. J. Geotech. Eng. 109,15-39 (1983).
The Design of Support for Underground Excavations 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103. 104. 105. 106. 107. 108. 109. 110. 111. 112. 113. 114.
347
Lombardi G. Dimensioning of tunnel linings with regards to constructional procedure. Tunnels and Tunneling 5, 340-351 (1973). Choquet P. Determination of the characteristic line of steel arch support. In Proc. 7th Int. Conf. Strata Control, Liège, Belgium, pp. 127-148 (1982). Landanyi B. and Gill D. E. Design of tunnel linings in a creeping rock. In Canadian Tunneling, pp. 39-50 (Ï990). Indraratna B. and Kaiser P. K. Design for grouted rock bolts based on the convergence control method. Int. J. Rock Mech. Min. Sei. ά Geomech. Abstr. 27, 269-281 (1990). Rock Engineering Software. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 25, 183-252 (1988). Brown E. T. Analytical and Computational Methods in Engineering Rock Mechanics, pp. 1-28. Allen & Unwin, London (1987). Lorig, L. J. Brady B. H. G. and Cundall P. A. Hybrid distinct element-boundary element analysis ofjointed rock. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 23, 303-312 (1986). Beer G. Application of 3-D boundary element and coupled analysis in geomechanics: case studies. Proc. 6th Int. Conf. Numerical Modelingin Geomechanics., Innsbruck, pp. 2209-2216 (1988). Beer G., Watson J. O. and Swoboda G. Three-dimensional analysis of tunnels using infinite boundary elements. Comp. & Geotech. 3, 37-58 (1987). Gnilsen R. Numerical methods. In Underground Structures: Design and Instrumentation (Edited by R. S. Sinha), pp. 84-128. Elsevier, Amsterdam (1989). Lorig L. Rock reinforcement: mechanical representation and use in finite difference schemes. In Proc. Int. Symp. Underground Engineering, New Delhi (1988). Brady B. and Lorig L. Analysis of rock reinforcement using finite difference methods. Comp. & Geotech. 5, 123-149 (1988). Szechy K. The Art of Tunnelling. Akademia Kiado, Budapest (1973). Kaiser P. K. and Barlow P. J. Rational assessment of tunnel liner capacity. Canadian Tunneling, pp. 31-44 (1986). Peck R. Advantages and limitations of the observational method in applied soil mechanics. Rankine lecture. Géotechnique 19, 171-187 (1969). Dunnicliff J. Underground excavations. In Geotechnical Instrumentation for Monitoring Field Performance, pp. 453-466. Wiley, Chichester (1988). Franklin J. and Chrzanowski A. Rock movements. In Mine Monitoring Manual (Edited by J. Franklin), Special Vol. 42. pp. 109-117. Canadian Institute of Mining and Metallurgy (1990). Franklin J., Belshaw D., Brown B., Cain P. and Choquet P. Support pressures, loads and strains. In Mine Monitoring Manual (Edited by J. Franklin), Special Vol. 42, pp. 118-123. Canadian Institute of Mining and Metallurgy (1990). Norris C. and Yearby M. Roof bolt developments. Colliery Guardian Coal International 229, 22-26 (1981). Sauer G. and Gold H. NATM ground support concepts and their effect on contracting practices. In Proc. Conf. Rapid Excavation and Tunneling, AIME, Colorado, pp. 67-86 (1989). Muller L. The reasons for unsuccessful applications of the New Austrian Tunnelling Method. In Tunnelling under Difficult Conditions (Edited by I. Kitamura), pp. 67-72. Pergamon, Oxford (1978). Muller L. Removing misconceptions on the New Austrian Tunneling Method. Tunnels and Tunneling 10, Oct., 29-32 (1978). Brown E. T. Putting the NATM into perspective. Tunnels and Tunneling 13, Nov, 13-17 (1981). Golser J. and Mussger K. The New Austrian Tunneling Method (NATM), contractual aspects. In Tunnelling under Difficult Conditions (Edited by I. Kitamura), pp. 387-392. Pergamon, Oxford (1978). Hoek E. and Wood D. Support in underground hard rock mines. In Proc. 13th Can. Symp. Rock Mech., Montreal, CIM Special Vol. 22, pp. 1-6 (1980). Whittaker B. N. and Frith R. C. Tunneling: Design, Stability and Construction. Institution of Mining and Metallurgy, London (1990). Zahary G. and Unrug K. Reinforced concrete as a shaft lining. In Proc. 8th Can. Symp. Rock Mech., Toronto, pp. 265-282. Energy, Mines and Resources, Ottawa (1973). Brekke T. L. Shotcrete in hard-rock tunneling. Bull. Assoc. Eng. Geol. 9, 241-264 (1972). Connell J. P. State of the art of shotcrete. In Underground Mining Methods Handbook (Edited by W. A. Hustrulid), pp. 1561-1566. AIME, New York (1982). Mason E. E. and Mason R. Shotcrete. In Tunnel Engineering Handbook (Edited by J. O. Bickel and T. R. Kuesel), pp. 335-353. Van Nostrand Reinhold, New York (1982). Rose D. Shotcrete for support of underground openings. In Underground Structures (Edited by R. S. Sinha), pp. 295-318. Elsevier, Amsterdam (1989). Heuer A. Use of shotcrete for underground structural support. In Proc. Conf. Engineering Foundation, South Berwick, ME, American Concrete Institute Publication SP-45 (1973). Fernandez-Delgado G., Cording E. J., Mahar J. W. and Van Sint Jan M. L. Thin shotcrete linings in loosening rock. In Proc. Conf. Rapid Excavation and Tunneling, Atlanta, GA, pp. 790-813. AIME, New York (1979). Amberg R. and Sala A. Shotcrete as permanent lining for the Furka Base Tunnel. Rock Mech. Rock Eng. 17, 1-14 (1984). Choquet P. Dimensionnement des cintres de soutènement de tunnel en tenant compte de la presence des assemblages des elements et de la raideur du garnissage. In Proc. Int. Congr. Large Underground Openings, Firenze, Italy, pp. 521-529 (1986). Birön C. and Arioglu E. Design of Supports in Mines, p. 248. Wiley, Chichester (1982). Mitri H. S. and Hassani F. P. Structural characteristics of coal mine steel supports. Int. J. Rock. Mech. Min. Sei. ά Geomech. Abstr. 27, 121-127 (1990). Jukes S. G., Hassani F. P. and Whittaker B. N. Characteristic of steel support systems for mine roadways. Part 1. Modelling theory, instrumentation and preliminary results. Mining Science and Technology 1, 43-58 (1983). Choquet P. Design of steel arch supports for gate roadways. CIM Bulletin 79, No. 891, 88-96 (1986). Choquet P. A failure criterion of steel arch supports for the interpretation of in situ monitoring results. In Rock Breaking and Mechanical Excavation. CIM Special Volume 30,14th Canadian Rock Mechanics Symposium (Edited by P. Baumgartner) (1984).
348
Support
115. Choquet P. Rock Bolting Practical Guide. Energy, Mines and Resources, Canada, SP88-15E p. 266 (1991). 116. Brown E. T. (Ed.) Suggested Method for Rockbolt Testing, Rock Characterization and Monitoring, ISRM Suggested Methods, Pergamon, Oxford (1981). 117. Ontario Ministry of Labour. Rockbolt Pull-Test, Equipment, Methods and Practices, Mining Health and Safety Branch, Sudbury (1983). 118. Xanthakos P. P. Ground Anchors and Anchored Structures, p. 686. Wiley, New York (1991). 119. Potvin Y., Hudyma M. and Miller H. D. S. Design guidelines for open stope support. CIM Bulletin 82, No. 926, 53-62 (1989). 120. Franklin J. A. and Woodfield P. F. Comparison of a polyester resin and a mechanical rockbolt anchor. Trans. Inst. Min. Metall. Eng. (London)&0, A91-A100, (1971). 121. Gerdeen J. G, Snyder V. W., Viegelahn G. L. and Parker J. Design Criteria for Roof Bolting Using Fully ResinGrouted Untensioned Bolts to Reinforce Bedded Mine Roof. USBM Open File Report 46-80, NTIS: PB80-180052, -180060, -180078, -180078, -180094 (5 volumes) (1980). 122. Bartels J. R. and Pappas D. M. Comparative laboratory evaluation of resin-grouted roof bolt elements. U.S. Bureau of Mines Report of Investigations 8924, p. 20 (1985). 123. Karabin G. J. and Delevec, W. J. Comparative evaluation of conventional and resin bolting systems. Information Report 1033, Mining Enforcement and Safety Administration, U.S. Dept. of the Interior (1976). 124. Ballivy G. and Martin A. The dimensioning of grouted anchors. In Proc. Int. Symp. Rock Bolting, Abisko (Edited by O. Stephansson), pp. 353-365 (1983). 125. Coates D. F. and Yu S. Rock anchor design mechanics. Research Report R-223, Energy, Mines and Resources Canada, CANMET, p. 13 (1970). 126. Brown D. G. Uplift capacity of grouted rock anchors. Q. Ontario Hydro Research 22, No. 4, 18-24 (1970). 127. Littlejohn G. S. and Bruce D. A. Rock anchors - State of the art, Part 1: Design, Part 2: Construction, Part 3: Stressing and Testing, Ground Eng. 8, Nos 3, 4, 5, 6 (1975) and 9, Nos 2, 3, 4 (1976). 128. Meyerhof G. (Ed.) Canadian Foundation Engineering Manual. 2nd edn. Canadian Geotechnical Society (1985). 129. Ballivy G. and Dupuis M. Laboratory and field evaluation of the bonding strength of grouted rock anchors. In Proc. 13th Can. Symp. Rock Mech., Toronto, CIM Special Volume 22, pp. 97-102. Canadian Institute of Mining and Metallurgy, Montreal (1980). 130. Ballivy G. Personal communication (1986). 131. Atlas Copco Canada Inc. Montreal. Pull Test Equipment, Swellex Technical Bulletin 5, (1982). 132. Wijk, G. and Skogberg B. The Swellex rock bolting system. In Proc. 14th Can. Symp. Rock Mech., CIM Special Volume 22, Vancouver (1982). 133. Ingersoll Rand. Technical Notice: Split Set Friction Rock Stabilizers for Underground Roof and Rib Support, Split Set Division, Princeton, NJ (1981). 134. Myrvang, A. and Hanssen T. H. Experience with friction rock bolts in Norway. In Proc. Int. Symp. Rock Bolting, Abisko (Edited by O. Stephansson), pp. 419-423 (1983). 135. Singh R. N. and Buddery P. S. An assessment of the efficiency of roof bolt anchorage based on laboratory and field experimentation. In Proc. Int. Symp. Rock Bolting, Abisko (Edited by O. Stephansson), pp. 445-457 (1983). 136. Einstein H. H. Design and analysis of underground structures in swelling and squeezing rocks. In Underground Structures (Edited by R. S. Sinha), pp. 203-260. Elsevier, Amsterdam (1989).
13 Development of Tunnel Support Philosophy ALAN M. MUIR WOOD Halcrow, London, UK
13.1
INTRODUCTION
349
13.2 AN HISTORICAL PERSPECTIVE
350
13.2.1 Traditional Support Systems 13.2.2 Evolution of Relationships Between Theory and Practice 13.2.3 The Great Leap Forwards 13.2.3.1 The phenomenon: essential elements 13.2.3.2 Geology 13.2.3.3 Geotechnology 13.2.3.4 Design of support 13.2.3.5 Development of support 13.2.3.6 Instrumentation 13.2.4 Holistic Approach
350 351 352 352 353 353 353 354 354 354
13.3 THE ROCK MEDIUM: GEOLOGY 13.3.1 Geological Data for Support Design 13.3.2 Site Investigation 13.3.3 Engineering Interpretation of Geology
354 354 355 356
13.4 PRESENT DAY APPROACHES TO DESIGN 13.4.1 Choice and Necessity 13.4.2 Design Concepts 13.4.2.1 Simplified stressIstrain models 13.4.2.2 Qualitative geological grading 13.4.3 The Observational Method 13.4.4 The Third Dimension 13.4.5 The Dimension of Time
357 357 357 358 358 360 361 363
13.5 CRITERIA FOR SUCCESS 13.5.1 Application of Principles 13.5.2 Practical Aspects of Support 13.5.3 Organizational and Procedural Aspects 13.5.3.1 Continuity 13.5.3.2 Flexibility 13.5.3.3 Quality assurance 13.5.3.4 Competence
363 363 365 366 366 366 367 367
13.6
ENVOI
367
13.7
REFERENCES
368
13.1
INTRODUCTION
Every major tunnel is unique; the most important feature of uniqueness is the state of the ground, but other major contributors are the scheme of construction and the form and timing of tunnel support. Progress in technique arises from understanding the significant aspects of experience, and
349
350
Support
from the distillation of this knowledge so that it may infuse the conceptual approach to a subsequent project with similarities to, and differences from, the past experience. The design and construction of tunnel support needs to be seen as part of a system or, more correctly, of two systems. In the first system, the learning process advances with time, as experience, both personal and that enjoyed vicariously, guides the designer. In the second system we are concerned with the interactions between geological investigations, the tunnel concept, design and construction, each of which interrelates with the others. One of the objectives of this chapter is to explore the criteria whose pursuit may lead towards optimal results, optimal that is in terms of good engineering or, its equivalent, value for money. There are divergences in the meaning attributed to certain phrases used in connection with tunnel support. The sense in which such terms are used in this chapter is usually defined at the first usage. Basic, however, to all discussion are the terms concerning support itself, defined as follows. (i) Primary support is the term used for the provision for short-term stability of a tunnel. Primary support need not all be established at one time; by design or by reaction to circumstances, it may be applied in stages while ensuring continuing stability. It may however, also form the permanent lining. (ii) Secondary support relates to the support provided for long-term stability and possibly for other purposes. Secondary support may be in the form of a lining, in concrete or other material. (iii) Formal support is a term used for support built to comply with a specified geometry, usually as an in situ concrete or as a segmental form of lining immediately behind a shield or (shielded) tunnel boring machine (TBM). (iv) Informal support describes support that does not comply with a particular stated finished geometry, e.g. rock-bolts, shotcrete, arches, ribs and lagging. (v) Incremental support describes a form of primary support that is planned for modification in response to observations on performance, i.e. incremental support based on the observational method (see Section 13.4). The main purpose of this chapter is to review recent developments in support strategy from the viewpoint of a practising engineer, to illustrate elements of the choice for any tunnel and the reasons for preference of a particular approach, advocating the maintenance at all times of an eclectic attitude. There is no universally applicable optimal method but there are sound common principles. A brief study of the recent history of tunnel support is helpful in deducing a rationale based on the understanding of the time in relation particularly to the knowledge of rock mechanics, the limitations of mechanical plant and of computing, and the craft tradition of tunneling. Today, an engineered system of support is to be considered as an art, drawing nevertheless, in its practical implementation, on procedures transmitted through time of a heuristic nature. Looking back at well-documented tunnel projects, we can identify reasons both for failure and for success, in relation to the schemes of construction. We can then enquire as to how such factors can be controlled at the present day for successful projects of a comparable type. It has to be admitted that surveys of conditions and quality control of many old tunnels leave a certain degree of conjecture around the cause for failure but it is usually possible to identify the main factors. 13.2 AN HISTORICAL PERSPECTIVE 13.2.1 Traditional Support Systems Tradition in technology always deserves respect, often repaying the effort of study by revealing results of learning, by trial and error, of practical value, in respects that are not susceptible to scientific calculation. The days of the European canal and early railway tunnels developed traditional methods of ground support, varying considerably across Europe but with sufficient regional variation for the names of several different forms of support by heavy timbering to have been preserved. Examination of each reveals the particular types of tunnel for which it was most appropriate. For illustration see, for example, Sandström [1] or Müller [2]. The dominant systems, by no means confined to the country of their titles, are described briefly below. The German system comprised a series of box headings within which successive sections of the sidewalls of an arch might be built from its footings upwards, a forerunner of the method of multiple drifts. Success depended upon the central dumpling continuing to resist side pressure on the arch, and also in supporting the top heading prior to relief by completing the arch. The Belgian system depended on the initial construction of a top heading, propped approximately to the level of the springing of an arch for a horseshoe section tunnel. This heading was then enlarged to each side to permit construction of the arch, which was progressively extended by under-pinning
Development of Tunnel Support Philosophy
351
from the side headings. The method presented the common problems of high cost which face all systems that depend on multiple drifts; it was, moreover, only practicable where rock loads were not heavy. The Austrian system depended upon a strongly constructed central bottom heading upon which a crown heading was constructed. Full-face excavation was heavily braced against the central headings with longitudinal poling boards built on timber bars carried on each frame of timbering. As the lining was constructed, the timbering was propped against each completed section to maintain stability. The method was able to withstand reasonably high ground pressures but was particularly extravagant in the use of timber. The English system depended upon the construction of a central top heading which allowed two crown bars to be hauled into place, the rear ends supported on a completed length of lining, and the forward ends propped within the central heading. Bars were then erected around the full face, with timber boards supported on each pair to exclude the ground. The system is economical in timber, permits construction of the arch in a full-face excavation, but is dependent upon relatively low ground pressures. The Austrian system is seen to be designed for the heaviest ground pressures, the English system for lighter pressures and for economic use of timbering. The systems, with local variants, were used with the expectation of, at least, ultimate success - throughout the main period of construction of canal and railway tunnels through the late eighteenth and the full nineteenth century. Meanwhile a few forward-looking engineers began to reflect upon the most economic means for supporting rock around a tunnel, attracting the least fraction of full rock load. So long as timber, a material with low modulus in relation to rock, remained the means for primary support there was little prospect of significant progress. The situation can be compared with that of the medieval architects and masons whose rules of thumb were applied to the, not always successful, design of buildings. The tunnelers were encountering practical problems of rock support with neither an adequate understanding of the forces at work nor the tools to put that understanding to good effect.
13.2.2 Evolution of Relationships Between Theory and Practice The substitution of steel arches for timbering introduced an urgent need to forecast the requisite strength of the members since, unlike timber, a steel arch, which fails by buckling and twisting of its web, provides a reducing support once the crippling load has been exceeded. One of the early attempts at estimating rock load is attributable to Kommerell, considering the height of a parabola/semi-ellipse above the tunnel as representing the limit of rock loosened by a stated amount of sinkage (expressed as a fraction of the height of the column of rock) at the level of the tunnel crown. By comparison with the masonry arch, subjected to vertical (V) and horizontal (H) loading, it is not unreasonable to suppose that a natural arch may be formed in rock in a comparable manner. In fact, for an elliptical cavity of axes 2a and 2c vertically and horizontally, in rock treated as an elastic continuum, the tangential stress on the vertical and horizontal axes are readily found (Muir Wood [3]) to be, respectively Pm = tf (1 + 2a/c) - V Pt{c) = V(l + 2c/a) - H whence, if a/c is in the ratio V/H, Pt(a) = Pt(c) = H + Fand the tangential stress is constant around the periphery of the cavity (Figure 1). This is not necessarily the most stable disposition for highly stressed rock [4]. The approach by Kommerell, as subsequently developed by Bierbaumer, tabulated by Terzaghi [5], and described by Kastner [6], considers the rock as analogous to a fragmented material in a long silo with vertical side walls. Kommerell [2] considers a kinematically possible form of failure which provides a basis for determining not only the arch load but also the lateral loading on the legs of arches in a horseshoe section tunnel (Figure 2). So long as rock was treated only as a cause for load and not as a supporting medium in itself, little further progress could be made. The high and irregular overbreak associated with drill-and-blast tunneling, until the introduction of more controlled blasting methods around about 1970, coupled with the uncertain degree of blocking, i.e. the continuity of packing, between the rock and the supporting arch, discouraged further advance in modeling the stability of the rock-support system. The absence of a developed science of rock mechanics also had the effect of separation between the engineers and the geologists.
352
Support "Ί
H
Çlc)*Vi\ + 2c/a)-H, radii of curvature 'c2/a,a2/c Pa,pc where H = I/,
etc.
. j
\v Figure 1 Stresses around an ellipse in an elastic continuum
Envelope of burden zone of loosened rock
Λ-ΑΤ„(θ + χ τ )
Figure 2
Rock load on a tunnel (after Terzaghi and Kommerell)
13.2.3 The Great Leap Forwards 13.2.3.1 The phenomenon: essential elements By far the most significant advance of rock support, indeed of tunnel design in general, has been the ability to design the rock, or more generally the ground, to contribute optimally to its own support. As in all advances in technology, success depends upon the conjunction of original thought with the practical means for fulfilment. For rock support the several factors may be simplified thus: (a) geology, (b) geotechnology, i.e. quantifiable geology, (c) mechanics of continuum and discontinuum, (d) design of appropriate means of support, (e) development of appropriate means of support, (f) instrumentation, (g) special construction plant, and (h) holistic features: contractual relationships, the 'enlightened purchaser', the observational method. Tunneling, as a traditional craft, presented special difficulties in the introduction of ideas of a revolutionary nature, reflecting the degree of simultaneous change required from several participants: the project promoters; the tunnel designers; the contractors; the skilled miners; the designers and manufacturers of plant and special equipment; the suppliers of specialist materials. It is not
Development of Tunnel Support Philosophy
353
therefore surprising to find that the ideas long preceded the achievement, more or less cogently expressed by many. Rziha in 1878 stated the objective in tunneling concisely as the anticipation rather than the acceptance of rock pressures; the substitution of mental energy for physical labor was presented as a simile (Sauer [7]). During the period of this experience there had been an unprecedented opportunity to observe the behavior of rock in numerous railway tunnels under construction, using several techniques of excavation and support. These included the Alpine Mont Cenis Tunnel and the beginning of the Gotthard tunnel (both based on the Belgian support system; see Section 13.2.1). This was the period of 'heroic tunneling' when the major tunnels encountered unforeseen and - with the techniques available at the time - truly unforeseeable rock pressure problems. As Rziha rightly observed, the increasingly heavy construction adopted as a solution was itself the partial cause of excessive stresses caused by heavy rock pressures (Sauer [7]). Several contributing factors to the successful application of light tunnel support from the 1950s onwards are summarized in the following paragraphs.
13.2.3.2 Geology Geology had been developing as an increasingly fragmented set of descriptive bases of knowledge, with only a few, ahead of their time, attempting to perceive the subject as a unified system of processes. Geophysics was developing as a separate specialist subject, attached to geology predominantly through the petroleum industry. Thus Talobre [8] complains that les orientations totalement divergentes de l'ingénieur et du géologue réduisaient considérablement l'eificacite de leur collaboration'. He saw in 1935 the imperative need to develop a new technology of rock founded on calculation and experiment, deriving, for the first time, numerical relationships for observable phenomena. While engineering geology had initially to be grafted on to a predominantly descriptive subject area, the wider revolution which has since transformed the isolated features of geology into a comprehensive discipline of earth sciences, highly permeated by the major sciences, now enables engineering geologists to perform their primary role more effectively, that of translating the relevant aspects of geology into statements of direct interest to the geotechnical engineer concerned with rock support. There remains considerable scope for improved perceptions of the engineering consequences of geological processes, most especially tectonics, and for a systematic database for the vast experience from rock excavation, which remains only partially accessible to each practitioner.
13.2.3.3 Geotechnology It would be confusing to imply a unified origin of the single discipline now designated as geotechnology. Soil mechanics developed as a stress-related subject during the 1930s. Rock mechanics, built on pragmatic ancestry associated with mining, did not develop as a recognizable subject based upon real rock until the 1950s, by which time soil mechanics was beginning more seriously to relate stress to strain. Rock considered as an ideal elasto-plastic continuum provided a simple means of relating stress to strain (integrated as the readily measured ratio of convergence) (Kastner [6]). Rock considered as a discontinuum could provide an alternative basis for the several criteria for stability. Subsequently, modeling techniques have permitted increasingly realistic concepts of rock as relatively stiff blocks bounded by joints in partial contact or containing a deformable filling {cf. de Broglie 'les concepts de continu et de discontinu, pousses a l'extrême et opposes l'un a l'autre, sont impuissants a traduire la realite') [8].
13.2.3.4 Design of support Two fundamental concepts might be considered to be contradictory. On the one hand, for relatively shallow tunnels in broken rock, the virtue of avoiding the unloosening of the rock was well understood by Brunei at the time of his first shield patent (1818) [1], through the early attempts at a mortar lining (Grimm, 1858) [7] and the general precepts set out by practical tunnel engineers. On the other hand, analysis for rock stabilization entailing plastic deformation was bound to be associated with convergence and hence with departure from an immediate incompressible form of support, leaving aside complications of time dependence. As will be subsequently developed, the apparent conflict is resolved by a correct definition of objectives.
354
Support
13.2.3.5 Development of support Rabcewicz [9] vividly describes features of traditional rock support which exacerbated the problems of stabilizing the ground. Masonry was preferred to concrete, the latter being considered as susceptible to damage when green. Space between lining and the support to the ground was drypacked, not grouted, and the space included timbering which could not be removed. Construction by multiple drifts for a typical length of tunnel occupied more than two months during which time appreciable movement of an incomplete arch length might occur, in advance of the long-term distortion attributable to voids, compression and decay of built-in timber, and to uneven rock/lining pressure. Application of the new approach to rock support demanded thin linings (relatively incompressible but offering little resistance to bending) in intimate contact with the rock. In the early 1950s such desiderata were satisfied by contemporaneous developments in concrete projected by compressed air (shotcrete) and in grouted or mechanically anchored rock-bolts and dowels originally developed for mining. The prospects for avoiding rock loosening were also advanced by improved control over blasting methods, and by increasing use of machine boring (by full-face machines, now known as TBMs, and by part-face machines, based on rock headers and boom cutters already widely used in mining). Thus, tunneling practice and methods were readily capable of development compatible with the new demand. Subsequent developments have shown a rapid response to new problems as they have been encountered or understood in circumstances where other aspects (see Section 13.5) have been favorable. 13.2.3.6 Instrumentation Techniques of site investigation entailing rock drilling were developing rapidly in the 1950s, spurred on by the needs of the oil industry. More cogently, relatively robust instruments were rapidly evolved and developed to meet perceived needs for observational data. As discussed in Section 13.4.3, appreciation of the contribution of tunnel support became increasingly based on observational techniques during construction and, for this vital need, relatively simple devices sufficed in the first instance. 13.2.4 Holistic Approach Consideration of the preceding paragraphs will indicate that by the 1950s new design methods were available and new techniques were being developed in parallel, in part stimulated by an increasing demand. There remained, however, a number of 'institutional' factors which needed to develop in a manner favorable to stimulating the simultaneous radical changes involved. First and foremost, there was the need for a champion of change, an influential party to the tunneling process capable of impressing upon others involved the practicability and the benefit of change. This initiative was most likely to come from the tunnel engineers but success depended upon a perceptive promoter capable of recognizing the potential, predominantly financial, benefits. (Since benefits would be expected to accrue with experience, the consenting promoter would be likely to be concerned with a project including major tunneling elements or with a series of tunneling projects, taking a long-term view, which may merit such a promoter to deserve the epithet of'the enlightened purchaser'.) As will be discussed subsequently, the new approach to tunnel support entails, and entailed predominantly in the initial stages, the principle of trial and modification based on the application of observational techniques; this feature introduces a necessary flexibility in the contractual relationships. In a negative respect, it could be held to be easier to supersede a traditional craft by a new technology when the former had been weakened by a desuetude - a major world war followed by immediate consequential years of low performance - followed by a rapid upsurge in demand. So the early 1950s favored the leap ahead, with the Snowy River Project, New South Wales tunneling from 1952 (Lang [10]) and developments in Austria (Rabcewicz [9]) making the most conspicuous contributions.
13.3 THE ROCK MEDIUM: GEOLOGY 13.3.1 Geological Data for Support Design An understanding of the geological environment is so fundamental to rock support that it calls for particular comment. The purpose of acquiring geological data is to describe the rock material,
Development of Tunnel Support Philosophy Tunnel features
Geology
i
Environment
I
i i i | I
x
i
355
Tradition and experience
-. ·
r ___J I I
.J Spectrum of Characteristics Increasing use of physical mechanics
Qualitative
RockLoad eg: Kommerell[2] Terzaghi[5]
Quantitative
Simple Zoning Deere[l3] Project-specific
~l
Mass Quality Barton etal.[16] Bieniawski [15]
I
1
Simple Models Continuum/ with Observation Discontinuum Cundall[2l] NAT M [ 9 ] ISOM (Section 13.4.3)
1
Elasto-plastic Models Talobre L 8 ] Kastner [ 6 ]
Figure 3 Taxonomy of rock support methods (indirectly after Einstein)
stratigraphy and structure so that the behavior of the rock may be predicted in relation to a specific (tunneling) project. It is also of course necessary to determine information on geohydrology. More exotic features such as temperature or the presence of inflammable or toxic elements are usually beyond the direct requirements for the design and construction of support. The design of the ground support requires geological information to be presented in a quantifiable form so far as is possible; hence the means for acquiring such information need to be refined. The choice of approach to support design across the empirical-analytical spectrum (see Figure 3) is largely constrained by the manner of presentation of the geological data. 13.3.2 Site Investigation Site investigation is defined as the activity of acquiring and analyzing data concerning the ground for the specific purpose of the (tunnel) project. Site investigation will start from a desk study of available data, with the acquisition of additional data where practicable from the mapping of exposures. The most useful information may come from records of other tunnels in the vicinity in comparable rock. Geological and geophysical exploration are then designed in a coordinated manner to be undertaken as ground investigation, the work often being conducted in stages as the need for further information becomes more clearly defined. For a properly coordinated project (see Section 13.5) the design of the site investigation develops in parallel with the planning and design of the scheme of tunneling, addressing specific problems whose solution will materially affect the cost of the project. Clearly, additional site investigation is only justified by the extent to which this may reduce project cost or reduce the uncertainty of project cost. In consequence, optimal site investigation also depends upon the contractual basis (see Section 13.5.3). On particular occasions, the preexisting knowledge about the ground may be adequate for the form of tunnel support to be established from the outset, but more generally a staged approach to site investigation serves to provide occasion for reflection upon the appropriate geological model(s) representing the current interpretation of available data, the prevailing uncertainties and the optimal means for their resolution. From time to time attempts are made to define appropriate magnitudes of site investigation in relation to the size of the project. There are no such general rules, since the extent (in meters drilled, for example) of investigation or the cost ratio of investigation and project depend upon geology, topography, the specific needs of the project and the inherent problems of tunneling. Legget [11], for example, indicates a range of 0.3-2.0% but there are examples of good (and bad) practice outside these limits in each direction. The evolution of a particularly economic form of tunnel support may depend upon highly detailed site investigation whereas heroic tunneling, conceived as battling against the forces mobilized by unpredictable nature, may benefit little from more than rudimentary exploration of the ground. Again, depending upon the contractual basis, more or less benefit may be derived from exploration of the ground during the initial stages of the project. Experience from full-scale or near full-scale excavation may serve to calibrate data derived from site investigation, thus reducing the
356
Support
extent of inference in interpretation. While trial adits or shafts may permit such calibration, they need not only to penetrate the rock types that are to be tunneled through but ideally to encounter these in conditions matching those of the project, e.g. degree of weathering, jointing patterns, stress fields and tectonics, extent and head of water. Interpretation in respect of support needs must yet take account of different rates of advance, size, means of excavation and of other divergences between adit and tunnel. A general account of the essential elements of a good site investigation is given by Legget and Hatheway [11]. Although many examples are dated - the senior author's experience extends over 65 years - the principles remain unchanged and the advice is exemplary, being summarized as: (i) systematic working from the general to the particular; (ii) determination of the means to suit the objectives; (iii) the development of observational prowess; (iv) high standards of supervision, logging of results and preservation of original data in samples, photographs and cores. Several practical features of ground excavation merit special emphasis. (a) Surface exposures of a rock may be quite unrepresentative of the rock at depth. Where the exposure has been weakened by weathering, the impression will be pessimistic; where, however, exposure leads to induration, e.g. by the formation by evaporation of a silcrete or calcrete, the impression will be optimistic. The interpretation of the nature of fractured rock and gouge of a fault zone may be particularly critical, as experienced for the Litani project in Lebanon [12]. (b) The dominant characteristics of the mechanical behavior of a fairly competent rock relate to joint geometry, nature and filling. These features therefore deserve special emphasis since they depend upon information most readily lost by erosion by the drilling fluid or by problems in core recovery. If critical information depends upon boreholes, integral (orientated) cores may be justified or the results of imprint packers or CCTV pictures may be useful. (c) The curation of rock cores deserves far more attention than it customarily receives, since the purposes and duration for which the cores may be used are often unforeseen at the time of their recovery. If cores are expected reasonably to represent their condition at the time of drilling, appropriate measures against moisture change, frost and high temperatures will be needed. Thinly bedded rocks will inevitably delaminate soon after recovery, and immediate (colour) photography is the only means to preserve a reliable visual record. (d) Full records of the torque, thrust and rate of penetration of the drill, and of the behavior of the drill fluid, may provide invaluable complementary information on the degree of continuity of the rock across zones of loss of core. 13.3.3 Engineering Interpretation of Geology In considering the diagram of different approaches to ground support (Figure 3), it is axiomatic that the greater the reliance upon analysis, the more quantified must be the inferences from the geological data. Thus, Terzaghi [5] depended upon a qualitative assessment of the rock condition, by the geologist; Deere [13] required only the relatively straightforward use of core logging techniques, while the more complex rock mass classifications depend upon the data being expressed in greater detail and quantitatively to a certain degree (Table 1). Ultimately continuum (or discontinuum) models require direct information to provide the basis for the constitutive equations for the behavior of the rock (and/or the jointing) material. The most reliable data in such respects are obtainable only from in situ testing, preferably pressuremeter tests, since these may be designed most directly to represent the effect of tunneling at the depth tested, always making allowance for the effect of scale, particularly important in jointed competent rock. Increasing emphasis is being placed upon the determination of in situ states of stress. The direct association of stress distribution with local topography has been well understood for many years and an increasing body of knowledge is developing on regional states of stress. The validity of a design basis for tunnel support depends on a reasonable assessment of stress not only transverse to, but also along, the tunnel. What is largely lacking at the present day is an adequate attempt to relate the jointing, weathering, alteration and warping of rock to the underlying causative tectonic history. The engineering geologist should naturally start from here in making quality assessments along the route of a tunnel. Regard should be given, particularly for schistose or thinly bedded rock, to the orthotropic physical characteristics to be expected. For a rock composed of layers of thickness tu i 2 , etc., of modulus El9 E2, etc., it is readily shown that mass modulus Eb along the bedding is given by Eh = (J51i1 + Ε2ί2...)Λ*ι + h + . · . )
Development of Tunnel Support Philosophy
357
while modulus En transverse to the bedding is given by 1/En = (tJEx + t2/E2 . . .)l(tx + t2 + · · · ) i.e. the arithmetic and geometric mean values, respectively. Simple estimates of the nature of interbedded rocks will provide a basis for such ratios between orthotropic elastic moduli and, by similar reasoning, hydraulic conductivities, necessary in predicting and interpreting tunnel behavior. The potential for delamination of thinly bedded strata, for example, can then be assessed. An understanding of the regional and local geological history will throw light upon the condition of the rock and, particularly, the nature of jointing. Tight discontinuous tension joints may approach the strength in shear of the intact rock; slickensided shear joints will have, depending on continuity and alignment, little more than the residual strength as determined from tests on intact rock samples. The tunnel engineer is directly interested in the mineralogy of rocks only in relation to the effects of change, e.g. whether swelling by hydration or breakdown by oxidation may occur over the period of interest for the project. Indirectly, variation in mineral composition may well be one factor of use in zoning the rock for differences in expected behavior. Geohydrology plays a part in the direct determination of stability of the rock locally to the tunnel face, in the variation of loading around a tunnel and in the need for special expedients practically to assemble tunnel support and in determining needs to incorporate waterproofing, a consideration that may dominate the scheme of construction. A particular concern for weak rock support is the possibility of zones of high permeability, containing water at high pressure, close to the tunnel giving rise to concentrated rock stress and the risk of irruption. The engineering geologist must have a clear understanding of the manner in which information he (or she) provides will be used in design and construction. Only thus will he be able to advise on the validity of data and the extent to which correlation may be appropriate in making estimates in a project-wide interpretation of data. The engineering geologist, as the interpreter between many specialist branches of the earth sciences and the engineer, must have a very catholic approach to his functions. He will need for example to be able to advise upon appropriate systems of geophysical and other remote forms of sensing. Cross-hole seismic data may prove informative in mapping qualitative variation in rock structure and type; in areas of particular concern such results may be interpreted by tomography as a 'body-scan'. 13.4 PRESENT DAY APPROACHES TO DESIGN 13.4.1 Choice and Necessity It is illusory to suppose that there could ever be a unique preferred systematic approach to the design of rock support. Apart from the infinite variability of the rock and its condition, there must continue to be regard for successful local experience, resources and tradition. There are nevertheless a set of precepts which do have general application, departures only being justified by particular circumstances. For example, the tunnel element of a project may be on too small a scale to justify development of unfamiliar techniques; expected variability of the ground may defy systematic analysis; administrative problems preclude reliance upon observations and hence rule out best practice; pressure of water or weak ground may require a closed face TBM and thus support, as a lining, erected without access to the ground. There have been suggestions that the approach to support design may be based on a choice across a spectrum of empirical to quasi-analytical methods. In fact, the eclectic engineer may well draw upon several concepts simultaneously to help to obtain a feel for an unfamiliar situation, maintaining an ability to accept or reject certain of the individual elements of each. The element of necessity concerns the essential compatibility of the relationship of stress and strain between the rock and the support, a condition ordained by nature rather than man for the early tunnels. Clearly, the degree of tolerance in such respects differs markedly between the types of support. Arch supports may yield, shotcrete may rupture, rock-bolt anchorage may drag. 13.4.2 Design Concepts Certain approaches to support design should be interpreted as concepts only, in any taxonomic discussion (Figure 3); their limitations as design tools need to be clearly understood. Such concepts fall essentially into two categories: simplified stress/strain models and qualitative geological grading.
358 13.4.2.1
Support Simplified stress/strain models
At the simplest, such models consider the rock as a homogeneous isotropic elastic or perfectly elastic/plastic medium; to the support is attributed a comparably simple relationship. The most popular form of such a model presents the result as radial convergence opposed by radial confinement (or 'support demand' as suggested by Duffaut) [14]. This, as a concept, is a helpful illustration of the undoubted fact that a stiff support erected prior to relaxation of the rock around an advancing tunnel will attract a high proportion of pre-existing rock stress (Figure 4). There are fundamental limitations to the application of such a concept however, apart from any practical question of variability of the rock, including: (i) The relationship between convergence and radial support is not unique, apart from the trivial case of an elastic rock; the degree of support affects the degree of triaxial confinement of rock layers close to the excavation and in consequence their stress/strain behavior and thus the contribution of these layers to convergence. (ii) Convergence is generally not only time dependent, but also dependent upon the proximity to the tunnel face. In consequence a convergence line is drawn uniquely for a certain assumed rate of steady advance of the tunnel excavation and for uniform application of support along the adjacent length of tunnel. With acceptance of such limitations, the concept is helpful, particularly when such a plot as Figure 5 is used as a guide, based on the experience accumulated during the construction of a particular tunnel.
13.4.2.2
Qualitative geological grading
Table 1 lists a set of attributes of rock, and those of a tunnel to be driven through the rock, which may dominate the requirement for, and timing of, support. This table also indicates those factors adopted quantitatively and qualitatively by four grading systems. The systems of Bieniawski [15] and Barton et al [16] will each provide a qualitative assessment of support needs and potential problems; whether either could adequately subsume the particular circumstances to provide a selfsufficient basis for an economic design is far more dubious. Many comparisons have been made of predictions by such gradings and the support actually adopted for specific tunnels. Only, however, where this support proved inadequate, or where detailed monitoring had been undertaken, is it possible to establish a factor of safety. There is, moreover, more or less uncertainty in relation to the assessment of the qualitative factors for these tunnels. The presumption of a direct universal
Radial pressure
— Radial deformation Figure 4
Convergence- confinement diagram
359
Development of Tunnel Support Philosophy i
Time 1
1
1
1
Initial support: bolts,mesh,shotcrete (with incremental supplement as necessary) Initial support secure apart from local ravelling
Incremental support adequate \ Extending zone of overstressed tor swelling) ground Incremental support inadequate
Figure 5 Typical convergence/time curves for a tunnel Table 1 Rock and Tunnel Attributes Considered by Selected Qualitative Geological Grading Systems Attributes Intact rock Lithology Special features of mineralology Bedding Attitude Strength Modulus Variability Stress tensor Jointing Attitude of sets 1, 2, 3 Spacing of sets 1, 2, 3 Tightness of 1, 2, 3 Roughness of 1, 2, 3 Continuity of 1, 2, 3 Filling strength of 1, 2, 3 Filling modulus of 1, 2, 3 Filling thickness of 1, 2, 3 Water Hydraulic conductivity tensor Pressure Tunnel Height Width Means of excavation Modes of excavation (e.g. full face, sequential) Support features a
Bi*
Bab
qf Xe
Wic
q
X
q
q q
q
q
q
X
X
q q q q
q q
q
q
q X
q q
q
q q
X
X
Fr*
X
q q
X
X
q q
q q q
q q q q
Bi = Bieniawski [15]. bBa = Barton et ai [16]. cWi = Wickham et al. [17]. d Fr = Franklin [18]. e x = Quantitative use. f q = Qualitative use.
relationship between a single quotient for rock quality and stand-up time (i.e. the period of stability for the advancing face) is yet more questionable. In summary, qualitative geological grading provides a reasonably reliable first assessment of the degree of difficulty to be encountered in rock support. No system provides a self-sufficient basis for design; for particular circumstances neither safety nor economy is ensured by its application. The
360
Support
main virtue of such systems is to provide an aide-memoire to the tunnel designer of the many geotechnical factors of potential importance. Where a tunnel traverses a suite of rocks variable only in respect of a few measurable parameters, a locally constructed geological grading system may provide a useful basis for the provision of initial support and frequently serves as such for rock zoning systems. Such systems provide a basis, for example, for estimating support requirements at the beginning of a project, desirably subjected to revision with the benefit of experience, including the methodical observation of the characteristics concerned and the behavior of the support, as the project proceeds. Certain forms of rock tunneling, and the excavation of large caverns, lend themselves to the selection of favorable subdivision and sequencing of excavation and support. There is then considerable merit in using a relatively simple basis for quality assessment related to a particular scheme of construction.
13.4.3 The Observational Method The observational method, central to good practice in rock support, merits definition and description in this context [19]. Essentially the observational method applied to rock support entails the following. (i) The selection of a conceptual model to justify a scheme of construction and the initial provision of support, with prediction of observable criteria to establish adequacy, usually including records of convergence, in relation to extent of support, to time and to distance from the face (the time and distance from the face being capable of being combined where a rate of steady progress is achieved in uniform rock). (ii) A comparison of observation with prediction to establish the adequacy of the conceptual model as the basis for determining support needs; alternatively, the inadequacy of the conceptual model, which calls for modification in consequence. (iii) Stages (i) and (ii) are repeated at all sections selected for the purpose until the conceptual model, as modified, adequately predicts observation. Adoption of such an approach must presuppose the ability to supplement support while the observational process is proceeding, without risk of collapse. It further presupposes a fundamental confidence in the approach, which is expected to be based on a combination of comparable experience elsewhere and of analysis of the particular circumstances. Full benefit of the method requires the tunnel support to be readily capable of being installed in an incremental fashion. The secondary support of thefinishedtunnel may be designed with a factor of safety higher than that applicable to the primary support. In consequence, provided adequate regard is given to timedependent factors, safety of the ultimate structure is not compromised by a phase of support approaching fairly closely to a factor of safety of unity. The definition of a safety factor in such circumstances is a subject for debate. Essentially, referring to Figure 4, assume stability is achieved at point B with radial pressure ρλ for convergence uja with a support system which could safely develop pressure py for convergence u2/a. The corresponding self-support by the ground may be estimated from extension of the ground line from B to D as being reduced from P — px to P — p2 as convergence of the ground increases by the equivalent of uja to u2/a. The factor of safety is then given by F = [Py - (Pi - PiWPi A tunnel, traversing a single suite of rocks, provides an excellent opportunity for exercising a learning process in relation to the design of adequate rock support, and in zoning the rock in relation to varying needs for the initial degree of support to be provided. The observational method is the central feature of many examples of current good practice in rock support. Indeed, the present author has suggested 'Incremental Support based on the Observational Method (ISOM)' as a general title for the current practice particularly of Europe and Japan, including the numerous successful examples attributed to the New Austrian Tunneling Method (NATM) [9]. The engineer needs to determine for a particular set of circumstances the degree of importance that observation (leading to incremental support) is to play; for a complex degree of sequential excavation, it may dominate the execution of the work, whereas for rapid excavation of rock of predictable behavior, the need for incremental support as a consequence of observation may be expected to be exceptional.
Development of Tunnel Support Philosophy
361
13.4.4 The Third Dimension Stress and strain of the rock have been analyzed and discussed at great length by many authorities in two dimensions in a transverse section to a tunnel (or essentially reduced to one dimension for the convergence-confinement concept - occasionally plotted to a time dimension). In the vicinity of the tunnel face where the predominant deformation and stress redistributions occur as the stressed rock is dislodged in excavation, the third dimension assumes a vital role. The principal phenomena need to be described so that the practical support measures may be foreseen, although the situation usually defies analysis through its very complexity. For the simplest example of a tunnel in homogeneous elastic rock where N = 1, i.e. px = py = pz = p0, convergence may be sketched on a longitudinal section as in Figure 6. This already makes an assumption about the degree of support of the (vertical) face. It also assumes uniform support along the tunnel in a state of plane strain. If we consider the stress tensor at a level of the line representing the excavated surface of the tunnel, at some distance ahead of the face the major principal stress is tangential (pt). As the face is approached, shear stress between transverse planes causes a rotation of principal stresses until a point close to the face where principal stresses will be inclined at 45° to the tunnel axis. Along the cylindrical surface of the excavation, the major principal stress will be tangential (pt). Thus the stress tensor will have rotated about the tangential axis and also about an axis along the tunnel radius. As with so many of the conceptual models applied to tunneling, once these are developed to take account of the real rather than ideal characteristics of the rock, complexity obscures the message without necessarily aiding practical planning and design of support. The most relevant features illustrated by the simple model are as follows. (i) The variations in the stress tensor cause an associated degree of change of strain pattern which may affect the subsequent strength of the rock. (ii) Consideration of a diagram of shear strength of a jointed rock, such as that illustrated by Figure 7, indicates that jointing in widely different attitudes in relation to the tunnel axis may lead to stress adjustment and consequent loosening of the rock in the vicinity of the face. (iii) The dominant feature stabilizing rock around the face is that of 'arching' along the tunnel, more correctly that of 'doming' around the face. In rock, as in soft ground, the representation of the face as a hemispherical dome may reveal necessary conditions for equilibrium. Where effective support is provided close to the face, a particular form of statically stable system may be considered. Thus, Figure 8 represents a dome containing an angle of 270° with effective support at a distance 2a from the face of a tunnel of diameter 2a. For a spherical cavity in an elastic continuum, the tangential stress at the inner surface, 1.5p0, may be compared with that for a cylindrical hole, 2pQ. (iv) Advance of the tunnel face will throw an increasing load on tunnel support, the increment reducing to zero (ignoring time dependence) when the face is 2.5 to 3 diameters distant.
./
/
/
y " \
Ground radial stress v^atr=<7
\
Zone of transverse shear
o
Convergence U
Radial stress
U\
W T (Transverse shear)
15
Figure 6 Radial convergence and ground stresses in the vicinity of a tunnel face
Support
362
Slip on joint
Simple compressive stress, N
(
2c cos φ \
, I -sin<£ )
Angle between line of N and joint of strength c + a n tan φ
Figure 7 Failure criteria for'jointed rock subjected to simple compressive loading Notional dome of ground support
Figure 8 Concept of a relieving dome of rock contributing to the support of ground close to the tunnel face
(v) Initial ground stress in the direction of the tunnel axis is a vital contributor to stability in the vicinity of the face. If this is low, benefit of the third dimension in providing support in the vicinity of the face is correspondingly reduced. For a real tunnel, a modified form of this simple concept may help to determine the approximate timing and form of support and often, in consequence, the manner and sequence in which construction is to be undertaken. It is in such practical aspects that developments by the proponents of the New Austrian Tunneling Method have been most influential, for example in the use of a short bench or a top-heading-and-bench excavation (Figure 9) and in the use of inclined arches so that
Figure 9 Examples of means of providing support close to the face (a) by use of an inclined arch support and (b) by use of a short bench
Development of Tunnel Support Philosophy
363
early support in the crown of tunnel may be combined with a sloping face to control ravelling of the rock, coupled with early stabilization of the invert, completing the cylinder of lining. Figure 6 also demonstrates the illusion of a single convergence-confinement diagram, since the third dimension dominates the circumstances of load transfer to the support. Transfer of load to the support is initially controlled by support stiffness and by distance from the advancing face, all in relation to the particular rock, with the major contributor to rock support close to the face only representable by rock stress in the third dimension, or as shear between adjacent slices transverse to the tunnel. Aspects of this subject have been discussed by Sauer [7], Ward [20] and Kerisel [14]. Simple reasoning for a particular set of circumstances may indicate critical features. While reinforcement of initial support may well be achieved in an incremental fashion in the application of the observational method, any variation of the overall concept of construction found to be required may be more difficult and costly. The third dimension may dominate such consideration. Provided that support develops sufficient shear stress across joints or by other means avoids loosening of unstable rock elements, the third dimension will provide continuity between supports and an element of load sharing such that a consistent patterned support may need only to take account of the mean condition of the local rock, apart from exceptional states of faulting or extensive weakness. The suffer the support, the lower the degree to which the third dimension may be viewed as an ally. In many intact weak sedimentary rocks, the shear modulus is found to reduce sharply from a high initial value with increasing strain. This must have the effect of increasing the proportion of rock load transferred away from the excavation, by comparison with the assumption of a constant modulus, a particularly advantageous feature in the vicinity of the tunnel face. 13.4.5 The Dimension of Time In the analysis of ground support for tunnels in clay, time dependence may be related predominantly to the effects of changes in pore pressure on the pattern of stress and strain. It is probable that such a phenomenon may well explain much of the time-related behavior of weak sedimentary rocks, taking into account their orthotropic structural features (the permeability parallel to the bedding will be greater than that across the bedding, modulus lower across the bedding; see Section 13.3.3). For jointed rock the dominant feature concerns creep in the vicinity of the most highly stressed contact elements between rock fragments (Cundall [21]). Sulem et al. [22] started from empirical exponential expressions for time and face-distance dependence in an attempt to distinguish between the contribution of the two factors to convergence along a tunnel. While interesting in theory, in practice it is likely to be impossible to use such a relationship in a predictive manner since, initially, face-distance will greatly predominate. For a tunnel advancing at a steady rate in unchanging ground, the practical tunneler is concerned with the typical shapes of the convergence curves plotted against time or face-distance, which in these circumstances will be interchangeable (Figure 5). If failure or yield is measured in relation to the rate or duration of application of stress, it is well known that natural and artificial rock (concretes) accept a reduced stress with increasing time (or reduced rate of loading). Where creep leads to plastic deformation, the predominant feature will be the reduced modulus; where time causes reduction in failure strength, a much more serious risk of brittle failure may present itself. So it is important to have a qualitative understanding of the timedependent behavior of the particular rock and, for old deformed tunnels, of the support also.
13.5 CRITERIA FOR SUCCESS 13.5.1 Application of Principles At the present day, the practical tunneler is presented with a confusing and elaborate set of options and precepts in design models. However, as has already been stressed (Section 13.4.1), these models should be viewed as concepts, not as design methods, and are judged in so far as they help to illuminate the behavioral characteristics of the supported rock. Large caverns merit an exhaustive study of the characteristics of the surrounding rock in relation to planning of optimal geometry and in providing adequate safety margins against foreseeable failure mechanisms. The approach here is that of good comprehensive rock mechanics, using the most appropriate numerical representation of the rock and the support, in relation to critical intermediate phases of construction and of time.
364
Support
In general, for tunnels that traverse a greater diversity of rock in relation to the total value of the project, the approach must be more flexible, the expectation of the need for prompt response to uncertainty more predictable. Section 13.4 sets out criteria for evaluation of the rock from the available sources of information; it has been stressed that the cost of obtaining further information must always be justified by the consequential value to the project (and, for the farsighted, to future potential projects). Geological quality designations provide a check-list for some of the relevant parameters concerning the rock in relation to the particular project. These should only be looked upon as rough guides to support costs and should never obscure the need to consider specific behavioral features derived from more elementary principles. Structural models of an idealized nature should be used as conceptual guides to behavior as appropriate but should not be expected to reproduce closely the behavior of the real rock, which is variable, jointed, locally weathered, all to a project-specific degree. The attempt to devise too detailed a model will usually end by obscuring the main features of behavior without approaching the full complexity of a variable continuum-discontinuum. At this point there is a decision fork. For the familiar rock mass, a well-tried empirical approach may be appropriate, drawing upon successful parallel experience, having assessed the risks, including those which arise from differences between the present and previous projects. Rock support may, for such projects, play a subsidiary part in the overall scheme of construction, with for example the use of a TBM with a precast form of segmental lining designed to perform adequately in all respects in all forms of expected conditions of the rock. Where, however, rock support presents potential problems, or where potential economies may derive from a more rational basis of rock support, the approach should be by way of simple conceptual models recognizing the dominant features of the type of rock, e.g. the extent to which it should be considered as a continuum or discontinuum, the consequent dominance of jointing patterns, directional schistosity, and in situ stress. Hoek [23] drew attention to the question of scale in the behavior of a jointed rock; a rock that is designated as distinct blocks near to a small tunnel may, at a distance, be better considered as a totally jointed rock mass. A scheme of construction is then devised which provides, at all intermediate stages, for the expected range of rock behavior, particularly in relation to stability near the face, and near the transitional faces between sequential stages of excavation where appropriate. The use of a TBM introduces practical problems of access for providing support close to the face; its design and scheme of operation should address such issues. It is advisable to consider particular failure modes, especially for tunnels in weak or fragmented rock. Thus the lateral wedge mechanism illustrated by ABC in Figure 10 is a valuable means, as suitably modified, to identify the critical geometry for failure of a selected form of support, tested across a range of reasonable parameters for rock shear strength. V Vertical stress
imiiiiiiiii
Figure 10 Application of a critical wedge to the determination of support needs (after Rabcewicz and Sattler)
Development of Tunnel Support Philosophy
365
Monitoring systems are designed to record features deemed necessary to establish the adequacy of support or, on the contrary, to indicate requirement for modification or extension, applying the principles of the observational method. Initially, this monitoring may be of a highly planned and detailed nature in order to develop criteria for comparison with subsequent records, in a form of calibration. Trial headings or a trial length of tunnel may serve for such a purpose. Monitoring must never degenerate to a ritualistic performance of recording instrumental readings. It must be seen as an essential element in guiding the tunneling process. Records must be plotted promptly and systematically in such manner as to draw immediate attention to departures from expectation. 'Rogue' records, i.e. those assumed to be in error, should never be discarded without a check. The 'rogue' may be the most reliable predictor of problems ahead. The overall design philosophy must be understood by those responsible for monitoring so that changes in rock condition may lead to appropriate modification in the design of the monitoring procedures. The scheme of monitoring must take account of the proposed operations in construction to avoid unnecessary conflict or interference. On the other hand, if observations during a vital phase of construction are essential, the construction scheme must, from the outset, be built around this need. It is particularly vital that elements of plant and equipment be chosen as compatible with the observational plan. Their use thereafter should be controlled to ensure no premature relaxation of the scheme, a constant threat when consistency of observations may suggest the absence of potential problems, thereby disregarding the fundamental principle of the observational method. 13.5.2 Practical Aspects of Support Convergences will be the most consistent feature of any observational programme. The criteria for stability may be reduced to: (i) absolute magnitude of convergence; (ii) rate of change of convergence with time and the approach to an asymptotic value; and (iii) concern for critical positions of wedge failure in relation to patterns of sets of joints (see Section 13.5.1). Allowance needs to be made for the consequential effect, within a length of tunnel of 1 or 2 tunnel diameters, of any sudden change in the extent of support; thus a step change in support will cause a transitional change in expected convergence, depending on the influence relationship between support stress at one section and the associated convergence of the adjacent length of tunnel. For a shallow tunnel in a town, convergence may need to be limited by consideration of the consequential effects of settlement rather than the criterion of stability. The fundamental control of rock behavior as to whether this occurs as brittle failure or plastic yield depends upon the degree of containment, i.e. the limitation of JV, the ratio between maximum and minimum principal stresses. A shattered rock in a contained volume stressed in one direction will develop a ratio of N equating to Ka9 the active earth pressure coefficient. Hence the classical experiment described by Lang [10] of the integrity of rock fragments, in a bucket, stressed by a central bolt. Plastic yield will develop gradual convergence with redistribution of stress across a relatively wide area; resourceful and conscientious observation will permit remedial action to be undertaken in due time. Brittle failure on the other hand will lead to a violent increase in support loading with the likelihood of progressive failure before counteraction is possible. The ability to contain may invoke the stabilizing effect of the bulking of fracturing rock and thus limit the zone of overstress. One of the virtues of rock-bolting is that it immediately achieves a measure of confinement of the rock and therefore promotes, to a corresponding degree, the propensity of plastic yield of the zone of rock closest to the excavation, and otherwise most likely to fail in a brittle mode. This is an example of the fact that the type of support may itself affect the shape of the notional convergence curve for the rock. Stress-strain compatibility for the support must be ensured. For the foreseen extent of plastic yield of the rock, considerable extension of a rock-bolt may result. Should the anchorage be designed to yield, or should the bolt be deliberately destressed in whole or in part? We may note the deliberate longitudinal slots provided in the shotcrete for the Arlberg and Tauern [24] tunnels to permit a high degree of convergence. Convergence of rock stressed within the elastic range will cause some elongation of securely anchored bolts since, in the absence of bolting it can readily be shown that the product of u - r is constant, i.e. the convergence within the body of the rock is inversely proportional to the radius. The stiffness of rock arches depends upon the form of construction. Where the arches are blocked at intervals from the rock, sideways buckling may occur leading to reduced strength and low
366
Support
stiffness. Recent practice is, however, either to use lattice arches and provide a continuous bead of shotcrete between rib and rock or to use rolled steel section arches with a similar bead in a weak mortar contained within a porous bag (Craig [25]). The timing of the support may be the most essential element. Thus shotcrete to provide immediate containment close to*the face may be expected partially to fail in due course, but only after further support has been provided, an activity requiring more time or space for its execution. The old art of graphic statics, revealing the conditions for stability of arch structures (Heyman [26]), has been applied to tunnel structures, where at least it may reveal the requirement for passive rock pressures to achieve equilibrium. It is as well to be reminded, however, of the extent of variation of passive pressures around a support or a lining since calculable outward deflection must be entailed in developing such pressures, e.g. for an elliptical cavity in uniformly stressed elastic ground (Figure 1), tangential stress at the boundary is proportional to [l/(radius of curvature)] 2/3 [27]. From this follows the virtue of flexible, if fairly incompressible, support, represented by a thin shell. Local variation in pressure, or variation in the curvature of the surface of contact, will also lead to variations in loading. Consideration must be given to the manner in which this is to be achieved, e.g. by shear between the support and the rock or by direct bearing on a footing. Hence, the benefit of achieving a full ring to limit such problems is obvious; the particular problem of the stability of the vertical walls of a horseshoe shaped tunnel is also evident. Stability depends on the adequacy of passive loading and the absence of an excess of active loading (which may follow the mechanism illustrated by Figure 2).
13.5.3 Organizational and Procedural Aspects Rock support will form part, usually an important part, of a tunnel project involving other operations. Aspects of good practice, described above, must be reflected by the project organization and procedures. While there are many different successful forms of organization, each must satisfy certain overriding principles. 75.5.5./
Continuity
Good practice demands continuity across all stages of planning, studies and investigations, design and construction. A few examples for this need are given. (i) Site investigations must address those features of special interest to design and construction with the ability for modification as the increasing extent of information about the rock may lead to needs for variation of the investigation itself. (ii) Adoption of the observational method demands close coordination and cooperation between design and construction. (iii) Close cooperation between engineers and geologists must enable the latter to perceive all those aspects of information that may be relevant and, in particular, to be sensitive to previously unforeseen geological features that may call for reassessment of conceptual models or of practice in design and construction.
13.5.3.2
Flexibility
There may be occasions when justifiable confidence in a specific scheme of construction demands no provision for change. The more normal expectation is that change is to be expected in support, and this is to be encouraged to match changing needs in the interest of economy and safety. Organizational provisions, including terms of contracts where such apply, must be written with such flexibility in mind. To the engineer the benefits of this approach are evident; there are often others who understand principles of accountancy only in relation to fixed costs, known from the outset. It is only necessary to compare costs of comparable rock support, adopting flexible and inflexible approaches, to appreciate the high potential benefits of the former. This, however, is not the place to argue the issues but rather to warn that, unless the flexible approach is adopted, the virtues of the new capabilities for rational, economic rock support cannot be achieved. This is therefore an issue to be addressed at the outset of a project. There are different ways in which the risk relating to uncertainty of the rock may be shared between the parties to a contract. If the owner bears the risk, additional cost will be incurred in so far as the risk eventuates. If the contractor bears the risk, the
Development of Tunnel Support Philosophy
367
owner must pay the full cost, whatever the actual risk encountered. Decisions on variation in support or in construction method should be taken without undue pressure in relation to direct extra costs. The indirect consequential costs of not adapting to circumstances may constitute a hazard and may entail additional costs of a far greater order. It is an unfortunate fact that litigation costs in relation to project uncertainty, not appropriately reflected in the organization, represent an increasing burden on all concerned in tunneling projects. Theflexibleproject makes provision for variation without resort to law. The alternative, the 'brittle' project, has a predictably higher propensity for expensive accident. Where 'brittleness' is combined with lack of continuity, the probable cost of litigation increases exponentially with the number of parties involved. 13.5.3.3 Quality assurance The application of quality assurance (QA) to geotechnical engineering should recognize the inevitability of uncertainty and the need to concentrate upon the systematic exercise of responsibility for appropriate response in relation to powers and technical capabilities of the participants. Unfortunately, QA too readily degenerates into paper-bound procedures which not only complicate decisions for desirable modifications but tend to deaden the sensitivity of the participants to features not foreseen by the scheme of QA, and inhibit the imaginative thought and observational prowess which should be the distinctive features for success. The more appropriate provision is that of the Project Quality Plan (PQP), a process of continuous assessment of the nature and quality of the inputs (technology, management,finance)appropriate to the successful fulfilment of the project, with the clear definition of duties and capabilities of the principal parties and their interrelationships. Lines of command must be direct and short; the needs for variation in support will not wait upon prevarication or incompetence. 13.5.3.4 Competence It should not be necessary to stress the degree of skills to be contributed by each party involved in the design and construction of rock support, with their specialist advisers from thefieldsof academia and research. The new methods require new technical capabilities and understanding in their execution and in the management of aflexibleapproach. Participants in important projects of rock support need to be appointed for their capabilities and for the ability to cooperate with the other competent parties concerned, which requires a mutual awareness and appreciation of each others skills. There must be an adequate shared basis of understanding between all communicating parties. 13.6 ENVOI The message from this chapter is that of reconciling the complexity of the rock to be supported, with all its diversities of type, structure, discontinuity and weathering pattern, with the essentially simple concepts affecting stability in any particular circumstances. It may be as dangerous to oversimplify the former as it is confusing to over-elaborate the latter. As Ward [20] reminds us, much can be learned from relatively simple observational studies which marry the geological complexity to the conceptual model. The observational method is the central feature of the present day approach to rock support (Peck [28]; Muir Wood [19]). The observational method must always permit an 'escape route', i.e. there must be a practical course of action wherever observation indicates inadequacy of measures so far taken. Numerical modeling provides a useful tool in exploring the important features of a particular situation, including on occasion tests for sensitivity to variation of particular parameters. Simple elastic models based on boundary integrals have their application; complex finite element nonlinear models have theirs, but only where they satisfactorily reproduce the structural features. Deteriorating rock structure based on the application of fractals may well contribute in the future. Rock support may have a life requirement of months only or, in other circumstances, tens or even hundreds of years. There should be a clear understanding of such requirements and of the prevailing conditions that will control life expectancy. The fundamental understanding of tectonics, of rock mechanics and of the application of such understanding will continue to develop. The general principles set out in this chapter are believed to
Support
368
have a degree of permanence against all foreseeable developments. The tunnel engineer must ever remain eclectic in his (or her) approach, selecting methods of construction and systems of rock support best fitted to each particular situation.
13.7
REFERENCES
1. Sandström G. E. The History of Tunnelling, Barrie and Rockliff, London (1963). 2. Müller L. Der Felsbau. Band III: Tunnelbau, F. Enke, Stuttgart (1978). 3. Muir Wood A. M. Ground behaviour and support for mining and tunnelling. In Tunnelling "79, pp. xi-xxiii. Institution of Mining and Metallurgy, London (1979). 4. Jaeger J. C. and Cook N. G. W. Fundamentals of Rock Mechanics, Chapman and Hall, London (1979). 5. Terzaghi K. Load on tunnel supports. In Rock Tunnelling with Steel Supports (Edited by R. V. Proctor and T. L. White). Commercial Shearing & Stamping Co, Youngstown, OH (1961). 6. Kastner H. Statik des Tunnel- und Stollenbaues. Springer-Verlag, Berlin (1971). 7. Sauer G. When an invention is something new: from practice to theory in tunnelling. In Tunnelling '88, pp. 1-15. Institution of Mining and Metallurgy, London (1988). 8. Talobre J. A. La Mechanique des Roches. Dunod, Paris (1976). 9. Rabcewicz L. V. The New Austrian Tunnelling Method. Water Power 16, 453-457, 511-514 (1964), 17, 19-24 (1965). 10. Lang T. A. Theory and practice of rock-bolting. Trans Am. Inst. Min. Metall. Pet. Eng. 220, 333-348 (1961). 11. Legget R. F. and Hatheway A. W. Geology and Engineering. McGraw-Hill, New York (1988). 12. Muir Wood A. M. Tunnel hazards: UK Experience, Hazards in Tunnelling and on Falsework, pp. 47-59. Institution of Civil Engineers, London (1975). 13. Deere D. U. Technical description of rock cores for engineering purposes. Rock Mech Eng. Geol. 1, 17-22 (1964). 14. Association Française des Travaux en Souterrain (AFTES). Stabilité des tunnels par la Methode ConvergenceConfinement, Tunnels et Ouvrages Souterrains (Lyon), No. 32, pp. 67-143 (1979). 15. Bieniawski Z. T. Geomechanics classification of rock masses and its application to tunnelling. In Proc. 3rd Int. Congr. Rock Mech., Denver, vol. 11 A, pp. 27-32. National Academy of Sciences, Washington, DC (1974). 16. Barton N., Lien R. and Lunde J. Engineering classification of rock masses for the design of tunnel support. Rock Mech. 6, 189-236 (1974). 17. Wickham, G. E., Tiedmann H. R. and Skinner E. H. Support determination based on geologic predictions. In Proc. 1st Rapid Excavation and Tunneling Conference, New York, AIME, pp. 43-64 (1972). 18. Franklin J. A. Safety and economy in tunnelling. In Proc. 10th Can. Symp. Rock Mech., Kingston, Ontario, vol. 1, pp. 27-53 (1975). 19. Muir Wood A. M. The observational method revisited. In Proc. 10th SE Asian Geotechnical Conference, Taipei, vol. 2, pp. 37-42. (1990). 20. Ward W. H. Ground supports for tunnels in weak rocks. Geotechnique 28, 133-171 (1978). 21. Cundall P. A. Distinct element models of rock and soil structure. In Analytical and Computational Methods in Engineering Rock Mechanics (Edited by E. T. Brown), pp. 129-162 Allen and Unwin, London (1987). 22. Sulem J., Panet M. and Guenot A. An analytical solution for time-dependent displacements in a circular tunnel. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 24, 155-164 (1987). 23. Hoek E. Strength of jointed rock masses. Geotechnique 33, 187-223 (1983). 24. John M. Construction of the Arlberg Expressway Tunnel tube. Tunnels and Tunnelling 12 (5), 45-50,12 (6), 66-68 (1980). 25. Craig R. N. The Lewes Tunnel, Sussex, England. In Tunnelling '79, pp. 153-157. Institution of Mining and Metallurgy, London (1979). 26. Heyman J. The Masonry Arch, Ellis Horwood, Chichester (1982). 27. Brady B. H. G. and Brown E. T. Rock Mechanics for Underground Mining. Allen & Unwin, London (1985). 28. Peck R. B. The observational method in applied soil mechanics. Geotechnique 19, 171-187 (1969).
14 An Overview of Tunnel, Underground Excavation and Boreholes Collapse Mechanisms VINCENT MAURY Elf Aquitaine, Pau, France 14.1
INTRODUCTION
369
14.2 FAILURE AND INSTABILITY AROUND ISOLATED CAVERNS 14.2.1 Case of isolated caverns in homogeneous isotropic or anisotropic material rock 14.2.1.1 Case of isotropic tight material under isotropic stress perpendicular to axis 14.2.1.2 Case of anisotropic tight material 14.2.1.3 Case of porous material 14.2.1.4 Case of square, rectangular or straight sidewall galleries 14.2.2 Heterogeneous medium 14.2.2.1 Rock mass having one prevailing heterogeneity direction 14.2.2.2 Rock mass with several heterogeneity directions (bedding, fractures) 14.2.3 Particular case of invading material 14.2.4 Case of swelling and solution materials 14.2.4.1 Case of anhydrite and gypsum 14.2.4.2 Case of some argillaceous materials 14.2.4.3 Case of pyrites 14.2.5 Mechanisms of thermal origin 14.2.5.1 Case of dry materials 14.2.5.2 Case of porous materials 14.2.6 Instability mechanisms occurring in solution-mined caverns 14.2.6.1 True rock failure mechanisms 14.2.6.2 Abnormal, unexpected or hazardous behaviors 14.2.7 Mechanisms of dynamic origin FAILURE MECHANISMS RELATED TO ROCK MASS STRUCTURE: EFFECT OF WORKS UPON THE STATE OF STRESSES APPLIED 14.3.1 Sudden room and pillar collapses in mines and quarries 14.3.2 Instability due to fault or other discontinuity shearing
373 374 374 382 384 387 388 388 390 391 392 392 393 394 395 395 397 399 400 401 402
14.3
403 404 406
14.4 CONCLUSIONS ABOUT THE EXCAVATION PROJECT, SUPPORT SIZING/INTERNAL 14.5
PRESSURE SELECTION AND FUTURE INVESTIGATIONS
407
REFERENCES
409
14.1 INTRODUCTION This chapter summarizes a future publication (to be made available at BRGM's secretary, Orleans, France), entitled Failure Mechanisms Around Underground Openings, now in progress by the Failure Analysis Working Group of the French Committee on Rock Mechanics. Figures 1 and 2 show unsupported galleries that are perfectly stable. It can be challenging to present these figures at the beginning of a chapter dealing with rock failure mechanism. Gallery 369
370
Support
Figure 1 Storage gallery driven in 150 m deep chalk in the Paris basin (Geostock-Elf, France). The symmetrical black lines at the crown of the gallery are part of a flint bed, and are not failures
stability is indeed a challenge to the design and construction rules relating to underground works projects. Furthermore, the challenge is two-fold. On the one hand, the excavation walls and crown in 150-m deep chalk of average strength (Figure 1) showed no sign of failure or plasticity while chalk was being sujected to stresses equal to or theoretically higher than its strength. More startling is the fact that boring the gallery walls gave rise to no failure either, whereas tangential stresses twice as high as the monoaxial compression resistance of the chalk acted theoretically upon the walls of these boreholes. In Figure 2, the right-angled corners should also have failed, due to a high stress concentration. On the other hand Figure 3 is still more puzzling: the 5.80 m diameter tunnel bored through a very strong crystalline schist acts upon the rock within limits much lower than the rock resistance level. 1-2 years after normal boring, a generalized spalling appeared on that rock of presumably little plasticity or tendency to creep. A 25 cm thick and 8-10 km long concrete ring had to be set, reducing the useful diameter and increasing the cost of this water supply system. In the first project (underground storage), reinforcement and support were not needed. They would have caused the underground storage works to be economically unviable. In the second project (water supply gallery), works lining based on the knowledge of real deformation mechanisms and causes might have been less inconvenient.
An Overview of Tunnel, Underground Excavation and Boreholes Collapse Mechanisms
Figure 2 Chalk quarry (40-90 m deep) with four levels (by courtesy of Ent. Gosse, France)
In both cases, the wrong estimation of the real project conditions was due to a poor knowledge of. deformation, failure and instability mechanisms. Only in situ tests could give some indication about these mechanisms in the first case, leaving, however, theoretical questions unsolved. During the construction or operation of works (gallery, tunnel, cavern) lining disorders may happen even though the construction methods appear to be quite well suited. A tricky question arises, then, as to the origin, causes and mechanisms of these unpredicted disorders, i.e. deformation or failure. Observations are difficult since the opening is in operation and the lining makes rock access impossible. Unlined galleries and caverns offer a considerable advantage as the rock and failures are always visible even if all occurrences cannot be understood or predicted. As often occurs in earth science, observing phenomena under comparable but slightly different circumstances helps to identify mechanisms that have been neglected or overlooked so far. This chapter is devoted to the analysis and identification of mechanisms affecting tunnel and gallery stability so as to justify reinforcement. In this connection and based on the foregoing, the following infrequently considered types of excavations will also be dealt with. (i) Mined galleries and caverns used for storage: reinforcements should be minimized for cost reasons. On the other hand, once construction is over, safety requirements are reduced, and reinforcement should be optimized even at the cost of structure shape and geometrical tolerances.
371
372
Support
Figure 3 Gallery (5.88 m) bored in crystalline schists. Incipient spalling (by courtesy of Electricité de France)
(ii) Solution-mined caverns when using a soft process, such as solution mining, for the excavation of a practically impervious material (salt), caverns of several hundred meters in height and ten meters in diameter are supported only by their proper, sometimes variable, internal pressure. Much can be learned by analyzing their behavior and possible failure. (iii) Deep drilling: oil or (current) scientific purpose-oriented wells. Based on their study, the behavior of initially circular openings can be analyzed. These structures are subjected to varying pressure, temperature and flow conditions and are more or less efficiently supported depending on whether the drillingfluidpenetrates the rock. This is an essential point in the support of boreholes in the course of drilling. At a later stage, the borehole is cased with a steel casing, which can be the subject of deformations that may result in failure. Furthermore, rock failure can produce wellbore failure. (iv) In addition to oil wells, deep scientific purpose-oriented wells, where stability is so critical as to condition the structure construction, should be mentioned. (v) Tens and probably hundreds of thousands of kilometers of mined galleries driven for mining purposes also furnish good examples of instability. Wall deformation and/or failure is and must be admitted sometimes. Reinforcement is aimed at ensuring personnel safety and that the operation is carried out in the correct time interval. (vi) Natural caverns provide interesting examples of stability and sometimes of rapid evolution toward instability and caving, which helps in the understanding of some mechanisms. Identifying and understanding the failure mechanisms occurring in oil wells acquired a new significance in the 1980s. For a long time, well logging (single or dual caliper logging) had provided information on the borehole geometry. Boreholes appeared to feature (as evidenced by Gough and Bell [1]) typical failures (ovalization, breakout) indicative of the in situ state of stresses and anisotropy of horizontal components. Where direct measurements were few and costly, such
373 An Overview of Tunnel, Underground Excavation and Boreholes Collapse Mechanisms
methods were used for drawing up a world stress map (Zoback). See Section 14.2.1.1 for further details. Based on conventional tunnels and galleries, as well as on the aforementioned excavations that are less classical for tunnel designers and constructors, the mechanisms affecting isolated or unique galleries will be discussed in Section 14.2. Then, instability indirectly caused by tunnel or cavern drilling will be dealt with in Section 14.3. Finally, the main results concerning important mechanisms will be given since the support and reinforcement designs depend on these mechanisms. As a conclusion, the improvements to be made in rock mechanics for a better utilization in works design will be quoted. In this chapter, 'support' means the forces applied by: (i) the conventional civil and mining engineering ribs, rock bolts, shotcrete, etc; (ii) the fluid pressure acting upon rock walls (as in penstocks, solution-mined caverns, boreholes); and (iii) steel casing and lining relating to the various types of galleries, openings and boreholes. The word failure should also be explained. From an engineering standpoint, failure applies to a construction or operational failure disturbing or impeding the completion or operation of structure. From a rock mechanics standpoint, failure means: (i) occurrence of new discontinuities on walls (true rupture); (ii) recurrence of movements along preexisting discontinuities; and (iii) occurrence of undue continuous deformations leading to structure failure within the engineering standpoint. Comparisons can be made in the laboratory, where a yield value has to be arbitrarily set as a failure point for some triaxial tests. Both meanings of the word 'failure' involve unexpected behavior of rock and/or structure. This chapter benefited from discussions held under the ISRM Commission on 'Rock Failure Mechanisms around Underground Openings' (RFMUO) and from the assistance of the team of the French Committee on Rock Mechanics working on this subject. The first paragraph in Section 14.1 summarizes some of the conclusions given in the Commission's report presented at the Montreal Congress in 1987 [2].
14.2 FAILURE AND INSTABILITY AROUND ISOLATED CAVERNS As suggested by Fairhurst [3], one might think that the stability of such small structures as small-diameter boreholes and galleries essentially depends on the strength of the rock matrix while the stability of larger caverns depends on the rock mass structure, fractures, faults, bedding joints. As is frequently the case, the reciprocal is also true: (i) where the matrix subjected to various stresses is not strong enough to ensure the stability of big openings; and (ii) where the rock mass structure and discontinuities can affect the stability of small structures. Such mechanisms are the cause of problems that have not been identified as such in the past or that are unsolved, leading to construction problems, unsuitable supports and reinforcements. A breakdown of isolated cavern conditions is given below for problem identification purposes: (i) Homogeneous material, isotropic, anisotropic, tight, dry (Sections 14.2.1.1 and 14.2.1.2) porous material (Section 14.2.1.3) especially with high fluid pressure values in proximity to walls. Mechanisms occurring around square galleries will be mentioned (Section 14.2.1.4). (ii) Heterogeneous material in one direction as shown by stratification or alternating of beds (Section 14.2.2.1), or in two or more directions (Section 14.2.2.2). (iii) The particular case of material in flows coming from faults, joints and fractures (Section 14.2.3). (iv) Swelling material (Section 14.2.4), and material subjected to dissolution. (v) Thermal mechanisms (Section 14.2.5). (vi) Instability mechanisms affecting solution-mined caverns (Section 14.2.6). (vii) Failure of dynamic origin such as rockbursts in very deep excavations as observed in South Africa (Section 14.2.7). The breakdown above may seem arbitrary since the various mechanisms can combine with the various structure types. But, it can be helpful to analyze problems, separate variables, identify the essential variable or parameter and design the reinforcement or support.
374
Support
14.2.1 Case of Isolated Caverns in Homogeneous Isotropie or Anisotropie Material Rock All conditions being known (i.e. in situ stresses, material behavior, excavation geometry (circular)), can we predict the occurrence of fracture-shaped wall ruptures? We would be tempted to answer 'badly', but we had better say 'no'.
14.2,1.1 Case ofisotropic tight material under isotropic stress perpendicular to axis The theoretical approach to borehole wall stability relies on elastic models [4] or elastoplastic models. Brown et al. [5] give references for elastoplastic models. (i) Case of lateral isotropic stresses A brief history of the various types of models helps understand the current situation (1990). The initial authors (Fenner, Kastner, Mandel, Labasse, Morrison-Coates, Hobbs, Bray, Diest) considered stress distributions based on an assumed elastoplastic behavior of material, i.e. on Mohr-Coulomb failure criteria regardless of plastic deformations. Hobbs assumed that a sudden strength decrease followed the strength peak, and evidenced an area with reduced properties. Then, Salençon took a significant step forward by dealing with the Tresca and Mohr-Coulomb criteria together with the associated behavior law and plastic volume variation, including large deformations. In the 1960s and 1970s the studies carried out by Daemen and Fairhurst, Lombardi, Hendron and Ayer, Ladanyi, Egger, Panet, Korbin, Kennedy and Lindberg led to improvements regarding bilinear, peak and ultimate criteria. The so-called 'softening' corresponding to a strength decrease at the time of failure development, behavior laws and front displacement [6] were also better known. The effect of radial pressure (equivalent to support pressure) was dealt with by Florence and Schwer, Nguyen Minh Due and Berest. Schwartz and Einstein, Hoek and Brown [7] proposed empirical peak and residual criteria as well as various behavior laws. Risnes, Bratli and Horsrud [8] studied various conditions of plastic flow around a borehole accounting for the effects of longitudinal stresses (application to sand and chalk). Do all these approaches and models provide solutions to the sizing of supports to be installed in tunnels and galleries (see the chapter by Panet, Volume 1, Chapter 27), and more especially what is the internal borehole pressure required for stability? (a) A rough approximation using an elastic model should be attempted in all cases because it would provide an order of magnitude for the stresses acting upon the borehole walls and a first estimate of critical conditions. (b) In addressing the failure question, the in situ observations that are particularly disturbing as compared with the theoretical stress distributions should be taken into account. (c) A great number of researchers (Obert-Stephanson, Berest, Bergues, Nguyen, Haimson, Guenot, Simonyants, Geertsma, Darley, Gay) whose references are reviewed by Guenot [9] observed that the external load applied to a hollow cylinder could be increased (two to eight times) much beyond the theoretical failure point (half of the compressive strength). These unusual stability cases (as compared with test data under homogeneous stress conditions) should be accounted for by new theories. One should cash in on these possibilities for an appropriate and inexpensive sizing of reinforcements. (a) There are practically no continuous plastic deformation areas whether on a centimeter-scale hollow test cylinders or in galleries of some meters or tens of meters in diameter or height. But, for an initially homogeneous material, failure lines separating practically undamaged material blocks can be observed. Failure line spacing ranges from some millimeters or centimeters for laboratory tests to some decimeters or meters for galleries or tunnels. (b) Considering the above-mentioned fractured zone as a weakened but still continuous zone can be practical for computing purposes. An in situ calibration test with the application of loads to gallery areas is necessary to integrate in models equivalent ficticious mean values characteristic of the general material behavior. This is not at all easy and may lead to the selection of wrong parameters representative of neither matrix nor fracture properties - the only real values in such cases. For a new excavation project relating to a rock of a given behavior and subjected to roughly assessed stresses, the 'equivalent' properties of such a partially fractured zone cannot presumably be determined.
An Overview of Tunnel, Underground Excavation and Boreholes Collapse Mechanisms
375
(a) Wellbore
Gallery
σθ > crt > trr '■ Corresponding failure '■ Mode A (shear)
(b)
Figure 4 (a) Definition of stresses at the walls of a wellbore and gallery and (b) intersection failure lines parallel to intermediate stress
(c) Observing failures may lead to such conclusions and questions as reported by the ISRM Commission on 'Rock Failure Mechanisms around Underground Openings' [2]. These are summarized and completed below. Where σθ is tangential stress, ar is radial stress, σχ is axial or longitudinal stress exerted on a gallery or a borehole as shown in Figure 4, and assuming that the borehole or gallery follows one principal in situ stress and that a shear failure occurs with the maximum stress in a π/Α-φ/2 direction according to the bisector of the acute angle formed by the combined failure lines (line intersection following the intermediate stress), failure can have the following forms. The form shown in Fig. 4 with σν < σχ < σθ referred to as Mode A and more particularly as Al. Another form referred to as A2 with intermediate stress, au is expressed as (Figures 5 and 7) σθ < σ, < σΓ
Mode Al corresponds to what is observed in laboratories and sometimes in unlined galleries and tunnels at atmospheric pressure, with one system of spiral failure lines predominant over the other (Figure 6) in the cross section of the gallery or borehole. Mode A2, more difficult to observe as it corresponds to high internal pressures, may apply to galleries under pressure and to overpressure boreholes. While internal pressure has a stabilizing effect for Mode Al, it can have a damaging effect for conditions close to Mode A2, which may occur in the course of pressure tests or round tripping.
376
Support Possible in situ state of stress
O
0.36
1.00
o
2.00
2 77 3.00
0
~p
1
2
2.
3
P
Q_ _ Horizontal stress (isotropic) P Vertical stress Xbh_ Internal pressure iyb mud weight) P
Vertical stress
CaseC=0,<*>=28°
Figure 5 (a) Stability evaluation (vertical well) and shear failure modes (φ = 28°, C = 0) and (b) stability diagram (vertical hole); shear failure modes and related shapes (φ = 28°, C = 0, Coulomb's criterion) (after Guenot [9])
When the tangential stress, σθ, is the intermediate stress, we can have two modes, Bl and B2, expressed as follows Bl: στ < σθ < σι
Β2: σχ < σθ < στ
In this case, too, only Mode Bl is directly observable and Mode B2 involves high internal pressures. The shape of failures is toroidal. Observations made on some vertical shafts under other conditions (heterogeneous rock mass in which deformation is not plane as described in Section 14.2.2.1) may correspond to Mode Bl. Annular shaft failures mostly due to erosion or induced by failures of type Bl can be seen in the less resistant horizons. Similarly, there exist two modes of failure, Cl and C2, corresponding to the following stress patterns with σΓ as intermediate stress Cl: σθ < στ < σι
C2: σλ < σχ < σθ
The associated failures should correspond to helical surfaces appearing as spiral lines on the excavation walls, with σ, and σθ on the bisector of the acute angle formed by the failure lines for Cl and C2, respectively. Presumably, this is the reason why some of the major spiral-shaped failure lines - here induced by drilling operations - appear intermingled (geologists say 'combined') on sidewall logs. Such conditions may exist at times during pressure tests carried out in soil mechanics. Note these failure modes have been reported in the past [10, 11]. However, their practical effects remained unnoticed. They were analyzed again in the 1980s for studying the various states of stresses to which a borehole is subjected [9, 12]. Guenot [9] proposed an elegant representation of the various modes of occurring shear failures versus the in situ state of stress for a vertical hole. The horizontal component Q of the in situ stress is assumed isotropic, plotted as abscissa and scaled to P(Q/P) , vertical component. The internal pressure plotted as ordinate is also scaled to P, P being assumed to be due to overburden weight. Figure 5(a) is drawn for a rock friction angle φ and cohesion C (here φ = 28° and C = 0) (Mohr-Coulomb criterion). This kind of diagram can be drawn for the case of rocks with cohesion (C # 0) and interstitial pressures. Figure 5(b) shows failure patterns for areas outside stability zones.
An Overview of Tunnel, Underground Excavation and Boreholes Collapse Mechanisms
Figure 6 Mode Al failure of a thick wall cylinder under isotropic stress (limestone) (Elf Aquitaine, France)
However, in effective stress, the mud density cannot become larger than the sum of the overburden stress and the rock tensile stress without inducing a horizontal hydraulic fracture. The abovementioned diagrams are then limited by a horizontal line. Note that tensile stresses can be produced on side walls (σθ < Rt) beyond some internal pressure levels. These are the conditions of the hydraulic fracturing method used in the oil industry to stimulate the low-productionfields.Such failures are here sometimes referred to as 'D' mode, 'tensile' or even 'traction' mode. It was previously assumed that failures are induced by shearing. This is frequently confirmed by laboratory tests. In situ observations made in gold mines in South Africa by Stacey [13] and Ortlepp [14] show that side wall failures in excavations, galleries or small openings can be in the form of spalling parallel to the excavation area (Figures 7 and 8), as in uniaxial compression failure, with plane or curved failure surfaces perpendicular to the minor stress. This failure mode would give succeeding curved areas parallel to sidewalls (i.e. normal to minor isostatic). These authors proposed a limit extension strain criterion for conveniently assessing stability. Such failure appears to be quite well illustrated in galleries or openings bored in the sidewalls of galleries under anisotropic stresses. The South African observations relate to extremely brittle materials having an elastic behavior. Surprisingly, softer materials (artificial soft sandstone) were
377
378
Support
aa
σθ <σχ <
στ min Figure 7
tf^min
Failure modes: circular excavations in an isotropic lateral field
Usual assumptions· elastic stresses, Coulomb's criterion
Modified elastic stress distribution (confining pressure dependent moduli / J for instance) /
LI
Isotropic loading 1
Failure mechanism Extension Curved slabs
Shear True or 'pseudos' (Heterogeneity or not perfect isotropic loading) Maximum tangential Anisotropie loading
at the surface
stress:
inside the wall
Successive steps of stress concentration and failure line
Extension Successive failure curved/slabs
Shear True or pseudos Contradiction: Maximum stress concentration - maximum at A' - appearance of failure at "C" and 'C1' f inside the wallB*
=}
Figure 8
Stresses and incipient failure
An Overview of Tunnel Underground Excavation and Boreholes Collapse Mechanisms
379
observed by Périé [15, 16] exhibiting similar failures. Comparable observations were reported by Mastin [17] and Zoback [18]. There exist three possible forms of failure occurrences on gallery sidewalls. The six shear failure modes mentioned above, the true tensile mode of failure ('D'; a tensile horizontal failure mode in case of effective stresses) and the three tensional failure occurrences where failure is normal to minor isostatic (i.e. to the minor stress direction) (Figure 7). In that case, sometimes called 'tensional' mode, failures might be: (a) circular, perpendicular to radial stress, (b) planar, perpendicular to tangential stress, (c) planar, perpendicular to hole axis according to whether the minor stress is radial, tangential or axial. Périé showed failure modes Al, Bl and Cl, and combined A1B1, C1B1 in laboratory tests for cases where two equal stresses were applied. He confirmed Mode B as being a true shear failure (comparison of failure test with Brazilian test conducted on an electronic microscope). A few modes could not be evidenced due to testing limitations. However, note that theoretical predictions help understand observations that could not have been otherwise interpreted, or even noticed. Some of the fractures observed in the laboratory (see [19] for the case of an inclined hole versus principal stresses) can be thought to feature one of these two modes. Well log observations of failures similar to the modes described above are reported by Plumb [20]. These can now be considered as proven. (a) Comments on observations and predictions. Whether the stress distribution is assumed to be elastoplastic or elastic, in situ observations suggest a double paradox. The first one, in the case of an elastoplastic distribution, is that generally no plastic deformation can be noticed on spalls. This is not an inconvenience at the level of computation but is of greater concern to the designers of supports. This fact, daily observed in underground works, does not favor such models. The second paradox relates to the case of an elastic distribution. In this case, the maximum concentration of stresses is at the edge of the hole: Kiersch solution doubled as 2g in isotropic field ß, or 3ßi-<2 2 and 3 g 2 - 6 i i n anisotropic field in direction of Q2 and Qx lateral, with Q2 minimum and Qi maximum stress. Practically, what occurs? Under anisotropic stresses, failures can be clearly observed progressing in the Qi direction but not at the point of maximum stress concentration on the wall of the borehole. Failure is visible, symmetrical in respect to this point (diameter parallel to Q2) (see Figure 8: note that for the A2 mode, failure may not occur at C and C points as with the Al mode, but at the borehole wall). These observations lead to the following conclusions: (1) the maximum stress concentration occurs clearly in the Q2 direction but starts within the sidewalls, (2) such a concentration can evolve into a sidewall shear failure or into an extension failure 'occasionally' ending on sidewalls as a result of spall edge crushing. As in the case of a uniaxial compressive test, the mode of failure is either of shear type or extension type, depending on the material behavior. To fit the aforementioned observations and paradoxes with conventional approaches, Santarelli [21,22] calculated the stress distribution around the hole by varying the modulus stiffening with the confining pressure. He assumed a modulus (E) and a Poisson's ratio (v) depending on radial stress, σΓ, and increasing with it. He obtained the so-called 'hypoelastic' stress distribution featuring the following characteristics: (1) Maximum stress concentration can be within the sidewalls at a few percent radius distance from the periphery, depending on the 'stiffening' law of modulus. (2) Convergence is smaller than that of a purely elastic model, as reported by many authors. (3) Contrary to elastic or elastoplastic distributions, such models typically never feature a maximum tangential stress concentration at the sidewall, thus leading to elastic and reversible stress redistributions. Santarelli's model mentioned above (which can be based on invariants) accounts remarkably for an internal sidewall stress concentration, resulting in shear or tensional stress failures with material practically intact between the failure lines. One may say the question has been shunned for the real properties of the material remaining after the front progress must be included in such models. That is true, but the author thinks that the resulting damage—which is greatly variable and one of the causes of the scattering of observations—should be taken into account. Such a model can be used as a basis for in situ observations and new interpretations [23]. It has been applied to anisotropic stresses by Duncan Fama [2]. (b) Applying the bifurcation theory. The failures occurring on hollow cylinders subjected to external isotropic stresses in the laboratory arouse controversy. Does the failure occur on the whole periphery of the hole at one time? Or is it initially present at two, three or more points on the hole periphery? It is difficult to say.
380
Support
For some experimenters aware of the fact that a perfectly centered hole is difficult to put under fully isotropic stress, the stress field is never isotropic, which accounts for the occurrence of local failures prompted by material defects. For other people, as axisymmetric failures obtained in the laboratory are very few, even upon application of stresses as isotropic as possible, the failure process is different. In their view, the material behavior, boundary conditions and sample geometry combine to produce deformations that are not homogeneous but located at places where failure lines will occur later. Observation of these failures reveals they are somewhat regular. Regularity is by no means casual. It is comparable to that observed in geology. The bifurcation theory was used by Vardoulakis [24] for determining, among others, failure conditions around a circular hole under axisymmetric stress using various instability criteria. Normally, the deformation mode is determined from equilibrium equations through a solution in accordance with relevant boundary conditions. Such a solution is based on the axisymmetry of stresses, which helps solve the problem. However, for some behavior laws and boundary conditions, this hypothesis appears to be too restricting. In fact, there may exist irregular solutions assuming that not all stresses are symmetric. These solutions are called 'irregular' as compared with the 'regular' or trivial solution mentioned above. The bifurcation analysis determines the threshold of irregularity. For a well, Vardoulakis et al. [24] showed two types of possible deformation locations interpreted as foretokens of failure lines: (1) deformations along a surface, called the shear band; and (2) deformation related to those producing free surface instabilities. Such periodic deformations from kinds of waves on the walls of a borehole (they are referred to as buckling though being different from buckling within the strict geometric meaning). Depending on period, some portions of the walls of a borehole are radially strained, up to failure in places. It is worth mentioning that these two types of instability correspond to above-described in situ failure observations. This theory also provides an explanation to periodicity and indicates the limits of a regular solution. Based on a rigid plastic behavior law and on a nonlinear volume change law, Vardoulakis et al. predicted rather accurately the stress to be exerted on a hollow limestone cylinder to produce failure. Using this theory, the instability mode (shear band or surface) to occur first and the stress level instability at which failure occurs can be also predicted. This approach should help understand a scale effect frequently observed since stability is easier to secure in smaller than in larger size openings. The use of theory - not yet proven - is too recent to get an idea about all applications. This theory opens up a new considerable perspective in that a leap forward could be made in some stagnant fields of rock mechanics. Note that the application of this theory to failures on even surfaces can also be contemplated [25] and used for square gallery instabilities (sidewall buckling). (ii) Case of lateral anisotropic stresses Stress concentration is the same as under elastic conditions (with internal pressure equal to zero) : σΓ = 0
σθ = Ρ[Χχ + K2 - 2(KX - K 2 )cos20] σχ = P [ l - 2v(Kl - K 2 )cos20]
where Qi = K1P,Q2 = K2P, P represents major, minor horizontal and vertical in situ stresses and Θ represents the polar angle formed by the point under consideration on the hole periphery with the direction of Qv If an internal pressure is exerted in the hole, σθ can be obtained by addition and subtraction of this value in the preceding formulas (sometimes notations σΗ, ah or σ2,σΐ9 or ahmax, tfhmin»are usec* f° r ^ e lateral component. The same notations will be used when quoting the authors who have previously used them). Elastoplastic distributions are also proposed [26]. All observations relating to underground works are more concerned with stresses evolving into failure than with continuous plastic deformations that occur very rarely on the walls of galleries or boreholes. A high stress concentration (361-62) is obtained on the diameter in the direction of producing possible shear (or extension) mode failure on this diameter. On the diameter parallel to Qu tensile stresses (case of negative 362~6i value) can be induced, producing traction mode failure (if Rt = 0).
An Overview of Tunnel, Underground Excavation and Boreholes Collapse Mechanisms Shear (or extension) failure modes on diameter parallel to Q2 have been the subject of many laboratory studies. Borehole ovalization and other in situ analyses are considered at the present time as a preliminary step toward determination of in situ stress anisotropy. In fact, the possible failure modes are more complex. Conclusions that ovalization of break-outs is parallel to the minor in situ lateral stress component holds for almost all the in situ stress conditions but not strictly for all. In addition, complete analysis of mode of failure allows us to: (a) go further into the interpretation of observations, (b) gain a better understanding of the 'supporting' effect of internal pressure, and (c) obtain more accurate data on the in situ state of stresses. Figure 9 by Van Dyck [27] shows the shear and traction mode failure that may occur around a circular hole in a stress field P, parallel to hole axis with crH( = Q1) = XP and ö"h (= Qi) = ΝσΗ (= NQX) and internal pressure equal to that exerted by a water column equivalent in height to the depth considered (borehole or gallery under hydrostatic load). It appears that all failure modes can combine to give composite failures that are otherwise difficult to explain when appearing on the walls (or logs). Several conclusions can be drawn from this figure, which only applies to the internal pressure under consideration (yb = 1, i.e. d = 1/2.5 = 0.4 with d = 0.4 corresponding to the ordinate of the stability diagram of Figure 5). It is assumed below that only Modes A and B result in failures producing hole ovalization; this should be confirmed by systematic well logging or laboratory tests, but is quite feasible. Under this assumption, it can be seen that the inferred ovalization direction (breakout along minor stress) appears to hold true for nearly all cases except for extremely dissymmetric lateral stressfieldswhere combined Modes A and B may simultaneously occur in the direction of Q2 (A) and Qx (B). The conclusions drawn from ovalization analyses should be corrected for plot data taking into account the internal pressure in excavation. Recent observations seem to confirm these modes: Bl for vertical wells, C for oil wells ([19, 20]; not confirmed) combined shear and traction Mode (Al and D) for boreholes in the Paris Basin, combined Modes A and B [28]. Examination of cavings falls produced by failures occurring on the walls of a borehole can reveal typical forms that may help better define the borehole shapes appearing on logs. In Indonesia, rock falls characteristic of anisotropic Mode Bl could be observed by the French Total Oil Company. Still more characteristic rock falls are encountered in drilling in argillites: helical lamellae up to 8-10 cm in length, 1 cm in width and a few millimeters in thickness. The state of the art does not provide interpretation for regular and recurrent failures. Such failures relate to the most complex case with anisotropic material deformation and strength (see Section 14.2.1.2) subjected to an unknown stressfieldprobably disturbed thermally and by pore pressure action. Surprisingly, similar
(Vertical well, 7^ = I, total stress, C-0,
0 = 35°)
Zones 1 Stability zone 2 Ovalization in crh direction (Mode AI) 3 Mode CI failure in aH direction 4 Ovalization in <7h direction (Mode ΔΙ) and failure in
Figure 9 Stability diagram: anisotropic in situ stress tensor (after Van Dyck [27])
381
382
Support
Figure 10 Helical sidewall spalling in a 2 m high gallery (Hartebeestfontein, South Africa)
modes can be observed in machine-drilled headings of rigid, elastic and extremely brittle quartzite (Hartebeestfontein mine in South Africa; Figure 10) with little marked bedding, but weak cohesion of joint partings. (Hi) Case of anisotropic stresses in oblique direction with respect to excavation In this case elastic stress distributions must be recalculated using analytical formulas such as those given by Hiramatsu and Oka [29] or Fairhurst [30]. In industry, plots of the type proposed by Bradley [31] are used. Figure 11, similar to that given by Guenot [32], shows inclined well stability assessment using Model MGP2. Such a model calculates the internal pressure (and drilling fluid weight) required to avoid subjecting the walls of the well to critical stresses. It also gives the failure mode, the possible effect of thermal stresses, and effective stresses other than those proposed by Terzaghi (Biot's coefficient). In spite of some uncertainties, these tools are most helpful lor any project. They can be used as a guide to interpretation of first results, observations and/or failures. They help make calibrations in the absence of local data, as for example, on the state of stresses. In turn, they help give a better assessment of the in situ state of stress inferred from observations in several wells. 14.2.1.2 Case of anisotropic tight material For any information about the properties of such materials, theoretical distributions of stresses under elastic conditions, and other references relating to previous studies (Filon, Leknitskii, Niwa and Hirashima) see Chapter 17, in Volume 1, and two basic publications by Amadei [33, 34]. We shall limit ourselves to the following remarks about failures that may be produced in such materials. (i) The case of total anisotropy is still out of reach and beyond practical concern for the time being. (ii) The case of anisotropy with transverse isotropy seems to be more frequent and of first concern (bedded, stratified, sedimentary, metamorphic rock mass). Determination offiveelastic deformation constants is possible, at least in the laboratory. There is still a great delay in the determination of in situ anisotropic properties and effort should be made tofillthe gap, as noted below. The problem is more complex for anisotropic strength criteria where anisotropy depends on confining pressure (see Amadei [33, 34] for relevant references).
An Overview of Tunnel, Underground Excavation and Boreholes Collapse Mechanisms
383
(a)
2.35
1.60
1.75
Mud density
Figure 11 Borehole stability evaluation as a function of deviation and internal pressure, anisotropic field (a) borehole azimuth in direction Ql and (b) borehole azimuth in direction Q2. Marls: C = 7.3 MPa, φ = 38°, v = 0.2. Effective stress calculation, good cake quality. In situ stress: P = 2.6 x 3500 = 91 MPa, Ql σ„ = 0.9 P, Q2
(iii) Stress distributions around a circular hole under anisotropic elastic conditions are extremely complex for various reasons. (a) The in situ stress field orientation may be different from the axes of material anisotropy. (b) The condition of plane deformation is no longer satisfied in the general case. (c) Stress distributions depend on the elastic properties of materials. Fortunately, this is not the case with isotropic materials. (iv) Assuming all experimental and analytical problems are solved in the determination of properties including the magnitude of in situ stresses, comparison should be made with anisotropic criteria with respect to strength, which is still more complicated. Understanding the failure mechanisms to which the openings and boreholes drilled in such materials are subjected may seem Utopian. Is such an idea hopeless? The author thinks not for the reasons below. (i) This is an essential question in some industries. To develop an oil field, tens of wells have to be drilled through large intervals of argillaceous anisotropic overburden materials with extremely varied azimuths and inclinations over distances of several hundreds of meters. Drilling problems are frequent and costly, especially in offshore wells.
384
Support
In addition, anisotropy perniciously tends to cause boreholes to deviate from azimuth and assigned path. The so-called spontaneous deviation is greatly variable and unpredictable, though generally dextral. It can be observed in oil wells where it can be a cause of target missing. In civil engineering works it occurs over short intervals (40-100 m) leading to serious inconveniences (heading investigation, grouting curtains). (ii) Observations in underground excavations in anisotropic rocks could certainly be beneficial, but are complicated by the very nature of the rock and require much more extensive monitoring than for isotropic rocks. Effort in the field of rock mechanics should be focused also on operators and contractors who need to be convinced that these troubles can be avoided. Observing rock falls occurring during drilling shows systematic and typical forms of failures (Section 14.2.1.1.1). Similar data have been observed by many oil operators. (iii) Analysis of accurate failure conditions (internal pressure, temperature) in vertical wells (or parallel to the axis of isotropy) should be a first step toward the understanding of the state of in situ stresses. (iv) Comparison with the hole geometry inferred from logs can help better determine features (ovalization, shear lines, tensile failures). (v) Overburden sampling (not always readily available) is a means to determine material anisotropy and effect on stress distribution. Knowing, for example, a modulus 2-5 anisotropy has little effect under a given confining pressure and a critical effect beyond certain pressure values is important, and it would be worth going further into this subject by making use of the theoretical results obtained. Failure mechanisms occurring in anisotropic materials are certainly complex. Systematic observations suggest that in situ observation of simple cases will improve knowledge of the subject. In our present view, given that such material behavior is of concern to many industries (oil, waste storage, etc.), it would be convenient to have a field laboratory for analyzing and testing in situ such anisotropic materials as argillites and for testing and calibrating theoretical models. 14.2.1.3 Case of porous material Analyzing the failure mechanisms occurring in the aforementioned dry materials is essential to the understanding of basic phenomena and interpretation of laboratory tests carried out to determine the validity of certain theories and criteria. This investigation is also important in the case of works located above the water table or for openings driven in materials where fluid occurrence is insignificant or plays a minor role: salt, evaporites, argillites, tight eruptive rocks. These cases excepted, the other rocks are porous. To analyze the failure mechanisms having an effect on them, it is necessary to determine the stress distribution, there, based on the acting forces, in situ stresses and drainage (or injection) carried out through the gallery or borehole. In these early 1990s, this subject is again of considerable concern given the number of publications presented at the ISMR Symposium in Pau, 1989. As appears from these University or oil industryderived papers that evidence specific and particular failure mechanisms, the outlook is promising, even for civil engineering works (such as mines and deep underground structures). People not acquainted with the subject should refer to Detournay et al [35, 36] for a detailed theoretical approach, and relevant references. In short, fluid presence in a porous medium may have the following effects. (i) Pore pressure is generated by the action of external stresses. (ii) Bulk volumetric deformation produced by 'effective' stresses different from the external applied stresses. (iii) Pore pressure gradient acting as a volume force (to be included as such in equation of equilibrium relating to effective stresses). (iv) Dispersion of pore pressure according to a diffusion law. As a result, volumetric deformation depends both on drainage boundary conditions and on the possibility of drainage of the rock itself. Assuming drainage boundary conditions and a rock of given permeability, the volumetric deformation will depend on the loading rate (isotropic) since part of stress exerts pressure on the fluid, causing it to flow out from the material. Asfluidescape duration varies with the loading rate, the rock will look more 'rigid' when undrained. Except for specific boundary conditions, deformation and diffusion processes are generally combined. And, as stated by Detournay, 'they should never be handled of separately in rock-fluid modeling'. In the initial combined approach used by Biot based on the so-called poroelasticity, the behavior law was only contemplated. In Terzaghi's separate approach based on observations and test results
385 An Overview of Tunnel, Underground Excavation and Boreholes Collapse Mechanisms obtained for granular materials, boundary conditions are the essential characteristic considered (one-dimensional consolidation). Since that time, the following progress has been made: (i) Determination of generated pore pressure represented by Skempton's coefficient B (undrained conditions). (ii) Bulk volumetric deformation governed by the so-called effective stress that differs from the Terzaghi effective stress in that it is expressed by Biot's coefficient α (σ' = σ — ocP instead of σ — P for Terzaghi). (iii) Law of pore pressure diffusion coupled with the volumetric deformation variation rate. (iv) Associated pore volume change significant for fluid recovery applications. Based on the new formulations of Rice and Cleary [37], constitutive equations, equilibrium equations, Darcy law and fluid phase continuity equation, Detournay and Chang [35] arrive at the following conclusions (beware of signs and notations; this chapter uses those employed by Detournay: minus for compressive stresses, plus for tensile stresses). Under an in situ isotropic load, when borehole pressure is equal to pore pressure in the rock mass, deformation is deviatoric (Lame solution) with no poroelastic effect.
(a)
(b)
1.4
1.8
Radius, r/a
(d)
(C)
Radius, r/a
(e)
S / = ct/ad
Time, / Figure 12 Thermoporomechanic effects (a) isochrones of the pore pressure variation with radius at Θ = 0, π for mode 3 (v = 0.2, vu = 0.4, B = 0.8), (b) isochrones of the tangential stress with radius at 0 = 0, π for mode 3 (v = 0.2, vu = 0.4, B = 0.8), (c) isochrones of the tangential stress variation with radius for mode 2, (d) pore pressure history at various r/a for mode 2 and (e) radial displacement history at r/a = 1, Θ = 0, π for mode 3 (v = 0.2, vu = 0.4, B = 0.8) (after Detournay [26])
386
Support
(i) The resulting stresses and displacements do not vary with time. (ii) Pressure variation, δΡ, occurring instantaneously on the walls of borehole (Figure 12a and b) causes a total tangential stress variation, δσθ whose asymptotic value in time can be expressed as (5σβί°° = ηδΡ(1 + a2/r2)
in which
η = α(1 - 2v)/2(l - v)
This value is instantaneously attained on the wall (Figure 12c). The initially compressive stress (except on the walls of the borehole) changes later to a tensile stress, where there is a pressure decrease and if no additional stress is applied. Speaking of effective stresses (Terzaghi's for example a = 1) the 'effective' tangential wall stress is a compressive stress in that case δσθ + δΡ = (2η - 1)δΡ (α = 1) (i) Radial displacement in the medium considered (asymptotic value for infinite time) expressed as U =
'
αηδΡίτ α\ "2ß\2"r)
is nil on the walls and not nil inside the walls (Figure 12e). (ii) Pore pressure dissipation versus time is as shown in Figure 12d. For an in situ anisotropic loading when initial pore pressure is nil everywhere, the poroelastic effect produces a pore pressure no longer evenly distributed over the walls of the borehole. Pore pressure varies with azimuth (relative to minor stress). It can be expressed as follows as time t (of drilling) PfoO)tm0
= (4/3)50ß(l-hvu)(/l2/r2)cos2ö
where So is the deviatoric component of in situ stresses, B is the Skempton coefficient, Θ is the azimuth relative to minor stress (according to Detournay, minus sign for compressive stresses). As a result, there occurs: (i) A sharp pore pressure radial gradient in proximity to borehole walls due to pressure conditions in the borehole (zero pressure is included in the calculation mode proposed by Detournay corresponding to this loading case). Gradient lessens with time. (ii) A tangential flow gradient. (iii) Rock behavior far from the excavation: undrained since rock is drained on the walls (not inside). As a result far rock looks more rigid. The nearby wall area of the borehole is somewhat 'relieved' of stresses. As is apparent from the theoretical calculations, the (total) tangential stress produced in the case considered is at the time of drilling δσ*θ=0 + =
- ( l + ^JSocos20
On the walls of the borehole it is instantaneously reduced to - 4
l-vu
-Socos20
1—v
with maximum stress value inside the walls within a very short time after drilling, evolving into a regularly decreasing pattern. The essential point, in this case, is that maximum stresses may concentrate at one point within the walls of the borehole within a short interval after drilling, thus causing failures to occur inside the walls. Note that such stress concentrations are always located on a diameter perpendicular to maximum in situ compression, which gives additional support to the conclusions drawn from borehole ovalization analyses as to the direction of stresses perpendicular to the axis of a borehole. The theoretical analysis of radial displacements shows they are dependent on cos 20 and vary in time. Such displacements tend to increase the size of galleries in the direction of the minimum compressive stress, as reported by Carter and Booker [38]. For values of parameters mentioned in this paragraph refer to Section 14.2.1.5.
An Overview of Tunnel, Underground Excavation and Boreholes Collapse Mechanisms
387
Practically, as far as failure mechanisms are concerned, the preceding theoretical approach reveals important facts that are brought about by poroelastic effects: (i) In the event of isotropic stresses, stress concentration occurs on the wall immediately. (ii) Under an in situ deviatoric load, a maximum tangential stress concentration may occur within the wall, and cause failures at this point. Numerical values used by Detournay locate this point at 5-10% radius distance, which may be one explanation among others to the paradox mentioned above (see Section 14.2.1.1.i a Comments on observations and predictions) for saturated porous materials. (iii) In the event of anisotropic stresses, delayed failures can occur on walls, as the tangential stress varies in time. The aforementioned authors give variations, on the order of the in situ deviatoric component, i.e. sometimes considerable. Such mechanisms should be kept in mind in the event of failures that are unexplained or thought to be connected with 'creep' occurring in little creep-prone rocks. (i) Case of very high pore pressure in the vicinity of excavation The case of high pore pressure in areas close to excavations or within faults or bedding joints is discussed in Sections 14.2.3 and 14.3. Practically, there may be failures stimulated by a high pore pressure in the rock matrix. Section 14.2.7 gives examples of mine occurrences with gas pressure in sandstones (wall rock of coal beds) being the cause of sudden failures. Decreasing pressure is one of the miners' concerns. Note the same phenomenon occurs in oilfields[39] more especially in the form of matrix failures. Pore pressure may reach 70-80% of the geostatic stress (that is twice the value of hydrostatic pressure at the depth under consideration). In the case of a drilling fluid whose weight may range from 1 to 1.4 or 1.5, conditions are underbalanced as compared with the formation pore pressure. Effective tensile radial stresses can be produced. Such stresses are considered to be a cause of hole failure. Generally, such failures are the result of high pore pressure combined with other factors (in situ stresses, temperature). It is difficult to determine the role played by each of these causes. Laboratory investigations may help in such cases.
14.2.1.4 Case of square, rectangular or straight sidewall galleries In homogeneous materials, galleries may have such forms in the following cases: (i) mechanical working of a layer richer than but with mechanical properties comparable to the surrounding formations, e.g. salt mines, other evaporite rocks; (ii) specific extraction modes: chalk, limestone quarries; (iii) when required by the future use of excavation: vertical sidewalls for power plant. Many authors have given solutions to the problems of stress distribution around such openings and stress concentration (known to all mechanics, depending on a sharp or round corner: see [5,40, 41]). These excavation types can be affected by: (i) failures due to an excess tangential stress concentration at the corners producing shear stresses. Initially, these failures are local and do not pose any problems. But, if they extend, they may eventually jeopardize the restraint of the roof, sidewalls and invert while giving rise to the failures described below. Such corner failures can be observed in the German salt mine of Herfa-Neurode, among others. (ii) large extension failures parallel to sidewalls delineating meter-scale thick slabs. These failures are noticeable only when leading to excess buckling and failure of separated slabs. However, they can be detected by the hollow sound produced by a loose wall. Such failures are known to all underground miners. Some remarkable examples of these failures have been observed in salt mines of Louisiana (with 10-15 m high, 1-2 m thick spalling on vertical walls in 25-30 m high rooms) and at Louveciennes in the Paris district (with 3-4 m high, a few decimeters thick spalling in chalk galleries on 8 m high walls). Similar observations have been reported for large underground power plants in hard crystalline rock (Echaillon plant in the Alps). Such failures call for the following remarks: (i) They are perpendicular to the minor (radial) stress and result from extension. No sign of shearing can be seen.
388
Support
(ii) Their geometry - essentially thickness - is considered probably to depend on: (a) gallery slenderness (to be taken into account in observation interpretations); (b) state of in situ stresses; and (c) material behavior. In fact, the behavior parameters acting upon the start of failure and, thereby, on spalling thickness and extent are still unidentified. Practically, as far as the sizing of gallery supports is concerned, slightly curved walls may be a solution to all problems in that there will be no need for support and additional excavation will be very limited. Driving a 5 km gallery in soft chalk highlighted this phenomenon (Figure 1) [42]. Some geometrical changes (less sharp angles and slightly curved walls) can reduce considerably the need for supports in the future. Identical phenomena may affect straight working faces, thus producing explosions where there is no support. There, too, a slightly curved surface could prevent instability. There is still progress to be made before the conditions that determine the occurrence of tensional failures can be anticipated and understood. Failure mechanics and bifurcation approaches have opened up new perspectives considering the relationships that have been found between spalling/splitting and instability evolution in free surfaces [25, 43]. 14.2.2 Heterogeneous Medium In this section, the term 'heterogeneous medium' covers: (i) the rock mass consisting of continuous material with different and sometimes contrasting geomechanical properties, (ii) the rock mass consisting of homogeneous, possibly identical, material blocks or beds. As a result, 'heterogeneous medium' applies also to the fractured rock mass dealt with in the failure mechanisms described below. A rock mass with one prevailing heterogeneity direction as observed in bedded, sometimes alternated, sediments will be distinguished from those having several heterogeneity directions, that are essentially fractured materials characterized by two or more discontinuity patterns. In all cases, excavation stability or failure will depend on its geometry (cross section, direction, position) as compared with the geometry and properties of heterogeneities (direction, geomechanical properties) (see Figure 13). As all possible cases could not be dealt with here, we just tried to categorize the various types of rock mass, available tools, and significant related questions. 14.2.2.1 Rock mass having one prevailing heterogeneity direction This case is that of galleries or openings driven in sedimentary rocks. The following factors are involved in the possible failure mechanisms: (i) dip (of beds and discontinuities); (ii) gallery directions as compared with the direction (or cross-cut) of beds or discontinuities (gallery strikes and cross-cuts); (iii) the cross-sectional shape: continuous roof or indented by excavation; (iv) excavation width as compared with bed thickness or discontinuity spacing; (a)
(b)
Figure 13 Stable cavity shapes: position with respect to rock mass (a) Galleries in the direction of the bedding; pentagonal shape with minimum support and (b) cross-cut gallery (case of Gypsum Mine, St Pierre Martigues, France)
An Overview of Tunnel, Underground Excavation and Boreholes Collapse Mechanisms
389
(v) position selected in sedimentary series (board left) for example, in the roof to ensure stability with minimum support. In most cases, such decisions are made empirically based on a quantitative evaluation of the possible failure types or mechanisms. Two kinds of extreme conditions can be observed: (i) Failures due to the contrasting mechanical properties of materials with no 'failure' of boundary joints. A typical example is that of pillars in quarries of limestones having different properties. Failures occur in the more rigid horizons whose limit extension strains are smaller than the less rigid ones. (ii) Local movements (sliding, parting) along joints producing later changes governed by joint behavior. In that case, possible instability is in direct connection with the evolution of movement along joints (as blocks sliding over faces until falling). As an example, in the case of vertical bedding and of galleries driven along the bedding, stability first depends on the properties of the vertical bedding joints, then on the behavior of the roof slabs under in situ stresses. The iron mine at May-sur-Orne in the Normandy province in France [44] provides examples of stability with 70-90 m high, 5-7 m width and 80-10*0 m long rooms supported by a minimum number of horizontal pillars in each room (Figure 14). In some rooms with no pillars, vertical slab buckling failures occurred whereas the collapse could have been avoided by rock bolts. Failures were accompanied by an upward movement of collapsed structure to the so-formed cavity top, which was a source of trouble for operating. On the other hand, instability caused by a massive roof collapse in shallow vertical slate quarries under probably low horizontal stresses was reported. The slate joint and sidewall properties combined with state horizontal stresses to cause the whole roof to fall at once. In the case of horizontal bedding, the state of in situ stresses is also of particular concern: horizontal stresses are more numerous than anticipated. Gallery direction with respect to the minor and major horizontal stresses is a determining factor as it may be a cause of extremely varied failures [45]. For all cases, analysis and modeling tools applied to bending supports or slabs use conventional analytical calculations [46] or numerical calculations [47]. Cross-cutting (perpendicular to the heterogeneity direction) these materials does not give rise to many problems generally. In the case of oblique bedding when working along the bedding, the thickness and inertia of layers as well as the properties of joints are determining factors in the handling of the problem of supports (selection) and application of required safety precautions (temporary for mining galleries, permanent for public access structures). (α)
(b)
im.
(c)
Pillars
70-90 m
/
σ
Figure 14 Slab buckling failure and belt-shaped collapse, (a) Stability secured by a few pillars, (b) slab buckling roof failure and (c) belt-shaped collapse due to joint failure (case of an iron mine at May-Sur-Orne, France)
390
Support
The 5 m high, 8 m wide room and pillar gypsum mine is excavated in a monocline characterized by an alternating of decimeter-scale marl and gypsum beds at St-Pierre Martigue, France. Galleries driven along the bedding have a particular pentagonal shape (Figure 13). A limited rock bolt and wire-mesh support ensures bedding stability in the case of such uncommon patterns. Perpendicular cross-cut horseshoe-shaped galleries are stable by themselves (scaling and wire-meshing are necessary to prevent falls). It appears from these examples that selecting an admissible or suitable shape pays off in terms of support.
14.2.2.2
Rock mass with several heterogeneity directions (bedding, fractures)
In most cases, these materials are stratified sedimentary, crystalline or metamorphic rock mass featuring one or more fracture directions. Beside the numerous sedimentary mines, many storage caverns have been constructed in this type of rock (fractured type) with minimum support (data on similar problems encountered in underground storage projects are available in Rockstore 77, Subsurface 80 and Zacatecas Symposium 1985). Unsupported caverns 20,30 and even 50 m high can be excavated provided that the shape and direction of galleries should be appropriately selected. Identical problems are sometimes encountered in designing underground power plants. The case of several heterogeneity directions can result from the preceding situation where fractures induced by excavation are combined with a preexisting heterogeneity direction [48]. As well as the matrix failure mechanisms dealt with in Section 14.2.1, rock mass instability mechanisms can also occur in such materials. Rock mass stability depends on: (i) matrix properties and joint system: generally the matrix is very rigid compared to less reliable joints; (ii) state of in situ stresses (permanent uncertainty about it); (iii) gallery geometry, i.e. direction with respect to joint system, size and shape; (iv) excavation mode, namely quality of work; (v) position and evolution of hydrogeological seepage pattern around the gallery. The last two points are left out in some studies. The inconvenience of this omission is discussed in Section 14.2.3. What are the available tools that can be used for assessing stability conditions and predicting the failure risks to which these structures can be subjected as a result of the aforementioned occurrences? In the construction phase, visual monitoring is useful but difficult to perform in high caverns. Telemetering of convergences can be an appreciable aid though sudden rock mass instabilities can be missed. In the author's view, a stereographic projection analysis of all possible block equilibria determined from joint measurements should be systematically made as it may be a useful guide to in situ examinations. In addition to this watching and monitoring, an eye should be kept on the hydrogeologic pattern, joint pressure and possible changes in the hydraulic properties of the joint system. Serious accidents during the construction of underground structures may be due to insufficient watching or to unexpected events that cannot be avoided in the case of a deficient or unavailable monitoring system. At the stage of design and preliminary studies - on which most effort has to be focussed a sufficient amount of data should be available for an appropriate knowledge of the joint pattern. These data are essential in the determination of shapes, sizes, directions, excavation sequences best fitted for minimum supports and increased safety and dependability . Clients and contractors will have to be convinced that improving the knowledge of a geological framework will pay out more than 100 or 1000 times the not always obvious construction delays and extra costs. Among the theoretical tools used for evaluating works stability, two main approaches can be taken. The first approach focuses on the behavior of one individual block (or a small number of individual block masses) and analyzes its stability. Block description varies from one method to another. Blocks can be identified individually as such [49, 50] or by their faces, as in the case of Goodman's 'pyramids' [51] where they are defined as portions of a volume bounded by discontinuities set by a direction (at least in the initial version of their theory). In all cases, depending on geometric conditions and under the action of its weight, block can be displaced by direct falling, sliding over its proper faces (Coulomb's Law) or tilting [52].
An Overview of Tunnel, Underground Excavation and Boreholes Collapse Mechanisms
391
The effect of a block displacement upon adjacent stability is also approached by these methods (search of key block, iterative methods). Many problems have to be solved as to block convexity or lack of convexity, integration or nonintegration of boundary joint deformability (especially dilatation properties) or introduction of in situ stresses in stability analysis. But, initially, the essential question is to know whether a model-builder can work with a discontinuity pattern realistic enough and representative blocks of known geometry. One should be easily convinced that a statistical reasoning is advantageous and several simulations of a fractured rock mass are necessary [53] if it is demonstrated that an unstable block can be split into two equal stable blocks. The second approach relates to the comprehensive behavior of the rock masses cut by discontinuities and considered as a block assemblage. It is referred to as a discrete element method as proposed by Cundall and Hart [54]. This designation also involves a method known as the distinct element method, expressed as the UDEC code. These calculation methods are advantageous in that they allow for large block displacements possibly accompanied by a complete separation and parting of some rock portions from the surrounding rock. Contacts may be modified in the course of calculation and the resulting movements are followed step by step. Blocks are deemed to be deformable (discretization is made using a conventional grid and finite difference or finite element methods applied depending on codes). The distinct element method appears to be promising as it provides an increasingly number of successful applications. Fundamentals amount to contact (friction or parting) modeling, use of conventional equations of motion, and to a numerical solution based on dynamic relaxation. Note also how important it is to make a reliable geometric representation of a partially visible fractured block. These modeling modes are fitted for a limited number and extent of fractures. A rock mass featuring a great number of fractures is generally treated as a comparable continuous medium. But models are not expected to reproduce suitably too irregular fracturing conditions. After some 20 years of working on underground projects it is tempting to summarize below how designers feel about these new numerical tools: (i) These tools can be extremely valuable at a preliminary reflection stage of the project and in making decisions upon work design. (ii) They can show up outstandingly a lack of input data, denoting insufficient geological and hydrogeological knowledge and consequently point to areas that need to be further analyzed. (iii) Their use for excavation and support design should be limited because of the scatter of joint properties (in situ determination of properties is very difficult) and mostly of a poor knowledge of in situ stresses. The current capabilities of numerical models far exceed in situ data acquisition and corresponding effort and possibilities expected.
14.2.3 Particular Case of Invading Material In the case of high fluid pressure (as described in Section 14.2.1.3) associated with heterogeneous features such as fractures or faults, failures can occur and be accompanied by a hazardous invasion of large quantities of materials. This is not a typical failure mechanism but the result of high hydraulic gradients in the close vicinity of excavation. It appears from the analysis of various case histories that failure control should be based more on preventive site measures than on sophisticated mathematical approaches. Nevertheless, the author thinks that such significant phenomena from a practical standpoint cannot be omitted from the chapter dealing with failure mechanisms. Painful cases of such failure mechanisms are plenty in many countries. A very good illustration of them is given by what occurred in 1959-1960 in the Litani gallery in Lebanon [55] as summarized below. The Litani project involved driving a 4.40 m horseshoe-shaped gallery from several 'windows' under a locally 100 m thick overburden. At the time the heading was at 3207 m from the entrance, large water inflows occurred in 1959 invading the gallery invert and depositing sands. The initial 700 L/s flow rate decreased to 200 L/s as sand invasion covered a 1000 m gallery interval. Attempts were made to stop such inflows by installing tight doors, coffer dams, succeeding walls that failed one after the other as the gallery was invaded over increasingly large distances. At the end of January 1960, the last retaining wall failed causing the whole gallery to be filled with Aptian sandstone derived sand. The flow rate was estimated at 6000 L/s for the first hour and at 3000 L/s for
392
Support
the next 12 h before stabilizing at about 100 L/s for months. Pressure estimations attained 7.5 MPa. Resumption of tunneling was long, difficult and costly. The conclusions that can be drawn from this case are that excavation is difficult when nearby areas are subjected to high hydraulic gradients, flow-rate evolution is practically impossible to predict, precautions have to be taken, and control measures are necessary for areas believed to be invasion-prone upon heading. In practice: (i) Geological structures must be carefully watched, especially: (a) wide faults with crushed and uncemented unfills, (b) karstic rock intervals where water table upward movement can cause infills to 'overflow', (c) permeable beds (sandstone, sand) in series that may be subjected to a high hydraulic loading, (d) altered fracture system. (ii) Heading sounding is the best detection method. (iii) Injection must be remote enough from excavation to avoid having high gradients in proximity to excavation. (iv) The author thinks that sudden failure occurrences are a sufficient reason for stopping an excavation work while operating upstream, especially in karstic areas. Although this question does not strictly come within this chapter, it is too important to be overlooked. Rocks mechanics as well as rock engineering should rely on the experience gained in underground excavations.
14.2.4
Case of Swelling and Soluble Materials
The case of swelling and soluble materials is not best addressed by rock mechanics probably because physical and chemical phenomena are involved. However, temperature, pressure and in situ stresses also play a role in making chemists and mineralogists ill-at-ease. Due to the natural intermingling of materials and parameters, separating variables is essential to problem analysis. Schematically, there are three possible swelling materials or families of swelling materials: (i) anhydrite, (ii) shales, especially smectites, and (iii) pyrites. The following solution materials can be encountered locally associated with these materials: (i) gypsum, resulting from anhydrite hydration, (ii) salt and other evaporates, multiple cation salts. Generally, evaporite series feature alternating gypsum, marls, anhydritic or gypsum marls and mixed up materials, making analyses difficult.
14.2.4.1
Case of anhydrite and gypsum
The same failure mode is shown by nearly all case histories: invert uplift and excess convergency of the sidewall base (whether concreted or not), giving rise to large-scale delayed thrusts (examples of tunnels: Genevreuille, 1855-1858; Bozberg, 1871-1875; Krappelisberg, 1878-1880; Ricken, 1903-1908; Hanenstein, 1912-1915). These uplifts and other disturbances can develop very fast, e.g. 60 mm in 10 days at Genevreuille. Such development can be at a rate of several millimeters or tens of millimeters over several years. The origin of disturbances affecting sulfate rocks (anhydrite and gypsum) is commonly attributed to a volume increase produced by the classical anhydrite hydration to gypsum. This reaction shows that: (i) in a closed system, there is only a volume decrease (i.e. settlement equal to 9%); (ii) in an open system, swelling can theoretically occur (61%). Studies carried out in the 1960s-1970s: in the laboratory by Sahores [56] ; at the time of the Belchen Tunnel construction (in Switzerland) and at the time of repairing older tunnels such as reported by Kastner [57], Krause [58], Grob [59, 60], Götz [61], Kaiser [62] ; and as part of the geologic survey of these formations by Anrieh [63], and Fabre and Dayre [64], called this interpretation into question and led to the following conclusions. (i) Swelling could never be observed in an undisturbed massive anhydrite devoid of microfractures as in the case of the Fréjus tunnel, hydraulic galleries (EDF) or anhydrite mine at Billingham, Yorkshire (UK). In this abandoned mine, the 600-700 km long network of rooms 5-6 m high and pillar galleries form an empty space of 15 x 106 m 3 in good repair. After being partially inundated (in 1974), no swelling or other disturbance could be observed. Only anhydrite powder laid and wetted for a haulage way showed local swelling in the form of invert 'protuberances'.
An Overview of Tunnel, Underground Excavation and Boreholes Collapse Mechanisms (ii) To be hydrated, anhydrite must have a large surface of exchange with water. This occurs when anhydrite is thinly bedded or features microfractures. (iii) Maximum swelling pressure is about 2 MPa. (iv) High swelling pressures produced by the heavy disturbances observed in sulfate rock masses originate from the swelling of clay minerals (corrensite) as shown by Henke [65] for the Wagenburg tunnel. (v) The disturbances observed around works in gypsum rock are due to gypsum solubility. Any change in the drainage pattern, e.g. lowering of the water table as a result of works, produces void formation making the rock mass loose and prone to water invasion. As a conclusion, for an excavation project concerning the aforementioned materials, one would be tempted to say that there is no risk at all in the presence of pure and massive anhydrite. The excavation process must be cautiously applied as rock must not be fractured. Problems appear to be due to minerals (especially clay minerals) other than anhydrite. Gypsum and anhydrite proportions should be determined when both minerals are present (at a depth less than 1000 m). The dilatable clay mineral content should be checked. Attention should be drawn to the clay mineral distribution in rock masses as well as to rock fracturing and permeability. Due to the lack of quantitative results, microfracturing and permeability limits able to cause swelling are not yet available. Case studies could be helpful. 14.2.4.2 Case of some argillaceous materials Among the various types of clay minerals, the smectite family includes minerals that can dilate in the presence of water, in the case of pressure or stress differences or chemical variations affecting the medium in which they are enclosed. Swelling - whether a volume or pressure increase - can be quite significant and exceed the anhydrite to gypsum change. Fabre and Robert [66] measured pressures of 8 MPa in a heaving marl in a tunnel through the Jura Mountains. Nevertheless, there are great differences between the results obtained from various laboratories and even between those obtained from one laboratory for tests carried out on close, seemingly identical samples. Without starting a complex physicochemical analysis of these minerals, remember they consist of phyllosilicates. The spaces between the various folia can be modified by water invasion or withdrawal. Due to their constituents, smectites can be easily hydrated or dehydrated. Water invasion or withdrawal may result in a substantial interfolium space modification leading to a significant macroscopic dilation or shrinkage. Smectite depositional conditions can be extremely variable, thus accounting for the variety of their enclosing rocks. It is necessary to distinguish the clay mineral that is part of the rock mass (fault infill, crushed zone) from the clay mineral distributed over a series rock, especially in the case of shales included in thick overburden (of oil fields) to be crossed through. When shales occur only in a portion of the works as fault infills, pressure and stress variations can cause them to swell. As a result, there can be displacements in the rock mass. Following the Hesendal tunnel (Norway) being driven through a granitic basement, there was a rock slide between two 10-12 cm thick clay veins. A careful exploration of adjacent areas, even of those deemed to be of minor interest, a minéralogie analysis and early processing of data during construction may be effective ways to avoid such slides. When minerals are present and scattered over the rock mass or are the major rock constituents, bear in mind they can occur in a great variety of rocks: limestones, marls, molasses, granites and other deeply altered eruptive rocks, and argillites as shown by the analysis of disturbances recorded in motor, railroad and other tunnels in France. This also applies to thick shale and argillite overburdens to be crossed through by development wells drilled in gas and oil fields with diverse deviation angles and azimuths. This question is of considerable economic significance. Two different approaches to this problem are taken in underground civil engineering and petroleum engineering. This is because galleries and tunnels are driven and maintained at atmospheric pressure before lining while in oil wells the drilling fluid exerting pressure on the walls of the borehole may permeate through these walls, thus producing physical and chemical effects on the formation shales. The civil engineering approach is essentially aimed at integrating the effects of stress variations induced by excavation. The swelling of hydrated material is determined in the laboratory through an oedometer test variation [67]. The following relation is obtained ôh P — = /clog— h *P0
393
394
Support
where ôh/h is the relative variation of sample height, and P and P0 are the final and initial stresses to which the sample is subjected. Based on the Grob method the so-obtained swelling potential is assigned to each point of the rock mass subjected to stress variations. Addition of elementary displacements gives the final displacement. The ISRM Commission on swelling rocks (see the report by Einstein [68], coordinator) proposed a standard approach and three tests for determining swelling: (i) measuring axial swelling pressure using an oedometer, (ii) measuring axial and radial free swelling, and (iii) measuring axial swelling versus axial stress during a standard Huder-Amberg test. Various approaches [69-71] tried to make allowance for the three-dimensional aspect of the phenomenon, i.e. the effect of stress variation in a given direction on deformations in the other two directions. As a comment, one can wonder whether the approach is representative of real conditions, especially how can one be sure the dehydration-rehydration process with a new stress application in the laboratory reproduces the in situ behavior of a saturated material subjected to pressure and stress variations? Furthermore, note the deficiency of experimental laws and particularly the lack of a three-dimensional behavior law that would help simulate properly the swelling events and associated causes. The petroleum approach involved two essential points more directly concerned with application. The first point involves technological research on drilling fluids. Research focuses on how to obtain the best cake (solid particles deposited on borehole walls by drilling mud), i.e. as tight as possible so that the borehole pressure may efficiently support the walls of the borehole and consequently prevent fluid from permeating through the walls in order to avoid reservoir contamination by drilling fluid. The same purpose must now be achieved in the case of shales, to avoid or minimize chemical exchanges between shales and filtrate (a small quantity of fluid passing through the cake). Note that cake deposition in soft, little indurated or subcompacted shales is questioned. Cake occurrence is more probable in indurated shales (argillites). The second point involves research on the physical and chemical equilibrium of mud and filtrate with formation shales. The purpose is to have the aqueous phase of drilling fluid in equilibrium with shale interstitial water in order to prevent either shale swelling or dehydration as a result of interstitial or interfolium water migration. As an example, it was observed when drilling in the ultra-thick shales of the Gulf of Guinea and the North Sea that the drilling fluids having a lesser salt content than the water occurring in shales could hydrate the shales. Conversely, too high a salt content of drilling fluids produced shale dehydration and disintegration. It should be recommended to watch the condition of cuttings so as to be aware of the equilibrium state of the aqueous phase of mud with respect to shale water. Softened cuttings reveal a lack of salt while dry and brittle cuttings indicate salt excess [72]. This chapter does not aim to draw up a balance of the studies carried out in this field by oil companies. Refer to Darley [73], and to Chenevert's [74] and Steiger's [75] communications presented at the Pau Symposium (1989) for bibliographic information about the existing works on this subject. In short, shales of the smectite family can be encountered locally or scattered all over large clay areas. They can be a source of considerable problems in tunneling or oil-well drilling. Failures in such argillaceous rock masses can result from a combination of various factors: physicochemical factors, deformability, mechanical strength anisotropy, stresses applied in drilling, and unpredictable high pore pressures that are difficult to determine. The occurrence of an incipient failure of mechanical origin accentuates the potential causes and problems of instability. Appreciable progress still remains to be made to cause this field of rock mechanics to be more helpful in drilling.
14.2.4.3
Case of pyrites
Pyrite oxidation produces the swelling of numerous materials, especially marls, shales and argillites (Fourmaintraux, D., pers. comm., 1989) [76]. Whereas physical and chemical considerations have been thoroughly investigated, there are relatively few works to date on mechanical effects [77, 78]. This phenomenon can be observed in the cores taken from oil boreholes and trench walls. It is a source of troubles for tunneling [79].
An Overview of Tunnel, Underground Excavation and Boreholes Collapse Mechanisms In fact, even in small quantity (1-5%), pyrite has a framboidal shape. It features microscopic (1 μηι) pentagonal dodecahedra so that it has a very large specific surface area and high reactivity. It can also feature larger aggregates. Pyrites can oxidize into sulfates (as gypsum and others) due to pressure, temperature changes and solution physicochemistry (and, in some cases, bacterial presence). Such crystalline aggregates can lead to a complete disorganization of rock structure even if pyrite occurs in such small amounts as to be considered as an accessory mineral. The mechanisms that cause these in situ phenomena are not known. One does not know whether it can occur in the course of drilling or under what conditions. It is important to determine the real origin of such phenomena, otherwise clays could be blamed for events wholly or partially unrelated. Further investigation is necessary to identify this phenomenon and determine the subsequent failure modes that might originate from it. 14.2.5 Mechanisms of Thermal Origin Focusing on thermal effects in underground works is a relatively recent attitude. Analysis of thermal effects on solution-mined caverns started in the 1960s-1970s [80]. Upon heating the storage structure in which the heavy fuel oil had to be maintained liquid and pumpable, it was necessary to allow for the thermal effects on the gallery. In this connection, in situ heating tests were performed [81]. Nuclear waste storage projects were to have a considerable impact in that they initiated the analysis of thermal effects on gallery sidewalls, rock mass permeability and stored material rheologies [82]. Although thermal effects andfluidfloware governed by identical equations, the theoretical case of dry materials exclusive offluidswill be discussed in Section 14.2.5.1 for simplification purposes. The applications subsequent to this simplified approach and the results obtained for unprecedented mechanisms will be discussed below. Section 14.2.5.2 deals with the case of porous materials. It first recalls the main theoretical results relating to stress distributions and possible combined thermal and hydraulic effects. It ends with showing the perpectives these approaches have opened up for the so-far unexplained failure mechanisms. Applying these theories requires measuring parameters. Measurements have just started in the early 1990s. Ongoing studies of thermal effects in underground works are and will be of prime value in current and future applications of rock mechanics, especially: (i) Nuclear and other waste burial with no damage to rock mass, storage works (rooms and pillars) and access openings (shafts and galleries). (ii) Deep underground works: deep tunnels and mines (Alpine tunnel projects, South African, American, Indian mines, among others). (iii) Scientific purpose oriented wells (8000, 10000 and over 12000 m); 12065 m Kola well and 8080 m Sarty well in USSR; KTB project in Germany [83-85]. (iv) Oil wells undergo phenomena that can be prejudicial to their stability at depths as shallow as 1500 or 2000 m. For deep wells (deeper than 3000,5000 and 6000 m) subjected to highfluidpressures and very high temperatures (over 100 MPa and 180°C at 5000 m) the problem is increasingly difficult to solve. Understanding stress distributions and failure modes occurring in these underground structures is necessary to make appropriate decisions upon supports, reinforcements to be installed or internal pressure to apply. 14.2.5.1 Case of dry materials Upon heating or cooling (<5<9), the walls of a circular borehole are subjected to tangential and longitudinal compressive or traction stresses, without the radial stress being affected. Thermoelastic analysis gives the following known result δσθ =
Ε<χδΘ , 1— ν
δσχ =
ΕαδΘ
, 1— ν
δστ =
0
where δσ is expressed in MPa ( < 0 traction, > 0 compression), E is the Young's modulus of rock (in MPa), v is the Poisson coefficient, a is the linear dilatation coefficient (in °C _1), δθ is the temperature increase (or decrease) (in °C) for isotropic rocks.
395
396
Support
Note that the induced stress magnitudes are greater than expected and the relevant consequences. For rocks with average to high rigidity values (E = 40000 MPa, v = 0.25, a = 8 to 10 x 10" 6o C _1 ), 1 °C is equivalent to a stress increase or decrease by 0.4-0.5 MPa. More rigid rocks such as dolomites, limestones, and eruptive rocks may have greater dilatation coefficients and higher moduli so that the aforementioned equivalent value can be 1-1.2 MPa. The salt case (a « 4 0 x 10" 6o C" l ) is worth considering in spite of plastic behavior on walls. This phenomenon can have significant effects on failure modes: (i) Underground works (especially oil wells) may be cooled by circulating drilling fluids at 20, 30 or 50 °C in proximity to or behind the displacement front. The state of tangential and axial compressive stresses acting on these works is relieved by this cooling, thus giving rise to considerable tensile stresses of about 8-20 MPa. At the end of cooling (e.g. when mud circulation is stopped) stresses are restored by the action of subsequent heating to their initial state as imposed by geostatic conditions. For a rock (of medium strength) resisting both stress applications, the failure may be delayed. In the past, this failure mode was wrongly believed to be due to creeping; in fact, it is produced by a return to thermal equilibrium. This failure mode was identified and reported as early as 1987 [12]. Its analysis and assessment have been described in a report on an oil well in southwest France by Guenot [86, 87]. (ii) At depths of several thousand meters, the walls of the borehole are cooled by the drilling fluid flowing downward. The drilling fluid temperature increases upon passage through the bit and contact with the rock. When going upward through the annulus, the drilling fluid is in thermal equilibrium with rock at the so-called neutral point (Figure 15). Above this point, rock is heated by fluid and subjected to thermal stresses. Such stresses may produce delayed failures typical of the upper part of the open-hole that are not yet fully understood by oil operators. In the case of shales, the cause of these failures is attributed to creeping (they occur a long time after fluid front passage) while they are merely of thermal origin due to delayed stresses. Knowing the actuating mechanism helps find a proper cure (i.e. by adjusting the internal pressure increase or the drilling mud weight) for limiting the tangential compression increase liable to produce failure (assuming a Mode Al stress distribution). Wells have been drilled with no problem when drilling at 1500 and 2500 m depths. However, stability problems (failure in the upper open hole) at these depths occurred when the bit reached deeper formations at 3500 or 5000 m. Operators had to pull the drillstring to the level of 1500-2500 m. But, in the meantime, temperature increased in the lower portion of the borehole producing failures in the lower part of the open hole. As a result of these troubles, the drilling of the borehole had to be resumed over a large interval. This is a costly operation that can be avoided once the failure mechanism is identified. Based on a Mohr diagram, Figure 16 shows the effect of temperature increases and decreases. Note the foregoing essentially applies to maximum tangential stresses (Al) common in drilling. Mud temperature in annulus (°C) 50 70
500,
1
Drilling at 800 m
2 3 4
Drilling at 1300 m Drilling at 2000 m Following 10h circulation stopped at 2000 m
At the start of drilling, borehole walls are heated.
ft 1500h Neutral point
Then, below the neutral point, walls are cooled. The upper open hole is heated. The neutral point depth lowers as the borehole depth increases.
Rock mass
temperature
Figure 15 Temperature evolution during drilling (after Guenot [9])
An Overview of Tunnel, Underground Excavation and Boreholes Collapse Mechanisms
Criterion
Criterion
/
\\
-y^
Mud weight increasing-Mode AI Stabilizing
σ
σ
θ
θ
Cooling-Mode AI Stabilizing
Criterion
Mud weight increasing - Mode CI Destabilizing Figure 16
397
Criterion
Cooling - Mode CI Destabilizing
Temperature influence on Al and Cl stability
Results can be contrasting in the case of a different stress distribution (Cl). It is clear how important it is to determine the magnitude of in situ stresses. The aforementioned phenomena can be well observed in boreholes due to the significant temperature increases and decreases undergone by the walls of the borehole. Perhaps we should wonder whether otherwise-unexplained delayed failures affecting deep galleries or tunnels should not be studied from this viewpoint.
14.2.5.2
Case of porous materials
Thermoporoelastic effects were formulated by MacTigue [88] following the poroelasticity formulation by Rice and Cleary [37]. In 1988 and 1989, Coussy [89,90] developed a general theory of thermomechanics in saturated porous media based on thermodynamics equations (mass conservation, movement quantity, first and second principles). For a material having a linear saturated elastic behavior, he inferred the general equations of thermoporoelasticity in a saturated medium made up of one compressible solid and fluid. He showed that the parameters, below, associated with the law of thermoporoelastic behavior could be determined from tests on standard drained materials (varying fluid mass) and standard undrained materials (constant fluid mass): (i) elasticity moduli for drained (Eh) and undrained (Eu) cases; (ii) volumetric incompressibility moduli for drained (Kh) and undrained (Ku) cases, or, which amounts to the same thing, Poisson coefficients for drained (vb) and undrained (vu) cases. Normally the two shear modulus values should be equal and are checked to ensure they are similar; (iii) dilatation coefficients for drained (ab) and undrained cases (au); (iv) specific heat (C) of the whole material (fluid and solid are not differentiated). Using these coefficients, the classical Biot (a) and Skempton (B) coefficients can be calculated as well as the latent heat, L, or amount of heat required to maintain temperature constant in isothermal, isochore tests. Note that these parameters can be determined with no assumed differentiation between the fluid and solid phases. But, determination of Biot coefficient, a, from the relation a = 1-
398
Support
is based on the additional assumption that material is characterized by linear elasticity and matrix isotropy (with a volumetric modulus Ks\ which can be ignored in handling porous medium problems. It is more advisable to calculate parameter a (and 5 , if necessary) from drained and undrained elastic material constants and later to verify through a new calculation if the (unnecessary) hypothesis that the matrix is homogeneous is confirmed. Similarly, based on the MacTigue formula, the global macroscopic dilatation coefficients can be related with the coefficients of fluid dilatation, a fl , and solid dilatation, a s as follows au-ab =
Βφ0((χ{ι-(χ,)
where φ0 is initial porosity. This relation is based on a differentiation between fluid and solid, which is not necessary in the case under consideration. Coussy [89, 90] and Charlez [91] prefer to use the following equation au = a b (l - B) + B (1 - φ0) as + Βφ0(χη in which porosity may not be constant. Setting aside the relations derived from the behavior law, 17 unknowns: 6 stresses, 6 deformations, 3 displacements, 1 pressure, 1 temperature and consequently 17 equations are to be used for handling the whole problem. In addition to the classical equations of elasticity (6 of behavior, 3 of equilibrium, 6 of compatibility) there are two equations of hydraulic and thermal diffusion. They involve additional parameters: permeability K, fluid viscosity μ (or absolute one-phase permeability Κ/μ) and the thermal conductivity X of the solid. Without going into detail and just as an indication, the two equations can be written as follows K „ -ΨΡ
ÔP ôekk δΤ = a,—- + α2— + α 3 —
ΧψΤ
δΡ &> kk δΤ = bY — + b2-^ + b3— + b4Q grad T ot ot ot
μ
δί
δί
δί
Coefficients a( and b( are functions of above-mentioned parameters. They account for the various coupled mechanical, hydraulic, thermal effects. This theory has been applied to the case of boreholes and wells drilled in porous materials subjected to plane deformations as a result of pressure and temperature variations. As Coussy gives a detailed description of it [90], only the main results obtained are summarized below. They are concerned with the failure modes that can be recorded on the walls of a borehole. (i) In addition to a coupled hydraulic and mechanical action as mentioned in Section 14.2.1.3, there can be some coupled effects of thermal origin, e.g. effects of applied pressure upon fluid or solid temperature, and effects of applied temperature upon fluid or solid pressure. (Theoretically the two phases are not differentiated.) (ii) Practically, in the case of liquid fluids (the case of gas fields may not be identical) the thermodynamic effect pressure changes have on liquid temperature (sometimes called PVT effect) is negligible. (iii) For relatively permeable materials (permeability higher than 1 mD), the hydraulic pressure diffusion governed by k/μ is more rapid than the thermal diffusion governed by thermal conductivity, X. These two phenomena are not coupled. (iv) The results of the effects are quite different in the case of low permeability materials and/or viscous fluids (permeability not greater than one microdarcy, as in the case of indurated argillites). As an example, a sudden temperature increase can cause a fluid to dilate more than the solid framework, thus generating pore overpressure within the walls of a borehole. As a result, maximum concentration of delayed stresses occurs inside the walls of a borehole within a few percent of radius distance under common hard rock conditions. Therefore, the possibility of a delayed failure occurrence still exists. Such failure starting from within the walls of a borehole is not connected with creeping but results from the transient thermal and hydraulic conditions to which the borehole walls are subjected. Similarly, a sudden temperature decrease can produce a pore pressure reduction inside the walls of a borehole and, thereby, produce a drastic change in equilibrium conditions. One cannot help making a comparison with consolidation observed in soil mechanics formerly considered and referred to as 'primary creeping' though being quite different from a true creeping.
An Overview of Tunnel, Underground Excavation and Boreholes Collapse Mechanisms
399
Table 1 Characteristics of Vosges Sandstone
Eu EB
KU KB V
u
vB *s
a (Biot) B au (dilat.) aB a
solid
Pconf = 1 MPa
Pconf = 15MPa
25 300 M Pa 26100 26300 15000
31000 30600 20200 16800
theor. 0.22 0.24 0.18 0.20 40200 MPa 0.55 0.30
0.34 0.21 39 700MPa 0.67 0.66 13.8 x 10" 6 10.3 x 10" 6 10.9 x 1(T 6
The real occurrence of such failures, which is difficult to observe, cannot be determined as there are no case histories and very few laboratory test results available. However, their possible occurrence is relied on in observations and tests, which would not otherwise be understandable. Due to the scarcity of known values, the parameters mentioned in this paragraph can be assigned orders of magnitude. Elasticity moduli, Eh and Eu (under drained and undrained conditions) are generally close to each other. Eu can be greater than Eb by a few percent only. Volumetric moduli, Kh and Ku feature a 15-30% difference. The Poisson ratios inferred from these measured values can differ significantly or be very close. Therefore direct measurements are not recommended save for checking purposes. The Biot coefficient derived from measured values is about 0.7-0.8 for rocks with 20-30% porosity. It ranges from 0.5 to 0.7 for 10-20% porosity. Same variations occur for the Skempton coefficient. Dilatation coefficients for drained rocks are generally in the range of 8 to 12 x 1 0 " 6 o C _ 1 . For undrained rocks, values can be 10-30% higher (the MacTigue relation is not always valid). Rock thermal conductivity values are relatively little scattered, close to 2 W/m °C. As an example, Table 1 gives data measured in sandstones from the Vosges sandstones in France with 18.5% porosity, and confining pressures of 1 MPa and 15 MPa. It is useful to know in which range of values common rock characteristics can be associated. As an example, Charlez [91] gives the following relations for common rock characteristics: volumetric moduli from 3000 to 30000 MPa; Poisson coefficient from 0.2 to 0.4; Biot coefficient from 0.1 to 1; dilatation 1 0 " 5 o C - 1 ; thermal conductivity from 1 to 2 W/m°C. th (hydraulic diffusivity characteristic time) ij (thermal diffusivity characteristic time)
thermal diffusivity hydraulic diffusivity
μ (cP) k (mD)
If this ratio is of the order of 1 or larger, then an increase of temperature AT (°C) will induce an increase of pore pressure ΔΡ (Pa) given by ΔΡ (Pa) - 400
M(cP) k (mD)
AT(°C)
As is apparent from the second relation, when the permeability is equal to 0.001 mD, a 1 °C rise in temperature produces a substantial pressure increase (here 0.4 MPa). Analysis of the in situ failures whose occurrence may be due to these phenomena, and analysis of abnormal wellbore instability caused by temperature decrease remain a must. 14.2.6
Instability Mechanisms Occurring in Solution-Mined Caverns
Caverns are mined by dissolution in salt rock with a view to producing salt in the form of brine and to storing liquid and hydrocarbon products (based on hydrocarbon tightness of salt). Solution mining can be performed from a single well by injecting fresh water through tubing and recovering the brine in the annulus between tubing and casing in the case of a direct solution-mining. A reverse procedure called reverse solution-mining is possible.
400
Support
The so-obtained caverns are pear- or flower pot-shaped. They can reach 60-80 m in diameter and 300-400 m in height (large enough to accommodate the Eiffel Tower). Volume is equal to or greater than 1 Mm 3 . Salt rock mass mining is characterized by the excavation of a set of such caverns hydraulically independent from one another but whose proximity is determined by mechanical interaction (internal pressure). Solution-mining can also be carried out from one or several wells referred to as injectors. These wells are used for fracturing the salt rock mass in the direction of one or several other wells referred to as producers. This is the track solution mining method using a single, dual or multiple track(s). In the case of a dual track method, caverns (100-200 m long, 100-300 m high, and some tens of meters broad) shaped as more or less elliptic cylinders are obtained. In fact, the shape of these caverns is ill-known. In the case of multiple track method, working is in the form of a solutionmining heading that compares favorably with a long wall mining. The depth of brine-producing caverns ranges from shallow (100-400 m) to deeper levels (1500-2000 m). Excavations are made in homogeneous salt rock mass (diapirs, layered salt mass) and in heterogeneous salt rock mass (layered salt mass including marls, anhydrites, dolomites and evaporites). Storage caverns are always individual units set together as a pattern. However, they should not be too far apart for an appropriate utilization of available space. They are generally located at deeper levels than brine-producing caverns and in more homogeneous rock mass. There exist various projects and achievements aimed at converting brine-producing caverns to storage rooms, especially in Louisiana, as part of the US Federal Storage Program (Bryand Mound, West Hackberry caverns; see Furiga and Smith [92]). Technical and statistical data on such caverns and related problems are available in the reports on salt Symposia, in the publications of the Solution Mining Research Institute, Thorns and Gehle [93] and Aubertin [94]. Only unexpected failures or behaviors occurring in such caverns are dealt with here. Due to the construction method used (solution-mining), such caverns are under pressurized conditions during the whole operating period. This internal pressure, variations in which have to be limited at times, has an actual supporting effect. As such, it is a true design and performance criterion. Sizing rules based on salt properties and internal pressure are complied with in the solution-mining industry [95]. See the general report on the ISMR Symposium of Zacatecas [96] for a summary of mechanical rock requirements to be met in the construction of such caverns. Such caverns can feature true rock ruptures with the occurrence of fracture lines as well as unexpected behaviors (as excess deformation, creeping). They are not rock failures in the literal sense of the word but events that may cause serious damage to the site or environment. For this reason, they are considered as a kind of 'failure'. As these phenomena may occur at the same time as a result of an ill-assessed geomechanical behavior, they will both be mentioned.
14.2.6.1
True rock failure mechanisms
The most dramatic failure is the roof collapse, whether purposely produced (salt mines in the Lorraine province, NE France) or accidental (Louisiana, Michigan, USA; Canada, etc.). It can be accompanied by surface sinkholes when occurring on top of caverns of karstic origin. Sinkholes can be of tremendous size: 60-100 m and even 200 m in diameter and several tens of meters in height. In NE France where such failures are due to a human decision, sinkholes are about 200 m in height and collapses occur in the case of 130 m diameter caverns and upon failure of a hard rock (such as Beaumont dolomite). The collapse process involves a gradual alteration of marls underlying the dolomite followed by a sudden dolomite failure under the load of the overburden. Shear strength at the edge of the vertical chimney on top of cavity is too low to support material column. The fact that failure is clearly noticeable in the case where cavern pressure is little different from hydrostatic suggests another cause of failure (as detailed in Section 14.2.3). In some cases, chimney ascents can be observed up to the surface with no bulking over large intervals (exceeding 100 m) for small size cavities. Analyses of specific cases should be made to support the view that a whole roof collapse can occur at once by the action of hydraulic pressure on the roof and be followed by an upward movement to the surface upon a sudden pressure change in the cavern (as in the case of karst emptying). The shapes of the so-called brecciapipes are indicative of such processes. Note that the specific design of storage caverns (geometry, internal pressure, depth) can prevent them from being affected by such failures.
An Overview of Tunnel, Underground Excavation and Boreholes Collapse Mechanisms
401
Wall failures are another mode of true failure leading to tubing failures and loss of solutionmining equipment (tubing and casing). Unfortunately, they are practically impossible to identify for the observed occurrences may be connected with the falling of insoluble materials. Sonar logging and inspection do not help determine their origin.
14.2.6.2
Abnormal, unexpected or hazardous behaviors
Storage caverns are subjected to two types of pressure variations. (i) For liquid hydrocarbon storage, production tubing is full of brine and cavern pressure is conditioned by the height of the column of products stored above the brine-hydrocarbon interface (the specific gravity of brine is 1.2 ; the specific gravity of hydrocarbon is about 0.85). (ii) For gas and compressed air storage (Huntorff, Germany) pressure variations must be as large as possible for operating. However, an upper pressure limit is necessary to avoid formation fracturing at shoe level. Admissible pressure limits (in M Pa) range from 0.16 to 0.18 H (H = depth in m) depending on the weight of overburden and applicable regulations. There should be lower pressure limits, too, to avoid salt creeping. On these lines, the specific behaviors of caverns will be discussed below. Shape measurements are not practical and it is difficult to follow their changes. Sonar logging and inspection are accurate to within 5%. Uncertainties in the count of stored products cumulate after several operations. As a result, the unknown range of error is an obstacle to the accurate determination of quantities. However, some conclusions can be drawn. Caverns for liquid storage are affected by convergences of a few percent. Dubois et al. [80, 95] analyzed the thermal behavior of caverns, especially salt creeping and brine heating. He showed a gradual return to equilibrium in these caverns characterized by viscoelastic and plastic phases of salt behavior depending on internal pressure. Such behavior is probably typical of thousands of liquid storage caverns. Unexpected behavior - not strictly failures speaking - could be observed in three cases. They are interesting in that they feature excess deformations. These cases, Eminence salt dome (USA), Kiehl (Germany) and Tersanne (France), are described below. (i) Examples of unexpected behavior The case of Kiehl (Germany) [96] involves a solution-mined cavern, between 1300 and 1500 m, whose original volume was 50000 m 3 , of which 20% was made up of insolubles. Following fluid displacements in the wells and sudden pressure changes, pumping troubles and abnormal free space reduction (7500 to 2500 m 3) were observed. Theses events could not be explained by the classical laws of flow or thermal action. Besides the unavailability of reliable data on these caverns, there was a high speed of convergence (100% per year after the first 23 h of pumping), and a high susceptibility to internal pressure and to the rate of pressure variation. Note also the roof collapse in this cavern. The Gaz de France mined caverns at Tersanne, France [97], between 1400 and 1550 m with a 90000 m 3 volume, were subjected to annual pressure variations of 22 to 8 MPa. Over 10 years (1970-1980) 25-30% volume losses were observed (60000 m 3 measured). There is an excess convergence due to an excess pressure differential affecting the in situ salt behavior. Interfacial measurements suggest the cavern bottom was probably more deformed than the upper portion, presumably because of depth (refer to Langer [98] and Berest [99, 100] for all questions about the rheological behavior of salt not discussed in this chapter). Eminence salt dome mined caverns, Mississippi, USA [101-103], between 1700 and 2000 m, had an initial diameter of 30 m. After several operating cycles or internal pressure variations (internal pressure decreased to 6 MPa), there were 40% volume losses, i.e. more than had been expected. Volume losses were subsequent to an upraise of the cavern bottom by 40-50 m. As is apparent from this case, convergence values are high as soon as the difference between geostatic and internal pressure exceeds 20 MPa and the convergence rate is proportionally greater than this difference. The last type of behavior examined is the one determined by salt creeping and brine heating; this is known as cavern fracturing. Practically, this is an important case as it results in the abandonment of all mined caverns and in the certainty that brine flow into the environment and surface aquifers will be avoided. It is very difficult to understand what occurs in such cases. Pressure build-ups and sensitivity to pressure variations are carefully and thoroughly analyzed for dual track mined caverns in order to determine abandonment conditions (e.g. salt mine caverns at Vauvert, Atochem, France). For information about the rheological behavior of salts refer to Chapter 6, Volume 3 of this publication (Langer, Fine, Charpentier) before giving attention to the following conclusions.
402
Support
(i) Prevailing temperature effect: from 20 to 100 °C,flowvelocity increases by one to two orders of magnitude. (ii) Zero influence of the mean stress is confirmed by fine observations. (iii) The probability for no viscoplastic limit is still questionable. (iv) There exists an exponential function (with exponent, n = 3, 4 or 5 for some salts) relating the deformation rate to the second invariant, J2, of the stress tensor. (v) The laws of flow vary according to whether salt is associated with hydrocarbons or brine. In the presence of brine, classical flow phenomena are followed by accelerated dissolution recrystallization [104]. (vi) Failures are very difficult to observe except when ending on the surface. Fracturing due to pressure build-up is of particular and practical concern in the abandonment of caverns. An in situ follow-up study is made to determine abandonment conditions. (vii) Diverse mechanisms are certainly involved. They will have to be individually analyzed before proposing behavior laws for salt (and other evaporites). Attempts are now being made in this field [105]. 14.2.7 Mechanisms of Dynamic Origin The only mechanisms discussed are ones subjected to dynamic stresses. Specialists and authors group various phenomena observed in different geological and mining environments into this category. Though their differentiation is sometimes difficult and arbitrary, one can distinguish three kinds of phenomena: (i) Phenomena generated by local stress adjustments in close proximity to excavation. These are referred to as 'strain bursts' and 'seam bursts' (equivalent to the French expressions 'coups de terrain' and 'coups de couche', respectively; as observed in some coal mines in France [106]). The 'heading burst' (equivalent to the French 'coup de front'), reported to have occurred in the Mont Blanc Tunnel in the Alps by Panet [6], may be added to these phenomena. There is no visible or observed relationship with liquid or gas fluid pressure, in this case. The case of very high water pressures in faults or fractures, leading to gallery invasion is dealt with in Section 14.2.3. Such phenomena are closely investigated and well documented in reports by Ortlepp [107, 108] and Stacey [13, 109] on the South African gold mines, and Hagan [110] on counteracting means and processes. These sometimes sudden and damaging stress releases referred to as strain bursts produce violent shocks in the vicinity of workings and are generally accompanied by just some limited seismic events. Ortlepp and Moore [14] describe such a phenomenon in a South African coal mine (which is a rare event) as a result of a very sudden roof collapse. The morphological study of fracture (mirror zones) shows propagation velocities of 2000 m/s. Such phenomena have also been observed in Poland, USA, etc. (ii) Phenomena due to gas or fluid pressure releases, dealt with in Section 14.2.1.3. Gas can be enclosed in the pore spaces of the rock or adsorbed as is the case with coal (or hydrocarbons). In the latter case, gas can be released in large quantities (in the form of methane, carbon dioxide), thus giving rise to large-scale phenomena such as explosions and material movements (e.g. gas pockets in some salt mines of Louisiana in the USA, and instantaneous outbursts ('dégagements instantanés') in French galleries [111, 112]. The outburst of gas in the coal mine of Sunagawa in Japan reported by Sato and Fuji (FLMW) as well as the explosions described and interpreted by Sato and Itakura [113] fall within this category of phenomena. These authors indicate the significance of the 97 outbursts of gas that have occurred for 40 years in the Japanese mining districts. In their view, gas pressure is too low to generate the spalling-producing failure (pressure is much lower than the rock tensile strength). According to these authors, spalling is the result of the transient stresses produced by blasting (and spall thickness reportedly depends on the rock tensile strength). The author thinks that similar phenomena are possible in the course of oil well drilling or during gas production tests in which sudden pressure drops occur. Such failures were observed to occur in a gas well drilled in extremely hard dolomite in southwest France. This well (depth 5000 m) was filled with breakouts over an interval of several hundreds of meters as a result of a sudden pressure drop of 20 M Pa in the course of a production test. (iii) Phenomena that are difficult to identify: mining produces stress concentrations that can be theoretically determined in proximity to heading. But the stress modifications produced at greater distance on faults, fractures, and bedding are more difficult to assess. The so-modified state of
An Overview of Tunnel, Underground Excavation and Boreholes Collapse Mechanisms stresses can be released extremely suddenly to produce quakes of significant magnitude in the neighboring area (up to a 100 m distance). There are dramatic examples of such phenomena in Canada, USA and South Africa where they are called 'rockbursts' [108] and analyzed as seismic events [114]. Ryder [115] describes an approach to these mechanisms based on the excess shear stress (ESS) release concept. By action of these mechanisms, the gallery or opening is subjected to an extremely violent dynamic stress producing, at times, a real explosion of the side wall rock [116,117]. This phenomenon alone is dealt with in this paragraph, though the various types of phenomena are not always easy to discriminate. Improvements in the analysis of such phenomena could be perceived at two Symposia on Rockbursts and Seismicity Mines (in Johannesburg, 1984 and Minnesota, 1988) and at an international workshop (Fred Leighton Memorial Workshop) preliminary to the ISMR Congress held in Montreal in 1987. Improving measurement, assessment and quantification were the first steps toward a better analysis. In this connection, the following techniques were concerned: (i) Sound wave and microseismic monitoring techniques (papers by Hardy, Young, Hanson, Quesnel, Mottahed in the Fred Leighton Memorial Workshop (FLMW), and Mathieu [118]). As the mechanisms involved are ill-known, measurements cannot yet be reliably used for predictions. (ii) Spectral analysis techniques developed in seismology for near field movements. They help determine the location of the event, the type of movements, size (radius) of seismic sources, energy of deformation associated with the seismic moment, stress decrease at the time of event, even if these quantities only indicate the order of magnitude. The following results can be quoted: (i) Discrimination of the various source mechanisms (case of South African gold mines reported by MacGarr (FLMW) characterized by fracture shearing at Klerkdoop and mass shearing at Carltonville based on the significant differences in seismic and mechanical parameters between the two types. Rudajev et al [119] showed the importance of the dilatant component in focal mechanisms corresponding to local events. A classification is proposed by Hasegawa et al (FLMW) of several mechanisms induced in Canadian Mines by excavations and associated P- and S-waves. (ii) Relationship between water movements and microseismic activity (Heick and Flak (FLMW)) in a potash mine in Germany and hydrogeologic activity areas. (iii) Relationship with seismic activity and exploitation [120-123]. This is a point that is essential to us. It illustrates the mechanism given as an example in Section 14.3 below. Deformations - even those of a very small extent - produced by works, occur along fractures, faults or bedding joints, thus increasing the permeability thereof. The energy associated with these small-scale movements is of little significance, as are the resulting sound emission and seismic activity. The insufficiently water-supplied fractures behave as if drained and are subjected to important shearing. When fractures (occasionally large portions of them) are filled with water and under pressure, there can be a sudden shear stress release resulting in a seismic event of the size of a natural earthquake. Finite element modeling has shown that a similar phenomenon could be obtained for a reverse fault in the absence of water due to a pressure decrease in the course of operating associated with high horizontal stresses [124]. Note that some of the collapses described in Section 14.3 have generated seismic events recorded by remote stations. Regarding the failures visible in the South African galleries we visited in the Western Deep Level Mines (Gold Mines in South Africa), where the failures can be said to have produced the almost entire destruction of the gallery. Identifying the basic cause of such a complete destruction is difficult since, normally, the emitted wave lengths are over 200 m and a 4-5 m gallery should not be subjected to such damaging movement differentials. An extended survey of this topic has been made by Dowding [125]. The cause can be attributed either to the excavation size (length or breadth) if it is equal to the wave length or to local reflection effects, difficult to investigate in detail. Installing gallery supports can hardly counteract the damaging and sometimes deadly effects of such mechanisms. 14.3 FAILURE MECHANISMS RELATED TO ROCK MASS STRUCTURE: EFFECT OF WORKS UPON THE STATE OF STRESSES APPLIED The previous sections deal with the failure mechanisms produced by stresses directly acting upon the rock mass during the construction or operation of a structure. Analyzing case histories of
403
404
Support
collapsing caverns and borehole instability showed up 'indirect' mechanisms, that is, cases where driving proper caused no matrix or joint instability but deformations generating a new loading process and/or leading to failures. In most cases, the resulting failures were delayed, sudden, unexpected and affected an initially stable structure. Such failures and indirect (or relaying) mechanisms appear to be of particular concern in thefieldof underground works engineering: they can cause death; they are costly; and until their combined causes are identified, they cannot be understood. Taking the geological structure of the rock mass into account is essential here, since discontinuities and heterogeneities play a major role. The approach to the rock mass deformation system must be taken as a whole, and based on a synthetic observation. An example of such observation (made during several years in the neighborhood of a coal mine) is given by Gramberg [126]. Two types of mechanisms are presented below. They are selected from case history analyses relating to mines and quarries on the one hand, and to oil well drilling and development, on the other hand. 14.3.1 Sudden Room and Pillar Collapses in Mines and Quarries Frequently, in underground mines and quarries, orthogonal galleries are driven using the socalled 'room and pillar' method. Pillar sizing is generally conservative with a safety factor of 3-5 relative to the weight to be borne (tributary area theory). Many such sites are stable for a long time, at least until the extraction of the remaining pillars (failure controlled as in the case of some iron mines in the Lorraine province in northeast France). Surface site stability may fail when the selected extraction rate is high enough to cause a controlled failure of pillars (as in some potash mines of New Mexico, USA). But when mining varied materials - coal, iron ores, natural calcareous building stone, cement stone, chalk, argillite (excavation for underground storage) - a sudden sometimes unpredictable collapse may affect simultaneously the whole site, generally from the underground area to the surface. It can be extremely hazardous and harmful to miners (such collapses are sometimes called 'spontaneous'). Tincelin [127] furnished the geometrical characteristics of the Lorraine iron mines where more than six collapses occurred over 20 years. Comparable failures in limestone, chalk and shale excavating sites have led the author [42] to propose the following failure mechanism (see Figure 17): (i) The immediate roof of such excavations is a good quality slab, whose bolting is not necessary. Miners are careful to leave it undamaged. (ii) The breadth and length of the working panel range from 100 to 150 m in the cases mentioned. The roof slab is subjected to sagging without any damage. (iii) Shear stresses act upon the bedding joints of the roof, thus increasing considerably their horizontal permeability. (iv) The gradual lowering of the deep roof pore pressure is changed, and full hydrostatic pore pressure is exerted on the immediate roof. Consequently, upon this pressure increase, excavation stability depends on pressure drainage through the immediate roof. (v) In the case of a low-permeability roof practically left undamaged by miners, the hydrostatic pressure, normal or in excess, produced by the rising of the water table will apply to the whole roof, causing its collapse and failure (Figure 18). The above-mentioned publication [42] shows that all reported observations relating to the analyzed case histories match, thus giving support to the mechanism proposed. Note that a series of in situ permeability measurements made above an underground limestone quarry - worked in Central Rock, Kentucky, USA as part of the US Federal Storage Program - gave a horizontal permeability about 100 times higher than vertical permeability, which gives additional support to the proposed solution. Note also that the primary roof deformation mechanism can have multiple origins: elastic behavior of pillars under stress (iron mines, resumption of invert excavation in a shale quarry in the USA), plastic deformation of thinned pillars (salt mine), punched footwalls, and softened material (water saturated byfloodingas in chalk quarries). The deformation itself may be harmless, but the subsequent effects acting more or less upon the state of stresses, and especially on hydraulic pressure may cause deformation to evolve into failure - here the change is extremely sudden. Examinations of old documents and observations relating to abandoned quarries in the Paris district made it possible not only to confirm the deformation mechanism but also to draw further conclusions, which are described below.
405 An Overview of Tunnel, Underground Excavation and Boreholes Collapse Mechanisms
(a)
IOm I p~~
Collapsed beds
(I)
Collapsed rooms
T
Coal (South Africa)
Gypsum (France)
30 m
90 m
3m
4m
(b)
20-30m
Roof of new collapsed gallery Initial gallery 4m Cross section
> 1000 m
Figure 17 Vertical or horizontal collapse with no bulking (a) collapsing rooms and (b) roof collapse in a quarry (Delporte quarry, North France)
(i) Reports on collapses that occurred 100 years ago distinguish two kinds of collapses: (a) gradual, even if they occur within a few hours, days, months or even years; and (b) sudden, with no premonitory sign. The latter can affect very small areas. Though it cannot be concluded with certainty as to the type of deformation given the pieces of information available, these failures may well suggest secondary mechanisms of hydraulic origin. The author thinks sinkholes (observed in gypsum quarries) going upward from a 70-90 m depth (on top of 4-5 m high gallery intersections) to the surface may well have caused such deformations. Obviously, allowance should be made for an appropriate support able to control them. Cameron-Clarke [128] made similar observations above room-and-pillar coal mines at shallow depths (30 m). He was surprised at observing roof collapses in a 3 m high gallery, producing a sinkhole (or 'pot-hole') at the surface. This occurrence is sudden, produced by a total collapse of the whole chimney, without any bulking. In the author's view, a hydraulic mechanism cannot be excluded (Figure 17a). (ii) The examination of an abandoned quarry in Middle Lutetian limestone showed a typical roof collapsing. Isolated galleries had been cut through this initially 4 m high quarry (worked thickness). These galleries remained stable during the whole working period. Following site abandonment, it was observed that the immediate roof had collapsed over 4-4.5 m without bulking (Figure 17b), thus causing the gallery to translate toward the top of this interval. It is possible to visit this gallery at present. The former sidewall top is now at the bottom, close to the footwall of the new gallery profile. But the most original feature of this immediate roof collapsing is that it extends over several hundreds of meters. The immediate roof has fallen as a whole, without any bulking (as far as we know). This can be explained only by the action of fluid pressure simultaneously exerted on the whole structure. The collapse of immediate roof slabs consisting of marls and limestones (of Upper Lutetian) caused the bedding joints to befilledwith water without drainage being possible. As soon as water invaded the open holding-joint above the roof slab, the resulting pressure destabilized the roof. Although data are lacking, it can be said that the deformation did not extend upward probably because of the high resisting upper slabs. It is clear that a limited drainage system would have acted more efficiently than heavy supports in that case. (iii) In this connection, the cylindrical chimneys observed above collapsed natural karstic caves (breccia pipes) may well have been caused by too rapid a karst emptying.
406
Support (α)
τ
τ
.—»•"Hydrostatic pressure
τ
100-200 m «5 m
T
_Kv¥ fc>.^- "^37
l&.fl
ML-
1„
(b)
T
"——
rei
T
IF
— , — π
y*
-frt —
ϋΐ__
Hydrostatic pressure
»
JH~~,~\
A
-
3Ξ
f/y
(c)
Figure 18 Roof collapsing (a) low mining rate: gradual displacement of the water table towards the immediate roof, (b) high mining rate: parting of roof slab, permeability increase and total hydrostatic pressure on the immediate roof and (c) the whole roof collapses
14.3.2 Instability due to Fault or Other Discontinuity Shearing Analyzing why an oil well under completion was 'lost' enabled us to evidence an indirect failure mechanism of primary concern. In fact, this indirect mechanism may have a direct effect in the course of drilling, thus becoming a direct instability mechanism (not yet identified as such among the borehole stability problems, in spite of its simplicity). Sometimes one has to start by studying complex cases in order to understand common and simple conditions! Summarizing the study case [129] (Figure 19), oil wells are lined with telescopic pipes (casing) aimed at isolating water, high pressure, and little producing horizons. The casing set in the Pont d'As 5 well in southwest France had been cemented opposite to a fault at 2400 m, and then to a high pressure hydrocarbon-bearing horizon at 3500 m (horizon pressure 60% higher than hydrostatic). The first internal pressure decrease to atmospheric pressure posed no problem. The second one, a few months later, produced a sudden collapse of both casing and tubing at the fault level. The analysis of data showed that a poor cementing produced pressure build up coming from the 3500 m horizons within the 2400 m fault. Under the release of the effective normal stress, the fault no longer stabilized, sheared and ovalized the casing, thus reducing the casing strength by 40%. Upon the second internal pressure decrease, casing and tubing collapsed. This mechanism is interesting for it can occur in the course of drilling and give rise to borehole stability problems. For safety reasons, mud weight is generally of a slightly higher value than the assumed pressure of in situfluids.The drilling mud leaves a millimeter scale film of solids (cake) on the walls of a borehole. Cake prevents drilling fluid losses to the borehole walls. When drilling in a fault zone or in any discontinuity (e.g. fracture or bedding joint) more permeable than the rock matrix, if no precautions are taken, the drilling mud will invade the fault and, thus, increase the initial fault pressure conditions. Pressure build-up will cause the fault to move on a millimeter or centimeter scale and thus to shear the casing. As shown by casing deformation measurements in
An Overview of Tunnel, Underground Excavation and Boreholes Collapse Mechanisms 09 0 9 86
10.12.86
20.12.86
06.03.87
Fault j 2 3 8 0 ml Air drilling |—2718 m Mud' drilling k>349lm 103^4501 3905 m
Casing lOV«inches
h
390lmqj;P7H 4 x 5 inches Packer
4480m-i§Zi_" Liner 6.V· inches 5027 m
Air drilling, 12 '/»inches through flysch Significant caving noted while crossing fault (2380-2440m) At 2718 m, stuck in hole Mud drilling started 1.15-1.25 weight Mud weight increased to 1.65
||
11 Liner 5 inches J—5250 m
10 V« inches casting installed to 3901 m
8Ve inches liner installed 3394-4542 m
Well completed, filled with fluid
Considerable gas observed in mud displaced during cementation operation
6 Va inches liner installed 4480-5027 m
Movements on the pressurized fault, initiated casing ovalization and distortion
After 48 h, gas flow observed between 13 Vfc and 10 V» inches casings
Entered reservoir with 5 V i inches bit, total loss of circulation at 5140 m Hole filled with water level controlled at 2700m depth 10 V4 inches casing partially empty
Difficulties lead to recompletion activities and dramatic lowering of fluid level to 2700 m Sudden final collapse of 10 V« inches casing in the highly pressurized fault zone
Gas inflow along cemented 10 V4 inches casing, gas entered fault under high pressure Good cement job above fault ( 2 3 0 0 - 2 4 0 0 m) trapped gas in fault region, raising pressures
Figure 19 Casing and tubing collapse due to fault movements
another adjacent well, the movements observed in a geological kink are in the order of one inch (inch = 25.4 mm) over a 300 m interval in a 12 1/4 inches well. It is worth noting that the same failure mechanism may occur in an uncased well. A point is made of recalling this since very few people in charge of drilling operations are aware of this possible occurrence that may have far-reaching consequences in practice [130]. In the course of drilling, borehole misalignment may cause the drillstring to bump against the socreated 'steps'. Upon POOH (pulling out of hole), from bottom to surface, the stabilizers (which have the same size as the bit and hole) will be stopped by these 'steps'. The origin of some wellbore stability problems faced by drillers can be found there: drillstring stopped upon RIH (running in hole) because of a hole misalignment, overpull upon POOH. Such problems are encountered locally. It may happen that the drillstring gets stuck and has to be left in the hole. In some such cases the well is abandoned. The most serious problem relates to the borehole support. Naturally and by experience one is inclined to increase the drilling mud weight to stabilize the borehole. This is fully justified in the case of type Al and Bl mechanisms described in Section 14.2.1.1. But this solution is hazardous in the case of fault shearing: if the fault behaves as in 'undrained conditions', the pore pressure increase reduces the normal effective stress on the fault, thus resulting in a shear displacement. Curiously, some highly experienced drillers know they would better go ahead under certain circumstances with the same drilling parameters (mud weight). It is a wise attitude that is fully justified here. Surprisingly, such a simple far-reaching mechanism has long been ignored. This illustrates how the knowledge and identification of mechanisms may lead to different and even opposing measurements in the determination of supports to be installed or the pressure prescribed for stability purposes. 14.4 CONCLUSIONS ABOUT THE EXCAVATION PROJECT, SUPPORT SIZING/ INTERNAL PRESSURE SELECTION AND FUTURE INVESTIGATIONS Supporting and reinforcing underground works as tunnels, power plants, storage and other galleries is a matter of considerable significance from the standpoint of safety and economy. The problem is of the s^me order in the oil industry where the drilling of deep boreholes from heavy
407
408
Support
offshore platforms may be surprisingly stopped due to a few centimeters or decimeters wellbore instability. A break of one day in drilling (costing US$ 50000-100000 in 1990) is equivalent to many engineering months. Here, too, improving supporting pressure and casing size determination pays off. As a matter of fact, ultra-deep drilling programs will now depend on the ability to control hole stability problems. To determine excavation stability, design the relevant supports and reinforcements and/or evaluate the internal pressure required, it is necessary to understand the mechanisms that may originate failures. This holds for ribs, bolts, concrete, internal pressure selection or borehole casing. Significant results have already been obtained in thisfield,due among others, to the effort made for discriminating and identifying failure mechanisms, before analysis and processing. There are two main categories of direct mechanisms affecting both large-size excavations (e.g. galleries and tunnels) and small-size excavations (e.g. deep oil wells stability). These direct mechanisms are: (i) rock matrix failure and deformation mechanisms, and (ii) discontinuity, joint, fracture-related mechanisms. As far as rock matrix is concerned: (i) Various types of failures can occur on the periphery of a circular excavation: shear, extension, traction failures, depending on the wall stresses. The resulting observations may help determine the state of in situ stresses. (ii) The origin of differences between extension and shear failures or transition from one mode to another must be understood and approached. (iii) Shear failures can occur upon internal pressure increases preceding fracturing. Pressure increases have a stabilizing effect only under certain in situ stress conditions. (iv) It generally appears from observations that failures can be initiated within sidewalls and later evolve into extension or shear failures. (v) Theoretical approaches (using pressure-dependent elastic moduli or bifurcation-theory) allow for this possibility. (vi) Original types of failures can be encountered in saturated porous materials subjected to anisotropic stresses as a result of a poroelastic effect. There, too, stress concentration is inside the wall. (vii) Thermal effects can be significant, especially in oil wells. They produce delayed failures, whose cause is wrongly attributed to creeping. Their influence in deep mining and tunneling should be kept in mind. (viii) Physicochemical effects on failure mechanisms are not yet well understood, determined and processed, perhaps because of several interacting mechanisms. (ix) Little information is available on failures occurring in anisotropic materials. Systematic observations might improve knowledge of materials featuring transverse isotropy by giving an idea of the ranges of prevailing anisotropic parameters. (x) Designers should be cautious about recurrent alteration phenomena, swelling or soluble materials. The shale problem is still unsolved since the in situ behavior conditions are difficult to reproduce in the laboratory. (xi) Designers should be cautious about hydraulic disturbances, inadvertently, unduly and casually caused in proximity to works. As far as joints are concerned: (i) Significant theoretical improvements have been made in the determination of rock mass or fractured zone stability. New models can be useful tools in excavation design, orientation and geometry. At the construction level, remember that there are still numerous unknowns: the state of stresses, scatter of joint properties leading to questionable and even delusive results. (ii) Close attention must be drawn to variations in hydraulic properties and joint pressure induced by otherwise troubleless deformations. Pore pressure increases may produce shearing movements and borehole misalignments. (iii) Such changes originate the so-called 'indirect' mechanisms. As seen above, initial and later stresses pose no problems excepting permeability variations. Too high a hydraulic load in the close vicinity of excavation may produce a sudden excavation wall failure. Limited drainage is more efficient than heavy supports in such cases. While theoretical models are valuable tools that enable us to predict unprecedented failure modes, improvements have been made in case analyses leading to the identification of mechanisms whose occurrence had not been previously contemplated. Gathering data on failure cases is an essential preliminary to progress. The ISRM Commission on Failure Mechanisms around Underground Openings proposes to edit data on failure cases [131]. Computerization is under way [132].
An Overview of Tunnel, Underground Excavation and Boreholes Collapse Mechanisms To end the discussion on such a delicate and promising subject of rock mechanics, let us repeat how important experience exchanges between industries are. Mines provide certainly the best observations. However, note that deep wells and underground storage structures undergo new mechanisms. Only well-documented observations will enable us to decide upon the supports to be used when discussing the tools proposed by theoreticians. ACKNOWLEDGEMENTS The author thanks the whole Failure Analysis Working Group of the French Committee on Rock Mechanics, particularly J. P. Piguet, for their advice. Thanks go also to A. Robert, J. Hy-Billiot, N. Kessler, P. Colin, P. Charlez, P. Berest, A. Guenot, F. Santarelli, D. Fourmaintraux, J. L. Maire, J. de Lautrec and to all members of the South African ISMR team for their opinions and advice in their fields of competence. 14.5 REFERENCES 1. Gough D. I. and Bell J. S. Stress orientation from borehole wall fracture with examples from Colorado, east Texas, and northern Canada. Can. J. Earth Sei. 19, 1358-1370 (1982). 2. Maury V. Observations, researches and recent results about failure mechanisms around single openings. Report of the Rock Failure Mecahnism around Underground Openings ISRM Comm. In Proc. 6th Int. Soc. Rock Mech. Congr., Montreal (Edited by G. Herget and S. Vangpaisal), vol. 2, pp. 1119-1128. Balkema, Rotterdam (French and English versions available on request) (1987). 3. Fairhurst C. General report: Deformation, yield, rupture and stability of excavations at depth in rock. In Proc. Int. Soc. Rock Mech. Congr. {Rock at Great Depth), Pau (Edited by V. Maury and D. Fourmaintraux), vol. 3, pp. 1103-1114 (1989). 4. Kirsch G. Die Theorie du Elastizität und die Bedürfnisse der festigkeitslehre, Veit. Ver. Deut. Ing. 42 (28), 797-807 (1898). 5. Brown E. T., Bray J. W., Ladanyi B. and Hoek E. Ground response curve for rock tunnels. J. Geotech. Eng. Div. Am. Soc. Civ. Eng. 109, 15-39 (1983). 6. Panet M. Quelques problèmes de mécanique des roches posés par le Tunnel du Mont-Blanc. Bull. Liaison Lab. Routiers Ponts Chaussées 42, 115-145 (1969). 7. Hoek E. and Brown E. T. Underground Excavations in Rock, p. 527. Institute of Mining and Metallurgy, London (1980). 8. Risnes R., Bratli R. K. and Horsrud P. Sand arching - a case study. Paper EUR 310, p. 11 (1981). 9. Guenot A. Contraintes et ruptures autour des forages pétroliers. In Proc. 6th Congr. Int. Soc. Rock Mech., Montreal (Edited by G. Herget and S. Vangpaisal), vol. 1, pp. 109-118. Balkema, Rotterdam (1987). 10. King L. V. On the limiting strength of rocks under conditions of stress existing in the earth's interior. J. Geol. 20,19-138 (1912). 11. Jaeger J. C. and Cook N. G. W. Fundamentals of Rock Mechanics, 3rd edn. Chapman and Hall, London (1979). 12. Maury V. and Sauzay J.-M. Borehole instability: Case history, Rock mechanics approach, and Results. Paper 16051 presented at Soc. Pet. Eng./Int. Assoc. Drilling Contractors, Conf. New Orleans, pp. 11-24 (1987). 13. Stacey T. R. and De Jongh C. L. Stress fracturing around a deep-level bored tunnel. J. S. Afr. Inst. Min. Metall. 78(5), 124-133 (1977). 14. Ortlepp W. D. and Moore M. A. Underground observation of high propagation rate, extension fractures. In Proc. 6th Congr. Int. Soc. Rock Mech., Montreal (Edited by G. Herget and S. Vangpaisal), vol. 2, pp. 1173-1180. Balkema, Rotterdam (1987). 15. Périe P. -J. Laboratory investigation of rock fracture around boreholes. PhD thesis, University of California, Berkeley (1990). 16. Périe P.-J. and Goodman R. E. Evidence of new failure patterns in a thick-walled experiment. In Proc. 12th Energysource Tech. Conf. Exhibition/Am. Soc. Mech. Eng. PD 22, pp. 23-27. ASME, New York (1989). 17. Mastin L. Development of boreholes breakouts in sandstone. MS thesis, Stanford University (1984). 18. Zoback M. D., Moos D, Mastin L. G. and Anderson R. N. Wellbore breakouts and in situ stress. J. Geophys. Res. 90, 5523-5530 (1985). 19. Baumgartner J., Carvalho J. and McLennan J. Fracturing deviated boreholes: An experimental laboratory approach. In Proc. Int. Soc. Rock Mech. Congr. (Rock at Great Depth), Pau (Edited by V. Maury and D. Fourmaintraux), vol. 2, pp. 929-937. (1989). 20. Plumb R. A. Fracture patterns associated with incipient wellbore breakouts. In Proc. Int. Soc. Rock Mech. Congr. (Rock at Great Depth), Pau (Edited by V. Maury and D. Fourmaintraux), vol. 2, pp. 761-768. (1989). 21. Santarelli F. J., Brown E. T. and Maury V. Analysis of borehole stresses using pressure dependent, linear elasticity. Int. J. Rock. Mech. Min. Sei. & Geomech. Abstr. 23, 445-449 (1986). 22. Santarelli F. J. Theoretical and experimental investigation of the stability of the axisymmetric wellbore. PhD thesis, University of London (1987). 23. Martin C. D. Failure observations and in situ stress domain at the Underground Research Laboratory. In Proc. Int. Soc. Rock Mech. Congr. (Rock at Great Depth), Pau (Edited by V. Maury and D. Fourmaintraux), vol. 2, pp. 719-727. (1989). 24. Vardoulakis J., Sulem J. and Guenot A. Borehole instability as bifurcation phenomena. Int. J. Rock Mech. Min. Sei. 25, 159-170(1988).
409
410 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64.
Support Papamichos E. and Vardoulakis I. The coupling effect of surface instabilities and surface parallel Griffith cracks in rock. In Proc. Int. Soc. Rock Mech. Congr. (Rock at Great Depth), Pau (Edited by V. Maury and D. Fourmaintraux), vol. 1, pp. 481-487 (1989). Detournay E. and St. John C. M. Design chart for a deep circular tunnel under non-uniform loading. Rock. Mech. Rock Eng. 21, 119-137(1988). Van Dyck V. Ovalisation en champs anisotropes. Application à la détermination des contraintes en parois de puits. Report Elf-Aquitaine DPRO/DREA (1985). Maloney S. and Kaiser P. K. Results of borehole breakout simulation tests. In Proc. Int. Soc. Rock Mech. Congr. (Rock at Great Depth), Pau (Edited by V. Maury and D. Fourmaintraux), vol. 2, pp. 745-753. (1989). Hiramatsu Y. and Oka Y. Determination of the stress in rock unaffected by boreholes and drifts from measured strains or deformations. Int. J. Rock. Mech. Min. Sei. 5, 337-353 (1968). Fairhurst C. Methods of determining in situ rock stress at great depths. Technical Report 1-68, Missouri River Div. Corps of Eng. (1968). Bradley W. B. Failure of inclined boreholes. J. Energy Resour. Technol. 101(4), 232-239; Oil & Gas J. 92-102 (1979). Guenot A. Borehole breakouts and stress fields. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 26, 185-195 (1989). Amadei B. Rock anisotropy and the theory of stress measurement. Lecture Notes in Engineering 2 (Edited by C. A. Brebbia and S. A. Orszag) p. 478. Springer (1983). Amadei B. Underground excavations in anisotropic rock masses. In Proc. Int. Soc. Rock Mech. Symp. Design and Performance of Underground Excavations, Cambridge (Edited by E. T. Brown and J. A. Hudson), pp. 27-37. British Geotechnical Society, London (1984). Detournay E. and Cheng A. H.-D. Poroelastic response of a borehole in a non-hydrostatic stress field. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 25, 171-182 (1988). Detournay E., Cheng A. H.-D., Roegiers J.-C. and McLennan J. D. Poroelastic considerations in in situ stress determination by hydraulic fracturing. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 26(6), 507-515 (1989). Rice J. R. and Cleary M.. P. Some basic stress diffusion solutions for fluid-saturated elastic porous media with compressible constituents. J. Geophys. Res. 14, 227-241 (1976). Carter J. and Booker J. R. Elastic consolidation around a deep circular tunnel. Int. J. Solids Struct. 18, 1059-1074 (1982). Guenot A. General report: Instability problems at great depth drilling boreholes and wells. In Proc. Int. Soc. Rock Mech. Congr. (Rock at Great Depth), Pau (Edited by V. Maury and D. Fourmaintraux), vol. 3, pp. 1199-1219 (1989). Obert L. and Duvall W. Rock Mechanics and the Design of Structures. Wiley, New York (1967). Poulos H. G. and Davis E. H. Elastic Solutions for Soil and Rock Mechanics, pp. 229-249. Wiley, New York (1974). Maury V. Effondrements spontanés. Synthèse d'observations et possibilité de mécanisme initiateur par mise en charge hydraulique. Rev. Ind. Miner 61(10), 511-522 (1979). Bazant Z. P. L'instabilité en milieu continu et la résistance en compression. Bull. Rilem 35, 99-102 (1967). Maury V. Stabilité, contraintes et exploitation de la mine de May-Sur-Orne comme stockage souterrain Rydrocartury. In Proc. 3rd Int. Soc. Rock Mech. Congr. vol. II.B, pp. 1294-1301. (1974). Gale W. J. and Blackwood R. L. Stress distribution and rock failure around coal mine roadways. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 24, 165-173 (1987). Timoshenko S. R. and Goodier J. N. Theory of Elasticity, 3rd edn. International Student Edition, McGraw-Hill, Maidenhead, UK (1982). Tang D. H. Y. and Peng S. S. Reinforcement analysis and design of mechanical roof bolting systems in horizontally bedded mine roofs. Int. J. Min. Geol. Eng. 3(1), 1-25 (1985). Ennour S., Piguet J. P., Baroudi H., Asof M. and Fraisse H. Nouvelles méthodes de modélisation des massifs rocheux: utilisation pour l'étude de la stabilité d'excavations minières. 3ème Colloque Franco-Polonais, Wroclaw, In Studia Geotechnica Mechanica, vol X (314), pp. 13-25. (Edited by Z. Gegowicz) (1990). Warburton P. M. Vector stability analysis of an arbitrary polyhedral rock block with any number of free forces. Int. J. Rock. Mech. Min. Sei. & Geomech. Abstr. 18, 415-427 (1981). Heliot D. Generating a blocky rock mass. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 25, 127-138 (1988). Goodman R. E. and Shi G. H. Block Theory and Its Application to Rock Engineering, p. 338 Prentice-Hall, Englewood Cliff, NJ (1985). Lin D. and Fairhurst C. Static analysis of the stability of three dimensional blocky system around excavations in rock. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 25, 139-149 (1988). Baroudi H., Piguet J. P., Chambon C. and Asof M. Utilization of the block generator 'Resoblok' to complex geologic conditions in an open pit mine. In Proc. Int. Conf. Mech. Jointed and Faulted Rock, Vienne (1990). Cundall P. A. and Hart R. D. Numerical modeling of discontinua. Keynote address, 1st US Conf. Discrete Element Method (1989). Halwany S., Janod A. and Lajeat J. P. Le percement du Tunnel d'Awali de ΓAménagement du Litani au Liban. Science et Industrie, Paris (1965). Sahores J. Contribution à l'étude des phénomènes mécaniques accompagnant l'hydratation de l'anhydrite. Thèse, University Toulouse, Rev. matériaux de construction Pub. Tec. No. 126 (1962). Kästner Static des Tunnel und Stollenbaues auf der Grundlage Geomechanischer. Springer, Heidelberg (1971). Krause H. Sulphate rocks in Baden-Wurtemberg and their importance to civil engineering. Bull. Int. Assoc. Eng. Geol. 13, 45-49 (1976). Grob H. Schwelldruck im Belchentunnel. In Proc. Congr. Int. Soc. Rock Mech. Symp. Lucerne (Edited by H. Grob and K. Kovari). pp. 99-119. Balkema, Rotterdam (1972). Grob H. Swelling and heave in swiss tunnels. Bull. Int. Assoc. Eng. Geol. 13, 55-60 (1976). Götz Auswirkungen des Blahverholtens von Gesteinen auf die sohlhe Bungen im Tunnel Bau. Veröff. Bod. Feld., Univ. Karlsruhe N° 76 (1978). Kaiser P. K. Behaviour of anhydrite after addition of water. Bull. Int. Assoc. Eng. Geol. 13, 68 (1976). Anrieh H. Zur Frage der Versgipsung in den Sulphatlagern desmittleren Muschelkalks und Gipskeuper in Sudwestdeutschland. Neues Jb. Geol. und Paläont. 106, pp. 293-337. Stuttgart Schweizerbart (1958). Fabre D. and Dayre Propriétés géotechniques de gypses et anhydrites du Trias des Alpes de Savoie. Bull. Int. Assoc. Eng. Geol. 25, 91-98 (1982).
An Overview of Tunnel, Underground Excavation and Boreholes Collapse Mechanisms
411
65. Henke K. F. Magnitude and rate of heave in tunnels in calcium sulphate bearing rocks. Bull. Int. Assoc. Eng. Geol. 13, 61-64 (1976). 66. Fabre D. and Robert A. Rapport sur le gonflement. Groupe de travail sur la rupture. CFMR, BRGM, Orléans La Source (1987). 67. Huder J. Amberg G Quellung im Merge Opalinuston und Anhydrit. Schweizerische Bauzeitung 43, 975-980 (1970). 68. Einstein H. (Editor) Suggested methods for laboratory testing of argillaceous swelling rocks. ISRM Commission on swelling rock. Int. J. Rock Mech. Min. Sei. 26, 415-426 (1989). 69. Einstein H., Bischoff N. and Hofmann, E. Verhalten von stollensohlen im quellendem Mergel. In Proc. Int. Soc. Rock Mech. Symp. Underground openings. Lucern (Edited by H. Grob and K. Kovari), pp. 296-319. Balkema, Rotterdam (1972). 70. Einstein H. and Bischoff N. Dimensionnement des tunnels en roche gonflante. Tunnels et Ouvrages Souterrains 15, 109-119(1976). 71. Wittke W. Fundamentals for the design and construction of tunnels in swelling rock. In Proc. 4th Int. Congr. Rock Mech., Montreux, pp. 719-729. Balkema, Rotterdam (1979). 72. Geraut R. Le Forage des Argiles. Internal note EP/S/PRO/FIP 87.273.17 (Available on request to Elf Aquitaine) (1987). 73. Darley H. C. H. and Gray G. R. Composition and Properties of Drilling and Completion Fluids, 5th edn. Gulf Publishing, Houston (1988). 74. Chenevert M. E. Diffusion of water and ions into shales. In Proc. Int. Soc. Rock Mech. Congr. (Rock at Great Depth), Pau (Edited by V. Maury and D. Fourmaintraux), vol. 3, pp. 1177-1184 (1989). 75. Steiger R. P. and Leung P. K. Predictions of wellbore stability in shale formations at great depth. In Proc. Int. Soc. Rock Mech. Congr. {Rock at Great Depth), Pau (Edited by V. Maury and D. Fourmaintraux), vol. 3, pp. 1209-1218 (1989). 76. Fourmaintraux D. Projet Ontario-Hydro Wesleyville Generating Station. Rapport du Laboratoire Central des Ponts et Chaussées (1976). 77. Nixon P. J. Floor heave in buildings due to the use of pyritic shales as fill material. Chem. & Ind. 5, 160-164 (1978). 78. Prokopovich N. K. Engineering impact of oxidation of pyrite. Bull. Int. Assoc. Eng. Geol. 24, 140-144 (1987). 79. Deletie R., Hagenauer B., Quenee B. and Robinet A. Eclatements du revêtement bétonné d'une galerie alpine liés au gonflement des formations encaissantes. Mémoire Société Géologique de France. Nouvelle Série, pp. 181-187. Soc. Geol., Paris (1990). 80. Dubois D. Verification du régime élastique d'une cavité de stockage dans le sel â partir des mesures de température et de dilatation apparente des liquides stockés. Rev. Ind. Min. Ca. du Com. Francais de Méc. des Roches 4, 65-72 (1972). 81. Maury V. An example of underground storage in soft rocks: the chalk. Rockstore Conf. Stockholm (1977). 82. Come. B. The CEC bench mark INTERCLAY Results of pilot phase (January-June 1989) about the boom clay at Mol (B). Commission Europ. Com., EUR 12791, Brussels (1990). 83. Kessels W. Observation and interpretation of time-dependent behaviour of boreholes stability in the continental deep drilling pilot borehole. In Proc. Int. Soc. Rock Mech. Congr. (Rock at Great Depth), Pau (Edited by V. Maury and D. Fourmaintraux), vol. 3, pp. 1479-1486 (1989). 84. Natau O., Borm G. and Röckel T. Influence of lithology and geological structure on the stability of the KTB pilot hole. In Proc. Int. Soc. Rock Mech. Congr. (Rock at Great Depth), Pau (Edited by V. Maury and D. Fourmaintraux), vol. 3, pp. 1487-1490 (1989). 85. Mastin L., Heinemann B. B. and Fuchs K. Preliminary results of breakout analysis below 500 meters depth in the KTB pilot hole. In Proc. Int. Soc. Rock Mech. Congr. (Rock at Great Depth), Pau (Edited by V. Maury and D. Fourmaintraux), vol. 3, pp. 1491-1499 (1989). 86. Guenot A. and Maury V. Stabilité des forages profonds. In La Thermomécanique des Roches (Edited by P. Berest and Ph. Weber). Manuel et Méthodes, pp. 292-304. Bureau de Recherches Géologiques et Minieres, Orleans, France (1988). 87. Guenot A. and Santarelli F. Influence of mud temperature on deep borehole behaviour. In Proc. Int. Soc. Rock Mech. Congr. (Rock at Great Depth), Pau (Edited by V. Maury and D. Fourmaintraux), vol. 2, pp. 809-817 (1989). 88. MacTigue D. F. Thermoelastic response of fluid-saturated porous rock. J. Geophys. Res. 91, 9533-9542 (1986). 89. Coussy O. A general theory of thermo-poro-elasto-plasticity for saturated porous media. Transport in Porous Media 4, 281-293 (1989). 90. Coussy O. Mécanique des Milieux Poreux. Technip, Paris (1991). 91. Charlez Ph. Rock Mechanics - Theoretical Fundamentals vol. 1. Technip, Paris (1991). 92. Furiga R. D. and Smith R. E. The development of the United States Strategic Petroleum Reserve. Annales des Mines 47-62 (1983). 93. Gehle R. M. and Thorns R. L. Analysis of a cavern near theflankof a salt dome. In Proc. Symp. Salt, vol. 1, pp. 545-548 (1984). 94. Aubertin M., Gill D. E. and Ladanyi B. Le comportement rhéologique du sel: revue bibliographique. Rapport EPM/RT 87/31, Ecole Polytechnique de Montréal (1987). 95. Dubois D. and Maury V. Underground storage of hydrocarbons at Manosque. In Proc. 4th Symp. Salt, France, vol. 2, pp. 313-325. The Northern Ohio Geological Society, Cleveland OH (1974). 96. Kuhne G., Rohr W. U. and Sasse W. Kiel gas storage facility, the first city gas cavern in Germany. C. R. 12th Congres Mondial du Gaz, Nice NIGU, A28-73, p. 26 (1973). 97. Boucly P. and Legreneur J. Hydrocarbon storage in cavities leached out of salt formations. In Proc. Int. Symp. Subsurface Space Rockstore, Stockholm, vol. 1, pp. 251-259. Pergamon, Oxford (1980). 98. Langer M. Rheological behaviour of rock masses In Proc. 4th Int. Congr. Rock Mech., Montreux, vol. 3, pp. 29-62. Balkema, Rotterdam (1979). 99. Berest P. Problèmes soulevés par l'abandon des cavités de dissolution profondes dans le sel gemme. Journée d'Etudes sur la Geotechnique Appliquée aux Mines et Carrières (1987). 100. Berest P. Problèmes de mécanique associés au stockage souterrain. These de Doctorat de l'ENS des Mines de Paris (1989). 101. Serata S. Rock mechanics problems of solution cavities used for storage of gaseous and solid matters. Soc. Min. Eng. AIME, Preprint 73-AM-79 (1973). 102. Allen K. Eminence natural gas storage in salt comes of age. Trans. Soc. Min. Eng. AIME, 250, 276-279 (1971).
412 103. 104. 105. 106. 107. 108. 109. 110. 111. 112. 113. 114. 115. 116. 117. 118. 119. 120. 121. 122. 123. 124. 125. 126. 127. 128. 129. 130. 131. 132. 133. 134. 135.
Support Baar C. A. Applied Salt-Rock Mechanics, I. Developments in Geotechnical Engineering, 16-A, Elsevier, Amsterdam (1977). Spiers C , Urai J. L., Lister G. S., Boland J. N. and Zwart H. J. The influence of fluid-rock interaction on the rheology of salt rock. Final report, Com. Europ. Com., EUR 1039, Brussels (1986). Dusseault M. B. Saltrock behavior as an analogue to the behavior of great depth. In Proc. Int. Soc. Rock Mech. Congr. (Rock at Great Depth), Pau (Edited by V. Maury and D. Fourmaintraux), vol. 1, pp. 11-17 (1989). Josien J. P. and Revalor R. The fight against dynamic phenomena: French coal mines experiments. In Proc. 23rd Int. Conf. of Safety in Mines, pp. 531-540. Research Institute, Washington DC (1989). Ortlepp W. D. and Hagan T. O. Rockbursts: understanding and control - past, present and future. In Proc. 13th CMMI Cong. Aus. Inst. Min. Met., Victoria, vol. 3, pp. 77-84 (1986). Ortlepp W. D. The mechanism of a rockburst. In Proc. 19th U.S. Symp. Rock Mech. Lake Tahoe, NV, pp. 476-484. University of Nevada, Reno (1978). Stacey T. R. and Page C. H. Practical Handbook for Underground Rock Mechanics. Trans. Tec. Clausthal-Zellerfeld (1986). Hagan T. O. Mine design strategies to combat rockbursting at a deep South African gold mine. In Proc. 29th U.S. Symp. Rock Mech., (Edited by P. A. Cundall, R. L. Sterling and A. M. Starfield), pp. 249-261. Balkema, Rotterdam (1988). Josien J. P., Noirel J. F. and Piguet J. P. Gas outbursts and rock rupture in French coal mines. In Proc. Int. Soc. Rock Mech. Congr. (Rock at Great Depth), Pau (Edited by V. Maury and D. Fourmaintraux), vol. 3, pp. 1425-1431 (1989). Josien J. P. 4ème colloque sur la prévision et la prévention des coups de terrains et dégagements instantanés, rapport général thème 2. Ostrava, Czechoslovakia (in press). Sato K. and Itakura K. The occurrence and mechanism of outbursts in sandstone. In Proc. Int. Soc. Rock Mech. Congr. (Rock at Great Depth), Pau (Edited by V. Maury and D. Fourmaintraux), vol. 2, pp. 635-643 (1989). Spottiswood S. M. Source mechanisms of mine tremors at Blyvoruitzicht Gold Mine, rockbursts and seismicity in mines. S. Afr. Inst. Min. Metall., Johannesb., 29-38 (1984). Ryder J. A. Excess shear stress (ESS): An engineering criterion for assessing unstable slip and associated rockbursts hazards. In Proc. 6th Congr. Int. Soc. Rock Mech., Montreal (Edited by G. Herget and S. Vangpaisal), pp. 1211-1216. Balkema, Rotterdam (1987). Fernandez L. M. and Van der Heever P. K. Ground movements and damage accompanying a large seismie event in the Klerksdorp mining district. In Proc. 1st Int. Conf. on Rock Bursts and Seismicity in Mines, pp. 193-198. SAIMM, Johannesburg (1984). Revalor R., Josien J. P. and Magron A. Le comportement discontinu des roches à grandes profondeurs: le cas des coups de terrains miniers. In Proc. Int. Soc. Rock Mech. Congr. (Rock at Great Depth), Pau (Edited by V. Maury and D. Fourmaintraux), vol. 3, pp. 1431-1439 (1989). Mathieu E. Apport de la sismoacoustique pour la surveillance des chantiers miniers affectés de coups de terrains. Application au cas des tailles de l'unité d'exploitation de Provence. Thèse de Doctorat de l'I.N.P.L., Nancy (1989). Rudajev V., Teisseyre R., Kozack J. and Sileny J. Possible mechanism of rockbursts in coal mines. Pure Appl. Geophys. 124, 841-855 (1986). Revalor R., Josien J. P., Besson J. C. and Magron A. Seismic and seismoacoustic experiments applied to the prediction of rockburst in French coal mines. In Proc. 2nd Int. Symp. Rockburst and Seismicity in Mines, Minnesota, MN (Edited by C. Fairhurst), Balkema, Rotterdam (1988). Revalor R., De Chelette O. and Verstraete M. Detection of rock burst risk situation using seismoacoustic monitoring at the Provence collieries. Mining Science and Technology, 11-23 (1986). Ben Sliman K., Revalor R. and AI Heib M. Seismic monitoring: an help in the planning of Mine layouts in French coal mines subject to rockbursts. ISRM Int. Symp. Static and Dynamic, Swaziland (Edited by R. Brummer), Balkema, Rotterdam (1990). Gibowicz S. Seismicity in mines. Reprint from Pageoph. 129 (3/4), Birkhaeuser, Basel (1989). Revalor R. La lutte contre les phénomènes dynamiques à charbonnages de France. Journée d'Etude sur la Géotechnique Appliquée aux Industries Extractives, Paris (Available on request to Elf Aquitaine) (1988). Dowding C. H. Earthquake response of caverns: Empirical correlations and numerical modeling. In Proc. Rapid Excavation and Tunneling Conf, Soc. of Min. Eng. vol. 1, pp. 71-83. AIME, New York (1985). Roest J. P. and Gramberg J. The deterioration of a gallery along a coal panel: a case history. In Proc. Int. Soc. Rock Mech. Congr. (Rock at Great Depth), Pau (Edited by V. Maury and D. Fourmaintraux), vol. 2, pp. 703-711 (1989). Tincelin E. and Sinou M. Effondrements brutaux et généralisés - Coups de toit. Rev. Ind. Min. 44(4), 239-262 (1962). Cameron-Clarke I. S. The distribution and nature of surface features related to shallow undermining on the East Rand. In Sangorm Symp. Effects of Underground Mining on Surface, pp. 33-38. South African Group of ISRM (1986). Maury V. and Sauzay J. M. Rupture de puits par glissement sur faille: cas vécu, mécanisme, remède, conséquences. In Proc. Int. Soc. Rock Mech. Congr. (Rock at Great Depth), Pau (Edited by V. Maury and D. Fourmaintraux), vol. 3, pp. 871-883 (1989). Heliot D., Etchecoppar A. and Cheung P. New developments in fracture characterization from logs. In Proc. Int. Soc. Rock Mech. Congr. (Rock at Great Depth), Pau (Edited by V. Maury and D. Fourmaintraux), vol. 3, pp. 1471-1486 (1989). Stacey T. R. ISRM commission in rock failure mechanisms in underground openings (RFMUO). Standard data collection form for describing failures in underground openings. Int. J. Rock Mech. Min. & Geomech. Abstr., 29(4), 421-431 (1992). Kaiser P. K. and Yauzi S. ISRM commission on rock failure mechanism in underground openings. Computerized rock mass failure data collection system. Int. J. Rock Mech. Min. & Geomech. Abstr., 29(4), 431-446 (1992). Bergman M. (ed.) Storage in excavated caverns. Rockstore 77, Proc. 1st Int. Symp., Pergamon, Oxford (1977). Bergman M. (ed.) Subsurface space. Rockstore 80, Proc. Int. Symp., Pergamon, Oxford (1980). Maury V. General report: la mécanique des routes: une discipline de base pour le stockage souterrain, In the role of rock mechanics in extractions for mining and civil works. Proceeding of ISRM Symposium, Zacatecas, Mexico (Edited by R. Sanchez Trejo and R. de la Latta), pp. 553-572. Sociedad Mexicana de Mecanica de Rocas, Mexico (1986).
15 Overview of Rock Anchorages STUART LITTLEJOHN University of Bradford, UK 15.1
INTRODUCTION
414
15.2
DEFINITIONS
414
15.3
SITE INVESTIGATION
415 415 416 416 417
15.3.1 General 15.3.2 Field Sampling and Testing 15.33 Laboratory Testing 15.3.4 Investigation During Construction 15.4
417
DESIGN
417 417 417 419 421 424 425 425 425 426 427 427
15.4.1 General 15.4.2 Overall Stability 15.4.2.1 Excavations 15.4.2.2 Uplift capacity 15.4.3 Rock/Grout Interface 15.4.4 Grout/Tendon Interface 15.4.5 Materials and Components 15.4.5.1 Cementitious grouts 15.4.5.2 Resinous grouts 15.4.5.3 Tendon 15.4.5.4 Anchor head 15.4.6 Safety Factors 15.5
428
CORROSION PROTECTION
428 429 429
15.5.1 General 15.5.2 Principles of Protection 15.5.3 Protective Systems 15.6
432
CONSTRUCTION
15.6.1 15.6.2 15.6.3 15.6.4 15.6.5 15.6.6
432 433 433 434 436 436
General Drilling Tendon Grouting Anchor Head Stressing
15.7 TESTING 15.7.1 General 15.7.2 On-site Acceptance Tests 15.7.3 Proof Load-Time Data 15.7.4 Apparent Free Tendon Length 15.7.5 Short-term Service Behavior 15.7.6 Monitoring Service Behavior
437 437 437 439 439 439 440
15.8 CASE EXAMPLES 15.8.1 General 15.8.2 Static Performance During Service 15.8.3 Dynamic Response During Service
441 441 443 446
15.9
448
15.10
FINAL REMARKS
449
REFERENCES
413
414
Support
15.1 INTRODUCTION The earliest reports of anchoring bars into rock to secure a roof date from 1918 in the Mir Mine of Upper Silesia in Poland [1], and by 1926 faces of an inclined shaft, in Chustenice shales in Czechoslovakia, were secured against caving by grouted bars installed in a fan pattern [2]. In the field of civil engineering the history of rock anchorages dates from 1934 when Coyne pioneered their use during the raising of the Cheurfas Dam in Algeria [3]. Since 1934 there has been a startling increase in the use of anchorages throughout the world, and millions of rock anchorages have been installed ranging from bolts of a few tonnes capacity in ground reinforcement to prestressed tendons of several thousand tonnes capacity to resist concentrated loads in bridges. Today, anchorages may be employed to solve problems involving direct tension, sliding, overturning, dynamic load and ground prestressing, and the range of applications includes retaining walls, dry docks, concrete dams and spillways, tall buildings, pile testing, bridges, slope stabilization, tunnels, shafts and underground caverns [4]. This wide range of applications involving a variety of engineering disciplines perhaps best explains why ground anchorage development and usage have been so dramatic over the past 50 years. If reliable performances are to be maintained, a technical understanding and appraisal of rock anchorage systems is required by the practising engineer. The purpose of this chapter is to provide guidance on site and ground investigation requirements, design methods, corrosion protection measures, construction techniques and testing procedures for grout-injected rock anchorages. 15.2 DEFINITIONS A rock anchorage is considered to be an installation that is capable of transmitting an applied tensile load to a load-bearing rock stratum. The installation consists basically of an anchor head, a free anchor length and afixedanchor (see Figure 1). The term 'anchor' is used exclusively to denote a component of the anchorage, e.g. anchor head and free anchor length.
Figure 1 Ground anchorage nomenclature
415
Overview of Rock Anchorages
The anchor head is the component of a rock anchorage that is capable of transmitting the tensile load from the tendon to the surface of the ground or structure requiring support. The fixed anchor length is the designed length of the anchorage over which the tensile load is capable of being transmitted to the surrounding rock. A tendon, which usually consists of steel bar, strand or wire either singly or in groups, is the part of an anchorage that is capable of transmitting the tensile load from the fixed anchor to the anchor head. A rock bolt is a specific form of rock anchorage where a steel bar is fixed in rock and tensioned after installation. If the bar is untensioned the term 'rock dowel' is used. 15.3 SITE INVESTIGATION 15.3.1 General The ground is one structural component of the anchorage system, and the importance of a good quality site investigation cannot be over-emphasized. Investigations are most satisfactorily undertaken in a number of stages, namely (i) initial desk and field study, (ii) main field and laboratory investigation, and (iii) investigation during construction. The work required at any stage will be dependent on the nature of the project. For example, the consideration of a rock-bolt system for a limited-depth rock cut might be based predominantly on a visual field study, whilst that for a major retaining wall could require that the main field and laboratory investigation stages be split into preliminary and detailed phases. The planning and general requirements of site investigations are well understood and the purpose of this section is to highlight particular features which are relevant to rock anchorages. The geometry of a rock anchorage and its mode of operation require, in particular, a detailed knowledge of ground conditions local to the fixed anchor. Whilst there may be adequate data to indicate both the feasibility and advantages of an anchorage system, it is usual to find that there is insufficient detailed information to permit its economic design or construction. In this regard the
( c ) Sliding on faults or
( a ) Sliding on bedding or foliation planes
( b ) Sliding on joints
( d ) Toppling
(e)
Wedge failure
Figure 2 Principal failure modes in rock cuts and slopes (reproduced from BS 8081:1989 with permission of British Standards Institution)
416
Support
data required for the safe design of temporary anchorages is often similar to that necessary for permanent works. The broad aim of the investigation should be to determine, by the most economic means, the nature of the block of ground that is influenced by, or influences, the installation and behavior of the anchorages. Since rock anchorages are installed at various inclinations, lateral variations in ground properties should be investigated as thoroughly as the vertical variations. For structures such as an anchored retaining wall it is recommended that the maximum centers of investigation locations, e.g. boreholes, pits, etc., should not exceed 20 m unless a well-known 'solid' geological formation is encountered. In addition, the plan dimensions of the site need to be carefully defined so as to include the probablefixedanchor zone. Too often in practice, particularly for deep excavations, anchorages are installed beyond the site perimeter where there is a dearth of ground data. By contrast, the assessment of the stability of rock faces in slopes and tunnels may demand a detailed geological investigation involving the location of almost every anchorage or group. Discontinuity frequency and orientation data together with joint continuity and roughness can be vital in determining the size and shape of the rock mass mobilized at failure, which is important in any analysis of overall stability (see Figure 2). Where rock does not outcrop, the necessary spacing of the investigation locations will depend on the knowledge of the geology of the formation. Where rock is encountered during the borehole investigation, the depth of penetration should not be less than 5 m to ensure bedrock and possibly a greater depth to prove rock of the quality necessary to give adequate rock/grout bond strengths. 15.3.2 Field Sampling and Testing Emphasis should be placed on obtaining maximum core recovery, particularly in thefixedanchor zone, which in most rocks generally implies cores of not less than 55 mm (NX size) and, in weaker rocks, of larger diameter. Determination of the ground water conditions on the site will be essential for the overall design of the project, particularly where excavations are proposed. For rock anchorages, water levels are important in any analysis of water loss in boreholes which may dictate the need for waterproofing. Chemical analysis of the water may also be important in determining the appropriate type of cement for the tendon-bonding grout, and the degree of corrosion protection required for the tendon. With reference to field tests in boreholes, the standard penetration test is of value in obtaining a relative measure of the in situ quality of weak rocks. Although experience is required in the interpretation of such tests, the values have also been correlated with rock/grout skin frictions for design purposes [5, 6]. In weak rocks with a high mass porosity, radial load tests via self-boring pressuremeters may provide useful information on the degree of lateral restraint that the rock mass can mobilize against the 'bursting' mode of failure of the fixed anchor. The radial stress/strain characteristics of weak rocks are not measured routinely, but such information, may help to classify, more accurately, in the future, rocks such as weaklyfissuredmudstones which produce on occasions very low skin frictions at failure. Where the investigation proves rock strata which may lead to grout loss, then permeability tests supplemented by fabric description may be required to quantify the problem and assess the need for pregrouting during anchorage construction (see also Section 15.6.4). 15.3.3 Laboratory Testing Index tests such as point load, Schmidt hammer and sonic velocity tests are cheap and easily undertaken. The results often correlate approximately with other parameters and provide indications of rock quality. With regard to strength and deformability tests, the shear strength of a rock mass is commonly estimated from a detailed study of joint geometry and roughness together with a knowledge of the material characteristics [7, 8]. Whilst this rock mass shear strength is important in any overall stability analysis, the design of thefixedanchor in rock is usually based on borehole information and is often dependent upon a confining stress system local to the fixed anchor. As a consequence, the influence of joint systems may be less marked and the results of tests on intact rock cores can be useful, if carefully interpreted, e.g. unconfined compressive strength [9] and unconfined tensile strength [10]. In current anchorage practice, initial estimates of ultimate rock/grout bond (skin
Overview of Rock Anchorages
417
friction) are based on empirical correlations with rock core strength, in the absence offieldtest data or precedent practice (see also Section 15.4.3). Deformability is determined commonly by measuring stress/strain relationships in the unconfined compression test, bearing in mind that such tests provide a measure of the deformability characteristics of the rock material. The slake durability test apparatus [11] permits assessment of the susceptibility of weak rock to weathering or softening in the presence of water. The results of the test give an indication of the potential for loss of skin friction in thefixedanchor zone and the period for which a hole may be left open prior to installation of the anchorage tendon. Slake durability or swelling test results may also be used by the designer to ban the use of water flushing during the drilling of the fixed anchor. Since there are no correlations with field behavior, the laboratory test results are only indicative in nature. 15.3.4 Investigation During Construction During the course of anchorage installation, further evidence on the nature of the ground conditions is revealed by excavation or in the boreholes of the individual anchorages. This information should be recorded and appraised as an extension of the engineering geology work. No site investigation can explore ground conditions as frequently as the anchorage installation process can and any major variations in the ground conditions from those anticipated have to be recorded and their significance assessed. It should be emphasized, however, that production drilling associated with anchorage installation is not geared to investigate the ground in detail. All ground data obtained during anchorage drilling should be recorded and subjected to daily analysis. Such a system can act as an early warning device, should variation in strata levels or ground type require changes in design or installation method. Any adjacent activities on the site that may influence anchorage behavior should also be carefully monitored, recorded and their possible influence assessed at an early stage in the work. Such activities include ground water lowering, pilling, blasting and freezing.
15.4 DESIGN 15.4.1 General A grouted rock anchorage may fail in one or more of the following modes. (i) Failure within the rock mass. (ii) Failure of the rock/grout bond. (iii) Failure of the grout/tendon bond. (iv) Failure of the steel tendon or anchor head. For the design of a rock anchorage each mode of failure must be considered in order to ensure an adequate load factor of safety, having regard to magnitude and mode of loading, period of service and consequences of failure. 15.4.2 Overall Stability 15.4.2.1 Excavations The use of anchorages to ensure the stability of existing or new rock slopes and underground excavations in rock is well established in most classes of rock. Depending on specific circumstances, anchorages may be used as the sole means of providing stability or they may be used in conjunction with other forms of support such as sprayed concrete, steel ribs or concrete-retaining structures. The selection of anchorages in a given situation is often a matter of experience and judgement but comparison with previous practice, together with considerations of classification schemes and empirical guidelines can provide useful assistance [8, 12, 13]. Anchorages should be selected to suit the conditions encountered (see Table 1) and typical empirical design recommendations are summarized in Table 2 [14]. For underground excavations, the design approach generally adopted is to nominate an anchorage system prior to the start of excavation and subsequently to modify the level of reinforcement on the basis of observed conditions. The level of reinforcement provided varies widely,
Support
418
Table 1 Use and Rock Conditions for Reinforcement (reproduced from BS8081:1989 with permission of British Standards Institution) Reinforcement
Use and rock conditions
Rock dowels
Use: (a) to ensure stability of areas very near to the excavated surface; (b) to reinforce rock to be removed at a later stage; (c) for general support when installed very close to an advancing face, when tension is developed after installation; (d) for prereinforcement prior to an excavation. Rock: suitable for all types. In weak rocks sufficient length should be allowed to develop the tensile strength of the dowel. Use: for general support in all types of underground opening. Rock: (a) bolts with mechanical fixed anchors are suitable for use in hard rocks only; (b) bolts with grouted fixed anchors may be used in all rock types. In soft rocks, or where clay infilling has a tendency to line the drillholes, there may be insufficient anchorage capacity available for resin-grouted fixed anchors. Maximum support pressure*: 300 kNm" 2 .
Rock bolts
Rock anchorages
Use: for reinforcement of large openings which require high support pressures and long length of reinforcement. Generally used in combination with bolts, dowels or sprayed concrete. Rock: suitable for all rock types but care should be exercised in rocks with a combination of low RQDb, heavily jointed or crushed rock, smooth slickensided or filled joints, high water inflows, high in-situ stresses, swelling or squeezing rock (Barton et al. [12]). In weak rocks, anchorages appropriate to soils may be required. Minimum support pressure8: 200 kNm" 2 . Maximum support pressure: 600 kNm" 2 .
"Support pressure = support pressure available at time of installation. b RQD is the Rock Quality Designation. Table 2 Typical Empirical Design Recommendations (after US Army Corps of Engineers [14]) Parameter Minimum length and maximum spacing Minimum length
Maximum spacing
Minimum spacing Minimum average confining pressure Minimum average confining pressure at yield point of elementsc
Empirical rule Greatest of: (a) 2 x bolt spacing (b) 3 x thickness of critical and potentially unstable rock blocks8 (c) For elements above the springline: spans < 6 m: 0.5 x span spans between 18 and 30 m: 0.25 x span spans between 6 and 18 m: interpolate between 3 and 4.5 m (d) For elements below the springline: height < 18 m: as (c) above height > 18 m: 0.2 x height Least of: (a) 0.5 x bolt length (b) 1.5 x width of critical and potentially unstable rock blocks8 (c) 2.0 mb 0.9 to 1.2 m Greatest of: (a) Above springline: either pressure = vertical rock load of 0.2 x opening width or 40 kNm" 2 (b) Below springline: either pressure = vertical rock load of 0.1 x opening height or 40 kNm" 2 (c) At intersections: 2 x confining pressure determined aboved
8 Where joint spacing is close and span relatively large, the superposition of two reinforcement patterns may be appropriate (e.g. long heavy elements on wide centers to support the span, and shorter, lighter bolts on closer centres to stabilize the surface against ravelling). b Greater spacing than 2.0 m makes attachment of surface support elements (e.g. mesh or chain link mesh) difficult. c Assuming the elements behave in a ductile manner. d This reinforcement should be installed from the first opening excavated prior to forming the intersection. Stress concentrations are generally higher at intersections and rock blocks are free to move toward both openings.
Overview of Rock Anchorages ( a ) General support pattern
419
(b) Beam building, generally in laminated rock
Φ
Q 111
It
Φ
il)
( c ) Prevention of buckling failure of slab or rock block columns
Figure 3 Principal failure modes in underground excavations (after CIRIA [15])
depending on opening geometry, ground conditions and ultimate use. Typical modes of failure are indicated in Figure 3, together with an indication of the function of ground anchorages in maintaining stability [15]. Whilst initial assessment of anchorage support requirements may be made using empirical methods or classification schemes [8, 12, 16, 17], the detailed design of the anchorage system in an underground excavation should take careful account of the following elements. (i) Current practice and past experience. (ii) Observed behavior of excavated opening. (iii) Reinforcement of structurally controlled zones or blocks. (iv) Reinforcement of overstressed zones. (v) Anchorage characteristics (size, capacity, orientation, spacing, length, type, etc.). (vi) Three-dimensional geometry of the opening. (vii) Excavation sequence. (viii) Timing of installation of anchorages. (ix) Durability requirements. (x) Integration with other means of support, e.g. sprayed concrete. (xi) Quality control. Monitoring of rock conditions as well as movement is essential. This provides a comprehensive check of excavation behavior, a comparison of observed and predicted movements and a check on design assumptions. Although both tensioned (active) and untensioned (passive) anchorages are in common use, in general, it is recommended by the author that anchorages should be tensioned as soon as possible after installation. Tensioned anchorages strengthen the rock mass that forms a slope or the surrounds to an excavation, by increasing the shear resistance along discontinuities. This prevents the detachment of loose blocks and enhances the interlocking nature of the rock mass. A fully bonded anchorage, which is that normally used, provides restraint along the full free length, thus minimizing the dilation of joints. A decoupled free length is generally only used in cases where anchorages are to be restressed during service or where a substantial amount of movement is anticipated which may overstress the anchorage. 15.4.2.2 Uplift capacity For vertical or downward inclined anchorages subjected to external loads, the individual anchorages must be installed at a depth sufficient to resist safely the applied working load without failure occurring in the rock mass. Calculations on uplift capacity are based on simple cone or wedge
420
Support
mechanisms (see Figure 4). There is little experimental or practical evidence to substantiate the methods, but bearing in mind the traditional dearth of detailed information on the rock mass, the tensile or shear strength of the rock is seldom exploited and the conservative calculation is based on the effective weight of rock only. Field experience with vertical anchorages in rock [19] indicates that general failure (see Figure 4) with accompanying surface heave does not occur for slenderness ratios (h/D) in excess of 15, where h is the depth to the top of the fixed anchor and D is the diameter of the fixed anchor. For slender anchorages (h/D > 15) the failure mechanism in the ground tends to be local to the fixed anchor zone. Where groups of closely spaced anchorages have their fixed anchor zones located in the same rock horizon and the rock mass is horizontally bedded, the likelihood of laminar failure should be investigated. To avoid laminar failure, it may be necessary to incline the fixed anchors further apart in plan or stagger alternate fixed anchors at different depths in order to reduce the intensity of stress on any plane. In upward inclined anchorages where the resistance to withdrawal is primarily dependent on the mechanical properties of the rock mass, such properties should only be used when test results are ( a ) Geometry of cone l
1
y//$^y^À
fW/s&V/JW, \
/
h
\
^\Λ· \ P
60°or 90ö
IN \t J
1
>f
I 1L A
60° or 90°
\ V Λ I
H·—D
( I ) Load transferred by bond
(2) Load transferred by end plate
60° is used when rock mass is soft, heavily fissured or weathered; 90° is used in all other rock conditions
( b ) Interaction of cones for overall stability analysis
Figure 4
Assumed mechanisms of failure for uplift capacity (after Littlejohn and Bruce [18])
Table 3
Depth of Anchorage for Overall Stability (after Hobst and Zajic [20])
Rock type
Formula for depth of cone One anchorage
Irregular submerged fissured rock
Line of anchorages
sfrw
'Sound' homogeneous rock Irregular fissured rock
Remarks
Sf 2 to 4 Apex angle = 90° (assumed)
2.83τ« \J \y7rtan 2 0y 3
/[ ——-— ) \l\(y-y„)nt&n2'J
V V7stan07
Â
(y - y w ) s
tan
Φ'
421
Overview of Rock Anchorages
available for the particular site being considered. This philosophy also applies to downward inclined anchorages where shear parameters are incorporated. Table 3 [20] provides formulae for calculating the required depth of vertical anchorages, where: τ is the shear strength of rock (kN m" 2 ); S{ is the factor of safety against failure (a value of 2 to 3 is customary in current practice); s is the spacing of anchorages (m); φ' is the effective angle of friction across fractures in rock mass (degrees); Tw is the working load on anchorage (kN); y is the unit weight of rock (kN m" 3 ); and yw is the unit weight of water (kN m~3). 15.4.3 Rock/Grout Interface For straight-shafted anchorages in rock, designs are based on the assumption of uniform bond distribution, and the pull-out capacity of the fixed anchor (T( in kN) is estimated from equation (1) T( = nDLTult
(1)
where rult = ultimate bond or skin friction at the rock/grout interface, L = length of fixed anchor and D = diameter of fixed anchor. This approach is used in many countries such as France, Italy, Switzerland, UK, Australia, Canada and USA, although it is just as common to use Tworking in place of rult where a safety factor has already been incorporated. Equation (1) is based on the following simple assumptions. (i) Transfer of the load from thefixedanchor to the rock occurs by a uniformly distributed stress acting over the whole of the perimeter of the fixed anchor. (ii) The diameters of the borehole and the fixed anchor are identical. (iii) Failure takes place by sliding at the rock/grout interface (smooth borehole) or by shearing adjacent to the rock/grout interface in weaker medium (rough borehole). (iv) There are no discontinuities or inherent weakness planes along which failure can be induced. (v) There is no local debonding at the grout/rock interface. Where shear strength tests are carried out on representative samples of the rock mass, the maximum average working bond stress at the rock/grout interface should not exceed the minimum shear strength divided by the relevant safety factor (normally not less than 2). This approach applies primarily to soft rocks where the uniaxial compressive strength (UCS) is less than 7 N mm - 2 , and in which the holes have been drilled using a rotary-percussive technique. In the absence of shear strength data or field pull-out tests, ultimate bond stress is often taken as one-tenth of the uniaxial compressive strength of massive rocks (100% core recovery) up to a maximum value rult of 4.2 N mm" 2 . As confirmation, iult = 4.3 Nmm" 2 is indicated for design in hard coarse-grained sandstone by Canadian research [21]. In some rocks, particularly granular weathered varieties with a relatively low φ value, the assumption that rult equals 10% UCS may lead to an artificially low estimate of shear strength (see Figure 5). In such cases, the assumption that Tult equals 20-35% UCS may be justified. Bond values for cement-grouted anchorages, which have been recommended for a wide range of igneous, metamorphic and sedimentary rocks, are presented in Table 4 [5]. Where included, the 0.5
0.4
CO O 0.3 3
O.I
0
10
Figure 5
20
30
40
50
60
Effect of φ on Tult:UCS ratio
70
Table 4 Rock/Grout Bond Values Which have been Recommended for Design (after Littlejohn and Bruce [18]) Rock type
Igneous Medium hard basalt Weathered granite Basalt Granite Serpentine Granite and basalt Metamorphic Manhattan schist Slate and hard shale Calcareous sediments Limestone Chalk - Grades I-III (JV = SPT in blows/0.3 m) Tertiary limestone Chalk limestone Soft limestone Dolomitic limestone Arenaceous sediments Hard coarse-grained sandstone Weathered sandstone Well-cemented mudstones Bunter sandstone Bunter sandstone (UCS>2.0Nmnr2) Hard fine sandstone Sandstone
Ultimate bond (N mm"2)
Factor of safety
5.73 1.50-2.50 3.86 4.83 1.55 1.72-3.10
3-4 2.8-3.2 3.1-3.5 2.6-3.5 1.5-2.5
India - Rao (1964) Japan - Suzuki et al. (1972) Britain - Wycliffe-Jones (1974) Britain - Wycliffe-Jones (1974) Britain - Wycliffe-Jones (1974) USA - PCI (1974)
0.70
2.80 0.83-1.38
4.0 1.5-2.5
USA - White (1973) USA - PCI (1974)
1.00 0.005N
2.83 0.22-1.07 0.01 N
0.83-0.97 0.86-1.00
2.76 2.76 1.03-1.52 1.38-2.07
2.8 2.0 (Temporary) 3.0-4.0 (Permanent) 2.9-3.3 2.8-3.2 1.5-2.5 1.5-2.5
Working bond (N mm"2)
1.21-1.38 1.38-1.55 0.45-0.59
Weak rock Medium rock Strong rock
- Wycliffe-Jones (1974) - Wycliffe-Jones (1974) PCI (1974) PCI (1974)
Canada - Coates (1970) New Zealand - Irwin (1971) New Zealand - Irwin (1971) Britain - Littlejohn (1973) Britain - Littlejohn (1973)
2.24 0.83-1.73
2.7-3.3 1.5-2.5
Britain - Wycliffe-Jones (1974) USA - PCI (1974)
0.17-0.25 (0.45 cu) 0.35
3.0
0.10-0.14
0.37 0.21-0.83
2.7-3.7 1.5-2.5
Britain - Littlejohn (1970) cu = undrained cohesion Canada - Golder Brawner (1973) Britain - Wycliffe-Jones (1974) USA - PCI (1974)
Uniaxial compressive strength - 30 (up to a maximum value of 1.4 N mm" 2) 0.35-0.70 0.70-1.05 1.05-1.40
Uniaxial compressive strength - 10 (up to a maximum value of 4.2 N mm - 2 )
3
0.40 0.60 0.69-0.83
Wide variety of igneous and metamorphic rocks
1.05
Wide variety of rocks
0.98 0.50 0.70
Britain - Littlejohn (1972)
Australia - Koch (1972)
2
1.20-2.50
0.70
Concrete
Britain Britain USA USA -
3.0 2.0-2.5 3.0 3.0
Weak shale
General Competent rock (whereUCS>20NmnT2)
Switzerland - Losinger (1966) Britain - Littlejohn (1970)
0.69-0.85 0.69
Argillaceous sediments Keuper marl
Soft sandstone and shale Soft shale
1.75
2.45
Source
0.69
2.76
1.4
4.2 15-20% of grout crushing strength 1.38-2.76
2-2.5 (Temporary) 3 (Permanent) 4 3 3
1.5-2.5
Australia - Standard CA35 (1973) France - Fargeot (1972) Switzerland - Walther (1959) Switzerland - Comte (1965) Switzerland - Comte (1971) Italy - Mascardi (1973)
Canada - Golder Brawnér (1973) USA - White (1973) Australia - Longworth (1971)
USA - PCI (1974)
423
Overview of Rock Anchorages
factor of safety relates to the ultimate and working bond values, calculated assuming a uniform bond distribution. It is common to find that the magnitude of bond is simply assessed by experienced engineers and the value adopted for working bond stress often lies in the range 0.35-1.4 N m m - 2 . The Australian Code [22] states that whilst a value of 1.05 N m m " 2 has been used in a wide range of igneous and sedimentary rocks, site testing has permitted bond values of up to 2.1 N m m " 2 to be employed. In this connection the draft Czech Standard [23] concludes that since the estimation of bond magnitude and distribution is a complex problem, field anchorage tests should always be conducted to confirm bond values in design, as there is no efficient or reliable alternative. Certainly, a common procedure amongst anchorage designers is to arrive at estimates of permissible working bond values by factoring the value of the average ultimate bond calculated from test anchorages. In general, there is a scarcity of empirical design rules for the various categories of rocks, and, as shown in Table 4, too often bond values are quoted without provision of strength data, or a proper classification of the rock and cement grout. The degree of weathering of the rock is a major factor which affects not only the ultimate bond but also the load-deflection characteristics. Degree of weathering is seldom quantified but for design in weak or weathered rocks there are signs that the standard penetration test is being further exploited. For example, in weathered granite in Japan the magnitude of the ultimate bond has been determined [6] from equation (2) Tult
= 0.007 N + 0.12 (Nmrn"2)
(2)
where N = number of blows per 0.3 m. Similarly, equation (3) has been established for stiff/hard chalk [5] Tult = 0.01 N(Nram" 2 )
(3)
More recent case histories [24] in chalk indicate that pressure grouting can create higher skin frictions, e.g. Tult = 0.02 to 0.03 N (N mm" 2 ). However, it should be noted that N values in chalk are subject to considerable scatter [25] and proving tests (see Section 15.7.1) are recommended to verify design assumptions. With the exception of rock bolts, the fixed anchor length should not be less than 3 m (2 m in rock if the working load < 200 kN). Under certain conditions it is recognized that much shorter lengths would suffice, even after the application of à generous factor of safety. However, for a very short anchor the effect of any sudden drop in rock quality along the fixed anchor zone, and/or constructional errors or inefficiencies could induce a serious decrease in that anchor's capacity. As a result, a minimum length of 3 m is often specified. Where load is transferred primarily by bond or shear an upper limiting length exists beyond which the extra length is redundant unless the proximal end of the fixed anchor yields. In practice, fixed anchor lengths seldom exceed 10 m even in rocks with low skin frictions such as weak mudstones and shales. In Italy, experimental research [26] has been conducted into the distribution of stresses both along the fixed anchor and into the rock. From this work it is concluded that the active portion of the fixed anchor is independent of the total fixed anchor length, but dependent on its diameter and the mechanical properties of the surrounding rock, especially its modulus of elasticity. Figure 6 [26] illustrates the uneven bond distribution as calculated from strain gauge data. Both anchorages were installed in 120 mm diameter boreholes in marly limestone (E = 3 x 104 kN m" 2 ; t Load ( tonne) Fixed anchor length =5.9 m
~ 294
Έ
Fixed anchor length = Mm
2
<Λ ■σ
g
98
CD
1
2
3 Anchor length (m)
1
2
3
Figure 6 Distribution of bond along fixed anchor length (after Berardi [26])
424
Support 0
Figure 7
01
{Tx/T)wd2 02 03
04
05
Variation of bond stress with depth along the rock/grout interface of a fixed anchor (after Coates and Yu [27])
UCS = 100 N mm" 2 approx.). Other results show that the bond distributions are more uniform for high values of £grout : ETOCk9 and nonuniform for low values of this ratio, i.e. for rock of high elastic modulus. These findings have also been predicted by Canadian researchers [27] (see Figure 7). It may be concluded that the distribution of the bond mobilized at the rock/grout interface is unlikely to be uniform unless the rock is weak. Nonuniformity applies to most rocks where ^grout'^rockisless than 10. Although it would appear from evidence presented that the assumptions made in relation to uniform bond distribution are not strictly accurate, it is noteworthy that few failures are encountered at the rock/grout interface and new designs are often based on the successful completion of former projects; that is, former 'working' bond values are reemployed or slightly modified depending on the judgement of the designer. 15.4.4 Grout/Tendon Interface Three mechanisms of bond, namely adhesion, friction and mechanical interlock, are widely recognized but recommendations pertaining to grout/tendon bond values in practice commonly take no account of the length or type of tendon, or the strength and cover of the surrounding grout [18]. For these reasons it is still advisable to measure experimentally the embedment length for untried field conditions. As an initial guide for cement-bonded tendons, the ultimate bond stress assumed to be uniform over the tendon bond length (see Figure 1) should not exceed: (i) 1.0 N mm - 2 for clean plain wire or plain bar; (ii) 1.5 N mm - 2 for clean crimped wire; (iii) 2.0 N mm - 2 for clean strand or deformed bar; (iv) 3.0 N mm" 2 for locally noded strands [28]. The above values are based on a minimum grout compressive strength of 30 N mm" 2 prior to stressing. They may be applied to single unit or parallel or multi-unit tendons, provided that the clear spacing is not less than 5 mm [19, 28]. For noded strands and tendons that can mobilize mechanical interlock or the shear strength of the grout, the minimum spacing criterion does not apply. For resinous grouts, ultimate bond values should be obtained from proving tests in the absence of relevant documented test data. The tendon bond length should not be less than the following, unless full-scale tests confirm that shorter bond lengths are acceptable: (i) 3 m where the tendon is homed and bonded in situ; and (ii) 2 m where the tendon is bonded under factory-controlled conditions. For low-capacity rock bolts, shorter grouted tendon bond lengths are used but the design should still consider the ultimate bond stresses above and be verified by proving tests. Bond strength can be significantly affected by the surface condition of the tendon, particularly when loose or lubricant materials are present at the interface. The surface of the tendons should
Overview of Rock Anchorages
425
Strain,Six I 0 3 )
Figure 8 Strain distribution along tendon in fixed anchor zone of a 2200 kN capacity anchorage (after Müller [29])
therefore be free from loose rust, soil, paint, grease, soap or other lubricants. A light film of rust on the tendon is not harmful and may improve bond. On the other hand, tendons showing signs of pitting should not be used. For resin-bonded rock bolts, the ratio of area of bar to area of borehole should be in accordance with the resin manufacturer's recommendations, since for resin capsules the annular space between the bar and the borehole is critical to allow efficient mixing of the capsule contents. For all other anchorages, it is recommended that the tendon area should not exceed 15% of the borehole area for parallel multi-unit tendons and 20% of the borehole area for single-unit tendons or suitably noded multistrand tendons, in order to minimize debonding [19, 28]. Figure 8 [29] illustrates the deboiiding which occurs in a high-capacity anchorage as the ductile tendon transfers stress to the brittle cement grout.
15.4.5 Materials and Components 15.4.5.1
Cementitious grouts
All conventional hydraulic cements, namely ordinary, rapid-hardening, sulfate-resisting and lowheat Portland are acceptable, but to avoid stress corrosion of the steel tendon, the total chloride content of the grout derived from all sources should not exceed 0.1% by weight of cement. To ensure that the cement grout has good bond and shear strength, the mix should attain an unconfined compressive strength of 40 N m m " 2 at 28 days. The bleeding of the tendon bonding grout should generally not exceed 2% of the volume 3 h after mixing. Higher values may be permitted in the case of permeable ground where the bleed water is filtered from the grout during injection under pressure. Given these design considerations, the water:cement ratio seldom exceeds 0.45 for rock anchorages. Admixtures should only be used if tests have shown that their addition enhances the properties of the grout, e.g. by improving workability or durability, reducing bleed or shrinkage, or increasing the rate of strength development. Detailed guidance on the design of cement-based grouts is available elsewhere [30].
15.4.5.2
Resinous grouts
Epoxy and polyester resins are most commonly used in capsules for rock bolting or tendon encapsulations for the protection of the bond length. For anchorages, ultimate compressive and tensile strengths in excess of 75 N m m " 2 and 15 N m m " 2 , respectively, are recommended traditionally for efficient load transfer, but post-gelation shrinkage, elastic modulus and percentage extension at failure are also important in relation to crack elimination in protection systems. Postgelation shrinkage should preferably be nil and not more than 5% otherwise debonding can occur, which in turn creates a potential corrosion hazard through the formation of leakage paths. To match the ductility of steel tendon it would appear that percentage extension of the resin at failure can be in the range 1 to 1.5%, but currently a minimum elastic modulus that will guarantee negligible creep under high service loads ( > 500 kN) cannot be recommended. As a consequence, where resin grout
426
Support
is proposed for tendon bonding, full-scale tests (including sample sectioning) should be carried out prior to anchorage installation to prove the efficiency of the grout mix. Given that the resin/hardener reaction is highly exothermic it can be beneficial both technically and economically to use inert fillers. Many crushed minerals are suitable but for capsule or encapsulation grouts, fillers should be graded with 100% passing a 200 μιη sieve. Resins are now used routinely for the rapid installation of rock bolts in underground excavations using a single or two speed resin system, but in thefieldof high capacity anchorages resins have not displaced the cheaper cement-based grouts. 15.4.5.3 Tendon Tendons usually consist of steel bar, strand or wire, either singly or in groups. For rock anchorages, Table 5 includes typical data for prestressing steel that may be used in tendon design. For such high strength steels the loss of prestress due to relaxation is small. For rock bolts or dowels not requiring high strength, steel of reinforcing quality in either smooth or deformed bar may be used as an alternative. Under normal circumstances working loads should not exceed 62.5% and 50% of the characteristic strength of the tendon for temporary and permanent works, respectively. To distribute load to the rock more uniformly, strands of different length are sometimes used within thefixedanchor zone. When these strands are stressed simultaneously, displacements at the anchor head are the same for all strands, and thus the strains and hence stresses differ in individual strands. In such cases the stress in the shortest strand should limit the acceptable working load. If the design requires uniform stresses within the tendon, mono-strand stressing is essential. Centralizers should be provided on all tendons to ensure that the tendon is centered in the grout column. Centralizers should provide within the borehole a minimum grout cover of 10 mm at the centralizer, and should be fitted at centers according to the angle of the rock anchorage and the possible sag between points of support in order to provide a minimum grout cover of 5 mm to the tendon. For rock reinforcement systems using a bar spun into resin, a centralizer may be used to retain the bar in the center of the hole and to retain resin in the fixed anchor section in up hole configurations. Spacers should be provided in thefixedanchor length of all parallel multi-unit tendons to ensure separation of not less than 5 mm between the individual components of the tendon and thus the Table 5
Typical Sizes and Specified Characteristic Strengths for Prestressing Tendon
Design (reproduced from BS 8081:1989 with permission of British Standards Institution) Type of steel
Non-alloy steel Wire 7-Wire strand 7-Wire drawn strand Low alloy steel bar Grade 1030/835
Grade 1230/1080 Stainless steel Wire Bar
Nominal diameter (mm)
7.0 12.9 15.2 15.7 12.7 15.2 18.0
Specified characteristic strength (kN)
60.4 186 232 265 209 300 380
Nominal steel area (mm2)
38.5 100 139 150 112 165 223
26.5 32 36 40 25 32 36
568 830 1048 1300 600 990 1252
552 804 1018 1257 491 804 1018
7 25 32 40
44.3 491 804 1257
38.5 491 804 1257
427
Overview of Rock Anchorages
effective penetration of grout to provide adequate bond. The spacer should not be compressible nor cause decoupling and a minimum of three spacers should be provided in each fixed anchor length. 15.4.5.4 Anchor head The anchor head normally consists of a stressing head in which the tendon is anchored and a bearing plate by which the tendon force is transferred to the structure or excavation. The stressing head should be designed to permit the tendon to be stressed and anchored at any force up to 80% of the characteristic tendon strength and should permit force adjustment up or down during the initial stressing phase. Monitoring requirements during service will dictate the need for a normal or restressable head. The stressing head should also permit an angular deviation of ± 5° from the axial position of the tendon without having an adverse influence on the ultimate load carrying capacity of the anchor head. Bearing plates for high-capacity anchorages are normally designed to a national structural code, but for low-capacity rock bolts the bearing or face plates may be flat, dished, square, triangular or circular, and designed to be rigid or to deform, depending on requirements. Table 6 [15] illustrates typical plate sizes for rock bolt applications, where it is generally sufficient for the plate to bear directly onto the rock surface. Alternatively, for irregular surfaces or to improve the spread of load onto the rock the plates may be bedded onto resin or mortar pads. 15.4.6 Safety Factors The traditional aim in design is to make a structure equally strong in all its parts so that, when purposely overloaded to cause failure, each part will collapse simultaneously. Have you heard of the wonderful one hoss shay, That was built in such a logical way? It ran for a hundred years to a day, And then, of a sudden it. . . .. . went to pieces all at once,All at once, and NOTHING FIRSTJust as bubbles do when they burst. The Deacon's Masterpiece, by Dr Oliver Wendell Holmes Thus for each potential failure mechanism a safety factor must be chosen having regard to how accurately the relevant characteristics are known, whether the system is temporary or permanent, i.e. service life, and the consequences if failure occurs, i.e. danger to public safety and cost of structural damage. Since the minimum safety factor is applied to those anchorage components known with the greatest degree of accuracy, the minimum values used in practice invariably apply to the characteristic strength of the tendon or anchor head and thereby encourage a ductile failure. Suitable safety factors are listed in Table 7. In regard to failure within the rock or at the rock/grout interface of the fixed anchor, load safety factors (Sf) generally range from 2 to 4, where Sf is defined as the ultimate load (rf) divided by the working load (!TW). T{ may be regarded as the maximum load attained when thefixedanchor can be withdrawn steadily, e.g. creep in a highly weathered rock, or the maximum load attained prior to a Table 6 Typical Dimensions of Rock Bolt Face Plates (after CIRIA [15]) Working load of bolt (kN)
80 150 300
Size of plate (length of side or diameter) (mm)
Thickness (mm)
125--150 150--200 200--250
7 10 12
428
Support
Table 7 Minimum Safety Factors Recommended for Design of Individual Anchorages (reproduced from BS 8081:1989 with Permission of British Standards Institution) Anchorage category
Minimum safety factor' Tendon
Temporary anchorages where the service life is less than six months and failure would have no serious consequences and would not endanger public safety, e.g. short-term pile test loading using anchorages as a reaction system. Temporary anchorages with a service life of say up to two years where, although the consequences of failure are quite serious, there is no danger to public safety without adequate warning, e.g. retaining wall tieback. Permanent anchorages and temporary anchorages where corrosion risk is high and/ or the consequences of failure are serious, e.g. main cables of a suspension bridge or as a reaction for lifting heavy structural members. a
Grout/tendon or Ground/grout interface grout/encapsulation interface
Proof load factor
1.40
2.0
2.0
1.10
1.60
2.5a
2.5a
1.25
2.00
3.0b
3.0a
1.50
Minimum value of 2.0 may be used if full-scale field tests are available. May need to be raised to 4.0 to limit ground creep.
b
Notes. (1) In current practice the safety factor of an anchorage is the ratio of the ultimate load to design load. Table 7 above defines minimum safety factors at all the major component interfaces of an anchorage system. (2) Minimum safety factors for the ground/grout interface generally lie between 2.5 and 4.0. However, it is permissible to vary these, should full-scale field tests (trial anchorage tests) provide sufficient additional information to permit a reduction. (3) The safety factors applied to the ground/grout interface are invariably higher compared with the tendon values, the additional magnitude representing a margin of uncertainty.
sudden failure and loss of load, e.g. loss of bond in a strong competent rock. As more poor quality rock has been exploited by anchorages, so safety factors have steadily increased in value. It is also fair to say that engineers today are less tolerant of individual anchorage failures, and whereas a 5% failure rate was common in the 1960s, concern is quickly expressed today whenever the figure exceeds 1%. On the other hand, with reference to overall stability and uplift capacity in particular, the load safety factor has been reduced in practice down to 1.6 where the analysis is judged to be conservative, e.g. simple, weight of cone calculation but shear restraint is assumed to exist [31]. 15.5 CORROSION PROTECTION 15.5.1 General Out of millions of prestressed ground anchorages that have been installed around the world, 35 case histories of failure by tendon corrosion have been recorded [32], some of which were protected only by cement grout cover. Invariably the corrosion has been localized and failures have occurred after service of only a few weeks to many years. As a consequence, it is considered that all permanent anchorages and temporary anchorages exposed to aggressive conditions should be protected, the degree of protection depending primarily on factors such as consequence of failure, aggressivity of the environment and cost of protection. The object of design against corrosion is to ensure that during the design life of the ground anchorage the probability of unacceptable corrosion occurring is small. Various degrees of protection are possible, and for corrosion resistance, the anchorage should be protected overall, as partial protection of the tendon may only induce more severe corrosion on the unprotected part.
Overview of Rock Anchorages Table 8
429
Proposed Classes of Protection for Rock Anchor-
ages (after Fédération Internationale de la Précontrainte [32]) Anchorage category
Class of protection
Temporary
Temporary without protection Temporary with single protection Temporary with double protection
Permanent
Permanent with single protection Permanent with double protection
Choice of the class of protection (see Table 8) should be the responsibility of the designer. By definition, single protection implies that one physical barrier against corrosion is provided for the tendon prior to installation. Double protection implies the supply of two barriers where the purpose of the outer second barrier is to protect the inner barrier against the possibility of damage during tendon handling and placement. 15.5.2 Principles of Protection Protective systems should aim to exclude a moist gaseous atmosphere around the metal by totally enclosing it within an impervious covering or sheath. Cement grout injected in situ to bond the tendon to the rock does not constitute a part of a protective system because the grout quality and integrity cannot be assured. Furthermore, fluid materials that become brittle on hardening crack in service as the structure suffers differential strains, the onset of cracking depending upon tensile strength and ductility. Nonhardening fluid materials such as greases also have limitations as corrosion protection media, for the following reasons. (i) Fluids are susceptible to drying out, which is usually accompanied by shrinkage and a change in chemical properties. (ii) Fluids are liable to leakage if even slight damage is sustained by their containment sheaths. (iii) Fluids having virtually no shear strength are easily displaced and removed from the metal they are meant to protect. (iv) Even in ideal conditions their long-term chemical stability, e.g. susceptibility to oxidation, is not known with confidence. These aspects require that nonhardening materials are themselves protected or contained by a moisture-proof, robust form of sheathing, which must itself be resistant to corrosion. Nevertheless, nonhardening fluids such as grease fulfil an essential role in corrosion protection systems, in that they act as afillerto exclude the atmosphere from the surface of a steel tendon, create the correct electrochemical environment and reduce friction in the free length. Whilst a layer of grease is not considered acceptable as one of the physical barriers required in the decoupled free length of a double corrosion protection system, grease is acceptable as a protective barrier in a restressable anchor head, since the grease can be replaced or replenished. Use of thicker metal sections for the tendon, with sacrificial area in lieu of physical barriers, gives no effective protection, as corrosion is rarely uniform and extends most rapidly and preferentially at localized pits or surface irregularities. Noncorrodible metals may be used for anchorage components, subject to verifying their electrochemical behavior relative to other components, and stress corrosion characteristics in appropriate environments. 15.5.3 Protective Systems There is a variety of protective coatings or coverings. The principles of protection are the same for all parts of the anchorage, but different detailed treatments are necessary for the tendon bond length, tendon free length and anchor head.
430
Support
In the free length, protection is achieved generally by either injection of solidifying fluids to enclose the tendon or by preapplied coatings, or by a combination of both, depending on circumstances. The protective system should permit reasonably uninhibited extension of the tendon during stressing, and thereafter, if the anchorage is restressable. Greased and sheathed tendons are a popular solution in such circumstances (see Figure 9). Sacrificial metallic coatings for high strength steel ( > 1040 N mm"2) should not be used when such coatings can cause part of the steel tendon to act as a cathode in an uncontrolled manner in a galvanitic process. The bond length requires the same degree of protection as the free length. In addition, the protective elements have all to be capable of transmitting high tendon stresses to the ground. This requires strength and deformability characteristics that have to be checked structurally. The deformation of individual elements of the corrosion protection system should not be such as to allow continuing creep nor expose the tendon bond length through cracking. The requirements of no creep and no cracking are in conflict and few materials are available that can comply with them under the intensity of stress around the fixed anchor. Certain materials, notably epoxy or polyester resins, have appropriate strength, ductility and resistance to corrosion. They may be substituted for cementitious grouts but are more expensive. When used to encapsulate bond lengths of tendon in combination with plastics ducts, the compatibility of elastic properties of the anchorage components has to be examined to minimize decoupling or debonding of the resin from the duct. To ensure effective load transfer between duct and grout, ducts are corrugated. The pitch of corrugations should be within six and 12 times the duct wall thickness and the amplitude of corrugation should not be less than three times the wall thickness. The minimum wall thickness is 0.8 mm, but consideration of material type, method of installation and service required may demand a greater thickness. Duct material should be impervious to fluids. Typical examples of double protection arrangements for the bond length of bar and strand tendons are shown in Figures 10 and 11. Drilled hole grouted solid Plastics binding tape Tendons comprise 10 strands each,greased and then sheathed in polypropylene. Minimum thickness of plastics coating = 0.8mm Figure 9
Typical free length detail for single protection of strand tendon (after Littlejohn [5])
Individual polypropylene sheath around grease-coated strand
Flexible sacrificial group sheath
Two concentric high-strength plastic corrugated ducts
Lead in shoe
Typical Longitudinal Section Through Encapsulation(Showing Two Strands Only) Strand locating High-strength non-shrink tape \ encapsulation cement grout
Individual strand sheath (7 wire strand) Section A-A 3 strand system Figure 10
Spacer Section B-B 3 strand system
Strand deformation around king wire (3 per strand)
Section C-C 5 strand system
Section D-D 8 strand system
Typical double protection of bond length of strand tendon using a double corrugated sheath and cement grout (after Fédération Internationale de la Précontrainte [32])
431
Overview of Rock Anchorages pSmooth plastics tubing
Enlarged view V - V
(dimension is in millimeters)
Figure 11 Typical double protection of bond length of smooth or ribbed bar tendon using a double corrugated sheath (reproduced from BS 8081:1989 with permission of British Standards Institution)
Unlike fixed anchors, anchor heads cannot be wholly prefabricated. Because of the strain in the tendon associated with prestressing, friction grips for strand and locking nuts on bars cannot fix the tendon until extension has been achieved. All existing locking arrangements require bare wire, strand or bar on which to grip, and any preformed corrosion protection of the tendon has to be removed. This leaves two sections of the tendon, above and below the bearing plate (outer head and inner head, respectively), which require separate protective measures in addition to the protection of the bearing plate itself. If the environment is aggressive, early protection of the anchor head is recommended for both temporary and permanent anchorages. The essence of inner head protection is to provide an effective overlap with the free length protection, to protect the short exposed length of tendon below the plate and to isolate the short section of the exposed tendon passing through the plate. In satisfying these recommendations, the protective measures have to allow free movement of the tendon that in certain instances may be solved by the use of a telescopic duct. Cement grouts are generally considered unsuitable for inner head protection. Primary grout should not be in contact with the structure and where a weak, lowbleed secondary grout is required to fill the void above the primary grout, it may be subject to cracking during structural movement. Grease-based corrosion protection compounds, or similar ductile materials immiscible with water, may be required. They may be preplaced or injected, and should be fully contained within surrounding ducts and retained by an end seal. Outer head protection of the bare tendon, the friction grips or the locking nuts above the bearing plate generally falls into two categories, controlled by whether the anchorage is restressable or not. Where restressability is called for, both the anchor head cap and the contents should be removable to allow access to an adequate length of tendon for restressing. Clearly these requirements will vary depending on the stressing and locking system employed. Grease is the most commonly used material within plastics or steel caps. Alternatives include corrosion-resistant grease-impregnated tape and heat shrink sleeving. Where restressability is not a requirement of the anchorage, then the cap and its contents are not required to be removable. Thus resins or other setting sealants may be used and a mechanical coupling between the cap and the bearing plate is not essential. Where the anchor head is to be totally enclosed by the structure, the outer head components may be encased in dense concrete as an alternative protection, given adequate cover. The bearing plate and other essential exposed steel components at the anchor head should be painted with bitumastic or other protective materials, prior to being brought to site. Steel surfaces should be cleaned of all rust and deleterious matter prior to priming, e.g. by blast cleaning. The
432
Support All steel components of gussets,ducts, bearing plates and caps coated with two coats of pitch epoxy
Free length classified as single protection,since grease is discounted as a protective barrier
Figure 12 Typical restressable anchor head detail for double protection of strand tendon (after Fédération Internationale de la Précontrainte [32])
Figure 13 Typical detensionable anchor head for double protection of bar tendon (after Fédération Internationale de la Précontrainte [32])
coatings should be compatible with the materials selected for both inner head and outer head protection. Bearing plates on concrete structures may be set in a seating formed of concrete, cement, epoxy or polyester mortar or alternatively may be seated direct onto a cast in steel plate. Typical examples of double protection arrangements for the anchor head are illustrated in Figures 12 and 13. 15.6 CONSTRUCTION 15.6.1 General During rock anchorage construction, the method of drilling (with or withoutflushing),the tendon installation, the grouting system and the time period of these operations may influence the capacity of the anchorage. Anchorage construction should be carried out in a manner whereby the validity of
Overview of Rock Anchorages
433
design assumptions is maintained, and a method statement detailing all operations, including drilling and grouting plant information, should be prepared prior to site anchorage work. Anchorage work is specialized and should always be carried out under the supervision of experienced personnel.
15.6.2
Drilling
Any drilling procedure may be employed that can supply a stable hole that is within the permitted tolerances and free of obstructions in order to accommodate the tendon easily. Drilling necessarily disturbs the ground and the method should be chosen relative to the ground conditions to cause either the minimum of disturbance or the disturbance most beneficial to the anchorage capacity. Care should be taken not to use high pressures with any flushing media, in order to minimize the risk of hydrofracture of the surrounding ground, particularly in built-up areas. In this connection, an open return within the borehole is desirable to limit pressures and it also permits the driller to monitor major changes in ground type from the drill cuttings or flush. Unless otherwise specified, the drill hole entry point should be positioned within a tolerance of ± 75 mm. The drilled hole should have a diameter not less than the specified diameter, and allowances for swelling may be necessary if the hole is open for several hours in, for example, overconsolidated marls. For a specified alignment at entry point, the hole should be drilled to an angle tolerance of ± 2.5°; unless, for closely spaced anchorages, such a tolerance could lead to interference of fixed anchor zones, in which case the inclination of alternate anchorages should be staggered. Rock anchorages should have a minimum inclination of approximately 10° to the horizontal to facilitate grouting. Assuming an acceptable initial alignment, overall drill hole deviations of 1 in 30 should be anticipated. On occasions, ground conditions may dictate a relaxation of this tolerance and for downward and upward inclined holes, it is probable that the vertical deviations will be higher than lateral deviations. After each hole has been drilled to its full length and thoroughly flushed out to remove any loose material, the hole should be probed to ascertain whether collapse of material has occurred and whether it will prevent the tendon being installed completely. For downward inclined holes, up to 1 m of overdrill may be added to cater for detritus that cannot be removed. Tendon installation and grouting should be carried out on the same day as drilling of the fixed anchor length, since a delay between completion of drilling and grouting can have serious consequences due to ground deterioration, particularly in overconsolidated, fissured mudstones and shales. During the drilling operations, all changes in ground type should be recorded together with notes on water levels encountered, drilling rates, flushing losses or gains, and stoppages.
15.6.3 Tendon Ideally, tendon steel in the bare condition should be stored indoors in clean dry conditions, but if left outdoors such steel should be stacked off the ground and be completely covered by a waterproof tarpaulin that is supported and fastened clear of the stack so as to permit circulation of air and avoid condensation. Bare or coated tendons should not be dragged across abrasive surfaces or through surface soil, and only fiber rope or webbing slings should be used for lifting coated tendons. In the event of damage, tendon which is kinked or sharply bent should be rejected because load-extension characteristics may be adversely affected. Over the bond length, bar tendons, multi-unit tendons and encapsulations should be centralized in the borehole to ensure a minimum grout cover to the tendon or encapsulation of 5 mm between centralized locations and 10 mm at centralizer locations (see Figure 14). For multi-unit tendons where the applied tensile load is transferred by bond, spacers should ensure a minimum clear spacing of 5 mm. Given tendons with local or general nodes that provide mechanical interlock, occasional contact between tendon units is permissible. A minimum of three spacers should be provided in each fixed anchor length, and both centralizers and spacers should be provided at centers according to the inclination and stiffness of the tendon, in order to provide the minimum clear spacing of cover (see also Section 15.4.5.3).
434
Support
Cross section
.Generally 1.0 to 3.0m (controlled by stiffness of encapsulation)
Borehole
Min 5mm grout cover between centralizers
Longitudinal section Figure 14 Typical encapsulation centralizer
At the bottom of the tendon, use of a sleeve or nose cone will minimize the risk of tendon or borehole damage during homing. Immediately prior to installation the tendon should be carefully inspected for damage to components and corrosion, after which the tendon should be lowered at a steady controlled rate. For heavy tendons weighing in excess of 200 kg, approximately, mechanical handling equipment should be employed, as manual operations can be difficult and hazardous. The use of a funnelled entry pipe at the top of a cased hole is also recommended to avoid tendon damage as it is installed past the sharp edge of the top of the casing. On occasion, particularly at the start of a contract, the tendon may be withdrawn after the installation operation, in order to judge the efficiency of the centralizer and spacer units and also to observe damage, distortion or the presence of smear, e.g. in chalk. Where significant distortion or smear is observed, improvements in relation to the fixing or design of the centralizers, or the borehole flushing method may be necessary. 15.6.4 Grouting Grouting performs one or more of the following functions. (i) It forms thefixedanchor in order that the applied load may be transferred from the tendon to the surrounding rock. (ii) It augments the protection of the tendon against corrosion. (iii) It strengthens the rock immediately adjacent to the fixed anchor in order to enhance anchorage capacity. (iv) It seals the rock immediately adjacent to the fixed anchor in order to limit loss of grout.
Overview of Rock Anchorages
435
The need for functions (iii) and/or (iv) should be highlighted by the ground investigation and/or as a result of pregrouting or water testing. To check that the loss of grout over the fixed anchor length is insignificant during injection for anchorages in permeable rock, it is normally adequate to observe a controlled grout flow rate coupled with a back pressure. The efficiency of fixed anchor grouting can be finally checked by monitoring the response of the rock to further injection when the back pressure should be quickly restored. Where pressure grouting is not carried out as part of routine anchorage construction, a falling head grout test can be used where the borehole is prefilled with grout typically having vv/c = 0.4-0.6, and the grout level observed until it becomes steady. If the level continues to fall it should be topped up and after sufficient stiffening of the grout (but prior to hardening), the borehole should be redrilled and retested. The test may be applied to the entire borehole or restricted to the fixed anchor length by packer or casing over the free anchor length. The likelihood of cement-grout loss in rock can also be assessed from an analysis of a water injection test [33], although the test is particularly rigorous and interpretation of the results demands care, since the leakage may be due to a single fracture of consequence, or many micro fractures which might not accept cement. Routinely, a falling head test is applied to the borehole or the fixed anchor length, and pregrouting is not required if the water loss is less than 5 L min" * at an excess head of 0.1 M Pa (one atmosphere) over a period of 10 min. Where there is a measured water gain under artesian conditions, care should be taken to counteract this flow by the application of a back pressure prior to grouting. If the flow cannot be stabilized in this way, pregrouting is required irrespective of the magnitude of the water gain. The acceptable water loss in current US recommendations [34] is 0.49 mL per millimeter diameter of borehole per meter of depth at an excess head of 0.034 M Pa, and again care is required in interpretation, since for rock with occasional but significant fractures the borehole depth and diameter are, strictly speaking, not relevant to the estimation of potential grout loss. In general, if the grout volume exceeds three times the borehole volume for injection pressures less than total overburden pressure, then general void filling is indicated which is beyond routine anchorage construction. This extra grout merits additional payment. For the preparation of cement grout, batching of the dry materials should be by mass, and mixing should be carried out mechanically for at least 2 min in order to obtain a homogeneous mix. Thereafter, the grout should be kept in continuous movement, e.g. slow agitation in a storage tank. As soon as practicable after mixing, the grout should be pumped to its final position, and it is undesirable to use the grout after a period equivalent to its initial setting time (see Figure 15). High speed colloidal mixers (1000 revolutions min" 1 minimum) and paddle mixers (150 revolutions min" l minimum) are permissible for mixing neat cement grouts, although the former mixer is preferred in water-bearing ground conditions since dilution is minimized. Pumps should be of the positive displacement type, capable of exerting discharge pressures of at least 1000 kNm" 2 , and rotary screw (constant pressure) or reciprocating ram and piston (fluctuating pressure) pumps are acceptable in practice. Before grouting, all air in the pump and line should be expelled, and the suction circuit of the pump should be airtight. During grouting, the level of grout in the supply tank should not be drawn down below the crown of the exit pipe, as otherwise air will be injected. An injection pressure of 20 kN m" 2 per meter depth of ground is common in practice. Where high pressures that could hydrofracture the ground are permitted, careful monitoring of grout pressure and quantity over the fixed anchor length is recommended. If, on completion of grouting, the fluid
Initia I vicat set
Time (h)
Figure 15 Setting times for ordinary Portland (Type I) cement grouts at 18°C (after Littlejohn [30])
436
Support
grout remains adjacent to the anchored structure then the shaft grout should beflushedback 1-2 m to avoid a strut effect during stressing. In regard to quality controls, emphasis should be placed on those tests that permit grout to be assessed prior to injection. As a routine, initial fluidity byflowcone or flow trough, density by mud balance and bleed by 1000 mL graduated cylinder (75 mm diameter) should be measured daily along with 100 mm cube samples for later crushing at 7 and 28 days, say. These quality controls relate to grout batching and mixing and the tests do not attempt to simulate the properties of the grout in situ. For example, water loss from grout, when injected under pressure into permeable sandstone, creates an in situ strength greater than the cube strength for similar curing conditions. Records relating to each grouting operation should be compiled, e.g. age of constituents, air temperature, grouting pressure, quantity of grout injected and details of samples and tests, as appropriate. 15.6.5 Anchor Head The stressing head and bearing plate should be assembled concentrically with the tendon to an accuracy of + 10 mm and should be positioned not more than 5° from the tendon axis. After final grouting or satisfactory testing, cutting of the tendon should be done without heat, e.g. by a disc-cutter, in which case the cut should not be closer than one tendon unit diameter from the face of the holding wedge or nut. Projecting tendons, whether stressed or not, should be protected against accidental damage. This protection is not common in practice and if individual tendon components are mechanically damaged, e.g. kinking of strand, then these components should be considered redundant, when assessing a safe anchorage capacity, unless tests confirm adequacy. 15.6.6 Stressing Stressing is required to fulfill two functions. (i) To tension the tendon and to anchor it at its secure load. (ii) To ascertain and record the behavior of the anchorage so that it can be compared with the behavior of control anchorages, subjected to on-site suitability tests. A stressing operation means an activity involving thefittingof the jack assembly on to the anchor head, the loading or unloading of the anchorage including cyclic loading where specified, followed by the complete removal of the jack assembly from the anchor head. Stressing and recording should be carried out by experienced personnel under the control of a suitably qualified supervisor, since any significant variation in procedure can invalidate comparison with control anchorages. At the present time, equipment calibration is not carried out regularly and discrepancies between jack and load cell readings are not uncommon on site. Jacks should be calibrated at least every year, using properly designed test equipment with an absolute accuracy not exceeding 0.5%. The calibration should cover the load rising and load falling modes over the full working range of the jack, so that the friction hysterisis is known when repeated loading cycles are being carried out on the tendon. Load cells should be calibrated after every 200 stressings or after every 60 days in use, whichever is the more frequent, unless complementary pressure gauges used simultaneously indicate no significant variation, in which case the interval between calibrations may be extended up to a maximum of one year. Pressure gauges should also be calibrated regularly, e.g. after every 100 stressings or after every 30 days, whichever is the more frequent, against properly maintained master gauges, or whenever the field gauges have been subjected to shock. If a group of three gauges is employed this frequency does not apply. On every contract the method of tensioning to be used and the sequence of stressing should be specified at the planning stage. In general, no tendon should be stressed at any time beyond either 80% of the characteristic strength (equivalent to 80% GUTS in US) or 95% of the characteristic 0.1% proof strength. In addition, for cement-grouted fixed anchors, stressing should not commence until the grout has attained a crushing strength of at least 30 N mm" 2 . However, in sensitive rock, e.g. chalk or mudstone, which may be weakened by water softening or disturbance during anchorage construction, it may be necessary to stipulate a minimum number of days before stressing. Details of all forces, displacements, seating and other losses observed during stressing and the times at which the data were monitored should be recorded for every anchorage.
437
Overview of Rock Anchorages
Finally, it is worth noting that when a stressing operation is the start point for future time-related load measurements, stressing should be concluded with a check-lift load measurement. During stressing, safety precautions are essential and operatives and observers should stand to one side of the tensioning equipment and never pass behind when it is under load. Notices should also be displayed stating 'DANGER - Tensioning in Progress' or similar wording.
15.7 TESTING 15.7.1 General There are three classes of tests for all anchorages as follows: (i) proving tests, (ii) on-site suitability tests, and (iii) on-site acceptance tests. Proving tests may be required to demonstrate or investigate in advance of the installation of working anchorages, the quality and adequacy of the design in relation to rock conditions and materials used, and the levels of safety that the design provides. The tests may be more rigorous than on-site suitability tests and the results, therefore, cannot always be directly compared, e.g. where short fixed anchors of different lengths are installed and tested, ideally to failure. On-site suitability tests are carried out on anchorages constructed under identical conditions as the working anchorages and loaded in the same way to the same level. These may be carried out in advance of the main contract or on selected working anchorages during the course of construction. The period of monitoring should be sufficient to ensure that prestress or creep fluctuations stabilize within tolerable limits. These tests indicate the results that should be obtained from the working anchorages. On-site acceptance tests are carried out on all anchorages and demonstrate the short-term ability of the anchorage to support a load that is greater than the design working load and the efficiency of load transmission to the fixed anchor zone. A proper comparison of the short-term service results with those of the on-site suitability tests provides a guide to longer-term behavior. 15.7.2 On-site Acceptance Tests Every anchorage used on a contract should be subjected to an acceptance test. As a principle, acceptance testing should comprise standard procedures and acceptance criteria which are independent of ground type, and should be of short duration. In this regard the maximum proof loads are dictated by Table 7, but acceptable load increments and minimum periods of observation have gradually been reduced over the years to save time and money (see Table 9). At each stage of loading, the displacement should be recorded at the beginning and end of each period, and for proof loads the minimum period of 1 min is extended to at least 15 min with an intermediate displacement at 5 min, so that any tendency to creep can be monitored. Table 9
Recommended Load Increments and Minimum Periods of Observation for On-site Acceptance Tests (after Littlejohn [35])
Temporary anchorages load increment (% Tw) 1st load cycle*
(%)
10 50 100 125 100 50 10 a
Permanent anchorages load increment (% Tw)
2nd load cycle
1st load cycle*
2nd load cycle
(%)
Minimum period of observation (min)
10 50 100 125 100 50 10
10 50 100 150 100 50 10
10 50 100 150 100 50 10
1 1 1 15 1 1 1
(%)
(%)
For this load cycle, which often includes extraneous nonrecoverable movements such as wedge 'pull-in', bearing plate settlement and initial fixed anchor displacement, there is no pause other than that necessary for the recording of displacement data.
438
Support
In order to establish the seat of load transfer within the anchorage, the apparent free length of the tendon may be calculated from the load-elastic displacement curve over the range of 10% 7W to 125% Tw (temporary anchorages) or 10% Tw to 150% Tw (permanent anchorages), using the manufacturer's value of elastic modulus and allowing for such effects as temperature and bedding of the anchor head. It is normally adequate simply to record the ambient temperature during the test, unless the monitoring equipment or anchored structure is known or observed to be temperature sensitive. The free length analysis should be based on the results obtained during the destressing stage of the second or any subsequent unloading cycle, otherwise extraneous nonrecoverable movements may mask the reproducible behavior of the anchorage in service (see Figure 16). For simplicity in practice the following equation is employed Apparent free tendon length =
AtEsAXc
(4)
where At is the cross section of the tendon, £ s is the manufacturer's elastic modulus for the tendon unit, AXe is the elastic displacement of the tendon (AXC is equated to the displacement monitored at proof load minus the displacement at datum load, i.e. 10% JTW say, after allowing for structural movement) and T is the proof load minus datum load. On completion of the second cycle, the anchorage is reloaded in one operation to 110% Tw say, and locked-off, after which the load is reread to establish the initial residual load. This moment represents zero time for monitoring load/displacement-time behavior during service. Where loss of load is monitored accurately using load cells with a relative accuracy of 0.5%, readings can be attempted within the first 50 min. Where monitoring involves a stressing operation, e.g. lift-off check without load cell, an accuracy of less than 5% is unlikely and longer observation periods of 1 day and beyond are required. Where displacement-time data are required, a dial gauge/tripod system (see Figure 17) is suitable for short duration testing, given that the tripod base should be surveyed accurately for movement. In practice, dial gauges reading to 0.01 mm are commonly used during the test, and where movement of the tripod base is anticipated, its position is checked before and after the test to an accuracy of 1 mm. 110% design free length or design free length plus 5 0 % tendon bond length
\
Design free length 1
9 0 % design free length
—Δ
Elastic displacement Displacement of tendon at anchor head
Figure 16 Acceptance criteria for displacement of tendon at anchor head (after Littlejohn [35])
Overview of Rock Anchorages
439
Figure 17 Typical method of measuring tendon displacement using a dial gauge
For the testing procedures outlined above, acceptance criteria based on proof load-time data, apparent free tendon length, and short-term service behavior, are proposed for temporary and permanent anchorages. These criteria are discussed in the following paragraphs. 15.7.3 Proof Load-Time Data If the proof load has not reduced during the 15 min observation period by more than 5% after allowing for any movement of the anchored structure, the anchorage may be deemed satisfactory. If a greater loss of prestress is recorded, the anchorage should be subject to two further proof load cycles and the behaviour recorded. If the 5% criterion is not exceeded on both occasions the anchorage may be deemed satisfactory. If the 5% criterion is exceeded on either cycle the proof load should be reduced to a value at which compliance with the 5% criterion can be achieved. Thereafter, the anchorage may be accepted at a derated proof load, if appropriate. As an alternative to these recommendations, the proof load can be maintained by jacking and the anchor head monitored after 15 min, in which case the creep criterion is 5% AXe. For anchorages that have failed a proof load criterion, tendon unit stressing may help to ascertain location of failure, e.g. for a temporary anchorage, pull-out of individual tendon units may indicate debonding at the grout/tendon interface, whereas, if all tendon units hold their individual proof loads, attention is directed towards failure of the fixed anchor at the rock/grout interface. 15.7.4 Apparent Free Tendon Length The apparent free tendon length should be not less than 90% of the free length intended in the design, nor more than the intended free length plus 50% of tendon bond length or 110% of the intended free tendon length (see Figure 16). The latter upper limit takes account of relatively short encapsulated tendon bond lengths and fully decoupled tendons with an end plate or nut. Where the observed free tendon length falls outside the limits, a further two load cycles up to proof load should be carried out in order to gauge reproducibility of the load-displacement data. If the anchorage behaves consistently in an elastic manner, the anchorage need not be abandoned, provided the reason can be diagnosed and accepted. In this regard it is noteworthy that the E value of a long multistrand tendon may be less than the manufacturer's E value for a single strand, which has been measured over a short gauge length between rigid platens. A reduction in the manufacturer's E value of up to 10% should be allowed in any field diagnosis. 15.7.5 Short-term Service Behavior Using accurate load cell and logging equipment, the residual load may be monitored at 5,15 and 50 min. If the rate of load loss reduces to 1% or less per time interval for these specific observation
440
Support Table 10 Acceptance Criteria for Service Behavior at Residual Load (after Littlejohn [35]) Permissible loss of load (% initial residual load,
Period of observation (min)
(%) 1 2 3 4 5 6 7 8
5 15 50 150 500 1500 ( Ä 1 day) 5000 ( « 3 days) 15 000 ( « 1 0 days)
Permissible displacement (% of elastic extension Ae of tendon at initial residual load) (%) 1 2 3 4 5 6 7 8
periods after allowing for temperature (where necessary), structural movements and relaxation of the tendon, the anchorage may be deemed satisfactory. If the rate of load loss exceeds 1%, further readings may be taken at observation periods of up to 10 days (see Table 10). If, after 10 days, the anchorage fails to hold its load as given in Table 10, the anchorage is not satisfactory and following an investigation as to the cause of failure, the anchorage should be (i) abandoned and replaced, (ii) reduced in capacity, or (iii) subjected to a remedial stressing programme. Where prestress gains are recorded, monitoring should continue to ensure stabilization of près tress within a load increment of 10% Tw. Should the gain exceed 10% Tw, a careful analysis is required and it will be prudent to monitor the overall structure/ground/anchorage system. If, for example, overloading progressively increases due to insufficient anchorage capacity in design or failure of a slope, then additional support is required to stabilize the overall anchorage system. Destressing to working loads should be carried out as prestress values approach proof loads, accepting that movement may continue until additional support is provided. As an alternative to load monitoring, displacement-time data at the residual load may be obtained at the specific observation periods in Table 10, in which case the rate of displacement should reduce to 1% Ae or less per time interval. This value is the displacement equivalent to the amount of tendon shortening caused by a prestress loss of 1 % initial residual load, i.e. Ae =
Initial residual load x apparent free tendon length Area of tendon x elastic modulus of tendon
(5)
If the anchorages are to be used in the work and, on completion of the on-site acceptance test, the cumulative relaxation or creep has exceeded 5% initial residual load or 5% Ae, respectively, the anchorage should be restressed and locked-off at 110% Tw, say. This procedure ensures that a contingency overload is locked into the ground anchorage at the start of its service. As a general guide, either acceptance criterion for short-term service, i.e. rate of prestress loss or rate of displacement, may be applied quite independently for the common range of free tendon lengths. For short free tendon lengths ( < 5 m), loss of prestress becomes the more appropriate criterion, while for long free tendon lengths ( > 30 m) it is clear that creep displacement may be more important to limit and therefore more appropriate as an acceptance criterion. 15.7.6 Monitoring Service Behavior As for buildings, bridges and dams, monitoring of structure/ground/anchorage systems will be appropriate on occasions. In general, monitoring is recommended for important structures where the following circumstances apply. (i) Wherever the behavior of anchorages can be ascertained safely by monitoring the behavior of the structure as a whole, e.g. by precise surveying of movements. (ii) Wherever the malfunctioning of anchorages could endanger the structure and cause it to become a hazard to life or property, and where problems would not be detected before the structure became unserviceable other than by monitoring.
Overview of Rock Anchorages
441
(iii) When, due to the nature of the ground and/or the protective system, tendons cannot be bonded to the walls of their holes, so that breakage of a tendon at any point renders it ineffective throughout its length. (iv) Where anchorages are of a pattern that has not been proved adequately in advance, either by rigorous laboratory tests or by site performance under similar circumstances. (v) Where anchorages are in rock liable to creep. Two methods of monitoring are in common use, namely measurement of loads on individual anchorages or measurement of the performance of structures or excavations as a whole, the latter being preferable. When monitoring individual anchorages, the maximum loss or gain of prestress that can be tolerated during service should be indicated, taking into account the design of the works. Variations up to 10% of working load do not generally cause concern. Prestress losses greater than 10% should be investigated to ascertain cause and consequence, and for prestress gain, remedial action, which may involve partial destressing or additional anchorages, is recommended when the increases exceed 20% Tw and 40% Tw for temporary and permanent anchorages, respectively. In general, monitoring should initially be at short intervals of 3-6 months, with later tests at longer intervals depending on results. The number of anchorages to be monitored should be indicated by the designer of the works; 5-10% of the total is typical in current practice.
15.8 CASE EXAMPLES 15.8.1 General In many countries anchorages have established a permanent place in construction practice, but for engineers not yet fully familiar with modern anchoring technology, practical applications are outlined in this chapter to provide some perspective and encourage further exploitation. Anchorages, often in conjunction with a drainage system, can be used to improve the stability of existing slopes or allow steeper slopes to be excavated. Cliff stabilization by rock-bolt reinforcement to overcome spalling in localized regions is also common, and notable examples include Edinburgh Castle rock in Scotland and the American Falls at Niagara. During the construction of tunnels, galleries and caverns for underground power stations or oil storage, rock anchorages are used to improve the mechanical properties of the rock and thus stabilize the excavation against collapse or excessive convergence. Rock bolting of tunnels and galleries is routine practice throughout the world and examples of anchored caverns include: Roncovalgrande, Italy; El Toro, Chile (Figure 18); Sackingen, Germany; Drakensburg, South Africa; and Dinorwic, Wales. For the strengthening or raising of dams, anchorages may be used to apply a vertical or subvertical force through the structure in order that it can withstand existing or increased lateral water thrust. Anchoring is a less costly and faster method than alternatives, and installation can proceed without interference to the normal reservoir storage. Recent strengthening contracts include Milton Lake Dam in Ohio, Lalla Takerkoust Dam in Morocco (Figure 19) and Manly Dam in Australia. Applied during dam construction, anchorages give substantial savings in concrete and time, and pioneering projects include Swallow Falls in South Africa and Catagunya Dam in Tasmania. Where new dams are founded on strata with underlying weak layers, anchorages can also be used to resist sliding, e.g. Muda Dam in Malaysia (Figure 20) and the Newburgh Dam in Indiana, where 7500 kN anchorages compensated for the presence of a weak layer of coal and underclay in the local sandy shale. Bridge structures often call for the highest capacity anchorages, particularly in the case of longspan suspension bridges. At Dent bridge, Idaho, the cable anchorage block was prestressed into the ground using grout injection anchorages to provide a restraint of 156 MN. For the Forth road bridge in Scotland, each side tower, 54 m high, was prestressed into the underlying rock using a total anchorage capacity of 48 MN. Out with the major applications already described, individual concentrated loads may have to be resisted in certain circumstances, for example, moorings for cable railways, penstocks, gantry cranes and pile tests. Post-tensioned anchorages can offer a convenient solution where the prestress minimizes deformation under loading and for dynamic conditions the prestressing can be further increased to eliminate fatigue failure. During the building of the arch dam at Jiroft in Iran, two cable cranes, each of 200 kN capacity and 520 m length, were used. The suspension cables were secured by one fixed point (Figure 21), a
442
Support
Figure 18 Cross section of El Toro Cavern, Chile
- Existing grouting
Figure 19 Lalla Takerkoust Dam, Morocco
nchor head
Mudsrone beds
Anchorage
Figure 20 M uda Dam, Malaysia
Overview of Rock Anchorages
Figure 21
443
Cable crane anchorage at Jiroft Dam, Iran
concrete structure anchored into limestone using eight anchorages to resist a total tensile load of 6400 kN. With the increase in height of office and residential tower blocks, often associated with large diameter piles, traditional methods of pile and plate load testing using kentledge can be uneconomic or impracticable, particularly where test loads are high or the space available is restricted. In strong competent rock, standard anchorage systems can mobilize loads of up to 2 MN via ring reaction beams. A wide range of applications now exists for both temporary and permanent anchorages, but whilst anchorage technology has developed rapidly there is still a reluctance to invest in performance studies during service. An absence of problems may be the reason but the following examples of monitoring are included to illustrate the benefits that can be gained. 15.8.2 Static Performance During Service The advantages of monitoring include: (i) the engineer being able to feed back performance observations into future designs and thereby to optimize such parameters as overload allowances and load safety factors; and (ii) the prospective client being accurately and confidently informed of how anchorages installed at his expense will perform after installation. Furthermore, the data collection permits all parties to judge at the earliest possible stage whether anchorages being monitored are, in fact, acting satisfactorily. On a more general front, this form of monitoring may permit correlation of anchorage load and structural movement, and thereby lead to a better understanding of anchorage/ground/structure interaction. In the construction of the Submarine Refit Complex at HM Dockyard, Devonport, England, twin dry docks were constructed in an existing basin approximately 140 m2, and surrounded on three sides by mass concrete retaining walls founded directly on bedrock. Initially, the project featured the production of a dredged and dewatered basin some 18 m deep necessitating the construction of a cellular steel sheet pile cofferdam across the south of the basin, and the stabilization of the existing basin walls against overturning (see Figure 22). The method of ensuring wall stability was to install 330 No. 2000 kN anchorages in holes angled as near to the heel of the wall as possible, and founded in the underlying bedrock (see Figure 22, Sections A-A and B-B). The design, construction and stressing of the anchorages have been described by Littlejohn and Truman-Davies [31]. At an early stage in the anchoring contract, permission was granted to monitor the time-related performance of selected production anchorages. The study had two principal aims: (i) to investigate the actual anchorage loads during the crucial basin dewatering stage, and (ii) to provide a case history of the long-term behavior of permanent high-capacity rock anchorages. The site at Devonport is underlain by a series of géosynclinal Upper Devonian sediments, mainly in the form of hard grey, purple and dark blue slates, known locally as 'shillet'. The rock surface dips at an average of 3.5° from north-east to south-west across the site, and the uppermost 1.5 m or so is commonly recorded as very weathered andfissile,with frequent softer shale or clay bands. Generally the rock is tightly and strongly folded, due to its participation in the Amorican orogeny, and the cleavage dip varies from 60 to 80°. Very little geotechnical data were actually made available upon which to base design - core recoveries of 80-100%, and a submerged density of 1.28 Mg m" 3 . Some core samples were later obtained which enabled diametral point load tests to be conducted. The actual specimens were not
444
Support Anchors 2 7 4 - 2 7 6 - 2 0 0 0 kN anchors at 2.00 centers -
Π
- East w a l l Inclinometer 6 A
Inclinometer 5A
il m
Anchors 219-222
il
Inclinometer 4A
142 No. 2 0 0 0 kN underwater anchors
--^B
; □.νΛν.ϊΛϊθΛΪΛϊΛν, + + * + *■*·
«■♦ + «- + |
| Thrust block
Inclinometer 3A Thrust slab Anchors 5 1 - 4 9
Inclinometer 2 A
• Inclinometer IA West wall /
t~-^A
- 2 0 0 0 kN a n c h o r s - , at 1.00 centers L ^ B
/
-2000kNanchors at 1.50 centers
- 2 0 0 0 kN anchors at 1.00 c
4.24
N o r m a l Basin Level (NB L)
-13.60
0.00
Thrust block
Thrust block
-19.95 24.88
Section
A-A
Section
B-B
Figure 22 Layout of anchorages for the Devonport Submarine Refit Complex, highlighting the position of anchorages and inclinometers monitored during service (after Littlejohn and Bruce [36])
of ideal shape, due to the small angle between core axis and rock cleavage, and the very close separation of the cleavage planes. However, 12 tests gave values of Is in the range 0.45-0.97 N m m - 2 and an average of 0.67 N mm" 2 (moderately weak to moderately strong). This average value would relate to estimates of uniaxial compressive strength, elastic modulus and uniaxial tensile strength of 12.0,3.1 x 103, and 1.0 N mm" 2 respectively. The anistropy index ranged from 8 to 18, with a mean of 11. The salient features of the anchorages monitored may be summarized as follows. (i) The fixed anchor length was 8.0 m, with a nominal diameter of 140 mm as drilled by down-thehole percussive hammers, giving an average rock-grout bond at service load of approximately 600 kN m~ 2 . A factor of safety in excess of 3 against failure of the rock-grout bond was verified by one test anchorage.
445
Overview of Rock Anchorages
(ii) The tendons consisted mostly of 12 Dyform 15.2 mm strands, with a working stress of 55% characteristic strength and a steel section/borehole area ratio of 14.2%. Over the free length the strands were individually protected from corrosion, and debonded from the surcharge grout, by 1.5 mm wall thickness plastic sheath with grease infilling. (iii) Special spacer-centralizer units were located at 2 m centers in thefixedlength and the tendons were noded at intermediate distances. (iv) The tendons were homed mechanically into the holes, and then fully trémie grouted in one operation, with neat 0.45 w/c rapid-hardening Portland (Type III) cement grout. Ten anchorages were selected for monitoring and their overall performance is illustrated in Figures 23-25. Two distinct phases may be recognized in terms of rate of prestress loss. Phase I is reflected by a stabilizing, but fairly rapid loss with time, occurring within a period of 3000 h. Thereafter, a slower arid more uniform rate of prestress loss is observed (Phase II). Based on these limited results, it is recommended that where service performance is being studied, the duration of the study should cover completion of Phase I, and hopefully provide sufficient results at say monthly intervals to indicate a clear trend for Phase II, thereby permitting an extrapolation of the results to cover the service life of the anchorages. 9
0 "^ —
-^e0le^ie/oxc,on
Weeks after lock-off Figure 23 Long-term performance of west wall anchorages up to 18 500 h (after Littlejohn and Bruce [36])
222 o
25
Llheore^re/axa^on 219
——— 50
221
\
75 μ
220 1000h 1
< 100
3000h 1
10 0 0 0 h .1
1 50 Weeks after lock-off
1 25
_J 75
1 100
Figure 24 Long-term performance of north wall anchorages up to destressing at 14600 h (after Littlejohn and Bruce [36])
-o„_
St
■* 7^
5<» Ου
I §50
k;
Theoretical relaxation
\-
1000 h 1
3000 h 1 1 25
IOÖÖOTT-—— 1 1 1 50 75 Weeks after lock-off
276 274 275 1
100
Figure 25 Long-term performance of east wall anchorages up to 18 500 h (after Littlejohn and Bruce [36])
446
Support
The maximum prestress loss recorded was 4.7% at 3000 h when the rapid loss Phase I was complete and 7% after 33000 h. These values are reassuring bearing in mind the 10% over-load allowance commonly stipulated in anchorage practice. It may be generally concluded that the anchorages have functioned satisfactorily in terms of loadholding capacity during a crucial construction phase, and for the monitoring period of almost 4 years after stressing. 15.8.3 Dynamic Response During Service Where tunneling rock demands drill and blast methods of excavation, estimates of the distance from the tunnel face to a safe location for the installation of permanent rock bolts are usually based upon conservative distances derived from precedent practice or a limiting dynamic parameter, e.g. a peak particle velocity of 100 mm s"1. Wherever rock conditions demand bolt support within a specified 'safe' distance, the bolting is generally classified as temporary, because of lack of confidence concerning the potential breakdown of bond or bolt damage. In such circumstances costly duplication of bolting occurs as the permanent reinforcement advances behind the face, and the temporary bolts become redundant. The 660 m long Penmaenbach tunnel in North Wales was constructed by drilling and blasting through rhyolite, and as part of the primary support for the 12 m diameter tunnel, fully bonded twospeed resin bolts were installed and tensioned to 100 kN. A typical cross section of the tunnel is shown in Figure 26. The rhyolite, of Ordovician age, is slightly weathered, fine-grained, very strong material with narrow to wide fracturing (spacing typically 0.2 to > 0.5 m). The range of properties is listed in Table 11.
20°
I5e Spot bolting in zone B and F directed by the Engineer J
! Tunnel N Class I
Not ro scale
Figure 26 Typical bolt array related to rock class Table 11 Geotechnical Properties of the Rhyolite at Penmaenbach Tunnel in North Wales 'Intact' rock Bulk density Uniaxial compressive strength Point load index Static elastic modulus Dynamic elastic modulus Poisson's ratio Rock mass Static elastic modulus Dynamic elastic modulus
2.52-2.66 Mgm" 3 85-339 N mm - 2 2.3-10.2 N mm" 2 35-60 kN mm"2 68.5-75 kN mm"2 0.1-0.2 10-40 kN mm - 2 20-50 kN mm - 2
Table 12
Tunnel perimeter zone
Rock-bolt length (m)
Rock-bolt orientation
Rock-bolt fixed anchor length (m) Class I
A. (South side wall) B. (South side wall) C. (Crown)
D. (Crown) E. (Crown) F. (North side wall) G. (North side wall)
Support Requirements Related to Rock Class (a) Support requirements
Class II
Class III
Rock-bolt spacing (m) Class I
Class II
Class III
Additional isupport Class I
None
No rock bolts required 3.5
Horizontal
1.5
2.0
Spot bolting 2.0
1.5
3.5
Radial
1.5
2.0
2.5
1.0
7.0 3.5 3.5
Radial Radial Horizontal
2.0 1.5 1.5
3.0 2.0 2.0
No rock bolts required
Class II
1.5
2.5 1.5 2.5 1.5 Spot bolting 2.0
1.0 1.0 1.5
None None
None
25 mm Sprayed concrete None None
None
Class III 50 mm Sprayed concrete 50 mm Fabric reinforced sprayed concrete 50 mm Sprayed concrete
o«<* <3
<** Ξ © Ci
55 5J-
5
(b) Rock class definitions Class
I II
RQD
90-100% 60-90%
Discontinuity spacing
(CSIR) RMR range
(NGI) Q range
>0.5m 0.2-0.5 m
81-100 61-80
>20 20-4
è
448
Support 13% Increase
lifViÀ/VV^
iftnr"nlpil,r—ι„ί\,ί*ι1_|ιι,ι un
r-Vgn Β Γ \ , Α — a ^ a
nJ^IVv
8 % Decrease
20
30
40 Time (ms)
50
60
70
Figure 27 Load fluctuation with time for a single delay blast
The designers of the rock support at the Penmaenbach tunnel employed two well-established empirical methods, namely the NG1 tunneling quality index (Q) developed by Barton et al [37] and the CSIR rock mass rating (RMR) scheme proposed by Bieniawski [38]. Both methods were used to compare and check the recommended amount and type of permanent support required for stability. At Penmaenbach three classes of rock were established (Table 12), each with a standard form of support which involved combinations of spot bolting, patterned rock bolting, and sprayed concrete, with and without fabric reinforcement. In order to study the dynamic response of the support system, instantaneous and residual load changes together with axial accelerations were monitored on 6-m-long fully-bonded experimental bolts installed at distances of 20 m down to 0.7 m from the advancing tunnel face. Figure 27 shows dynamic load changes in a bolt post-tensioned to 100 kN and located 1.9 m from the tunnel face. The 76 ms activity was caused by a single delay which produced a maximum acceleration of 130 g (peak particle velocity of 345 mm s" 1 at 600 Hz). Overall, for accelerations ranging from 10 to 640 g, all bolt deformations were elastic and no significant residual load loss or resin/bolt debonding was registered, even when bolts were located within 1 m of the blast face. These results quantify the resilience of resin-bonded rock bolts subjected to close proximity blasting. The influence of prestress in reducing the dynamic loading on rock bolts has also been highlighted [39] and should encourage the use of tensioned bolts in circumstances where dynamic loading is anticipated. Attenuation relationships for peak particle velocity and peak dynamic load (expressed as a percentage of prestress load) have been established for the Penmaenbach tunnel, as follows: Peak Particle Velocity = 190 (x)~ 0615 Peak Dynamic Load = 4 ( x )
-0633
(6) (7)
where x = distance x (charge mass per delay)" 0,5 m k g - 0 · 5 . Such data, when collected for other tunnel projects, will provide a basis for a design predictive capacity which should lead to more cost-effective rock-bolt systems subjected to blasting. 15.9
FINAL REMARKS
Experience indicates that higher quality and more detailed ground investigations are required at the planning stage of many anchorage projects to permit their economic design and construction. In addition, there is a need to define clearly in contract documents the design responsibilities of the designer and specialist anchorage contractor in order to minimize contractual problems. In the field of permanent anchorages, corrosion protection ranges from double protection (implying two physical barriers) to simple cement grout cover. The latter solution is not considered acceptable when the safety of people and property in the event of anchorage failure is balanced against the cost of providing protection. The required degree of protection should be specified at the time of bid, and single protection should represent the minimum standard for permanent anchorages.
Overview of Rock Anchorages
449
Given the specialized nature of rock anchorage work and the wide variety of anchorage types and construction procedures, coupled with the variability of ground, more reliance should be placed on performance specifications related to choice of materials and acceptance testing of all anchorages, compared with control of construction. Such testing should involve proof loading to show a margin of safety, load-displacement analysis to confirm that the resistance to withdrawal is mobilized correctly in the fixed anchor zone, and short-term monitoring of the service behavior to ensure reliable performance in the long term. If reliable performances are to be maintained, a technical appraisal of anchorage systems is required by the practising engineer, in addition to routine comparisons on the basis of cost and duration of contract. Modern codes facilitate these technical appraisals but more importantly, the adoption of code recommendations should ensure both the safety and satisfactory performance of the anchorage system. Millions of anchorages have been installed successfully for temporary and permanent works throughout the world, the development in anchoring techniques having been dramatic over the past 40 years. With an absence of serious failures, there is a strong base upon which anchorage specialists can build and expand their market with confidence. There is no room for complacency however; engineers must rigorously apply high standards and much field development remains to be tackled.
ACKNOWLEDGEMENT Extracts from BS 8081:1989 are reproduced with the permission of the British Standards Institution. Complete copies can be obtained through national standard bodies. 15.10 REFERENCES 1. Thomas E. Stabilisation of rock by bolting. Reviews in Engineering Geology I & II, New York (1962). 2. Host'âk P. Zavesna Vystroj Suornikovâ a Laenovâ. UVR, Praha (1960). 3. Drouhin M. Consolidation du Barrage de Cheurfas par Tirants Métalliques Mis en Tenson. Annales des Ponts et Chaussées (Août, 1935). 4. Littlejohn G. S. The practical applications of ground anchorages. In Proc. 9th Federation International de la Précontrainte Congress, Stockholm (1982). 5. Littlejohn G. S. Soil anchors. In Proc. Conf. on Ground Engineering, pp. 33-44. Institution of Civil Engineers, London (1970). 6. Suzuki I., Hirakawa T., Morii K. and Kanenko K. Developpments nouveaux dans les foundations de pylons pour lignes des transports THT de Japon. In Int. Conf. des Grandes Réseaux Electriques à Haute Tension, Paper 21-01. (1972). 7. Hoek E. and Bray J. W. Rock Slope Engineering, 2nd edn. Institute of Mining and Metallurgy, London (1977). 8. Hoek E. and Brown E. T. Underground Excavation in Rock. Institute of Mining and Metallurgy, London (1981). 9. ISRM. Suggested Methods of Determining the Compressive Strength and Deformability of Rock Materials. ISRM, Lisbon (1978). 10. ISRM. Suggested Methods of Determining Tensile Strength of Rock Materials. ISRM, Lisbon (1977). 11. ISRM. Suggested Methods of Determining Water Content, Porosity, Density, Absorption and Related Properties and Swelling and Slake Durability Index Properties. ISRM, Lisbon (1978). 12. Barton N., Lien R. and Lunde J. Engineering classification of rock masses for the design of tunnel support. Rock Mech. 6, 189-236 (1974). 13. Farmer I. W. and Shelton P. D. Factors that affect underground rockbolt reinforcement systems design. Trans. Inst. Min. Metal 89 (A), A68-A83, A106 (1980). 14. US Army Corps of Engineers. Engineering and Design: Rock Reinforcement. Engineer Manual EMI 110-1-2907 (1980). 15. CIRIA. A Guide to the Use of Rock Reinforcement in Underground Excavations. Report 101, CIRIA, 6 Storey's Gate, London (1983). 16. Deere D. U. et al. Design of Tunnel Liners and Support Systems. US Dept. of Transportation and University of Illinois, Contract 3-0152 (1969). 17. Bieniawski Z. T. Geomechanics classification of rock masses and its application in tunnelling. In Proc. 3rd Congress, ISRM, Denver, Vol. IIA, pp. 27-32. (1974). 18. Littlejohn G. S. and Bruce D. A. Rock Anchors - State of the Art. Foundation Publications Ltd, Brentwood, Essex, (1977). 19. Bruce D. A. The design and performance of prestressed rock anchors with particular reference to load transfer mechanisms. PhD Thesis, Department of Engineering, University of Aberdeen, UK (1976). 20. Hobst L. and Zajic J. Anchoring in Rock and Soil. Elsevier, Amsterdam (1983). 21. Coates D. F. Rock Mechanics Principles. Dept. of Energy, Mines and Resources Mines Monograph No. 874, Ottawa (1970). 22. Standards Association of Australia. Ground anchorages. In Prestressed Concrete Code CA35, pp. 50-53. (1973). 23. Klein K. Draft standard for prestressed rock anchors. In Symposium on Rock Anchoring of Hydraulic Structures. Vir Dam 86-102 (1974). 24. Barley A. D. Ten thousand anchorages in rock. Ground Eng. 21 (6), 20-21, 23, 25-29; (7), 24-25, 27-35; (8), 35-37, 39 (1988).
450 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39.
Support Dennehy J. P. Correlating the SPT N value with chalk grade for some zones of the Upper Chalk. Geotechnique, 25, 610-614 (1975). Berardi G. Sul comportamento degli ancoraggi immersi in terreni diversi. Univ. of Genoa, Inst. Const. Sc. Series III, No. 60 (1967). Coates D. F. and Yu Y. S. Three dimensional stress distribution around a cylindrical hole and anchor. In Proc. 2nd Int. Conf. on Rock Mechanics, Belgrade, pp. 175-182. (1970). Barley A. D. A study and investigation of underreamed anchors, and associated load transfer mechanisms. M Sc Thesis, Department of Engineering, University of Aberdeen, UK (1978). Müller H. Erfahnungenmit Verankerungen system BBRV in fels und lockergesteinen. Schweizerische Bauzeitung, 84 (4), 77-82 (1966). Littlejohn G. S. Design of cement based grouts. In Proc. Geotechnical Engineering Speciality Conference, Grouting in Geotechnical Engineering, pp. 35-48. plus disc. 1 to 6. ASCE, New Orleans, LA (1982). Littlejohn G. S. and Truman-Da vies C. Ground anchors at Devonport Nuclear Complex. Ground Eng. 7 (6), 19-24 (1974). Fédération Internationale de la Précontrainte. Corrosion and Corrosion Protection of Prestressed Ground Anchorages. Thomas Telford, London (1986). Littlejohn G. S. Acceptable water flows for rock anchor grouting. Ground Eng., 8 (2), 46-48 (1975). Prestressed Concrete Institute. Recommendations for Prestressed Rock and Soil Anchors. Post-tensioning Institute, Phoenix, AZ (1986). Littlejohn G. S. Routine on-site acceptance tests for ground anchorages. Ground Eng., 24 (2), 37-43 (1991). Littlejohn G. S. and Bruce D. A. Long-term performance of high capacity rock anchors at Devonport. Ground Eng., 12 (7), 25-33 (1979). Barton N., Lien R. and Lunde J. Estimation of support requirements for underground excavations. In Proc. 16th U.S. Symp. Rock Mech., Minneapolis (Edited by C. Fairhurst and S. L. Crouch), pp. 163-177. ASCE, New York (1975). Bieniawski Z. J. Rock mass classification in rock engineering. In Proc. Symp. Exploration for Rock Engineering, Johannesburg, pp. 97-106. (1976). Littlejohn G. S., Rodger A. A., Mothersille D. K. V. and Holland D. C. Dynamic response of rock bolt systems. In Proc. Int. Conf. on Foundations and Tunnels, University of London, Vol. 2, pp. 57-64. (1989).
16 Rock Reinforcement-Technology, Testing, Design and Evaluation CHRISTOPHER R. WINDSOR and ALAN G. THOMPSON CSIRO, Nedlands, WA, Australia 16.1
INTRODUCTION
452
16.2 ROCK REINFORCEMENT TECHNIQUES 16.2.1 Terminology - Support or Reinforcement 16.2.2 The Load Transfer Concept for Reinforcement 16.2.3 Pre- and Post-excavation Reinforcement 16.2.4 Pre- and Post-tensioned Reinforcement 16.2.5 Permanent and Semipermanent Installations 16.3
452 452 453 454 454 457
REINFORCEMENT HARDWARE
457
16.3.1 Classification of Reinforcement Hardware 16.3.2 Continuous Mechanically Coupled Elements 16.3.3 Continuous Frictionally Coupled Elements 16.3.4 Discrete Mechanically and Frictionally Coupled Elements 16.3.5 Types of Grout 16.3.6 Auxiliary Fittings 16.3.7 Large-scale Reinforcing Elements 16.4 REINFORCEMENT ACTION 16.4.1 16.4.2 16.4.3 16.4.4
457 457 457 457 458 458 459 459 460 462 463 465
Types of Rock Behavior Axial Reinforcement Behavior Shear Response Modes Combined Response Mode
16.5 REINFORCEMENT TESTING PROCEDURES 16.5.1 Testing Requirements 16.5.2 Laboratory Axial Tension Testing 16.5.3 Laboratory Shear Testing 16.5.4 Field Axial Tension Testing 16.5.5 Field Shear Testing
465 465 466 467 468 468
16.6
468
REINFORCEMENT DESIGN
16.6.1 Continuous Rock Response 16.6.2 Discontinuous Rock Response 16.6.3 Reinforcement Interaction with Instability Mechanisms 16.6.4 Design Methods 16.6.4.1 Empirical methods 16.6.4.2 Analytical methods 16.6.4.3 Computational methods 16.6.4.4 Physical simulation methods
469 470 470 470 471 471 474 All
16.7 REINFORCEMENT PERFORMANCE EVALUATION 16.7.1 Philosophy 16.7.2 Performance-monitoring Instrumentation 16.7.3 Performance-monitoring Programs
479 479 480 480
16.8 SUMMARY AND CONCLUSIONS
481
16.9
482
REFERENCES
451
452
Support
16.1 INTRODUCTION Rock reinforcement is a specific technique within the general category of rock improvement methods. Rock improvement includes all techniques which seek to increase the strength or decrease the deformability characteristics of a rock mass. This includes methods such as injection of chemical or cementitious grouts, ground freezing, presplitting and drainage. In the case of rock reinforcement the prime objective is to improve the shear and tensile strength of the rock mass adjacent to surface and underground excavations. Further, reinforcement is often mistakenly taken to include all those methods which may be more properly called support. It is of benefit to the understanding of the mechanics of these techniques to distinguish between the concepts of support and reinforcement. Support is taken to include all methods which essentially provide surface restraint to the rock mass by installation of structural elements on the excavation boundary. By way of contrast, reinforcement is considered to include methods which modify the interior behavior of the rock mass by installation of structural elements within the rock mass. Support and reinforcement are essential components in the design of all surface and underground excavations and are often combined to provide an overall system. The techniques are required for both safety and productivity considerations in rock masses which are unstable in the required geometrical configurations. In the mining industry they enable more economical extraction of orebodies through steeper slopes in open pits or increased recovery rates from underground stopes. In the civil industry they enable increased excavation rates in tunnels and safer permanent surface and underground facilities. This chapter concentrates on reinforcing techniques as a priority compared with support techniques. Recent estimations suggest that some 500000000 reinforcing units are installed annually in the civil and mining industries throughout the world. It is suspected that a high proportion of these may not actually be achieving design expectations. This is due partly to practical difficulties associated with installation, partly to misconceptions and misinformation regarding the suitability of the many different devices and partly to the intractable nature of completing a formal engineering design. The solution to these problems lies in firstly understanding the terminology and fundamental behavior of the reinforcing elements. Secondly, a rational design approach that avoids both guesswork and unnecessary rigor is required. This approach comprises prediction of excavation response and selection of reinforcing element to suit the purpose of the excavation and the predicted instability mechanism. Finally, field assessment of installation quality and evaluation of the reinforcement scheme are required.
16.2 ROCK REINFORCEMENT TECHNIQUES 16.2.1 Terminology-Support or Reinforcement Aggressive mining strategies, design constraints on excavation geometry or unfavorable arrangements of rock stress, structure or material properties often lead to a requirement for some form of ground improvement technique. The various options for achieving this rock mass improvement in both surface and underground excavations are given in several key reference texts and conference proceedings [1-10]. The simple elegance of the rock reinforcement solution to this universal problem has led to a large number of apparently different reinforcing elements, auxiliary fittings, installation procedures and design philosophies. These have then been assembled in a variety of ways to address the different mechanisms of excavation instability. The penalty has been the generation of equally large amounts of terminology and jargon which often obscure the fundamental aspects needed for rational implementation of the technology. Reinforcement terminology includes words to describe the reinforcing elements, installation procedures and the philosophy behind the reinforcement scheme design. For example, some of this terminology includes: reinforcing elements (anchors, dowels, bolts, pins, nails, cables, tendons); installation procedures (pre- and post-reinforcement, pre- and post-tensioning, grouted and ungrouted, bonded and debonded, coupled and uncoupled, permanent and temporary reinforcement); and reinforcement scheme philosophy (strata reinforcement, rock support, cable doweling, rock anchoring, pattern reinforcement and spot bolting). A number of authors [11-14] have recognized this and attempted to define limited terms to describe the various reinforcement techniques. However, the terminology is still often mixed in all its variant forms. This has resulted in conflicting
Rock Reinforcement-Technology, Testing, Design and Evaluation
453
claims and misinformation for advantages and disadvantages when comparing the devices, the installation procedures and the reinforcement philosophies. To make matters worse these comparisons may even be made for distinctly different rock mass conditions and instability mechanisms. Consequently, considerable care is needed when interpreting the published information. A limited terminology is proposed which will simplify the selection of appropriate reinforcing devices, installation procedures and design philosophies for use in particular applications. The first task is to distinguish between support and reinforcement techniques. The terms support and reinforcement are often used interchangeably. However, it is useful to consider the two terms as being explicitly different due to the method by which they stabilize the rock adjacent to an excavation. Essentially, support is the application of a reactive force at the face of the excavation. Techniques and devices which may be considered as part of rock support includefill,timber, steel or precast concrete sets, shotcrete and props. In contrast with support, reinforcement is considered to be improvement of the overall rock mass properties from within the rock mass and will therefore include all devices installed in boreholes. 16.2.2 The Load Transfer Concept for Reinforcement The fundamental aspect in understanding reinforcement behavior and the action of the different devices and their effects on excavation stability is the load transfer concept. This concept is shown schematically in Figure 1 and can be visualized as being composed of three basic mechanisms. (i) Rock movement which requires load transfer from the unstable rock to the reinforcing element. (ii) Transfer of load via the reinforcing element from the unstable surface region to a stable interior region. (iii) Transfer of the reinforcing element load to the stable rock mass. The lengths of reinforcing elements have evolved to approximately three ranges which reflect the depth of the perceived unstable surface regions associated with different surface and underground excavations. These ranges are termed near surface (1.5-3 m long), medium depth (3-15 m long) and deep seated (> 15 m long). There are a wide variety of methods by which the load transfer between the rock and reinforcing element may be achieved and many reinforcing devices have been developed. However, the load transfer mechanisms for all these devices can be placed within one of the three categories defined in Figure 2. These categories are (a) continuous mechanically coupled (CMC), (b) continuous frictionally coupled (CFC) and (c) discrete mechanically and frictionally coupled (DM FC). The various types
(a)
Stable zone
II II II Discontinuity
^■\νχ
IZN
Unstable zone
ll
IL
^y/xv
I
^fr 4^ Surface
(b)
I hardware
.,-.,..^..::^V*...;:...,J, II II
-'s^V
- ' ·
[P
i
Figure 1 The reinforcement load transfer concept: (a) discrete load transfer and (b) continuous load transfer
454
Support Type
Cross section
Longitudinal view
(α)
(b)
(c)
B-B Figure 2 Categories of reinforcing element load transfer: (a) CMC, (b) CFC and (c) DMFC
of load transfer mechanisms are described in detail in Section 16.3. Some of the more common reinforcing devices are classified in Table 1. There are also factors related to installation which can optimize the load transfer and the performance of the reinforcing element in response to rock mass behavior. These include the timing of installation and the provision of an initial tension in the reinforcement and the procedures for semipermanent or permanent excavations. The options for a reinforcement category to achieve the desired installation conditions are summarized in Table 2.
16.2.3 Pre- and Post-excavation Reinforcement In many applications it has been found that there are substantial benefits in safety and productivity associated with prereinforcement of excavations [15-17]. Prereinforcement can prevent premature failure of the rock and provides safer working conditions for the installation of further reinforcement or support. Some advantages in overall reinforcement requirements are sometimes possible through postreinforcement or reinforcement at an appropriate time after the creation of the excavation. Some examples of the application of both pre- and post-reinforcement are shown in Figures 3,4 and 5 for slopes, tunnels and large underground excavations, respectively.
16.2.4 Pre- and Post-tensioned Reinforcement In some applications, it is desirable to provide the reinforcing element with an initial pretension. Posttensioning is the tensioning or retensioning of devices subsequent to installation. Further
Rock Reinforcement-Technology, Testing, Design and Evaluation
455
Table 1 Classification of Reinforcement Types Basic type
Subset
Description
Reference [22] [18]
Wooden dowel Plain bar Deformed bar Thread bar Pigtail bolt Paddle bolt Yielding bolt Perfobolt Sigbolt Fibreglass bolt Injection polymer bolt Birdcage strand
[24] 15], [10] [25] [26] [27] [28]
Long cement encapsulation
Multiwire cable Prestressing strand Destranded hoist rope Shear key
[15], [16] [29] [43] [30]
Continuous frictionally coupled
Friction
Split Set GD Rock Nail Swellex bolt Wedge-Pipe bolt Ramp bolt Pipe anchor
[31]
Discrete mechanically and frictionally coupled
Friction
Slot and wedge anchor Expansion shell anchor Plastic expansion anchor Swellex
[35] [35] [36] [32]
Resin encapsulation
Plain bar Deformed bar Thread bar Pigtail bolt Paddle bolt Fiberglass bolt Tube anchor Long tendons
[18]
Continuous mechanically coupled
Short cement/resin encapsulation
[23]
[32] [33] [5] [34]
[23] [27] [37] [29]
Table 2 Installation Options for Reinforcement Support or reinforcement
Timing of installation
Provision for tensioning
Category of reinforcement -CMC
-Untensioned -
-CFC -DMFC
- Prereinforcement -
-CMC -Tensioned-DMFC
Reinforcement-
-CMC -Tensioned-DMFC
-Postreinforcement -
-CMC - Untensioned-
-CFC -DMFC
CMC, continuous mechanically coupled. CFC, continuous frictionally coupled. DMFC, discrete mechanically and frictionally coupled.
456
Support
Radial postreinforcement
Radial pre reinforcement
Subparallel prereinforcement (shear keys)
Figure 3 Pre- and post-reinforcement in slopes Radial postreinforcement
Subparallel prereinforcement (spiling)
Figure 4 Pre- and post-reinforcement in tunnels
Radial prereinforcement
Subparallel prereinforcement Figure 5 Prereinforcement of large underground excavations
tension may develop with time as the rock mass moves due to subsequent excavation activity, stress changes or creep. This possibility must be explored and allowed for to avoid subsequent overstressing and rupture. For example, in some specific applications associated with rockburst prone conditions it has been found desirable to limit the amount of load transfer to the reinforcing element [19].
Rock Reinforcement - Technology, Testing, Design and Evaluation
457
16.2.5 Permanent and Semipermanent Installations The purpose and service life of an excavation dictate the required quality of reinforcement installation (e.g. [20, 21]). The requirements of support and reinforcement for mining and civil applications are often different. Mining applications generally require the excavations to be stable only for the time required to extract the ore in a localized area of the orebody. This period will vary from a few months to a few years and will be different for service and production excavations. Civil applications generally require a long term stability. The service life of the reinforcement must be defined prior to installation such that the correct device is chosen and any special procedures and quality assurance during installation can be performed. It is important to recognize that many reinforcing devices have been designed and developed for rapid installation and have only a limited service life. 16.3 REINFORCEMENT HARDWARE 16.3.1 Classification of Reinforcement Hardware A large number of devices have been developed and proposed for use as reinforcing elements. In general, these devices are solid bar or hollow tube elements that are installed within boreholes drilled into the rock. To assist with the comparison of performance and help evaluate new devices, a classification system is proposed in which there are only three basic types of devices. The classification scheme may appear an oversimplification but is in fact appropriate if the basic mechanisms of load transfer between the reinforcing element and the rock mass are to be understood. A selection of commonly used reinforcing devices have been placed within one of the three categories in Table 1. 16.3.2 Continuous Mechanically Coupled Elements Continuous mechanically coupled elements rely on a securing agent whichfillsthe annulus between the element and the borehole wall. This agent, known generally as grout, is usually placed in a fluid condition which requires somefinitesetting period before the element can begin service. This period varies between a few seconds and a number of hours depending on the type of grout used and this may well affect selection of the device. The major function of the grout is to provide a mechanism for load transfer between the rock and the reinforcing element. Reinforcing elements used in conjunction with grouts are often manufactured with variable cross sectional shapes. This variation causes a geometrical interference between the element and the grout and creates a mechanical key. When the geometrical interference extends over the length of the element, it is coupled continuously to the rock mass by way of the grout. 16.3.3 Continuous Frictionally Coupled Elements Continuous frictionally coupled elements behave somewhat similarly to continuous mechanically coupled elements. However, the reinforcing element is placed in direct contact with the rock. Load transfer results from friction between the element and the borehole and is limited by the radial prestress set up during installation. Any geometrical key that may be present occurs by chance due to borehole irregularities. This may be advantageous in some applications. Devices in this class consist of either expansion of an undersized section into a larger borehole (e.g. Swellex [32]) or contraction of an oversized section into a smaller borehole (e.g. Split Set [31]). Placement generally requires deforming the cross section of the element to suit the borehole. In some cases the size of the borehole is important and may be critical in terms of installation and performance. 16.3.4 Discrete Mechanically and Frictionally Coupled Elements These elements tend to be simple bars with either a deformed end region or provision at one end for attaching an expanding anchor. They provide either mechanical or frictional load transfer over this relatively short interval of their total element length. This interval is commonly known as the
458
Support
anchorage length and is usually limited to less than 500 mm for grouted anchorages and less than 200 mm for expansion anchorages. The anchorage must be sufficiently strong to mobilize the full material strength of the reinforcing element. Two examples of discrete frictionally coupled devices are the wedge bolt and expansion shell bolt [35]. Both these elements have anchorage assemblies that have a component of geometrical interference with the borehole and a component of frictional interaction. The expansion shell bolt is the most widely used. The mechanics of this device are basically to expand the anchor outwards against the borehole wall due to tension induced in the element during installation and in service. The strength of the anchorage may be limited by the strength of the rock and these devices are best suited to hard rock applications. The grouted anchorages have an advantage in a lower unit load transfer compared with the expansion shell or wedge bolt anchorages. This may be essential in softer rocks where transfer of high loads over a short length of borehole may initiate failure at the rock interface. It has been found that the major function of a grouted anchorage is to provide a mechanical keying effect between the bolt and the rock surface. There is in most cases very little if any adhesion between the grout and the rock and the grout and the bolt. In general, only resinous grouts can meet the high strength requirements for short anchorages. 16.3.5 Types of Grout Grouting materials can be broadly classified as being made of cementitious or resinous materials, with the latter being generally stronger. Cement grouts are suited to the longer reinforcing elements due to availability, ease of mixing and placement. Resinous materials are generally confined to the shorter elements due to the higher costs of the resin materials and placement difficulties in longer holes. A benefit of grouting is an improvement in resistance to corrosion which is common to all steel reinforcing devices. A number of additives can be used to aid the placement and improve the quality of cement grout and inhibit corrosion (e.g. [38, 39]). Resin-grouted reinforcement has found wide application in discrete load transfer elements where a two-part resin mixture is contained within cartridges which are broken and mixed during installation (e.g. [40]). Mix and set times can be reduced to less than 10 seconds for fast and productive installation. This is particularly important in cyclic and fast advance mining methods where the reinforcement installation process may be the critical factor in determining the rate of advance. In addition, the anchors can supply design capacity within a few minutes after installation which improves safety during mining advance. It is also worth noting that high early strength cement grouts are also available in cartridges [41]. 16.3.6 Auxiliary Fittings There are various forms of auxiliaryfittingsthat have evolved for use with the different reinforcing elements to cater for different rock mass conditions. Thefittingsare either attached to the reinforcing element at the rock mass boundary or used to modify the internal response of continuously coupled devices. Table 3 provides a summary of fittings which have been used to enhance reinforcement performance. Special equipment and installation procedures may be required for tensioned installations. External fittings can be attached to the reinforcing element to provide varying degrees of surface restraint to the rock mass surface. These support-typefittingsare usually in the form of plates, straps and mesh. External fittings are essential for the discrete coupled devices. Internal fittings are used predominantly with continuous mechanically coupled devices and comprise intermittent anchors (barrel and wedge anchors or swaged ferrules) and decoupling sleeves [44]. The combination of anchors and decoupling modifies the load transfer mechanism between the rock and the reinforcing element. These modifications are often necessary to manipulate the stiffness of the element to suit the expected deformation of the predicted rock mass failure mechanism. In most applications, the number of reinforcing elements in relation to the exposed excavation area is quite low. In closely jointed,fissileor weathered rock masses where unraveling may occur, stability may be improved by installing structural elements which span between the reinforcing elements. These supports may be arranged to form a diaphragm that provides a reactive stress to the free span area and promotes interlock of small blocks. This stress is usually very low for the straps and mesh used in mining excavations. However, this form of diaphragm has proved successful in
Rock Reinforcement-Technology, Testing, Design and Evaluation Table 3
Auxiliary Reinforcement Fittings
Fitting
Type
Reference
Mesh
Chain mesh Weld mesh Cable net
[10] [10] [42]
Surface plates and straps
Flat plate Domed plate Flat washer Tapered washer Spherical washer Domed washer (Brown) Butterfly plate Flat strap W-strap Cable sling
Anchors
459
Surface Internal
Spherical nut Barrel and wedge Barrel and wedge Swaged steel Swaged aluminum
[43] [10] [44] [44], [45] [44], [45] [43], [45]
preventing unraveling of small blocks. The concept of diaphragm support, but on a larger scale using reinforcement as anchors is very important and holds the key to supporting many of the larger, more complex failure mechanisms. This is especially true for surface excavations.
16.3.7 Large-scale Reinforcing Elements The large-scale reinforcing elements are generally greater than 15 m long. These long elements tend to be arranged in larger cross sectional areas to handle the greater volumes of unstable material that require reinforcement. They still conform with the proposed classification scheme but have been subdivided here into their primary modes of action. The high axial capacity elements are sometimes called ground anchors. They are used extensively in civil engineering and not so much in mining. They tend to be composed of large numbers of individual elements that act together as a composite reinforcing element. They are usually discretely coupled over a fairly long anchorage to enable prestressing and may be used to impart considerable load to the surface of the rock mass by way of large, rigid, built-in stressing blocks situated at a free surface. Multiple installations are often used in a uniform pattern in conjunction with a built-in rigid diaphragm consisting of a network of reinforced concrete members. These devices often play a critical role in maintaining stability and they are therefore subject to stringent installation quality assurance by proof-testing programs [21]. The high shear capacity elements are sometimes called shear keys. These may be in the form of universal steel sections, large diameter steel tubes [30] or railway lines cast in concrete to continuously couple them to the rock mass. These elements are mostly used as prereinforcement in surface excavations. They are often installed as a vertical palisade quite close to and approximately subparallel to the proposed boundary of the excavation. The displacements on discontinuities are directed primarily towards the excavation and the large cross sectional areas of the elements provide a high shear resistance to these transverse displacements.
16.4 REINFORCEMENT ACTION The interaction between the rock mass and reinforcing elements is very complex. This is due to the variable nature and the complicated failure mechanisms of rock masses and the mechanics of load transfer between the reinforcing element and the borehole. The mechanical behaviors of the three distinct classes of reinforcing element undergoing pure axial, shear and combined axial and shear displacements may all be studied in terms of the load transfer concept of reinforcement.
460
Support
16.4.1 Types of Rock Behavior Reinforcement is a ground improvement option where excavations are found or predicted to be unstable. Instability is usually synonymous with the formation of new fractures or the reactivation of existing discontinuities. By previous definition, the general mode of reinforcement behavior is one of modifying the internal strength and deformation characteristics of a rock mass. Thus reinforcement action is inextricably linked to the interaction of reinforcing elements with rock mass discontinuities. In massive rock, reinforcement acts following fracture formation, propagation and consequent displacement on these fractures. In stratified and jointed rock it is linked with displacement at preexisting discontinuities or displacement at any new discontinuities which may form. Thus, reinforcement either actively seeks to prevent the formation and propagation of new discontinuities or responds to the displacements that may occur during instability at preexisting discontinuities. This formal definition finds expression in many of the commonly used descriptions to illustrate reinforcement action within unstable mechanisms. Figure 6 shows a number of assumed reinforced rock mass responses associated with underground excavations. These include (a) suspension, (b) beam building, (c) arching and (d) keying. (α)
(b)
(c)
(d)
Figure 6 Types of rock behavior (a) suspension, (b) beam building, (c) arching and (d) keying
Rock Reinforcement-Technology, Testing, Design and Evaluation
461
Regardless of the level of appropriateness of these rather simple descriptions of reinforcement behavior, the above definition still holds. These descriptions also serve to illustrate the important role that the structural nature of the rock mass plays in mechanism formation and consequently the purpose of reinforcement designs that seek to arrest these mechanisms. In fact, many reinforcement schemes only operate when the rock mass attempts to fail. Thus the modes of displacement at a discontinuity define the modes of action of reinforcing elements crossing that discontinuity. These modes of displacement are given in Table 4 and these aspects are investigated in greater detail in the following sections. In practice the displacements at discontinuities are rarely as simple as suggested in Table 4. They tend to be combined in a truly three-dimensional manner which may consist of translation and rotation. The displacement may also be path dependent and could comprise multiple steps of these combined components, including reversals in direction. This is further complicated by the orientation of the reinforcing element in relation to the discontinuity and the direction of displacement on the discontinuity. However, even with these complications, the response of the reinforcing element may still be simply resolved into three prominent modes: pure axial, pure shear and combined axial and shear response. For simplicity, these modes are given in two-dimensional form for a dilating discontinuity and a shearing discontinuity in Figures 7 and 8. In both cases the reinforced block attempts to detach itself from the rock mass. The resulting response mode of each reinforcing element is controlled by the displacement vector of the block and the orientation relationship between the element and the discontinuity that it reinforces. The mechanical interaction of discontinuity displacement and reinforcement response in these modes for a single discontinuity is very complex. The response for this interaction cannot be solved simply due to a number of nonlinearities and path dependencies in both discontinuity behavior and reinforcement behavior. This is explored in more detail in Section 16.6 which includes the effects of multiple discontinuities and reinforcing elements. For simplicity, the problem is restricted at this stage to a single reinforcing element and one discontinuity.
Table 4 Discontinuity Displacement Modes and Reinforcement Response Discontinuity displacement mode
Reinforcing element mode
Dilational displacement Shear displacement Rotational displacement
Pure axial response Pure shear response Axial and bending response
Figure 7 Reinforcement action at a dilating discontinuity
462
Support
Figure 8 Reinforcement action at a shearing discontinuity
16.4.2 Axial Reinforcement Behavior For reinforced fiber composites, reinforced concrete and reinforced rock there are three basic load transfer mechanisms that operate in transferring load from the reinforcing element to the matrix or vice versa: adhesion, friction and mechanical interlock. These three factors have often been lumped into a single quantity called 'bond' and extensive work has been performed in studying bond in other disciplines [46]. However, in rock engineering it is preferable to examine the behavior of reinforcing systems with respect to each of the three different factors. The adhesion component is only relevant to the continuously and discretely coupled devices that use bonding agents to secure them into the borehole. However, there are two aspects which suggest that adhesion is of little significance in the overall determination of reinforcement performance in rock applications. Firstly, it relies on having surfaces which are suitable for bonding and there are a number of operational difficulties associated with maintaining the clean surface conditions required for optimum adhesion. Secondly, simple analysis [47] suggests that the axially directed shear stresses induced near the reinforcement/grout interface at low load levels quickly exceed the shear strength of the grout. Therefore, even if significant adhesive bond between the grout and reinforcement existed, failure would occur preferentially in the grout. The mechanical interlock component is relevant to all the classes of reinforcement. Mechanical interlock describes the keying effect caused by having a reinforcement surface profile that keys into the rock in the case of frictionally coupled devices or keys into the grout in the case of mechanically coupled devices. This geometrical interlock is provided by the borehole surface irregularities (i.e. roughness and rifling) in the frictional devices and by the reinforcing element surface geometry (i.e. deformed ribs, helical grooves or threads) in the mechanically coupled devices. This geometrical interlock ensures that material failure must occur rather than having a simple sliding mechanism occur at the interface. At this point it becomes evident that the major controlling factors are the rock strength for steel frictionally coupled devices and the grout strength for steel mechanically coupled devices. Following loading in excess of the weakest material strength, the only component of 'bond' remaining is the friction between the two failed interfaces. The frictional component is relevant to all classes of reinforcing elements and is probably the dominant component after a small initial displacement has taken place. When the reinforcement is loaded, very high shear stresses are generated at the reinforcement/grout or the reinforcement/rock interfaces. The maximum strength of the rock or grout is exceeded at relatively small displacements. Therefore, at reasonable design loads, some failure of the intact material at the interface must occur. With this concept, interface material strength and friction become the most important parameters in determining the minimum anchorage length to achieve design load. Since friction has now been deemed to be an important factor it remains to examine what factors are involved in friction. Accordingly, it is found that the axial transfer must depend on the coefficient of friction and the radial stress existing at the failure interface. Factors which may affect the coefficient of friction include the microroughness of the reinforcing element surface and the particle size of the grout. The roughness of the surface will change depending on the extent of corrosion. Light rusting will cause surface pitting which will increase the coefficient
Rock Reinforcement-Technology,
Testing, Design and Evaluation
463
of friction between the grout and the steel [48], but excessive corrosion will create a layer between the grout and the steel which will then have a detrimental effect on frictional strength. A design specification should call for clean reinforcement surface conditions, such that light rusting' cannot be interpreted in the field as either severely rusted or dirty. Factors which affect the level of radial stress include the installation process (i.e. curing of grouts, interference stress established in frictionally coupled devices), dilation/contraction/removal of material during interface shear failure and radial contraction of the reinforcing element under load due to the Poisson effect. In the case of continuous mechanically coupled grouted devices the question of the radial stress level following curing and its variation during shearing is currently unanswerable. However, the consequences of the other two factors can be inferred. If shearing of the grout is accompanied by dilation, then radial stress must increase. Anything which causes a reduction of the volume inside the grout annulus will cause a reduction in radial stress. Thus removal of material as shearing takes place or radial contraction (Poisson's effect) of the reinforcement with increased axial load will lead to a reduction of the radial stress. The latter effect will be particularly pronounced following the yield of say a multiple-wire cable element. In the case of continuous frictionally coupled devices the establishment of a radial stress at the interface of the reinforcing element and the borehole is the critical part of the element design. In the case of the Split Set, for example, the hole diameter will be crucial in establishing the initial level of radial stress at the interface.
16.4.3 Shear Response Modes Reinforcing systems are traditionally designed around their axial strength and stiffness. However, the shear performance is at least if not more important from a practical point of view. The mechanics of shear of the reinforcement must include all the factors mentioned for axial performance in addition to factors such as crushing of the rock due to bearing stresses and bending of the reinforcing element. In the case of continuous mechanically coupled devices crushing of the grout must also be included. A discretely coupled reinforcing element which intersects a discontinuity undergoing shear will initially provide little resistance other than the resolved component of axial load and stiffness. The larger part of the total shear resistance will only occur when sufficient displacement has occurred to cause the device to be tightly jammed between opposite sides of the borehole wall either side of the discontinuity. Consequently, these devices have a low initial shear stiffness and this may not be desirable because significant rock mass displacements may occur before some resistance is supplied by the reinforcement. In contrast, a continuously coupled reinforcing element which intersects a discontinuity undergoing shear will provide immediate resistance to shear movement. However, the shear stiffness and the peak shear capacity will depend on the cross sectional shape of the element and the strength and amount of reinforcing material within the hole. The continuous frictionally coupled devices are generally hollow in cross sectional shape and the method of installation usually requires that their geometry or material strength be arranged for installation purposes. Their behavior in shear is quite complex with the hollow cross sections becoming severely distorted. This makes any rational analyses to support physical tests very difficult. The basic mechanism is one of guillotining (pure shear) compared with the softer response of the continuous mechanically coupled devices. With the continuous mechanically coupled devices it is possible to make a number of simplifying assumptions with respect to their behavior in shear. A physical representation of a deformed reinforcing element at a discontinuity is shown in Figure 9. The important parameters that affect behavior in shear are: (i) the axial properties of the reinforcing element material; (ii) the bending properties of the reinforcing element material; (iii) the axial load transfer between reinforcing element and grout; (iv) the properties of grout/rock in response to crushing. Typical forms of the force and deformation responses for the various components of interaction between the reinforcing element and grout or rock are shown in Figure 10. Attempts have been made to solve this problem using closed form techniques (e.g. [49, 50]). However, the material and geometric nonlinearities mean that the problem must be solved using computational techniques. Some researchers have looked at fundamental theories to attempt to quantify the grout/rock crushing behavior (e.g. [51, 54]). It is also possible to design simple laboratory tests which provide
464
Support
Dilation
Figure 9 Physical representation of reinforcement behavior
(a)
Axial strain
Curvature
(d)
Lateral displacement
Displacement
Figure 10 Force-deformation responses for element P-axial force; M-bending force; T-friction force and N-bearing force shown in Figure 9
Rock Reinforcement-Technology, Testing, Design and Evaluation
465
important insights into the behavior one may expect in practice, especially for grout, and these are mentioned in Section 16.5.3. With this model and data for the forms of interaction it is possible to predict the reinforcement response to various components of discontinuity shear and dilation. This reinforcement response model is required for the computational techniques described in Section 16.6.7. Continuous mechanically coupled devices are usually installed in relatively large diameter boreholes to allow room for inclusion of the grout. The addition of the grout annulus tends to have a cushioning effect, allowing the reinforcing element to deform across the discontinuity. Consequently, these devices are initially 'soft' in shear during crushing and densification of the grout, during which the angle of the reinforcing element changes. When this occurs the element starts to act somewhat in tension rather than direct shear and the concept of combined response mode must be considered. 16.4.4 Combined Response Mode Very few reinforcing elements are subjected to pure shear or axial loading alone. The majority of reinforcing elements are loaded to produce combinations of axial and shear loadings. This may occur either as a consequence of the orientation of the reinforcing element in relation to the discontinuity and the displacement vector or, as we have seen for continuous mechanically coupled devices, by geometry changes during a pure displacement mode.
16.5 REINFORCEMENT-TESTING PROCEDURES 16.5.1 Testing Requirements The basic aim of testing is to determine the performance of the reinforcing element when it is installed in the rock mass. The testing is required to determine the ultimate strength of the installed reinforcing unit, the stiffness of its response to rock mass induced loading and to indicate the mechanics of load transfer between the reinforcing element and the rock mass. These factors need to% be quantified for the design of appropriate reinforcing schemes for various rock mass conditions. Testing can be conveniently divided between tests in the field and those in the laboratory. Field tests are used to quantify the in situ performance under specific loading conditions. A laboratory test in some cases may be used to achieve the same aim. However, a laboratory test may also be used to study the basic mechanics of reinforcement behavior. The knowledge of the basic parameters which affect the reinforcement behavior can be used to modify and enhance performance. Field and laboratory tests may be used to load and measure the response of reinforcing elements for a very limited set of geometrical and loading conditions. To obtain the maximum benefit from field and laboratory testing, some complementary fundamental analysis can be used to extend the results to the multitude of other conditions which cannot all be tested (e.g. [51]). These analysis techniques form the basis for developing appropriate models for use with the computational techniques such as the finite element and distinct element methods. Therefore, to obtain the maximum benefit from an assessment program, all three methods (field, laboratory and analytical) of defining reinforcement performance should be considered. A carefully instrumented in situ field exercise in which reinforcement and rock mass discontinuities are monitored is complementary to the three established methods of defining reinforcement performance. Three basic reinforcement categories have been defined. These different categories can be evaluated using similar testing arrangements. However, some devices are more amenable to laboratory evaluation than others. Other devices must be tested in thefieldto evaluate performance in particular rock types. It is convenient to classify testing of reinforcement into tests which produce either axial or shear loading. A number of practical constraints generally restrict shear testing to the laboratory. A field 'shear' test is possible but difficult and cost would preclude the execution of a significant number of tests to define behavior for different conditions. In practice, the tests and analysis are performed in the order field, laboratory and analysis. However, it is essential to know the fundamental mechanics of reinforcement behavior to define the testing program and to critically appraise the field or laboratory test results. The fundamental considerations define the types of tests to be conducted that will provide the most information. This is especially important for continuously coupled devices where the aim is to vary the embedment length to produce both slip and element rupture failure modes. The limitations imposed
466
Support
by the nature of field testing techniques dictate that these form a small subset of all the tests which may be performed in the laboratory. It is therefore appropriate to describe the types of laboratory tests which can be performed prior to describing the limited number of field tests. 16.5.2 Laboratory Axial Tension Testing In the past, field tests and trials were often the first step in the assessment and the commercialization of new devices. More recently, laboratory tests have been used to conduct the preliminary evaluation in recognition of the ease of doing these tests and the reliability of the information that is obtained. This perception of laboratory tests has evolved with the development of better laboratory testing techniques which are now more representative of field conditions than previously. However, practical results for application in design will usually be restricted to grouted systems where failure is initiated and propagated at the element/grout interface or in the grout. Early axial testing of reinforcement devices involved the reproduction of the field loading conditions in which one end of the reinforcing element was gripped by the jaws of the testing machine whilst the other end of the element was encapsulated within a steel tube, as shown in Figure 11(a). This type of test, called a single embedment axial tension test, had a number of problems. Failure was often initiated at the jaws of the testing machine. To assess the stiffness of response at a rock discontinuity, a correction was required for the free length of the element between the end of the encapsulation and the point at which the displacement was monitored. It has been found that a better method for the measurement of axial performance is the double embedment axial tension test [52], shown in Figure 11(b). In this type of test, the response of the reinforcing element at the interface between the two halves of the test specimen more closely represents the performance of a similar reinforcing element crossing a dilating discontinuity. The aims of this test are to determine the variation of strength and stiffness with change of embedment length. A typical result will be as shown in Figure 12. A variety of loading rates may be used to evaluate reinforcement for conditions appropriate to field conditions such as dynamic or long term constant loading [53]. These tests represent the reinforcement load conditions experienced during a rockburst or for an element with a block suspended from it, respectively. The latter type of test would identify any factors related to creep of the materials present in the reinforcement system. A load relaxation test, in which the load in a tensioned element is monitored, may also be used to evaluate creep. In addition to performing tests to determine overall reinforcement response, there are a number of very specific tests which may be performed which quantify the load transfer mechanisms for different reinforcing elements and grout materials. These usually involve testing of short lengths of reinforcing elements to quantify the material characteristics required for models of reinforcement axial load transfer and shear action as detailed in Section 16.4.3 and as defined in Figure 10.
Figure 11 (a) Single and (b) double embedment length laboratory tests
Rock Reinforcement- Technology, Testing, Design and Evaluation
467
Rupture of element
Increasing embedment length
Axial displacement Figure 12 Effect of embedment length for a continuously coupled reinforcing element
Shear displacement Figure 13 Typical variation of reinforcing element response in shear as the angle to a discontinuity is changed
16.5.3 Laboratory Shear Testing Shear tests are more difficult to perform than axial tests. However, once equipment has been established to enable shear of a reinforcing element, it is then possible to do a wide variety of tests which can provide a large amount of detailed information on the reinforcing effects of an element crossing a discontinuity. The variation in response as the angle of the element is varied relative to the plane of the shear surface is shown in Figure 13. The difference in reinforcing response is most marked for cases in which the initial displacement produces a compressive response in the reinforcing element. The earliest tests were block shear tests [54-56]. These tests provided useful information on the overall changes in behavior of the discontinuity caused by the presence of the reinforcing element. However, to evaluate the performance of the reinforcing element, they did require the overall response to be modified by an assumed unreinforced response for the discontinuity. The consistency of frictional response for sliding surfaces is a discipline in itself and subsequent investigations have shown that the reinforcement causes a number of modifications to the discontinuity behavior additional to its own response. An improved concept was developed by the CSIRO Rock Reinforcement Group [57] in which the response of the reinforcing element was isolated from the complicating issues of frictional sliding of rough interfaces. The device developed is shown schematically in Figure 14. The device has low frictional resistance needle roller bearings to provide a shear free from dilation. The resistance to shear is due to the reinforcing element coupled with a small amount of rolling friction. The dilational and contractional forces induced by the reinforcement are monitored by load cells. The distribution of these loads provides indications of the rotational effects produced by the reinforcing element during shearing.
468
Support
Figure 14 CSIRO laboratory shear testing device
16.5.4 Field Axial Tension Testing Axial tension field tests are used commonly to evaluate reinforcing systems for a limited set of conditions. The tests are of a single embedment configuration and, as pointed out for laboratory tests, the results need to be corrected for the free length between the loading point and the end of the encapsulation to determine the performance of the reinforcing element at the position where it crosses a discontinuity. The ISRM [58] has made recommendations for procedures to be adopted for afieldtest program for reinforcing devices tested in this manner. Figure 15 shows a typical field test in which both load and displacement are monitored electronically. There are a number of other field tests which may be required to quantify reinforcement performance for special conditions. For example, for tensioned installations, it is necessary to perform tests which can monitor the tension induced in the reinforcement during the stressing process and the residual stress following release of the applied load. Periodic checks with load cells are required to ensure that the stressing equipment characteristics do not deteriorate with time and use. Tensioned installations can also be monitored for load relaxation behavior to complement laboratory measurements. The field test will include all the effects on the reinforcing system caused by environmental factors such as groundwater and temperature. 16.5.5 Field Shear Testing It is possible to perform a block shear test of a reinforced natural discontinuity. However, as mentioned in Section 16.5.3 on laboratory shear testing, there are a number of practical difficulties with regard to the discontinuity properties which would make the test result unique and difficult to generalize for design purposes. In addition, the test would be expensive to set up and would only provide a single result. 16.6 REINFORCEMENT DESIGN Analysis and design are relatively straightforward in the engineering disciplines that deal with manufactured materials such as steel and concrete. When dealing with rock excavations, simplifications and idealizations often have to be introduced into the description of the problem such that an analysis or a design can be attempted. The difficulty in predicting, and sometimes understanding, the behavior of a rock mass is due to a number of complexities inherent in rock mechanics. Three complexities of rock masses have been defined by Bray [59] as being associated with the material, structure and analysis. However, two simplifying assumptions can be made to ease the design process; one concerns the structural nature of the rock mass and the second its response mode. The structural nature of the rock mass surrounding a surface or underground excavation can
Rock Reinforcement- Technology, Testing, Design and Evaluation
469
Figure 15 A field axial tension test
be broadly classified as being massive, stratified or jointed. The structural nature of the rock mass often has a marked effect on its behavior and will impinge on all stages and decisions in the excavation design process. For example, the chosen category will govern whether a particular design or analysis method is appropriate and will also dictate factors related to the type of reinforcement chosen, its length, capacity and the appropriate timing of installation. The rock mass behavior can be broadly classified into either continuous or discontinuous response. 16.6.1 Continuous Rock Response Continuous response may occur in all the rock mass categories before fracture initiation through intact material or failure along preexisting discontinuities. There is continuity of normal and shear stress on any given plane through the rock mass and the stress and strainfieldsmay be described by continuous mathematical expressions. The most common approach is to assume that the rock mass is homogeneous and isotropic and there is a linear elastic constitutive relationship between stress and strain for the rock model material. This approach has proven to be most useful in predicting zones of stress concentration and stress relaxation near excavations (Brown [60]).
470
Support
Excavations that respond in a continuous manner are quite easy to deal with from a design and analysis point of view and most importantly they do not usually require reinforcement to maintain stability. In fact it is difficult to reconcile what value reinforcement would have under these circumstances. 16.6.2 Discontinuous Rock Response Discontinuous response occurs in stratified or jointed rock along preexisting discontinuities and in massive rock due to the creation and propagation of fractures caused by stress changes during the excavation process. As displacement proceeds, stress is redistributed away from the discontinuities to more competent regions of the rock mass, leaving part of the rock mass destressed and sometimes unstable. The redistribution of stress results in irregular stress and strain fields that cannot now be adequately described by continuous mathematical expressions. The theories of continuum mechanics must be replaced by iterative computational schemes which take into account the nonlinear deformation of intact material, the creation of new discontinuities and displacement on these and any preexisting discontinuities. 16.6.3 Reinforcement Interaction with Instability Mechanisms In Section 16.4, it was shown that the apparently simple mechanism of a reinforcing element intersecting a single discontinuity was quite complex. In an ideal situation, the geometry, weight and displacement of an unstable region involving one discontinuity would be predictable and it would be possible to design a single reinforcing element response to maintain stability. In reality, the complexities of a rock mass make it very difficult to precisely predict the geometry and displacement of a failure mechanism. Furthermore, a multiplicity of either or both discontinuities and reinforcing elements are usually involved and consequently make the problem virtually indeterminate. A solution to the problem, if it could be solved, would only apply to the specific geometrical arrangement of rock structure, reinforcing elements, boundary conditions and material properties adopted. This makes reinforcement design for entire excavations more of an art than a science. The important questions the designer of an excavation must ask are as follows. (i) Can the excavation be created without risk of collapse? If not, can the geometry, extraction sequence or other operations such as blasting be rearranged to cancel the risk? (ii) What is the likely collapse mechanism and the extent of failure that must be dealt with? (iii) Is the reinforcement technique the best choice of ground improvement and, if so, could it be complemented by another technique? (iv) What are the best choices of reinforcing element, capacity, density, geometry and timing of installation? (v) Once installed, is there adequate quality assurance on installation? (vi) Has the reinforcement design solved the instability problem? (vii) If not, why and how can the design be suitably modified? There are of course other questions that must be asked concerning logistics and economics, for example. Support and reinforcement requirements may also be optimized by considering the effects of subsequent activities such as adjacent excavations and blasting practices. Therefore, the overall design of a reinforcement scheme is not just a case of solving for the components in point (iv). It is linked to all other excavation activities arid an evaluation of how these activities impinge on the overall stability must be attempted. The best designs will be those that approach the problem with consideration of all the above factors with a mixture of pragmatism, empiricism and rigor. Critical appraisal of the reinforcement performance will naturally lead to improved designs. 16.6.4 Design Methods There are several levels of sophistication and effort which may be used in the design of reinforced excavations. These range from looking at the success of reinforcement schemes in similar excavations to complex three-dimensional numerical analyses for the specific problem at hand. The methods available can be categorized into simple empirical methods, simple analytical methods, computational simulation methods and physical simulation methods.
Rock Reinforcement-Technology, 16.6.4.1
Testing, Design and Evaluation
All
Empirical methods
Design by precedent involves the review of reinforcing schemes used in the past for similar excavations and ground conditions. For example, Cording, Hendron and Deere [61] give a simple equation for a uniform 'support' pressure in terms of span, unit weight of rock and a multiplying factor for crowns and side walls in a large number of case studies on civil underground chambers. This uniform support pressure is assumed to be equal to the number of rock bolts per unit area multiplied by the pretension set in each rock bolt. Similarly, Lang [1] gives a number of tried and proven rules for permanent civil chambers for the rock bolt length and spacing in terms of the rock block width and the span of the crown or side wall. Thus, an estimate of rock bolt lengths, spacings and pretensions could be arrived at in a matter of minutes for the crown and walls of a chamber in blocky ground. Farmer and Shelton [62] have combined the experiences of previous workers into a single table of design rules for jointed rock masses. The next level of sophistication is the use of classification schemes. A number of classification schemes have been proposed for application in underground chambers, tunnels and slopes. Some of these, notably the CSIR Geomechanics Classification (Bieniawski [63] and Laubscher and Taylor [64]) and the NGI Tunneling Classification Scheme (Barton, Lien and Lunde [65]), have been modified and improved over the years into very useful and simple to use design tools. The schemes require the calculation of certain indices or rating values based on the particular rock mass conditions such as RQD, numbers of joint sets, joint strength parameters and stress reduction factors. These indices are then used to determine the associated recommendations summarized from observations on hundreds of case studies. An advantage of empirical methods is that the input information need only be relatively sparse. However, precedent methods may not be particularly reliable and should only be used as a guide to reinforcement requirements during feasibility studies. A greater amount of confidence can be attributed to a design derived from a classification scheme because greater amounts of information about the particular case are included in the design.
16.6.4.2
Analytical methods
It is possible for the excavation to behave as a continuum for the most part but with a transformation from continuous response to discontinuous response in restricted regions of the rock mass where the strength of the intact material or the strength on discontinuities is exceeded. When the failure mechanism is straightforward and the excavation shape simple, it may sometimes be possible to conduct simple analyses to define the region and extent of the failure zone and the probable displacement directions. With this information the length, capacity, pattern and geometry of a reinforcement scheme can be estimated.
(i)
Overstressed zones In massive rock, the failure regions are usually confined to the boundary of the excavation where shear failures occur due to excessive compressive stresses, or tensile failures can occur due to excessive tensile stresses. In many cases, the extent of each region shown in Figure 16 for massive, stratified and jointed rock can be predicted using simple elasticity theory. Methods to predict these regions are given by Hoek and Bray [3], Hoek and Brown [5] and Brady and Brown [9]. For example, in the case of a circular tunnel driven through rock in a biaxial stress field, the likely zones of instability can be predicted using the Kirsch solutions. In massive rock, zones of compressive and tensile failure that might occur at the boundary could be predicted as shown in Figure 16(a). In stratified rock the stresses may be resolved into shear and normal stresses on the bedding planes to predict regions of slip and flexure of the layers (Figure 16b). In jointed rock the stresses may be resolved onto the joints and the single plane of weakness theory (Jaeger [66]) applied in parts to give zones of slip on discontinuities and failure through intact material. A complicating issue is that once failure through intact material or displacement on discontinuities is initiated, stresses are redistributed away to other parts of the rock mass. This changes the problem somewhat and makes it difficult to estimate the depth of the failure zone in the rock mass. However, in most cases, the probable failure zone can be inferred. In the case of a tunnel in massive rock undergoing compressive failure a stable elliptical shape usually results.
472
Support (b)
(a)
Figure 16 Zones of instability around a tunnel: (a) zones of compressive and tensile failure in massive rock and (b) zones of layerflexureand slip in stratified rock (φ] is the friction angle on the joints)
(ii)
Simple and complex shaped individual blocks
Excavations in highly structured rock are often unstable due to the instability of some of the individual blocks of rock that make up the excavation boundary. In these circumstances the unstable blocks may translate or rotate into the excavation. There are a number of graphical and analytical single block analysis methods available for simple triangular wedges {e.g. Hoek and Bray [3], Hoek and Brown [5]), tetrahedral blocks {e.g. Priest [67]) and complex-shaped polyhedral blocks {e.g. Goodman and Shi [68], Warburton [69]). Most of the methods treat the block as a simple rigid body with continuous planar faces and ignore the effects of in situ stresses and the possibility of block rotation. Modern developments in this area have been reviewed by Warburton in Volume 2, Chapter 12. In general, all these methods require information on the structural geology of the rock mass and the geometry of the excavation and provide information on the individual blocks that may become unstable. The information on individual block size, weight, preferred movement direction and the factor of safety against translation can be used to assess the lengths, orientations, densities and capacities of the reinforcements required to maintain stability. Unfortunately, most methods are limited to the blocks directly adjacent to the excavation under the assumption that if these individual blocks can be made stable then the rock mass is also stable.
(Hi) Simple and complex shaped regions There are a number of very simple and reliable methods which predict simple- and complex-shaped regions that become unstable, Included in this group are the rotational slip analysis methods used in slope stability studies [3]. Equally simple methods exist for the unstable volumes of rock that may occur above tunnels and underground chambers. In general, the problem is solved by searching for a limiting equilibrium surface in the rock mass. The reinforcement may then be designed to maintain equilibrium at the interface between the two regions with the required lengths and directions of the reinforcement being given by the geometry of the unstable region and its movement direction, respectively. In many circumstances in structured rock, the individual blocks that become unstable may sometimes lead to more complex sequential or multiple-block collapses. Some of the methods for individual blocks can be adapted to follow the successive translations of single blocks which may form one large complex void. For example, the methods proposed by Goodman and Shi [68] and Warburton [69] can be used to predict the total unstable region as well as the individual blocks.
Rock Reinforcement-Technology, Testing, Design and Evaluation
473
The basic problem with these methods is that as the collapse of the block assembly becomes more complex, the effect of ignoring rotation as a possible failure mode becomes more severe, especially when the effects of reinforcement are to be included. A method for the analysis of single arbitraryshaped blocks which includes rotational displacements has been developed by Thompson [70]. The analysis is also able to incorporate specifically placed reinforcing elements (Figure 17).
Figure 17 Reinforcement of an arbitrary-shaped block
(iv) Rock reinforcement interaction In some special cases the concept of rock-support interaction or rock-reinforcement interaction may be used to estimate reinforcement requirements. This analytical procedure has been developed from work originally proposed by Daemen [71] and is really only suitable for underground excavations of a simple shape in which the assumption of plane strain is acceptable. The procedure could be generalized but would need to be conducted on a computer. The approach shown schematically in Figure 18 consists of predicting the 'ground characteristic' line on a graph of boundary displacement versus pressure on the boundary. The ground characteristic line represents the pressure required to limit deformation of the excavation boundary. This curve is then intersected by the 'support reaction line' which represents the pressure supplied to restrict and slow further deformation of the rock mass. The objective is to allow a sufficient amount of deformation to occur that will mobilize the inherent strength of the rock mass but insufficient to allow unraveling and complete loosening. Clearly, the timing of installation in terms of displacement and the stiffness of the reinforcement and support is very important. A
Support reaction line (a) Too stiff (b) Correct stiffness-correct timing
(c) Too flexible
(d) Incorrect timing
Radial displacement
Figure 18 The ground characteristic line and reinforcement interaction
474
Support
An important advance in this procedure, made by Ladanyi [72], enabled the extent of the plastic or fractured zone to be predicted around a circular tunnel in rock subject to a hydrostatic stress (Figure 19). By formulating this axisymmetrical problem into one differential equation of equilibrium a solution may be found for the zone of fracturing that develops around the excavation. This unstable zone is described by the radius of the elastoplastic boundary in terms of the rock properties and an internal pressure on the tunnel boundary. The procedure has subsequently been extended to include more realistic rock behavior and to predict the ground characteristic line. Solution techniques particularly amenable to simple calculations have been given by Brown et al. [73] and Gyenge [74] for tunnels and shafts.
Elastic zone
Figure 19 Fracture zone in massive rock (after Ladanyi [72])
The rock-support interaction analyses do not take into account the effect of reinforcement within the rock mass but rather the effect of an external pressure created on the boundary by the collective effect of surface hardware and any supporting elements. The support reaction lines for rockreinforcing elements fitted with surface hardware are sometimes nonlinear and must be defined by testing. (v)
Summary of analytical design methods
The analytical methods are commonly cast in two-dimensional form and make a number of sometimes restrictive assumptions. Some care must be exercised to ensure that the appropriate mechanism is being analyzed. Regardless of their shortcomings, these methods do provide an answer for the general shape and extent of the unstable zone within the rock surrounding the excavation. This allows the required capacity and the length of reinforcement to be estimated. 16.6.4.3
Computational methods
It is not intended to discuss the various computational methods in detail. Table 5 provides a summary of the major classes of computational methods (boundary integral, finite difference, finite element and distinct (or discrete) element) which may be used to analyze complex excavation stability problems. The capabilities and limitations of these methods are outlined elsewhere in Comprehensive Rock Engineering. It is only intended to identify the requirements for quantifying the effects of reinforcement on excavation behavior and the methods which can potentially achieve these. It was described in Section 16.6.1 how the rock mass response could be considered to be essentially continuous in most regions of the rock mass with the exceptions being close to excavations where fracture initiation or movement on preexisting discontinuities could occur. In general, the regions remote from excavations can be adequately assumed to be continuous but morerefined models need to be considered for the rock near the excavation boundary.
Rock Reinforcement-Technology, Testing, Design and Evaluation
475
Table 5 Computational Methods Method Boundary element Finite difference Finite element Distinct element
Rock mass type
Reinforcement model
Continuum Discontinuum Continuum Discontinuum Continuum Discontinuum
Equivalent material Limited explicit Equivalent material Explicit Equivalent material Explicit
Continuum Discontinuum
Equivalent material Explicit
Application Stress analysis Stress analysis Stress analysis Stress analysis Stress analysis Stress analysis Modes of displacement Stress analysis Stress analysis Modes of displacement
It has been observed that the effects of reinforcement on stress and displacement distributions are only noticeable when discontinuities are present in the computational analyses. The rock mass model must allow for mobility of the rock mass in order to reproduce the failure mechanisms of rock masses and quantify the modifying effects of reinforcement schemes. Analysis for the effects of reinforcement can only be achieved by the use of a computational method which either implicitly or explicitly can incorporate the effects of discontinuities. The first step is to identify the computational methods which can model the mechanisms of failure for the different classes of rock mass. Computational methods exist which have a few discrete discontinuities, simple layered models or an arbitrary distribution of rock structures. The remaining step is to then identify which of these methods can be used to incorporate the modifying effects of reinforcement. There are basically two methods by which the effects of discontinuities can be modeled within the computational schemes. Either the discontinuities are assumed to be present with a uniform distribution throughout or are specifically located as interfaces between regions of the rock mass. In the former case, procedures have been developed to derive equivalent continuum parameters which reflect the orientation and spacing of the discontinuities (e.g. [75]). The equivalent materials approach will generally be valid where discontinuities remain closed or undergo only small dilations. For these cases any of the continuum analysis schemes can be used. For cases where large dilational and shear displacements are possible, methods which include explicit discontinuities are required. The locations of specific discontinuities can be incorporated into all the major classes of methods but the number and distribution of discontinuities varies with the method. For limited numbers of discontinuities in massive rock, the boundary integral approach developed by Crotty [76] may be appropriate. For arbitrary discontinuous rock, the choice is basically betweenfiniteelement programs with discontinuity elements or distinct element programs. With both the equivalent material and specific discontinuity schemes, reinforcing elements can be defined either implicitly or explicitly. The modifying effects of a uniform reinforcement scheme on a rock mass with a uniform distribution of discontinuities can be modeled using the procedures developed by Gerrard and Pande [75] (Figure 20) or reinforcing elements can be located in specified positions. For boundary integral or finite element methods which may model specific discontinuities, it is also possible to modify the discontinuity properties in a similar way to the equivalent material models. This approach achieves a modification in the strength and stiffness of the discontinuity. It is found that the most noticeable change will be to modify the tensile strength. Only small improvements will be noticed in the shear strength and stiffness of closed discontinuities. This is in agreement with the concept described previously which suggested that the rock mass behaves like a continuum except in regions affected by the excavation. Earlier sections have also shown that the general response of reinforcing elements is complex. The equivalent models have a number of deficiencies in that they can only hope to represent simple approximations to the actual behavior of reinforcing elements. Furthermore they cannot reproduce one of the important rock mass response mechanisms of block rotation and the modification to this mode of deformation caused by reinforcement. Rotational rock mass failure modes such as block toppling or release are important in jointed rock. Both these modes can only occur when there are sufficient intersecting discontinuities in the model. There are only two computational methods which can achieve these modes. These methods are thefiniteelement method (coupled with the ability to model discontinuities by interface or joint elements) and the distinct element method, which by its nature has the capability to deal with
476
Support Load applied and prestress
Load applied and prestress
Rock I material First set | of passive reinforcement
Last set | of passive reinforcement
■ First set of ' interfaces
i Last set of 1 interfaces
■ First set of • discontinuities
First set of discontinuities
Second set of discontinuities
. Last set of I discontinuities
Load applied and prestress
Q
Passive reinforcement set
£ ]f
Interface for reinforcement set
Load applied and prestress
Figure 20 An equivalent material model of reinforced rock (after Gerrard and Pande [75])
modeling arbitrary discontinuities. The efficiencies of both these methods have been improved by coupling the discontinuous discretization near excavations with the boundary integral method to represent the assumed continuous regions distant from the excavation (e.g. Beer [77], Lorig and Brady [78]). It therefore remains to identify the requirements for the reinforcement model for these two types of computational methods. The earliest reinforcement models were in the form of bar elements in finite element methods. These were generally used to model discretely coupled reinforcing elements connecting specific nodes within the finite element mesh. Early applications clearly demonstrated that reinforcement did not modify the behavior of continuum models. They did not attempt to model the shear effects. The development of the joint finite element and distinct element model created a need for reinforcement models which could model both the axial and shear response modes. A number of relatively complex models have been attempted. Lorig [79] recognized that these could be approximated by two decoupled equivalent stiffness springs (Figure 21). One spring represents the axial response across a discontinuity and which rotates during shear. The other spring represents the shear response and isfixedin the direction transverse to the original reinforcement axial direction. This approach is capable of representing the basic combined response modes of reinforcement but requires careful selection of the characteristic length of the axial spring and appropriate forms of the load-displacement characteristics for each of the springs. An analysis to demonstrate the hybrid distinct element/boundary integral method was performed to determine the modifying effects of reinforcement in a layered crown of a large underground opening. The results are shown in Figure 22 and clearly show the changes in both the displacement profile and the creation of zones of compressive stress adjacent to the surface anchors. Both the hybrid finite element/boundary integral and distinct element/boundary integral computational schemes offer the best potential for the future as aids for the analysis of reinforcement schemes in discontinuous rock. In time, better reinforcement models will be developed which are compatible with the computational method and which reflect the characteristics of the different categories and types of reinforcing elements. The computational efficiency of the distinct element method coupled with improved computing technology will reduce the execution times for this method. The comments made above have simply been about techniques. The subject of reinforcement design is truly three-dimensional and the limitations of the use of two-dimensional analysis approximations should be assessed carefully.
All
Rock Reinforcement-Technology, Testing, Design and Evaluation
16.6.4.4 Physical simulation methods These methods are not widely used nowadays due to the high cost and skill associated with model manufacture and testing. It may also reflect the notion that much of this work can now be completed more rapidly and cheaply using numerical modeling. The critical problem in physical experiments on rock models, especially when reinforcing elements are involved, concerns the requirement to (a)
(b)
i
Shear displacement She
Axial spring Discontinuity Shear spring
Figure 21 Spring model for reinforcement (after Lorig [79])
(a) x I x
xy
\ X
x\* -Ί
( x]x
^ h * "N
TTJT frT~7
^ \
A<'WM Γ r 4 iff]
£x
^
\r A W
n
^ ^
_L
ON hC 0\ X X X^ % X X Λ X X X
&
—K Λ X
/
x
IVx x X NJ*^ K
Vhk
A**
>s^
xK
A
Ά
#
2l
ΚΛ ν
Figure 22 a
/
A
Δ ΔX MAS
"*J^^«j^^*f*"·
/
/
*,
478
Support (b)
Figure 22 Example of discrete element analysis of (a) unreinforced and (b) reinforced underground excavation
maintain similitude between the prototype and the model [80]. Physical modeling in jointed rock is particularly difficult because of its three-dimensional nature of response and the fact that most models are severe simplifications of the actual prototype. Regardless of the number of problems, physical modeling techniques have proven particularly useful in assessing reinforcement requirements and preliminary excavation designs. Physical modeling falls into approximately three classes: equivalent models, centrifuge models and photoelastic models. Collectively these models have led to a number of propositions regarding the interaction of the reinforcement with the rock mass. In equivalent material modeling the excavation is scaled down to a manageable size and the process of excavation or loading due to the stress field is simulated in a test frame. The work described by Lang [1] typifies the early modeling work conducted for permanent civil excavations. These experiments range from the famous 'reinforced upside-down bucket of gravel' to reinforced multiple-block models of underground power stations. This modeling led to the concept of 'keying' and many of the empirical reinforcement design rules still being used today. Centrifuge modeling is quite rare in rock mechanics studies and very few published cases are to be found where the effect of reinforcement has been included in a model study. A notable exception is the important work conducted by Panek [81] to study the fundamental behavior of discretely coupled elements in laminated roof strata above underground openings. This work led to the postulation of the friction effect and the suspension effect by which reinforcement may improve the stability of laminated strata. Photoelasticity [82] is a modeling technique used in rock mechanics to infer the stresses around excavations by constructing a model of the excavation in a suitable transparent material. The effect of reinforcing elements on the stress distribution can be introduced by loading the model with tensioned wires and observing the fringe patterns in polarized light. This technique is classically
Rock Reinforcement-Technology,
Testing, Design and Evaluation
479
(b)
(α)
//s=l.5
//s =2.0
(c)
Figure 23 Photoelastic model showing zones of compression caused by tensioned reinforcing elements (after Lang [35])
described in [1] for work conducted by Worotnicki over 30 years ago to determine the behavior of reinforced crowns in massive, layered and jointed rock. The jointed rock models were among the first studies on the behavior of reinforcement in a discontinuum. A single photoelastic model given in Figure 23 graphically illustrates the effect that the spacing of pretensioned elements has on the induced photoelastic stress pattern for a monolithic layer in the crown of an underground excavation. Later work conducted on the effect of reinforcement around circular holes has been conducted with similar care in St Petersburg, in particular UNIMI (All Union Scientific Research Institute of Mine Geomechanics and Survey [83]. Physical modeling has played an important role in the understanding of rock mechanics problems and the interaction of reinforcement. In general, the results can now be predicted by theoretical or numerical investigations. 16.7 16.7.1
REINFORCEMENT PERFORMANCE EVALUATION Philosophy
The philosophy for reinforcement evaluation is to review its performance at all stages from design to completion of excavation and for some time thereafter. The details of the philosophy of how to monitor the installed performance of reinforcement have been described in detail in Chapter 9 of
480
Support
Volume 5 of Comprehensive Rock Engineering. The monitoring of performance comprises only one part of an evaluation. Other important factors are the ease of installation, whether the performance was in agreement with predictions and whether there are better systems which could achieve the same result with improved productivity and economy. Performance evaluation is a crucial phase in the design process and should form the basis for ongoing refinement of reinforcement practices. 16.7.2 Performance-monitoring Instrumentation The techniques available to evaluate the performance of reinforcement are very limited. The techniques Vary in the precision and accuracy with which loads are indicated and they can be broadly classified as being qualitative or quantitative. Within the quantitative category of load indication, instruments can further be classified as measuring load or measuring strain. Table 6 gives a summary of the techniques available for reinforcement evaluation and some of the instruments which have been developed within each of the categories. The most common approaches for quantitative measurement are the use of either cells that measure loads at surface anchors or to use gauges that measure the strain at discrete points, or over discrete lengths, of the reinforcing element. The first approach only requires a knowledge of the cell characteristics which are usually obtained from laboratory calibrations. Most cells are designed to operate in the linear elastic range of the material and have high repeatability and sensitivity. Load cells are generally associated with the measurement of load transfer in discrete anchored devices. These types of reinforcing devices have also been monitored by strain gauges installed on the shank of the bolt between the anchor and the face plates. Conversely, the measurement of strain requires a knowledge of the reinforcing element material properties. The relationship between stress and strain will become nonlinear at high loads. This relationship becomes important when converting the indicated quantity to load for comparison with the capacity of the reinforcing element. Summaries of the techniques and commercial instruments available for reinforcement have been written by Moy [11], Rosso [84] and Maleki, Hardy and Brest van Kempen [85]. Also, some novel prototype instruments used by the writers in full-scale reinforcement design and evaluation exercises are described in detail in Volume 5, Chapter 9. 16.7.3 Performance-monitoring Programs Volume 5, Chapter 9 stresses the importance of rock mass monitoring (especially at discontinuities) in association with reinforcement-monitoring instrumentation. The best philosophy for Table 6 Instruments for Application to Reinforcement Load Measurement Description Strain measurement (Short base length)
Application
Device
(Long base length) Strain cell
Rock bolt shank Tendon wire Tendon (averaged) Tendon average
Extensometers
Deduction
RWCSG [86] RWCSC [86] YCSC [86] RWGSG [86]
Rock bolt nut Tendon surface anchor Tendon internal anchor Surface nut or anchor Surface nut or anchor
Hollow load cell [87], [90] Hollow load cell Hollow load cell Glotzl Perard [88]
Torque wrench Deformable washers Rock bolt
ISRM [58]
Load cell measurement (Strain gauge or vibrating wire) Hydraulic Photoelastic
Strain gauge
Indirect assessment Boltometer [89]
RWCSG, resistance wire cable strain gauge. RWCSC, resistance wire cable strain cell. RWGSG, resistance wire grout strain gauge. YCSC, yoke cable strain cell.
Rock Reinforcement-Technology,
Testing, Design and Evaluation
481
monitoring programs is the use of large numbers of inexpensive instruments as opposed to the use of smaller numbers of high cost instruments. The instruments must also be installed so that the measurements of reinforcement performance are complementary to the rock mass behavior measurements [90, 91]. The lack of instrumented case studies is one major cause for the poor understanding of reinforcement action and its relationship with rock mass behavior. Case studies are required in which information can be used to evaluate reinforcement performance in relationship to the rock mass displacement on discontinuities. The reason for the lack of information is the unavailability of suitable instrumentation for measuring discontinuity movements and deformations of many of the reinforcing devices. Many of the instrumented field studies on reinforcement have been with discretely anchored rock bolts such as expansion shell or resin-anchored deformed bar. The instrumented studies have generally involved short (less than 2 m long) devices and the results are easily interpreted. By contrast, continuously coupled devices are far more difficult to instrument and the results are also difficult to interpret. The cement-grouted devices are usually installed in long lengths which complicates the process of selecting appropriate points at which to monitor rock mass displacements and reinforcement load. The writers are not aware of any published information obtained for instrumented, short, continuous frictionally coupled devices.
16.8 SUMMARY AND CONCLUSIONS This chapter has defined a limited terminology which should be used in relation to the general discipline of ground improvement techniques. In particular, support and reinforcement have been deemed to be different techniques within this discipline. Further, reinforcing devices have been classified into three major categories which are based on their method of load transfer between the particular reinforcing device and the borehole. This enables a better understanding of how each of the many reinforcing devices behaves and aids in the comparison of performance of reinforcing devices within one category or between devices in different categories. A comprehensive table of categories of devices has been presented. Devices which are developed in the future can easily be added to this table. The limited terminology can also be used to improve the understanding of reinforcement action and aids in the planning of field and laboratory testing programs to quantify reinforcement properties. The types of tests and their relevance to particular categories of reinforcing devices have also been described and evaluated. The limitations of the axial loading test and the importance of shear tests have been presented. The double embedment test has been identified as being the most appropriate test for reinforcement evaluation in the laboratory. The area in which significant advances can be expected in the future is with respect to improved methods and techniques for reinforcement design. The review of design and analysis methods has identified a number of limitations with current procedures. These limitations are caused by the general intractability of the design problem to analysis and the requirement for the more appropriate techniques to have specific excavation geometries in rock masses with a defined arrangement of discontinuities. Methods which attempt some variability in the rock mass model and the effects of overall reinforcing schemes are available but they cannot reproduce the actual mechanics of the stabilizing effects of reinforcement in a failed rock mass. The innovation of new reinforcing techniques should, where at all possible, be accompanied by a thorough performance evaluation program. A philosophy to be adopted in planning a suitable program has been proposed in which reinforcement performance is inextricably linked to rock mass behavior. This requires the use of instrumentation schemes in which the reinforcement and rock mass discontinuity monitoring are complementary so that reinforcement performance can be explained in terms of the rock mass deformations. The method of reinforcement design using experience, trial and observation will remain the most appropriate for some time. This process uses all the available information from theory, testing, analysis and personal or reported experience to make judgments as to the preferred reinforcing scheme for particular rock mass conditions and excavation geometries. The future should see this apparently ad hoc approach become more or less formalized. This will make reinforcement design more accessible to the general rock mechanics community faced with this responsibility and not the domain of a few experienced practitioners who have, from necessity, achieved their expertise over a period of several years. Reinforcement and support are too important not to have the best technology applied for optimum productivity, economy and safety.
Support
482 16.9
REFERENCES
1. Lang T. A. Theory and practice of rock bolting. Trans. Am. Inst. Min. Metall. Pet. Eng. 220, 333-348 (1961). 2. Barron K., Coates D. F. and Gyenge M. Artificial support of rock slopes. Research Report R228, CANMET, Energy, Mines and Resources, Mining Research Centre (1970). 3. Hoek E. and Bray J. W. Rock Slope Engineering (2nd edn. 1977, 3rd edn. 1981), p. 358. Institution of Mining and Metallurgy, London (1974). 4. Schach R., Garshol K. and Heltzen A. M. Rock Bolting: A Practical Handbook, p. 84. Pergamon Press, Oxford (1979). 5. Hoek E. and Brown E. T. Underground Excavations in Rock, p. 527. Institution of Mining and Metallurgy, London (1980). 6. Hobst L. and Zajic J. Anchoring in rock. In Developments in Geotechnical Engineering, vol. 13, p. 390. Elsevier, Amsterdam (1983). 7. Douglas T. H. and Arthur L. J. A guide to the use of rock reinforcement in underground excavations. CIRIA Report No. 3. CIRIA, London (1983). 8. Stephansson O. (Ed.) Proc. Int. Symp. Rock Bolting - Theory and Application in Mining and Underground Construction, Abisko, Sweden, p. 404. Balkema, Rotterdam (1983). 9. Brady B. H. G. and Brown E. T. Rock Mechanics for Underground Mining, p. 527. Allen & Unwin, London (1985). 10. Franklin J. A. and Dusseault M. B. Rock Engineering, p. 600. McGraw-Hill, New York (1989). 11. Moy D. The design and monitoring of rock bolt and dowel systems in underground excavations. Ph.D. Thesis, Imperial College, University of London (1975). 12. Scott J. J. Interior rock reinforcement fixtures, state-of-the-art. In Proc. 21st U.S. Symp. Rock Mech. Rolla, MO (Edited by D. A. Summers), pp. 79-87. University of Missouri (1980). 13. Stillborg B. Professional Users Handbook for Rock Bolting, p. 145. Trans Tech Publications, Sweden (1986). 14. Hoek E. and Wood D. Rock support. World Tunnelling and Subsurface Excavation, 2, 131-134 (1989). 15. Clifford R. L. Long rockbolt support at New Broken Hill Consolidated Limited. In Proc. Aust. Inst. Min. Metall. No. 251, pp. 21-26. AusIMM, Melbourne (1974). 16. Palmer W. T., Bailey S. G. and Fuller P. G. Experience with preplaced supports in timber and cut andfillstopes. In Proc. AMIRA Tech. Meeting, Wollongong, pp. 45-71. AMIRA, Melbourne (1976). 17. Fuller P. G. Pre-reinforcement of cut and fill stopes. In Applications of Rock Mechanics to Cut and Fill Mining (Edited by O. Stephansson and M. J. Jones), pp. 55-63. Institution of Mining and Metallurgy, London (1981). 18. Lang T. A. Rock reinforcement. Bull. Assoc. Eng. Geol. 9, 215-239 (1972). 19. Ortlepp W. D. An empirical determination of the effectiveness of rockbolt support under impulse loading. In Proc. Int. Symp. Large Permanent Underground Openings, Oslo (Edited by T. L. Brekke and F. A. Jorstad), pp. 197-205. Scandinavian University Books, Oslo, Norway (1970). 20. Pender E. B., Hosking A. E. D. and Mattner R. H. Grouted rock bolts for permanent support of major underground works. J. Inst. Eng. Aust. 35, 129-150 (1963). 21. Littlejohn G. S. and Bruce D. A. Rock anchors: state-of-the-art. Ground Eng. 8(3), 25-32 (May 1975); 8(4), 41-48 (July 1975); 8(5), 34-35 (Sept 1975); 8(6), 36-45 (Nov. 1975); 9(2), 20-29 (March 1976); 9(3), 55-60 (May 1976); 9(4), 33-44 (July 1976). 22. Parker H., Hargreaves R. P. N., Ramseil R. and Hawkins A. Strata reinforcement using the technique of longhole resin dowelling. Min. Eng. 154, 519-532 (1973). 23. Dywidag Systems International. Dywidag - threadbar rock bolts for underground excavations and stabilisation of rock formations (company pamphlet) (1982). 24. Ortlepp W. D. A yielding rock bolt. Research Organization Bulletin of the Chamber of Mines South Africa 14, August, 6-8 (1968). 25. SIG. New generation rockbolt systems. Tunnels and Tunnelling/Machinery, Plant and Equipment Issue 13, August, 29 (1981). 26. Reed J. J. Rock reinforcement and stabilization-bolting.alternatives. Min. Congr. J. 12, 40-45 (1974). 27. Habberstad J. K., Thompson R., Bates R. and Wallace G. Fibreglass reinforced polymeric rock bolts. In Proc. Rapid Excavation Tunnelling Conf. San Francisco (Edited by H. C. Pattison and E. D'Appolonia), pp. 977-993. Port City Press, Baltimore (1974). 28. Hutchins W. R., Bywater S., Thompson A. G. and Windsor C. R. A versatile grouted cable dowel reinforcing system for rock. In The AusIMM Proc. vol. 295(1), pp. 25-29. AusIMM, Melbourne (1990). 29. Hunt R. E. B. and Askew J. E. Installation and design guidelines for cable dowel ground support at Z C / N B H C In Proc. Underground Operators Conf, Broken Hill, NSW, pp. 113-122. AusIMM, Melbourne (1977). 30. Matthews S. M., Thompson A. G., Windsor C. R. and O'Bryan P. R. A novel reinforcing system for large rock caverns in blocky rock masses. In Proc. Int. Symp. Large Rock Caverns, ISRM, Helsinki, Finland (Edited by K. Saari), vol. 2, pp. 1541-1552. Pergamon Press, Oxford (1986). 31. Scott J. J. Friction rock stabilisers - a new rock reinforcement method. In Proc. 17th U.S. Symp. Rock Mech., Snowbird, UT pp. 242-249. (1976). 32. Wijk G. and Skogberg B. The Swellex bolting system. In Proc. 14th Can. Rock Mech. Symp. Voncouver Paper No. 20. Can. Inst. Min. Metall., Vancouver (1982). 33. Bangzao L. Rockbolting for excavations in sensitive urban areas. In Proc. Conf. Rock Eng. and Excavation in an Urban Environment, Hong Kong, pp. 197-209. Institution of Mining and Metallurgy, London (1986). 34. Nikolaev N. Oval shaped frictional pipe anchor. In Proc. Int. Symp. Rock Bolting - Theory and Application in Mining and Underground Construction (Edited by O. Stephansson), Abisko, Sweden, pp. 337-340. Balkema, Rotterdam (1983). 35. Lang T. A. Underground experience in the Snowy Mountains-Australia. In Proc. 2nd Int. Conf. Protective Construction, Santa Monica (Edited by J. J. O'Sullivan), vol. 2, pp. 767-853. Rand Corp., CA (1959). 36. Ingersoll-Rand. Introducing Dyna-Rok and Dyna-Rok Plus anchors-a new generation of roof support systems (company pamphlet) (1989). 37. Hagedorn H. Synthetic resin grouted tube anchors. Rock Mech. Rock Eng. 16, 143-149 (1983). 38. Shaw J. D. N. Cementitious grouts. Civ. Eng. May, 62-70 (1980).
Rock Reinforcement- Technology, Testing, Design and Evaluation 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77.
483
Robins N. S. and Milodowski A. E. Q. J. Eng. Geol. 19, 175-181 (1986). McCormick J. A., Hollop E. D. and Debevec W. J. Survey of resin bolting installations. Min. Congr. J. 60, 28-32 (1974). Breakthrough in roofbolting using full column cementitious anchoring. Min. J. 296, Methods and Machines, p. 276 (1981). Davidge G. R., Martin T. A. and Steed C. M. Lacing support trial at Strathcona Mine. In 2nd Int. Symp. Rockbursts and Seismicity in Mines, Minnesota, USA, pp. 363-367 (1988). Bywater S. and Fuller P. G. Cable support of lead open stope hangingwalls at Mount Isa Mines Limited. In Proc. Int. Symp. Rock Bolting - Theory and Application in Mining and Underground Construction, Abisko, Sweden (Edited by O. Stephansson), pp. 342-357. Balkema, Rotterdam (1983). Matthews S. M., Tillmann V. T. and Worotnicki G. A modified cable bolt system for the support of underground openings. In Proc. Aust. Inst. Min. Metall. Underground Operators Conf. Broken Hill, pp. 243-255 (1983). Thompson A. G., Matthews S. M., Windsor C. R., Bywater S. and Tillmann V. H. Innovations in rock reinforcement technology in the Australian mining industry. In Proc. 6th Int. Congr. Rock Mech. Montreal, Canada (Edited by G. Herget and S. Vangpaisal), vol. 2, pp. 1275-1278. Balkema, Rotterdam (1987). Bartos P. (Ed.) Bond in Concrete. In Proc. Int. Conf. Bond in Concrete, Paisley College of Technology, Scotland, p. 466. Applied Science Publ. London (1982). Farmer I. W. Stress distribution along a resin grouted rock anchor. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 12, 347-351 (1975). Fuller P. G. and Cox R. H. T. Mechanics of load transfer from steel tendons to cement based grout. In Proc. 5th Australian Conf. Mech. of Struct, and Mat. Melbourne, pp. 189-203 (1975). Fairhurst C. and Singh B. Roof bolting in horizontally laminated rock. Eng. Min. J. February, pp. 80-90 (1974). Gerdeen J. C , Snyder V. W., Viegelahn G. L. and Parker J. Design criteria for roof bolting plans using resin-grouted nontensioned bolts to reinforce bedded mine roof. U.S. Dept. of the Interior, Bureau of Mines, OFR 46-80 (1980). Dight P. M. Improvements to the stability of rock slopes in open pits. PhD. Thesis, Monash University, Australia (1983). Pells P. J. N. The behaviour of fully bonded rock bolts. In Proc. 3rd Int. Congr. Rock Mech., Denver, vol. 2B, pp. 1212-1217. NAS, Washington, DC (1974). Vos E. and Reinhardt H. W. Influence of loading rate on bond behaviour of reinforcing steel and prestressing strands. Mater. Constr. 15(85), 3-10 (1982). Dulacska H. Dowel action of reinforcement crossing cracks in concrete. J. Am. Concr. Inst. 69(12), 754-757 (1972). Haas C. Shear resistance of rock bolts. Trans. Am. Inst. Min. Metall. Pet. Eng. 260, 32-41 (1976). Bjurstrom S. Shear strength of hard rock joints reinforced by grouted untensioned bolts. In Proc. 3rd Int. Congr. Rock Mech. Denver, vol. 2B, pp. 1194-1199. NAS, Washington, DC (1974). Windsor C. R., Thompson A. G. and Choi S. K. Rock reinforcement research for hard rock mining. In Proc. WAS M Conference '88 - R & D For the Minerals Industry, Kalgoorlie, pp. 113-121 (1988). International Society for Rock Mechanics Commission on Standardization of Laboratory and Field Tests 1974. Suggested methods for rockbolt testing. Reprinted in Rock Characterization, Testing and Monitoring (ISRM Suggested Methods) (Edited by E. T. Brown), pp. 161-168. Pergamon Press, Oxford (1981). Bray J. W. The teaching of rock mechanics to mining engineers. Min. Eng. 126, 224-237 (1967). Brown E. T. Fracture of rock under uniform biaxial compression. In Proc. 3rd Int. Congr. Rock Mech. Denver, vol. 2A, pp. 111-117. NAS, Washington, DC (1974). Cording E. J., Hendron A. J. and Deere D. U. Rock engineering for underground caverns. In Proc. Symp. Underground Rock Chambers, Phoenix, AZ, pp. 567-600. ASCE, New York (1971). Farmer I. W. and Shelton P. D. Review of underground rock reinforcement systems. Trans. Inst. Min. Metall. 89, A68-83 (1980). Bieniawski Z. T. Engineering classification of jointed rock masses. Trans. S. Afr. Inst. Civ. Eng. 15, 335-344 (1973). Laubscher D. H. and Taylor H. W. The importance of geomechanics classification of jointed rock masses in mining operation. In Exploration for Rock Engineering (Edited by Z. T. Bieniawski), vol. 1, pp. 119-128. Balkema, Cape Town (1976). Barton N., Lien R. and Lunde J. Estimation of support requirements for underground excavations. Rock Mech. 6, 189-239 (1974). Jaeger J. C. Shear failure of anisotropic rock. Geol. Mag. 97, 65-72 (1960). Priest S. D. Hemispherical Projection Methods in Rock Mechanics, p. 124. Allen & Unwin, London (1985). Goodman R. E. and Shi H. Block Theory and Its Application to Rock Engineering, p. 338. Prentice-Hall, Englewood Cliff, NJ (1985). Warburton P. M. Applications of a new computer model for reconstructing rock geometry, analysing single block stability and identifying keystones. In Proc. 5th Int. Congr. Rock Mech., Melbourne, pp. F225-230. Australian Geomechanics Society, Melbourne (1983). Thompson A. G. Analysis of a reinforced arbitrary shaped block. In Proc. Int. Conf. Computational Techniques and Applications, Brisbane CTAC-89 (Edited by W. L. Hogarth and B. J. Noye), pp. 723-730. Hemisphere Publ. Corp., New York (1990). Daemen J. J. K. Problems in tunnel support mechanics. Underground Space, 1, 163-172 (1977). Ladanyi B. Use of the long-term strength concept in the determination of ground pressure on tunnel linings. In Advances in Rock Mechanics, Proc. 3rd Int. Congr. Rock Mech., Denver, vol. 2B, pp. 1150-1156. NAS, Washington, DC (1974). Brown E. T., Bray J. W., Ladanyi B. and Hoek E. Characteristic line calculations for rock tunnels. J. Geotech. Eng. 109, 15-39 (1983). Gyenge M. Parametric study of variables of support design. CANMET, Division Report MRP/MRL 83-67(TR). Energy, Mines and Resources, Canada (1983). Gerrard C. M. and Pande G. Numerical modelling of reinforced jointed rock masses - I Theory. Computers and Geotechnics 1, 293-318 (1985). Crotty J. M. User's manual for BITEMJ - two-dimensional stress analysis for piecewise homogeneous solids with structural discontinuities. CSIRO Division of Geomechanics, Australia (1983). Beer G. B E F E - a combined, boundary element finite computer program. Adv. Eng. Software 6(2), 103-109 (1983).
484 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91.
Support Lorig L. J. and Brady B. H. G. A hybrid discrete boundary element method of stress analysis. In Design and Performance of Underground Excavations, Proc. ISRM Symp., Cambridge, UK (Edited by E. T. Brown and J. A. Hudson), pp. 105-112. British Geotech. Soc, London (1982). Lorig L. J. A simple numerical representation of fully bonded passive rock reinforcement for hard rocks. Computers and Geotechnics 1, 79-97 (1985). Mandel J. Tests on reduced scale models in soil and rock mechanics, a study of the conditions of similitude. Rev. Ind. Miner 44(9), (1962). Panek L. A. Design for bolting stratified roof. Trans. Am. Inst. Min. Eng. 229, 113-119 (1964). Coker E. G. and Filon L. N. G. A Treatise on Photo-Elasticity (2nd edn.), p. 720. Cambridge University Press, Cambridge (1957). Filatov N. A. and Ievlev G. A. Aspects of studying roof bolting operation using polarized-light optical method. In Proc. Int. Symp. Rock Bolting - Theory and Application in Mining and Underground Construction, Abisko, Sweden (Edited by O. Stephansson), pp. 115-121. Balkema, Rotterdam (1983). Rosso R. S. An investigation of the effects of hardened washers on the uniformity of roof bolt tension and resulting ground control in an underground mine. U.S. Dept. of the Interior, Bureau of Mines, Open File Report 40-78 (1977). Maleki H. N., Hardy M. P. and Brest van Kempen C. J. H. Evaluation of roof bolt tension measuring techniques. In Proc. 26th U.S. Symp. Rock Mech., Rapid City, SD (Edited by E. Ashworth), pp. 425-437. Balkema, Rotterdam (1985). Windsor C. R. Application of instrumentation in underground excavation, rock reinforcement and rock deformation. In Geotechnical Field Instrumentation, vol. 1, pp. 14.1-14.15. Australian Geomechanics Society/IEAust (1987). Windsor C. R., Thompson A. G. and Cadby G. W. Monitoring rock-support interaction around tunnels. In Proc. 6th Aust. Tunnelling Conf. Melbourne, vol. 2, pp. 173-182. (1987). Perard Torque Tension Limited. Rock bolting systems for mining and civil engineering (company pamphlet) (1984). Thurner H. F. Non-destructive test method for rock bolts. In 3rd Int. Congr. Int. Assoc. of Eng. Geol. Madrid, Spain, vol. 10, pp. 309-312. (1978). Dejean M. and Raffoux J. F. Monitoring of rock bolting and of its effectiveness. In Rock Bolting, Revue de VIndustrie Minerale, Saint-Etienne, GEDIM, pp. 153-164. (1980). Windsor C. R. and Worotnicki G. Monitoring reinforced rock mass performance. In Proc. Int. Symp. Large Rock Caverns, ISRM, Helsinki, Finland (Edited by K. Saari), vol. 2, pp. 1087-1098. Pergamon Press, Oxford (1986).
17 Rock Mass Response to Large Blast Hole Open Stoping BENGT L. STILLBORG Luleâ University of Technology, Sweden 17.1
INTRODUCTION
485
17.2 SITE CHARACTERIZATION 17.2.1 General Geology 17.2.2 Diamond Drilling and Core Logging 17.2.3 Structural Mapping 17.2.4 Mechanical Properties of Rock 17.2.5 In Situ Stresses 17.2.6 Groundwater conditions
486 486 486 487 488 488 491
17.3
491
MINE LAYOUT DESIGN
491 492 492 493 499 500 502
17.3.1 Mining Considerations 17.3.2 Mining Sequence 17.3.3 Stability Analysis for Detailed Layout Design 17.3.4 Design of the Crown Pillar 17.3.5 Design and Support of the Loading Level 17.3.6 Hanging-wall Support 17.3.7 Foot-wall Support 17.4 INSTRUMENTATION 17.4.1 17.5
502
Displacement Measurements of the Hanging-wall, Crown and Transverse Pillars
ROCK MASS BEHAVIOR DURING MINING
17.5.1 17.5.2 17.5.3
502 504 504 508 508
The Crown The Loading Level The Hanging-wall
17.6
SUMMARY AND CONCLUSIONS
510
17.7
REFERENCES
511
17.1
INTRODUCTION
The activities of the Swedish Mining Research Foundation, (1981-1986) were primarily confined to the Luossavaara mine in Kiruna where the Research Mine was operated. The Luossavaara orebody has been mined since the end of the 19th century, but the mine was closed in 1973 due to a recession in the iron ore market. Research carried out at the mine in the fields of rock mechanics, mining engineering and mining equipment engineering aimed to introduce and develop new technology for the Swedish mining industry. In particular the rock mechanics program involved three major components. (i) A geomechanics preinvestigation scheme focusing on methods applicable to mining. (ii) A rock reinforcement programme directed towards prereinforcement. (iii) An extensive instrumentation and monitoring scheme for observation of rock behavior during mining.
485
486
Support
A number of factors influence the choice of mining layout and mining sequence in a particular orebody. In the Research Mine Programme, it was decided that an open stope technique would be used, together with large hole drilling, 162 mm (6.5 in). This decision was made before the start of the project, based on recent trends in world mining. Application of large-scale techniques in underground mining have shown very good economic results. Furthermore, all additional considerations in the mine design were based on research programs in rock mechanics, mining engineering and mining equipment engineering. From a rock mechanics viewpoint, the ideal situation would be to conduct a thorough geomechanical investigation. Based on this, an 'optimum' mining layout from the stability viewpoint would be proposed. However, this Optimum' layout can never be fully implemented in practice. Due consideration must be given to practical mining engineering, economics, machinery and production requirements. Furthermore, a geomechanical investigation activity has no 'end point'. New developments should be added to the investigation, so that current information can be updated continuously. An economic study of the proposed open stope mining indicated that best return could be achieved with stope heights of at least 75 m. Stope dimensions which are as large as possible provide the greatest chance to yield useful research information, as the stability of the excavations approaches their limit. Thus a decision was made to conduct a geomechanical preinvestigation programme, studying a stope height of 100 m over a 220 m section of the orebody.
17.2 SITE CHARACTERIZATION 17.2.1 General Geology The Luossavaara orebody is 1200 m long, with an average width of 23 m. It is known to extend down to depths of approximately 500 m below surface level. The ore strikes roughly NNE, and dips at about 60° towards the east (Figure 1). It is a high grade magnetite deposit, containing an average of 66% Fe, with low phosphorus and alkali contents, called B ore. The Luossavaara orebody lies between a syenite porphyry foot-wall and a quartz bearing porphyry hanging-wall, [1]. The northern part of the orebody was mined within the Research Mine programme. The section considered had a length of 220 m and a width of 5-35 m. The contacts between the ore and the porphyries were, in general, easily delineated, and often defined by a chloritized schistose layer. An 'ore breccia' also occurred in this section of the orebody, in the syenite porphyry foot-wall. The ore breccia closest to the orebody was high grade, with 40-60% Fe content, but was not mined. 17.2.2 Diamond Drilling and Core Logging The purpose of the exploration drilling was the following. (i) Locate and determine the extension of major crush zones, fault planes and other dominant structures. (ii) Delineate ore binderies from chemical analysis on core samples, ore reserve estimation and ore quality designation. (iii) Define the extension and the thickness of a hanging-wall contact zone. (iv) Determine RQD values (Rock Quality Designation) on cores. (v) Collect rock samples and cores for mechanical properties testing. (vi) Sample details of rock texture and structure. The RQD values determined from the diamond drilling programme can be summarized as follows: (i) hanging-wall, in general, was approx. 75%, but deteriorated approaching the ore contact; (ii) ore, approx. 95%, this was in general much less jointed than the hanging-wall; and (iii) foot-wall values were between those of the hanging-wall and ore, approx. 85%. A mean joint frequency (joints m" 1 ) was calculated from underground mapping: hanging-wall 4.5 ± 0.3; ore 2.1 ± 0.2; and foot-wall 3.8 + 0.3. These results confirmed the indications given by the RQD results on cores. The northern region of the orebody is intersected by a fault plane (Figure 1). The fault forms the boundary of the orebody to the north. The boreholes associated with the fault exhibit a relatively higher joint density than other regions of the hanging-wall. This is also true for southern regions of the hanging-wall. There is a higher joint density in the hanging-wall near the ore contact. The joint frequency exceeds the mean hanging-wall value by 30% at distances less than 10 m from the hanging-wall
Rock Mass Response to Large Blast Hole Open Stoping
I
I Quartz -bearing porphyry ^ H
\ \
487
Ore
Faults
Figure 1 The geology of the Luossavaara area, after Parâk [1]
contact. This high frequency is concentrated mainly within 2 m of the hanging-wall contact. There, the joint frequency exceeded the mean value by as much as 50%. The joints observed were not always open. Rather, they often corresponded to foliated structures in the rock, opening up when the core sample was taken out of the barrel. This phenomenon may well have great importance to stope hanging-wall stability. 17.2.3 Structural Mapping The results of structural mapping underground are summarized in stereo contours (Figure 2). The following remarks can be made. (i) A major proportion (47%) of the joints were randomly orientated throughout the mining area. Seven groups of joints were delineated. However, the foot-wall possessed the most regular formation of joints, compared to the hanging-wall and ore. (ii) The orebody was the least jointed area in the mine. A major proportion (57%) of the joints were randomly orientated. Four groups of joints were delineated. (iii) The hanging-wall was the most heavily jointed area in the mine. The major proportion of the joints were randomly orientated. Eight sets of fairly well defined groups of joints were distinguished, but these still represented the minority (31%) of all joints. In general, the joints were very limited in extent. 5% of the joints mapped had an in-plane area exceeding the cross-sectional area of the drifts, approximately 5 x 4 m2. Extensive cross checking failed to identify any more consistent jointing patterns. The only major structure found and confirmed was the fault in the northern region of the orebody.
Support
488 Foot-wall
Ore
Hanging-wall Legend E ^ j l 2 - 3 % Poles 1 1
3 - 4 % Poles
l l l l l 4 - 5 % Poles | H
I 5 - 6 % Poles 6 - 7 % Poles Ore projection
Figure 2 Result of structural mapping underground. Foot-wall, ore and hanging-wall plot of poles of joints
17.2.4 Mechanical Properties of Rock A number of different rock properties were determined for different rock types. A summary of the results is presented in Table 1. Referring to Table 1 the following can be concluded. (i) The results of the uniaxial compressive strength display a considerable scatter. This reflects the variations of the rock, even within one and the same rock type. (ii) The Young's modulus of the quartz-bearing porphyry in the hanging-wall is significantly higher than that of the ore. (iii) Except in the high phosphorus ore, the Poisson's ratio is much the same throughout the mining area. 17.2.5 In Situ Stresses The location of in situ stress measurements is shown in Figure 3. The results of the measurements are shown in Figure 4. The result can be summarized according to [2] as follows.
489
Rock Mass Response to Large Blast Hole Open Stoping Table 1 Mechanical Properties of Intact Rock, Core Sample Testing. Ore
Foot-wall B
D
Hanging-wall contact
Hanging-wall
Rock property
Syenite porphyry
Breccia
Low phosphorus <0.1%, Fe > 60%
High phosphorus > 0.3%, Fe > 60%
Chloritized quartz-bearing porphyry
Quartz-bearing porphyry
Young's modulus (GPa)
57.3 + 23.9 10
59.7 + 11.3 8
38.2 + 15.6 12
34.5 + 6.9 3
38.4 + 13.8 9
56.9 + 15.8 92
68.7 + 51.7 7 14.6 + 6.6 13 0.15 + 0.07 11
52.8 ± 57.6 7 14.1 + 9.0 3 0.16 + 0.08 5
100.4 + 20.9 2 11.6 1 0.11+0.04 9
—
19.6 + 6.8 9 17.2 + 12.2 3 0.16 + 0.07 8
109.6 + 65.8 137 20.4 + 8.0 64 0.16 + 0.05 80
Uniaxial compressive strength (MPa) Tensile strength (MPa) Poisson's ratio Fracture toughness (MNm" ( 3 / 2 ) )
1.44 1
— 0.06 + 0.03 3
2.08 1
1.84 + 0.28 8
Results are given as mean value ± standard deviation number of tests
Outcrop of body
265m level, plan view
Figure 3 Location of boreholes for rock stress measurements (close up shows details of location and borehole length)
(i) In the foot-wall, the maximum principal stress σχ forms an angle of 40° to the orebody and is flat dipping. The directions of σ2 and σ 3 are inconsistently defined. The magnitude of σ1 is approximately 10 MPa and those of σ2 and σ 3 are 5 MPa and 3 MPa, respectively. (ii) In the orebody, the maximum principal stress σχ forms an angle of 45° to the orebody and is flat dipping. The directions of σ 2 and σ 3 are inconsistently defined. The magnitude of at is approximately 10 MPa and those of σ2 and σ 3 are 5 MPa and 3 MPa, respectively.
490
Support Ore
Foot-wall
Hanging-wall Legend •
Maximum principal stress
o
Intermediate principal stress
*
Minimum principal stress Orebody
Mean Value of Principal Stress Intermediate
Maximum Location
Mag (MPo)
Bearing
Mag.
Dip
(Μρ*,
Bearing
Minimum
Dip
Mag *
( M p
Foot-wall
10.4
165°
7°
5.4
3.3
Ore
9.6
331°
17°
4.9
2.7
Hanging-wall
8.3
200°
34°
5.2
2.9
Bearing
Dip
Figure 4 Results of stress measurements. The principal stresses in the foot-wall, ore and hanging-wall. Stereographic projection in the lower reference hemisphere, Wulff net. Table: mean value of principal stresses
(iii) In the hanging-wall, the maximum principal stress, σχ is almost parallel to the orebody, dipping 30° to the south, whilst the directions of σ2 and σ 3 are inconsistently defined. The magnitude of σχ has been accurately determined, being 8 MPa; but those of σ2 and σ 3 are not so well defined. However, they are estimated to be 5 MPa and 3 MPa, respectively.
Rock Mass Response to Large Blast Hole Open Stoping
491
From the above results, the following can be concluded. (i) The measured magnitudes of the stresses are consistent with the measurement locations and rock types. (ii) The maximum principal stress is nearly horizontal, and is approximately twice the intermediate principal stress. This is consistent with other results obtained in the region [3]. (iii) The directions of the stresses are not consistent in the three rock types: an unexpected result. This is probably a localized effect, since measurements were confined to the 265 m level in the mine at relatively shallow depths (143m below the mean surface height). However, this situation is frequently encountered. It is perhaps typical in the experience of stress measurement in northern Sweden.
17.2.6 Ground water Conditions Water which flows into the mining area is assumed to originate from three sources: atmospheric precipitation through the caved material at the outcrop of the orebody, water from the nearby Luossajärvi Lake (Figure 1), and migration of groundwater from joints and faults. From these, percolation of water through the outcrop of the orebody is by far the most dominant source of flow of water into the mining area. The water that penetrates the caved material will ultimately be distributed throughout the mining area, by percolation through the joints in the rock mass. From structural mapping underground, including the investigation of joints carrying water, the following conclusions were made. (i) A very small fraction of the total number of joints mapped (1-1.5%) contain any water. (ii) All joints carrying water are steeply dipping, and are randomly oriented. (iii) Water is most likely to penetrate all structures emanating on the surface in the spring when the snow melts and large volumes of water are released.
17.3 MINE LAYOUT DESIGN The final mine layout and stoping sequence are intimately related to both mining and rock mechanics considerations. Based on overall rock mass strength and ore grades, the general type of mining method is chosen. In the case of the Luossavaara orebody, bulk mining open stoping methods were considered suitable. Once this had been decided upon, further detailed layout involved an iterative process of examining various layout alternatives from the point of view of both stability and mining practicality. As with many aspects of rock mechanics, however, some aspects of the final detailed design could not be carried out in advance due to the complexity of the problem and the lack of appropriate design methods. To resolve this latter problem, final layout decisions had to be based on close monitoring of rock mass behavior during mining. This section outlines the mining and rock mechanics considerations of layout design made prior to mining.
17.3.1 Mining Considerations The Luossavaara deposit was intended to be a test site for both mining and rock mechanics, and therefore a new mining method was developed. This new method borrowed features from opencast techniques, VCR mining, conventional sublevel open stoping and sublevel caving. The concept was to develop aflexiblemining method that was both large-scale and selective, and could therefore be applied to a wide variety of deposits. For economic reasons therefore, the following constraints were imposed on the mine layout. (i) A top drill level would be used for downhole drilling. (ii) The selected drill level must enable the drilling of a uniform blasthole drill pattern. This would allow a uniform drilling density and fragmentation, minimum drilling volume, ease of drill setup, and simple planning and execution of charging work. (iii) A minimum blasthole length of 75 m. (iv) A layout of the loading level that permits both conventional and continuous loading systems. (v) Four stopes in production simultaneously to provide the necessary production tonnage. Based on these considerations, a mining sequence was developed.
492
Support
17.3.2 Mining Sequence Referring to Figure 5, the proposed mining sequence can be summarized as follows. Primary stage 1. Four primary stopes should be mined under stable conditions. Secondary stage 2. Mining of the two intermediate pillars by blasting rows sequentially and/or mass blasting. Third stage 3. Blasting of the crown above stopes A, B, C and D. Fourth stage 4. Mining of the remaining central pillar and the crown together with mining of the trough undercut remnant. Apart from the four primary stopes, the mining sequence described above is veryflexible.It can easily be exchanged for an alternative mining sequence, should rock mechanical measurements indicate that alternative plans are necessary. Within the primary stopes, the extraction sequence was chosen to demonstrate a selective mining capability, and also to ensure maximum stability of the hanging-wall. The initial opening was created by driving a raise in the production hole pattern, followed by conventional benching of the production holes around the raise until a sufficiently large opening was created to blast large bench rounds. The stability of the hanging-wall was considered to be of prime importance due to the adjacent waste rock and hanging-wall contact being chlorite-rich and jointed. For this reason it was decided to mine the foot-wall sectionfirst,followed by progressive mining towards the hanging-wall contact. This would be beneficial for two reasons. Firstly, initial mining would affect and weaken the hanging-wall contact and therefore impede energy transfer into the hanging-wall in subsequent blasts. This would reduce blast damage. Secondly, the size of the charge could be progressively reduced as mining operations approached the hanging-wall. This mining sequence is shown in Figure 6. 17.3.3 Stability Analysis for Detailed Layout Design Stability must be considered from a global scale down to a local scale. These different scales as they apply to the Luossavaara orebody are shown in Figure 7. (i) Global instability will occur if a major structure or portion of it, for example, the hanging-wall or the crown-pillar, would collapse and so ruin any further possibility of a controlled mining process. This must be prevented by a reinforced or unreinforced crown pillar giving the hanging-wall sufficient support during the mining of the stopes. (ii) Regional stability must be considered when decisions regarding the stope height, pillar dimensions and general layout of drill and loading levels are to be made. The hanging-wall between pillars has to be considered to prevent extensive backbreaking, which would result in serious dilution and even major, global instability. (iii) Local stability is related to details of drill and loading levels, blast damage and unfavorable orientation of discontinuities causing fallouts in the drifts and ore passes. The economic impact of
Figure 5 Schematic illustration of proposed mining layout and sequence of mining. Left: longitudinal, idealized vertical section. Right: idealized, typical vertical cross section. 1. Primary stopes. 2. Secondary mining of intermediate pillar between stopes A and B. 3. Third stage of mining, mass blasting of crown above stopes A and B and pillar 1
Rock Mass Response to Large Blast Hole Open Stoping
493
Horizontal section
Figure 6
Mining sequence in primary stopes, selective mining
this type of instability problem may not be hazardous to the mining venture. However, the importance of safety considerations should not be underestimated. Crown pillar and loading level stability can be assessed explicitly using either analytical or numerical methods. The results of these analyses will be described as they influence the support design in those areas. Hanging-wall stability, on the other hand, cannot reliably be assessed analytically. Part of the research programme was therefore directed at examining the effect of different support strategies on hanging-wall stability. This latter aspect of the mine layout will be addressed in more detail in the section dealing with rock mass monitoring. (Section 17.4). 17.3.4 Design of the Crown Pillar The crown pillar separates the previously mined upper sections of the orebody from the lower stopes to be mined as part of the research programme (see Figure 7). The crown pillar not only acts as a support member to provide global stability, but also as a working platform from which drilling is carried out. Because of the mining method chosen, the cross-cuts from which drilling were to be carried out were relatively closely spaced (5 m wide x 4 m high with center to center spacing of 10 m). In assessing the stability of the crown pillar, therefore, it was decided to consider the drilling crosscuts as a horizontal slot dividing the crown pillar into an upper and a lower section (see Figure 7). During mining, the thickness of the top crown pillar would be progressively reduced. Two approaches were used in determining the dimensions of the crown pillar. Firstly, a simple analytical method was used. This consisted of the two-dimensional hanging-wall caving model, using the solution developed by Brown and Ferguson [4]. This model is shown in Figure 8. Using the relevant rock mass strength and geometrical parameters, it was concluded that hanging-wall stability
494
Support
Figure 7 Schematic cross section of an open stope mining layout demonstrating examples of ( 1 ) global, (2) regional, and (3) local instability
would be ensured if the entire crown pillar thickness of at least 5 m was used. In practice this is a small figure considering the 16 m width of the orebody, blast-induced fracturing and the like. It was decided that the top crown pillar should therefore have a minimum dimension of 10 m. The bottom crown pillar was more complicated to design, mainly due to a lack of published guidelines. The approach adopted was therefore to make an initial estimate using finite element stress analyses, and in practice to use support and monitoring. The stability of the crown pillar can be assessed by comparing the strength at various points in the pillar to the stress. The finite element program was used to obtain the stress distribution in crown pillars of different sizes, and to calculate the stability in terms of the well known ratio of strength to stress - the factor of safety. The method of determining stress by this method was quite standard in that a linear elasticity was assumed. Strength, however, was determined in the following manner, after [5]. Firstly, it is well known that large volumes of rock have a lower strength than small laboratory specimens. This volume dependence of strength is determined using the following equation
where σ0 = unconfined compressive pillar strength,
where σρ = confined pillar strength, σ 3 = minimum principal stress, q = (1 + sin0 b )/(l — sin0 b ), and b = base friction angle of joints in the pillar.
Rock Mass Response to Large Blast Hole Open Stoping
495
-"300
Hx '■ Mining depth at which initial failure occurs H2- Mining depth at which subsequent failure occurs Hc- Depth of caved material P : Force simulating the support of the crown pillar 5 : Width of orebody T '· Thrust on failure plane due to caved material Tc : Thrust on foot-wall due to caved material V '■ Water pressure force in tension crack W : Weight of wedge of sliding rock Wc ■ Weight of caved material X '■ Thickness of crown pillar U : Water pressure force on failure surface Zx '■ Depth of intial tension crack Z2: Depth of subsequent tension crack Zm : Depth of water in tension crack a '■ Dip of upper ground surface (shown positive in Figure 13) 0 : Inclination of 7" to normal to failure surface φ„ ■ Friction angle between caved and undisturbed rock fa : Dip of orebody j* b : Angle of break ψ9{. Dip of initial failure plane Ψ&. Dip of subsequent failure plane
Figure 8 Theoretical model for analysis of failure induced by caving, after Brown and Ferguson [4]
The formula for strength is based on the well-known Mohr-Coulomb failure criterion. Finally, the factor of safety F is determined by strength stress The finite element model was constructed to portray the full stope geometry. A detail of the crown pillar section showing the different components of the crown pillar is shown in Figure 9, and the corresponding rock properties are listed in Table 2. A number of model geometries based on Figure 9 were run, in which the thickness of the bottom crown pillar, H2, was varied. An example of
496
Support
Figure 9 Geomechanical model of the crown pillar. Numbers refer to rock types listed in Table 2
Table 2 Rock Properties for Numerical Modeling of the Crown Pillar Rock type
(1) Quartz-bearing porphyry (2) Foliated quartz-bearing porphyry (3) Chloritized quartz-bearing porphyry (4) Ore (5) Breccia (6) Caved material
Young's modulus (GPa)
Uniaxial compressive strength (MPa)
Poissorfs ratio
Unit weight (kgm" 3 )
57
110
0.16
2700
38
110
0.16
2700
13 38 60 0.030
20 100 52
0.10 0.11 0.16 0.25
2400 4970 3600 2400
—
the results of this analysis, in terms of strength and factor of safety contours in the crown pillar, is shown in Figure 10. In Figure 10, it can be seen that stability varies considerably throughout the pillar. It was decided that an overall assessment of the pillar stability could be obtained by considering the factor of safety at the pillar midpoint. In terms of the width to height ratio of the bottom crown pillar, the strength and factor of safety from the finite element analyses are summarized in Figure 11. These results show that for the bottom crown pillar, both strength and factor of safety increase with increasing width to height ratio, whereas in the top crown pillar, factor of safety increases while strength decreases. The reason for this somewhat unusual behavior in the top crown pillar is caused by a decrease in the minor principal stress component as the bottom crown pillar is increased in size, thereby decreasing strength. The factor of safety used in this analysis has direct impact on the function of the crown pillar. Agapito and Hardy [6] claim that safety factors of 0.8-1.0 may be appropriate for yielding pillars. Safety factors of 1.2-1.3 are needed for full overburden support systems and safety factors of 1.4-1.6 for barrier pillars. If we apply their approach to the crown pillar at Luossavaara we reach the following conclusion: the large dimension of the bottom crown pillar allows us to accept certain amount of yielding and we can allow for a safety factor of O.ß. From the diagram of safety factor versus height to width ratio for the midpoint of the pillar (Figure 11) we obtain HJW = 0.7. For an ore width of 20 m this will give a minimum pillar thickness of H2 = 14 m. However, this is true for the midpoint of the pillar. If we analyze the isograph of the safety factor for a bottom crown pillar with H2 = 15 m (Figure 10), the safety factor is less than 0.75 for the bottom central parts of the pillar. Hence, if we allow for minor failure and maybe limited amounts of rock falls from the roof of the large open stope, a bottom crown pillar of 15 m height is an optimum design. As cable
497
Rock Mass Response to Large Blast Hole Open Stoping (a)
Luossavaara Reseach Mine Pillar strength analyses
Attributes Parallel 0.000 0.000 I .000
U
Scale
10
0
I—
-J
Contour levels A 0.000E + 0 0 5.000E + 0 0 Ι.ΟΟΟΕ + ΟΙ I . 5 0 0 E + 0I 2.000E+0I 2.500E + 0 I 3.0Ο0Ε + 0 Ι 3.500E + 0 I 4.000E + 0I 4.500E+0I 5.000E + 0 I Min I . 9 I IE + 01 Max 6 . 7 0 IE + 01 Bottom crown pillar 15 m
(b)
Luossavaara Reseach Mine safety factor
Björsta data AB Baspl V3.0 10 Jan 85
Attributes Parallel 0.000 0.000 1.000 Scale 0 i
L. 10
—I
Contour levels A 5.000E-0I 7.500E-0I B I.OOOE + OO C I.250E + 0 0 D I.500E + 0 0 E 2.000E + 0 0 F 3.000E + 0 0 G 4.000E + 0 0 H 5.000E + 0 0 I Min 5 . 3 8 5 E - 0 I Max 4.071 E + 0 2 Bottom crown pillar 15 m
Björsta data AB Baspl V3.0 9 Jan 85
Figure 10 Pillar strength and safety factor of crown pillar for H2 = 15 m, after Stephansson [5]: (a) pillar strength, (b) factor of safety
reinforcements are installed in the bottom crown pillar we can assume that these will prevent failure and rock falls. A safety factor of 1.2-1.3 is needed for pillars to carry the weight of the overburden of a mine as stated by Agapito and Hardy [6]. A safety factor of 1.3 for pillars in general is suggested by Hoek and Brown [7]. For this analysis we use a safety factor of 1.25 as a design criterion. From the diagram in Figure 11, which shows safety factor versus pillar height to width ratio, a safety factor of 1.25 for the midpillar gives a ratio of 1.25. Hence, for a top crown pillar of 10 m thickness and an ore width of 20 m the thickness of the bottom crown pillar is H2 = 15 m, i.e. the same dimension as obtained in the previous section with the assumption of a yielding bottom crown pillar. The isographs of the safety
498
Support (α)
Bottom crown pillar
a.
Έ 30
S
20h
/ . /
o
CO
0.5 1.0 1.5 2.0 Height to width ratio, Hz/W D Pillar strength • Safety factor
(b)
Top crown pillar 40
S. §
30
20
0.5 1.0 1.5 2.0 Height to width ratio,(HX+HJ/W
2.5
Figure 11 Pillar strength and safety factor as a function of pillar height to width ratio for the midpoint of the pillar, after Stephansson [5]: (a) bottom crown pillar, (b) top crown pillar
factor for the top crown pillar with H2 = 15 m show a safety factor of 1.25 for most of the upper part of the pillar and safety factors of 4-5 at the drilling level. Further mining of the bottom crown pillar beyond 15 m will reduce the safe factor of the top crown pillar to become 0.5 when the bottom crown pillar is diminished. In conclusion, this analysis shows that an optimum design of the crown pillar at 265 m level of Luossavaara Research Mine for the case of a 10 m high top crown pillar is a 15 m high bottom crown pillar. This design is recommended for the final stage of mining and will give a yielding bottom crown pillar with a midspan safety factor of 0.8 and a full overburden support with a safety factor of 1.25 for the top crown pillar. As with many rock mechanics design procedures, the reliability of the pillar stability calculations remains unknown until direct experience has been obtained. In view of this, a number of additional considerations were made with regard to the implementation of the design and mining procedures adopted. These will be briefly described. The first consideration was to introduce controlled blasting. Blasting is a means of introducing fractures to the rock mass if it is not carefully controlled. It was decided that in all development work, smooth wall blasting techniques would be used to minimize overbreak. In practice, this eliminated the need for systematic bolting, and only spot bolting was required. Secondly, an additional measure of support was given to the bottom crown pillar by introducing cable bolts. This is shown schematically in Figure 12. In practice, the final thickness of the bottom crown pillar was not specified. This was made possible by the mining method in which ore was extracted in an upward direction. By monitoring the degree of fracturing in the bottom crown pillar, and measuring displacements, it was possible to make the decision on pillar thickness as mining progressed, based on the real pillar behavior. The instrumentation programme will be described in detail in Section 17.4.
Rock Mass Response to Large Blast Hole Open Stoping
499
Top crown pillar
Bottom crown pillar
Figure 12 Cross section of crown pillar including cable bolt reinforcement of bottom crown pillar
17.3.5 Design and Support of the Loading Level Stability of the loading level is of utmost importance to prevent any unnecessary interruptions to ore production. In the planning stage, there were two possible locations being considered for the loading level haulage and cross-cuts, either in the foot-wall or the hanging-wall. For ore recovery reasons, the hanging-wall location was favored from a mining engineering point of view, but based on rock mechanics considerations, a foot-wall location was chosen. As with the crown pillar design, this stability analysis was carried out using a two-dimensional finite element program in which the multiple cross-cuts on the loading level were considered as a continuous horizontal slot. To ensure further a high degree of stability, three additional strategies were adopted. Thefirstwas smooth wall blasting as in the drilling level. Secondly, support was installed as soon as possible during development of the cross-cuts and the drawpoints. Finally, the drawpoints were excavated in a particular sequence designed to facilitate the prereinforcement of critical areas. The latter two aspects will be described in more detail below. Figure 13 shows a cross section through the loading level illustrating the major components of support and excavation sequence. The most important part of the drawpoint to control throughout
c
^
Zero reodlng
Profil·
830618
DST-8
^
) ""
— *
Prof «col· Displ seal· 60 cm l—1
2 mm
Level 370 m
Foot-wall
4m
\
I 2 4 m II
I' / I
I
4 m \y\iL/ i > 1 ,1 Λι Λ Cross-cut
4 m—^~~ JÇy{\ 4m ■
Scram -T
T
T'
3m
Blow area *
\
Drawpoint for continuous loading
*
\
Figure 13 Cross section through the loading level. In the scram, a convergence measurement profile is illustrated. Typical convergence results for the scram after completion of the third stage of mining are also shown
500
Support
mining is the brow, which controls the flow of ore during the drawing operation. To assist in supporting this area as well as possible, the through drive in the orebody and the cross-cuts were excavated leaving an unmined plug beneath the brow (see the hatched area of Figure 13). Fully grouted rebars were then installed in the brow in three fans from the cross-cut. From the trough undercut drive, two fans of fully grouted 38 mm discarded steel winder ropes were installed. The flexible steel ropes were used in this area in preference to rebars because of their greater ease of installation in confined areas. After the brow had been extensively presupported as shown in the figure, the plug was blasted. Practical experience indicated that grouted reinforcing bars provided better support in drawpoint brows. Since the foot-wall adit (the scram) and the intersections between the scram and the drawpoints were not subjected to the same kind of dynamic loading and abrasion as the brows, the lighter and more easily installed Swellex bolts were used. The hanging-wall drift was prereinforced using the same procedure as the corresponding foot-wall adit (the scram), but placed inside the orebody (Figure 13). A hanging-wall drift located in the ore was, from a stability point of view, considered to be a calculated risk. The drift was still considered necessary for recovery of the trough undercut remnant. For research purposes, the installation of a continuous loading system in tne hanging-wall drift was also planned. 17.3.6 Hanging-wall Support The general philosophy adopted for the stability of the hanging-wall was as follows. (i) In the primary stage of mining (mining of the four primary stopes), the conditions of the hanging-wall should be stable, and the dilution of the ore by rock coming from the hanging-wall should be kept at a minimum (local stability). (ii) In the secondary stage of mining (mining of the intermediate transverse pillars), the regional stability of the hanging-wall should be maintained, so that the overall stability of the hanging-wall as Stope B
Stope C
Stope D
Stope A
280 m
Trough undercut remnant
Stope D
Pillar 3
Stope C
Pillar 2
Stope B
Pillar I
Stope A
Figure 14 Theoretical support pattern at hanging-wall contact, the perspective looks up towards the hanging-wall interface from the empty stopes. Theoretical support pattern (horizontal cross section) of bottom crown 285 m level is displayed in the lower part of figure
Rock Mass Response to Large Blast Hole Open Stoping
501
well as the recovery of the crown pillar are not jeopardized. However, it was understood that a higher rate of dilution was to be expected. (iii) In the third stage of mining (mass blasting of the crown), the hanging-wall was expected to fail and perhaps bring the failure up to the surface. In the primary stage of mining, support of the hanging-wall was provided by the crown pillar, the transverse pillars between stopes, the trough undercut remnant and the preplaced cable reinforcement system. The crown improves the regional stability of the hanging-wall. The trough undercut remnant reduces the actual exposed hanging-wall height. Nevertheless, the exposed hanging-wall area of stope B was approximately 2600 m2 at an orebody dip of 60°. The crown and the transverse pillar (natural support members) provided primary support to the hanging-wall by preventing collapse of major portions of the hanging-wall. The cable bolting provided local stability, in that the cables prevented blocks of rocks from failing due to gravity loading, which would dilute the ore. Local stability of the hanging-wall between pillars was achieved by prereinforcement using fully grouted cables. For practical application of an even support pattern at the hanging-wall contact, two hanging-wall drifts have been developed at a constant distance of 30 m from the hanging-wall interface at the 265 m level and on an intermediate level at 320 m. The drifts were planned for access to the hanging-wall for cable bolting, and extensive instrumentation of the hanging-wall. The distance between the hanging-wall contact and drifts is 1.5 times of the average width of the orebody. This, in practice is considered to be the distance necessary in order that the drift will not be affected by the excavation of the opening. No special support action was necessary here, since our numerical modeling predicted that excessive stress would not be encountered in regions around these drifts. Three different patterns of support have been designed for stopes B, C and D (Figure 14). Stope A has no cable reinforcement and serves as a reference to the reinforced stopes. Details of the support patterns are based on three different strategies, based on the desired effect of the support. Stope B (Figures 14 and 15) is reinforced in a 3 x 3 m pattern, at the hanging-wall contact. Every second fan has been shifted 1.5 m sideways giving the regular pattern of reinforcement shown in Figure 14. The cable bolting is confined to the central part of the exposed hanging-wall surface and does not give Stope B 250 my \ \
N
\ Top \ crown pillar \
X
265 m
\
365 m
I y Load level v
I Trough . /undercut^ I remnant \
Figure 15 Vertical cross section of theoretical support pattern of stope B. Dashed lines indicate position of adjacent fan
502
Support
complete cover all the way down to the trough undercut remnant. The idea behind this design is to investigate the action of a beam type of support, and whether this gives any stabilizing effect to the unreinforced surrounding rock. A beam action giving support to surrounding rock is of interest. When the larger stopes originate in secondary stages of mining, these will only be artificially supported (if no additional support is introduced) by reinforced beams. The quality of support obtained in stope B, a total of 6200 m of cables in the hanging-wall, is determined by the comparatively poor hanging-wall conditions in the southern part of the mining area. Stope C will have a support pattern of 6 x 6 m covering the full exposure of the hanging-wall. The support of stope C includes less than half the cable bolting used for stope B. The effect of an evenly distributed pattern of reinforcement will be studied in stope C, taking a mean joint spacing of the hanging-wall of approximately 4.5 m. The reinforcement of stope D is a combination of the beam action support of stope B and the support pattern for stope C (but less dense) reduced to 6 x 3 m, with no zigzagging of the cable fans (Figure 14). 17.3.7 Foot-wall Support The stability of the foot-wall is of minor importance. A number of the foot-wall discontinuities dip at an angle less than 62°, subparallel to the orebody. They do not appear to be continuous and they have high internal friction. The general dip of the orebody will stabilize the foot-wall during mining. Any dilution from the foot-wall will be high grade ore breccia, unlike the hanging-wall plain country rock. For these reasons, nothing was done to stabilize the foot-wall to provide local stability. Regional stability of the hanging-wall as well as the foot-wall will be provided by the transverse pillars.
17.4 INSTRUMENTATION The detailed design of various aspects of the mine layout described in the previous sections were necessarily based on idealizations of the real conditions. When implementing such designs it is important to ensure that the real rock mass response is behaving according to the predictions, and that no unforeseen events are occurring. Monitoring rock mass behavior by means of various instruments is therefore an important part of any large rock mechanics program. At the Luossavaara mine, the purpose of the instrumentation program was to monitor rock mass behavior during progressive extraction of the ore. Areas of particular interest were: (i) hanging-wall, (ii) transverse pillars, and (iii) crown pillar (bottom). After extracting the ore in the stopes, the crown pillars were to be mass blasted. The effect of this major change in regional support was also of great interest. Numerous instruments were installed for monitoring purposes, but the basic quantities measured in all cases were: (i) displacements, (ii) absolute stress and stress changes, and (iii) ground velocity (microseismic events). The type and layout of the instruments in this program is of interest, and will be briefly described prior to presenting the results from monitoring. 17.4.1 Displacement Measurements of the Hanging-wall, Crown and Transverse Pillars Rock mass displacements were monitored using both rod and wire extensometers. Details of these instruments and their layout are shown in Figures 16 and 17. For the multiple point rod extensometers, both manual and remote read-out were used. Both read-out systems had a measuring range of approximately 200 mm. Due to design differences in the two measuring systems, the accuracy of the readings differed, being + 0.9 mm for the remote read-out system and ± 0.2 mm for the manual reading system (for a typical 30 m long seven rod instrument). Data from the two readout systems were stored in a computer located on the surface, which made data manipulation relatively simple. Wire extensometers have less accuracy than rod extensometers, but have the benefit of tolerating more lateral movement of the rock mass with respect to the axis of the borehole in which they are installed. These extensometers were therefore only placed in those areas where large movements were expected, which as shown in Figure 17 is in the hanging-wall. Remote reading was also used on the wire extensometers.
Rock Mass Response to Large Blast Hole Open Stoping
503
Figure 16 Location of rod extensometers in mining layout. Close-ups show details of instrument: (a) in-hole section, (b) remote read-out head, (c) manual read-out system
While extensometers provide a general appreciation of rock mass deformations, they only provide a measure of displacement over the interval between adjacent anchors. The resolution is therefore limited and the cause of the displacement, i.e. one or many fractures, cannot be determined. To overcome this problem, an instrument known as a 'conductive strip' was developed, details of which are shown in Figure 18. The instrument is basically a contiguous series of 1 m long conductive strips. If the borehole undergoes an excessive amount of shear or tension, a given circuit will break, the location of which is determined by means of a portable circuit scanning instrument. Once the strip is broken, it cannot be read again, but the zone of displacement is then determined. Reading of the strips was made only occasionally, guided by movements registered with the extensometers.
504
Support
Figure 17 Location of wire extensometers in mining layout. Close-ups show details of instrument: (a) complete instrument, (b) remote read-out head
17.5 ROCK MASS BEHAVIOR DURING MINING This section describes the rock mass behavior during mining, based on instrument readings and from general observations. The particular areas to be discussed are the crown, the loading level and the hanging-wall. 17.5.1 The Crown In the extraction of stopes B and C (Figure 5), which were the first two stopes to be mined out, the stope roofs were mined domed shaped (Figure 19). It was believed that this would enhance the stability of the bottom crown pillar as it was mined thinner. The actual result of this measure was not experienced until the last rounds of stopes B and pillar 1 were mined out. In pillar 1 and the last rounds of stope B, the stope roof profile was changed from domed shaped to a flat roof perpendicular to the dip of the orebody, as shown in Figure 19.
Rock Mass Response to Large Blast Hole Open Stoping
505
Interfacing unit 'board'
Figure 18 Location of rod and wire extensometers in mining layout, including shear strips. Close-ups show details of instrument: (a) in-hole section of shear strip, (b) read-out unit for shear strips
As the roof form was changed, the stability of the roof improved significantly. An overbreak of 2-4 m was often observed when the dome-shaped stope roof was used. The overbreak was measured accurately by using the blastholes. The corresponding overbreak for the flat surface stope roof, perpendicular to the dip of the orebody, was reduced to almost zero. The explanation for this is to be found in weak hanging-wall contact, the structures in the crown, as well as the fact that the major principal stress direction was oriented approximately along the strike of the orebody. These factors together were unfavorable to the stability of the dome shaped roof. In addition, the effective roof area is reduced significantly with a flat roof. This reduces the probability of gravity-induced block caving from the roof. The effect of the change in the stope roof geometry was detected by both measurements of blasthole length after each blast and also by observations in the draw points. Blocks of ore holding intact pieces of boreholes would indicate overbreak. However, as the stope height increased, the stability of the crown was also determined by rod extensometers installed from the drill level down into the ore. In the last rounds, only the rod extensometers were used to determine the stability of the crown continuously, even in the time between blasts.
506
Support
Figure 19 Shape of the roof section. Changes from dome shaped, initially, to a flat roof perpendicular to the dip of the orebody. This improved the stability of the crown significantly
Extensometer
No. I anchor G No. 7 anchor
Hanging-wall
Loading level No. 7 anchor lost
Ω. No. 3 No. 2 The complete anchor lost anchor lost exto. lost
No. 5 and 6 No. 4 anchor lost anchor lost
^ o c o E E
No. I anchor
No. 4 anchor No. 4,5 and 6 anchor
No.2 anchor J_ D
E
F
Sequence of mining
Figure 20 A schematic illustration of the mining sequence of stope B, together with the corresponding response of one of the extensometers in the crown pillar
Rock Mass Response to Large Blast Hole Open Stoping
507
In stopes A and B and pillar 1, mining ceased at a bottom crown pillar thickness of 14 m. Since the extensometers installed in the bottom crown had a length of 24 m, mining continued up through the extensometers. This created no problem for the function of the extensometers. Readings could be obtained without any interruption, until the entire crown was mass blasted in the third stage of mining. In Figures 20 and 21 the typical response of two of the rod extensometers installed in the bottom crown pillar of stope B and pillar 1, respectively, is shown. The response of the extensometers is correlated with the approximate location of the lower limit of the bottom crown. It is obvious from Figure 20 that the crown was not subjected to any significant downward displacements as a result of the excavation of the stope. In the second stage of mining, the bottom crown area increased, covering the roof sections of stopes A and B as well as pillar 1. This resulted in displacements of the bottom crown of the pillar 1 as shown in Figure 21. The magnitude of the displacements remained small, however, and did not cause any visible instability to the bottom crown section. One important conclusion, obtained from the extensometer readings, was that the cable reinforcement of the bottom crown pillar of stope B was of no use. From the viewpoint of the stability, it could have been omitted. Based on this experience, cable reinforcement of the bottom crown of pillar 1 was not used. The confinement given to the crown by the hanging-wall and the foot-wall, and the flat roof linked with the favorable orientation of the rock mass structures in the crown gave very good stability to the crown, even in an advanced stage of mining. However, since stope B was the Extensometer
No.I anchor
No. 7 anchor
Hanging-wall
O
Loading level
Scale
No. 5,6 and 7 anchor lost
20 m
The complete exto. last
100 U No. 4 anchor
"r 10 r-
A No. 3 anchor
5 c
No. 7 No. 6 No. 5 No. 4 No. 3
I
^ -σ
E
No. 2 I
1 -i a. 3
No. 7 anchor
A No. 2 anchor
IT
No. I anchor
Sequence of mining
-2
Figure 21 A schematic illustration of the mining sequences of pillar 1, together with the corresponding response of one of the extensometers in the crown pillar
508
Support
first stope to be mined out, the cable reinforcement of the bottom crown pillar, together with the instrument readings provided a system which gave a feeling of security to the miners using the crown as a working platform. The final decision to mass blast the crown was not based on any possible instability of the crown and/or the fact that it was subjected to time-dependent displacements. The bottom crown pillar could have been mined thinner, before it was finally mass blasted. All rock mechanics experience gained from the excavation of stopes A and B and pillar 1 supported this argument. However, in a research mine operation, a number of interests (sometimes conflicting) must be considered. The successful control of a large mass blast of the crown was of interest to the mining engineering program. 17.5.2 The Loading Level The precautions taken at the loading level, to maintain the stability during mining, were very successful. No stability problems were encountered at the foot-wall side of the orebody. Mucking of the ore could be conducted without interruptions for additional reinforcement of the rock. Even the brow, commonly a weak link in this kind of mining system, showed no sign of instability that called for additional reinforcement. Convergence and extensometer instruments were used to monitor the rock mass response to mining. In Figure 13 a convergence measurement profile in the scram is illustrated, together with the final deformations of the scram profile as a result of mining. It is evident from Figure 13 that the deformations, however small, still clearly indicate that the scram was subjected to some sort of rotation as a result of mining. The hanging-wall drift was however severely damaged by the large hole blasting and had to be abandoned before it was ever used. A major cause of the stability problems was that the blasthole length with respect to the location of the hanging-wall drift was not accurately determined. Some blastholes therefore ended up too close to the drift. Dynamic loading from several blasts and falling ore also gave rise to the problems encountered. It is therefore believed that placement of a hangingwall drift inside an orebody, in the mining situation described above, is not advisable, unless very special precautions are taken to prevent stability problems. 17.5.3 The Hanging-wall The stability of the hanging-wall was monitored using both conventional extensometers and the so-called sliding micrometer. Blast damage was assessed using cross-hole seismics and a vibration monitoring system. The sliding micrometer produces displacement readings along the complete length of the instrument at intervals of 1 m, with an accuracy of ±0.003 mm over a base length of 1 m. The results obtained by the instrument, over three years of use, proved most reliable. The instrument can therefore be highly recommended when accurate readings of displacements in rock are required. All of the combined instrumentation gave a very good feedback system to the mining activity and its impact on the surrounding rock. In the primary stage of mining, support of the hanging-wall was provided by the crown pillar, the transverse pillars between stopes, the trough undercut remnant as well as the preplaced cable reinforcement system. The crown improves the regional stability of the hanging-wall. The trough undercut remnant reduces the actual exposed hanging-wall height. Nevertheless, the exposed hanging-wall area of stope B was approximately 2600 m2 at an orebody dip of 60°. The crown and the transverse pillars (natural support members) provided primary support to the hanging-wall by preventing collapse of major portions of the hanging-wall. The cable bolting provided local stability, in that the cables prevented blocks of rocks from failing due to gravity loading, which would dilute the ore. An illustration of the displacements of the hanging-wall of stope B, as a result of the completion of the mining of stope B is given in Figure 22. In the secondary stage of mining (mining of the transverse pillar between stopes A and B, pillar 1), the hanging-wall remained very stable. Only isolated minor sections of the hanging-wall faced stability problems that caused any ore dilution. These results were very encouraging, bearing in mind that the exposed hanging-wall area, after completion of the secondary stage of mining exceeded 5600 m2. From the instrument readings it was evident that the magnitude of displacement was limited. The magnitudes of the displacements were significant only at the weak hanging-wall
Rock Mass Response to Large Blast Hole Open Stoping
509
Figure 22 Illustration of the displacements (measured using a sliding micrometer, readings taken at 1 m intervals) of the hanging-wall of stope B as a result of the completion of the mining of (i) the primary stopes A and B ( ), (ii) the secondary mining of intermediate pillar 1 ( ), (iii) the third stage of mining, mass blasting of crown above stopes A and B and pillar 1, ( ) (note the different scales of the diagrams)
contact. It was not possible to evaluate the effect of the different cable bolting patterns used in the hanging-wall of stope B and pillar 1, in detail. However, it was believed that the regular pattern of cables covering the complete hanging-wall area of stope B resulted in a lower rate of dilution than was obtained from the hanging-wall of pillar 1. Pillar 1 had its reinforcement confined to the central part of the exposed hanging-wall surface. Details of the cable bolting layout are given in Section 17.3.6. A comparison of the results of stope B and pillar 1 with those from the unreinforced hangingwall of stope A and from consideration of the quantities of hanging-wall rock to be mucked at the drawpoints from respective openings clearly indicated that the cable reinforcement played an active role in limiting the dilution of the ore by rock from the hanging-wall. An illustration of the displacements of the hanging-wall of stope B as a result of the completion of mining the transverse pillar 1 (the secondary stage of mining), is given in Figure 22. It is evident from Figure 22 that the rock surrounding the orebody is affected for at least 100 m (5 x the width of the orebody) out from the orebody. This result was also indicated by the finite element modeling of the orebody and the surrounding rock, preceding the mining trial. Again the importance of the drilling and blasting procedure should be recognized so that blast damage to the surrounding rock and unnecessary ore dilution can be avoided. The final mass blast of the crown above stopes A and B and pillar 1 (the third stage of mining), included some 292 000 tons of ore. The recovery of this last support of the hanging-wall (at the southern end of the orebody), resulted in some significant displacement of the hanging-wall, an effect that extended at least 100 m out into the hanging-wall. An illustration of the displacements of the hanging-wall of stope B as a result of the mass blast of the crown, is given in Figure 22. The blast also resulted in a raveling failure of the caved rock above the crown. This failure extended up to the surface. No significant new fractures (tension cracks) could however be observed in the hanging-wall at the surface outcrop, although some initially loose parts of the hanging-wall failed as a result of the blast.
510
Support
The sudden unloading of the hanging-wall by the mass blast of the crown could also be registered by several stress monitoring gauges (HI-cells) that were installed before the blast. The increase in the stress level was particularly noticeable in the remaining central pillar (pillar 2), where the increase in the magnitude of the maximum principal stress was of the order of 10 MPa from a premining level of 11 MPa. The direction of the stress field was also aligned perpendicular to the orebody. Initially, the maximum principal stress direction had been approximately parallel to the orebody. However, the overall effect of the mass blast on the rock mass surrounding the orebody was small throughout the mine. The mass blast was therefore very successful from the point of view of stability. The combined rate of dilution for the total of 1.32 million tons of ore product during the mining was approximately 4%. It remains the opinion of the author that the main reason for the very stable conditions encountered in the primary stage of mining (almost no dilution was recorded by rock from the hanging-wall), was the drilling and blasting procedure adopted. The use of an advanced drill-rig for the blasthole, enabled the boreholes to be drilled very accurately. This, in combination with the system of continuous sampling of the cuttings allowed the mining engineers both to avoid charging holes or portions of holes that were located outside the orebody, and also to make special charging arrangements for boreholes close to the hanging-wall contact.
17.6 SUMMARY AND CONCLUSIONS (i) Due to the imprecise nature of existing design criteria for mine layout design and stability assessment, it is not possible to arrive at a final design prior to the commencement of mining. In these circumstances, the best strategy is to adopt a mining method in which the details of the layout (crown pillar thickness, etc.) do not have to be specified prior to the final stages of mining, and in which monitoring can assist in making this final decision. (ii) A detailed instrumentation program is required to provide enough information to enable the rock mass behavior to be reliably monitored. The observations must be sufficient, and extensive enough, to allow stability to be reliably assessed. (iii) A multistage method is best suited for the 'design by monitoring' strategy. It allows experience gained in stable primary mining conditions to be applied in optimizing ore extraction in latter stages of mining. (iv) The results of numerical and analytical modeling assist in the design process - they are not the final answer as they are based on idealizations and must be refined. (v) Because of the uncertainties involved in designing a mining layout and sequence, it is strongly recommended that a mining trial be incorporated into the initial development work. Experience from the trial can be used to calibrate and validate the initial models. Careful monitoring of the trial is very important to learn as much as possible during the early stages of a mine's life. (vi) In an open stoping method, blastholes drilled parallel to the hanging-wall produce more stable conditions than fan-drilled holes. The amount of development work associated with hanging-wall parallel drilling is also smaller. (vii) Careful control of blasting is vital to minimize damage to the walls of production and service excavations. In this regard, careful attention should be paid to drilling accuracy and to reducing charge weight near these areas. (viii) Regional support in the open stoping method must be provided by pillars, as artificial support (cable bolts, or other types of rockbolts) cannot provide stabilization on a large scale. (ix) Prereinforcement of rock is the best way of preserving its integrity during later mining activities. Also, consideration of the development sequence is important in this regard to ensure timely installation of support. (x) Cable bolts are an ideal support element to use in open stoping. Long lengths are relatively simple to install. Installation of cable bolts prior to mining is essential as they are passive elements, and carry load only as a result of rock deformation. Cable bolts are particularly useful in supporting open stope hanging-walls. (xi) When selecting displacement monitoring instruments, guidance can be provided by numerical modeling results. In areas where small displacements are anticipated, but which may be of importance to stability (such as the crown pillar), extensometers of high resolution should be used, such as rod extensometers. In areas where large deformations are expected, lower resolution instruments can be used, but they must have a higher resistance to being damaged by deformations. Wire extensometers are practical in this latter context.
Rock Mass Response to Large Blast Hole Open Stoping
511
17.7 REFERENCES 1. Parâk T. The origin of the Kiruna iron ores. Ph.D. Thesis, Stockholm, Sweden (1975). 2. Leijon B. and Stillborg B. A comparative study between two rock stress measurement techniques at Luossavaara Mine. Rock Mech. Rock Eng. 19, 143-163 (1986). 3. Leijon B. Bergspänningsmätningar i LKAB:s Gruvor 1979-1980. Teknisk Rapport 1981: 67T. University of Technology, Luleâ, Sweden (1981) (In Swedish). 4. Brown E. T. and Ferguson G. A. Prediction of progressive hanging-wall caving, Gath's Mine, Rhodesia. Trans. Inst. Min. Metall. Sect. A 88, A 92-105 (1979). 5. Stephansson O. Pillar design for large hole open stoping. In Proc. Int. Symp. Large Scale Underground Mining, Luleâ, Sweden, pp. 185-197(1985). 6. Agapito J. F. and Hardy M. P. Induced horizontal stress method of pillar design in oil shale. In Proc. 15th Oil Shale Symp. Colorado School of Mines, Golden, USA, pp. 179-191 (1982). 7 Hoek E. and Brown E. T. Underground Excavations in Rock, p. 314. Institution of Mining and Metallurgy, London (1980).
18 Coal Mine Support Systems BARRY N. WHITTAKER Formerly of University of Leeds, UK 18.1 INTRODUCTION 18.1.1 Geological Setting and Interaction with Mining 18.1.2 Underground Coal Mining Extraction Processes 18.1.3 Role and Importance of Support in Coal Mining Operations
514 514 514 514
18.2
515
METHODS OF WORKING AND MINE LAYOUT ASPECTS
18.2.1 18.2.2 18.2.3 18.3
515 515 515
Principal Methods of Working Room and Pillar Mining Longwall Mining Layouts
STRENGTH AND DESIGN OF PILLARS FOR ROOM AND PILLAR MINING
18.3.1 18.3.2 18.3.3 18.3.4
Strength of Coal Pillars Pillar Height to Width Ratio Influence on Strength Role of Barrier Pillars in Room and Pillar Layouts Strength of Rectangular Coal Pillars
519 519 519 520 521
18.4 CHAIN PILLAR DESIGN CONSIDERATIONS FOR LONGWALL MINING 18.4.1 Nature of Chain Pillar Loading: Single Line of Square Pillars 18.4.2 Nature of Chain Pillar Loading: Double Line of Square Pillars 18.4.3 Chain Pillar Loading: Single Line of Rectangular Pillars 18.4.4 Chain Pillar Loading: Double Line of Rectangular Pillars
521
18.5 BARRIER PILLARS BETWEEN SUCCESSIVE LONGWALL FACES 18.5.1 Reasons for Barrier Pillars Between Longwall Faces 18.5.2 Gate Roadway Closure Associated with Longwall Working 18.5.3 Barrier Pillar Influence on Stability
523 523 525 525
18.6
526
LONG WALL STRATA PRESSURE ABUTMENTS AND THEIR INFLUENCE ON STABILITY
18.6.1 18.6.2 18.7
Strata Pressure Abutments Associated with a Longwall Face Critical Zone for Development Headings Adjacent to Previous Longwall Extraction
LONG WALL ADVANCE SYSTEM: GATE ROADWAY STABILITY
18.7.1 Influence of Position of Formation of Gate Roadway 18.7.2 Gate Roadway Support 18.7.3 Gate Roadway Support Resistance and Closure Aspects
521 523 523 523
526 527 528 528 529 530
18.8 SUPPORT OF COAL MINING TUNNELS 18.8.1 Comparison of Main Support Types 18.8.2 Rigid Steel Support Systems in Mine Roadways 18.8.3 Yielding Steel Support Systems in Mine Roadways 18.8.4 Rectangular Steel Support Systems in Mine Roadways 18.8.5 Rock Bolting Support Systems in Mine Roadways
530 530 531 532 533 535
18.9 SUPPORT OF LONG WALL FACES 18.9.1 Support Principles in Relation to Longwall Operation 18.9.2 Generation of Strata Loading on Face Supports 18.9.3 Loading on Face Supports: UK Coal Industry Approach 18.9.4 Strata Loading Determination in Level Seams 18.9.5 Strata Loading Determination in Steep Seams 18.9.6 Influence of Support Design on Roof Control
536 536 537 538 538 538 540
18.10 CONCLUSION
541
18.11
541
REFERENCES
513
514
Support
18.1 INTRODUCTION 18.1.1 Geological Setting and Interaction with Mining Underground coal mining operations involve the systematic extraction of coal deposits which are usually tabular but can be encountered as an irregular massive deposit in some parts of the world. Coal is commonly encountered as a seam deposit which can be flat or inclined. Most coal seams exhibit a fairly consistent thickness although this may not always be the case especially where geological erosion has occurred during the formative period of the coal deposit. Tectonic movements can also result in coal seams exhibiting varying thicknesses especially in association with markedly folded strata. Coal deposits occur in sedimentary rock formations by virtue of the nature of their origin. A further important factor is the manner in which such deposits successively congregate to form economically recoverable mineral reserves. The rock types forming part of the coal mining deposit are virtually always of the sedimentary type also. These are usually mudstones, shales, sandstones, limestones and marls with varying combinations of each depending upon the geological age, location and the conditions which prevailed during the early formation of the coal seams. In some countries, notably in Europe, there is a predominance of mudstones and shales with significantly less proportions of the other common sedimentary rock types in the Carboniferous Coal Measures, whilst in India, there is a predominance of sandstones associated with the strata accompanying the coal seams of similar age to those referred to in Europe. Coal deposits differ widely in age, and consequently this has an important influence on the type of sedimentary rocks which occur as part of the overall sequence. Some coal deposits are also associated with igneous or metamorphic rock types especially as basement rocks. The most common occurrence of igneous rocks within coal-bearing strata, however, is as a result of intrusion, especially as sills and dykes. Coal measures frequently exhibit the effects of earlier geological movements and other processes. Faulting and jointing patterns exhibit marked regularity to the extent that such features can be taken into account during the planning stages when determining the direction of working and support type be employed. The effect of geological pressures explains the variations in coal rank (quality as related to carbon content) as encountered with some coals of the same age. 18.1.2 Underground Coal Mining Extraction Processes The coal extraction process is based on mechanical cutting employing machines which rapidly excavate the coal in situ and incorporate loading onto a coal clearance system which can be a conveyor, free-steered vehicle or scraper mechanism, etc. In steeply inclined seams, however, gravity may play an important role in the coal clearance process. In some countries, however, drilling and blasting or even hand-working of coal may still prevail and can be encountered on small-scale projects. A major consideration in the design of the coal mining system, is the choice of method of working. For underground coal mining, the choice generally lies between room and pillar operation or that of longwall extraction. There are variations of these systems which are frequently given local names, whilst various combinations of the two principal methods also occur. A major feature of the room and pillar method is the formation of pillars which are aimed at giving support to the overburden and consequently planned to prevent deliberate caving of the overlying beds and inducement of a widely spread subsidence trough. As a consequence, coal is left in the form of support pillars which can sterilize significant quantities of the coal reserves. The longwall method of coal extraction generally deliberately induces caving of the roof beds over the extracted area and results in the formation of a subsidence trough at the surface. This method of working allows a greater percentage (50-100%) of the coal reserves to be recovered as compared to the room and pillar method which generally achieves up to 50% extraction but frequently only 30-40% overall, although in exceptional conditions, it can be greater. 18.1.3 Role and Importance of Support in Coal Mining Operations Support is an important factor which governs the degree of success of all coal mining systems. In the case of room and pillar operation, the size of the coal pillars is a principal factor in the design of the mine for its safe operation since the pillars need to provide effective support to the overburden.
Coal Mine Support Systems
515
The rooms and intersections of this method of working need to be supported to prevent collapse, at least during the operational life of the mine. Where coal pillar extraction forms part of the method of working, special consideration needs to be given to the support of the workings at the edges of the depillaring operations. The longwall method of extraction involves the systematic removal of the seam and deliberate caving behind the working face. This mining operation is only feasible with effective support of the face line with a purposely designed support system which controls the movement of a canopy of undermined roof beds which are allowed to collapse, or converge, behind the face line as the longwall moves progressively forwards after successive cuts are taken. Roadways serving the longwall extraction need effective support in order to ensure security of access and egress at all times by the face personnel, in addition to the need to maintain the openings for ventilation and transportation purposes. Support of coal mine operations is governed by legislation in several countries. Additionally, the history of coal mining has witnessed the effects of insufficient or ineffective support in terms of lower standards of safety. The important role played by support systems in allowing coal extraction to be carried out whilst promoting safe standards has been clearly demonstrated in recent years and is now part of modern mining practice. Selection and design of support systems in coal mines require consideration to be given to a wide range of factors, ranging from geological setting, method of coal extraction and overall working plan, to that of the degree and duration of the support required. 18.2 METHODS OF WORKING AND MINE LAYOUT ASPECTS 18.2.1 Principal Methods of Working Peng [1] has given a detailed account of coal mine ground control and discussed aspects of underground mine design and layout. Peng has focused attention on underground design and support practices which prevail in the North American coalfields, and gives examples of typical layout features. The main underground coal mining methods are categorized by Peng under room and pillar mining and longwall mining. 18.2.2 Room and Pillar Mining Figure 1 shows a typical coal mine room and pillar layout. The mine is developed by driving main headings, from which flank districts are formed and separated by barrier pillars. The main heading developments frequently have rectangular-shaped pillars in order to reduce the number of intersections, thereby increasing the potential rate of development. Rectangular pillars are also used for reasons of increased stability of such workings, in view of their need to provide security of access and egress over significantly longer periods of time than the flank districts. The room and pillar flank developments usually employ square pillars and four-way intersection configurations in view of attaining the maximum extraction with ease of operation in developing the district. Staggering of room intersections to create T-configurations considerably increases the stability of the junctions against the risk of collapse. A further important feature associated with room and pillar mine layouts is the adoption of the chevron configuration to ease the problem of free-steered machines negotiating corners in restricted space conditions. Where pillar recovery is carried out, the continuous miner is employed to remove lifts in a set sequence. Coal fenders, or coal pillar stumps, representing a yielding support system are left to provide limited local support during the pillar extraction operation. Wood props and cribs (woodenlatticed structures providing increased areal support over that of props) are also used to assist local roof control especially in providing support and encouraging caving of the roof along preselected breaking-off lines. Pillar extraction poses special roof-support problems where continuous miners require large spans of exposed and unsupported roof areas. The caving of the roof in pillar extraction areas is often unpredictable, and can result in large roof cave in dislodging the props at the edge of the depillaring area. There is a significant measure of uncertainty regarding roof control in pillar extraction operations, and this can lead to substantial losses of coal reserves. Pillar extraction with continuous miners is hazardous and calls for appreciable vigilance especially-during impending roof cave in which may be difficult to detect by the machine operator.
516
Support (α)
Figure 1 (a) Room and pillar coal mine layout aspects, (i) Illustrating general layout with development headings from which flank room and pillar mining districts are formed, (ii) The development headings are shown to have rectangular pillars to facilitate rapid development and reduce the number of intersections which, as a result, promotes stability, (iii) Theflankroom and pillar mining districts are separated by barrier pillars to isolate each district. This promotes safety against collapsed roof areas in depillared workings encroaching into adjacent current workings. It is also advantageous to isolate such districts if there is a risk of spontaneous heating of the coal, or other hazardous conditions, (iv) Depillaring can take different forms with continuous miners. Supplementary supports using props and cribs are essential during depillaring in order to give protection against roof collapses occurring within the working area, (b) Room and pillar coal mine layout aspects. Staggering of pillars allows formation of T-junctions which are significantly more stable than four-way intersections, (c) Chevron room and pillar layouts ease movement around the pillar corners of free-steered vehicles in the relatively restricted space conditions. The shorter diagonal of the pillar governs the pillar strength. Additionally, failure of the slender pillar corners can create problems.
18.2.3 Longwall Mining Layouts Longwall mining can be placed under two broad categories: advance or retreat working. Most European coal mining countries employ longwall advance mining, particularly for reasons such that the retreat headings may not be sufficiently stable at significant mining depths in relatively weak strata conditions. There is a need for rapidly developed headings with longwall retreat mining, which results in the invariable adoption of rectangular-profiled roadways. Such a profile is more vulnerable and sensitive to strata-loading effects, than by comparison to the arched profile. Consequently, this has been a limiting factor regarding the wider adoption of the retreat method. Rock bolting increases the scope for wider application of retreat mining owing to this form of support allowing stability to be substantially improved in rectangular openings by comparison to that attained with props and girders. The support operation of a rectangular profile is greatly simplified and, consequently, eases the face-end operations with longwall retreat working. Increased and more consistent coal production rates are attained with longwall retreat mining and, consequently, its adoption should be considered as the primary choice. It follows that careful assessment should be made in respect of selecting a support design which gives satisfactory roadway stability consistent with the expectations of longwall retreat mining. Longwall advance mining will continue to be employed in those conditions where the success of retreat mining may be in doubt owing to strata pressure effects or other factors. Figure 2 shows a comparison of common longwall layouts employed in various countries. Pillars invariably play an important role in controlling the effects of strata pressure abutments interacting with successive longwall faces which can cause increased roadway closure and decreased stability in the proximity of the second longwall face.
Coal Mine Support Systems (a)
517
(b)
(d) (c)
(e)
(f)
(h)
Figure 2 Longwall layout configurations (after Whittaker [2]). (a) Longwall advance with wide rib pillar, (b) longwall advance with narrow rib pillar and ribside pack, (c) longwall advance with reuse of previous gate, (d) longwall retreat with rib pillar, (e) Z-system of longwall operations, (f ) longwall retreat with total extraction, (g) longwall retreat with pilot heading, and (h) longwall retreat with intervening chain pillars.
Choice of layout needs to take into consideration the physical environment of the seam to be worked in addition to mining and geological constraints. No one system is universally predominant. The systems shown in Figure 2 give an appreciation of the range of layouts together with associated factors influencing their choice of application. Figure 2(a) Longwall advance with wide rib pillar, (i) This layout which is common practice in UK coalfields affords protection of both gate roadways on each side of the rib pillar, (ii) The rib
518
Support
pillar width should be of the order 0.08-0.1 x depth below the surface, based on research findings of Whittaker and Singh [3]. (iii) The rib pillar provides protection for longwall isolation Figure 2(b) Longwall advance with narrow rib pillar and ribside pack, (i) Allows excellent recovery of reserves but can introduce increased complexity into the longwall operation, (ii) A narrow strip of solid coal promotes roof control around the face end and is usually not less than 3-5 m wide; the pack is generally 2-3 m wide. Figure 2(c) Longwall advance with reuse of previous gate, (i) The gate being reused has already experienced closure behind the first face due to subsidence effects and stress redistribution. Consequently, its support system needs to be designed to cater for the increased degree of closure, (ii) Further closure is experienced by virtue of the second face's front abutment pressure zone and subsequent goaf settlement behind the second face. As a consequence of these three distinct closure phases, the roadway is likely to experience substantial closure/instability which calls for special attention being given to the initial design and support. Figure 2(d) Longwall retreat with rib pillar, (i) This form of layout is commonly applied for retreat mining in Europe; it incorporates a high degree of stability for roadway protection and face-end support operations, (ii) The gates serving the second face are frequently of rectangular profile in order to allow high development rates to be achieved, although arch profiled roadways are occasionally used where increased stability is desirable but this reduces the development rate, (iii) The width of pillar between successive faces requires to take account of the strata pressure abutment zones. For rectangular headings, a rib pillar width of 0.1 x depth surface is found to give adequate protection whilst for arch profiled headings an offset distance of 0.05 x depth as been employed with success as the latter roadway is located within the yield zone of the flank strata pressure abutment zone of the first face. Figure 2(e) Z-system of longwall operation, (i) This is a combination of longwall advance and retreat mining, which allows partial proving of the area prior to longwall extraction, (ii) Reuse of the gate roadway is possible as a second means of access, although it is usually abandoned owing to the excessive closure it experiences behind the longwall which is in effect in the centre of a goaf region, (iii) Overall extraction is high with economical use of gate roadway formation and usage, (iv) There is increased blocking out development work by comparison to other forms of longwall layout. Figure 2(f) Longwall retreat with total extraction, (i) This method which has been used extensively in the USSR and Germany is strongly favored in steep seam mining situations. The lower gate facilitates the discharge of coal from the longwall into the roadway without disturbing the roadway supports in steep seam conditions, (ii) This method has benefits in working seams subjected to spontaneous heating problems in view of minimizing the amount of coal left in the extracted areas, (iii) Coal pillars are not generally left and consequently reduces the effects of interaction between coal seams lying within close proximity to each other. Figure 2(g) Longwall retreat with pilot heading, (i) This layout is favored in deep mining situations particularly where the roadway supports need to be able to withstand high strata loading pressures and significant deformation, (ii) The development heading is located in a highly stressed situation and consequently needs to be supported, usually by steel arches to achieve adequate stability, although rock bolting can improve stability in such conditions, (iii) The rib edge is usually extensively fractured and requires special support considerations particularly at the cross-cut intersections, (iv) The chain of small crush pillars commonly has a width of 8-12 m with the cross-cut centers being at about 30 m. The pilot heading permits convenient access to the mining operations and eases congestion around the face end. Figure 2(h) Longwall retreat with intervening chain pillars, (i) This method is employed in the USA and other countries particularly in respect of convenience of development with multientries employing continuous miners which allow rapid development coupled with highly efficient mining operations, (ii) The pillars need to be designed to permit increased strata loading resulting from caving of the goaf regions on both sides of the lines of chain pillars, (iii) Additional support measures are normally required within about 30 m of the face line and especially at junctions owing to the travelling strata pressure abutment zone activating rock movement in such localities. The application of longwall retreat mining with intervening chain pillars is a layout favored in the USA and some other countries. It is particularly attractive in respect of convenience of development by virtue of employing multientries with continuous miners. The multientries are advantageous in respect of ease of access and egress, but can be disadvantageous if the roof is prone to collapse owing to the large number of room intersections. The staggering of intersections to form T-junctions and use of supplementary supports within the strata pressure abutment zone traveling ahead of the face line, can significantly assist in promoting stability and reducing the risk of roof collapses as this can considerably hinder longwall operation.
Coal Mine Support Systems
519
18.3 STRENGTH AND DESIGN OF PILLARS FOR ROOM AND PILLAR MINING 18.3.1
Strength of Coal Pillars
The strength and design of coal pillars has been discussed by Salamon and Munro [4], Salamon [5] and Whittaker [6]. Salamon reported the findings of an investigation which covered 125 case studies of which 27 represented collapsed pillar areas in South African coal mines. The depth range was 20-220 m, room height range 1.2-5.5 m, pillar width range 2.7-21.3 m, percentage extraction 37—91%, and pillar width over height ratio 0.9-8.8. Following an analysis of the stable and collapsed coal pillar cases in these 125 room and pillar mines, Salamon established a formula which expresses the factor of safety of the pillar against collapse in terms the principal physical and mining factors. This basic formula is expressed here in SI units and is given as equation (1). Figure 3 illustrates the symbols pertaining to equation (1) S = [(7180νν°·46)//ι°·66]/{[23//(νν + B) 2]/w2}
(1)
where S = factor of safety against collapse of pillar, w = width of square pillar (m), h = pillar height (m), H = depth of overburden (m), and B = room width (m). Pillar strength = (7180 w°'46)//i°'66(kNm-2) Pillar stress = ygH(w + B)2/w2 = 23J/(w + ß) 2 /vv 2 (kNnr 2 ) for y = 2350 k g m " 3 and g = 9.81 m s" 2 . When the factor of safety S = 1, then this represents the critical condition between the collapse (for S < 1) and stable (for S > 1) states for the pillar. Field results reported by Salamon [5] indicated a degree of overlap generally being encountered for S < 1.3, whilst one collapsed case was reported to have occurred at between S = 1.4-1.5.
18.3.2
Pillar Height to Width Ratio Influence on Strength
The pillar height to width (smaller side dimension in the case of rectangular pillars) ratio plays a major role in the pillar's strength-deformation behavior particularly after the limit of proportionality on its stress-strain characteristic. The results presented in Figure 4 show that, for a pillar height to width ratio of 1:4, the pillar has a rising strength characteristic after the limit of
Figure 3 Strata loading of a square pillar in a typical room and pillar mine layout: illustrating the symbols used in respect of equation (1).
520
Support m x p x L (m) 0.025x0.1x0.15 (m)
Pillar vertical strain, -— Figure 4 Load-deformation characteristics for a rectangular pillar of different heights (after Whittaker [7])
5000
0
0.25
0.5
0.75
1.0
Pillar height / width ratio,
1.25
m/p
Figure 5 Influence of height/width ratio on pillar strength (after Whittaker [7])
proportionality is reached. Conversely, for ratios of 1:2 and 1:1 the pillar strength characteristic decreased after reaching the limit of proportionality. The experimental results referred to here were obtained using small scale pillars of plaster of Paris material. The influence of pillar height to width ratio is clearly evident in Figure 5. Squat pillars appear to achieve considerable strength by virtue of achieving a constrained core which may have failed earlier and resulted in reconsolidation.
18.3.3 Role of Barrier Pillars in Room and Pillar Layouts When designing room and pillar layouts, barrier pillars should be incorporated as these encourage the overlying beds to bridge across extracted areas, and thereby reduce the vertical loading carried by the pillars. This method has been observed by the author to be highly effective in a number of room and pillar coal mines in different countries, especially in the USA, where barrier pillars form the main support elements of a given layout. As a consequence, the room and pillar districts are correspondingly more effectively protected against pillar collapses occurring. Should
Coal Mine Support Systems
521
pillar collapse occur in such a room and pillar district, then the progressive collapse process would be arrested by the barrier pillars. 18.3.4 Strength of Rectangular Coal Pillars The pillar strength factor of safety formula given as equation (1) is based on square pillar configurations. Salamon [5] has recommended that where rectangular pillars are used, then the smaller pillar side dimension should be employed in equation (1). This design recommendation has been confirmed by experimental work carried out by Whittaker [7]. Figure 6 shows the results of experimental work carried out on model rectangular pillars and demonstrates that the pillar strength is in effect mainly governed by the smaller pillar side dimension. 18.4 CHAIN PILLAR DESIGN CONSIDERATIONS FOR LONGWALL MINING 18.4.1 Nature of Chain Pillar Loading: Single Line of Square Pillars Chain pillar design aspects have been discussed by Whittaker [7] who extended the basis of load transfer from longwall extracted areas to intervening pillars. Loading of continuous rib pillars located between longwall extractions has been discussed by King and Whittaker [8]. In the case of chain pillars, however, their stability aspects need to be considered individually. Load is assumed to be transferred from an adjacent goaf area as shown in Figure 7(a). The engineering basis of the assumptions used here is consistent with subsidence compatibility. With reference to Figure 7(a) (a) Pillar height, m ■ 0.1 m
1400 1200 1000 800
Strength based on p
600 CVJ E
z
w
x:
o Actual test value
200^ 0
■ -^
J
0.5
I
1.0
I
1.5
ÉT a>
I
2.5
I
I
3.0
3.5
I
4.0
I
4.5
L
5.0
L/p
Pillar length/width,
(b)
2 £
I
2.0
2000 1800 1600 1400 1200 1000^ 200 "0
I
0.5
I
1.0
I
1.5
I
2.0
I
2.5
I
3.0
I
3.5
I
4.0
I
4.5
I
5.0
Pillar length/width, L/p
Figure 6 Pillar strength in relation to increasing ratio of pillar side dimensions, (a) Comparison of actual and predicted pillar strengths (m = 0.1 m), and (b) comparison of actual and predicted pillar strengths (m = 0.05 m) (after Whittaker [7])
Support
522 (α)
ip+B)
h-
(2p+£)
H
Figure 7 General representation of (a) single and (b) double rows of square chain pillars (after Whittaker [7])
which represents a single line of chain pillars, then the pillar load is given by equation (2) Pillar load = 9.81y[(p + w)h - (w2cot<£)/4](p + B)
(2)
where w/h < 2 tan φ. With respect to equation (2), the total volume of rock resting on the pillar is [(p + w)h - (w2 cot )/4] (p + B) whereby the goaf triangle is assumed to be completely relieved. As a consequence, its loading is transferred to the extracted region and the triangular section is governed by an angle of shear φ which has been indicated to have been approximately 31° for UK coal measures rocks in view of its compatibility with subsidence principles. Equation (1) corresponds to a subcritical subsidence condition. For critical and supercritical mining subsidence conditions, the goaf triangle will reach the surface boundary, so that the chain pillar will be loaded by an inverted frustum of a cone. The new pillar loading situation is given by equation (3) Pillar load = 9My\_ph + (fc2tan)](p + B)
(3)
where w/h > 2 tan φ. The respective average pillar stresses (aav) for these two conditions are given by equations (4) and (5) for single rows of square pillars )/4][(p + £)/p2]
(4)
Coal Mine Support Systems
523
where w/h < 2 tan φ, and *av = 9Mylph + (/i2tan0)][(p + 5)/p2 ]
(5)
where w/h ^ 2 tan φ. 18.4.2
Nature of Chain Pillar Loading: Double Line of Square Pillars
The mining situation assumed here refers to that shown in Figure 7(b) and the only physical changes are (i) the effective width of the chain pillar support area becomes (2p + B) and (ii) the load is borne by two pillars of plan area 2p2. The respective average pillar stresses are given by equations (6) and (7) for the double row of square pillars
(6)
where w/h < 2 tan φ, and ffav = 9.81y [(2p + Bh + h2 tan φ)(ρ + ß)/2p2]
(7)
where w/h > 2 tan φ. The symbols used in equations (2) to (7) are defined as follows: w = width of longwall face (m), h = depth below surface (m), y = average density of overburden (kg m~ 3 ), φ = average angle of shear (°), p = pillar width (m), B = bord (heading or room) width (m), and σΛν = average pillar stress. 18.4.3
Chain Pillar Loading: Single Line of Rectangular Pillars
Figure 8(a) shows a representation of this mining situation for which the average pillar stress (σ3ν) is given by equations (8) and (9) *av = 9.81y[(p + W - (w2cot fl/4][(L + B)/pL]
(8)
where w/h < 2 tan φ, and tfav = 9.81y[(pfc + fc2tan ψ)(Ζ, + B)/pL]
(9)
where w/h > 2 tan φ. 18.4.4
Chain Pillar Loading: Double Line of Rectangular Pillars
The average pillar stress (σ3ν) for this configuration of rectangular pillars is established similarly to that given in the previous section. Equations (10) and (11) give
(10)
where w/h < 2 tan φ, and *av = 9.81y[(2p + B)h + h2 tan φ] [(L + B)/2pL]
(11)
where w/h > 2 tan φ. 18.5 BARRIER PILLARS BETWEEN SUCCESSIVE LONGWALL FACES 18.5.1
Reasons for Barrier Pillars Between Longwall Faces
Barrier pillars are commonly left between successive longwall faces mainly for reasons of: (i) the pillar affords roadway protection against the effects of excessive closure, (ii) the control of surface subsidence, (iii) the isolation of successive longwall extractions which is especially important where fire and gas hazards need consideration, and (iv) the increased control of the immediate roof strata
524
Support (α)
β
μ
"1
.LÎ 5
A
.
(b)
h Figure 8
(ZP + B)
H
Rectangular chain pillar configuration, (a) single row and (b) double row (after Whittaker [7])
adjacent to geological disturbances such as faults or where massive sandstones or igneous intrusives present special problems to the current longwall working [3]. Barrier pillars are also required by law for purposes of providing adequate and safe isolation of a mine from an adjacent mine which may have been abandoned and could be flooded or represent some other form of hazard. The current legal requirement in UK coal mines is for a barrier pillar of 37 m minimum width to be left between mines. Most longwall layouts employ barrier pillars (the term 'rib pillar' is also commonly used) for the purpose of achieving improved strata control in the immediate proximity of the gate roadway adjacent to the pillar. The question frequently arises, however, as to the width of pillar necessary to achieve an acceptable degree of control over strata movement and subsequent closure effects on the roadways which, of course, is related to the economics of longwall extraction. Leaving excessively large barrier pillars results in loss of valuable coal reserves by virtue of sterilization. There are occasions, however, where designing a layout with a narrow pillar which has
525
Coal Mine Support Systems
been purposely designed to crush (commonly referred to as a 'crush pillar') during the second longwall face working phase, can give acceptable mining conditions, providing the roadway support system is appropriately designed [3]. 18.5.2 Gate Roadway Closure Associated with Longwall Working Figure 9 shows the two basic designs of European longwall layout for advance and retreat working, together with observed closure phases associated with the face operation, stress readjustment and caving/goaf settlement. The observations were made in the UK Midland Coalfields. The initial closure phase for longwall advance working does depend, however, on the position of formation of the gate roadway. If the roadway were formed ahead of the faceline, as with the advanced heading technique, then significantly more closure would be experienced. Conversely, forming the roadway appreciably behind the face line would result in less roadway closure owing to the formation point being located in destressed ground and further from the position of greatest strata pressure activity, i.e. at the faceline. Figure 9 shows a final closure phase which is usually linked with time-dependent deformation, depth below surface and rock types; it does not necessarily occur in all situations, especially in shallow conditions and where strong rocks exist immediately above and below the coal seam. 18.5.3 Barrier Pillar Influence on Stability The results embodied in Figure 9 cover a depth range of 100-900 m, face width range of 130-275 m and rib pillar width range of 0-400 m [3]. These results were also used to examine the influence of width of barrier pillar on pillar stress and gate roadway closure. Table 1 presents representative geological strength data of the associated rock types involved with the longwall roadway situations studied by Whittaker and Singh [3] in UK coalfields. Barrier pillar loading has been investigated and subsidence theory applied to the rib pillar situation in order to relate strata pressure arching across longwall extractions, thereby producing loading of the adjacent barrier pillars [8, 9]. The basic assumptions regarding loading of a barrier
(a)
v,. Wmmammm Longwall advancing
r*-Final closure-^ phase |—— Initial closure phase—+\ 121.4 ± 78.4 m Initial closure
_i
Distance from face line
Final closure
t_
(b)
1
Longwall
I*-Initial closure —< phase 50.9 ± 4 5 5 m
retreating
Initial closure
Distance from face line Figure 9
Illustrating development of gate roadway closure for (a) longwall advance method and (b) longwall retreat method (after Whittaker and Singh [3])
526
Support
Table 1 Representative Geological Strength Data of the Associated Rock Types with Longwall Roadways Studied. General Cyclothem Sequence of Carboniferous Coal Measures Strata Encountered in British Coalfields (after Whittaker and Singh [3])
Rock type Sandstone Siltstone Mudstone Coal Seatearth Sandstone (next sequence)
(General thickness (m) Range Ao.
Uniaxial comp. strength (MPa)a Range Ao.
Tensile Young's strength (MPa)b modulus (GPa)c Range Av. Range Ao.
Poissorfs ratiod Av. Range
4 2.5 6 1.5 1
80 50 30 28 25
5 4 3 <2 ?e
0.2 0.25 0.25 0.3 ?
0-70 0-30 0-70 0-9 0-3
30-150 30-80 15-70 10-50 10-40
3-10 2-10 2-5 \-A ?-4
35 40 20 25 15
20-100 20-140 5-40 5-40 ?-30
0.1-0.25 0.2-0.3 0.15-0.3 0.1-0.4 7-0.4
a
Uniaxial compressive strength performed on cylindrical specimens, 38 mm dia x 76 mm. bTensile strength determined using Brazilian discs of 38 mm dia x 19 mm. cYoung's modulus determined using electrical strain gauges mounted on 38 mm dia x 76 mm cylinders in uniaxial compression. dPoisson's ratio determined using electrical strain gauges as in footnote c. e Where ? is given this signifies material too weak to give satisfactory test results on samples tested.
pillar between successive longwall faces are similar to those discussed earlier in connection with chain pillar design, see section 18.4. The average stress (aav) acting on a barrier pillar has been reported earlier [3, 9] and is given by equations (12) and (13) 9 81v 2 *ν = TT^liCiP cot φ)/4]ρ} I L/ΙΎη + Λ. w) ΛΑΛDn - _ (w iu^r 1000p2X
(12)
*av = j ^ L {plpD + D2 tan ]}
(13)
σ
for w/D < 2 tan φ, and
for w/D > 2tan0. Equations (12) and (13) have been used to examine the role played by longwall width (w) on barrier pillar stress. Figure 10 (a) shows average pillar stress related to pillar width as a function of depth below the surface. This illustration clearly demonstrates a significant change in the average pillar stress when the pillar width is less than 0.1 x depth. The investigation into the influence of barrier pillar width with respect to stability aspects was extended to take into account the resulting gate roadway closure. The results given in Figure 10 (b) represent a depth range of 200-400 m in respect of longwall advancing faces. The roadway closure is expressed in terms of the relative change in height (CH) and width (Cw) compared to the original dimensions as constructed. These stability results further confirm that a barrier pillar width of 0.1 x depth is a reasonable design value for longwall advance mining layouts. Whittaker and Singh [3] have also reported similar conclusions for a depth range of 400-600 m and examined results from situations as deep as 900 m. They have presented confirmatory results for a wide range of conditions. 18.6 LONGWALL STRATA PRESSURE ABUTMENTS AND THEIR INFLUENCE ON STABILITY 18.6.1 Strata Pressure Abutments Associated with a Longwall Face Longwall mining gives rise to a redistribution of strata stresses around the excavation. The longwall face has a strata pressure abutment which travels with the face line and is commonly referred to as the 'front abutment'. There is also a similar concentration of stress along the sides of the longwall, and these are commonly referred to as 'flank abutments'. Figure 11 shows the distribution of vertical stress produced by a longwall extraction. Although an advancing longwall face is shown, the same stress redistribution occurs with the retreating situation. The return to coverload stress within the extracted region behind the longwall depends upon thé width of face to depth below surface ratio and on the caving/subsidence characteristics of the immediate roof and overlying beds. A return to coverload stress generally occurs in European
527
Coal Mine Support Systems (b) 2.0
~
1.5
Longwall face data range.
"\Δ
Λ
E .
Depth range 2 0 0 - 4 0 0 m Mean depth 0 = 335 ± 55 m
\ Δ
1.0
Face width w 1 3 4 - 2 7 5 m Δ Mean width w - 207 ± 42 m
\Δ
" Δ
^ Δ
\
0.5 h-
υ
1 0
1
0.1
a
1
0.2
0.3
0.4
1
1
0.3
0.4
0.5
p/D
2.0 H
Ό
ID
\
D
a>
\
\ \
0.5
0
k
D 1000 m 800 600 400
û_
8 0 \-
lb
60 I-
..
P
0.2
0.1
o
h""H
o\
^ntttttttPi " '
V) V) l_ V)
σ
40
'5.
20
cr
0
p/D
\
\o
0.2
0.3
p/D Figure 10 (a) Rib pillar average stress in relation to pillar width expressed in terms of depth below surface, (b) Gate roadway stability and rib pillar stress plotted against rib pillar width as a function of depth (longwall depth range = 200-400 m) (after Whittaker and Singh [3])
longwall situations where the extraction width to depth ratio exceeds around 1.2. Figure 11 shows an analysis for the average depth of longwall working in the UK, and was performed using a stress balance assumption of compatibility before and after mining. The front abutment strata pressure zone can give rise to coal spalling (sloughing) from the face particularly in thick seams and may require special support measures to control it. The flank abutment strata pressure zone often has an influence on the stability of gate roadways constructed immediately adjacent to the solid rib edge. This may require consideration to be given to the design of support for the gate roadway and its positioning in relation to the rib edge. 18.6.2 Critical Zone for Development Headings Adjacent to Previous Longwall Extraction Figure 12 shows the distribution of the flank strata pressure abutment with respect to the solid ribside and reference to a 'critical zone' which is known to produce deleterious effects to headings located within this region. This is essentially a high strata pressure zone induced by the adjacent longwall extraction and can create stability problems to coal headings. In a number of cases, consideration is given to avoiding locating headings in such zones either by their positioning very
528
Support Face width Depth Width/depth Cover load (/?)
Section XX
_L
=fc
J.
60 80 100 Distance from face line(m) 2 2.:.53 4 4 j Face cen center line
-80
-60 - 4 0
, \p
1.5
40
■2IOm -470 m ■ 0.447 - 10.8 MN m"M1550psO
-20
Cover load
200
Section YY
Figure 11 Distribution of vertical stress produced by longwall extraction showing front and flank abutment zones (after Whittaker and Singh [9])
0.04
0.06
0.08
0.10
0.12
0.14
0.18
Distance from ribside (xdepth)
Figure 12 Longwall strata pressure abutment adjacent to longwall extraction showing position of critical zone for subsequent retreat mining development headings (after Whittaker [2])
close to the ribside edge or at a significant distance into the solid and thereby in effect creating a barrier pillar. The critical zone is indicated to be generally defined as being located at 0.01-0.06 x depth as measured from the ribside [2]. Where headings are located in such a zone, however, due consideration should be given to selection of a support system which caters for the increased stress anticipated. 18.7 LONGWALL ADVANCE SYSTEM: GATE ROADWAY STABILITY 18.7.1 Influence of Position of Formation of Gate Roadway Reference has been made earlier in this chapter in Section 18.6 to the influence of strata pressure abutments influencing the stability of gate roadways in longwall advance mining. Figure 13 shows
Coal Mine Support Systems
529
(a) (b)
(c)
(d)
Figure 13 Illustrating longwall advancing basic face-end designs for formation of gate roadway in relation to face line position: (a) advanced heading; (b) half heading; (c) conventional; (d) in-line system (after Whittaker and Singh [10])
four basic layouts relating to the position of formation of the gate roadway. The advanced heading method results in the roadway being formed within the front abutment strata pressure zone and consequently gives rise to substantially increased roadway closure which needs to be taken into account when considering its mining advantages. The half-heading method of roadway formation allows stability to be substantially improved owing to the fact that the roadway is formed in a destressed zone and is located partially in the solid ribside. The conventional position gives a compromise in terms of gate roadway stability although it tends to impede the mining operations. The in-line system favors mechanization regarding roadway formation particularly in thick coal seams, but its stability may suffer due to its formation in an active strata pressure zone and where pack construction may pose difficulties [10]. 18.7.2 Gate Roadway Support Figure 14 illustrates the general strata deformation and loading characteristic associated with a gate roadway serving a longwall advancing face [11]. The roadway support system needs to effectively control the strata deformation which arises due to the redistribution of strata stresses and the caving of the roof behind the longwall. The gateside pack plays an important role in respect of: (i) providing a strata control function in terms of carrying a significant proportion of the load and allowing bridging to take place between the pack and the solid ribside, (ii) the pack should effectively create sufficient resistance to promote caving of the roof on the goafside, (iii) the pack should prevent encroachment of the caved waste into the gate roadway, (iv) the pack should also serve as an effective seal to prevent leakage of ventilation current short circuiting across the goaf region behind the face line, and (v) where applicable, the pack area is regarded as a convenient site for disposal of unwanted strata debris as usually occurs at ripping lips. Gateside pack construction can be achieved by various methods ranging from a loose handbuilt structure to that of a cementitiously bonded material which is effectively placed by pumping action into a containing bag or other form of containment to form a strong structure. Figure 15 shows reported pack stress values relating to different forms of pack construction as a function of distance behind the face line [12]. The mechanical packing systems generally give a compromise between hand packing and monolithic packing in terms of the final strength of the structure. Monolithic packing enables roof support to be achieved earlier and more effectively than other systems and consequently gives rise to superior levels of roadway stability behind the working face, see Figure 15(b).
Support
530 \
Cover load oV) ΙοααΊηα
K (Y U1U U U U1 Nr,r"D.u.Tr
5» /
Low pressure zone
x x x x x x x x x-ir x x x x x x x x x x x x x x x x x x x Potential extrusion zone
(b)
Figure 14 Illustrating main closure features of a gate roadway serving a longwall a,d^ncing extraction, (a) Gate roadway ITZ.l ^ i . t r i v «ft«· and (b) (b) gate sate roadway roadway condition 63 m from face (after Whittaker and Hodgkmson [11]) condition Jimmediately after formation formation and
18.7.3 Gate Roadway Support Resistance and Closure Aspects Gate roadways in longwall advance mining experience significant closure by virtue of the effects of caving behind the longwall. The arched steel girder support system has predominated in such situations by virtue of its ability to accept significant deformation whilst still retaining appreciable supporting resistance. Figure 16 serves to indicate the effective support resistance for a wide range of gate roadway situations which have experienced varying degrees of roadway closure. The support resistance is predominantly within the range 0.0^0.06 MPa although higher support resistances of UD to 0.1 MPa have been employed in UK coalfields [10]. . Exceptional circumstances have arisen in UK coalfields where the effective support resistance in the roadway serving the longwall face has been substantially greater than the values quoted above. An example is given in Figure 17 of a pilot heading serving a longwall retreat face where substantially increased support resistance gave significantly improved stability. 18.8 SUPPORT OF COAL MINING TUNNELS 18.8.1 Comparison of Main Support Types There is a wide range of support types used in coal mining tunnels and the selection is influenced by such factors as life of the tunnel, durability, rock pressure, degree of stability required and that of cost. Coal mining tunnels are usually characterized by their association with weak rocks which are a cte d f . .^f^. susceptible to deformation under moderate stresses and, in some cases, when Table 2 gives a comparison of the principal support methods and systems used in UK coal mine ^ A l ï e X ^ h a s T e Î f o c u s e d on the design of circular tunnels in Coal Measures rock conditions which prevail in UK coal mines [13]. Pre- and post-failure strength and deformation properties of
531
Coal Mine Support Systems 10 — i
(a)
;Äk wmwtm Ë
20
40
60
80
100
Distance behind face, x (m) 30 i -
(b)
^,0/ÊÊÊÊÊÊS
,ocW\nQi ■
*** 2
.JA
M M
0
20
40
60
_L 80
100
Distance behind face, x (m)
Figure 15 Stress and roadway closure aspects of gateside packing systems in longwall advanced mining: (a) stress build-up in pack for hand packing, mechanical packing and monolithic packing; (b) roadway closure in relation to packing type (after Woodley and Osborne [12])
the rocks have been investigated to provide an engineering database for the purpose of designing the tunnel support system. A commonly encountered problem in coal measures strata is that of the banded character of the rocks which give rise to different degrees of closure into the excavation. By taking account of the stress field and the rock deformational properties, due allowance can be made for the resulting closure and its compatibility with the support strength-deformation characteristics [13]. Concrete supports have been used in mining tunnels where a combination of high strength, durability and a measure of flexibility have been required. They are particularly attractive for application in those situations where conventional steel supports would be unable adequately to control ground movement. Their cost is significantly greater than steel supports such as the common steel arch type and, consequently, their application has been limited to special situations such as longlife drivages at major mines. Concrete linings provide a continuous support system and, as a consequence, can eliminate subsequent dinting and back ripping in roadway maintenance operations. Although monolithic concrete linings have been used in some coal mine tunnels, the preference is generally for segmental linings which include packing between segments for the purpose of incorporating a measure of yield and for providingflexibilityfor the tunnel profile to adjust to the stress field or geological setting. 18.8.2 Rigid Steel Support Systems in Mine Roadways The UK coal industry mainly employs steel supports in mine roadways owing to such advantages as ease and speed of support operation, and the effectiveness which such supports possess in
Support
532
C.C« Correlation coefficient
Arched _ <7 ΛΛsteel girder
k « load factor k = I (point load at crown to produce plastic hinges)
Ί* ' » I —
i · · ·· 0.02
0.04
·
0.06
0.08
0.1
g (χΑΉΜΡα)
I'JLSL·^-^ ^0.5
0.02
0.04
0.06
0.08
<7(x*)(MPa) 1.0 Γ-
0.5
|_
A-AP
•
C.C. = 0.074
0
-#
0.42 + 0.96 <7 ·
_L
0.02
·
♦—
·+£*Ζ 0.04
·
·
0.06
· ··
#
j 0.08
0.1
Effective support resistance, q (x/r)(MPa)
Figure 16 Gate roadway vertical and cross-sectional closure in relation to effective support resistance associated with the roadway, (after Whittaker and Singh [10])
providing stability in a wide range of conditions during the life of the mine [13]. Steel arch support systems can undergo significant deformation and still provide effective support of openings in those conditions subjected to high levels of rock stress and deformation [14,15]. Additionally steel arches have contributed significantly to reducing accidents from falls of ground in mine roadways [16]. The steel arch type of support provides a natural shape which is advantageous in permitting improved accommodation and control of strata displacements into the excavation. It would appear that the most valuable advantage of the steel arch system of roadway support lies in its ability to continue offering significant supporting resistance even after appreciable deformation has occurred. Steel arch support systems are generally used without any attempt to provide constraint to floor movement. As a consequence, there is a frequent need for some subsequent dinting operation for the purpose of maintaining adequate roadway height. Rigid steel arches of the H-section type continue to find considerable favor in UK coal mines. The main reason for this would appear to be because of its ability to provide effective resistance to formation of plastic hinges. This particular feature offers improved supporting strength over other sections. The H-section also has marked advantages in respect of resisting twisting and lateral bending. This promotes its supporting ability under conditions of high deformation and particularly in the gate roadways of advancing longwall faces. 18.8.3 Yielding Steel Support Systems in Mine Roadways Yielding steel arch support systems have been used in UK coal mines for many years. The main yielding feature incorporated has been that of accommodating vertical movement usually by a special design of yielding leg. This has allowed resistance to be offered and support provided to the
Coal Mine Support Systems
533
Previous extraction
40
60
80
Distance from face line, x (m)
100
120
120 2h
yyy/£z^/yyyyy \'tfj\ D = 595 m
Figure 17 Illustrating influence of pilot heading length and support resistance on vertical closure (CH) within the heading of a longwall retreat face (after Whittaker and Singh [10])
roadway without resulting in excessive damage to the support system. A major disadvantage is that of the occurrence of significant lateral movements at floor level which can impair the yielding mechanism of the support and give rise to localized buckling. Incorporation of the yielding mechanism within the steel support as is achieved by using the V-section design enables the support to accommodate both vertical and horizontal displacements with less risk of the yielding mechanism becoming impaired by comparison to the earlier system referred to above. V-section yielding steel arches are easy to erect, and indeed recover when required, in conditions where the supports are fairly lightly loaded and deformed. This form of yielding support system is particularly advantageous in promoting roadway stability especially in view of its ability to continue providing effective support over a wide range of closure conditions. The result is that of improved support to the strata and less overall closure whilst retaining a support structure in its most effective supporting condition [17,18]. Table 3 shows the comparison between V-section yielding and H-section rigid steel support systems. 18.8.4 Rectangular Steel Support Systems in Mine Roadways Rectangular standing supports in the form of sets of RSJ roof beams and end props probably represent the simplest form of support system used in coal mine roadways. They are extensively used in retreat mining where the conditions favor their application. The range of application appears to be limited to reasonably competent rock conditions or where low strata pressures are encountered. Rectangular supports have been designed to give a compromise between the simple rectangular sets consisting of a girder resting on props and the more rigid arch girder. Some designs incorporate a cambered, or straight, rigid roof girder attached to legs by means of specially designed plates.
Support
534 Table 2
General Classification of Support Methods Employed in UK Coal Mine Roadways (after Whittaker, Baxter and Townley [15])
Support type 1 Natural
Geological conditions favoring choice (i) Limited to competent rock, e.g. massive sandstones. (ii) Very rare in UK coal mines owing to lack of suitable natural conditions
2 Rock bolting
(i) Stratified rock formations assist in promoting beam action with bolting. (ii) Weak plastic type rocks require special anchorages. (iii) Rock bolting can contribute to promoting stability in most coal mining conditions.
3 Rectangular steel sets
(i) In-seam mining as with fairly thick coal seams. (ii) Roof beam should have a measure of natural strength which promotes effective spanning. (iii) Low stress conditions assist its success, e.g. shallow workings.
Mining applications
Operational comments
(i) Used in some shaft bottom locations where suitable rock conditions occur. (ii) Occasionally encountered away from active mining areas.
Conditions occur so rarely that it is not generally considered as an option.
(i) Rectangular openings lend themselves to installation of bolting particularly room and pillar working. (ii) Retreat face developments using rectangular roadways. (iii) Effective as supplementary support as in reinforcement work.
(i) Rock bolting must be applied effectively to obtain best results and calls for high standard of installation and practice. (ii) Positive anchorage is of major importance.
(i) Retreat mining developments, (ii) Short-life headings, e.g. face start line.
(i) Simple to transport and quick to install, (ii) Assists high speed developments.
(iii) Geological conditions favoring maintenance of flat roof, e.g. sandstone roofs.
(iii) Susceptible to high strata stresses and large-scale deformations can induce premature failure of support.
4 Rigid arch sets
Gate roadways with significant deformation and high strata stresses.
Most roadways in coal mines especially where other support systems likely to fail.
Proven by years of experience as a very reliable system of support in heavy ground conditions and highly stressed situations.
5 Yielding steel arch sets
(i) Roadways with mainly vertical strata movement.
Several roadways in coal mines except where heavy lateral movements are likely.
(i) High lateral movements can disrupt yielding mechanism and lead to early failure of support. (ii) Difficult to incorporate wide range of struts.
(i) Favors rapid drivage with full face excavation in most coal mine geology conditions, (ii) Mainly used for roadways required for life of mine, (iii) Spine roadways rather than gates and short-term developments
(i) Strong natural support in shape and placed effectively in contact with strata, (ii) Promotes natural stability.
(i) Very strong form of support but cannot accept appreciable yield before cracking. (ii) Long-life roadways in shaft bottoms and some major developments such as spine roadways.
(i) Adequate and effective grouting/backfilling between segtments and rock most desirable. (ii) Point loading can rapidly decrease strength of lining.
(ii) Where weak rocks encountered and a measure of yield is desirable. 6 Circular steel
(i) Heavy ground conditions. (ii) Hydrostatic stress conditions of significant magnitude. (iii) Conditions requiring a measure of yield.
7 Concrete segmental
(i) Heavy ground conditions.
(ii) Hydrostatic stress conditions of significant magnitude.
535
Coal Mine Support Systems Table 2
Continued
Support type
Geological conditions favoring choice
8 Monolithic concrete
(i) General mining geology conditions as encountered in UK. (ii) Low lateral stresses required. (iii) Needs to be fairly free from active strata stresses due to geology or mining.
9 Sprayed concrete linings
Table 3
Most geological conditions including minor water flows where immediate support desirable since it can be applied remotely.
Mining applications
Operational comments
(i) Rectangular roadways in shaft bottom areas needing large cross section and smooth sides. (ii) Underground construction chamber and workshop. (iii) Bunker housing etc.
Ground needs to be relatively undisturbed after construction.
(i) As an immediate temporary support to prevent detachment of small slabs shortly after excavation, and prior to erection of permanent support. (ii) Effective when used in conjunction with rock bolting and mesh and even standing supports.
(i) Specialized operation requiring skilled application.
(ii) Can be applied at any time during tunnel support operations.
Comparison between V-section yielding and H-section Rigid Steel Arch Supports (after Whittaker, Baxter and Townley [15])
V-section yielding supports 1. More expensive due mainly to complex connecting elements 2. Highly resistant to out-of-plane deformation due to comparable section moduli in x and y planes, therefore less need for heavy struts 3. Capable of significant yielding before plastic deformation hinges occur 4. Can be easily dismantled and reerected 5. Flexible size range; supports easily adapted to many strata conrol situations owing to their versatility 6. Excellent joint strength due to the overlapping nature of yielding joints 7. Yield load can be set at higher levels therefore system suited to wide ranging strata loading conditions 8. Difficult to clamp struts and other fittings to the section 9. Their cost has probably discouraged their wider application in the UK coalfields 10. The yielding mechanism is prone to variable yield loads; clamp seizures can be a major problem in certain conditions
H-section rigid supports Less expensive; cheaper connecting elements in the form of fishplates H-section prone to out-of-plane deformation as the ratio of selection moduli in x and y planes is around 3:1 respectively A degree of yield can be catered for by the use of stilts Once deformed, supports are difficult to remove and reerect Flexible size range in both section and support size Weak joint strength especially when supports are severely deformed Lower limit to yielding load can be achieved by incorporating yielding stilts Comparatively easy to strut and attach clamps and various fittings Cheapness and simplicity coupled with familiarity of a well proven support system continues to promote its wide acceptance in UK coal mines Yield loads using stilts can be reasonably confidently predicted
There is a wide range of these superior designs for operation in rectangular roadways and this has considerably increased the scope of application of this profile to longwall mining. 18.8.5 Rock Bolting Support Systems in Mine Roadways Rock bolting support techniques continue to find favor worldwide in rectangular coal mine drivages where strata conditions are suitable for their application. Rock bolting supports lends itself
536
Support
to rectangular mine drivages. This support system improves roadway conditions and promotes stability by virtue of allowing the rock to support itself. The application of rock bolting to weak strata conditions has also had considerable success, but this has called for special attention to be given to employing superior anchorages such as the full column bonded rock bolt, and suitable plates and mesh arrangements at the mouth of the borehole. Rock bolting is probably the most commonly applied support system in North American coal mines in addition tofindingwide use in other countries. The success of its application depends much on its systematic application following fairly rigid guidelines concerning spacing and positioning of bolts within support patterns. In some countries its operation needs to be monitored by suitable instrumentation mainly based on observing bed separation/rock displacement and/or load or strain measurement within bolted sections of the roadway. This has enabled refinements to be made concerning the improved application of rock bolting to the support of coal mine roadways. Additionally rock bolts have proved successful in improving support in large coal mining excavations such as shafts and underground stations, in addition to lending themselves to achieving improved ground control where rock reinforcement has been required. For example in recovery after falls of ground have impeded the mining operations. Various aspects of support and design have been discussed in relation to the application of rock reinforcement to underground excavations [19-23]. In France and Germany, coal mining tunnels have used rock bolting successfully as a sole means of support. The success of such installations has been found to be generally dependent upon using steel mesh with the rock bolts and, in several situations, reliance on monitoring roof stability by means of displacement measurements using wires or rods in special boreholes. Some mining tunnels driven in coal seams and supported only by rock bolts, steel straps and steel mesh have been considered to be of a short-term application. The possibility of corrosion problems affecting this form of support needs to be taken into account in relation to the timescale of the mining operations. Due account should also be taken that rock reinforcement does not generally allow readily visible means of inspecting the installation quality. As a consequence, support installation needs to be of a high standard and carried out under close supervision.
18.9 SUPPORT OF LONGWALL FACES 18.9.1 Support Principles in Relation to Longwall Operation The main principle of applying support to achieve effective roof control on a longwall face is that of the application of sufficient supporting pressure to maintain a canopy of relieved roof strata over the working area and subsequently allow such strata to be safely sheared and caved behind the face supports. Figure 18 shows a photograph of an exposed cross section of a coal face. This photograph was made possible after the current working face stopped and the hydraulic-powered supports were withdrawn and replaced with wood props and bars. A roadway was then driven along the old face line and, at the point where the photograph was taken, the width of the drivage was increased in order to construct a machine hall. The roof above the coal seam consisted of sandy siltstone and sandstone strata whose strength was in the range 50-100 MPa UCS (uniaxial compressive strength). The extracted seam height on the longwall was 0.94 m whilst the depth below surface was 600 m. The actual excavation width shown by the photograph corresponds to about 6 m whilst the height is approximately 3.5 m. It is evident from this photograph that the roof behaved as a set of sheared blocks. The angle of shear measured to the vertical was around 26°. The general character of sheared roof blocks suggests that they can be justifiably assumed to behave as a composite block for the purpose of carrying out support-loading calculations. Figure 19 shows a line diagram of a cross section across a longwall face supported by a powered roof support. A detached roof block of twice the extracted seam height is assumed to be carried by the longwall support. On the assumption that the roof block shears, caves and increases its volume through bulking action by 50%, then the broken material will provide support to the upper roof beds and thus promote spanning across the working area. The height of the roof block has been investigated by several authorities. The present author has carried out detailed observations on longwall faces and also related the bulking factor to observed density values for the caved material in order to determine the maximum likely height of the detached roof block [24].
Coal Mine Support Systems
537
Figure 18 A cross section of a longwall coal face, showing the origin, extent and inclination of induced fractures as a result of longwall mining (reproduced with permission of the Institution of Mining Engineers)
Figure 19 Illustrating the principle of loading face supports by formation of a roof block (after Whittaker [24])
18.9.2 Generation of Strata Loading on Face Supports Earlier work carried out in the UK has supported the concept of using a detached roof block as the means of representing the generation of strata loading on longwall face supports [25-27]. Important fundamental work has been carried out in the USA on longwall support design and operation with particular reference to caving of roof strata under a wide range of geological conditions, and this work has also given recognition to the fundamental representation of support loading by means of a detached roof block [28, 29]. Although a roof block height of twice the extracted seam height has been extensively used in longwall face support design studies and calculations, basic research has indicated that a roof block
538
Support
height of four times the extraction height may be more relevant under some conditions [24]. For most basic calculations, however, a roof block height of twice the extracted seam height is generally employed. 18.9.3 Loading on Face Supports: UK Coal Industry Approach The UK coal industry employs an Instruction [30] for determining the setting and yield load resistances for powered roof supports associated with longwall faces. The Instruction has been based on the UK coal industry's experience in the application of such supports in coal mines covering depths as great as 1200 m and extracted seam heights of up to 3.5 m. This particular Instruction puts forward roof loading resistance values for différent parts of a longwall face. The basic assumption used in deriving the loading value for the face area is that of considering a rectangular roof block with vertical sides and whose height is twice the extracted seam height, i.e. 2H. The roof block density is assumed to be 2.5 t m~3. The roof block load can then be expressed as (base area) (2H) (2.5 t m" 3 ) which is given in the Instruction as 5H (t m" 2 ) where H is the extracted seam height (m). It is then necessary to take into account an operational contingency factor. This is required in view of the roof support needing to be released for moving forward to its new setting position, and under such conditions, it is assumed that its load is equally shared by the two adjacent supports on each side of the released support. As a consequence, the setting load becomes 1.5 [5H (t m~2)] which is given in the Instruction as Ί.5Η t m" 2 . A loading factor of two is adopted for determining the yield load of the face support, i.e. 15iitm- 2 . The UK guidelines require the same setting and yield load resistances for the buttress area (between the pack and the goaf) as for the face area. The pack and roadhead areas are required to have setting and yield load resistances of at least 5H and 10H t m" 2 respectively. Where massive sandstone conditions form the immediate roof of the coal seam, then due account needs to be taken of the possibility of significantly overhanging roof beds causing increased loading of the roof support. As a consequence, the setting and yield load resistances need to be correspondingly higher to cater for the increased loading conditions.
18.9.4 Strata Loading Determination in Level Seams The work of Wilson [26] considered the equilibrium of the roof block acting on the longwall supports to require a balancing force (5) when the centre of mass of the block was either in front or behind the two rows of supporting props as indicated in Figure 20. In this particular situation, the roof block is assumed to have sheared at an angle Θ measured relative to the vertical. The basic calculation simply involves determination of the support reactions R{ and Rb under minimum loading conditions. This fundamental concept has been extended to consider basic support-loading values which are obtained for various values of Θ in respect of extracted seam heights of 2 m, 4 m and 6 m respectively, and the corresponding values are given in Figure 21. Details of these analyses are given by Whittaker [24].
18.9.5 Strata Loading Determination in Steep Seams The roof block principle also allows an assessment to be made of longwall support loading in steeply inclined seams. The longwall face is usually worked along the strike of the seam so that the supports are correspondingly on a line which is almost on full dip, as is shown in Figure 22. Under these conditions, the roof block is in a position which will readily encourage it to slide off the supports and down the gradient unless effective control is achieved. As a consequence, the roof block needs to be held in equilibrium by the longwall support. The relationship between the resultant thrust of the support and the forces associated with the roof block have been discussed by Wilson [26] who derived equation (14) R = ^ [ c o s a + (sina)/ji]
(14)
539
Coal Mine Support Systems (α)
(b)
. ^Htond + ^zW+d + c)
Validity D< \KH
Validity ran Θ+ D + d+cXD
+d
D > l/z{Hlar\9+D
+ d+c)
(C)
Validity */2{Hiane
+ D + d+c)>D
+ d
Figure 20 Illustrating basic assumptions for analysis of support loading in level seams using the Wilson Concept [25] (after Whittaker [24]) //-4m
Figure 21 Variation in longwall face support leg loading with respect to roof block geometry and extracted seam height (after Whittaker [24])
Support
540 0.6 r
û
'2.6m
0.5 h
2.0 m
0.4
U
m
0.2
D o 0.1
T t
J 1
10
η^
l.lm 0.5m
1
20
1
1
1
1
1
1
I
30
40
50
60
70
80
90
Seam gradient
a°
Figure 22 Showing variation in support loading in relation to extracted seam height for a longwall support for increasing seam gradient (after Whittaker [24])
where R = the resultant thrust acting on the support, W = the weight of the roof block, a = the seam inclination, and μ = the coefficient of friction at the rock interface. Equation (14) has been used to determine data for a longwall situation for a seam gradient which has been varied from level to vertical. The graphs show a marked increase in support thrust over that required for flat seam working. Clearly, there is the additional role played by the roof support in maintaining the roof block in equilibrium against gravitational forces which are acting to cause it to slide down the gradient, and such forces are not necessarily in evidence in flat seam conditions. These results indicate the importance of a high-setting load for steeply inclined conditions in order to ensure roof block equilibrium is achieved during the early stages of extraction. In a corresponding situation in a flat seam it can be argued that setting load is of less significance from the safety point of view, although this is questionable as a high setting load, in any longwall situation whether it be flat or inclined, is highly desirable for achieving early and effective roof control. 18.9.6 Influence of Support Design on Roof Control Longwall face support design has seen remarkable changes in the last few years. Highly advanced hydraulically powered supports are employed in most longwall operations in developed countries. The emergence of the shield support has allowed even greater progress to be achieved, especially in respect of allowing longwall supports to achieve greater compactness whilst having the capability to accept even greater support loads. Shield supports have a linkage mechanism (usually of the lemniscate type) which allows the lateral thrust exerted to the support by the differential movement between roof and floor to be carried by the linkage and not transmitted as a lateral force to the hydraulic legs, as was a problem with conventional powered roof supports. Shield supports also have superior canopy designs which give improved roof control especially in friable conditions. The rear design of shield supports also caters much better in respect of preventing caved materialflushinginto the working area as had been a problem with conventional supports in the past. The new designs of shield support have allowed more difficult roof conditions to be successfully worked, in addition to extending the extracted seam height possible to be taken in one lift. Shield supports offer increased flexibility for working thick seams in multiple lifts. Shield supports have considerably increased the scope of longwall coal mining, particularly in respect of caving strong roof beds. Support design needs to take into account the possibility of periodic weighting as occurs with irregular caving of massive sandstone roof beds. This problem has been encountered in China, India and Australia. Support resistance needs to be correspondingly high and the longwall dimensions selected to allow effective roof control to be achieved. Under such irregular caving conditions, the face support area should be kept to a minimum, and one-web back operation avoided if at all possible. Some longwalls have needed to employ high pressure water infusion of the immediate roof ahead of the working face, for the purpose of weakening the rocks and allowing them to cave more readily.
Coal Mine Support Systems
541
This method has been employed successfully in China, where irregular and secondary caving behind a longwall taking a 3 m extracted seam height, had previously been causing rapid and excessive convergence on the face. This resulted in some of the supports becoming damaged due to loss of travel capacity on the support legs. Under such caving conditions, the supports in the central area of the longwall are exposed to greater risks regarding encountering rapid convergence, and consequently calls for special operational attention to be given to this part of the longwall. 18.10 CONCLUSION Coal mine support systems have witnessed remarkable changes in their long history of development. The first half of the 20th century saw the domination of wood as the primary means of support of coal mining roadways and faces. Records show that such support measures were also accompanied by high accident rates and low standards of safety. The application of steel supports to underground coal mine rooms, roadways and working faces resulted in substantially improved standards of support practice and safety. The improved strength and supporting properties of steel over wood led to significant advances being made regarding improved roof control and stability for different forms of opening in coal mines. There is a wide range of support types available for use in coal mines at the present day. The different support types allow various excavation conditions to be catered for in respect of the duration of application of the support, durability and the magnitude of the stresses affecting the opening. Rock bolting clearly has a major role to play in providing support in mine roadways particularly in the future where support cost is a major consideration. Due account should be taken of the fact that rock bolts have much to offer in respect of improving the stability of different mine openings providing the installation standards are comparable with the equally desirable high design standards. Instrumentation as a means of monitoring performance of such openings also has a vital role to play in such situations. Coal pillars form a major support element in most coal mine layouts. Their design needs to be fully appreciated in order to ensure high standards of safety are achieved coupled with the efficient extraction of coal from the mine. Longwall face operation has been transformed as a result of the wide scale application of advanced support systems such as the shield type. Such supports have resulted in substantial advances being made in support technology to the extent of allowing more difficult mining conditions to be safely and efficiently worked. 18.11 REFERENCES 1. Peng S. S. Coal Mine Ground Control, p. 491. Wiley, New York (1986). 2. Whittaker B. N. A review of progress with longwall mine design and layout. In Proc. State-of-the-Art of Ground Control in Longwall Mining Subsidence and Mining Subsidence, Hawaii (Edited by Y. P. Chugh and M. Karmis), pp. 77-84. SME, New York (1982). 3. Whittaker B. N. and Singh R. N. Stability of longwall mining gate roadways in relation to rib pillar size. Technical Note. Int. J. Rock Mech. Min. Sei. ά Geomech. Abstr. 18, 331-334 (1981). 4. Salamon M. D. G. and Munro A. H. A study of the strength of coal pillars. J. S. Afr. Inst. Min. Metall. 68, 55-67 (1967). 5. Salamon M. D. G. A method of designing bord and pillar workings. J. S. Afr. Inst. Min. Metall. 68, 68-78 (1967). 6. Whittaker B. N. An appraisal of strata control practice. Min. Eng. (London) 134, 9-24 (1974). 7. Whittaker B. N. Chain pillar design considerations with reference to longwall mining. Mining Department Magazine, University of Nottingham, 35, 65-76 (1983). 8. King H. J. and Whittaker B.N. A review of current knowledge on roadway behaviour. In Proc. Symp. Roadway Strata Control, Paper No.6, pp. 73-87. Institution of Mining Engineers, London (1971). 9. Whittaker B. N. and Singh R. N. Design and stability of pillars in longwall mining. Min. Eng. (London) 139(214), 59-70 (1979). 10. Whittaker B. N. and Singh R. N. Deformational behaviour of longwall gate roadways. Min. Sei. TechnoL, 1, 275-284 (1984). 11. Whittaker B. N. and Hodgkinson D. R. The influence of size on gate roadway stability. Min. Eng. (London) 130(124), 203-214 (1971). 12. Woodley J. N. L. and Osborne B. A. MRDE experience with pump packing. Min. Eng. (London) 140(231), 437^443 (1980). 13. Whittaker B. N., Carter M. R., Kapusniak S. S. and Townley A. J. Design and selection of support systems in mine roadways and tunnels with reference to UK coalfields. In Proc. 9th Plenary Scientific Session of the International Bureau of Strata Mechanics. In Mining Systems Adjusted to High Rock Pressure Conditions (Edited by A. Kidybinski and M. Kwasniewski), pp. 209-223. Balkema, Rotterdam (1986). 14. Whittaker B. N. and Frith R. C. Tunnelling: Design, Stability and Construction, p. 460. Institution of Mining and Metallurgy, London (1990).
542
Support
15. Whittaker B. N., Baxter N. G. and Townley A. J. Stability behaviour of steel and concrete systems for coal mine roadways. In Proc. Turkish 5th Coal Mining Congress, Zonguldak, Turkey, pp. 317-335 (1986). 16. Carver J., Luxmore S., Havieson I. A. and Jones H. D. Safety and health in coal mine tunnel drivage. In Tunnelling "76, pp. 85-96. Institution of Mining and Metallurgy, London (1976). 17. Jukes S. G., Hassani F. P. and Whittaker B. N. Characteristics of steel arch support systems for mine roadways. Min. Sei. Technol. 1, 43-58 (1983). 18. Whittaker B. N. and Ambrose D. Strength behaviour of steel arch supports with reference to loading distribution and joint position. Min. Sei. Technol. 3, 267-275 (1986). 19. Douglas T. H. and Arthur L. J. A Guide to the Use of Rock Reinforcement in Underground Excavations, CIRIA Report No. 101, p. 74. (1983). 20. Farmer I. W. and Shelton P. D. Factors that affect underground rock bolt reinforcement systems design. Trans. Inst. Min. Metall. 89, A68-A83 (1980). 21. BischoffJ. A. and Smart J. D. A method of computing a rock reinforcement system which is structurally equivalent to an internal support system. In Proc. 16th U.S. Symp. Rock Mech, Minneapolis, MN (Edited by C. Fairhurst and S. L. Crouch), pp. 179-184. ASCE, New York (1977). 22. Bennett G. H. and Scott J. J. Tunnel and shaft support with the split set friction rock stabiliser. In RETC Proc. vol. 1, chap. 38, pp. 656-664. (1979). 23. Raffoux J. F. and Dejean M. J. Rock bolting and time dependent behaviour of strata. In Tunnelling '79, pp. 175-181. Institution of Mining and Metallurgy, London (1979). 24. Whittaker B. N. A review of the contribution made by powered roof supports to longwall mining. Mining Department Magazine, University of Nottingham, 28, 21-33 (1976). 25. Wilson A. H. Conclusions from recent strata control measurements made by the Mining Research Establishment. Min. Eng. (London) 43, 367-378 (1964). 26. Wilson A. H. Support load requirements on longwall faces. Min. Eng. (London) 173, 479-488 (1975). 27. Ashwin D. P., Campbell S. G., Kimble J. D., Haskayne J. D., Moore J. F. A. and Shepherd R. Some fundamental aspects of face powered supports. Min. Eng. (London) 119, 659-671 (1970). 28. Peng S. S. and Chiang H. S. Longwall Mining, p. 708. Wiley, New York (1984). 29. Peng S. S., Chiang H. S. and Lu D. F. Roof behavior and support requirements for the shield supported longwall faces. In Proc. Int. Symp. Longwall Ground Control and Mine Subsidence, pp. 107-130. SME-AIME (1982). 30. The Use of Powered Supports on Longwall Faces, p. 3. The National Coal Board, London, UK, Mining Department Instruction, PI/1982/6 (1982).
19 Back Analysis in Rock Engineering SHUNSUKE SAKURAI Kobe University, Japan 19.1
INTRODUCTION
543
19.2
BACK ANALYSIS AND ORDINARY ANALYSIS
544
19.3
MODELING OF ROCK MASS
546
19.4
BACK ANALYSIS PROCEDURES
547
19.5 STABILITY ASSESSMENT OF UNDERGROUND OPENINGS 19.5.1 Direct Strain Evaluation Technique 19.5.2 Mathematical Formulation of Back Analysis 19.5.3 Case Study 19.5.3.1 Brief description of the tunnel and instrumentation 19.5.3.2 Back analysis of measured displacements
548 548 548 550 550 550
19.6 BACK ANALYSIS FOR DEFORMATIONAL BEHAVIOR OF JOINTED ROCK 19.6.1 Introduction 19.6.2 Constitutive Equation 19.6.3 Case Studies 19.6.3.1 Large underground cavern 19.6.3.2 Shallow tunnel
552 552 553 556 556 556
19.7 CUT SLOPES 19.7.1 Introduction 19.7.2 Constitutive Equation 19.7.3 Determination of Mechanical Constants and Initial Stress 19.7.4 Case Studies 19.7.4.1 Case A 19.7.4.2 Case B
558 558 559 562 563 563 565
19.8
CONCLUDING REMARKS
568
19.9
REFERENCES
568
19.1
INTRODUCTION
Various types of numerical analyses such as the Finite Element Method (FEM), the Boundary Element Method (BEM) and the Distinct Element Method (DEM), etc. have rapidly developed in rock mechanics during the last decade. They have already been extensively used in engineering practices for designing rock structures such as tunnels, underground caverns, slopes, dam foundations and so on. Still, it is not an easy task to predict the mechanical behavior of such structures with sufficient accuracy. The reliability of the prediction entirely depends on the accuracy of the input data used in analysis, i.e. whether or not the input data is accurate enough to represent the real mechanical properties of rock. However, it is extremely difficult to quantitatively determine geological structures, the geomechanical properties of in situ rocks, initial states of stress, underground water tables and permeability, etc. in a sufficiently accurate manner. So, it is not surprising that the real behaviors of structures often differ from the ones predicted, even though careful geological survey investigations, rock testing and sophisticated computer analyses have been done. 543
544
Back Analysis Monitoring
In order to overcome this difficulty, field measurements are performed during the construction of the structures, not only for monitoring the stability of the structures, but also for reevaluating the input data of the geological and geomechanical parameters which were used in the design analysis. The réévaluation can be done in such a way that the discrepancies between the real and predicted behaviors of the structures are reduced to a minimum. If necessary, the original design and the construction method are modified in order to achieve a rational structural design. This design/construction method, adjusted byfieldmeasurements, was named the 'observational procedure' by Terzaghi and Peck [1]. In this procedure, however, a question may arise concerning how to quantitatively interpret the field measurement results for assessing the original design and construction method. A technique called 'back analysis' is a key to answering this question. It can bridge the gap between prediction and reality. In this chapter, back analysis methods are described and their applicability for engineering practice is demonstrated in a few case studies. 19.2 BACK ANALYSIS AND ORDINARY ANALYSIS Back analysis is generally defined as a technique which can provide the controlling parameters of a system by analyzing its output behavior. In back analysis of rock engineering problems, force conditions, such as external loads and/or rock pressures, and mechanical properties of rock, such as modulus of elasticity, Poisson's ratio, cohesion, internal friction angle, etc., are identified from displacement, strain and pressure measured during and/or after construction. This identifying procedure is the reverse of an ordinary analysis method, where the force conditions and mechanical properties are the input data for determining displacement, stress and strain. Thus, this reverse procedure is called 'back analysis'. The relationship between ordinary analysis and back analysis is shown in Figure 1. In geotechnical engineering, back analysis is being more and more extensively used and has been recognized as a powerful tool for assessingfieldmeasurement data. However, it should be mentioned that back analysis is not a new technique and it has been used among geotechnical engineers for quite a long time. One example is a plate bearing test for determining the modulus of elasticity. In this test, settlement of a plate is measured when applying a load, P. The modulus of elasticity, E, is then back calculated from the measured value of settlement, <5, by using the following equation, which is derived by the theory of elasticity on the assumption of a homogeneous isotropic linear elastic body
E
(1 - v2)P
(1)
- Ssr-
where v is Poisson's ratio and a is the radius of the loading plate. It can be easily seen that this procedure of determining the modulus of elasticity in a plate bearing test is a back analysis. Even in uniaxial compressive tests, the modulus of elasticity is determined from measured strain assuming a linear elastic model. This is also a process of back analysis. It must be emphasized that back analysis is not simply the reverse calculation of ordinary analysis, because, particularly in the modeling of rock, the basic assumption could differ. In ordinary analysis, a mechanical model is usually assumed in such a way that rock is approximately represented by a certain model such as elastic, elasto-plastic, visco-plastic or discrete block models, etc. When a mechanical model is assumed, the values of the mechanical constants of the model can be determined by performing laboratory tests and/or in situ tests. These values are then used as input data in an ordinary analysis to calculate the resulting quantities for displacement, stress and strain. It must be stated that this result provides a unique solution, at least for the given model. In back analysis, on the other hand, displacement, strain and/or pressure values are first obtained by field measurements and then a mechanical model is assumed. The mechanical constants of the model and external forces can then be back calculated by using field measurement results as input Load /external pressure Mechanical properties Geometry/boundary conditions
Ordinary analysis Back analysis
Displacement Stress Strain
Figure 1 Relationship between ordinary analysis and back analysis
545
Back Analysis in Rock Engineering
data. Consequently, it is obvious that the back calculated values of the mechanical constants and external forces depend entirely on the assumed model. For instance, if an elastic model is assumed, then modulus of elasticity and Poisson's ratio will be obtained. However, using identical input data, if an elasto-plastic model is assumed, then a cohesion and internal friction angle will be obtained in addition to modulus of elasticity and Poisson's ratio. So, it can be seen that if different models are (û)
Ordinary
analysis Input
Modeling
Mechanical parameters E, v, c, φ1 . · · External force
Back
Ordinary analysis
-Uniqueness
-Assumption(b)
Results
data
Displacement stress strain
is guaranteed-
analysis Input
Results
Mechanical parameters E, v, c, φ, · · · External force
Back analysis
Model ing
data
Displacement pressure stress strain
—Assumption-Uniqueness
is not guaranteed
Figure 2 Comparison between (a) ordinary and (b) back analysis (Start) [Exploration |-
Figure 3 Procedure for assessing the stability of rock structures and adequacy of design/construction methods
546
Back Analysis Monitoring
assumed, the same input, that is field measurement data, will provide different results. This shows that the results of back analysis are entirely based on which mechanical model is assumed. The difference between ordinary and back analysis is shown in Figure 2. It can be seen in this figure that in ordinary analysis a uniqueness of result is guaranteed, at least for given input data, even if the real behavior of rock is not the same as the one represented by the model. However, in back analysis, uniqueness is not guaranteed because modeling is carried out after obtaining input data. This implies that, in order to guarantee a unique result in back analysis, great care should be taken with the model. The model must be as precise as possible to represent the real behavior of the rock. So, it is obvious that modeling in back analysis is more important than in ordinary analysis. It must be stressed that, if possible, in back analysis, a model should not just be assumed, but should be back calculated from the actual field measurement data. The aim of back analysis, as far as engineering practice is concerned, is not merely to identify the mechanical model together with the values of mechanical constants and external forces. Its final target should be to assess the design/construction methods during construction of the structure; that is, the back analyzed results are used to reevaluate the input data adopted in the original design and to assess the adequacy of the original design/construction methods from safe and economic points of view. The procedure of assessing the stability of structures and the adequacy of the design/construction methods is illustrated in Figure 3. Considering the above discussion of the relationship among back analysis, field measurements and design/construction methods, it is easily understood that back analysis should always be used in conjunction with the construction process. In this aspect, back analysis is entirely different from ordinary analysis which is carried out before the construction phase begins. 19.3 MODELING OF ROCK MASS The ground on or in which structures are built is classified into three groups: (a) continuous, (b) discontinuous and (c) pseudo-continuous types, as shown in Figure 4. Type (a) classification may be used for ground consisting of intact rock without joints or soil. Type (b) represents jointed rock masses. Type (c) classification is for highly fractured and/or weathered rock masses. The overall behavior of Type (c) ground may be similar to that of a continuous body. Thus, it is called a pseudo-continuous type of ground. The mechanical behavior of Type (a) ground can be analyzed by means of the theory of continuum mechanics, while discontinuous approaches proposed by Cundall [2], Kawai [3], and Goodman and Shi [4], can be used for analyzing the behavior of Type (b) ground. Various types of joint elements and interface elements proposed in finite element analysis may also be useful for Type (b). A discontinuous approach similar to that used for Type (b) can, of course, be adopted for Type (c) ground. However, it is almost impossible to explore the location, dimension and mechanical characteristics of all joint systems. This means that a discontinuous approach is not appropriate in engineering practice. Moreover, it seems that this type of ground behaves, in general, just like a continuous body. A continuum mechanics approach can therefore be adopted even for Type (c). It should be noted, however, that the effect of the discontinuities must be adequately taken into account in modeling Type (c) ground as a continuous body. This can result in developing a continuous model which is equivalent to the discontinuous and jointed rock masses. Considering the above discussions, Types (a) and (c) ground can be modeled as continuous bodies, while Type (b) ground should, as precisely as possible, be modeled in such a way that each joint is taken into account individually in a discontinuous model. At any rate, after modeling the ground, all the values of the mechanical constants and geometry of joints for Type (b) can then be determined by back analysis. (α)
Figure 4
(b)
(c)
Classification of the ground: (a) continuous type, (b) discontinuous type, (c) pseudo-continuous type
Back Analysis in Rock Engineering
547
19.4 BACK ANALYSIS PROCEDURES Various back analysis procedures have been extensively developed in geomechanics, ranging from the simple elastic problem to far more complex nonlinear problems. Gioda and Sakurai [5] summarized these procedures with particular reference to the interpretation of the results of field measurements. In general, back analysis procedures may be roughly classified into two categories: the inverse approach and the direct approach [6]. In the inverse approach, the mathematical formulation is just the reverse of that in ordinary analysis, although the governing equations are identical. It must be remembered that the number of measured values should be greater than the number of unknown parameters, so that optimization techniques can be used to back calculate the unknowns. In the case of the ground represented by a simple mechanical model with simple geometrical configurations, closed form solutions in the theory of elasticity and plasticity may be used. On the other hand, for ground with arbitrary shapes under more complex geological and geomechanical conditions, the finite element method (FEM) seems to be more promising. For example, Kavanagh [7] proposed a back analysis formulation based on FEM which may make it possible to obtain the material constants not only for isotropic materials, but also for inhomogeneous and anisotropic materials from both measured displacement and strain. Gioda [8] made a modification of Kavanagh's algorithm to back calculate both the bulk and shear moduli by using static condensation and the least squares method. In order to obtain the material constants, the displacement measurements alone are sufficient. However, for identifying the load conditions in addition to the material constants, the measurement of not only displacements, but also the values of loads and pressures is necessary. For this, a unique and general approach for back calculating the load parameters was formulated by Gioda and Jurina [9]. Sakurai and Takeuchi [10] formulated a back analysis algorithm based on the inverse method. By assuming homogeneous and isotropic linear elastic materials, it can determine the strain distribution around a tunnel based on the data from a limited number of measured displacements. This algorithm was extended by Feng and Lewis [11] in such a way that the algorithm could take into account a linearly changing ground stress field, a material nonlinearity and the effect of the advancing tunnel face with respect to the installation time of the monitoring devices. Sakurai and Ine [12] also proposed an anisotropic material model which can easily represent the behavior of discontinuous rocks. It should be pointed out that the inverse approach of back analysis has a great advantage in engineering practice, because, in general, no iteration is required so computing time can be reduced. However, trouble is often encountered with the inverse approach in obtaining a numerically stable solution for widely scattered values of measurement data which are commonly found in geotechnical engineering problems. There is also difficulty when applying this method to nonlinear problems. On the other hand, the direct approach is based on an iterative optimization procedure which corrects the trial values of unknown parameters in such a way that the discrepancy between the measured and computed quantities is minimized. The advantages of this approach are: (i) that it may be applied to nonlinear problems without having to rely on a complex mathematical background and (ii) that standard algorithms of mathematical programming, such as Simplex and Rosenbrock methods, can be used. Gioda and Maier [13] demonstrated the applicability of a direct method to back calculate the nonlinear material parameters and the load conditions using a numerical example of a pressure tunnel test. Cividini et al [14] stated that the direct approach can be used for determining time dependent material constants by the use of convergence displacement measurement data taken at various stages of tunnel construction. Yang and Sterling [15] proposed a unique back analysis method for determining both the in situ stresses and the elastic properties of rock mass using the fictitious stress boundary element method. The back analysis of the in situ stresses in a nonlinear material was formulated by Zhang et al [16] by using an iterative back analysis algorithm on the basis of the boundary element method. As mentioned before, the direct approach has a great advantage because it can be easily applied to nonlinear, elasto-plastic material problems. However, this method requires rather time-consuming computations since a large amount of iteration is usually involved. The inverse and direct methods are both based on a deterministic concept and provide precise values of material constants and load parameters. However, it is often difficult to determine these values precisely because of the various kinds of uncertainties which are usually involved in rock engineering problems. In overcoming this difficulty, a stochastic approach which is capable of taking these uncertainties into account is preferable. The most advantageous feature of this approach is that the final results are expressed in statistical terms, such as means and variances.
548
Back Analysis Monitoring
Among the various stochastic methods, the Bayesian approach seems to be preferable. The pioneer work on back analysis by means of the Bayesian approach was carried out by Asaoka and Matsuo [17] for parameter identification in consolidation problems. Maier et al. [18] applied thé Bayesian approach for identifying the yield limits of a piecewise linear structural model under static loading. Cividini et al. [19] demonstrated that the Bayesian method can be well applied in parameter identification problems. Another approach of stochastic back analysis is the Kaiman filter parameter identification method. Murakami and Hasegawa [20] initially applied the Kaiman filter method to consolidation problems and later formed an iterative algorithm to determine the best measurement location for back calculating the modulus of elasticity under given load conditions [21]. 19.5 STABILITY ASSESSMENT OF UNDERGROUND OPENINGS 19.5.1 Direct Strain Evaluation Technique For monitoring the stability of underground openings such as tunnels, underground powerhouses and oil storage caverns, etc. field measurements are commonly carried out during construction. These measurement results must be properly interpreted without delay. For this purpose, Sakurai [22] proposed the Direct Strain Evaluation Technique (DSET), which can allow for a quantitative interpretation of displacement measurement results. The basic idea of this technique is to assess the stability of underground openings by comparing the strain occurring in the rock surrounding the openings with allowable strain. Sakurai [22] defined critical strain as the ratio of uniaxial compressive strength to modulus of elasticity, which may be adopted as an allowable value of strain. The critical strain is generally independent of the joints. Therefore, even for jointed rock masses, the critical strain can be determined from laboratory experiments carried out on intact rocks (Sakurai [23]). If the number of data figures in displacement measurements is sufficiently large, strain can be determined directly from the measured displacements by the use of the kinematic relationship. This approach is certainly advantageous because no information is needed with respect to the load conditions and in situ material properties of the rock. In practice, however, the number of extensometers, inclinometers and convergence meters installed around the openings is generally limited and not sufficient for obtaining an overall view of the strain distributions. In order to overcome this difficulty, Sakurai and Takeuchi [10] proposed an inverse back analysis method. In this method, the initial stress and in situ material properties of rock are first back calculated from measured displacements and then used as input data for an ordinary finite element analysis to determine the strain distribution around the underground openings. The two different procedures for obtaining the strain distribution are shown in Figure 5. 19.5.2 Mathematical Formulation of Back Analysis In formulating mathematical equations, the following assumptions are made. (i) The mechanical characteristics of rocks are expressed by a linear homogeneous isotropic elastic model so that the material constants are modulus of elasticity and Poisson's ratio. However, since Poisson's ratio has no great influence on the results of back analysis, an adequate value can be chosen.
Back analysis Normalized initial- stress
Kinematic relationship Ordinary
analysis
Strain distributions
Figure 5 Flow chart for determining strain distributions from measured displacements
Back Analysis in Rock Engineering
549
(ii) The elastic constants of the lining are assumed to be known. (iii) The initial stress is uniformly distributed throughout the rock being excavated. In Finite Element analysis, for excavation problems of underground openings, the genuine rock pressure can be taken into account by applying equivalent nodal force {P0} at the excavation surface. This force is determined by {P0} = £ [ * ] > ( > } d t ; + [ ΙΝ]τ{ρ}άυ
(2)
where [ N ] and [/?] are the matrices of the element shape functions and their derivatives, respectively. {σ0} is the initial state of stress, {/>} is the vector of the body force components due to gravity and v is the volume of the excavation element. A two-dimensional formulation [24] is presented here, although it is easily extended to three-dimensional cases. Hence, the initial stress components are as follows { σ ο } = {σ χ 0 σνο txyo}T
(3)
The relationship between nodal force {P} and nodal displacement {u} is expressed by the wellknown equation [*]{«} = {P}
(4)
where [ΛΓ] denotes the stiffness matrix of the assembled finite element system. For a lined underground opening with moduli of elasticity £R and EL of the rock and lining, respectively, the stiffness matrix is expressed as [*] = £R[Jif*]
(5)
where [**] = [JTR] + Ä[ITL] R = EJEK in which [ # R ] represents the stiffness matrix for the finite element model of the ground where ER = 1, and [tf L ] for the lining where £ L = 1. When determining these stiffness matrices, Poisson's ratio of the rock and lining must be assumed. The finite element mesh must be chosen in such a way that the measuring points coincide with the nodes of the mesh. Substituting equation (5) and the equivalent nodal force given by equation (2) into equation (4), we obtain the following equation provided that the second term of equation (2) is disregarded
[**]{«} = σ,ο/ΜΡι} + °,O/ER{PI} + Txy0/EK{P3}
(6)
where {Pt} (i = 1-3) denotes the equivalent nodal forces corresponding to the components of the unit initial stress. Substituting σχ0/Εκ = 1 and ay0/EK = rxy0/ER = 0 into equation (6) gives [**]{«,} = {l\}
(7)
Solving equation (7), we obtain displacement {ux} at the nodal points due to the initial stress component, σχ0/Εκ = 1 only. By following a similar procedure, displacements {uy} and {uxy}, caused by the other components of unit initial stress ay0/EK = 1 and Txy0/ER = 1, respectively, are obtained. Considering displacements {ux}9 {uy} and {uxy} due to each component of initial stress, the following equation is derived
[flWsW
(8)
where [Λ] = [{"*} {·,}{«„}] {σθ} = {<7X0/£R
{σ0} is called the normalized initial stress.
σ
νθ/Εκ
Tjcyo/^R}
550
Back Analysis Monitoring
Displacement {u} may be split into two parts, namely, measured displacements {«χ} and unknown displacements {i/2}. Therefore Μ1]{σ0} = {«1}
(9)
Matrix [A x ] is uniquely defined when Poisson's ratio for the ground and lining materials are given and parameter R also is given. It is noted that in equation (9) the measurements of the relative displacement between two measurement points as well as absolute displacements can be used as input data [10]. If the number of data for measured displacements is greater than three, the normalized initial stress can then be determined by an optimization procedure. If the least squares method is adopted, the normalized initial stress is uniquely determined from the measured displacements as follows {*o} = ( [ ^ ι ] Τ Μ ι ] Γ 1 Μ ι ] Τ > ι } = [ ^ ] { « ι }
(10)
From the normalized initial stress, we can calculate the modulus of elasticity and initial stress by assuming that the vertical component of initial stress is equal to the overburden pressure. In the case of lined underground openings, an appropriate value must be assumed for parameter R (i.e. /?! ). Then, the back analysis procedure which provides the modulus of elasticity as described above is followed. Using this modulus of elasticity, a new value of parameter R (i.e. R2) is determined. If R2 is not close enough to Rl9 this calculation is repeated until sufficient agreement between assumed and calculated values of R is obtained. After determining the normalized initial stress, the displacements at all the nodal points can be calculated by equation (8). The strain in each element, therefore, can be obtained by use of the following relationship between strain and the displacements {«} = [*]{«}
(11)
where matrix \_B] is a function of the location of the nodal points alone. It should be noted here that strain distribution is uniquely determined from the normalized initial stress. This means that it is not necessary to know both the modulus of elasticity and initial stress, but the ratio of these two alone is enough to determine strain distribution. If necessary, however, the values of modulus of elasticity and initial stress can be derived from the normalized initial stress by assuming that the vertical component of initial stress is equal to the overburden pressure. This idea has been extended by Shimizu and Sakurai [25] to derive a three-dimensional back analysis method in which both three-dimensional initial stress and modulus of elasticity are determined. The basic principle of the method is the same as the one presented in Section 19.5.2. However, a boundary element method is used in the formulation. This method was applied successfully to back calculate both the three-dimensional initial stress and modulus of elasticity during the construction of an underground hydropower plant cavern (Sakurai and Shimizu [26]).
19.5.3 Case Study [27] 19.5.3.1 Brief description of the tunnel and instrumentation A double-lane highway tunnel was constructed at the maximum depth of 25 m. The ground in which the tunnel was excavated consisted of heavily weathered granite which was a nearly soil-like material. Little underground water came out at the tunnel face. Excavation of the tunnel was carried out by adopting a top heading method. Support measures consisted of a combination of shotcrete, rock bolts and steel ribs, depending on the geological conditions. Displacement measurements were performed at several different tunnel sections, where a convergence meter, multirod extensometers, and a Sliding micrometer, developed at ETH, Zurich, were used. The settlement at the tunnel crown was also measured by an ordinary surveying instrument. The places where the instruments were installed are shown in Figure 6. 19.5.3.2 Back analysis of measured displacements The back analysis procedure described in Section 19.5.2 was used for determining the normalized initial stress and for obtaining strain distribution around the tunnel. In the back analysis, four cases
Back Analysis in Rock Engineering m&w
Surface
v&ss**
551
-TSK&A
I®
©
Convergence
©
Extensometer
©
Sliding micrometer (ISETH) 5m
-H
Figure 6
Installation of instruments
Table 1 Results of Back Analysis Case I Input data <7χθ/Ε
Gyo/E τ
χγθ/Ε
σχ0 (MPa) Oyo ( M P a )
xxy0 (MPa) E (MPa) v (assumed)
Convergence - 0.977 x 10 " 3 - 0.669 x 10" 3 - 0.243 x 10 " 3 - 0.686 - 0.471 -0.171 703.6
0.3
Case 2 Convergence + extensometer - 0.123 x l O " 2 - 0.978 x 10" 3 - 0.957 x 10 " 4 - 0.592 - 0.471 -0.046 481.3
0.3
Extensometer
Case 4 Extensometer + Sliding micrometer
- 0.221 xlO" 2 -0.178xl0-2 0.245 x 10"3
-0.233xl0-2 -0.233x10-2 0.206 x 10"3
Case 3
- 0.585 - 0.471 - 0.065 265.1
0.3
- 0.471 - 0.471 -0.041 201.7
0.3
were considered which all depended on the use of measured displacements for input data. They were: (i) convergence measurements alone, (ii) both convergence and extensometer measurements, (iii) extensometer measurements alone and (iv) both extensometer and Sliding micrometer measurements. The results of the back analysis are summarized in Table 1. Assuming that the vertical component of initial stress is equal to the overburden pressure, the other components of initial stress and modulus of elasticity can be separated from the value of the normalized initial stress. When the initial stress and the modulus of elasticity are determined, they are used as input data for an ordinary finite element analysis to determine stress, strain and displacement distributions throughout the ground. The comparison between calculated and measured displacements proves the accuracy of back analysis, as shown in Figures 7 and 8. Figure 7 is for a case in which only the data of convergence measurements are used as input data. Of course, a good agreement for convergence measurements was obtained, while a large discrepancy can be seen in the extensometer measurements. On the other hand, the results shown in Figure 8 indicate that a good agreement exists between the computed and measured values of extensometer measurements, because the extensometer measurement results alone have been used for this back analysis. It can be seen from these figures, that although relatively large displacements occurred in the ground surrounding the tunnel, small convergence values were measured. This may be due to the fact that convergence was restricted by the high rigidity of the support structures. The displacements calculated through the back analysis by using both extensometer and Sliding micrometer measurement data (Case 4), are also compared to the measured values, as shown in Figure 9. It is seen from this figure that there is a good agreement between the two, and it demonstrates that a large extent of the ground deforms continuously, just like a homogeneous
552
Back Analysis Monitoring
Measurement line number 0 T
(mm) 1
10 Γ
Extensometer (mm)
Figure 7 Comparison between measured and back calculated displacements (Input data = Convergence)
isotropic elastic material. In this case, the modulus of elasticity becomes the smallest among all the cases shown in Table 1. It is also of interest to know that if data for the measured displacements for a large area of the ground are used in the back analysis, the back analyzed modulus of elasticity decreases. In order to assess the stability of the tunnel, the strain occurring in the ground was compared with the critical strain of ground materials on the basis of the direct strain evaluation technique. The strain distribution was calculated for each case of the back analysis. One of the results for Case 3 is shown in Figure 10, as contour lines of maximum shear strain. This figure shows that the largest value of maximum shear strain reaches about 0.4%, and ranges from 0.2 to 0.3% in the ground at a distance of 5 m from the tunnel surface where the rock bolts are installed. On the other hand, in laboratory tests, the critical strain proposed by Sakurai [22] was evaluated at approximately 1.2%. Therefore, the strain occurring in the tunnel's surrounding ground was still smaller than the critical strain. This means that the stability of the tunnel was guaranteed. 19.6 BACK ANALYSIS FOR DEFORMATIONAL BEHAVIOR OF JOINTED ROCK 19.6.1 Introduction The discontinuous deformation of jointed rock may be classified roughly into the following two modes: (i) spalling of joints and (ii) sliding along a particular slip plane. It is, of course, possible to represent these two deformational behaviors of jointed rock by using two different mechanical models. As far as back analysis is concerned, however, a single model capable of analyzing these different modes of deformation is preferable because, as mentioned earlier, a mechanical model should not be assumed in back analysis, but it should be determined uniquely from field measurement results. In this respect, a single model is easy to handle in order to identify the real behavior of jointed rock.
Back Analysis in Rock Engineering
553
[gigll Measurement
E E 20 |
| Back analysis
? 101O
o
©
©
©
©
Measurement line number
Figure 8 Comparison between measured and back calculated displacements (Input data = Extensometer)
In this section, a single mechanical model is described, which can represent the two different modes of deformation simply by changing the values of mechanical constants. Therefore, determining the mechanical constants by back analysis means that the mechanical model of jointed rock has been clearly identified. 19.6.2 Constitutive Equation In order to represent the deformational behavior of discontinuous jointed rock, the following constitutive equation is introduced in the framework of continuum mechanics. Consider a small element which contains a discontinuous plane taken from jointed rock masses as shown in Figure 11. Assuming the element as an equivalent continuous body, the incremental stress and strain relationship can be expressed in the x'-y' local coordinate system as follows (12)
where nv2(l + v j
0
nv2(l + v j
1 — vj
0
0
0
w(l -n\\) [/>']
(l +
Vl)(l_Vl-2iiv*)
m(l + v J i l - V i
(13) -2nv\)*
Hence, it is transformed into the x-y global coordinates as follows W = [D]{«}
(14)
Back Analysis Monitoring
554
Surface
—x—Measurement -o-Back analysis
Vertical initial stress a y 0 =-4.8kg c m - 2
Figure 9
Comparison between measured and back calculated displacements (Input data = Both extensometer and Sliding micrometer)
where [Ζ)] = [Γ][Ζ)'][Γ] Τ
(15)
[ J ] is a transformation matrix expressed as cos2 a
in =
2
sin a sin a cos a
sin2 a 2
cos a
— 2sin a cos a 2sinacosa
(16)
— sin a cos a cos2 a — sin2 a
where a is the angle between the x' and x axes. It should be noted that equation (14) can represent various types of deformational behaviors of the materials from an isotropic elastic to a nonelastic anisotropic behavior. They can be represented simply by changing the values, m and n, which are defined as the anisotropic parameters (Sakurai and Ine [12]). If n = 1, m — 1/2(1 + v) and v1 = v2 = v, then rocks behave as an isotropic elastic material. A discontinuous behavior can then be represented by using the values of n and m which are different from the ones of isotropic materials. This discontinuous behavior may be due to the fact that the
Back Analysis in Rock Engineering
555
Scale
Figure 10 Contour line maximum shear strain (Input data = Extensometer) (b)
t t>
Spoiling
'
1
x
»
Joint Continuum body
4>"
:
Ç>
Figure 11 Modeling for discontinuous deformation
joints existing in rock tend to open or slide. Therefore, the anisotropic parameters are determined so that the constitutive equation can represent the spalling of discontinuous planes and/or the sliding along slip surfaces. The values of the anisotropic parameters are defined as follows. (i) Spalling of the discontinuous plane: Spalling of the discontinuous plane shown in Figure 11 can be represented by increasing the anisotropic parameter n, that is, by reducing the value of E2 against Ex. Poisson's ratio v2 is taken to be zero because spalling in the direction of the y' axis makes no movement in the x' axis. In this case, the other anisotropic parameter m must be taken as m = l / 2 ( l + v1). (ii) Sliding along the discontinuity: When sliding occurs along the discontinuity parallel to the x' axis, the anisotropic parameter m can be reduced to a small value i.e. m < 1/2(1 + Vi ), while n = 1.0 and Vx = v2 are assumed. Anisotropic parameters for different behavior of jointed rock masses are summarized in Table 2. It may be noteworthy to mention that plastic flow is also represented by introducing the anisotropic parameters (Sakurai, Ine and Shinji [28]). It should be stated here, that there is no need to assume a mechanical model in back analysis, provided that the anisotropic parameters are introduced. For instance, if a small value of m is
556
Back Analysis Monitoring Table 2
Anisotropie Parameters for Different Behavior of Jointed Rock Mass
Isotropie material
n = 1.0
m = 1/2(1 + v)
vx = v2
Spalling of joints Sliding along joints
n > 1.0 n = 1.0
m=l/2(l+V!) m < 1/2(1 + v)
v2 = 0 v1 = v2
obtained in a certain zone from performing a back analysis on measurement results, then it can be seen that sliding tends to occur in this zone and the direction of the sliding is known as the value of a. This means that a mechanical model can be derived by back analysis, as well as by mechanical constants.
19.6.3 Case Studies 19.6.3.1
Large underground cavern
A large underground cavern for a hydroelectric power plant was constructed in rock which consisted of tuff breccia and andésite. Several small shear zones existed at the site of the power house. Careful observations and field measurements were carried out during excavation for monitoring the stability of the cavern and verifying the adequacy of the design and the construction method. A cross-section of the cavern is shown in Figure 12. The length of the cavern is 253 m. One of the locations where extensometers were installed is shown in Figure 13. The measurement results are also given in this figure. The finite element mesh is shown in Figure 14. The mechanical constants, including the anisotropic parameters and the initial states of stress, were back calculated so as to minimize the following value δ = £(i4-tC)2-*min.
(17)
i=l
where u™ and u* are measured and computed displacements, respectively. M is the number of measurement points. (For this minimization analysis, computer programs such as Simplex, Rosenbrock, etc. supplied in the program library, can be used.) The results from the back analysis indicate that the sliding and spalling zones appear around the cavern as shown in Figure 15, and the values of anisotropic parameters m and n in the sliding and spalling zones are m = 0.038 and n = 20, respectively (Sakurai and Tanigawa [29]). The occurrence of the sliding and spalling zones seems to be reasonable, considering the geological conditions involved (see Figure 16). The back calculated initial stresses and modulus of elasticity are given in Figure 17. The displacement distribution is then calculated by using all the back calculated values of the mechanical constants and initial stresses as input data in an ordinary finite element analysis. The results for comparing them with the measured values are also shown in Figure 17. The maximum shear strain distribution around the cavern is given in Figure 18.
19.6.3.2
Shallow tunnel
A double-truck railway tunnel of shallow depth was constructed underneath a highly developed urban area. The ground in which the tunnel was located consisted of fine grain sand deposits. Both the tunnel diameter and the height of overburden are approximately 10 m. Extensometers were installed from the ground surface before tunnel excavation so that the total displacements due to excavation could be measured. The ground surface settlements were also measured. The back analysis, taking into account the anisotropic parameters, was carried out in order to determine the deformational mechanism of the ground as well as the mechanical constants. The finite element mesh used in the back analysis is shown in Figure 19. The shaded zone indicates the loosened zone in which the slip planes are mobilized. The back analysis results of the mechanical constants and initial stresses are given in Figure 20. The displacements can then be calculated. The results are indicated in Figure 20, where the measured values are also shown for comparison. From this figure, it can be seen that there is a good correlation between the calculated and measured
Back Analysis in Rock Engineering
557
Figure 12 Cross-section of the cavern EL 117.21m
25 20 15 10 5 Penstock side
(m)
Figure 13 Location of multirod extensometers and displacements measured
Figure 14 Finite element mesh
displacements. The maximum shear strain distribution is shown in Figure 21. The stability of the tunnel was then assessed by comparing this maximum shear strain with the allowable value of strain. The maximum shear strain distribution occurring around a circular tunnel which was obtained by direct field measurements is shown in Figure 22 for reference (Hansmire and Cording [30]). It is interesting that a similar strain distribution appears in both Figures 21 and 22.
558
Back Analysis Monitoring Zone(2) (spoiling)
Zone (sliding)
Figure 15 Sliding and spalling zones occurring around the cavern
A
Penstock side
/
A
EL 117. 21 m
Shear zone Δ Tuff Breccia A Lapilli Tuff
Figure 16 Geological conditions
19.7 CUT SLOPES 19.7.1 Introduction As already mentioned, it is of extreme importance in back analysis that a mechanical model should not be assumed beforehand, but should be determined uniquely from the results of field measurements. In order to determine the model uniquely, it is recommended that the model accepted in back analysis should include all the modes of deformation and represent any type of deformational mode only by changing the parameters of the model. If this sort of mechanical model is introduced, the deformational mode of rock can be identified simply by determining the parameters of the model. This means that by using this model it is possible to determine uniquely the mechanical model representing the real behavior of rock by back analysis of field measurement results. In the deformation of cut slopes, the deformational mode is classified into three different groups: (a) elastic, (b) sliding and (c) toppling, as shown in Figure 23. Therefore, the mechanical model for analyzing cut slope problems must be one which includes all three deformational modes as a potential, and one or more of the modes will be derived by changing the parameters of the model.
559
Back Analysis in Rock Engineering -0.000457 °χο/ε' σγ0/Ε = -0.000683 τ
/Ε
χγο
-
0.000250
σιο/£"*= -0.000296 σ 2 0 /£*= - 0 . 0 0 0 8 4 4 0 = 32.8°
— ο — Measured χ
Calculated No. 2
10 15 (m)
η2 -J ι (cm)
20
No. 4
5
10 5 20 (m)
m* 0.038
•ύθ 25 -,2
H I (cm) 00
25
fl*20.0
Figure 17 Results of back analysis and comparison between measured and calculated displacements
( Unit : % )
Figure 18 Maximum shear strain distribution
As far as engineering practice is concerned, the model should be as simple as possible, so that it can easily be applied to design analysis. In this section, a back analysis method for cut slope problems is described. The mechanical model used in this back analysis is capable of dealing with all three deformational behaviors, that is, elastic, sliding and toppling deformations. 19.7.2 Constitutive Equation The deformational behavior of discontinuous materials generally indicates the smallest shear rigidity in the direction parallel to the potential slip surface. Considering this deformational behavior, an anisotropic constitutive equation of jointed rock was proposed by Sakurai et al. [31].
Back Analysis Monitoring
560
Scale I
3.5m
1
Figure 19 Finite element mesh and loosened zone
2
( cm) 0-2
(cm) x
10
Back analysis Measurements
σχ0/Ε
-0.7793 x I0"2
σ
-0.1475 x I0"2
T
χθ'£
1 xyo/E
0.1044 x I0~ 2
m ELX'E
|1
εί2/ε
0.06 10 60
j
Figure 20 Results of back analysis and comparison between measured and calculated displacements
Let the local coordinate system x'-y' be taken as shown in Figure 24, where x' axis is parallel to the direction of the potential slip planes. Then, the constitutive equation is described as follows (18)
561
Back Analysis in Rock Engineering
□■ 0.100
0.150
0.200
0.500
1.000
¥>' ' - 's d V7—~7
Λ
Scale
max = 4 . 0 0 %
Figure 21 Maximum shear strain Ground surface
20ft 6m
(Unit : % )
Figure 22 Maximum shear strain distribution (after Hansmire and Cording [30]) (a)
(b)
(c)
Figure 23 Deformation modes of cut slopes: (a) elastic, (b) sliding, (c) toppling
μ
3.5 m
Back Analysis Monitoring
562
yx
A/
Horizontal back slope
Base line /
^ ^ "Λ Base angle Θ
Toe of slope Figure 24 Coordinate system
Original ground surface
Figure 25
Layered elements of zones parallel to the base line
where
[*>'] =
1 - v - 2v 2
1 -v
v
0
v
1- v
0
0
0
m(l - v - 2v 2 )
(19)
{σχ> Qy' τΧ'γ' }T and {εχ> ey> γχΎ }T are stress and strain in the x'-y' coordinate system, respectively. E is the modulus of elasticity, v is Poisson's ratio, and m is an anisotropic parameter. When m = 1/2(1 + v), the equation changes to the one for an isotropic material. It should be stated that the constitutive equation is identical to equation (12), in which the anisotropic parameter n = 1 is introduced. When the constitutive equation for the local coordinate system is known, it is easy to extend it to the x-y global coordinate system. 19.7.3 Determination of Mechanical Constants and Initial Stress In the analysis of cut slope problems, rock above a base line is divided into N layered elements of zones parallel to the base line, as shown in Figure 25. The base line is defined as a line under which
Back Analysis in Rock Engineering
563
no displacement occurs at all. The model may be extended to more general slope problems with a curved failure plane using curved layers. It is assumed that each layer has a different value of m and a, but the same value for E and v. It is also assumed that the material below the base line behaves as a homogeneous isotropic material, where only two material constants, E0 and v0, exist. The location and inclination of the base line can be evaluated through a careful investigation of observations/measurements taken during the cutting of slopes. The number of layers above the base line can also be estimated by considering the results of measurements, depending on the displacement distributions along the vertical axis. The material constants (£, v, mx,. . ., mN), as well as the initial stress existing in the ground prior to slope excavation, are obtained so as to minimize the error function presented in equation (17). 19.7.4 Case Studies 19.7.4.1 Case A A cut slope appeared adjacent to the portal of a highway tunnel. The stability of the slope became a serious problem and therefore field measurements were performed to monitor the slope during excavation. The casing tube for a borehole inclinometer was installed prior to excavation 2 m apart from the slope surface and 9 m below the floor of the excavation, as shown in Figure 26. The geological formation of the ground consists of nearly horizontal layers of sand and gravel. The displacements due to the cutting of the slopes were measured by the inclinometer and the results were used for a back analysis to determine the initial stress and mechanical constants, including the anisotropic parameter. In this case study it is assumed that the base line is a straight line and the zone above it is divided into three layers, as shown in Figure 27. Thefiniteelement mesh in this zone is also indicated in thisfigure.Each layer may have a different value for the anisotropic parameters. It is noted that the angle a of each layer is not necessarily the same as the angle of the base line. All the material constants, including the anisotropic parameters as well as the initial stress existing before excavation, are then back calculated from measured displacements. The results are as follows (Kondoh and Shinji [32]) ml = 0.385 (isotropic)
σχ0/Εκ
= - 0.274 x l O - 2
m2 = 0.385 (isotropic)
σγ0/Εκ
= - 0.478 x 10" 2
m3 = 0.025
τχνο/Εκ
= - 0.113 x l O - 2
v = 0.3
(assumed)
where σχ0, ay0, rxy0 are the components of initial stress acting at the toe of the slope.
EL 60.0 m
EL 50.0 m
EL 4 0 . 0 m
I
L
Figure 26 Configuration of slope and location of inclinometer
564
Back Analysis Monitoring
Figure 27 Triple layered zone above a base line
Figure 28 Comparison between measured and calculated displacements
Once all these values are known, calculation of the displacements by means of an ordinary finite element analysis can be done. Then they can be compared with the measured values to verify the accuracy of the back analysis. Figure 28 illustrates this comparison and shows that a good agreement exists between the measured and calculated displacements. In this figure, the results obtained by assuming an isotropic elastic material are also shown for reference. It is seen from this figure that only a small discrepancy appears between the results obtained by the isotropic model and those obtained by the anisotropic model. This means that the behavior of this cut slope is similar to that of isotropic elastic materials. Thus, the slope is classified as being of an elastic type. However,
565
Back Analysis in Rock Engineering
0.500
1.000
Κο-,Ί I
7.0 m
1.500
2.000
2.500
Maximum shear strain (%) Figure 29 Maximum shear strain distribution
Figure 30 Front view of vertical cutting wall
the maximum shear strain distribution shown in Figure 29 demonstrates that the potential sliding surface seemingly starts to occur, although it is not too serious.
19.7.4.2 CaseB In this case study, rock mass was cut vertically by reinforcing it with rock bolts and shotcrete sprayed on the free surface. The rock mass consisted of granite and some parts of the rock were heavily weathered.
566
Back Analysis Monitoring
1&**\
KH2
/=28m
Inclinometers
Figure 31 Cross-section D-D, and location of measuring instruments Table 3 Back Analyzed Initial Stress and Mechanical Constants σχ0/Ε γ/Ε τχγ0/Ε
- 0.142 x 10"2 0.734 x 10" 6 - 0.288 x HT 3
y (assumed) E
22.54 (kNm -3 ) 306.9 (MPa) 0.07
displacement
8.0 m
Figure 32 Calculated displacement vectors
The purpose of this cutting was to investigate the possibilities of vertical cutting for construction of a foundation for a suspension bridge (Shiraishi et al [33]). The maximum height of the cutting was 25 m. The front view and a cross-section of the cutting are shown in Figures 30 and 31. The cross-section shown here is one of the principal measuring sections where extensometers and inclinometers were installed. The horizontal displacements on the top surface were measured by using an invar wire extensometer and the settlements were measured by using an ordinary surveying technique. The movement of the cutting surface was also measured by an optical surveying system. The displacements measured during cutting were back analyzed to obtain the mechanical constants of the rock masses. The back analyzed initial stress and mechanical constants are shown in Table 3. The calculated displacement vectors are shown in Figure 32. A comparison of calculated displacements with measured ones is given in Figure 33. It is seen from this figure that there was a good agreement between the calculated and measured values of displacements. The maximum
567
Back Analysis in Rock Engineering Extensometer
24 22 20 ΙΘ (m) 16 14 12 10 8 6 4 2 0
K
16 14 12 10 8 6 4 2 0 N (m) —EXO
\i•P
— · —
Measured
—o
Calculated
KH2
M
I I I I I 2 I 0
9.0 m
Inclinometer (cm)
i
Ψ*
30 L
r
28 [ 26Î
L
KT5
24 22 20 (m)
\
IQl '6 14 12
1
14 12 10 8 6 4 2 0 Extensometers (m) 9.0 m
J J
10 8 6 4 2 0
N •
■' 1
1
O
9.0 m
Incl inometer (cm)
Figure 33 Comparison of calculated displacements with measured values: (a) inclinometer KH2 and extensometer EXO, (b) inclinometer KT5, (c) extensometer EXl and EX2
0.200 0.300 0.400 0.500 0.600 (%) I
I I»I*!*K»!·!«!'
Figure 34 Maximum shear strain distribution
568
Back Analysis Monitoring
shear strain distribution is shown in Figure 34. It is understood from these results that the deformational behavior of this slope is a toppling type, and the mechanical model described here can also be applicable for a back analysis of toppling type deformation. The factor of safety was then calculated by using back analyzed material constants given in Table 3. Since the procedure of the calculation can be found elsewhere (Sakurai [34]), only the results are given here. The back analyzed cohesion and internal friction angle are c = 0.1 MPa and φ = 30°, respectively. Thus, the factor of safety becomes FS = 1.7. 19.8 CONCLUDING REMARKS Emphasis must be placed on the field measurements carried out during the construction of rock structures such as tunnels, caverns and cut slopes, etc. They are of extreme importance for achieving rational design and construction of the structures. The field measurement results must also be properly interpreted for assessing the design/construction methods. If necessary, the design/construction methods should be modified during construction without delay to ensure safety and economy. For the interpretation offieldmeasurement results, back analysis is a powerful tool which can assess the design parameters in such a way that the input data adopted in the original design are reevaluated on the basis of back analysis results. So the primary aim of back analysis is not only to identify the material constants and external forces, but also to assess the adequacy of the original design/construction methods. This implies that back analysis should be considered as an important element in conjunction with the construction process. In this chapter, the back analysis methods proposed by the author and his coworkers have been presented. The methods are formulated on the basis of continuum mechanics, so that they can be applied to an engineering problem associated with continuous and pseudo-continuous types of rock. It should again be emphasized that in back analysis, an important yet difficult task is the determination of a mechanical model to represent the real behavior of rock. The mechanical model should not be assumed, but should be determined by a back analysis. ACKNOWLEDGEMENTS The results presented here are mainly based on work carried out by various coworkers and students at Kobe University. The author wishes to thank all of those people. The author also thanks Mr N. Shimizu, Research Associate, and Mr I. Kawashima, Graduate Student, both of Kobe University, for their help in preparing this manuscript. Special thanks also to staff members Ms G. Patten and Ms B. Salisbury for proofreading and typing this manuscript. 19.9 REFERENCES 1. Terzaghi K. and Peck R. B. Soil Mechanics in Engineering Practice, pp. 627-632. Wiley, New York (1948). 2. Cundall P. A. A computer model for simulating progressive large-scale movements in blocky rock systems. In Proc. Symp. Int. Soc. Rock Mech. Nancy Vol. 1, Paper II-8 (1971). 3. Kawai T. Some considerations on the finite element method. Int. J. Numer. Methods Eng. 16, 81-120 (1980). 4. Goodman R. E. and Shi G.-H. Block Theory and its Application to Rock Engineering, p. 338. Prentice-Hall, New Jersey (1985). 5. Gioda G. and Sakurai S. Back analysis procedures for the interpretation of field measurements in geomechanics. Int. J. Numer. Anal. Methods Geomech. 11, 555-583 (1987). 6. Cividini A., Jurina L. and Gioda G. Some aspects of 'characterization' problems in geomechanics. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 18, 487-503 (1981). 7. Kavanagh K. T. Experiment versus analysis: Computational techniques for the description of static material response. Int. J. Numer. Methods Eng. 5, 503-515 (1973). 8. Gioda G. Indirect identification of the average elastic characteristics of rock masses. In Proc. Int. Conf. Structural Foundations on Rock, Sydney, pp. 65-73 (1980). 9. Gioda G. and Jurina L. Numerical identification of soil-structure interaction pressures. Int. J. Numer. Anal. Methods Geomech. 5, 33-56 (1981). 10. Sakurai S. and Takeuchi K. Back analysis of measured displacements of tunnels. Rock Mech. Rock Eng. 16, 173-180 (1983). 11. Feng Z. L. and Lewis R. W. Optimal estimation of in situ ground stresses from displacement measurement. Int. J. Numer. Anal. Methods Geomech. 11, 391-408 (1987).
Back Analysis in Rock Engineering
569
12. Sakurai S. and Ine T. Strain analysis of jointed rock masses for monitoring the stability of underground openings. In Proc. Int. Symp. Computer and Physical Modeling in Geotechnical Engineering, Bangkok, pp. 221-228 (1989). 13. Gioda G. and Maier G. Direct search solution of an inverse problem in elastoplasticity: Identification of cohesion, friction angle and in situ stress by pressure tunnel tests. Int. J. Numer. Methods in Eng. 15, 1823-1848 (1980). 14. Cividini A., Gioda G. and Barla G. Calibration of a rheological material model on the basis offieldmeasurements. In Proc. 5th Int. Conf. Numer. Methods in Geomech. Nagoya (Edited by Z. Einsenstein) pp. 1621-1628 (1985). 15. Yang L. and Sterling R. L. Back analysis of rock tunnel using boundary element method, J. Geotech. Eng. Div. Am. Soc. Civ. Eng. 115, 1163-1169 (1989). 16. Zhang De-Cheng, Gao Xiang-wei and Zheng Yingren. Back analysis method of elastoplastic BEM in strain space. In Proc. 6th Int. Conf. Numer. Methods in Geomech. Innsbruck (Edited by G. Swoboda), pp. 981-986. Balkema, Rotterdam (1988). 17. Asaoka A. and Matsuo M. Bayesian approach to inverse problem in consolidation and its application to settlement prediction. In Proc. 3rd Int. Conf. Numerical Methods in Geomechanics, Aachen (Edited by W. Wittke) Vol. 1, pp. 115-123. Balkema, Rotterdam (1979). 18. Maier G., Nappi A. and Cividini A. Statistical identification of yield limits in piecewise linear structural models. In Proc. Int. Conf. Computational Methods and Experimental Measurements, Washington DC. pp. 812-829 (1982). 19. Cividini A., Maier G. and Nappi A. Parameter estimation of a static geotechnical model using a Bayes' approach. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 20, 215-226 (1983). 20. Murakami A. and Hasegawa T. Observational prediction of settlement using Kaiman filter theory. In Proc. 5th Int. Conf. Numerical Methods in Geomech. Nagoya (Edited by T. Kawamoto) Vol. 3, pp. 1637-1643 (1985). 21. Murakami A. and Hasegawa T. Back analysis using Kaiman filter-finite elements and optimal location of observed points. In Proc. 6th Int. Conf. Numer. Methods Geomech. Innsbruck (Edited by G. Swoboda) pp. 2051-2058. Balkema, Rotterdam (1988). 22. Sakurai S. Direct strain evaluation technique in construction of underground openings. In Proc. 22nd U.S. Symp. Rock Mech. Boston, MA (Edited by H. H. Einstein) pp. 278-282. MIT (1981). 23. Sakurai S. Displacement measurements associated with the design of underground openings. In Proc. Int. Symp. Field Measurements in Geomechanics, Zurich Vol. 2, pp. 1163-1178 (1983). 24. Sakurai S. and Shinji M. A monitoring system for the excavation of underground openings based on microcomputers. In Proc. ISRM Symp. Design and Performance of Underground Excavations, Cambridge, pp. 471-476 (1984). 25. Shimizu N. and Sakurai S. Application of boundary element method for back analysis associated with tunneling problems. In Proc. 5th Int. Conf Boundary Elements, Hiroshima, pp. 645-654 (1983). 26. Sakurai S. and Shimizu N. Initial stress back analyzed from displacements due to underground excavations. In Proc. Int. Symp. Rock Stress and Rock Stress Measurements, Stockholm pp. 679-686 (1986). 27. Noami H., Nagano S. and Sakurai S. The monitoring of a tunnel excavated in shallow depth. In Proc. 2nd Int. Symp. Field Measurements in Geomech. Kobe pp. 851-859 (1987). 28. Sakurai S., Ine T. and Shinji M. Finite element analysis of discontinuous geological materials in association with field observations. In Proc. 6th Int. Conf. Numerical Methods in Geomech. Innsbruck (Edited by G. Swoboda) Vol. 3, pp. 2029-2034 (1988). 29. Sakurai S. and Tanigawa M. Back analysis of deformation measurements in a large underground cavern considering the influence of discontinuity of rocks (in Japanese). In Proc. Japan Society of Civil Engineers, 403/VI-10, pp. 75-84 (1989). 30. Hansmire, W. H. and Cording, E. J. Soil tunnel test section: Case history summary. J. Geotech. Eng. Div. Am. Soc. Civ. Eng. Ill, 1301-1320 (1985). 31. Sakurai S., Deeswasmongkol N. and Shinji M. Back analysis for determining material characteristics in cut slopes. In Proc. Int. Symp. Engineering in Complex Rock Formations, Beijing pp. 770-776 (1986). 32. Kondoh T. and Shinji M. Back analysis of assessing for slope stability based on displacement measurements. In Proc. Int. Symp. Engineering in Complex Rock Formations, Beijing pp. 809-815 (1986). 33. Shiraishi T., Hirai Y. and Inoue S. Field test on earth retaining for Kurushima Bridges (in Japanese), Honshi Technical Report, Honshu-Shikoku Bridge Authority, Japan, Vol. 14. No. 55, pp. 25-33 (1990). 34. Sakurai S. Monitoring the stability of cut slopes. In Proc. Mine Planning and Equipment Selection, Calgary pp. 269-274. Balkema, Rotterdam (1990).
20 Decision Making in Tunneling Based on Field Measurements KALMAN KOVARI and CHRISTIAN AMSTAD Swiss Federal Institute of Technology, Zürich, Switzerland 20.1
INTRODUCTION
20.2 THE STRUCTURAL BEHAVIOR OF UNDERGROUND OPENINGS 20.2.1 20.2.2 20.2.3 20.2.4
Rock Conditions The Initial State of Stress in the Ground Dimensions and Shapes of Underground Openings Method of Construction and Support Measures
571 572 572 573 574 575
20.3
THE PROCESS OF DECISION MAKING IN TUNNELING
577
20.4
FUNDAMENTALS OF FIELD INSTRUMENTATION
578
20.4.1 The Purpose of Field Measurements 20.4.1.1 Check on the safety 20.4.1.2 The investigation of material behavior 20.4.1.3 Verification of the effectiveness of a particular constructional method 20.4.1.4 Comparison of theoretical studies with observed behavior 20.4.2 The Measured Physical Quantities 20.4.3 Principles for Field Measurements 20.5
DECISION MAKING IN TUNNELING BASED ON FIELD MEASUREMENTS: CASE HISTORIES
20.5.1 Decision Making in Tunneling Based on Convergence Measurements 20.5.1.1 Tunnels of the Imigrantes Highway 20.5.1.2 Pressure tunnel with prestressed concrete lining 20.5.2 Rock Pressure Determination by Measuring the Changes in Curvature and the Strain Along the Tunnel Lining 20.5.3 Decision Making for Tunneling in Swelling Rock Based on the Monitoring of Ground Displacements 20.5.3.1 The swelling process in the vicinity of a tunnel 20.5.3.2 Characteristic line for swelling rock 20.5.3.3 Constructive countermeasures 20.5.3.4 Tunnel design with yielding support 20.5.4 Decision Making in Subway Tunneling 20.5.4.1 Strain profiles in the subsoil due to changes in pore water pressure 20.5.4.2 Interaction between adjacent tunnels and the effect of compressed air
578 578 579 580 580 580 580 581 581 582 585 588 588 588 592 594 595 600 601 602
20.6
SUMMARY AND CONCLUSIONS
604
20.7
REFERENCES
605
20.1
INTRODUCTION
The design of underground openings like tunnels, subways and chambers in soil or rock was in the past almost purely a matter of experience. In the last two decades, however, new methods of site investigation, systematic measurements in the field and computational methods have been introduced as powerful design aids in order to arrive at safe and economical structures. In fact, the increasing worldwide activity in the construction of underground openings and the frequency of large projects even under difficult geotechnical conditions call for a continual improvement in design 571
572
Back Analysis Monitoring
principles. The basic cause for the development of displacements in the ground around the opening or for the occurrence of rock and earth pressure phenomena is the disturbance of the stressfieldin the virgin rock or soil due to the creation of the opening. Each step in the excavation process involves a redistribution of stresses and strains in the ground, thus transforming the primary state of stress and strain into the secondary state. Temporary and permanent support like anchoring and tunnel lining have the task of restoring a new state of equilibrium,firstlyfor the construction period, and secondly for the service life of the structure. In many cases a new equilibrium state is required under the rigorous condition of limited displacements around the openings; for instance, in subway construction, settlements of buildings and traffic surfaces have to be kept to a minimum. 20.2 THE STRUCTURAL BEHAVIOR OF UNDERGROUND OPENINGS The tunnel support (lining, anchoring, etc.) and the surrounding rock form a unit (Figure 1) which is looked upon as the actual structure in tunneling [1]. In practice, the behavior of this structure is often characterized by the nature of the rock pressure, i.e. the effective contact stress between the ground and the lining. The magnitude, distribution and time variation of the rock pressure are important indicators of the sort of problem arising in tunneling. The deformations of the tunnel section and the displacements in the rock together with their time-dependent characteristics, however, are also good indicators and in many cases are practically the only indicators for the behavior of the structure. The protection of the opening against rockfall, keeping the rock pressure under control and limiting the deformations in the most economical way often present the main problems in tunneling. For the solutions of these problems, it must be kept in mind that the behavior of an underground opening depends essentially on the groups of factors shown in Figure 2. 20.2.1 Rock Conditions The scope of the problems which may arise in tunneling is best illustrated by the fact that tunnels may have to be driven through completely cohesionless soil, hard rock mass or through any intermediate type between these two extremes. The materials in tunneling are not chosen, as in some other branches of structural engineering; rather, they are encountered. Their mechanical properties are determined by means of geological surveys and soil and rock mechanics investigations. As far as possible this information should be obtained well in advance of construction. Generally drill holes or Rock anchors
Tunnel lining
^Invert arch Figure 1 Tunnel support and rock, forming a structural unit
Rock conditions
¥
Initial stresses Structural behavior
-+Dimension and shape
Method of excavation
Support measures
Figure 2 Factors influencing the structural behavior of a tunnel
Decision Making in Tunneling Based on Field Measurements
573
adits give access to the material in the area of the planned underground opening. Often, important information is gathered from outcrops on the surface, as well as by using the experience gained from previous constructions under similar geotechnical conditions. The rock properties on the scale of specimen size together with the structure of the rock mass determine properties on the scale of the construction. The rock structure is given by stratification, schistosity and jointing. The latter constitute actual or potential surfaces of separation or slip. Therefore, their frequency and orientation in space are generally of great importance. The material tests in the laboratory comprise soil mechanics investigations, uniaxial and triaxial compression tests and frequently direct shear tests on surfaces of weakness. Load tests in boreholes or even trial sections in tunnels or chambers on a reduced or on full scale can, in certain cases, be applied with advantage as further methods of investigation. Of the many aspects that are important for the geological conditions only two are given special mention here, namely the presence of water and the rock types containing clay or anhydrite. Water inflow in even relatively small quantities into the opening may substantially affect the progress of excavation. The water may reduce the strength of the material by decreasing its cohesion or by the development of pore pressure decreasing the effective normal stresses. When tunneling in saturated soils, special measures, often very expensive, must be taken in order to prevent infiltration and to stabilize the ground, for example grouting, jet grouting, groundwater lowering, utilization of compressed air and hydroshield or ground freezing techniques. Rocks containing clay or anhydrite give rise to special problems in tunneling. Such rocks, e.g. marlstones and anhydrite, can swell, i.e. increase considerably their volume due to absorption of water, whereby a substantial amount of heave in the bottom of the tunnel may occur. The tunnel lining (invert arch), in resisting the heave, may be subjected to high swelling pressures. In tunneling practice, unconstrained heave of up to 70 cm may occur [2] and swelling pressures of up to 3.5 MN m" 2 have been reported to act on the invert arch [3]. Many of the unexpected difficulties that arise in tunneling can be traced back to an inadequate knowledge of the material properties. The actual rock conditions are often, in fact, first known as the underground opening is under construction. This is specially true for deep tunnels, for which borehole explorations, either for technical or economic reasons, are out of the question or else can only be carried out to a very limited extent. Also, one only has to think of the possible variability of the material with respect to its pétrographie composition and its structure (jointing, etc.), then it becomes evident that it is especially important to determine the ranges in which the rock mass behavior may be expected to vary. Here, not only statical but also purely constructional considerations can be important. The greater the degree of mechanization in the method of construction, the more important possible extreme cases in the material occurrence become. For instance, when using the shield tunneling method in soils, if the cutting edge comes up against occasional boulders, a big time delay in construction may result, which leads to increased costs. Turning to another example, the economical application of a full face boring machine with anchors and shotcrete support is not only limited by poor rock quality (too short a stand-up time of the rock, insufficient thrust for the advance of the machine) but also in certain circumstances by a very hard, massive rock. The more uncertain the geotechnical predictions or variable the rock conditions, the more adaptable the constructional method has to be. 20.2.2 The Initial State of Stress in the Ground Due to gravitational forces and possible tectonic influences, the rock is already stressed before the underground opening is excavated. Thus, one speaks of an initial or primary state of stress, which, of course, is different from location to location (Figure 3). There are two ways in which the initial stresses may give rise to difficulties in tunneling. Firstly, the material in the vicinity of the opening often reacts to the changes in the stressfieldby failure and creep processes, which may lead either to the closure of the opening or, if it is hindered, to the development of rock pressure. Secondly, in hard rock at great depths the much feared phenomenon of rock burst may occur. This is characterized by the explosive-like separation of plate-shaped pieces of rock often of considerable size, which may endanger the lives of the people working in the tunnel. The mechanism of rock burst has not, as yet, been adequately investigated. All that is known with certainty is that the orientation of the tunnel axis in relation to the directions of the principal stresses of the initial state of stress plays an important role. The stress tensor in the rock cannot be determined theoretically because of the changing topographical conditions, the generally complex structure of the rock mass and its nonlinear
574
Back Analysis Monitoring
Figure 3 Initial state of stress in the rock
stress-strain relationship and the tectonic forces which may still be active today. Stress measurements in situ are only successful if the rock in the immediate vicinity of the measuring point can be assumed to be elastic, isotropic and homogeneous [4]. Unfortunately, these conditions are as a rule only fulfilled in those cases in which knowledge of the initial stresses due to excellent rock strength is only of secondary importance. Thus, with regard to the magnitude and direction of the principal stresses, we are left with little more than suppositions. For a more or less horizontal surface terrain it is justifiable to assume that the vertical normal stress in the initial state is approximately equal to the overburden stress of the overlying rock or soil. No generally valid statement can be made about the horizontal normal stress component. It can vary from a small fraction to a multiple of the vertical stress. The lower and upper limits for the relationship between the horizontal and the vertical normal stresses may be assessed by the failure condition of the material in the sense of the active and passive earth pressures. It may be noted that the greater the tendency for the material to creep and the greater the overburden pressure, the closer the initial stresses approach a hydrostatic stress condition. Tunnels located in slopes or beneath the bottom of a deep valley require special attention with regard to the initial state of stress. 20.2.3 Dimensions and Shapes of Underground Openings The relationship between the span of the opening and the average joint spacing is in many cases decisive for stability considerations (Figure 4). With increasing span D, or D/d respectively, the influence of the jointing becomes more marked and the probability of an unfavorable joint combination, which could give rise to a rockfall, increases. Thus, in the case of a subway through jointed rock the construction of stations generally requires special considerations, even when the single track tubes might be left completely unsupported. In the particular case of soil with no cohesion the vertical pressure on the tunnel lining in the roof increases with increasing span of the tunnel, the ratio of the span of the tunnel to the height of the overburden being also an important factor. If this ratio is less than one it is not possible to develop a noticeable arching effect in the soil, not even in heavily jointed rock. Especially large dimensions in the construction of tunnels or rock chambers are, from the point of view of safety and economy, only possible by imparting a special shape to the profile. A good example illustrating this point is a chamber in the form of a vertical cylinder with a spherical closure (Figure 5). Statically, this shape is very favorable, for horizontally we have the effect of a closed ring and a double arching action exists at the roof closure. Cavities of this form and with dimensions of about H = 80 m, D = 45 m are at present planned for underground
Figure 4 Influence of the span D on the stability in jointed rock
Decision Making in Tunneling Based on Field Measurements
575
Figure 5 Large underground chamber with statically favorable shape
Figure 6 Possible adaptation of shape to increasing rock pressure
nuclear power plants. The shape of a section is also important in the case of a tunnel. However, as a design parameter it is in many instances not given the attention it deserves. Should rock conditions be encountered in which high rock pressure is expected, the shape of the profile should be selected in such a way that an arching action in the rock and tunnel lining may be developed. In railway tunnels, for instance, this can be achieved by choosing shapes as shown in Figure 6. 20.2.4 Method of Construction and Support Measures The method by which the opening is excavated along its longitudinal direction and in its cross section can have a significant influence on the development of the rock pressure and the displacements in the surrounding rock. In the case of a tunnel the profile can be excavated in a full face operation or by dividing the section into different parts and excavating it in different sequences (heading and bench method, multiple drift method, etc.). Difficulties of various kinds can be overcome more easily when working in smaller cross sections. When the rock conditions require it, the profile must be excavated in two or more stages (Figure 7), whereby staging is also employed in the direction of the tunnel axis. The first stage of excavation is in many cases well in advance of the works for enlarging the section to the full profile, thus providing a useful means of rock exploration. Depending on whether the problem is to control the rock pressure or to limit the displacements in the neighborhood of the tunnel, various constructional procedures may be chosen along the axis of the tunnel. This is illustrated by practical examples, one for a subway construction and the other for a deep tunnel, both driven through a soft rock. For the cross section one can in both cases proceed according to Figure 8. For the subway tunnel in Figure 8(a), in order to avoid undesirable settlements of buildings in its vicinity, the invert arch should be placed as quickly as possible. The time required to complete a full ring may be only a matter of days or a couple of weeks. Thus, at a distance of less than one tunnel diameter a closed ring is formed which is statically extremely efficient. In a tunnel situated at great depth (Figure 8b), where high rock pressures can be developed, considerable deformations may be deliberately permitted using aflexibletemporary support to keep them under control. In any case, it is impossible to prevent the deformations completely even when using a stiff lining, since the pressures that would occur may be of the order of magnitude of the initial stresses (in a depth of 1000 m there would be an overburden pressure of about 30 MN m" 2 in the rock). Thus, with the protection of aflexible,temporary support, one allows radial displacements of the sides of the opening of up to 50 cm or more, which in some cases may take a year to develop. Any further deformation that might occur can then be safely prevented using a suitable closed ring shaped
576
Back Analysis Monitoring (α)
(c)
(b)
Figure 7 Examples of multiple stage excavation in the tunnel section: (a) pilot tunnel with boring machine, (b) head and bench, and (c) side drift tunneling method
(a) s
y
S
,
S
J
>
s
s
s
,
,
,
,
s
s
,
,
I
/ / / S V / / / / S / / / i
Invert
~4-IOm 1-2-3 days
Invert -Temporary support
(b) ^Permanent support
Invert ~ 0 . 5 - 2 years
Figure 8 Placement of the invert arch: (a) subway close to the face and (b) deep tunnel at great distance from the face
Figure 9 Two different blasting techniques applied in the same rock at the construction of adjacent roadway tunnels
Decision Making in Tunneling Based on Field Measurements
577
permanent lining. This may follow the working face of the heading in a distance of a few kilometers. With regard to the conventional methods of excavation only the elementary requirement of carefully controlled blasting, which causes the least disturbance of the surrounding rock, is mentioned here. The rock should not be unnecessarily loosened by blasting, as this would result in a considerable loss of strength. In many instances, heavy lining is necessary only because of poor blasting work. Such a case is shown in Figure 9 together with a tunnel in the same rock but with a smooth rock surface and with no support at all. The indisputable advantage of blast-free mechanical excavation methods is that they do not affect the in situ rock quality around the opening. In summarizing the above, it generally holds true that the method of excavation and the type of support system (rigid orflexible)as well as the time and place of its installation have a profound influence on the behavior of the underground opening. 20.3 THE PROCESS OF DECISION MAKING IN TUNNELING In order to obtain a safe and economical structure the engineer has to make decisions on the following items: (i) location, alignment, shape and size of the opening; (ii) method of excavation, both in the section and in the longitudinal direction; (iii) support measures, temporary and permanent; and (iv) dewatering, ground improvement, etc. Decisions are required prior to, during and, in exceptional cases, also after construction. It must be emphasized that the decisions are not only a matter of purely theoretical consideration but in many cases they are somewhat restricted by contractual aspects. The technical criteria (Figure 10) for correct decisions may basically originate from the safety of the opening during construction and during its service life or from displacement restrictions and in some cases from both. The sources of information for the structural decisions are: (i) geological explorations,fieldtests; (ii) laboratory investigations; (iii) statical computations; (iv)fieldmeasurement; and (v) the engineer's own experience. Again the flow of information generally extends from the initiation of the project up to its completion. Modern tunneling is characterized by the systematic use of all sources of information in a balanced manner. A clear understanding of the factors influencing the behavior of an underground opening under specific conditions can only derive from the engineer's own experience and from his theoretical knowledge. Experience manifests itself in good structural judgement. Together with laboratory investigations, statical computations and field measurements it forms the basis for decision making, both at the planning and the construction stage. To what extent such modern aids should be applied on a given project depends solely on the nature of the problems that arise. In the following, an attempt will be made to give an up to date survey of the possibilities and limitations offieldobservation techniques. Computational methods as a design aid in tunneling have been discussed elsewhere [5]. Here we only want to point out that by means of statical computations an analytical prediction of the structural behavior of the opening is obtained. The interrelationship between the various factors, for instance rock properties, shape and dimensions of the opening, initial state of stress, etc., may be clearly seen in the calculated results. But although these results are available at the design stage, they are subject to great uncertainties. Measurements, on the other hand, enable its behavior to be observed directly, without the actual mechanism which gives rise to its behavior necessarily being completely illuminated. The measurements are usually carried out during the constructional phase and if carefully planned and executed they give a true picture of the behavior of the structure. From
Criteria
Displacements
Safety
Settlements Rockfall
Breakdown
Collapse Figure 10
Closure
Criteria for decision making
578
Back Analysis Monitoring
these considerations it is obvious that computations and measurements complement each other and only when combined are they capable of leading to a correct explanation of the structural performance in complex geotechnical situations.
20.4
FUNDAMENTALS OF FIELD INSTRUMENTATION
The practical significance of systematic measurements for a given project depends upon the extent to which the results of the continuous observations are able to influence the constructional work. This point is well illustrated by means of two examples. The first one concerns the case of shield tunneling with lining segments. Here, the most important constructional decisions, for instance deciding upon the shield diameterbased on the anticipated soil movements and lining deformations, or designing the segments themselves, have to be made well before the start of the construction. The observation of the actual deformations of the tunnel profile, the movements of the surrounding ground or settlements at the ground surface mainly have the function of checking the structural behavior with regards to a satisfactory design and proper execution of the works. In this way shortcomings arising in backfilling the space between the rings and the ground or concerning insufficient support of the tunnel face can be detected. Using a tunneling method with shotcrete and anchoring as a support, which may in cases of favorable ground conditions also be envisaged in subway construction, extensive measurements can really serve as feedback signals for the constructional process. Here, on the basis of careful statical computations, a concept is worked out for the excavation sequences both in the cross section and along the axis, and for the corresponding support measures. If the measurements indicate a substantial deviation from the anticipated behavior of the structure, then the most important corrective measures in the construction can still be applied. The above comparison of the two methods of construction restricted itself to the possibilities of influencing the tunneling process by a proper use of measurements and should in no way be regarded as a general evaluation of the two methods. Which of the two methods of construction should be applied in a particular case is decided, of course, by economy and the attainable progress in advancing the tunnel.
20.4.1
The Purpose of Field Measurements
In general the real purpose of field measurements lies in the optimization of the design and execution of underground structures. In other words, the aim is to obtain adequate safety for a minimum of cost expenditure, whereby the manifold influence of the construction time is also included in the costs. This does not exclude, however, the conscious decision to accept a calculated risk. Since the problem of optimization is very varied, the immediate objective of the measurements themselves may be concerned with quite different aspects, the most important of which are: (i) the safety control; (ii) the investigation of material properties and possibly the determination of the initial state of stress; (iii) the verification of structural response to a specific method of construction; and (iv) the comparison of theoretical predictions with the actual structural behavior. As a general rule, the above classification of the objectives of measurement is not rigid. It is intended to indicate the main emphases. It should be noted that with the same program of measurement usually several aim's are envisaged. The most important thing is that the concept, the execution and the interpretation of the measurements are adjusted to suit the needs of the problem in hand.
20.4.1.1
Check on the safety
As a rule, completed underground structures exhibit an excessive safety. On the other hand during construction a variety of tunnel hazards may occur which emphasize the importance of safety considerations. Since it is very difficult, however, to quantify a safety concept, the tendency in tunneling is often to speak of safety simply in a qualitative sense. Systematic measurements can provide a great deal of help here too, since, for example, using observed deformations it can be estimated if the structure or its parts are reaching or have already reached a condition of stable equilibrium, or if instabilities or inadmissibly large deformations are to be expected. Measurements can serve therefore as a possible warning system enabling preventive measures to be introduced in proper time. The correct interpretation of the observations, i.e. the establishment of warning levels,
Decision Making in Tunneling Based on Field Measurements
579
may, however, present a difficult problem when the displacements increase steadily in time but with a decreasing rate. If only small deformations are permitted in the vicinity of a tunnel, as is often the case in subway construction, then not only the safety of the underground opening itself but above all that of the neighboring structures is of prime interest. Systematic displacement measurements are most frequently employed for safety checks. 20.4.1.2 The investigation of material behavior The deformational properties of the material on a small scale can be estimated using tests such as a borehole dilatometer or the loading plate of a flat jack. In the tests an active loading is applied to the rock and the resulting deformation is measured. From the observed load-deformation diagram and with the aid of the theory of elasticity (with very simplified assumptions) a so-called deformation modulus of the surrounding rock is estimated. An essentially different concept of measurement is based on the realization that by excavating underground openings, such as galleries, tunnels or caverns, the rock mass is unloaded on the scale of the structure itself. To be more exact, it is a question of changing from the initial to the secondary state of stress, which is accompanied, of course, by deformations. By measuring these deformations and with an assumption regarding the initial state of stress it is possible with the aid of a suitable computational model to calculate the 'global deformability modulus' for the rock, which may yield an important indication of the overall rock quality. Although this method of back analysis has its limitations, it gives useful information about the in situ deformation characteristics of the material on the scale of the structure itself. In many cases not only a quantitative assessment of the deformation properties is sought, but also a technological characterization based on measurements. One might, for instance, want to find out the nature of rock pressure, which is to be expected in a particular formation and under given conditions (dimension of opening, height of overburden, method of construction, etc). For this purpose measurements in access tunnels, drifts, trial headings, etc. are advisable. From the amount, time variation and spatial distribution of the measured displacements at the boundary of the tunnel excavation and in the rock some clues for the nature of the present or anticipated rock pressure can be gained. In a situation with loosening pressure, large deformations are generally observed in the area of the roof, which usually can be brought to a standstill in a short time with just temporary support measures (Figure 11a). In the case of genuine rock pressure the displacement field is fairly uniform around the opening and stretches far into the surrounding rock (Figure lib). The deformations continue to increase over a long period of time (years) and in many cases do not stop until a permanent lining has been constructed. The third type of rock pressure, namely swelling pressure, only occurs in rock containing clay minerals (illite, montmorillonite) or anhydrite. The volume increase (swelling) due to absorption of water might reach such proportions as to render the structure inoperative, if no special precautions are taken. Experience shows that swelling is confined to the area of the bottom of the tunnel (Figure lie), and the resulting deformations exhibit the character of genuine rock pressure. Field measurements also provide useful indications here to estimate the swelling potential of the surrounding rock or the swelling pressure, if the deformations are prevented by an invert arch construction.
(α)
(b)
(c)
Figure 11 Typical displacement fields associated with different types of rock pressure: (a) loosening pressure, (b) genuine rock pressure, and (c) swelling pressure
580 20.4.13
Back Analysis Monitoring Verification of the effectiveness of a particular constructional method
The optimum execution of a particular constructional concept can only be achieved, in many cases, if individual aspects like the span of the unsupported roof section, the enlargement of the cross section, the arrangement and the capacity of temporary supports, the time for introducing the permanent lining, etc. are determined on the basis of in situ measurements. The greater the uncertainty of the geotechnical prediction, whether it is due to inadequate site investigation or to the absence of sufficient experience in working in the given rock, the greater the flexibility that one should have to be able to make correct engineering decisions during construction. By means of a suitable monitoring program and statical considerations one can then check the effectiveness of the specific constructional measures decided upon, and thus, while preserving adequate safety, the object can be more economically constructed.
20.4.1.4 Comparison of theoretical studies with observed behavior Here, primarily the verification of the theoretically assumed behavior mechanisms is implied. The selection of the physical quantities to be measured and the arrangement of the instruments are based on careful preliminary investigations of a theoretical nature. The computational results do not agree numerically, as a rule, with the measured values, but by varying the parameters and with the aid of several computer runs a better agreement can be achieved. However, if fundamental deviations between theory and reality occur this indicates that some factors, which because of too great a simplification of the model were left out of consideration, are in fact of greater significance than was originally assumed. One only has to think of the time effect, for instance, which is neglected in the usual assumption of an elasto-plastic continuum for the rock mass, or the influence of a complex three-dimensional state of stress, which cannot be considered in a conventional plane strain analysis, which is generally used in statical analysis. The above classification of the objectives of measurement is only intended to point out the most essential features, as in many cases several aims are envisaged with the same measurement program.
20.4.2 The Measured Physical Quantities Depending on the particular problems, the observations most frequently refer to one or to a group of the following physical quantities: (i) strains; (ii) relative displacements; (iii) absolute displacements; (iv) changes in curvature (in tunnel lining); (v) stresses in lining and in rock mass; (vi) rock or earth pressures on tunnel lining, forces in rock anchors; and (vii) piezometric heads. When planning a measuring program some sound principles have to be followed in order to obtain useful results for practical purposes with a minimum of cost expenditure.
20.4.3 Principles for Field Measurements The main principles for field instrumentations and field measurements are as follows. (i) Correct formulation of the structural problem, the solution of which requires observations. (ii) Selection of the most sensitive physical quantities. (iii) Assessment of the order of magnitude of the measured quantities; conclusions with regard to required accuracy. (iv) Selection of measuring techniques, instruments, location of measuring sections, reading program. (v) Assessment of possible sources of error in the readings well in advance. (vi) Application of monitoring with overlapping results for complex situations. (vii) Employment of reliable instruments and competent personnel only. (viii) Continuous data processing, establishment of tentative emergency levels, correct flow of information. Experience shows that when observing these principles,fieldmonitoring really turns out to be an invaluable aid in the design and execution of underground works.
Decision Making in Tunneling Based on Field Measurements
581
20.5 DECISION MAKING IN TUNNELING BASED ON FIELD MEASUREMENTS: CASE HISTORIES With the help of examples chosen from tunneling practice, the basic considerations given above will be further discussed in the following sections. When dealing with case histories the actual problems arising in the various projects will be briefly formulated, the applied monitoring technique referred to and the relevance of the obtained results to the constructional problems discussed. The measuring techniques used and the associated instruments are described in the literature [9,13-17]. As a rule, the authors give preference to displacement measurements (convergence of the opening or movements in the rock), since in a mathematical sense they represent integrated quantities and are basically not subject to local effects. Stresses, strains or changes in curvature, on the other hand, are differential quantities, whose validity is limited to local regions. When being measured, therefore, they should be observed at several successive points, so as to obtain their distribution over a sufficiently great area. In this way the predictive value even of differential quantities can be substantially improved. 20.5.1 Decision Making in Tunneling Based on Convergence Measurements The measurement of convergences, i.e. of the changes in distance between two points of the excavation or lining surface, is one of the simplest and least expensive operations. In Figure 12 typical applications are shown for a tunnel with different construction sequences and for a circular tunnel section. Figure 13 indicates how the complete distortion of the cross section may be determined by a mesh of measuring lengths. The displacement vectors wf and vt of a point are referred to an arbitrarily selected kinematical system A-B. Generally, three displacement vectors should be known or fixed. It is advantageous to introduce in a mesh some control lengths V as indicated in Figure 13. In such a manner the reliability of the individual readings can be checked. Using a computer program for data processing the mesh can be adjusted by the method of least squares, thus increasing overall accuracy. In many instances such simple measurements are carried out merely to ascertain whether a state of stable equilibrium has already been reached, will be reached, or instabilities are to be expected.
(a)
Figure 12 Convergence measurements with typical arrangements of the measuring lengths: (a) tunnel with different construction sequences and (b) control of diameter change in a gallery
-Kinematic system of reference Figure 13 Determination of the complete deflection of a tunnel lining by a mesh of individual convergence measurements
582
Back Analysis Monitoring
20.5.1.1 Tunnels of the /migrantes Highway This three lane 55 km long highway connecting Sao Paulo with Santos in Brazil involved the construction of altogether 18 viaducts and 11 tunnels with the total length of 3825 m [6]. The tunnels - most of them slope tunnels - were constructed under difficult topographical and geomechanical conditions [7] and excavated simultaneously by different contractors. The tunnels present a considerable size of profile with a sectional area of 120 m2. The general sequence of excavation is shown in Figure 14. The convergence measurements [8] with the distometer [12] were expected to give information about (i) the type of rock pressure which may develop in different sections of a particular tunnel (it was important to identify the type of possible rock pressure phenomenon); and (ii) the stability of the slopes affected by tunnel construction. Different arrangements of the measuring lengths were used. The observations had to cover all stages of the construction, as shown in Figure 14. Special attention was paid to the behavior of the temporary lining in the calotte, which consisted of steel ribs with or without shotcrete. Figure 15 shows an example of readings in the case of a local instability, which occurred in tunnel TA-4 near to the face. By means of monitoring, this instability could be detected in its very early stage, thus permitting the installation of emergency supports formed of wooden timbers (Figure 16). The monitoring of the permanent lining in the calotte during core removal and side wall construction (Figure 17) was of great interest, too. In some cases, the effect of these constructional measures on the readings could be clearly observed (Figure 18). When interpreting the readings of all 21 monitored measuring sections in seven tunnels the following points were kept in mind. (i) The permanent lining forms together with the surrounding rock essentially a three-dimensional structure. This statement holds especially true when considering an asymmetrical excavation process and a step by step side wall construction. Simplification to a two-dimensional case is allowed whenever uniform conditions prevail in the vicinity of a measuring section. (ii) The type of rock pressure phenomenon and the supporting effect of the lining may be estimated from the order of magnitude and the rate of the deformations. (iii) Due to time limitations a useful back analysis could not be carried out. Systematic field measurements in the Imigrantes Tunnels have shown quantitatively that no exceptional rock pressure phenomenon occurred in the tunnels. The rock was self-supporting as it had been supposed at the time of the first site inspection. Only loosening pressure has occurred. It was clear that the permanent lining of the tunnel was considerably overdesigned. Although the
4 Central cut
5 Pit excavation
6 Side wall concrete
7 I n v e r t excavation and concrete
Figure 14 Imigrantes Highway - construction sequences of the tunnels [7]
Decision Making in Tunneling Based on Field Measurements ^
583
Time (days) 14
18
22
26
30
Placing of permanent lining
Figure 15 Indication of failure in the tunnel TA-4 by the distometer [12] reading, Imigrantes Highway [7]
Figure 16 Emergency support in tunnel TA-4
measurements were initiated at a very advanced constructional stage, important savings in concrete and steel could still be achieved. The readings further demonstrated that the core removal could take place in one working unit and not in short trenches as it was originally assumed, thus speeding up the excavation and making it less expensive. On the other hand, the problems related to slope stabilities in the portal zones deserved some attention. The large dimensions of the tunnel sections, the unfavorable alignment of the tunnels in steep slopes with slight overburden in the portal zones and some critical cuts for the 'service road' have raised the problem of slope stability. During the first site inspection it could easily be recognized that the problem of slope stability in the portal zones had in some cases been overestimated but in other cases almost completely ignored. By means of an adequate measuring program considerable savings in time, effort and money could have been achieved for this type of problem, too.
584
Back Analysis Monitoring
Pit excavation for wall construction, Imigrantes Highway [7]
Figure 17
E E ~
12
E E
-8
σ> c o
E Φ φ
1 c
o
c
cH
rfl
± Ί
mW
Ö58I
-20
90
30
Time (days)
Figure 18 Influence of the core removal and the side wall construction on the control lengths. Tunnel TA-9, Imigrantes Highway [7]
Decision Making in Tunneling Based on Field Measurements 20,5.1.2
585
Pressure tunnel with prestressed concrete lining
For the construction of the hydroelectric power scheme Grimsel-Oberaar in the Swiss Alps, the pressure tunnel in igneous rock was lined with prestressed concrete. This method of construction is especially suitable if the tunnel is subjected to a relatively low internal pressure, i.e. a value not much in excess of 1.5 MN m~ 2 [10]. The internal diameter of the tunnel is 6.8 m, the depth of the concrete lining 0.4 m and the internal pressure 0.75 MN m" 2 . The cross section of the tunnel with the cable guides at position 1 of the stressing location is shown in Figure 19(a). The cables (system VSL) were laid alternatingly at intervals of 20 cm in the positions 1 to 4 (Figure 19b). The breaking capacity for each cable was 1547 kN. The function of the prestressing was to prevent cracking in the concrete due to internal pressure, in order to ensure that the pressure tunnel remains leakproof. The interaction of the concrete with the rock was thoroughly investigated at the planning stage. Parametric studies with the finite element program RHEO-STAUB [11] were carried out in cooperation with the design engineers and the contractors. The aim of these computations was to throw some light on the question of whether the interaction of the concrete lining with the rock could hinder the desired build-up of compressive stresses in the concrete. The interaction was simulated as the embedment of an elastic ring in an elastic medium. The Young's modulus of the medium corresponded to the deformation modulus of the rock. The loading was given by the forces due to the prestressing of a single cable. In order to eliminate tensile stresses between the ring and the elastic medium the computations were carried out iteratively. As a result of these parametric studies using a simplified computation model, any apprehensions about the possible transmission of the prestress effect from the concrete to the rock could be dispelled. It was decided to check this result by in situ measurements as well and to investigate further the behavior of the prestressed ring for the alternating cable positions. In particular the deformation of the concrete ring and the separation of the ring from the surrounding rock had to be evaluated. By adjusting the computational model better to the actual conditions in situ it should also be possible to test the theoretical predictions against the real behavior of the structure. Two measuring sections 40 m apart were fitted out with eight distometer measuring lengths (Figure 20). Based on preliminary computations the anchor position of the extensometer in a depth of 2.2 m could be regarded as a fixed point. The measuring lengths between the diametrically opposed measuring heads of the extensometer have solely a control function. Readings were taken before applying prestress and in steps of 25%, 50% and 100% of the maximum stress. The changes in diameter for the loading case of 100% prestress force are given in Table 1. A comparison of the results of the convergence measurements using the distometer [12] with those of the single point extensometer shows that the anchor position of the extensometer is in fact a fixed point. The radial deformations of the concrete lining in two measuring sections for the last stage of loading (100% prestress) are shown in Figure 21. The variation of the deformations between the individual measuring points is unknown. The curves in this figure are based on arbitrary estimates and serve simply the purpose of giving a visual representation of a possible ring deformation. In reality, the concrete lining is a long cylindrical shell, which rests in places on the rock, and it is quite possible that in certain sections the lining is not in contact with the rock over its whole circumference. This could be the case in measuring section 2. The formation of a gap between
Figure 19
Prestressed concrete lining for a pressure tunnel: (a) layout of the tendon in position 1 and (b) positions of the tendon in subsequent sections
Back Analysis Monitoring
586
Anchor Extenso meter Measuring head
Concrete lining
Figure 20 Layout of the extensometers and the convergence measuring lengths Table 1 Change in Diameter, δ Θ
<5Ε - ^D
<5Ε
<5D
(mm)
(mm)
(%)
<5D
Measuring section 1
0 π/4 π/2 3π/4
5.85 3.97 9.95 4.04
5.26 3.93 9.83 4.57
11 1 1 12
Measuring section 2
0 π/4 π/2 3π/4
3.65 1.44 12.90 3.41
3.50 1.70 12.74 3.50
4 15 1 3
δΕ, extensometer measurements; <5D, distometer measurements.
Measuring section 2 Measuring section I
° Extensometer readings ~7
7
7~
Figure 21 Results of the extensometer measurements for the loading case 100% prestress force
the lining and the rock is of practical significance in preventing a transfer of the prestress force to the rock. Having carried out the prestressing along the whole length of the tunnel and after the completion of the measurements this gap wasfilledby grouting. If one plots the observed changes in diameter as a function of the angle Θ (Figure 22), reliable deformation curves can be obtained due to
Decision Making in Tunneling Based on Field Measurements
587
rr/Z
Figure 22
Figure 23
Distometer measurements at three different levels of prestress force in measuring section 2
Results of computations and observations for the loading case of 100% prestress force
Figure 24
The prestressed concrete lining under construction
the large number of measurements taken. It was attempted, therefore, to select the assumptions for a computer model, such that the theoretical and experimental curves come together as close as possible. The distribution of the computed values shown in Figure 23 is based on the following assumptions. (i) The support positions of the elastic ring are at θχ = — 30° and θ2 = 210°. (ii) The Young's moduli for concrete and rock are Ec — 15 000 MN m" 2 and ER = 2000 MN m" 2 respectively.
588
Back Analysis Monitoring
(iii) The influence of different cable positions is obtained by superimposing the effects of the individual cables. The agreement achieved in this example between computation and observation is considered to be reasonable. One has to realize that the three-dimensional structure has been simplified to two dimensions for the sake of the computation. Also, the concrete behavior was assumed to be linear elastic, although (as evident from Figure 22) its behavior is in reality distinctly nonlinear, e.g. the deformations at 100% prestress are more than double those at 75% prestress. Figure 24 shows the prestressed concrete lining under construction. 20.5.2 Rock Pressure Determination by Measuring the Changes in Curvature and the Strain Along the Tunnel Lining Rock pressure causes the lining of a tunnel to deform. An alternative method to determine the deflection of the lining is given by systematic measurements of both change in curvature and strains along the intrados of the tunnel lining in consecutive points. As an advantage, the internal forces (bending moment and normal force) and the rock pressure also can be calculated [9], provided that the material properties of the lining material are properly defined. A more detailed description and practical applications of this method can be found in Chapter 24 of this volume. 20.5.3 Decision Making for Tunneling in Swelling Rock Based on the Monitoring of Ground Displacements Rocks containing clay minerals and anhydrite increase in volume when they come into contact with water; this phenomenon is referred to as the swelling of these rocks. In tunnel construction, swelling of rocks manifests itself as a heave of the tunnel floor or as pressure on the invert arch (Figure 25). When the lining remains intact, a heave of the entire opening can occur, where the crown as well as the floor experience an upward displacement. In some cases, the pressure resulting from the swelling of the surrounding rock leads to a failure of the invert arch. In view of the good number of underground construction projects that are planned to be carried out in rock formations with a swelling potential, it is of significant economic necessity to develop viable methods of countering swelling pressures and swelling deformations. For a successful design of a tunnel in swelling rock, field measurements are an extremely important source of information. That is the reason to present the different elements and steps for the design of tunnels situated in heavy swelling rocks in more detail. For this problem, the systematic use of all sources of information such as field measurements, laboratory investigations, statical computations and last but not least the engineer's own experience is demanded. 20.5.3.1 The swelling process in the vicinity of a tunnel Our presentation of the swelling mechanism is based on the observations of Terzaghi [21], which state that the swelling of the rock is brought about by changes in the state of stress resulting from the excavation. A relief of the stress component normal to the surface of the excavation takes place in the vicinity of the opening; at the surface, this component reduces to zero. Other factors must be
Heave of the tunnel floor
Figure 25
Pressure on the tunnel lining
Effects of swelling in tunneling
Decision Making in Tunneling Based on Field Measurements
589
relevant, however, since we observe the swelling only in the area of the tunnel floor. The principal aspects to consider have to do with the water conductivity of the rock, i.e. those factors which have an influence on the water seepage. Computational methods which take only the stress relief but not the movement of water into consideration would predict swelling not only in the floor but also in the walls and in the roof of the tunnel; this is clearly in conflict with observations made in thefield.The water thatflowstowards the excavated area tends to move towards the floor of the opening (effect of gravity), leaving the upper rock zones untouched. Thus, the swelling which would be expected in these zones as a result of the stress relief does not take place. Other authors [27, 28] have proposed formulations that take into account the local three-dimensional state of stress. The results of such computations also depend highly on the initial state of stress and in particular on the relation between the horizontal and vertical normal stress components. Neither the initial stresses nor the stress redistribution resulting from the excavation, however, can be measured in swelling rock formations at the present state of measuring technology. In a recent research work, Anagnostou [38] proposed a new numerical model that takes account of the movement of water inside the rock mass and the interaction of the pore water with the rock matrix. It allows the simulation of time-dependent effects. The swelling rock is modeled by elasto-plastic constitutive equations. The effects of the swelling of the surrounding rock on the structure are in most cases monitored in thefieldby means of leveling and convergence measurements. Borehole extensometers are also often called for to obtain a picture of the deformations in the rock surrounding the tunnel. It has been possible for some time now to follow the in situ swelling of rock with great accuracy and detail by using a sliding micrometer (Figure 26). This instrument allows us to measure the strain profile along the entire length of boreholes. Along a vertical borehole, ring-shaped measuring marks connected by hard PVC casings arefixedat intervals of 1 m by means of grout (Figure 26). To take a reading beginning at the mouth of the borehole, the probe is lowered in sliding position (Figure 27a) and fixed in the two measuring marks (Figure 27b). There, a reading of the axial LDT gauge is taken and sent to a hand-held computer. The probe is then traversed in a stepwise manner down along the borehole taking readings at every step in a similar manner. After reaching the bottom of the measuring line, a repetition of all readings is made on the return way of the probe. This allows an immediate check of accuracy by the computer in the field. Spurious readings can be detected while still at the site and the readings can be repeated if necessary. Thefieldaccuracy of this portable type of probe is better than + 5 x 10 ~6. Since the standard length over which each strain measurement is made is 1.0 m, strain is usually expressed in mm m" 1 . The change in relative distance between two consecutive measuring points can thus be determined with an accuracy of + 5 μπι. A detailed description of this instrument is given elsewhere [14, 15]. Several examples of strain measurements in road tunnels through formations of swelling rock are presented in the following. The Pfaender Tunnel near Bregenz in Austria [18] traverses interbedded layers of marl and sandstone over a distance of about 4.4 km. The tunnel floor, which was supported only by a cover of
Figure 26 (a) Operation of the sliding micrometer probe in the borehole using a cable winch, (b) Readout unit with data terminal
590
Back Analysis Monitoring
Figure 27 The sliding micrometer borehole probe. Schematic view of the instrument in (a) sliding position and (b) measuring position in the borehole
shotcrete, started heaving during construction. An anchoring system, consisting of corrosion resistant, pretensioned anchors with a design strength of 1000 kN, was called for. The length of these anchors is 10 m with a fixed anchor length of 5 m. Along with the anchoring work, two boreholes, each 18 m in length, were drilled at two locations in the tunnel, and equipped with tubing for strain measurements using a sliding micrometer. Readings have been taken regularly since 1980 and are presented in Figure 28. A layer about 2 m thick, which is situated directly under the tunnel floor, shows compressive strains. This layer is located between the heads of the pretensioned anchors and a larger, underlying mass of swelling material. Strains resulting from swelling of the rock decrease with depth, and reach zero about 12 m under the tunnel floor. The deformations, which are recorded as a function of time, are tending to level off. A total heave of only 1 mm was observed over the nine year measuring period, which demonstrates the effectiveness of the anchoring. Figure 29 shows the results for the second borehole located about 1.9 km from the first. In this case, a 2 m thick swelling layer located directly under the tunnel floor was identified. This layer caused a heave of the floor of 5 mm. We now turn to the observations made in the T8 (Sonceboz-Biel) tunnel in Switzerland (Figure 30), which traverses layers of marl of high strength. The strain measurements in the borehole could be made only up to a level of 1.5 m under the floor. Displacements here are also remarkably small, although noticeable heaving was recorded right after excavation. A decrease of strain with increasing distance from the excavation is also noted here. A further example of an observed swelling process in the vicinity of an excavation is the Belchen Tunnel of the N2 highway (Switzerland). Long stretches of the tunnel run though formations of
>—i
-1.0
1
-0.5
1
0
1
0.5
1—..
1.0
Extension Compression €(mm rrf1)
ÜJ
I—i—:
1980
i
85
i
1989
Time (year)
Figure 28 Pfaender Tunnel: (a) strain profile ε in the rock and (b) time dependence of strains in layers A and B in measuring section 1
591
Decision Making in Tunneling Based on Field Measurements (α)
f~5mm II
II
(b)
1989
Section C E E
Im
JE
vu c o
-3.0
-2.0
_i_
-1.0
-2.0h
1980
0
Figure 29
1989
85
Extension, € (mm nrf')
Time (year)
Pfaender Tunnel: (a) strain profile ε in the rock and (b) time dependence of strains in layer C in measuring section 2
(a)
(b) Sections A and B £ E E
I
-0.8
_ ^Ί\
w c o
c
a> +-
X
-0.8
Extension, € (mm nrf') Figure 30
^/^
-0.4
> /
/>-
Γ^
s'
^
B
1983
J — ^ 1987
Time (years)
Tunnel T8: (a) strain profile ε in the rock and (b) time dependence of strains in layers A and B
Opalinus clay and rock containing anhydrite (Jurassic formations, ref. 'Gipskeuper'), both of which are well known for their particularly pronounced swelling characteristics. Detailed reports have been made of the damages which occurred during construction [20]. An extensive study including sliding micrometer measurements was ordered in 1986 to check the condition of the tunnel 18 years after its completion. Ongoing swelling was still noted in four measuring sections, an example of which is given in Figure 31. The strains due to swelling over a four year period can be seen in this figure to reach values of approximately 3.5 mm m"1; this corresponds to an extension of about 1% per year. The increase in swelling pressure on the invert arch that goes along with these increments in strain and the cause of the gaps in the strain profile are not known. For the latter, it is assumed that the material in those zones did not swell, or that local swelling has reached a maximum value under the present stress state. It is also conceivable that groundwater had not yet found access to this area. Summing up our present insights on the effects of swelling in tunnel construction gives the following. (i) Swelling processes take place - to an extent which is relevant to practice - only in the floor area of excavations. (ii) The swelling strain decreases with increasing distance from the excavation, reaching zero at a distance of about one diameter from the bottom of the opening.
592
Back Analysis Monitoring
Figure 31 Belchen Tunnel: strain profiles in the Gipskeuper for two subsequent measuring periods of six months each (18 years after the completion of the tunnel)
(iii) Swelling processes can take place more or less rapidly, depending on the facility of access the water has to the floor area and on the water conductivity of the rock. In some cases, swelling processes continue over several decades after the completion of an underground opening.
20.5.3.2
Characteristic line for swelling rock
The characteristic line procedure was proposed by Lombardi for the quantitative description of the interaction between the invert arch and swelling rocks [25,26]. 'Characteristic line' here refers to the relation between the heave of the tunnel floor l/ a and the lining resistance Pa (Figure 37). The lining resistance is the pressure the lining exerts on the surrounding rock. The reaction to it is the swelling pressure, i.e. the pressure exerted by the rock on the lining, e.g. on the invert arch. We have proposed a simple procedure for the determination of the characteristic line that does not require consideration of initial stresses [32]. Initial stresses as well as stresses resulting from the excavation can be neglected by presupposing the knowledge of the strain profile ε(ζ) of the swelling rock beneath the invert arch (Figure 32). In homogeneous rock, the maximal swelling strain value ea is shown at the excavation surface with the strain decreasing to zero at about one tunnel diameter D from the opening. Along with this fundamental assumption on the strain ε(ζ), we use a second assumption, as other authors do [20, 27, 28], namely the 'swelling rule' for the rock [32]. Not only individual laboratory tests but the entire stratigraphy below the tunnel floor must be considered in the determination of the parameters a and b (Figure 32). If only few oedometer test results on samples of small size are available, the sandwich structure of the rock must be taken into account for the determination of the parameters a and b. Representative swelling parameters a and b can be estimated on the scale of the rock formation, if the stiffness and swelling behavior of the individual layers are known. Given the assumptions in Figure 32, the floor heave i/ a for rock that is homogeneous with respect to swelling can be obtained from the integration of ε(ζ). One thus obtains the general form l/ a = kDsa
where k is a shape factor for the ε(ζ) strain curve. Three examples of possible strain distributions are represented in Figure 33. Using the so-called log-normal 'swelling rule' (Figure 32), already mentioned by Terzaghi [37], we find the characteristic line of a swelling rock formation in the general form C/a =
Α-Β\ο%ΡΛ
The characteristic line thus derived exhibits basically the same behavior as the 'swelling rule' curve for the oedometer tests. The unrestrained floor heave L/ao corresponds to the unconstrained strain due to swelling. The maximal value of the lining resistance corresponds to the swelling pressure. Theoretically, the characteristic line represents equilibrium conditions if no more measurable displacements can be observed. Two interesting, practical conclusions can be drawn from the
Decision Making in Tunneling Based on Field Measurements Assumption I
593
Assumption 2
D
^-—Γ-^Λ° < Depth ■
*
=
)*
(
*
Swelling pressure ( log
Uo=f
e-a-b
log σ
Figure 32 The two fundamental assumptions for the determination of the characteristic line of the rock: the strain profile in the tunnel floor and the swelling rule for the rock
logarithmic relation shown in Figure 34. On the one hand, a small allowed floor heave ( + Ai/ a ) results in a relatively large decrease in swelling pressure ( — APa); also, conversely, a relatively small lining resistance ( + APa) causes a significant decrease of the floor heave ( — A(7a). Tests and measurements in the field must be made to confirm whether rock with a swelling potential in the floor area of an excavation behaves in accordance with the characteristic line in Figure 34. Individual observations which, according to this figure, confirm the effect of a relatively small lining resistance on the heave due to swelling are already available [32]. Ua = kD€a
* = l/2
z A- = 1 / 3
Figure 33 The shape factor k and the floor heave Ua for different strain profiles
U^A-B
log Pa
Resistance of the structure, Pa Figure 34
Attributes of the characteristic line for small floor heave ( + Δ Ι / 8 ) and low lining resistance ( + ΔΡ 8 )
594
Back Analysis Monitoring
20.5.3.3 Constructive countermeasures The design solutions shown in Figure 35 can be used to counter the undesired effects of the swelling rock [22]. Using a resistant invert arch or an anchoring system aims at restricting heaving of the tunnel floor. In this case, significant swelling pressures can result. We know of only one example of a large scale anchoring system, namely the Pfaender Tunnel near Bregenz, which was mentioned earlier. The use of an open space under the roadway plate, on the other hand, is aimed at avoiding the build-up of swelling pressures [26]. The solution using an invert arch on yielding supports is a compromise between both extremes. Here, both swelling and the development of some swelling pressure are allowed in a controlled way. The constructive design, i.e. the dimensioning of the individual structural support elements, is made based on two criteria [30]. On the one hand, enough bearing capacity must be available to prevent failure under swelling pressures. On the other hand, deformations (especially heaving of the floor) must be kept within acceptable limits (Figure 36). In some cases, like in tunnels for high speed trains, where strong limits are imposed on warping and alignment of the rails, the second criterion may be decisive. In this case, the heaving by sections of the entire tunnel lining or its twisting can lead to more serious problems than higher swelling pressures. The particularities of the various design measures and the interaction of the lining with the surrounding rock can be illustrated with the characteristic line. Figure 37 shows the log-normal characteristic line derived above. Let us start by examining the case of an invert arch. The heave of the tunnel floor is strongly constrained by the great stiffness of the arch. As a result, a swelling pressure is imposed on the arch; this pressure may approach the extreme value P* (Figure 37). This is particularly the case if the invert arch is placed under dry conditions before any swelling takes place. The other extreme is the open tunnel floor. The implementation of a yielding support system between the invert arch and the rock results in values of Pa and Ua which depend on the deformation properties of the construction elements (Figure 37). The solution to be implemented for a given project should be in accordance with the swelling parameters of the rock and the design criteria (Figure 36). An invert arch seems to be the simplest and most economical solution for smaller extreme values of P*, which can be taken up by an arch with a thickness of 0.4-0.5 m and a moderate reinforcement of the concrete. The open tunnel floor should be considered only where the maximal heave of the floor remains small. In this case, the lining lacks the strength of the statically advantageous closed ring form. There is also a risk of the side walls caving in if the heave of the floor is too large. The construction method using an invert
Invert arch
Anchoring system
Open space
Yielding support
Figure 35 Design measures used in swelling rocks
Failure
Heaving of the floor
Figure 36 Design criteria in swelling rocks
Decision Making in Tunneling Based on Field Measurements
595
0 B (Open floor)
Resistance of the structure, Pa
Figure 37 Log-normal characteristic lines for different design measures
arch on a yielding support should in general be the optimal solution for higher values of P*. We therefore examine this construction method in greater depth here. 20.5.3.4 Tunnel design with yielding support The fact that a reduction of swelling pressure on the invert arch results in increasing deformations of the rock is already obvious from Terzaghi's observations [21]. The concept of a yielding support using for example 'a compressible layer between the lining and the rock' was proposed in 1972 [20]; the first constructive implementation of the concept, however, was made by Lombardi [25]. In 1978, Kovari and Amstad realized a yielding support for the T8 tunnel (Switzerland), using low strength support pads of lightweight concrete having heights up to 300 mm [19]. These pads with square cross sections were installed at intervals of 2.0 m (Figure 38), and the foam plates between
Figure 38 Construction of low strength support pads in tunnel T8
596
Back Analysis Monitoring
them served mainly as rock-side formwork for pouring of the concrete. The pads have two principal functions. Firstly, they serve as foundation for the invert arch, even if there is no swelling of the underlying rock. The additive (Leca) used in the lightweight concrete is resistant against chemical decay, so that rotting of the pads does not need to be considered. The second role of the pads is to allow for heaving of the rock under the tunnel floor to a certain, predetermined extent, while keeping the swelling load on the invert arckbelow a given design value. The design of the pads was based on an estimate of the floor heave that would take place without any constraint. In Figure 39 a comparison of the measured swelling heaves in the various measuring sections is shown. In this semilogarithmic representation the time variation of the movements during the period of observation can be approximated very well by straight lines. The extrapolation, therefore, to a further 100 years seems to be reasonable, provided that the conditions of the construction remain the same. The authors are aware of the great uncertainties involved in such an extrapolation. However, it is felt that the displacements obtained in this way are on the safe side. From these field measurements and considerations it is concluded that the heave of the flat bottom of the tunnel without the restraint of the invert arch could reach up to 20 cm depending on the height of overburden and mineralogical composition. A second example of the implementation of a yielding support between the invert arch and the underlying rock is the Freudenstein Tunnel of the German Federal Railways. This 6.8 km long railway tunnel is part of the new high speed section between Mannheim and Stuttgart [30, 31]. The tunnel traverses the strongly leached Gipskeuper formation in the eastern section on a length of 2.3 km. This formation consists of alternating layers of water-bearing and variably weakened rock. The following section passes through unfractured, unleached strata containing layers of anhydrite and clay with high swelling potentials. The leached and the unleached strata are separated by the socalled 'gypsum level'. Leaching takes place in this interface at a relatively slow rate. The tunnel intersects the anhydrite and gypsum level over fairly large stretches. The angle between the tunnel axis and the gypsum level is relatively small, and, as a result, the tunnel floor is often situated in zones of rock with a high swelling potential, while the crown is located in leached, almost loose rock. The phreatic surface in the formation is about 60 m over the tunnel. In the following discussion, only the 4.1 km long, western section of the tunnel in the unleached Gipskeuper formation will be considered. One particularity of this tunnel is evident from the description given above: the swelling potential of the surrounding rock would present no problem if water movement from the leached to the unleached areas around the tunnel could be safely prevented over the planned, 100 year lifetime of the project. Even with sophisticated sealing measures, however, the water cannot be fully prevented from seeping along the tunnel, from the wet to the dry areas. The designer decided to call for the statically advantageous circular cross section for those parts of the tunnel lying in swelling rock. The laboratory tests conducted during the design phase of the project showed that the first estimates of the maximal swelling pressure had been too optimistic. As a result, the thickness of the lining and the amount of reinforcement had to be continuously readjusted to the increasing values of
4 û O Q I20hE
a> o σ Q.
oÇ^-^
M6 2 - ^ i IP4
I
10
(days) IOO (years)
T i m e , log t
Figure 39
Comparison of the swelling heave with time extrapolation in the various measuring sections
Decision Making in Tunneling Based on Field Measurements
597
swelling pressure. The result was a design with a thickness of the lining of 1.70 m in the floor and 1.00 m in the crown (Figure 40). In view of laboratory tests giving values of 30% for the maximal swelling strain, and over 6 Nmm" 2 for the swelling pressures, even the strongest circular section could not be accepted as being sufficiently resistant. A greater worry yet was the expected heave of the entire tunnel lining, which would make the usefulness of the structure itself questionable, even if the lining could resist the swelling pressure. The proposal worked out [32] with a yielding zone between the tunnel floor and the underlying rock presented a viable alternative. Not only could the maximal expected swelling pressure (and consequently the values of uplift) in the tunnel be significantly reduced, but a flattened tunnel floor design could also be used instead of the circular cross section (the former presenting great technical advantages in execution over the latter). The new project was then fully worked out and opened to call for tenders. The profile and the dimensions of the structure are based on the following assumptions. (i) The rock is homogeneous in the floor area and behaves according to the 'swelling rule' in Figure 32, Its swelling parameters are: ε = 20% (swelling strain) and σ* = 6 N mm - 2 (swelling pressure). (ii) The shape factorfe,based on the measured strain distribution in the rock (Figure 33), is taken to be 1/3 (parabolic distribution). (iii) The yielding support zone should on the one hand provide a sufficient capacity for the pouring of the concrete for the invert arch and on the other hand have a compressibility of up to 30% under a swelling pressure of 0.5 N mm" 2 . (iv) B 35 concrete and BSt 500 reinforcement were used for the tunnel lining. The calculations done according to these assumptions lead to the profile in Figure 40. Notable differences in the dimensions (e.g. curvature and thickness of the concrete lining) can be seen upon comparison with the original circular profile (without yielding supports). It was intended to install the support pads (Figure 44) of lightweight concrete to provide an extra support for the lining that would compress as well in cases of greater swelling and in order to limit the forces transferred to the vault. Finally, the support pads had to be made of B 25 concrete because of the doubts concerning the longtime stability of the pads surrounded by the extreme aggressivity of the groundwater. Upon request of the owner, a test gallery was set up near the west entrance of the tunnel for further in situ tests; these tests are expected to answer some of the questions still remaining on the swelling of sulfatic rocks. In four sectors of the test gallery, each floor plate (Figure 41) was pretensioned with a constant load P of 0.1,0.25,0.5 and 0.75 MN m" 2 . The floor heave Ua, three years after watering the floor area of the gallery, is presented in Figure 42. These values are not the equilibrium results, because the swelling process still continues nearly linear with time. The strain distribution of the swelling rock under the pretensioned plate (P = 0.1 MN m" 2 ) is shown in Figure 43. Three years after watering the floor area, the swelling process concentrates in the rock layer near the floor plate. These first impressions of the in situ behavior of the swelling rock confirm the laboratory results. They show the extreme swelling capacity of the Gipskeuper rock formation in the Freudenstein Tunnel.
With
—
172 m 2 38m
3
1
m"
Yielding support
-
Without
Excavation
207m2
(+20%)
Interior lining
58m3
m"'(+50%)
Yielding support
Figure 40 Freudenstein Tunnel: tunnel sections with and without yielding support in the Gipskeuper rock formation with high swelling potential (swelling strain > 20%, swelling pressure > 6 N mm -2 )
Back Analysis Monitoring
598
77~ Extensometer
f
Sliding micrometer
Figure 41 Example of a cross section in the test gallery with an anchored floor plate and the layout for field observations (Freudenstein Tunnel)
i 1
100
$
80
-\
60
-
>
Ό Φ JO
e—3 years \ after watering
40
O
o 20 H
II
II
0.25
0.5
1 111*1 1 0.75
log P(MN m"2)
Figure 42 Measured log-normal characteristic line for the Gipskeuper rock formation, three years after watering (results from the test gallery in the Freudenstein Tunnel)
Ir 1987
h 1990
H
Ï -60
-L -40
_L -20
Watering from boreholes
Im Im
0
Extension,« (mm m"')
Figure 43 Distribution of the swelling strain in the Gipskeuper rock formation below thefloorplate. Pretension of the invert plate with a constant load of 0.1 MN m" 2 , three years after watering (Freudenstein Tunnel)
Decision Making in Tunneling Based on Field Measurements
599
The following conclusions can be drawn concerning the design of tunnels in swelling rock, based on field measurements. (i) The swelling process in rock formations can be monitored exactly by means of continuous strain measurements along selected boreholes using the sliding micrometer. These measurements give in situ information on the swelling potential of individual layers, on the decrease of the swelling strain with increasing distance from the opening and on the time-dependent development of the swelling. (ii) A characteristic line for the 'homogeneous' swelling rock can be determined, based on two simple and verifiable assumptions. Firstly, the calculation is based on a 'swelling rule' that is represented for rocks containing anhydrite as well as clay rocks by a straight line in a log-normal graph. The second assumption concerns the distribution of swelling strains in the floor area of a cavity. The validity of the simplifications in the determination of the characteristic lines can be checked through direct field observation. The initial state of stress in the surrounding rock and the
(b)
Figure 44 Construction of the yielding support in the Freudenstein Maintunnel
600
Back Analysis Monitoring
one resulting from excavation thus do not need to be considered, because their effects are included in the assumed strain distribution. (iii) The implementation of yielding supports seems to be the safest and the most economical solution when the swelling potential of the rock through which the tunnel is built exceeds a certain limiting value. For projects in which the extent offloorheave is limited by the use of the opening (e.g. in the construction of tunnels for traffic) this yielding support zone is placed between the invert arch and the underlying rock surface. In rock formations with high swelling potentials (e.g. Gipskeuper) the yielding support system (Figure 44) represents a constructive solution that ensures a long-term use of the facility, in that it allows a certain amount of controlled heave while providing structural resistance. 20.5.4 Decision Making in Subway Tunneling Two examples from the subway in Munich are presented here to show the importance and usefulness offieldmeasurements for decision making in subway tunneling. The planned network for the Munich subway system has a total length of about 100 km with 106 stations. Since 1974 the tunneling with shotcrete has been of growing importance, resulting in a stretch of 21 km using this method [33]. Economy and safety are being given great attention and therefore field measurements always accompany the construction procedure. Figure 45 represents a typical geological section showing the two major formations, i.e. the quaternary deposit consisting of gravel, sand and the tertiary marl, frequently referred to as 'Flinzmergel', below it. The latter has a varying appearance consisting of stiff or even hard clays, clayey silts, marl, marlstones and fine to medium grained sand. The groundwater in the quaternary formation is as a rule not connected with the water in the tertiary ones. There the pore water pressure can also be very different in adjacent sand lenses sometimes showing an artesian character. The clays and marls are nearly impermeable, offering a reliable protection against the water in the quaternary formation providing the thickness of the marl layer above the tunnel roof is not less than 2 to 3 m. In the cases discussed below this condition was always fulfilled. The method of excavation for a single track tunnel is the head and bench method (Figure 46a). Emphasis is placed on shotcreting the invert very close to the head (2 to 4 m) and in a short time span of 1 to 2 days only. In this way a statically favorable action against ground deformations and surface settlement is produced immediately. The same principle is applied to the double track cross section (Figures 46b and 46c). Here, the first half of the tunnel is excavated and supported as a single track tunnel. The enlargement to the full cross section follows in a distance of approximately 15 m and again in head and bench operation. If water-bearing sand layers are encountered special measures must be taken. They may involve decreasing of the piezometric head by drainage wells and also application of compressed air as an additional measure. The tertiary sands are generally rather compact so that they are stable at the face provided that no excessive water pressure prevails. If compressed air is applied, the whole section is constructed using shotcrete as temporary support. After the completion of the section, atmospheric conditions are restored. The shotcrete lining resists the outside water pressure until the final reinforced concrete lining is constructed. This
Figure 45
Stratification of subsoil indicating hydrological conditions
Decision Making in Tunneling Based on Field Measurements (a)
601
(b)
(c)
Figure 46 Method of excavation with shotcrete support: (a) single track tunnel, sectional area 38 m2, (b) twin track tunnel, sectional area 80 m2, and (c) photograph of twin track tunnel
procedure has proved to be very successful, being safe and having a reducing influence upon ground deformations. 20.5.4.1 Strain profiles in the subsoil due to changes in pore water pressure In the case being discussed here, the groundwater in the tertiary sand formation was dewatered by conventional wells while the groundwater in the overlying quaternary gravel was maintained at its initial level. Decrease of pore water pressure in soils increases the effective normal stress [21] which in turn leads to compression of the material. To optimize the dewatering measures and to control the differential settlements in the different layers of the ground, strain profiles where measured with the sliding micrometer in different sections of the subway line 5/9 [34]. In Figure 47 the measured compression strains along two boreholes having a depth of 38 m are shown. The corresponding borehole logs show the start of the tertiary formation approximately at 8 m depth in both cases, whereas the stratification is different.
602
Back Analysis Monitoring Compression, €. (mm m"') strain (χΙΟ' 3 )
Compression, £
(mm m"')
Geological log ♦
0.2 1
>»
Ground surface
0.4 1 '
• · o. .-. ϋο.
l_ O
c a>
■1-
σ
o3
4 0
É
o° o
g& •'•:*j·
8
··.*!· ^-*>^±
Ξ^
:ο ·ν
)
α\-'
ËÏ *-; 7
**«& ΐ££ *fî??T
sa^
·_» "~ Ξ Τ • ·*"
JΊ mm w\ 1
12
(
Έ
1111. A/7,=0
1
4-M··.
>» o
>* £*
16
20
O
o ; c a> °"Π·· o ■ & . ·. '
_
3
O
^ \
1 1 m\ 1
JA,
·-·
III
—_
12
«_» _
16
>» ^ ·-· r* k_
σ
■1-
É
1
■MMJ 1
W2
».0^?. ·«·
—_τ-
ρΔ/? 2
1 llll
)
4
Δ/7,
?4
y^-:J J ,~ ££
20
O
vw •.'..'^r··
24
·ΐ^ϊ· 2 8 >— J
■M \
.JT_
·—·—
28
:;ij. ■*T"
32
36
T i "
\
\wÊ'\
32
:~v
36
f^3 — Mea suring line I
-Measuring line 2
Figure 47 Strain distribution ε along two vertical measuring lines caused by partial dewatering of tertiary formation ( W = observation well)
In the tertiary formation, conventional and vacuum wells were operated and their effects observed by open standpipe water level recorders. In Figure 47 the changes in piezometric heads are indicated by Ahi for the different observation wells designated by W{. When interpreting the measured strain distribution one has to bear in mind that apart from the details of the geology of that area also the efficiency of the pumping operation is decisive. The accumulated strains along measuring line 1 resulted in a surface settlement of about 3 mm and those of the measuring line 2 yielded 4 mm. Such surface settlements occur before the tunnel construction. Therefore, one has to instrument the boreholes to take readings well in advance. 20.5.4.2 Interaction between adjacent tunnels and the effect of compressed air The interaction of adjacent tunnels and its effect on settlement is influenced by various factors such as the shape, span and depth of the tunnels and also by the distance between them, the method of excavation, the rate of advance, the characteristics of the subsoil and finally the groundwater conditions. Obviously the prediction of ground settlements by computational methods has major shortcomings in such complex situations. If only a limited stretch of a subway line is subjected to severe restrictions on permissible settlements, different constructional measures can be tested before the critical area is reached by the tunnels. This was the case in Munich when undertunneling old houses with low overburden near to Odeonsplatz'. The tunnel section between the starting shaft and the critical area was approximately 350 m, offering a unique possibility for trial sections and an accompanying monitoring program. Along the trial stretch there were no buildings, services or major roads and therefore no severe limitations on permitted settlements. Two basically different constructional measures were tested with respect to their capability to reduce deformation. The first measure consisted of applying compressed air to control pore water pressure in the ground. The second proposal involved the excavation of the two track tunnel in five different stages (Figure 48c) instead of the commonly applied four stages (Figure 48b).
Decision Making in Tunneling Based on Field Measurements Compression strain, € (mm m"1)
Compression strain,^(mm m" ) (a)
0
0.4
0
0.4 0 8
1.2
603
0
0.4
0
0.4 0.8
1.2
1.6
Quaternary Tertiary
Quaternary Tertiary
0.4 —i
0.8 r~
0 0.4 0.8 j — i 1
1.2 1
1.6 r-
Quaternary Tertiary
Excavation tunnel I
Excavation tunnel Π
Figure 48 Strain distribution ε caused by the excavation of tunnels I and II: (a) measuring section MQ 21, atmospheric conditions, (b) measuring section MQ 20, compressed air, and (c) measuring section MQ 19, compressed air
In order to assess the most effective method of construction and to establish the distribution and intensity of the ground settlements prior to the arrival of the tunnels at the critical area, three measuring sections were installed. From the results of the comprehensive measuring program only the strain profiles measured with the sliding micrometer will be discussed here. The diagrams shown in Figure 48 reveal interesting details of the ground deformations caused by the construction of two parallel tunnels (I and II) using different excavation procedures under atmospheric and
604
Back Analysis Monitoring
compressed air conditions. The distance between the three measuring sections was great enough to exclude interference but also small enough for the assumption of uniform ground conditions. The first section of the tunnel starting from the shaft was excavated in four steps under atmospheric conditions (Figure 48a). Next, provisions were made for compressed air application using the same four stage excavation procedure (Figure 48b). In the following section compressed air application was maintained but the method of excavation was made in five stages in the cross section (Figure 48c). In this way the benefits resulting from a more sophisticated method of excavation and from compressed air application could clearly be assessed. From Figure 48 it can be concluded that using compressed air results in markedly smaller ground deformations when compared with atmospheric conditions. On the other hand, no reduction in ground deformations can be observed due to the more sophisticated excavation method shown in Figure 48(c). Based on the unambiguous results from the trial construction sections a sound decision could be made regarding the method of construction to be applied when undertunneling the critical city area. In fact, the compressed air application (Figure 48b) was most successful throughout the whole construction section. Two additional phenomena observed during the measuring campaign deserve to be mentioned. These are the 'pillar effect', i.e. the compression of the ground between the two tunnels, and the change of the pore water conditions due to the drop of compressed air pressure to atmospheric air pressure during 24 hours. The 'pillar effect' can be seen clearly from all three cases (Figures 48a, b and c), whereas the effect of a drop of air overpressure from p = 0.8 bar to p = 0 bar can be seen in Figures 48(b) and 48(c). 20.6 SUMMARY AND CONCLUSIONS The successful design of underground openings is based on different sources of information. The most important among them are geological explorations, soil and rock mechanics investigations, statical computations and field measurements during construction. The way to make use of computer programs and the criteria for the interpretation of the results obtained are still the subject of discussion. This is the main reason for the lack of standard design procedures in tunneling. The inherent weak elements in purely theoretical considerations can, however, be compensated for by direct field observations and the sound engineering experience of the designer. Depending on the design problem, it may be necessary to make decisions well before the start of the construction. In this case, the observation of the actual deformations of the tunnel profile, the movements of the surrounding ground or the settlements at the ground surface during the excavation of the tunnel mainly have the function of checking the structural behavior with regard to satisfactory design and proper execution of the works. In contrast, using the shotcreting method with anchors or steel grid support, which may in many cases also be applied in subway construction, continuous measurements inside the tunnel and in the subsoil can serve as feedback signals for the constructional process. On the basis of careful statical computations a concept is worked out for the excavation sequences both in the cross section and along the axis with the corresponding support measures. If the measurements indicate a substantial deviation from the anticipated behavior of the structure, the most important corrective measures in the construction can still be applied. The basic idea offieldmeasurements lies in the optimization of the design and construction of the underground structures. In other words, the aim is to obtain adequate safety for a minimum of cost expenditure, whereby the manifold influence of the construction time is also included in the expenditure. This does not, however, exclude the conscious decision to accept a calculated risk. Since the problem of optimization is very varied, the immediate objectives of the individual measurements may be concerned with quite different aspects, the most important of which are as follows. (i) The investigation of the global material properties of the rock. (ii) The determination of the type and quantity of rock pressure (loosening pressure, genuine rock pressure and swelling pressure). (iii) The safety control of the structure. (iv) The verification of structural response to a specific method of construction. (v) The control of the effectiveness of particular support measures. (vi) The comparison of theoretical predictions with the actual structural behavior. As a general rule the above classification of the objectives of measurement is not rigid. It is intended to indicate the main emphases. It should be noted that usually the same program of measurements has several aims. The most important thing is that the concept, the execution and interpretation of the measurements are adjusted to suit the needs of the problem in hand.
Decision Making in Tunneling Based on Field Measurements
605
Field measurements are now recognized worldwide as an indispensable aid for correct decision making in tunneling. They often form the link between theory and the engineering practice. Successful measurements require both a thorough understanding of the specific problems arising in tunneling and a close familiarity with instrument techniques. In this chapter an attempt was made to show, on the one hand, the significance of monitoring by the discussion of some case histories and, on the other hand, to give information on new developments in measuring techniques.
20.7 REFERENCES 1. Kovari K. Basic considerations on the design of underground openings. Period 3 Int. Assoc. Bridge and Structural Eng. (1979). 2. Golta A. Schwellvorgänge im Planum schweizerischer Bahntunnel. Rock Mech. suppl. 5, 231-243 (1976). 3. Huder J. and Amberg G. Quellung in Mergel, Opalinuston und Anhydrit Schweiz. Bauztg. 43, 975-980 (1970). 4. Grob H., Kovari K. and Amstad C. Sources of error in the determination of in situ stresses. Tectonophysics 29, 29-39 (1975). 5. Kovari K. The elasto-plastic analysis in the design practice of underground openings. In Finite Elements in Geomechanics (Edited by G. Gudehus), Chapter 12. Wiley, London (1977). 6. Proc. I migrantes Seminar, Re vista Construçao Pesada No. 1481, Sao Paulo (1976). 7. Fiasco J. Problemas Diversos na Implantacao de Tuneis (Imigrantes Seminar), Revista Construçao Pesada No. 65, Sao Paulo (1976). 8. Carvalho O. S. and Kovari K. Displacement measurements as a means for safe and economical tunnel design. In Proc. Int. Symp. Field Measurements in Rock Mechanics, Zurich, pp. 709-721. Balkema, Rotterdam (1977). 9. Kovari K., Amstad C. and Fritz P. Integrated measuring technique for rock pressure determination. In Proc. Int. Symp. Field Measurements in Rock Mechanics, Zurich, pp. 289-316. Balkema, Rotterdam (1977). 10. Matt P., Thurnherr F. and Uherkovich I. Vorgespannte Durckstollen. Schweiz. Bauztg. 96, 63-72 (1978). 11. Kovari K., Hagedorn H. and Fritz P. Parametric studies as a design aid in tunneling. In Proc. 2nd Int. Conf. Numerical Methods in Geomechanics, Blacksburg, VA (Edited by C. S. Desai). ASCE, New York (1976). 12. Kovari K., Amstad C. and Grob H. Messung von Verschiebungen und Deformationen an Bauwerken mit dem Distometer-ISETH. Schweiz. Bauztg. 36, 819-825 (1974). 13. Kovari K. and Koppel J. Head distribution monitoring with the sliding piezometer system 'Piezodex'. In Proc. 2nd Int. Symp. Field Measurements in Geomechanics, Japan (1987). 14. Kovari K. and Amstad C. Fundamentals of deformation measurements. In Proc. Int. Symp. Field Measurements in Geomechanics, Zurich, pp. 219-239. (1983). 15. Kovari K., Amstad C. and Koppel J. New developments in the instrumentation of underground openings. In Proc. 4th Rapid Exavation and Tunneling Conference, Atlanta, GA, pp. 817-837. (1979). 16. Koppel J., Amstad C. and Kovari K. The measurement of displacement vectors with the 'TRIVEC Borehole Probe. In Proc. Int. Symp. Field Measurements in Geomechanics, Zurich (1983). 17. Amstad C, Koppel J. and Kovari K. Trivec-measurements in geotechnical engineering. In Proc. 2nd Int. Symp. Field Measurements in Geomechanics, Japan, pp. 17-32. (1987). 18. John M. and Wogrin J. Geotechnische Auswertung des Richtstollens für den Vollausbruch am Beispiel Pfändertunnel. Rock Mech. suppl. 8, 173-194 (1979). 19. Kovari K. and Amstad C. Field instrumentation in tunneling as a practical design aid. In Proc. 4th Int. Congr. Rock Mech., Montreux, vol. 2, pp. 311-318. Balkema, Rotterdam (1979). 20. Grob H. Schwelldruck im Belchentunnel. Ber. Int. Symp. für Untertagbau, Luzern, pp. 99-119. (1972). 21. Terzaghi K. Introduction to tunnel geology. In Rock Tunneling with Steel Support (Edited by R. Proctor and T. White), Youngstown Printing Co., OH (1968). 22. Kovari K., Madsen F. T. and Amstad C. Tunneling with yielding supports in swelling rocks. In Proc. Int. Symp. Weak Rock, Tokyo, pp. 1019-1026. (1981). 23. Einfalt H., Ergebnisse der Untersuchungen an den Calciumsulfatmineralien aus dem Gipskeuper Stuttgarts. Forschungsberichte, Strassenbau und Strassenverkehrstechnik 184, 95-108 (1975). 24. Henke K. F., Kaiser W. and Nagel D. Geomechanische Untersuchungen im Gipskeuper. Forschungsberichte, Strassenbau und Strassenverkehrstechnik 184, 149-184 (1975). 25. Lombardi G. Rock mechanics at the CERN proton-antiproton facilities. In Proc. 4th Int. Congr. Rock Mech., Montreux, vol. 3, pp. 433-436. Balkema, Rotterdam (1979). 26. Lombardi G. Underground openings in swelling rock. In Proc. 1st Nat. Conf. Case Histories in Geotechnical Engineering, Lahore (1984). 27. Wittke W. and Rissler P. Bemessung der Auskleidung von Hohlräumen in quellendem Gebirge nach den Finite Element Methode. Veröff. der RWTH Aachen 2, 7-46 (1976). 28. Einstein H. H., Bischoff N. and Hofmann E. Verhalten von Stollensohlen in quellendem Mergel. Ber. Int. Symp. für Untertagbau, Luzern, pp. 296-319. (1972). 29. Henke K. F., Kaiser W. and Beiche H. Verhalten von Tunnelbauwerken in quellfähigen Schichten des Gipskeupers. Ber. 2. Nat. Tagung Ing. Geolog. Fellbach, BRD, pp. 135-142. (1979). 30. Prommersberger G., Bokemeyer R. Erkundungsstollen Freudenstein-Tunnel. Sonderausgabe 1. Int. Tunnelbau Symp. Bauma, München (1986). 31. Kuhnhenn K., Prommersberger G. Der Freudensteintunnel, Tunnelbau in schwellfähigem Gebirge, Forschung und Praxis, Nr. 33, STUVA (1989). 32. Kovari K., Amstad C. and Anagnostou G. Design and construction methods - Tunneling in swelling rock. In Proc. 29th U.S. Symp. Rock Mech., Minneapolis, MN (Edited by P. A. Cundall, R. L. Sterling and A. M. Starfield), pp. 17-32. Balkema, Rotterdam (1988).
606 33. 34. 35. 36. 37. 38.
Back Analysis Monitoring Hochmuth W., Krischke A. and Weber J. Subway construction in Munich. Developments in tunneling with shotcrete support, Rock Mech. Rock Eng. 20, 1-38 (1987). Amstad C. and Kovari K. Strain monitoring in the subsoil of the Munich subway. In Proc. 2nd Conf. Mass Transportation in Asia, Singapore, pp. 255-271. (1984). Kovari K., Amstad C. and Grob H. Displacement measurements of high accuracy in underground openings. In Proc. 3rd Congr. Int. Soc. Rock Mech., Denver. NAS, Washington DC (1974). Amstad C. and Koeppel J. A multihead borehole rod-extensometer design. In Proc. Int. Symp. Field Measurements in Rock Mechanics, Zurich, pp. 429-436. Balkema, Rotterdam (1977). Terzaghi K. Erdbaumechanik auf Bodenphysikalischer Grundlage. Franz Deuticke, Leipzig (1925). Anagnostou G. Untersuchungen zur Statik der Tunnel in quellfähigem Gebirge. Dissertation 9553, Swiss Federal Institute of Technology (1992).
21 Deformation Monitoring for Stability Assessment of Underground Openings PETER K. KAISER Laurentian University, Sudbury, Ontario, Canada 21.1
INTRODUCTION
607
21.2 MONITORING FOR STABILITY ASSESSMENT 21.2.1 Why Monitor for Stability Assessment? 21.2.2 Ingredients of a Successful Monitoring Program 21.2.3 Types of Monitoring 21.2.4 Concept of Safety for Underground Openings 21.2.4.1 Safety margin assessment by deformation monitoring
608 608 610 610 610 611
21.3 ROCK MASS FAILURE MECHANISMS 21.3.1 Failure Initiation, Propagation and Collapse 21.3.2 Failure Modes Dominated by Weaknesses in Rock 21.3.3 Concept of Monitoring for Failure Detection
614 614 615 617
21.4 DEFORMATION MONITORING 21.4.1 Introduction 21.4.2 Visual or Qualitative Monitoring 21.4.3 Use of Deformation Magnitudes 21.4.3.1 Monitoring of global versus local response 21.4.3.2 Placement of instruments to observe local failures 21.4.3.3 Convergence to assess support performance 21.4.4 Use of Deformation Rates 21.4.4.1 Extent of yield zone from deformation rates 21.4.4.2 Support effectiveness from deformation rates 21.4.4.3 Assessment of mobilized safety margin from deformation rates 21.4.5 Miscellaneous Considerations for Displacement Monitoring 21.4.5.1 Required accuracy 21.4.5.2 Required duration and frequency of readings 21.4.5.3 Displacement monitoring layout 21.4.5.4 Combination of stress change and displacement measurements 21.4.5.5 Monitoring data interpretation
618 618 619 619 619 621 621 622 623 623 623 624 624 624 624 626 627
21.5
CONCLUSIONS
627
21.6
REFERENCES
628
21.1 INTRODUCTION Underground construction methods are rapidly changing and new technologies allow us to excavate larger openings at much greater depths. In association with advances in construction technology, there are increasing demands for more accurate predictions and assessments of ground behavior by the ground control or geotechnical engineer. Monitoring has become a fundamental requirement for assessing the stability of many underground openings and for quantifying the risk of unacceptable rock response. Monitoring consists of obtaining field measurements and observations over time for a number of functions. 607
608
Back Analysis Monitoring
(i) To assess the safety (stability) of an opening. (ii) To confirm the assumed or expected rock mass behavior. (iii) To improve understanding of fundamental rock mass behavior and failure processes. (iv) To obtain data for design and analysis. (v) To provide data for numerical model verification. (vi) To confirm excavation performance predictions. (vii) To allow extrapolation for the prediction of long-term rock response. (viii) To provide quality control data. (ix) To supply data to assist in modifying and improving excavation design and construction procedures, including remedial measures. (x) To evaluate effects of remedial work or changes in construction procedures. The underground excavation environment is often very complex, demands large up-front investments and contains many uncertainties. Whitman [1] stated in his Terzaghi Lecture that 'all risk cannot be eliminated nor calculated with sufficient accuracy'. Fortunately, field monitoring provides a means to reduce risk to a desirable or acceptable level. However, risk reduction through stability assessment can only be achieved if we gain a better understanding of the rock mass response. Observations of the actual ground behavior, qualitative (visual) or quantitative (measured) in nature, are prerequisites for a better understanding. Information collected during construction permits more-rational decisions and appropriate actions with respect to rock support, excavation sequencing, rate of excavation, etc. Monitoring can detect imminent failures and, by giving advanced warning, allows time to implement remedial measures or changes in the excavation and support procedures. This Observational design approach' which incorporates field observations was promoted by Peck [2] and has proven effective for minimizing risk. Today, risk can be reduced by a careful monitoring program without creating excessive extra costs or much disruption to the construction schedule. Today, a carefully designed and implemented monitoring program should be a component of any construction project. However, a monitoring program is only functional if it is well planned and managed, the results are analyzed to provide feedback, the knowledge gained is applied and the resulting benefits are verified. A successful observational design project requires good coordination, communication and cooperation between all parties involved in the construction process. Furthermore, an effective monitoring program has to be designed with a well-defined purpose and clear objectives such that the types and locations of instruments, the timing of installation and the frequency of readings can be rationalized. The flow chart in Figure 1 schematically illustrates how field measurements fit into the underground excavation process consisting of collecting, processing and using information to produce a safe underground opening. Adding information is the intent of monitoring. In geotechnical engineering, information is commonly collected during the site investigation phase and then processed for application to a specific project. However, it is seldom possible to collect sufficient information from an economical site investigation program because of the inherent geological complexities of a rock mass. Obviously, more information can be obtained if the collection process is extended by continuously updating the database as construction proceeds. In situations where the engineer believes that insufficient information was collected before construction, the observational design approach is not only desirable but essential to provide feedback for the decision-making process. The scope of this chapter is to provide guidance for a rational application of deformation monitoring in underground construction.
21.2 MONITORING FOR STABILITY ASSESSMENT 21.2.1 Why Monitor for Stability Assessment? The goal of engineering is to design and construct reliable structures. The fact that a structure in rock has been designed with adequate site investigation, with sufficient performance modeling and measures taken to circumvent failures, does not mean that the risk of failure has been completely eliminated. A very important purpose of any monitoring program is to ensure that rare, undesirable events or conditions are not encountered. However, if a failure is detected, monitoring must provide sufficient warning and information for remedial measures to be designed.
Deformation Monitoring for Stability Assessment of Underground Openings
609
Underground excavation Design/construction process Preliminary investigations (feasibility..) Site investigation phase (rook mass properties, in situ stress field...)
Pilot excavation monitoring
Design methods
H AorBl Selection (or modification) of: (i) Excavation technique and sequence (ii) Support type and dimensions
V)
C
0)
Direct path
Indirect path
ichS-i , , \.
,
|Design criteria | Analytical or numerical simulation
E
Final design
I
Design model
Mining of underground opening
Empirical design by classification
X
Performance prediction model
I
IBehavior prediction! »
Field measurements
Optimization
Safety evaluation
Future benefits
Feedback Or
I Feedback
Figure 1 Monitoring and underground excavation
If viewed as a preventive measure, monitoring constitutes one of the most important components of risk reduction and should be guided by the principle: prevent rare, undesirable events or identify and help to alleviate them. A sound engineering design without monitoring cannot provide an economic defence against rare events the laws of occurrence of which are not well understood. Hence, monitoring must attempt to detect unexpected ground behavior and provide insight to allow diversion and eventual control of failure processes. Since rare events by their nature are not likely to be encountered, it must be realized that a monitoring system normally confirms adequate performance of an excavation. In order to detect unlikely failures, the lowest level of monitoring must provide wide coverage and must be inexpensive. The difficulty in predicting rare failure events should never be viewed as a deterrent to monitoring. Instead, monitoring must be viewed as a cost effective line of defence. Monitoring is needed to detect adequate as well as inadequate conditions, for the refinement of a design and for long-term extrapolations to ensure maintenance-free and safe performance. Monitoring always delivers more and new information and reduces risk by improving the database on which engineering decisions are made. In rock engineering, field observations provide the best source of information because no simplifying assumptions are required, size effects are not neglected and unknown geological complexities are automatically considered. Any decision-making process depends largely on the amount and quality of the available information. Monitoring is the process of gathering sufficient information for engineering design and decision making as well as to provide insight into what cannot be otherwise seen. Furthermore, monitoring constitutes a source of information for gaining experience, by educating or training staff to understand the dominant rock behavior and failure modes. Experienced staff will make fewer mistakes and thereby help to reduce risk. The cost of monitoring and the related loss of production are frequently raised as primary arguments against monitoring. Unfortunately, it is difficult to measure the value of a monitoring program because economic, social, human and technical considerations must be balanced. The benefits of a carefully planned monitoring program have been demonstrated by many, including
610
Back Analysis Monitoring
Lane [3] for tunneling. Even though monitoring may be expensive, it constitutes a cost effective approach to rock engineering because each underground excavation is used as a laboratory to verify design and numerical performance models. Without monitoring many challenging and innovative projects would never have been attempted or would not have been completed successfully. A welljustified and properly managed monitoring program must have a clear purpose, must be well designed and implemented and its findings must be used. 21.2.2 Ingredients of a Successful Monitoring Program Unsuccessful monitoring programs often lack one or several of the following key components. (i) The monitoring program has no clear objective. (ii) The instrumentation is poorly designed or installed. (iii) The instrumentation is installed in the wrong location or too late. (iv) The frequency or duration of data recording is inadequate. (v) The collected data are not or cannot be properly analyzed. (vi) Useful results are not used to modify the construction process. For a well-designed monitoring program, these deficiencies must be eliminated and the following additional ingredients required for a successful application of the observational design approach by performance monitoring must be considered. (i) A high degree of uncertainty or major difficulties in terms of economic or safety considerations must be expected, i.e. ground control problems must be envisaged and little experience for their resolution should be available. (ii) Access for monitoring must be possible with minimal interference with the construction or production process. (iii) Hiding of ground control problems and related, undesirable consequences (e.g. by placing the blame elsewhere) must not be possible. (iv) Cooperative, open-minded technical and managerial staff from all contributing parties must be involved, eager to learn and willing to react. (v) Immediate feedback must be provided for successful implementation of findings into an evolving plan of remedial measures. Severe deficiencies in any of these aspects will reduce the value of observations and likely render the observational approach useless. 21.2.3 Types of Monitoring Monitoring as an ongoing surveillance strategy for changes in ground and excavation behavior can be made by qualitative or quantitative means, such as visual inspection, instrumentation for direct indicators (deformations, stresses, stress changes, etc.) and instrumentation for indirect indicators (acoustic emissions by microseismic monitoring, etc.). A discussion of instrument design and selection is beyond the scope of this chapter but the choice of the monitoring method, instrument type, array layout, etc. depends on the excavation geometry, rock mass characteristics and monitoring purpose. However, several outstanding texts contain discussions on instrument design and selection, namely Hanna [4], Dunnicliff [5] and the Mine Monitoring Manual [6]. This chapter deals primarily with one direct indicator, i.e. deformation monitoring, because ground and support deformations are frequently measured and provide much insight into the ground response and excavation performance. Deformation records often contain substantial diagnostic information but the interpretation process is complex and deserves special attention. A wide variety of instrumentation is available for deformation monitoring, including convergencerecording devices, extensometers and deflectometers. Instrumentation requirements depend on the purpose of a monitoring program. 21.2.4 Concept of Safety for Underground Openings The level of safety of an underground opening may be described by the safety margin, defined as the difference between the support capacity C (provided by a combination of artificial support and rock mass strength) and the demand D due to gravity loads, in situ stresses or mining-induced stress
Deformation Monitoring for Stability Assessment of Underground Openings
611
changes. A safe opening performs adequately, providing a low probability of failure with an adequate safety margin. Underground openings are truly interactive structures. Their stability depends on the amount of deformation that is allowed before a new equilibrium is established after excavation. During the rock removal process, the natural ground support is gradually removed and replaced by an artificial support after some initial deformation us (Figure 2) has occurred. The safety margin is clearly related in some manner to the accumulated deformation before and after support interaction. At the equilibrium point, the mean demand must be less than the mean capacity but the distributions of demand and capacity may overlap and, hence, a finite probability of failure exists. Monitoring for stability evaluations should assess whether the safety margin is acceptable. In an attempt to solve statistically the interactive problem illustrated by Figure 2, Matsuo and Kawamura [7] have treated the ultimate support pressure as a fictitious support pressure equal to the support capacity (independent of displacement) and have described the ground pressure by the convergence curve as a displacement-dependent demand D(u). They then calculated the probability of failure at the equilibrium point as the probability that the factor of safety FS = C/D(u) be less than unity. Unfortunately, the safety margin concept was not adopted and the displacement prior to installation of the support (us) was neglected. Nevertheless, this approach demonstrates that once the equilibrium point is determined, monitoring should reveal whether there is too much overlap between the capacity and demand distributions. Once an equilibrium has been reached, at zero rate of deformation for time-independent ground, an adequate safety margin must be provided artificially by additional support capacity (AC). The safety margin is then S = (C — D = 0) + AC = AC. The ultimate safety margin is selected by the designer in providing support beyond that required to establish an equilibrium (e.g. by adding a secondary tunnel lining). Whereas the capacity of a supplemental support and, hence, the safety margin can be determined relatively easily, an absolute factor of safety relating total capacity and demand, FS = (C + AC)/D — 1 + (AC/D), cannot be determined because the demand D is unknown at the point of equilibrium. The safety margin is therefore a much more meaningful measure of safety in underground construction than a factor of safety.
21.2.4.1 Safety margin assessment by deformation monitoring The assessment of the safety margin before failure initiation would be relatively straightforward if the load or stress could be measured directly and related to the strength of the support. In an unsupported opening, the stress (or demand) could be compared with an assumed or a predicted rock mass or support strength (capacity), and an increase in stress (before failure initiation) would generally reflect an increase in risk. However, during failure propagation with associated stress transfer processes both increases and decreases in stress can be observed. For example, near a tunnel in yielding ground the tangential stress decreases at the wall, whereas it increases at some distance from the wall. Due to these stress redistribution processes, the safety margin cannot be easily determined by stress or stress change measurements. Ground deformations, reflecting the integrated effect of all stress changes, increase continuously and should provide an appropriate tool for risk or safety assessment. Intuitively, it is expected that the amount of deformation or the deformation rate should reflect the safety margin as it represents a measure of the spread between capacity and demand (Figure 2).
0)
Confinement curve (mean support strength)
CL
o
Q. Û.
Convergence curve (mean ground pressure) u%
Ground displacement, u
Figure 2 Safety assessment concept for an interactive, displacement-dependent case of an underground opening
612
Back Analysis Monitoring
The nature of the ground response (elastic, perfectly plastic or elastic viscoplastic, etc.) will dictate the type of information that can be obtained from deformation measurements. In an attempt to establish some monitoring guidelines, let us first consider the statically determined problem of an initially supported block of rock that is allowed to slide on an elastic, plastic or viscoplastic interface. For an elastic, perfectly plastic interface, the block behaves elastically in the prefailure range when the safety margin is positive (S > 0) and plastically after failure initiation (for S = 0). In the elastic range, the displacements are a function of the mobilized demand or the load increment due to support removal. The rate of deformation during steady state (called work stoppages) is zero at a constant demand and support force. Therefore, deformation rates during excavation or work stoppages in nonyielded, elastic, perfectly plastic (time-independent) materials must theoretically be zero in the prefailure range and very large as soon as yielding is initiated. Deformation rates recorded during excavation stoppages allow separation of the influence of loading and failure. Increasing or high deformation rates during work stoppages are clear indicators that failure is imminent (S = 0). This logic leads to a first set of important guidelines for deformation monitoring for the assessment of the risk of failure or the safety margin. The monitoring guidelines are as follows. Deformation rates rather than magnitudes of deformation should be used to detect failure and deformation rates must be recorded during excavation stoppages. In elastic, perfectly plastic materials and statically determined problems, the magnitude of the safety margin cannot be assessed by deformation monitoring. Deformation rates can only detect when failure is reached. Rapidly increasing or high deformation rates during work stoppages are clear indicators that the safety margin must be zero. Most geotechnical materials are not perfectly plastic and a transition zone moving from zero to high deformation rates can normally be expected. A meaningful, finite rate can be defined in practice to represent the transition from S > 0 to S = 0. A sliding block on an elastic, viscoplastic interface can temporarily sustain loads (demands) in excess of the long-term yield limit and it will deform at a rate related to the amount by which the yield limit is exceeded. If the safety margin is defined relative to the long-term yield limit, as defined above, a negative safety margin (safety margin deficit) may exist and the deformation rate will be proportional to the magnitude of this deficit. For this situation, the deformation rate provides a clear indicator of the amount by which the long-term yield point has been exceeded, i.e. deformation rates are indicators of the magnitude of the safety margin deficit in viscoplastic materials and statically determined problems. The equilibrium condition, for which S = 0, can then be found by interpolation and the observed deformation rate can be used to determine a safe working stress level or to design remedial measures. Additional monitoring guidelines evolve from the following consideration. Deformation rates do not permit the assessment of the magnitude of a positive safety margin but provide information about the safety margin deficit, that is, how far a state is from reaching a new equilibrium (at S = 0). The logic of safety margin assessment (presented at the beginning of Section 21.2.4) and the principle of the last-mentioned monitoring guideline are commonly applied to determine the capacity of ground anchors, where the failure load (at S = 0) is defined as a creep rate of 2 mm per log cycle of time (see ref. [8]). A safety margin is then provided by selectively reducing the demand per support element (negative AD) by providing additional anchors. The factor of safety is, in this case, defined as FS = C/(D — AD), with D = C signifying measured failure load. The same principle may be applied to any other statically determined stability problem where the total demand remains constant (planar, two-dimensional slope failure; gravity-driven wedge failure in tunnel roof; etc.). This rationale is, however, not directly applicable to statically indeterminate problems, such as three-dimensional slope failures, stress-driven wedge failures or yielding underground openings. In these situations, illustrated schematically by the confined sliding block model in Figure 3, the capacity does not remain constant after yield initiation because additional resistances (capacity C2 in Figure 3) can be mobilized, for example, by arching or by activating an artificial support after some initial displacements (us in Figure 3). The capacity is no more a steadily increasing function of the deformation. High, theoretically infinite rates of deformation are observed when S = 0 or when the change in S (dS/du) is zero. Hence, high rates do not necessarily imply that the total ultimate safety margin (51 + 2) is low or decreasing. Furthermore, zero rates after high deformation rate periods may be observed during the mobilization of additional resistances when the ultimate safety margin is further reduced. As demonstrated earlier, magnitudes or rates of deformation do not permit an assessment of the magnitude of a positive safety margin because the capacity is not known and cannot be related to the demand. This can be nicely demonstrated by the example of a deep excavation in clay shale for the
Deformation Monitoring for Stability Assessment of Underground Openings
613
D
Sliding block model
or
D
C
_7
>
h*—
u.
-
c Cz
u
S V
Equivalent mechanical model
i
5lt2
\
////////
\s'
\ U
A
▲ ou dt
C2/2
C2/2
\
Figure 3 Schematic diagram illustrating the relationship between safety margin and deformation rate for elastic, perfectly plastic interface with support (£> = demand, Cx and C 2 == components of total capacity, S = safety margin, u = displacement)
Edmonton Convention Centre. Balanko et al [8] plotted (similar to Figure 4) the volume of excavated rock (representing the magnitude of force F in Figure 3) against the observed excavation wall movement and concluded from the observed linear relationship after excavation of more than 100 000 m3 of rock \ . . that the movements were of an elastic nature and should slow down after excavation completion'. Figure 4, presenting the same data, is essentially a plot of increasing demand and, hence, ultimate safety margin S against deformation. However, the location of the zero point (S = 0) is not known. Whereas it is correct to state that the excavation responded in a linear manner to the excavation process, this plot does not permit an assessment of the magnitude of the actual safety margin or the risk of failure. Even if this relationship is nonlinear (as actually evidenced in Figure 4 by the kink at V = 90 000 m3), this does not necessarily imply that failure is imminent. The safety margin could be far from or close to zero. This graph only indicates the obvious: the safety margin decreases as the excavation proceeds. The proximity of this excavation to failure could have been assessed if sufficient deformation rate measurements during excavation stoppages had been taken and if these rates were used to determine a potential safety margin deficit, for example by comparing observed rates with empirically established critical rates. However, because of the statically indeterminate nature of this large, anchored excavation with two lateral abutments, even increasing rates or rates in excess of some established limits would not necessarily have reflected a diminishing ultimate safety margin (as explained earlier, Figure 3). Even though the monitoring program at the Edmonton Convention Centre was inadequate to establish a safety margin, it nicely demonstrates that field observations are of great value even if the safety margin cannot be determined. It provided the designers with confidence as the deformations were clearly related to the excavation process and did not show any signs of progressive failure or instability. Underground openings are statically indeterminate and deform under the in situ stress field in a truly interactive manner, as illustrated schematically by Figure 3. The deformations are strongly affected by the excavation rate (volume of rock removed), the activated support pressure and the nature of the rock mass. The mobilized capacity of the supporting ring of ground gradually increases as the excavation face advances and the artificial support is mobilized. For an elastic, perfectly plastic material, thefindingsdiscussed earlier for statically indeterminate structures (Figure 3) apply and rates monitored during excavation stoppages must theoretically be zero or high if the safety margin is positive or zero, respectively. During advance, the deformation rate is related to the rate of advance and the extent of induced yielding (see Section 21.4.4 and [9]). In elastic, viscoplastic ground, it is possible that the demand temporarily exceeds the capacity. If the loading rate on a statically indeterminate structure is relatively high, the demand may exceed
614
Back Analysis Monitoring 0 *
E
50
O
Demand,/^ 100
E Kw Fitted by Balankoe'/tf/. [8]
150
^
Safety margin,5
200
„
H.
Capaciîy,C If C- constant, unknown in |then S = C - / ( l / e x ) magnitude
I
2 Ε5 Λ0 b^
Unknown baseline for safety margin i
'
* ι
L,A 15
20
25
30
35
Lateral displacement (mm)
Figure 4 Excavation volume versus tiltmeter deflection (lateral displacement) at Edmonton Convention Centre (after Balanko et al [8])
the currently mobilized capacity (as demonstrated by the convergence curve in Figure 2). The deformation rate, measured during advance stoppages, is then an indicator of the magnitude of the temporary safety margin deficit. When equilibrium is reached, the deformation rate must be zero and the safety margin is also zero. The ultimate safety margin is again unknown. With respect to monitoring underground openings, this discussion leads to several additional monitoring guidelines. In elastoplastic ground, the deformation rate is strongly related to the rate of excavation. The rate component due to advance must be separated from rates caused by timedependent properties or by yield zone propagation (see Section 21.4.4). Rates during advance must reflect the extent of the yield zone induced by the excavation process because the safety margin deficit must be larger in heavily yielding ground. Furthermore, in viscoplastic ground, the deformation rate during work stoppages also reflects the amount by which the currently mobilized capacity has been exceeded. Despite the many limitations for use of deformation data as previously discussed, adequate information in underground construction and rock engineering can in general only be gained if qualitative and quantitative monitoring techniques are employed to update continuously the knowledge base of the design engineer. Because the design of underground openings is almost exclusively concerned with stability assessment, it is necessary to develop a monitoring program based on a proper hypothesis of a relevant and kinematically possible failure mechanism. The following review of rock mass failure mechanisms is intended to provide some guidance for the selection of the most appropriate monitoring program including instrument type and sensor layout. 21.3 ROCK MASS FAILURE MECHANISMS 21.3.1 Failure Initiation, Propagation and Collapse Rock mass failures are common occurrences in underground construction and mining. The existence of failed or failing rock is seldom a problem as long as the failure process is understood, its location and extent are known and proper measures for its control are taken. For the design of a proper monitoring program, it is necessary to differentiate between failure initiation, propagation and ultimate collapse, and to group processes or mechanisms as well as causes and contributing factors [10]. For example, the processes that control rupture initiation may not be the same as those that dominate failure propagation and the transition from one to another behavioral mode must be evaluated. The final collapse situation often differs drastically from the initiation and propagation stages and the level of associated risk depends on the type of potential failure mode. Figure 5 illustrates the sequence and characteristics of the rock mass failure process. (a) Initiation. Rock mass failures may initiate in compression, by shear, in tension, or by buckling. Initiation of failure occurs if the capacity of the rock is exceeded locally due to stress concentrations. It often starts from a point at the opening wall but recent work [11,12] suggest that failure initiation
Deformation Monitoring for Stability Assessment of Underground Openings Nonviolent
Violent
Time-dependent process
me-mdependent process
With warning
Without warning
Unstable
Stable Time-dependent process
Acceptable Controllable Desirable
Figure 5
615
Time-independent process
Unacceptable out of control Undesirable
Rock mass failure development sequence
may also start from inside the rock mass in rock with a confining pressure dependent modulus or with imperfections. In shear, three modes of failure initiation may be encountered [13] depending on the orientation of the stress deviator near the wall. The location of initiation depends largely on the far field stress ratio and the shape of the opening. It would often be advantageous and safest if failure initiation could be prevented but this is seldom economical unless the field stresses are relatively low. Monitoring in the initiation phase must detect zones of unacceptable stress concentrations. (b) Propagation. Failure may also propagate in compression, by shear, in tension, or as a combination of several failure processes. Failure propagation occurs when the rock capacity has been exceeded locally and stresses must be transferred from failing ground to stable ground away from the point of failure initiation. It is often impossible or undesirable to prevent failure propagation. As a matter of fact, methods such as destressing, by blasting or by the creation of relief openings, are attempts to achieve the beneficial effects of failure propagation in a controlled manner. The main objective of ground control in an underground operation is to keep the propagation process under control. This can be achieved by increasing the rock mass strength (using reinforcement, confinement by support or backfill, etc) or by minimizing the extent of stress concentration zones (using alternative excavation sequences or rates of excavation, destressing, changing the shape or size of an opening, etc). Monitoring in the propagation phase should detect where and how failure is propagating such that the failure process can be controlled and remedial measures implemented if necessary. (c) Collapse. The process of collapse may be induced intentionally, as in mining, or it may develop when failure propagation is allowed to continue in an uncontrolled manner. In the context of this discussion, collapse is understood as the process of partial or complete disintegration of a rock mass. It may lead to a fall of ground, to the complete closure of an opening or to excessive propagation of a failure zone to the ground surface or to a nearby opening. Monitoring in the collapse phase must detect how far and how rapidly failure propagates, and whether ground control measures are effective. 21.3.2
Failure Modes Dominated by Weaknesses in Rock
Once a potential cause of failure has been identified, it is necessary to understand how failure might propagate. Shear failure through the rock mass, extension failure, slip on discontinuities or weaknesses (fault slip), bed separation, kinkband formation or combinations of these modes may lead to a ductile or gradual failure propagation (loosening, raveling, squeezing, slaking, swelling, etc) or to a brittle, instantaneous propagation (falls, slabbing, buckling, bursting, etc).
616
Back Analysis Monitoring
Only under special, relatively rare circumstances of homogeneous ground with uniform strength can rock failure be properly described by continuum, elasticity or plasticity models. Weaknesses or imperfections in a rock mass such as fractures or discontinuities can cause marked behavioral deviations [14, 15]. The size of the area of the failure zone may be increased, nonsymmetric and noncontinuous failure zone patterns may develop, even under hydrostatic loading of a circular opening, and combined instability mechanisms may dominate. Even though the overall rock mass may be sufficiently strong to avoid collapse, failure may often initiate from stress concentrations inside the rock mass leading to stress redistribution processes due to slip on a weakness plane where shear stresses exceed the local strength. The stress pattern in the rock mass is then significantly altered by the orientation, extent and properties of these weaknesses and failure of intact rock with otherwise adequate strength may be induced. Several examples demonstrating the propagation of failure induced by weaknesses in a nonuniform stress field (K0 = ajay = 0.5) are presented in Figure 6. In nature, imperfections are seldom uniformly distributed and may or may not intersect an opening. Finite element simulations of openings in rock with local weaknesses causing failure propagation were performed by Kwong [16]. The rock was assumed to be elastic, strain-weakening, brittle plastic with an instantaneous strength loss after peak. Figure 6 demonstrates that local weaknesses may initiate a propagation process (a-d) leading to narrow shear zones which eventually cause a block of nonyielded (elastic) rock to move into the opening. The three cases (III-V) with weaknesses at different locations lead to very similar failure modes that strongly deviate from those predicted by a conventional plasticity model (case II). Evidence in support of this failure mechanism can often be found in underground construction, e.g. at the Arlberg tunnel [17].
Figure 6
Yield patterns for five rock mass configurations (after Kaiser and Kwong [12])
617
Deformation Monitoring for Stability Assessment of Underground Openings
In reality, failure of a rock mass is almost always dominated by pre-existing or newly created weaknesses and can be grouped into two classes, as explained below (Figure 7). (a) Brittle failures. A kinematically possible failure mechanism is established as soon as a state of limiting equilibrium is reached, i.e. when failure is initiated. This is a brittle, instantaneous failure mode because the three stages of initiation, propagation and local collapse occur simultaneously. Deformation monitoring can provide little warning of this type of failure, called 'undetectable' failures in this context. (b) Ductile failures. Slip along a weakness does not immediately lead to an instability because a kinematically acceptable mechanism is not created unless yielding propagates sufficiently to release a block of nonyielded rock. Failure occurs only with some warning. Instabilities induced by local weaknesses belong to this type of failure mode and are 'detectable' by deformation monitoring. Consequently, rock mass failures may also be grouped into failures with little or no warning (brittle modes where initiation and collapse occur simultaneously) and those which display a gradual deformation process. Deformation monitoring is ideally suited to identify and understand failures that typically occur over a period of time and are characterized by a process of failure initiation, propagation and ultimate collapse. Ground falls, on the other hand, are often controlled by geological structures and are generally associated with large deformations which are almost instantaneous. As a consequence, insufficient time is available for their detection and for the implementation of remedial measures. The ductile failure mode is commonly encountered near shallow tunnels in relatively weak ground. Wong and Kaiser [18] have grouped soft ground behavioral modes in the space of normalized support pressure and field stress ratio. When yielding starts it may initially be localized and, if the support pressure is reduced, localized yield zones may expand creating a global, continuous yield zone. Ultimately, collapse starts when gravitational forces dominate and loosening occurs. This type of yield propagation can be easily detected by monitoring. The same development can be encountered in heavily fractured rock or in highly stressed rock masses at depth. In summary, behavior modes of underground openings can be structured as shown in Figure 8, and the following deliberations will deal with 'detectable' failures. 21.3.3 Concept of Monitoring for Failure Detection As mentioned before, for the development of a proper monitoring program it is essential to start with realistic hypotheses of relevant and kinematically possible failure mechanisms. This involves Ductile failure modes
Brittle failure mode Initial geological structure
(ii) Tensile rupture
| & Keyblock
(M> Yielding y
of intaCt r
^—"fêUV'w
( i ) Slip on geological \ ^ ^ X ^ structure ^ ~ ^
(i)
Slip
Figure 7 Brittle and ductile failure modes Behavior modes
Stable (nonyielding)
Unstable (yielding)
Ductile
Other modes
Global yielding
Localized shearing
Detectable
Brittle
Gravity ( loosening)
Bursting
Undetectable
Figure 8 Behavior modes of underground openings
Other modes
, on , geolog.cal structure
°°k
618
Back Analysis Monitoring
the identification and location of the three phases of the failure process (initiation, propagation and collapse) because the factors controlling each phase differ. The appropriate methods of detection and measures of control vary accordingly. Monitoring detects those failures that are detectable. Ductile failure modes can best be detected by monitoring and the risk of brittle failures must be reduced by eliminating them through preventive measures. For example, standardized ground support with pattern bolts and mesh or a thin shotcrete skin can be used to prevent these modes of instability. Alternatively, potential falls of ground can be identified by visual inspection (e.g. structural mapping) and handled by application of limit equilibrium [19] or key block methods [20]. Nevertheless, monitoring may assist in identifying the critical conditions favoring the development of sudden structurally controlled failures. In short, it is extremely important to recognize and accept that monitoring cannot eliminate all risk but can often reduce risk effectively. The interpretation of field observations is like solving a puzzle. The information contained in an individual piece (record) is often of little value unless the various pieces fit together properly. Hence, a monitoring program must be laid out in such a manner that the collected data can eventually be combined to arrive at a conclusive overall picture of the rock mass behavior. How to link individual pieces of information to provide immediate and conclusive feedback must receive special attention during the development phase of a monitoring program. A staged instrumentation and data interpretation approach is often most advantageous. Inexpensive, less accurate measurements should be spread over a wide area to detect spatial variations or regions of odd behavior such that focused, more accurate instruments can then be placed in strategic locations. This implies that some instruments should only be installed after data from the lowest level of monitoring provides some guidance. In practice, this highly desired flexibility can only be accomplished if contract documents reflect this concept and are formulated properly. The task of selecting the best instrument location and orientation will be discussed in subsequent sections using several practical examples.
21.4 DEFORMATION MONITORING 21.4.1
Introduction
From the previous discussions it follows that deformation measurements should be considered during a monitoring program for the following reasons: (i) to find the location of failure initiation; (ii) to identify the mode of failure initiation and propagation; (iii) to verify the cause of failure; and (iv) to assess the effectiveness of remedial measures. Furthermore, deformation monitoring may be needed to: (i) determine design parameters for design improvements or more accurate performance predictions; and (ii) extrapolate to assess long-term stability by separating excavation-induced and time-dependent processes. Deformation monitoring methods can be grouped into: (i) visual observations; (ii) surface (wall) displacement monitoring by surveying or convergence measurements; and (iii) deep-seated ground deformation monitoring with extensometers, such as rod or wire extensometers and the BOrehole Fracture monitoring EXtensometer (BOF-EX by Rocktest Ltd. [21]), or sliding micrometers and inclinometers (slope indicators, horizontal deflectometers, Trivec [22], etc.). The use of these methods and some aspects of data interpretation are discussed in the following sections. As discussed earlier, all stages of a monitoring program must have a specific objective and each monitoring phase must be guided by a hypothesis of expected ground behavior. Frequently, the only intent of monitoring is to confirm that the desired conditions actually exist and that assumed, undesirable and potentially risky situations are not encountered. Lack of understanding this goal may lead to excessively expensive monitoring programs or to the unjustified conclusion that monitoring was not needed because little was measured. The fact that only insignificant movements were recorded does not mean that the findings are of no value-they confirm that a stable condition was achieved.
Deformation Monitoring for Stability Assessment of Underground Openings
619
21.4.2 Visual or Qualitative Monitoring Visual observations constitute the cheapest and often most productive means of monitoring. Frequently, only qualitative information is needed to detect undesirable situations and to implement appropriate remedial measures. Thus, the most important purposes of visual monitoring are: (i) identification and assessment of highly stressed or failing ground; and (ii) identification and understanding of the failure mechanism. Borehole breakouts, crushing of intact rock, shears in rock near opening walls, slabbing of walls, popping of roofs, etc. are all indicators of relatively high compressive stresses and, if the rock mass strength is known or can be estimated, the magnitude of the in situ stress can be inferred. Critical tensile stress conditions can often be detected from joint separations, excessive overbreaks or falls of ground. Observations of distress may also provide an estimate of the principal stress direction and stress ratio. Visual monitoring for stability assessment starts with the identification of critical rock structures (joints, bedding planes, faults, etc). Valuable information for this purpose is routinely recorded by the site geologist and simple rock classification systems provide a basis for communication as well as a quick means of assessing the no-support limits (maximum unsupported spans) or establishing conventional support requirements. If instability modes are encountered, visual inspection can provide information about the direction of movement as well as the extent and type of the failure mode. Loose rock can be easily detected by the sound of a scaling bar and detailed displacement monitoring is not needed. Such relatively simple observations are extremely valuable for risk reduction and can provide sufficient information for the implementation of effective remedial measures and may eliminate the need for quantitative monitoring. For complex problems visual observations are seldom sufficient and the magnitude and spatial distribution of deformations as well as the rate of deformation may be needed to identify the extent and shape of a failure zone [12]. Both the magnitude and rate of deformation should be monitored because they provide complementary insight into the failure process. Furthermore, because displacements are influenced by support interaction they can be related to the effectiveness of rock reinforcements in yielding ground [23]. The deformation rate is affected by the rate of advance or rate of excavation, the extent of the yield zone, the time-dependent deformation of the ground, the support interaction, and most importantly by the safety margin deficit. These aspects are discussed separately in the following sections.
21.4.3 Use of Deformation Magnitudes 21.4.3.1 Monitoring of global versus local response Wall movements reflect the cumulative effect of ground deformation and failure due to an induced stress change. Consequently, convergence measurements provide an excellent indicator of the overall ground response but, as independent observations, seldom contain sufficient information to identify the cause of failure or the failure process. For example, relatively large springline deformations could be caused by a high horizontal stress before yield initiation, by a high vertical stress with dilation in a yield zone at the springline, or by an instability of a rock wedge at the springline. The shape and extent of yielding near a circular opening in five rock mass configurations were presented earlier in Figure 6. The corresponding radial deformations calculated by finite element analyses [10] are presented in Figure 9. Figures 9(a) and 9(b) present the convergence curves for roof-to-floor and springline-to-springline. Figure 9(c) presents the radial strain development for four extensometer sections (A-D) for case V only (see Figure 6). A detailed study of these simulated deformation curves reveals that local measurements at an appropriate location are needed to identify the location of yielding and the resulting mode of failure. For example, the pushing-in of a wedge of nonyielded rock created by two local rock mass weaknesses or a single weakness plane near the springline (case V, Figure 6) is only detected by a horizontal extensometer at D. The convergence records (II-V) demonstrate that nonelastic deformations associated with yielding occur in all four cases. The nonyielding, elastic roof is displaced in a nonlinear manner for cases II-V. Furthermore, these openings deform by almost equal amounts at a fictitious support pressure of about 50%. Remedial measures, such as roof bolts to restrain excessive roof deformations, would be completely ineffective as the cause of deformation is found at the springline. The actual mode of failure cannot be identified by the convergence measurements alone.
620
Back Analysis Monitoring
(a)
V
_L _L
0.2
0.4
0.6
— O ■ a *
>**
L
_L il 0.8
1.0
1.2
J
1.4
1.6
(I) Elastic (II) Plastic (III) Symmetrical WE (EC) Non symmetrical WE ( 2 ) I WE and IWJ
L
1.8
J
2.0
0
(c)
1 c
a>
3
3
4
5
6
7
8
9
10
II
12
•\
_c
0)
2
100
S 80 S o
I
Normalized displacement at springline, u/ue
Normalized displacement at roof,i//i/ e
60 \-
b
b-x.
"0
♦ Location A O Location B
20
I
♦A —
> 0.5
1.0
1.5
2.0
2.5
A
_
Case 3£
^ 3.0
Average radial strain (%)
Figure 9
Radial displacements caused by failure modes shown in Figure 6: (a) convergence of roof; (b) convergence of springline; (c) extensometer readings at roof and springline (Kaiser [10])
Several other interesting observations deserve some comments. In case II, the roof and springline respond by a dramatic increase in convergence as soon as yielding is initiated and both locations provide a clear warning signal of yield initiation and propagation. Yielding is reflected in case III by a gradual increase in the nonelastic component of deformation and this would be difficult to detect unless the elastic response was known from other monitoring locations in elastic ground. In both cases (II and III), large displacements accumulate during a relatively minor decrease in support pressure. Therefore, instability would suddenly occur and would be rather difficult to control once the yield propagation stage was reached. Case IV is similar to case III but monitoring data from the springline could initially be misconstrued as elastic rock response. For this case, the roof convergence only reflects yield initiation and propagation at the springline. The convergence at the springline is small until sudden and excessive deformations occur at a fictitious support pressure of about 65%. The frequently encountered situation of case V is most interesting. Both the nonyielding roof and the springline deform almost identically. This could be misinterpreted as being caused by a uniform stressfieldand an axisymmetric yield zone. As shown in Figure 6, this is clearly not the case and the true failure mode cannot be determined from these convergence measurements alone. Conversely, the records of radial strain from four simulated extensometers (Figure 9c) clearly identify that the roof responds elastically (location A) and that an elastic wedge (at B), created by localized shear (at C), is moving into the opening. The rock outside the weakness (at D) again responds elastically. Consequently, local displacement observations are absolutely essential for a proper identification of the failure mode induced by rock mass imperfections. The importance of local measurements for failure mode identification can be further demonstrated on data from one of four Washuuzan tunnels (Japan 1983; unpublished data). Measurements of convergence and rock bolt strain recorded during a work stoppage in this tunnel (Figure 10a) show a sudden, albeit small, increase of all but one (number 2) of the convergence records between 75 and 85 days. Without local measurements, it would be impossible to identify conclusively the cause of this unexpected increase in convergence. Simultaneous measurements of axial strains (forces) in many rock bolts of the same cross section showed little load build-up (Figure 10b). However, one
Deformation Monitoring for Stability Assessment of Underground Openings
621
Elapsed time (days)
0
20
40
60
80
100
120
140
0
Elapsed time (days)
20
40
60
80
100 120
140 160
Elapsed Time (days)
Figure 10 Measured convergence and bolt loads at Washuuzan tunnel station 248 + 60 (1983, unpublished): (a) wall convergences; (b) and (c) bolt loads (in metric tonnes) at two locations, determined by bolt strain measurements
bolt at the right springline experienced a sudden load increase at 85 days (Figure 10c). This single, local measurement permitted the location of the source of movement and provided sufficient information to allow implementation of minor remedial measures consisting of spot bolting of this local instability (tensile fracture or shear) at a depth of between 0.75-1.5 m from the tunnel wall. These measurements were of great assistance in the design and risk assessment of this tunnel. In summary, convergence measurements are good indicators of instability but localized measurements are needed for failure mode identification.
21.4.3.2
Placement of instruments to observe local failures
It is impossible to instrument fully the entire rock volume affected by an excavation. In a paper by de Mello [24] it was stated that \ . . any planning of instrumentation and interpretation automatically pre-supposes an anticipated model of theoretical behaviour (physical and mathematical) . . Λ Consequently, the location of an instrument must be selected on the basis of one or several hypothetical failure modes. The layout of an instrumentation package for safety assessment is only as good as the hypotheses on which the design is based. An understanding of all possible instability mechanisms is a compulsory prerequisite for safety-related monitoring. Once a failure mode hypothesis has been established, it is normally a simple matter to identify the best instrument type, location, orientation and anchor length. For example, the benefit of a well-placed instrument for local monitoring is nicely demonstrated by the bolt strain measurements discussed earlier (Figure 10) from the Washuuzan tunnel.
21.4.3.3
Convergence to assess support performance
Artificial supports in underground openings act as a support pressure or as a rock reinforcement. The support reduces the wall convergence by resisting the ground pressure and by improving the effective rock mass properties. Consequently, the effectiveness of a support must be reflected in the wall convergence and it should be assessed in terms of convergence or convergence rate reductions (see also Section 21.4.4.2). The use of convergence measurements as part of the observational tunnel design approach for the selection of the density and length of fully grouted bolts has been described by Indraratna and
622
Back Analysis Monitoring
Kaiser [23]. In detailed studies of bolted model tunnels in homogeneous and jointed artificial rocks, an analytical method for convergence-controlled bolt design was developed and verified [25, 26]. This study showed that the normalized convergence or convergence ratio seems to be linearly related to the bolt density parameter. The normalized convergence was defined as the ratio of the total displacement of the reinforced tunnel wall to the displacement of the unreinforced wall. The bolt density parameter relates the bolt length and spacing to the tunnel size [25]. Based on these studies, the support density required to restrain deformations to a desired level can be found by extrapolation from convergence measurements. Unfortunately, this approach is so far only directly applicable to axisymmetrical situations. 21.4.4 Use of Deformation Rates Barlow [27] demonstrated that deformation rates can be used most effectively for the purpose of field data interpretation if the current rate, at some time after the tunnel face passes the measurement section, is normalized to the maximum rate recorded near the excavation face. This normalized deformation rate is an effective measure for assessing: (i) the extent of the yield zone; (ii) the effect of remedial measures or the support effectiveness [9]; and (iii) the mobilized safety margin. The deformation rate close to the tunnel face is not much affected by the rock reinforcement or the support placed in the excavation but rather reflects the ultimate tunnel convergence. Hence, the normalized displacement rate is essentially independent of the magnitude of the ultimate convergence and is an indicator of the current gap between demand and mobilized capacity. Barlow [27] expanded the method introduced by Guenot et al [28] for separating timedependent from excavation rate dependent components of deformation and developed an analytical procedure for field data analysis. An illustrative example of the application of the normalized convergence rate for the three purposes listed above is given in Figure 11 with data from the Enassan tunnel (Japan). The observed convergences recorded over a period of 200 days during the excavation of this tunnel were presented by I to [29] and analyzed by Barlow [27]. Figure 11 presents the normalized displacement rate for the roof at station A. Indicated on this figure are the times when the bench and the invert were excavated during the staged NATM excavation and support process. Standard 9 m long supplemental grouted bolt anchors were installed three times, once after bench excavation (four per ring) and twice after the invert was excavated (eight and 15 per ring). This additional support was applied in an effort to control excessively large roof settlements ( > 0.8 m). The sudden changes in the convergence rate curves reflect the nonsteady nature of the excavation process. The displacement rate increases almost instantaneously as soon as the tunnel is advanced and drops off gradually when the advance is stopped.
R/a = 1.5 - — R/a propagates from 1.5 to 4 A — A Smoothed data - roof ^mm Excavation stoppage
- R/o = \.'b*
4
*
J
10°
Number of extra rock bolts ( 9 m length )
I I
[•«-Heading 0
_J
L 20
(-«Bench 40
60
[««-Invert I 80
100
120
140
160
180
200
Time (days)
Figure 11 Measured roof displacement fitted by predicted displacement rates assuming: (i) constant extent of yield zone of R/a = 1.5 (full line) and (ii) propagating yield zone to R/a = 4 during bench excavation (dashed line) (after Barlow [27])
Deformation Monitoring for Stability Assessment of Underground Openings
623
21.4.4.1 Extent of yield zone from deformation rates The observed convergence rates are shown in Figure 11 by connected triangles and two predicted rates are shown by the full and dashed lines. Until the bench is excavated at 40 days the curve fit is excellent, assuming a yield zone of 1.5 times the tunnel radius (R/a = 1.5). Subsequently, the observed rates do not drop as rapidly as expected (compare with the full line). As a matter of fact, the observed rates remain almost constant from 40-70 days during the advance of the bench. This is indicative of a propagating plastic zone. By trial and error, the radius of the yield zone (R) was established and it was found that it must have expanded to about four tunnel radii (dashed line for R/a = 4). It is interesting to note that monitoring could have established the size of this yield zone within 10-20 days after bench excavation and could have been used to design the supplemental support. Since the resulting yield zone is in excess of 20 m, it is evident that very long anchors would have been needed to stabilize such a deep yield zone. It is also evident from Figure 11 that the slope of the normalized rate plot is related to the extent of the yield zone. In slightly yielding ground (R/a = 1.5), the normalized rate should have dropped to about 7% (full line) in 20 days after bench excavation while the tunnel was advanced at an average rate of about 1 m day -1 . In heavily yielding ground (R/a — 4), the rate should have dropped to about 20% (dashed line) over the same time period (e.g. between 44-64 days). The measured normalized rate dropped only 50% (triangles), indicating that a new stable equilibrium was not yet reached after 20 days. This example demonstrates that the normalized displacement rate plot provides a useful means of evaluating tunnel performance in yielding ground. This method of displacement rate assessment also fulfills one of the most important monitoring requirements, namely provision of immediate feedback. However, this approach requires a high data-recording frequency. For example, displacement rates should be recorded daily during tunnel advance and during the early stages of excavation stoppages.
21.4.4.2 Support effectiveness from deformation rates Barlow and Kaiser [9] demonstrated that an effective support system could lead to a normalized displacement rate reduction of more than one order of magnitude. A careful study of Figure 11 reveals that none of the extra bolting efforts (four plus eight plus 15 bolts of 9 m length) caused any significant displacement rate reduction. Hence, these bolts contributed little to the control of roof settlements. Indraratna and Kaiser [26] arrived at the same conclusion by considering the magnitude of the displacement rather than deformation rates. The benefit of the rate approach is the immediate feedback before the magnitude of the ultimate deformation is known. Only 10 to 20 monitoring days would have been required at the Enassan tunnel to determine the inadequacy of the initial and supplemental bolting for the control of yield zone propagation after the bench excavation.
21.4.4.3 Assessment of mobilized safety margin from deformation rates The normalized deformation rate recorded during excavation stoppages is a measure of the mobilized safety margin. For the Enassan tunnel, the deformation rates during excavation stoppages (lower bound values on Figure 11) are slightly more than one order of magnitude lower than the peak rates recorded during advance. The rate during stoppages should theoretically be zero for elastoplastic rock. In viscous rock, a nonzero rate reflects a safety margin deficit (see Section 21.2.4). The predicted rate during stoppages (see Figure 11) is about three times higher for R/a = 4 than for R/a = 1.5. Obviously, the latter case with less-yielded rock would be closer to a stable equilibrium and the safety margin deficit would be smaller. Nevertheless, the measured rates decrease gradually with time, indicating that a stable condition is being approached. Without further investigations of other tunnel records, it is not possible to provide quantitative limits or guidelines for selecting safe, normalized displacement rates. However, based on a qualitative assessment of laboratory test results and measurements from some case histories, it seems that a new equilibrium between demand and capacity has been reached when the normalized deformation rates recorded during work stoppages drop to less than 0.5-0.2% ( 5 x l 0 ~ 3 t o 2 x l 0 ~ 3 in Figure 11). These rates are not applicable for swelling ground or rocks with dominant, timedependent deformation behavior such as rock salt. Based on the normalized deformation rate limits given above, a stable equilibrium was achieved at the Enassan tunnel after about 150 days. It
624
Back Analysis Monitoring
appears that the remedial measures implemented after 95 days, i.e. the installation of 15 additional rock bolts per meter of tunnel, or time-dependent stress redistribution processes reduced and eventually eliminated the safety margin deficit. If the yield zone had been confined by a more effective initial support system to a radial extent of R/a = 1.5, then the same safety margin (normalized deformation rate of 0.5-0.2%) could have been reached after about 80 days (see full line in Figure 11). The following practical monitoring guidelines for deformation rate monitoring evolve from this case history and from the work by Barlow [27]. The absolute deformation rate is a poor indicator for the assessment of the safety of an underground opening in yielding ground. However, the normalized deformation rate, if recorded during excavation stoppages, provides a useful measure of the current safety margin deficit and can be applied effectively in the observational design approach.
21.4.5 Miscellaneous Considerations for Displacement Monitoring 21.4.5.1
Required accuracy
Today's technologies permit recording of deformations with sufficient accuracy for any risk assessment related purpose. In general, an instrument should be capable of detecting but not accurately measuring the elastic rock response. Furthermore, the measurements should be accurate enough for calculation of a deformation rate and to provide a basis for comparison with data from locations experiencing large deformations due to yielding. The accuracy for risk assessment purposes seldom needs to be as high as for the back-analysis of rock mass properties. Comparatively high accuracies would, however, be required if rock strain measurements were to be used to establish the field stress level or for comparison with critical strains ([30] or [31]).
21.4.5.2
Required duration and frequency of readings
The duration and frequency of readings depend on the monitoring purpose, the rock mass properties, the advance or excavation rate and many other factors. Data should be collected as frequently as is practically and economically possible. As illustrated in Section 21.4.4, valuable information is contained in the deformation rate. This requires much higher data collection frequencies than for the determination of the ultimate deformation magnitude. Furthermore, a high data collection frequency is required when the deformation rate is high (for example, near the face, during advance, and when mining-induced stress changes occur). Despite the problems of interference with the construction process, it is extremely important that as many data as possible are collected, especially during work stoppages. On many projects, valuable information is lost because insufficient data are collected. The monitoring program of the Enassan tunnel provided excellent and sufficiently frequent measurements for the determination of the extent of the yield zone and the verification of the effectiveness of remedial support measures (Section 21.4.4), and can be used as a guide for comparable projects.
21.4.5.3
Displacement monitoring layout
Pelli [32] conducted extensive three-dimensional numerical simulations of a circular tunnel advanced by a TBM to investigate the effect of excavation face position, stress field orientation, rock anisotropy and nonlinearity on monitoring data [33, 34]. This work revealed an extreme sensitivity of extensometer measurements to the actual tunnel face position relative to the extensometer installation point, particularly in a nonuniform stress field and in anisotropic rock. Some typical normalized displacement profiles (deepest anchor assumed to be fixed at infinity) are presented in Figure 12. (The normalized displacement is defined as uTE/apw and K is the ratio of horizontal to vertical, total stress; ur = radial displacement; E = Young's modulus; a = tunnel radius; and Pv = vertical field stress. In Figure 12 p a = axial field stress; ph and pH = minor and major horizontal field stresses, respectively; x = distance from tunnel face; and (x/2a)ms = location where measurements are taken.) Figures 12(a) and 12(b) illustrate the effect of the stress ratio on the total displacement profile. For K = 2, essentially no displacement occurs at the crown at R = 3a (Figure 12b), whereas a significant percentage of displacement is expected at more than five tunnel radii at the springline (Figure 12a).
625
Deformation Monitoring for Stability Assessment of Underground Openings (a)
(b)
t K
U/2tf)ms = 0
#\~ο50
Uf\075
Ö -0.50 I p 2
3
4
5
Normalized distance from tunnel wall, R/a
1
6
jj σ
E
-0.75
1
v
-3.0
Crown
_J
2 3 4 5 Normalized distance from tunnel wall, R/a
2 3 4 5 Normalized distance from tunnel wall, R/a
Figure 12 Normalized radial displacement profiles recorded by extensometers at various distances from the tunnel face (K = 2): (a) total displacements at springline if installed ahead of the face and readings taken at x/2a = 0-3; (b) total displacements at crown if installed ahead of the face and readings taken at x/2a = 0-3; and (c) partial displacements at crown if installed or zero reading taken at (x/2a)0 = 0-0.5 (after Pelli [32])
These displacement distributions must be considered when selecting the location of the deepest anchor point in a nonuniform stress field (see also Figure 14). Because radial extensometers can only be placed at some distance behind the face in deep tunnels, only part of the total deformation is recorded (called partial displacement). This is illustrated by Figure 12(c) for the crown. If the initial reading was taken at half or one tunnel radius behind the face l(x/2a)0 = 0.25 and 0.5, respectively], zero or even small compressive strains would be recorded. A record of zero straining could be misinterpreted if viewed in isolation or interpreted as a malfunctioning extensometer. An example showing very low straining in the direction perpendicular to the major principal stress was recorded at the Underground Research Laboratory, Pinawa, Canada [35]. An even more drastic situation is presented in Figure 13 for an extensometer in transverse isotropic rock {E2/E1 = 10). With vertical bedding and the strata parallel to the tunnel (case 2), significant compression would be recorded at the roof. In contrast, extensometers in the crown of a tunnel in horizontally bedded (case 1) or vertically bedded (case 3) rock with the strata perpendicular to the tunnel axis show extension strains. In summary, instruments should only be placed where sufficiently large displacements are expected and the effect of the advancing face must be considered. The length of the extensometer must be chosen by relating the sensitivity of an extensometer type to the expected displacement field. For example, Figure 14 provides for a circular opening in linear elastic rock the limits of the zone containing 67% of the total displacement recorded at the tunnel wall for various stress ratios (K = oJay). For K = 1,67% of all deformation occurs within two tunnel radii from the tunnel wall (between R/a = 1 and R/a = 3). The contours presented in Figure 14 provide the depth at which an anchor point of an extensometer would have to be placed to record 67% of the total rock mass
626
Back Analysis Monitoring
Normalized distance from tunnel wall, R/a
Figure 13 Normalized radial displacement profiles recorded by extensometers placed in the crown at the tunnel face for three transverse isotropic cases (see inset) (K = 2) (after Pelli [32])
T
Figure 14 Limits of zones containing 67% of the total radial displacements (including displacements ahead of the face) near a circular opening in elastic rock (K = ah/ay)
deformation. It follows from this rather simple model that extensometers in nonuniform stress fields (K φ 1) should be placed at various depths, i.e. slightly longer extensometers than for K = 1 should be used in the direction of the major principal stress and significantly shorter extensometers in the direction of the minor principal stress. In strongly nonuniform stress fields (2 < K < 4), little or no deformation can be recorded in the direction of the minor principal stress (vertical for Figure 14). With respect to convergence measurements, Pelli [32] demonstrated that the magnitude of the axial stress, in addition to the location of the convergence pin installation point relative to the face, significantly alters the radial wall displacement profile, as illustrated by Figure 15. Extensometer measurements show the same sensitivity to axial stress. 21.4.5.4 Combination of stress change and displacement measurements No deformation occurs without a change in stress and the stress changes are related to the deformations by the deformation properties of the rock mass. Hence, if deformation measurements are combined with stress change measurements, the deformation properties can be back-analyzed, particularly if the rock mass can be assumed to behave in an elastic manner. The need for a simultaneous determination of stress changes and displacements was justified on the basis of the convergence/confinement method by Korpach and Kaiser [36]. Furthermore, Pelli [32] showed
Deformation Monitoring for Stability Assessment of Underground Openings \
Uj
Or—«
Ô
-0.4
£ E o
-06L 0
z
627
I
2
3
4
Normalized distance to tunnel face,
5
6
x/Za
Figure 15 Convergence distribution recorded along a tunnel for pins placed at (x/2a)0 = 0.25 from the tunnel face {K = 2). Horizontal stress in the direction of the tunnel axis variable between pa = 0-4pv (after Pelli [32])
that the radial stresses change rapidly near an advancing face, whereas the tangential stresses change more gradually and the shear stresses reverse near the tunnel face. Consequently, it is almost impossible, in practice, to predict accurately the actual stress change at a particular instrument location and it should be measured while deformation measurements are taken. In particular, the axial stress changes near the wall affect displacement measurements drastically ( [32] and Figure 15). Furthermore, the actual stress change recorded during deformation monitoring with instruments installed near the tunnel face is normally very small. This was demonstrated during the sinking of a shaft [36], where stress change measurements indicated that only about 10% of the total stress change occurred while the extensometers were read. Hence, stress change measurements are essential for a conclusive interpretation of displacement records for the back-analysis of rock mass deformation properties. 21.4.5.5 Monitoring data interpretation The process of monitoring data interpretation, or back-analysis, is often complex and is discussed by others in this volume. A detailed assessment and interpretation of a typical monitoring package installed during the advance of a tunnel excavated by a tunnel-boring machine has been performed by Pelli [32]. His findings [33, 34] demonstrate that it is very difficult to interpret measurements if a monitoring program is incomplete and certain essential observations are missing. A review of many conventional instrument arrays shows that severe deficiencies frequently exist in monitoring programs. As discussed earlier, a conclusive data interpretation is only possible if the instrumentation package is designed to collect all essential pieces of information. The guidelines provided throughout this chapter are intended to eliminate some of these deficiencies. 21.5 CONCLUSIONS Monitoring provides an economic means for reducing the risk of failure in underground construction and it constitutes an essential component of modern rock engineering. Without monitoring many daring and innovative projects would never have been attempted or successfully completed. An effective and well-managed monitoring program must be directed to satisfy a set of monitoring objectives. These objectives should be established before construction begins. Immediate feedback must be provided and used to reap the benefits of a monitoring program. A reliable hypothesis of the expected rock mass behavior or potential failure modes constitutes the basis for designing a meaningful monitoring program and facilitates rational data interpretation. The choice of instrument types, instrument locations and orientations, recording frequencies, etc. depends on the objectives of a monitoring program as well as the suspected ground response. Behavior mode identification is the primary task of monitoring for stability assessment and monitoring should be executed in stages, starting with a relatively crude instrumentation package
628
Back Analysis Monitoring
covering a wide area. Pattern monitoring for the purpose of stability assessment is only encouraged for this level of monitoring. Instruments can best be chosen for specific purposes and located at strategic positions based on thefindingsof this lowest level of monitoring and on the hypotheses of several likely behavior modes. Several monitoring guidelines have been presented throughout this chapter to assist in the development of optimal monitoring programs. Visual observations of rock or support deformation and failure are often the cheapest and most productive means for detecting undesirable situations or for confirming adequate excavation performance. Quantitative observations are, however, needed to determine the cause of unacceptable behavior. In general, convergence records are good indicators of yield initiation and propagation but seldom provide sufficient insight to establish the cause of failure. The exact location of failure initiation and the failure mode can often only be identified by local deformation measurements. Local observations are frequently essential for a conclusive interpretation of field measurements, but a monitoring program must be laid out in such a manner that the collected data can eventually be combined to arrive at a conclusive overall picture of the rock mass behavior. Normalized deformation rates rather than the magnitudes of deformation should be used for ongoing performance assessments. They provide a measure of the safety margin deficit and immediate feedback when remedial measures can be implemented as part of the regular construction process and are most effective. This demands higher than normal recording frequencies. Ideally, automated deformation-measuring systems should be installed wherever possible. The top priority of any data interpretation should be to confirm an assumed mode of behavior. For this purpose, measurements from carefully selected and well-positioned instruments can be compared with predictions from analytical or numerical models or with empirically established limits. Frequently, the only intent of monitoring is to confirm that the desired conditions actually exist and that assumed, undesirable and potentially risky situations are not encountered. Not understanding this goal may lead to excessively expensive monitoring programs or to the unjustified conclusion that monitoring was not needed because little was measured. Because rock masses are seldom homogeneous, weaknesses and discontinuities cannot be ignored during the assessment of underground opening stability. However, sufficient yielding must occur, often along more than one weakness, to create a kinematically acceptable failure mode. Monitoring must attempt to detect failure initiation on such imperfections in the rock mass. Back-analyses based on continuum models, neglecting weaknesses and associated rupture modes, can often be misleading. ACKNOWLEDGEMENTS Much of the content of this chapter was written for the 12th Canadian Geotechnical Society Colloquium presented at the Geotechnical Conference in 1987. The financial support from the Canadian Geotechnical Society for the preparation of the colloquium lecture was much appreciated. Most of the research was conducted with support from the Natural Sciences and Engineering Research Council of Canada (NSERC) and was executed by my former graduate students J. P. Barlow, A. Guenot, S. Maloney, B. Indraratna, A. Kwong, F. Pelli and R. Wong. Without this support and their contributions this chapter could not have been written. My colleagues D. R. McCreath, J. Simmons and D. Tannant have assisted through their constructive comments on various versions of this chapter and their contributions are thankfully acknowledged.
21.6 REFERENCES 1. Whitman R. V. Evaluating calculated risk in geotechnical engineering. 17th Terzaghi Lecture. J. Geotech. Eng. Div.t Am. Soc. Civ. Eng. 110, 145-188 (1984). 2. Peck R. B. Advantages and limitations of the observational method in applied soil mechanics. Geotechnique 19,171-187 (1969). 3. Lane K. S. Field test sections save cost in tunnel support. Report from Underground Construction Research Council, ASCE, p. 95. ASCE, New York (1975). 4. Hanna T. H. Field Instrumentation in Geotechnical Engineering, p. 843. Trans Tech, Clausthal-Zellerfeld (1985). 5. Dunnicliff J. Geotechnical Instrumentation for Monitoring Field Performance, p. 577. Wiley-Interscience, New York (1988).
Deformation Monitoring for Stability Assessment of Underground Openings 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36.
629
Franklin J. A. (Ed.) Mine Monitoring Manual, p. 156. The Canadian Institute of Mining and Metallurgy, Special Volume 42 (1990). Matsuo M. and Kawamura K. Reliability based design of supporting system for NATM. In Proc. 4th Int. Conf. Application of Statistics and Probability in Soil and Structural Engineering, Firenze, Italy, pp. 1517-1530 (1983). Balanko L. A., Morgenstern N. R. and Yacyshyn R. Tangent pile wall for Edmonton Convention Centre. In Application of Walls to Landslide Control Problems (Edited by R. B. Reeves), pp. 108-123. ASCE, New York (1982). Barlow J. P. and Kaiser P. K. Interpretation of tunnel convergence measurements. In Proc. 6th Int. Congr. Rock Mech., Montreal (Edited by G. Herget and S. Vangpaisal), vol. 1, pp. 787-792. Balkema, Rotterdam (1987). Kaiser P. K. Detection of rock mass rupture modes. In Proc. 6th Int. Congr. Rock Mech., Montreal (Edited by G. Herget and S. Vangpaisal), vol. 3, Panel Discussion, pp. 1457-1461. Balkema, Rotterdam (1987). Santarelli F. J., Brown E. T. and Maury V. Analysis of borehole stresses using pressure-dependent, linear elasticity. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 23, 445^49 (1986). Kaiser P. K. and Kwong A. Stability of openings in rock with imperfections. In Proc. 29th U.S. Symp. Rock Mech., Minneapolis, MN (Edited by P. A. Cundall, R. L. Sterling and A. M. Starfield), pp. 735-738. Balkema, Rotterdam (1988). Wong R. and Kaiser P. K. Design and performance evaluation of vertical shafts: rational shaft design method and verification of design method. Can. Geotech. J. 25, 320-337 (1988). Makurat A., Barton N., Vik G., Chryssanthakis P. and Monsen K. Jointed rock mass modelling. In Proc. Int. Conf. Rock Joints, Loen, Norway (Edited by N. Barton and O. Stephansson), pp. 647-656 (1990). Kaiser P. K. and Maloney S. Factors influencing the stability of deep boreholes. In Proc. 6th Int. Congr. Rock Mech., Montreal (Edited by G. Herget and S. Vangpaisal), vol. 1, pp. 675-680. Balkema, Rotterdam (1987). Kwong A. Borehole and Tunnel Stability in Rock with Anisotropie Strength and Imperfections, p. 162. Ph.D. Thesis, University of Alberta, Edmonton, Canada (1990). Vilanek J. Der Arlberg Strassentunnel und die Zufahrtsrampen - Baudokumentation, p. 697. Herausgeber: Arlberg Strassentunnel Aktiengesellschaft, Innsbruck, Austria (1981). Wong R. and Kaiser P. K. Ground behaviour near soft ground tunnels. In Proc. ITA Conf. Large Underground Openings, Firenze, Italy, pp. 942-951 (1986). Hoek E. and Brown E. T. Underground Excavations in Rock, p. 527. Institution of Mining and Metallurgy, London (1980). Goodman R. E. and Gen-hua Shi. Block Theory and its Application to Rock Engineering, p. 338. Prentice-Hall, Englewood Cliffs, NJ (1985). Thompson P. M., Kozak E. T. and Martin C. D. Rock displacement instrumentation and coupled hydraulic pressure/rock displacement instrumentation for use in stiff crystalline rock. In Proc. NEA Workshop Excavation Response in Geological Repositories for Radioactive Waste, Winnipeg, pp. 257-270 (1988). Koppel J., Amstad Ch. and Kovari K. The measurement of displacement vectors with the TRI VEC borehole probe. In Proc. Int. Symp. Field Measurements in Geomechanics, Zürich, pp. 209-218 (1983). Indraratna B. and Kaiser P. K. Control of tunnel convergence by grouted bolts. In Proc. Conf. Rapid Excavation and Tunneling, vol. 1, chap. 22, pp. 329-348 (1987). de Mello V. F. B. Reflections on design decisions of practical significance to embankment dams. 17th Rankine Lecture. Geotechnique 27, 281-355 (1977). Indraratna B. and Kaiser P. K. Analytical model for the design of rock bolts. Int. J. Numer. Anal. Meth. Geomech. 14, 227-251 (1990). Indraratna B. and Kaiser P. K. Design for grouted rock bolts based on convergence control method. Int. J. Rock Mech. Min. Sei. ά Geomech. Abstr. 27, 269-290 (1990). Barlow J. P. Interpretation of Tunnel Convergence Measurements, p. 235. M.Sc. Thesis, University of Alberta, Edmonton, Canada (1986). Guenot A., Panet M. and Sulem J. A new aspect in tunnel closure interpretation. In Proc. 26th U.S. Symp. Rock Mech., Rapid City, SD (Edited by E. Ashworth), pp. 455-460. Balkema, Rotterdam (1985). Ito Y. Design and construction by NATM through Chogiezawa fault zone for Enassan tunnel on central motorway (in Japanese). Tunnels Underground 14, 7-14 (1983). Sakurai S. Direct strain evaluation technique in construction of underground openings. In Proc. 22nd U.S. Symp. Rock Mech., Cambridge, MA (Edited by H. H. Einstein), pp. 278-282. MIT Press, Cambridge, MA (1981). Stacey T. R. A simple extension strain criterion for fracture of brittle rock. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 18,469^74(1981). Pelli F. Near Face Behaviour of Deep Tunnels, p. 406. Ph.D. Thesis, University of Alberta, Edmonton, Canada (1987). Pelli F., Kaiser P. K. and Morgenstern N. R. The influence of near face behaviour on monitoring of deep tunnels. Can. Geotech. J. 28(2), 226-238 (1990). Pelli F., Kaiser P. K. and Morgenstern N. R. An interpretation of ground movements recorded during construction of the Donkin-Morien tunnel. Can. Geotech. J. 28(2), 239-254 (1990). Lang P. A. Room 209 excavation response test in the underground research laboratory. In Proc. NEA Workshop Excavation Response in Geological Repositories for Radioactive Waste, Winnipeg, pp. 295-330 (1988). Korpach D. R. and Kaiser P. K. Use of stress change measurements to assess performance of underground excavations. In Proc. Int. Symp. Prediction and Performance in Geotechnical Engineering, Calgary, pp. 319-328 (1987).
22 Rock Mass Behavior During Large-scale Cavern Excavation SATOSHI HIBINO and MUTSUMI MOTOJIMA Central Research Institute of Electric Power Industry, Chiba-ken, Japan 22.1
INTRODUCTION
631
22.2 CAVERN SHAPES, ROCK CONDITIONS AND MEASUREMENT ITEMS 22.2.1 Cavern Shapes and Mechanical Properties of the Rock Mass 22.2.2 Measurement Items
632 632 632
22.3
635
ROCK MASS BEHAVIOR AROUND CAVERNS DURING EXCAVATION
22.3.1 22.3.2 22.3.3 22.3.4 22.3.5 22.3.6 22.3.7
Subsidence Characteristics and Relaxed Zones of Ceiling Rocks Rock Deformation Characteristics of Walls Stresses in Arched Concrete Linings and Their Distribution Stresses in Arched Concrete Linings and Rock Deformation Variation of Elastic Wave Velocity Joint Opening in the Rock Mass Variation of Permeability
22.4 CHARACTERISTIC FEATURES OF ROCK BEHAVIOR 22.5
DESIGN OF ARCHED CONCRETE LININGS FOR CAVERNS
22.5.1 22.5.2
635 636 638 640 641 644 646 647 648 648 649
Role of the Arched Concrete Lining Design of the Arched Lining
22.6
CONCLUSIONS
650
22.7
REFERENCES
650
22.1 INTRODUCTION In the excavation of tunnels, caverns, slopes and others, stresses in rock masses will change and even cause collapses in some cases. Such rock mass behavior is strongly affected not only by the deformation characteristics, strength and ground pressure of the rock mass but also by the discontinuous and nonhomogeneous nature of the rock mass. Geological structures such as faults and joints constitute the discontinuity and they have an important influence on the behavior of the rock mass. In an attempt to explain the complicated behavior of the rock mass, it is necessary to carry out analyses, field observations, measurements and so on in various kinds of excavation work. In Japan, large-scale underground caverns were excavated at more than 20 sites for the construction of underground pumped storage power stations in the 1970s and 1980s. Prior to excavation work, excavation analyses and rock mass tests were conducted and various measurements of the rock mass were taken to ensure the safety of the excavation work. In this chapter, typical examples of rock mass behavior during the excavation of these caverns are discussed, and at the same time the characteristics of rock mass behavior, noted through field measurements at many sites, are clarified. In addition, the mechanism of rock mass behavior 631
632
Back Analysis Monitoring
estimated from these results is discussed, with consideration given to the design of arched concrete linings. 22.2 CAVERN SHAPES, ROCK CONDITIONS AND MEASUREMENT ITEMS 22.2.1 Cavern Shapes and Mechanical Properties of the Rock Mass The underground power stations treated in this chapter are located throughout Japan (Figure 1). The geological conditions on these sites are various. The caverns are, on average, about 50 m in height, about 25 m in width and about 100 m in length; the average volume is about 120 000 m3 (Table 1). As shown in Figure 2, most of the caverns are mushroom shaped, but two sites are rather more egg shaped. In the excavation of the caverns, the arched part (sections 1 and 2 in Figure 2) is excavated first, then rock bolts (B) are set and the arched concrete lining (L) is placed. After excavation of section 3, the main part (sections 4, 5, etc.) is excavated while reinforcing the rock walls using prestressed (PS) strands, if necessary. Where the condition of the rock mass is poor, working tunnels are first excavated in the arch. After shotcrete and rock bolts are set the arch is excavated; the excavation pattern of the arch depends on the geological conditions. The kinds of rock masses around the caverns are listed in Table 1. The Japanese archipelago has active crustal movements, and geological features are very complicated in the islands. In one site there can be a wide variety of rock kinds, and these rocks are affected by various stages of weathering. There are also many faults. Because of such geological complexity, a wide range in the elastic modulus and strength of the rock masses is a feature. The values of elastic modulus obtained from plate-bearing tests in the exploratory galleries were in the range of 10-20 GPa. The strengths (i.e. values of τ 0 in the failure envelope; τ/τ0 = (1 + σ/σ0) 2) were in the range of 1-2 MPa, obtained by rock-shearing tests at the sites. The initial ground pressures, measured by the stress relief method (overcoring method), were 5-10 MPa [1]. 22.2.2 Measurement Items Since the caverns are large in scale and geological conditions are not always good, the safety of the excavation work must be considered. Prior to excavation, geological surveys are conducted. As mentioned previously, however, crustal movements are active and geological structures are complicated. It is not a rare case where geological features abruptly change within a distance of 10 m. Thus,
Figure 1 Locations of the underground power stations (numbers correspond to the sites listed in Table 1)
Site
Kind of rock mass
1 Kisenyama 2 Niikappu 3 Okutataragi 4 Oohira 5 Nabara 6 Shintakase 7 Okuyoshino 8 Okuyahagi 9 Numazawa number 2 10 Tanbara 11 Arimine 12 Honkawa 13 Takami 14 Matano 15 Tenzan 16 Imaichi a
b
(1968) (1972) (1972) (1973) (1974) (1975) (1976) (1978) (1979) (1979) (1979) (1980) (1981) (1981) (1982) (1982)
Shale, sandstone, chert Schalstein Rhyolite, diabase Sandstone, slate Granite Granodiorite, diorite Shale, sandstone Granite Rhyolite Conglomerate Granite Black schist Schalstein Granite, porphyrite Granodiorite Sandstone, breccia
Size of cavern (m) Height Width Length 51.0 43.8 49.2 45.4 47.7 54.5 41.6 47.8 47.6 49.5 20.8 47.4 43.3 46.2 48.0 51.0
25.6 19.8 24.9 22.0 25.0 27.0 20.1 22.4 26.0 26.6 14.6 26.3 21.5 23.5 24.0 33.5
60.4 50.8 133.4 82.8 85.6 165 157.8 103.3 96.5 116.3 30 98 55 155.5 89.0 160.0
Size of arched lining* R/S Um) Um) 2.04 2.30 1.88 2.11 1.72 2.51 1.64 2.00 2.05 2.49 0.20 2.01 1.20 1.00 1.767 0.320 d
1.20 1.20 1.00 1.20 1.20 1.50 0.80 1.00 1.10 1.50 0.20 1.00 1.20 1.00 1.00 0.320
Refer to Figure 27. £ 0 = deformability, τ0 = cohesive strength, a = creep coefficient. "Depth of overburden. Anisotropy.
0.236 0.250 0.243 0.247 0.239 0.251 0.209 0.251 0.242 0.248
—
0.225 0.250 0.237 0.26
—
Ground pressure (MPa) τ σΛ h (mf *r 1.3 3.3 5.8 5.7 7.2 2.0 6.6 7.4 4.5 4.5 1.3 5.4 7.1 18.5 15.0 7.6
3.9 4.4 6.5 7.8 6.3 5.9 6.9 10.8 3.4 7.0 1.9 7.1 5.9 12.5 11.0 9.1
— 0.7
—
1.6 1.1 2.2 2.3 1.0 0.8 1.9
—
0.9 0.2 6.6 1.2 0
250 110 240 280 180 250 180 340 160 240 63 270 220 350 500 400
Properti es of rock imassb E0 (GPa) τ0 (MPa) 0L 6-12 24 3.5-10 10-29 3-9 14/7d 13/6 15/7 10 16-20 4/2 12/8 3-8 15-20 25 18
1.5-2.9 2.4 3.9^.9 1.0-2.5 0.5-1.5 3.1/1.3d 2.0/0.8 2.9/1.2 1.4 2.4-2.9 1.7/1.3 2.5/1.3 0.14 2.9 7.4 1.9
0.16 0.16 0.05 0.17 0.2 1.0 0.3 0.8 0.1 0.4 0.4 0.7 0.8 0.5 0.5 0.4
Rock Mass Behavior during Large-scale Cavern Excavation
Table 1 Outline of the Caverns and the Initial Conditions
633
Back Analysis Monitoring
634
Arch : 1 - 3 Main cavern : 4 , 5 , etc.
B: rock bolts L". concrete lining PS: prestressed reinforcement strands
Q
Draft tube side
Penstock side
10
m
20
30
Figure 2 Mushroom-shaped cavern and excavation steps
Arched concrete lining
i—i
4» » »
Reinforcement gauge
x
Concrete strain gauge
o
Thermometer ►
Figure 3 Layout of monitoring system
not only geological surveys but also various measurements for the assessment of safety, some of which are listed below, are indispensable. The main measurement items are as follows: (i) displacements in the rock mass; (ii) displacements between the walls of the caverns (convergence); (iii) strains and reinforcement stresses in lining concrete; and (iv) axial forces in the PS strands. Other measurements should be made of elastic wave velocity and permeability. Observation with a borehole TV camera can be useful. In Figure 3, an example of the layout of a monitoring system is shown. Rock displacement meters are installed at small intervals near the excavation walls, so that the sizes of relaxed zones can be detected. In order to improve the accuracy of the measurement, the concrete strain gauge is paired with the reinforcement gauge, and both of the meters are set in the arched concrete lining. The data from both gauges are cross-checked. In the case where faults are found near the caverns, rock displacement meters are set so that the movements of faults can be observed. Since the length of the cavern can be 100 m or more, measurements are carried out at several sections. An example of a rock displacement meter is displayed in Figure 4; however, the types of instruments to be used and their layout vary from site to site.
Rock Mass Behavior during Large-scale Cavern Excavation
635
Cement mortar
(Section A - A )
Figure 4 Extensometer
22.3 ROCK MASS BEHAVIOR AROUND CAVERNS DURING EXCAVATION 22.3.1 Subsidence Characteristics and Relaxed Zones of Ceiling Rocks An example of the variation in the subsidence in ceiling rocks caused by excavation is shown in Figure 5. This site corresponds to site number 7 in Table 1, and the rock mass mainly consists of sandstone and slate. In order to measure fully the subsidence of the rock mass during excavation, measuring instruments were set before the cavern excavation in the exploratory gallery located 50 m above the arch. What is noticeable is that almost all of the total subsidence was generated in the excavation of the arch, and the subsequent excavation of the main cavern did not increase the subsidence; on the contrary, there was a tendency toward decreasing subsidence. Major subsidences occurred particularly when the arch was excavated just below the area where the displacement meters were installed [2]. The behavior mentioned above can be explained as follows. The shape of the cavern in the excavation of the arch is horizontally long and vertically short. The relief of the vertical component of the ground pressure due to the excavation, therefore, is prominent, and ceiling rocks are pulled downwards. In the succeeding excavation of the main part of the cavern, the relief of the horizontal component of the ground pressure is large because the shape of the main cavern is vertically long and horizontally short. The walls of the main cavern, therefore, suffer deformation toward the center of the cavern, and both the ceiling rocks and the bottom rocks of the cavern are compressed in the horizontal direction, thereby causing the ceiling rock mass to suffer upward displacement contrary to subsidence. Both the measured and the calculated results are displayed in Figure 5. The forecast of rock subsidence by calculation agrees well with the trend of subsidence and with the measured amount of subsidence. Consequently, the forecast calculation method is proved to be valid. This forecast calculation is done prior to the excavation of the cavern. Through cross-checking between the results Excavation below the measuring location
Measurement E E
Calculation (with prestress) Calculation (without prestress)
Figure 5 History of the subsidence of the ceiling rocks (site 7)
636
Back Analysis Monitoring
of this forecast calculation and the results of measurements made during excavation, it is important to assess the safety of the work. The forecast calculation analysis method is described elsewhere [3]. Some distributions of subsidence in ceiling rocks after the completion of cavern excavation are displayed in Figure 6. The numbers beside each plot in the figure correspond to the site numbers listed in Table 1. What is characteristic in thisfigureis that almost all of the subsidence in each case occurred only in the rocks several meters above the arch, and little subsidence took place in the rocks much deeper than that. Therefore, the relaxed zone caused by the excavation in the rock can be estimated to be several meters thick (in the case of site number 13, the thickness of the zone was about 10 m because of the locally bad conditions in the ceiling rocks). Figure 7 shows the apparent strains obtained by dividing the amount of subsidence by the measuring length. In the neighborhood of the surface of the ceiling rock, a strain of 0.1-0.4% at the largest was obtained. The strains in the relaxed zones shown in Figure 6 were more than 0.1% or so; hence, there is the possibility that an apparent strain of more than 0.1% is an indicator of relaxation. 22.3.2 Rock Deformation Characteristics of Walls An example showing the changes in the relative horizontal displacements in the wall rocks with the progress of excavation is given in Figure 8. Rock displacement meters for these measurements can usually be installed only after the excavation of the main cavern has reached the desired
, Site number 50 H
40
i
£ 30 20
S
10
I 1SI
10 20
0
10
0
10
0
10
0
10 20 30 4 0
Relative subsidence of ceiling rock (mm) Figure 6
Distribution of the subsidence of the ceiling rocks
, Site number
13
50
~
E
40
10
30
E
13
12 14
20 h
o a> 10 h
L·.
I"
: 33 Γ25 =»40
013 Apparent strain of ceiling rock (x 10" ) Figure 7
Distribution of the apparent strains in the ceiling rocks
100
637
Rock Mass Behavior during Large-scale Cavern Excavation
installation depth. Measured values, therefore, refer only to changes in rock deformation due to excavation of the main cavern below the point at which these instruments are installed. (If there are drain galleries or exploratory galleries around the main cavern, and if the rock displacement meters can be installed beforehand through bore holes in the direction of the main cavern from these galleries, the total displacement owing to the full excavation of the main part of the cavern can be measured.) What is characteristic in Figure 8 is that the horizontal displacements of the cavern walls around the center (R-17 to R-19) were much larger than those higher up (R-10 to R-12), i.e. three to four times larger. There may be two reasons for this. Firstly, since rocks in the higher part of the cavern are located close to the arch their deformation is suppressed by the three-dimensional strengthening effect in the corners, while rocks in the central part experience a smaller surrounding restraint. Secondly, the whole cavern has a vertically long and horizontally narrow shape, which is mechanically unstable. If the cavern excavation was completed up to the main part (section 2), the ratio of the height to the width of the whole cavern would be close to one, thus making it mechanically stable. The amount of deformation would also be reduced. To excavate such a vertically narrow cavern it is therefore necessary to pay special attention to the stability of the cavern when the lower half is excavated. An example of the distribution of relative horizontal displacements at the final excavation stage is shown in Figure 9. In the case of relative subsidences in the ceiling rocks, what was characteristic was that subsidence was significant within the range of several meters from the rock surface, and in the rocks deeper than that little subsidence occurred. On the other hand, it is evident that horizontal displacement of the cavern walls occurred in relatively deep zones too. The measurements shown in Figure 9 refer to a case where the displacements were large compared with other measurements in Japan. In Figure 10, distributions of apparent strains obtained from the relative horizontal
g
R-12
10
R-ll
O 1978 Feb.
Jf
Mar.
Ό Φ
Apr.
1
Jul.
Jun.
May
Aug.
4
3
2
Sep.
Oct.
Excavation stage
5
Figure 8 Measured relative horizontal displacements obtained from extensometers (site 8) R-19
20 m
-^45.8
1Λ
L ψΐmL-
R-17
R-18
-( 35.4 —'145.4 45.4 mm
► 38.7 ^25.6
42.9 mm
50.0 mm
Underground cavern (plan) 28.5 mm
32.9 mm
18.2 mm
-V25.7 -j 20.1
■V 30.9 '23.9
i 18.2 ' 16.2
' 13.9
Ί8.0
I 0
R-10,11,12
-Λ40.3 H 45.0
i
h/5.1
I 20
«
4
+4\
+-*—►—►
-S*Î & R - I 7 , 18,19
I
w
Ix&>% ( Section )
L
m
Figure 9 Distribution of the measured horizontal displacements obtained from extensometers at thefinalexcavation stage (site 8)
638
Back Analysis Monitoring Penstock side Site number Draft tube side
6
37
<
^
15
7
'— 25
J6
.,
is
^ Extensometer
12
-^
13
r*1
25
30
L
20
14
10
L
L
23
^ - r - * · J
*-.
0
J
0
J
10
2_ I
20
L
30
Measuring length (m) Figure 10 Distribution of the apparent strains ( x 10" 4 ) in the walls
displacements at several sites are shown. There are some zones with strains of more than 0.1% at depths of 15-20 m, depending on the site, showing that it is possible that the wall rocks are more readily relaxed than are the ceiling rocks. 22.3.3 Stresses in Arched Concrete Linings and Their Distribution As described in Section 22.2.2, concrete strain meters and reinforcement meters are installed in the arched concrete linings. Figure 11 shows some measurements of stresses and temperatures in the concrete lining. It can be seen that the temperature of the concrete rose abruptly from about 20 °C (ambient temperature) to 50 °C shortly after it was put in place. This was due to the hydration of the concrete. The concrete subsequently cooled down with the lapse of time and soon returned to ambient temperature. The stress values indicated by the reinforcement meter obviously changed in proportion to the temperature of the concrete. Therefore, these values also included thermal stresses (temperature compensation type reinforcement meters were used). To show more clearly the relationship between temperature and stress in the reinforcement, the temperature change and one of the representative stress curves are shown in Figure 12. After placing the concrete in the arch it is usually about one month before the main cavern excavation is started. The temperature of the concrete returns to around ambient temperature during this period, and the seasonal variation of the temperature is rather small, about 10 °C. Accordingly, subsequent changes in reinforcement stresses can be regarded as being caused by the main cavern excavation. In considering the stability of arched concrete linings, stresses in the concrete are more important than stresses in the reinforcement. It is therefore necessary to estimate stresses in the concrete from the measured stresses in the reinforcement. The stress in the concrete can be determined from the equation ac = askEJEs (1) where ac and as are the stresses in the concrete lining and the reinforcement during the main cavern excavation, respectively; Ec and Es are moduli of elasticity of the concrete and the reinforcement,
Rock Mass Behavior during Large-scale Cavern Excavation
100 h
-\ 50
40
Q-
20
10
100 h
200 H
Nov.
Dec.
Jan.
Feb.
Mar.
Apr.
May
Jun.
Jul.
Aug. Sep.
Oct.
Excavation stage
Figure 11 Variation of the stress and the temperature in the concrete lining (site 8)
<^Ë^>
50 h 40 h
E 30
- Temperature ·
2Oè09>
S.
Δ σ ' = 6 9 (due to hydration)
50 h 100 h
*-
150 \200
g
I Nov. I Dec. I Jon. |Feb. |Mar. |Apr. [May | Jun. | Jul. JAug. | Sep. | Oct.
Excavation stage |
Figure 12 Variation of the stress and the temperature in the concrete lining (site 8)
639
640
Back Analysis Monitoring Location in arched concrete lining
Figure 13 Distribution of the stresses in the linings
respectively. The quantity k is a corrective coefficient for the shrinking and creeping of concrete and has the value 0.68. More details on equation (1) can be found in the literature [4]. In Figure 13 stress distributions from 11 sites are shown. Concrete lining stresses were usually measured in three or four, sometimes more, vertical sections of the cavern length at each site. The stresses are average values obtained from those measured by the reinforcement meters installed in the upper and lower parts of the concrete lining (refer to Figure 3) at all measuring sections. The stress distributions of each site show that the stresses at the crown are a little larger than those at the abutment, and it is evident that in igneous rocks the stresses tend to be more than twice as large as those in sedimentary rocks. As is clear from Table 1, in the sites consisting of igneous rocks the values of the initial ground pressure were larger than those in the sites consisting of sedimentary rocks by about 20%, which is a small difference compared to the ratio of over 2:1 for the lining stresses in such rocks. It can therefore be concluded that this behavior depends on the type of rock (refer to Section 22.3.4). The igneous rock sites in Table 1 are numbered 3, 5,6, 8,9,11,14 and 15, and the kinds of igneous rock studied were granite, granodiorite, rhyolite, diabase and diorite. These rocks usually have the characteristic of well-developed joints and are tentatively called hereafter 'jointy rock'. The sedimentary rock sites in Table 1 are numbered 1,2,4,7,10,13 and 16, and the kinds of sedimentary rock studied were shale, sandstone, schalstein, slate, conglomerate and breccia. These kinds of rock usually do not have such well-developed joints and are named 'nonjointy rock' accordingly. 22.3.4 Stresses in Arched Concrete Linings and Rock Deformation Figure 14 shows for several sites the relationship between the average stresses in the arched concrete lining (average of all measured stresses in that site) and the convergence of the cavern walls just under the arched lining abutments (or the sum of the displacements measured in both walls using rock displacement meters). It is evident that there is a strong correlation between the two measured values. This means that the stresses in the arched concrete lining occur because of horizontal compression in the lining as the wall rocks deform into the center of the cavern as the main cavern excavation progresses. The relationship between the stress in the arched concrete lining and the length of the cavern is displayed in Figure 15. This figure shows the following two characteristics. Firstly, the longer the cavern the larger the stresses in the lining. Secondly, the stresses in the concrete linings o£ jointy rocks are two to three times larger than those of nonjointy rocks. As for the face of a tunnel, there is the effect of three-dimensional restraint. At a distance of more than one tunnel diameter from the face, however, the three-dimensional effect of the face is small, and
Rock Mass Behavior during Large-scale Cavern Excavation £ 12
|-
O Jointy rock □
641
Site number
Nonjointy rock
£ 10
8 h o υ
6
■D 0)
S 4
0
10
20
30
40
Convergence between walls, I f (mm)
Figure 14
Relationship between stress and convergence
Figure 15 Relationship between the concrete stress and the cavern length
there is no major difference between treating the tunnel as a two-dimensional or three-dimensional structure. Thus, as the height of any one cavern is about 50 m, it can be supposed that the stresses in the linings may converge if the length of the cavern reaches more than 100 m. As shown in Figure 15, however, the stresses increase with the length of the cavern. The above-mentioned features reveal some very interesting kinds of behavior in rocks, and will be discussed again together with the results of other measurements in Section 22.4. 22.3.5
Variation of Elastic Wave Velocity
In the Shintakase underground power station (site number 6), measurements of elastic wave velocity, observations with a borehole TV camera and permeability tests were carried out in addition to the measurements of rock displacements and stresses in linings mentioned previously. The results of these measurements and observations are described in Sections 22.3.5 to 22.3.7. The main rock mass at this site is granodiorite. The elastic wave velocity measurements were conducted by Honsho and Motojima [5]. As shown in Figure 16, 12 boreholes were drilled downwards around the cavern from the exploratory gallery located at an elevation of 1054 m. Pick-ups were installed in nine of the
642
Back Analysis Monitoring \
y
\
\
V
>
\\
/ \
y \
\ Λ
\
VN
21
876\ 5 4 3
o Borehole for pick up (9) \ x Borehole for shot point (3) \ Δ Shot point inside power statton\ cavern \
\\
\
\
I
\
0
i
^
i
i
i
I
50
\
V
Figure 16 Locations of the boreholes for the elastic wave measurement (plan, elevation 1054 m) (reproduced from ref. 5 by permission of S. Honsho)
boreholes. Detonators (one to three) or 10-20 g of dynamite were set off in the blasting holes and the travel times of the elastic waves were measured. Figure 17 shows the initial elastic wave velocities before the excavation of the cavern. The average value Vpo = 4.53 kms" 1 was obtained from these results. There were two predominant joint systems in the rocks and they had the same concentration degree of 6.2%. The rock mass was considered to be macroscopically homogeneous, although there were slight differences in the average velocities in different directions. The elastic wave velocity measurements after the excavation of the arch showed no significant change in the velocity in the cavern walls. In the ceiling rocks, however, a remarkable decrease in velocity (maximum 29%) occurred. There were changes in velocity of 7% in the diagonally right, downward direction and 16% in the diagonally left, downward direction (almost orthogonal to the
I Borehole number 0
974 x'
Figure 17 Initial values of the elastic wave velocity (km s l) (reproduced from ref. 5 by permission of S. Honsho)
643
Rock Mass Behavior during Large-scale Cavern Excavation
predominant joint plane), indicating that the predominant joint had a great influence on the changes in the elastic wave velocity. During the excavation of the main part of the cavern, a decrease of up to 30% in the elastic wave velocity was seen in the cavern walls. Figure 18 shows the percentage changes when the cavern excavation was complete. The negative sign refers to a decrease. The velocity changes were apparent even at depths of 20-30 m from the cavern wall surface. In Figure 19 changes in the average elastic wave velocities measured at each elevation in the ceiling rocks are shown. What is noticeable is that when the arch was excavated there was a decrease in velocity of up to 20%; however, the velocity then increased considerably during the main cavern excavation. This trend agrees quite well with that discussed earlier for the subsidence of ceiling rocks (Figure 5).
2 Borehole number
ΐ 0
-1-5.ZJ
-
1-16.4;—*<
BoreholeJI'^
s-
" Borehole 5'
Power station
6.7jj
—
x
—
Shot point Pick-up
974 *:
Figure 18 Variation of the elastic wave velocity due to the excavation (reproduced from ref. 5 by permission of S. Honsho)
*^^
ω
-5
^ \
\ Elevât ion (m) 1049
\
A
/y^-j\
\io44\
/
/
\ *s
' S*^' —
/ s.y Ss
«-
-10
Ό - 15 h 0) "o c o
-20
\I039
.
**
,
N__/
---~"rr
-^^^
__y
1034
Arch excavation
Main cavern excavation
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
[
1973
1974
1975
|
1976
|
Figure 19 Variation of the elastic wave velocity above the arched concrete lining due to the excavation (reproduced from ref. 5 by permission of S. Honsho)
644
Back Analysis Monitoring
Changes in the elastic wave velocities in the cavern walls are shown in Figure 20. It is evident that the velocities decrease almost monotonously with the progress of the excavation. This trend corresponds with that for the displacements of cavern walls (Figure 8). Measurements of elastic wave velocity were continued for more than one year after the cavern excavation was completed. Looking at the changes during this period, as shown in Figures 19 and 20, there is a trend showing that the velocities decrease at first and then increase during later stages of the excavation. It is possible that rocks which have relaxed due to excavation are tightened again with the lapse of time. 22.3.6 Joint Opening in the Rock Mass A compact borehole TV camera (Figure 21) enables us to observe the surfaces of boreholes and to record the locations, strikes and openings of joints. For the purposes of observation with the borehole TV camera two holes (numbers 1 and 2), each having a diameter of 76 mm and a length of 25 m (measuring range 22 m), were drilled 7 m away from each other in parallel at an angle of 20.5° downward from the drain gallery toward the wall of the cavern to be excavated (Figure 22). The measurements were carried out by Hori and Miyakoshi Elevation lm> 974
? >»
0
o o
i > 5 o w o 0)
O
'^/ιο&'
X
/
-10
Total
' ' ^ y y^
-15
c o -20 σ
>
984
-5
1014 994^
\ ' Arch excavation
Main cavern excavation
^—-Î22?
^
1 1 1 1 1 1 1 1 1 11 11 1 1 1 1 1 1 1 1 1 1 11 1 1 M 1 11 1 1 1 1 1 1 1 1 1 1 1 1 1
1973
1974
1975
|
1976
|
Figure 20 Variation of the elastic wave velocity in the cavern walls due to the excavation (reproduced from ref. 5 by permission of S. Honsho) Monitoring unit
Figure 21 Borehole TV camera system
Rock Mass Behavior during Large-scale Cavern Excavation
645
[6]. The first measurement was conducted when the excavation of the main part of the cavern was complete to a depth of 1010 m, then three more measurements were performed until the total cavern excavation was complete. The results of the first observation showed that there were 11 joints in hole number 1 and seven joints in hole number 2 in the 22 m measuring range, and the average joint density in each hole was 0.5 joint m _ 1 and 0.3 joint m _ 1 , respectively. When the excavation was complete the number of joints in each hole had increased to 55 (hole number 1) and 45 (hole number 2), thus showing an increase in joint density of about 2 joints m" x . The variations in joint aperture during the excavation are listed in Table 2. The accumulated apertures during the excavation were 27 mm for hole number 1 and 21 mm for hole number 2. In this measurement two kinds of joint opening were observed: 'open joints' (11 open joints in hole number 1), the existence of which could be confirmed through the first observation with the TV camera, and 'micro joints' (44 micro joints in hole number 1), which could not be identified atfirst.It is apparent that the accumulated aperture in the case of the micro joints is far larger than that for the open joints, and 70-90% of the total aperture is due to the apertures of the micro joints. The convergence measurement c, performed near the boreholes for the TV camera observation (Figure 22), was 74 mm after the completion of the excavation. Now, supposing that half of the amount of convergence is equal to the displacement in the wall rocks, this accounts for a displacement of 37 mm. The accumulated joint aperture of 21-27 mm means that nearly 65% of the rock mass displacement is caused by joint opening-the effect of the joints on rock displacement is therefore quite remarkable. Observations were also made in a site consisting of sandstone and shale (site 7, Table 1). At this site three boreholes (each having a length of about 20 m) were drilled for observation with a borehole TV camera. The observations were carried out by Miyakoshi and Kakuta [7]. The numbers of open joints observed at the initial stage of the excavation were 18 (hole number 1), 20 (hole number 2) and 29 (hole number 3). The total number of joints which opened in each hole during the cavern excavation was: three, with an accumulated joint aperture of 3.2 mm, in hole number 1; six, with an accumulated joint aperture of 5.6 mm, in hole number 2; and seven, with an accumulated joint aperture of 4.7 mm, in hole number 3. When compared with the aforementioned site consisting of granodiorite, it is seen that for the sandstone/shale site the numbers of joints which opened and the accumulated aperture are both small.
1003 m -
- — 974 m
Figure 22
Table 2
Shintakase underground cavern and the locations of measurements
Variation of the Joint Aperture due to the Excavation (reproduced from ref. 6 with permission of Y. Hori) Hole number 1 Accumulated Ratio aperture (mm) (%)
Open joint Micro joint 3 Total a
7.63 19.37 27.00
28.2 71.8 100.0
Joint with an aperture of less than 0.25 mm at the initial stage.
Hole number 2 Accumulated Ratio aperture (mm) (%) 2.50 18.37 20.87
12.0 88.0 100.0
646
Back Analysis Monitoring
The measured displacement of the wall rocks was about 20 mm. The ratio of the joint aperture to the total displacement of the rock mass was nearly 25%, rather small compared with the ratio of nearly 65% obtained in the granodiorite site. It may not be appropriate to draw a conclusion from only these two examples; however, it is conceivable that in igneous rock such as granodiorite the deformation and relaxation of the rock are greatly influenced by the opening of joints, while in sedimentary rock such as sandstone the deformation caused by the strain change in the rock is more influential than the discontinuous displacement effected by joint opening. 22.3.7 Variation of Permeability Measurements of permeability were performed by Motojima [8] in the same holes used for observation with the borehole TV camera mentioned previously (Figure 22) and using the Lugeon test method (Figure 23). The length of the packer was 1 m and the measuring section was 2 m long. The water pressure was low (less than 1 kg cm"2) so as not to damage the rocks in the neighborhood of the boreholes. The first measurement was performed after the main part of the cavern was excavated down to 1010.5 m, and six more measurements were taken until the cavern excavation was complete.
mmtm
I
* * Upper packer Pressure gauge Thermocoupler
Lower packer
Figure 23 Measurement of permeability Borehole number
I
Borehole number 2 Initial measurement
60
10
FJTh
1
n
1
-
I
I 1 . j O.I
Measurement after excavation
1 1 1 J
l
_1 10
Ur "LrJ .j 20
Drain gallery
Drain gallery 0
10
Distance from the cavern wall
20
(m)
Figure 24 Variation of the Lugeon value distributions due to the excavation (reproduced from ref. 8 by permission of I. Motojima)
Rock Mass Behavior during Large-scale Cavern Excavation
647
In Figure 24 the distributions of the Lugeon values at the first measurement and at the completion of the cavern excavation are shown. In the first measurement the average Lugeon values for the two boreholes were 0.79 Lu (hole number 1) and 0.23 Lu (hole number 2). The initial Lugeon values were large in the neighborhood of the drain gallery, probably because of excavation in that area. After the cavern excavation was complete, the average Lugeon values for the two holes increased to 28.8 Lu (hole number 1) and 18.1 Lu (hole number 2), which are respectively 36 and 78 times higher than the initial values. In the neighborhood of the cavern wall the initial Lugeon values of 0.03 Lu and 0.1 Lu increased to values several hundred times larger, 30 Lu and 55 Lu, after the excavation was complete. The above-mentioned variations in the Lugeon values can be explained by consideration of the process of joint opening during the excavation, described in Section 22.3.6.
22.4
CHARACTERISTIC FEATURES OF ROCK BEHAVIOR
From the results of the various kinds of measurement described in Sections 22.3.1 to 22.3.7, the following conclusions can be drawn as characteristic features of rock behavior. (i) The rock mass deforms under external forces. The deformation roughly consists of two types of displacement: one is a discrete type of displacement caused by the opening of joints, tentatively called 'opening displacement', and the other is a continuous type of displacement caused by changes in strains, tentatively called 'strain displacement'. The generation of Opening displacement' is evident through the borehole TV camera, mentioned earlier. Both the variations in permeability and elastic wave velocity can be attributed to this type of displacement. The relative rock displacement shown in Figure 9 was 45.8 mm between the points A and B (separated by 8 m). The resulting strain, a relatively huge value of 5.7 x 10" 3 , could not be generated without Opening displacement'. (ii) Opening displacement' occurs during relaxation of the rock mass, while 'strain displacement' is generated at both the elastic and relaxed states of the rock. Rock masses may be classified into 'easy to relax' and 'difficult to relax' according to origin. In the case of jointy rock, such as igneous rock, the 'opening displacement' is predominant and causes the rocks to relax easily and also produces a large amount of displacement. On the other hand, in the case of nonjointy rock, such as sedimentary rock, the rock mass can be classified as 'difficult to relax', the Opening displacement' is small and the 'strain displacement' constitutes the greater part of the total displacement. Figure 15 shows results typical of the phenomena mentioned above; the stresses in jointy rocks are two to three times higher than those in nonjointy rocks. The horizontal component of the ground pressure ah is one of the important factors affecting the stresses; however, there is no evident relationship between oc and a h , as shown in Figure 14. (iii) In the case of jointy rock, it seems that the larger the scale of the excavation surface the larger becomes the Opening displacement', and the ratio of Opening displacement' to 'strain displacement' increases. In excavating a 'usual' tunnel this ratio is usually small compared to those obtained in the large-scale excavations of caverns treated in this chapter. (iv) 'Opening displacement', which causes discontinuous planes in the rock, leads to unstable rock structures. Rock structures in nonjointy rock, therefore, offer superior stability to those in jointy rock. (v) Among the many factors affecting the stability of the cavern, such as ground pressure, mechanical properties of the rock mass and others, the shape of the cavern plays a very important role. The caverns discussed in this chapter are tall and narrow, with a ratio of height to width of about two to one. Through the measurements performed on these caverns during excavation, the following characteristics were revealed. (a) The subsidence of the ceiling rock occurs only during the excavation of the arch, and no increase in subsidence occurs during the subsequent excavation of the main part of the cavern. The subsidence occurs mainly within the region of several meters from the surface of the ceiling rocks. Accordingly, the relaxed zones in the ceiling rocks are estimated to be several meters in extent, and it can be considered that the relaxation occurs during the excavation of the arch and that the relaxed zones increase little during the subsequent excavation of the main part of the cavern. (b) The walls of the cavern are deformed and relaxed during the excavation of the main part. The depths of the relaxed zones in the cavern walls are apt to be greater than those of the relaxed zones in the ceiling rocks. The depth of the main cavern excavation is about 50 m, about twice the width of the cavern, and the area of the excavation surface is therefore large, thereby facilitating 'opening displacement' and relaxation.
648
Back Analysis Monitoring
(c) During the excavation of the main part of the cavern the wall rocks deform inwards, and consequently stresses are produced in the arched concrete lining. The stresses are in direct proportionality to the horizontal displacements of the walls; that is, horizontal displacements of the walls produce stresses in the arched concrete lining. In designing the arched concrete lining, therefore, it is necessary to take the following forces into consideration: (i) external forces Fy due to the weight of the relaxed zone above the ceiling rocks. The zone is formed during the excavation of the arch and is several meters in depth; and (ii) external forces Fh due to horizontal displacements of the wall rocks, produced through relief of horizontal ground pressure in the walls during the excavation of the main part of the cavern. The forces Fh are usually far larger than the forces Fv. 22.5 DESIGN OF ARCHED CONCRETE LININGS FOR CAVERNS 22.5.1 Role of the Arched Concrete Lining The role of the arched concrete lining is to maintain the stability of the cavern, and it is necessary that the lining be constructed so as to work effectively in supporting external forces. The magnitudes of the external forces and the working mechanism vary according to whether the cavern is located near the ground surface or in a deep location. For a cavern in a shallow location (Figure 25), there is the possibility that the ceiling rocks can relax in the region up to the ground surface if the rock quality is poor. In such a case, a large external force Fv works vertically on the arched lining during excavation. On the other hand, in the case of a large-scale cavern in a deep location (Figure 26), the external forces Fh are far larger than the external forces F v , as mentioned in Section 22.4. Thus, the design of the arched concrete lining varies with the nature of these external forces. So, the arched lining in a shallow location is designed mainly to support the vertical external force Fv, and hence the curvature of the lining is comparatively large. The thickness of the arch abutment is greater than that of the crown so that the arch thrust force is transmitted evenly to the surrounding rock. On the other hand, the arched lining in a deeply located cavern is designed mainly to support the horizontal external forces Fh. Here a smaller curvature in the arched lining is acceptable, and there is no need to make the thickness of the arch abutment greater than that of the crown, i.e. an arched lining of constant thickness is conceivable. As the external forces Fy in this case are comparatively small, the arched lining may be relatively thin. Because of the above considerations, the shapes of arched linings designed for underground power stations in Japan have changed with the lapse of time, as shown in Table 1. The rise to span (R/S in Figure 27) ratios of the arched linings were originally designed at 0.25-0.23, but came to be designed at 0.209 at site 7. The thickness of the arch crown was conventionally about 1.2 m, but it was later designed at 0.8 m (site 7) or with a constant thickness at both the abutment and the crown (sites 13 and 14). (The designs were not always changed chronologically in Table 1, due to special conditions at each site.)
Figure 25 Forces experienced by a cavern located near a ground surface
Rock Mass Behavior during Large-scale Cavern Excavation
649
Relaxed zone
Release of initial stress
Power station cavern Figure 26 Forces experienced by a cavern located at great depth
Figure 27 Shape of an arched concrete lining
Figure 28 Lining by precast, reinforced concrete segments (PCRCS)
22.5.2 Design of the Arched Lining Considering the arched lining of a cavern in a deep location, the external forces Fy must be supported by the lining. As far as the external forces Fh are concerned, however, it is desirable to use a lining made with a contractile structure or soft material. This is because the forces Fh are produced by displacements of the wall rocks, and the stresses in the lining are proportional to its stiffness.
650
Back Analysis Monitoring
Under these conditions the lining structure shown in Figure 28 will be effective. This lining consists of PS strands and polygonal, precast, reinforced concrete segments (PCRCS). The PCRCS are similar to the shield segments used in shield tunnels. Hard rubber is attached around the periphery of each segment. The PCRCS arefixedonto the ceiling rocks using PS strands. If the rock surface is not smooth, ultrafast-setting cement isfilledbetween the surface and the PCRCS to provide a close contact. The vertical external forces Fv are supported by setting thefixingends of the strands in the rocks to greater depths than those of the relaxed zones. The horizontal external forces Fh can be sustained by deformation of the hard rubber attached around the PCRCS. The method using shotcrete and PS strands is also effective as it is currently applied. In the case of this method, however, the lining is constructed so as to support the external forces Fh directly, thereby causing cracks in the shotcrete under excessive force. In the case where it is difficult to form a smooth surface owing to joint systems or others, stress concentration occurs in the shotcrete lining and a local fracture results. The lining with PCRCS is also effective under these conditions. 22.6 CONCLUSIONS Many large-scale caverns were excavated in Japan for underground, pumped storage power stations in the 1970s and 1980s. Various kinds of rock tests and geological surveys were carried out and analyses for the forecasting of rock behavior during excavation were performed on the basis of these data. Furthermore, various kinds of measurement were taken during the excavation, and the construction works were carried out safely by securing the stability of the caverns through comparison of the results of the forecast analyses with the actual measurements. Some characteristics of rock behavior have been clarified through the results obtained at these sites. The major results are as follows. (i) Rock deformation consists of Opening displacement' and 'strain displacement'. In the case of jointy (e.g. igneous) rock the opening of joints causes the opening displacement to increase, forming relaxed zones. In the case of nonjointy (e.g. sedimentary) rock, on the other hand, the opening of joints is less pronounced and the rock mass does not relax so much. In the construction of a largescale cavern, therefore, nonjointy rock is superior in stability to jointy rock for samples of each type of rock having the same strength and deformability. (ii) In the excavation of a large-scale cavern with a height to width ratio of about two to one, the characteristic behavioral features of the rock are as follows. (a) Ceiling rocks relax for several meters in depth and subside due to the excavation of the arch. However, the relaxed zones above the arched lining do not increase in extent during the subsequent excavation of the main part of the cavern and subsidence proceeds no further. During the main excavation the wall rocks relax and deform towards the center of the cavern. Here, the relaxed zones are larger than those above the arched lining. (b) The stresses in the arched concrete lining increase in proportion to the horizontal displacement of the cavern walls during the main excavation. The horizontal displacement of the cavern walls varies greatly from one type of rock to another. Jointy (e.g. igneous) rock suffers deformation more readily than does nonjointy (e.g. sedimentary) rock. Stresses in the lining of jointy rock tend to be two to three times higher than those in the lining of nonjointy rock. (iii) The mechanism of the external forces acting on the arched concrete lining varies with the depth at which the cavern is located. For caverns at great depth, the vertical external forces are not so large, whereas the horizontal external forces are very large. In this instance it is desirable to avoid such a lining that would receive the horizontal external forces directly, and it is therefore advisable to adopt a lining formed from segments and PS strands.
22.7 REFERENCES 1. Kanagawa T., Hibino S., Ishida T., Hayashi M. and Kitahara Y. In situ stress measurements in the Japanese islands. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 23, 29-39 (1986). 2. Hibino S., Motojima M. and Kanagawa T. Behaviour of rocks around large caverns during excavation. In Proc. 5th Congr. Int. Soc. Rock Mech., Melbourne, vol. 2, pp. D199-D202. Balkema, Rotterdam (1983). 3. Hayashi M. and Hibino S. Visco-plastic analysis on progressive relaxation of underground excavation works. In Proc. 2nd Congr. Int. Soc. Rock Mech., Belgrade, pp. 565-576 (1970). Hayashi M., Kitahara Y. and Hibino S. Time-dependent stress analysis in underground structure in viscoplastic rock masses. In Proc. Int. Symp. Determination of Stresses in Rock Masses, Lisbon, pp. 145-156 (1969).
Rock Mass Behavior during Large-scale Cavern Excavation 4.
651
Hibino S., Hayashi M. and Motojima M. Behavior of anisotropic rock masses around large underground cavity during excavation works. Central Research Institute of Electric Power Industry, Report No. 379028, pp. 50-60 (1980). 5. Honsho S. and Motojima I. Velocity change during underground excavation at Shintakasegawa power station. Central Research Institute of Electric Power Industry, Report No. 379003, pp. 2-10 (1979). 6. Hori Y. and Miyakoshi K. Relaxation of rock masses during underground excavation at Shintakasegawa power station - observation by bore-hole TV. Central Research Institute of Electric Power Industry, Report No. 376528, pp. 8-12 (1977). 7. Miyakoshi K. and Kakuta T. Relaxation of rock masses during underground excavation at Okuyoshino power station - observation by bore-hole TV. Central Research Institute of Electric Power Industry, Report No. 377533, pp. 2-11 (1978). 8. Motojima I. Study on the permeability change of rock mass due to underground excavation. Central Research Institute of Electric Power Industry, Report No. 379009, pp. 11-16 (1979).
23 Predictive Calculation and Monitoring of Rock Stress and Displacement Induced by Ore Extraction YOSHIAKI MIZUTA Yamaguchi University, übe, Japan
23.1
INTRODUCTION
653
23.2
SITES, MINING PROCESS AND MINING PLAN
654
23.3
METHODS AND INSTRUMENTS
654
23.4 RELIABILITY O F THE STRESS DETERMINATION 23.4.1 Conventional Procedure for Three-dimensional Stress Determination by Hydraulic Fracturing 23.4.2 The Points at Issue in Three-dimensional Stress Determination by both Hydraulic and Double Fracturings 23.4.3 The Double-fracturing Technique and its Controversial Points 23.4.4 Results of Stress Measurements from Five Different Methods
658 658 658 659 661
23.5 MODELINGS O F BOUNDARY ELEMENT ANALYSES
662
23.5.1 23.5.2 23.5.3 23.5.4 23.5.5
DDM Modeling of Tabular Orebody Extraction DDM Modeling of Multiple-layer Mining DDM Modeling of the Interaction between the Pressure Capsule and the Surrounding Rock DDM Modeling of the Interaction between the Lune-shaped Flatjack and the Surrounding Rock Coupled FSM-DDM Modeling of the Behavior of Rock around a Pressurized Sleeve
662 662 662 662 663
23.6
MEASURED RESULTS
665
23.7
PREDICTED ROOF SINKAGE
666
23.8
DISCUSSION
668
23.9
REFERENCES
670
23.1
INTRODUCTION
In situ measurement of the rock deformations and stresses induced by the mining process have been or are being carried out in two mines. In the Yanahara mine, which is an underground iron mine, the induced displacements in the roof rock have been measured using connected vessel leveling systems and the rock stresses have been measured by means of flatjack and hydraulic fracturing techniques. The stress changes have been derived from pressure changes in pressure capsules. The measurements have been carried out over an extended period of time and displacement and stress change measurements are still being carried out. Besides the measurements, the induced roof displacements were predicted by numerical analysis using the boundary element method to form a comparison with the measured roof sinkages. In the Kokura mine, which is an underground limestone mine recently developed, the stress changes in the rock around the primary opening, induced by excavation of the secondary opening which is above or underneath the primary opening, are being monitored by means of pressure
653
654
Predictive Calculation and Monitoring of Rock Stress
capsules. The induced floor displacements are being measured using the LASDIS system (a displacement meter coupling a laser beam and two solar batteries). The initial rock stress and the tangential stress at the primary opening were measured, respectively, by the sleeve-fracturing method and the stress compensation method (using a lune-shaped flatjack). This chapter describes the outlines of the excavation processes, the measurement systems and the measurements obtained in relation to the progress of the excavation. 23.2
SITES, MINING PROCESS AND MINING PLAN
The areas in the lower deposit (370 m deep on average) of the Yanahara mine which were backfilled or newly mined up to the first half of 1990 are illustrated in Figures 1(a) and 1(b). The positions of the instruments for the four kinds of measurement are also shown in the figure. The initial rock stresses were measured twice at HI. Roof sinkage measurements were carried out over many rows and columns on various levels over a long time. Only the series of measuring points V l - V l l is shown in Figure 1(a). Wall stress measurements using lune-shaped flatjacks were carried out around point F. Rock stress measurements by way of hydraulic fracturing were carried out four times at H2 and three times at H3, and stress variation measurements by means of pressure capsules were carried out around point R and at points P1-P10. Recently, the big pillar along row 21, located in the center of the deposit, was partially extracted. Six extracted blocks are shown in Figure 1(b). Five of these blocks were backfilled; the sixth was mined upwards, effecting almost complete extraction of the block. The mining process up to 1989 and the mining plan employed subsequently are illustrated in Figure 2. The divided segments in the figure are related to the boundary elements used in the analyses for the prediction of the variation in roof sinkage (see Section 23.5). The mining plan for the Kokura mine is illustrated in Figure 3. The mined area at the 140 m level and the other three levels above and below triis, which are being mined or are about to be mined, are shown in the figure. The initial stress components were measured only around the main drift at the 140 m level. In the initial stress measurements using the sleeve-fracturing method a horizontal hole 15 m long with a radius of 104 mm was drilled, but the stress determination was carried out at point D, at a depth of 3 m below the surface of the wall. Overcoring using a diamond bit of 450 mm inner diameter and 485 mm outer diameter was also performed. The overcore was cut into slices in order to observe directly the primary and secondary fractures. The pressure capsules were inserted not far below the surface in the pillar and in the rock around the main drift, as shown in the figure. A laser beam was fixed at LA. The sensor containing two pairs of solar batteries, which detects the two displacement components (vertical and horizontal), was set at SB. Wall stress measurements using a lune-shaped flatjack were carried out at F v for the vertical component and at F H for the horizontal component. 23.3 METHODS AND INSTRUMENTS The methods employed in the measurements are as follows. (i) Initial stress measurement (a) Three-dimensional stress measurements were made using the hydraulic fracturing method (Y). (b) Two-dimensional stress measurements were made using the sleeve-fracturing method (K). (ii) Displacement measurement (a) Roof sinkage measurements were made using the connected vessel leveling method (Y). (b) Floor displacement measurements were recorded using the coupled laser beam and solar battery system (K). (Hi) Wall stress measurement (a) Stress compensation measurements were made using a lune-shaped flatjack (Y and K). (iv) Stress change measurement (a) Three-dimensional measurements were made using the hydraulic fracturing method (Y). (b) Apparent stress change measurements were made using pressure capsules (Y and K).
655
Back Analysis Monitoring (α) C9 section Level 21
R 2 , sec tion
370 m below surface
100
J
EP1
Ore
ΓχΠ
Backfill
ESS3
Dike
H
Hydrofracturing
F
Flatjack
P R V
Pressure capsule Recoverable pressure capsule Connected vessel system
(b) L27 section in R2I pillar, before 1987
£^IT""1 R2I section enlarged L24
L25
—
\l\ "
ΛΛ
SL26
—
SL27
—
L27
—
1%
L28
SL29 L29
Figure 1 (a) Outline of the lower deposit in the Yanahara mine, (b) Detail of mining progress to date in the R21 pillar
Predictive Calculation and Monitoring of Rock Stress
656
Coming extraction Under extraction 7/1987-12/1989 9/1983-6/1987 I/I98I-8/1983
mmmÊmmmmm
1/1978-12/1980 -12/1977
CIO Figure 2
Mining progress up to 1989 and mining plan after 1990 for the lower deposit in the Yanahara mine
LI55 section
ΕΠΠΒ Mined to date after measuring system set up D: double fracturing LA: laser source SB: solar battery sensor P: recoverable pressure capsule F: flatjack SH: shaft »ΧνΛνΉ l·"""""··^ Llbb—L D FP P LI40— 1 · · · · • SB · || LA J LI25—!_ __E ■^ζ^" LI 10 — I
i;i:i;1; ..ι.ι.ι^ΗΐΜΠίΙΐΙΐ SH
sS
t!;:': :·::·::::·::·::::·::::·::::·::::·::·:·::^
Figure 3
Mining progress to date and mining plan for the future in the east block of the Kokura mine
The letters in parentheses represent where the measurements were carried out: Y and K represent the Yanahara mine and the Kokura mine, respectively. The procedures for stress determination by hydraulic fracturing and sleeve fracturing are both described in the literature [1-5]. However, the sleeve-fracturing method is sometimes called the diametral deformation method [3] and the double-fracturing method [4].
Back Analysis Monitoring
657
In wall stress measurements by the stress compensation method, a lune-shaped slot was cut in the rock using a diamond-edged wheelsaw and a lune-shaped flatjack was put into the slot to achieve direct contact with the rock. Figure 4 shows the displacement distribution of the rock surface induced by the lune-shaped slot. The distribution was computed numerically by a three-dimensional, elastic analysis in which the displacement discontinuity method (DDM) was employed. In this type of measurement, the distance between point C and a steel bar connected with the rock at the points A and B is monitored to find the oil pressure which balances the rock stress component perpendicular to the slot. The advantage of this procedure is that the distance variation can be monitored during both slotting and pressurizing, although, as can be seen from Figure 4, the distance change detected is about 40% ofthat which can be measured conventionally, i.e. the change in the distance between C and its symmetrical point C with respect to the slot. The conventional pressure capsule is composed of a long, rectangular flatjack with metal fittings as the sensor, and for the purposes of measurement it is buried with cement mortar in a drill hole. Such pressure capsules were used in positions P1-P10 shown in Figure 1. However, around the point R in Figure 1 and in all the measuring points in the Kokura mine another type of pressure capsule was used [6]. Such a capsule is simply put into the drill hole with the metal fittings to achieve direct contact with the rock. This type of pressure capsule is therefore recoverable. The principle of displacement measurement by the laser beam and solar battery system is shown in Figure 5. The two solar batteries are fixed in a plate on the floor and the laser spot hits the plate in
7.5 cm
Figure 4 The calculated distribution of rock surface displacement induced by a lune-shaped slot Displacement
Solar battery A
^
^
(■"""
^ïier
)
Solar battery B
Servomotor Figure 5 Schematic diagram of displacement measurement by the LASDIS system
658
Predictive Calculation and Monitoring of Rock Stress
the center between the batteries. When the plate moves relative to the laser spot, a difference between the outputs of the two batteries occurs and a servomotor moves the plate so that the output difference is compensated. This movement corresponds to a displacement component of the floor. This system, named LASDIS, was developed through the cooperation between Nishimatsu Construction Co. Ltd. and Takara Denken Ltd. 23.4 RELIABILITY OF THE STRESS DETERMINATION 23.4.1 Conventional Procedure for Three-dimensional Stress Determination by Hydraulic Fracturing Figure 6 is a typical pressure-time record obtained from a hydraulic fracturing test. In the case of either a longitudinal or a transverse fracture the normal stress component ση, perpendicular to the fracture plane, is determined by the instantaneous shut-in pressure Ps =
(1)
P.
The secondary breakdown pressure P sb in the case of a longitudinal fracture is put into the following equation + 3
(2)
where axi and ayi are, respectively, the normal stress components parallel and perpendicular to the fracture plane produced in the borehole of ith inclination. If the fracture is confirmed as new, i.e. artificially produced, then the fracture direction may be determined by the stress state; thus *Xyi = 0
(3)
where xxyi is the shear stress component parallel or perpendicular to the fracture plane. Hence, if hydraulic fracturing tests are carried out in three or more boreholes, each with a different inclination, then the three-dimensional stress state can be determined from equations (1) to (3). 23.4.2 The Points at Issue in Three-dimensional Stress Determination by both Hydraulic and Double Fracturings As Figure 7 shows, in the case where the principal stress direction is inclined to the borehole axis, the angle 0M at which the maximum tangential stress σΜ occurs is different from the angle 0m at which the maximum hoop stress am occurs. It is well known that the fracture produced in such a case is apparently longitudinal, even though the direction of fracture initiation is inclined at an angle yM to
60
120
240
300
420
480
540
Time (s)
Figure 6
A pressure-time record (longitudinal fracture) obtained from hydraulic fracturing stress measurements in the Mozumi mine
Back Analysis Monitoring
659
Figure 7 Diagram of the difference in the angles at which the maximum hoop stress am and the maximum tangential stress σΜ occur
the vertical and thus the fracture is not a straight line. The difference in the angles 0M and 0m is great when the maximum difference between the magnitudes of the principal stresses is large and when the tensile strength of the rock is low. Consequently, Sakuma et al. developed a procedure for the strict back-calculation of the three-dimensional stress state, taking the angular position of σΜ and the inclination yM into consideration [7]. Table 1(a) shows two kinds of test data: one is a case where the differences between the principal stresses are small and the other is a case where these differences are large. The inclinations of the four boreholes are assumed as the cosines of the angles between any two borehole axes are relatively large. If the back-calculation is carried out by the procedure developed by Sakuma et a/., the stress states coincide with those assumed; however, some error occurs if the conventional method of calculation described earlier is employed. Table 1(b) shows the stress states calculated by the conventional procedure, where two values for the tensile strength of the rock are assumed. It can be seen from the table that the error in the calculated result becomes greater as the differences between the principal stresses increase and as the tensile strength of the rock decreases. This problem also occurs in the three-dimensional stress determination by the double-fracturing method. 23.4.3 The Double-fracturing Technique and its Controversial Points In double fracturing, in contrast to hydraulic fracturing, the fracture plane does not propagate at the point of fracture initiation, since no liquid is allowed to penetrate directly into the fracture. Instead, the fracture length is controlled by a balance between the hydraulic loading and the confining stresses of the ground. Furthermore, loading is controlled by constant flow into the sleeve and the pressure-time record is ideally a straight line, even after fracturing. Hence, four diametral deformations of the borehole boundary are measured through the total process of internal loading and unloading. Typical pressure-displacement curves are shown in Figure 8 for two of the four diametral displacements monitored. As the fracture length increases with a further increase in the loading pressure beyond the breakdown pressure P b , the angular distribution of the tangential stresses around the borehole boundary changes with the fracture length, and, eventually, when the loading
Predictive Calculation and Monitoring of Rock Stress
660
Table 1(a) Assumed Stress States and Inclinations of Four Boreholes
Magnitudes (MPa)
Principal stress directions (°) Azimuth* Dip"
Case 1
1.0 0.9 0.8
0.0 0.0 90.0
0.0 90.0 90.0
Case 2
1.0 0.7 0.3
0.0 0.0 90.0
0.0 90.0 90.0
a
Borehole inclinations (°) Dip*
Azimuth*
60.0 30.0 10.0 45.0
30.0 80.0 60.0 135.0
Dip is downward from the horizon. bAzimuth is from north to east. Table 1(b) Stress States Calculated by the Conventional Method Tensile strength of rock, T (MPa)
Principal stresses Magnitude Direction (°) (MPa) Dip Azimuth
1.0
0.996 0.905 0.802
0.0 3.5 86.5
1.2 91.6 91.3
0.5
0.986 0.900 0.797
1.7 2.7 86.8
3.2 93.3 60.4
1.0
0.878 0.693 0.269
1.2 7.6 82.3
3.0 93.1 84.0
0.5
0.813 0.682 0.245
2.9 10.4 79.2
8.4 98.9 83.2
Case 1
Case 2
Displacement of channel 3 (m)
Displacement of channel 4 (m)
Figure 8 A pressure-displacement record obtained from double-fracturing stress measurements in the Mozumi mine
pressure becomes Pi, the tangential stresses at the points distant from the primary fracture approach the tensile strength there and secondary fracture initiation occurs. It is a weak point of double fracturing, as shown in Figure 8, that ambiguity often exists in the determination of the values of Pi, Ps2b, Ph and Psh, where P sb and Ps2b are, respectively, the pressures at the instances when the primary and secondary fractures are just reopened.
661
Back Analysis Monitoring
Equation (2) can also be used in the case of double fracturing, being based on the primary fracture. Besides equation (2), for double fracturing the following equation is valid, based on the secondary fracture αμ, 0)Ps2b 4- β(λ, Θ)σγ + γ(λ, θ)σχ = 0
(4)
where λ is the ratio of the primary fracture length to the borehole radius. The quantities α, β and y are the stress concentration factors at angle Θ to the primary fracture position on the borehole boundary, where the maximum tensile tangential stress occurs. The variations in α, β and y with increase in fracture length have been calculated by the fictitious stress method (FSM) coupled with the DDM. The FSM-DDM analysis is described in Section 23.5.5, and the results are available in the literature [4].
23.4.4 Results of Stress Measurements from Five Different Methods A deep underground cavern for a new experiment is due to be excavated at 1000 m depth in the Mozumi mine by the Kamioka Mining & Smelting Co. Ltd. A plant for the same purpose has already been constructed roughly on the same level and in the same mine; the plant is named KAMIOKANDE (KAMIOKA Nucléon Decay Experiment). The new cavern will be about 10 times the capacity of the old one and will effectively hold 27.5 times more equipment; thus, the new cavern is named SUPER KAMIOKANDE. Five kinds of measurement of the initial stress were carried out before the excavation procedure was designed. The stress measurement techniques employed were hydraulic fracturing, double fracturing, two kinds of overcoring and acoustic emission (AE). One of the overcoring techniques employed is based on the principle of stress relief in the hemispherical bottom of the borehole over which 16 strain gauges are connected [8]. The other overcoring technique is based on stress relief in the borehole inclusion of cement mortar with eight strain gauges, each set up in a different direction [9]. The AE method employed is based on laboratory loading tests [10]. The specimens were taken from the cores in the orientations of the maximum and minimum principal stresses, as determined by an inclusion-type strain meter during overcoring. The stress states determined by thesefivedifferent methods are represented in Table 2. Note that in the three-dimensional stress determinations by both hydraulic and double fracturing the conventional method of calculation of the results was employed. It can be seen from Table 2 that the principal stress directions measured by the different methods are in reasonable agreement, although the directions of the intermediate and minimum principal stresses obtained from the OC-H measurement are essentially reversed in comparison with those obtained from the other methods. Note also that the magnitudes of the stresses are significantly different from each other. If we think about the real stress state, taking the overburden load and experience of the safe excavation of the KAMIOKANDE cavern into consideration, it seems that the maximum principal stress is given by the overcoring results and the minimum principal stress is given by the fracturing results.
Table 2 Comparison of the Results from Five Kinds of Measurement of the Initial Stress
σχ (MPa) σ2 (MPa) σ 3 (MPa)
OC-H*
OC-Ib
AEC
HFd
DFe
35.6 (205°/52°)f 9.7 (54°/34°) 2.4 (314°/14°)
28.5 (202°/54°) 13.6 (301°/6°) 4.2 (35°/35°)
28.4
19.2 (212°/43°) 9.4 (335°/31°) 4.7 (87°/32°)
17.9 (210°/52°) 13.0 (310°/8°) 11.0 (46°/37°)
— 4.8
OC-H = overcoring over hemispherical borehole bottom. bOC-I = overcoring over borehole inclusion. CAE = acoustic emission method. d HF = hydraulic fracturing method. eDF = double-fracturing method. fAzimuth/dip in all cases.
662
Predictive Calculation and Monitoring of Rock Stress
23.5 MODELINGS OF BOUNDARY ELEMENT ANALYSES 23.5.1 DDM Modeling of Tabular Orebody Extraction As Figure 1 shows, the lower deposit of the Yanahara mine is not tabular in a strict sense. However, the following are confirmed from strict analysis with respect to a spheroidal opening [11] : (i) the convergence distribution between the roof and the floor of a spheroidal opening whose span/height ratio (k) is 2 resembles that of a very flat opening (k = 20); and (ii) for a spheroidal model of 200 m in span and 60 m in height, displacements at points 20 m above the roof boundary resemble those of the roof boundary itself. Hence, the lower deposit of the Yanahara mine, massive in real terms, is assumed to be tabular in the predictive calculation of the model represented in Figure 2. The elastic constants for the orebody can then be quite different from those used for the rock around the orebody in the calculation. The calculation method can be found in the literature [12]. 23.5.2 DDM Modeling of Multiple-layer Mining The Kokura mine is a limestone mine. This means that the material unmined is the same limestone. Any opening in the Kokura mine is 5 m in height, i.e. not like a slot. However, it is confirmed from the two-dimensional calculation [13] that the stress and displacement distributions in the unmined area are very similar in both the accurate model and the approximate model in which any opening is represented by a horizontal slot at the center line of that opening. Therefore, quasithree-dimensional, predictive calculations of the model represented in Figure 3, where all the openings are assumed to be tabular, were carried out. Table 3 shows the calculated variation of the vertical stress component in a pillar into which pressure capsules were placed, for the period from March 1988 to July 1990. It can be seen from the table that the pillar stress slightly decreases as mining progresses over the period of measurement. 23.5.3 DDM Modeling of the Interaction between the Pressure Capsule and the Surrounding Rock Figure 9(a) shows the model used to calculate the convergence distribution, induced by the vertical rock stress component ay9 of the boundaries between the metal fittings and the flatjack, in the case where no flatjack is present. The model shown in Figure 9(b) is used to calculate the expansion distribution of the boundaries, induced by fluid pressure p in the flatjack. Figure 10 shows the results calculated from the model shown in Figure 9. It is presumed from the figure that roughly half of the change in oy appears as a change in the pressure p when the angle 0, shown in Figure 9(a), is 18°. 23.5.4 DDM Modeling of the Interaction between the Lune-shaped Flatjack and the Surrounding Rock Figure 11 is the model used to investigate the influence of the size difference between the luneshaped slot and theflatjackon the interaction between theflatjackand the surrounding rock. In this model all the boundaries, including the approximation of the lune-shaped slot, are divided into Table 3 Calculated Variation of the Vertical Stress in a Pillar Containing Pressure Capsules Stress (MPa)
Period 1988
1990
March August September October November December June
18.3 18.4 18.0 17.8 17.7 17.6 16.5
Back Analysis Monitoring ( a )
σ,
663
IIHilll!
ttttttttf Figure 9 DDM model for calculation of the interaction between a recoverable pressure capsule and the rock around it -|6
0.5 Position,
x/R
Figure 10 Calculated results for capsule-rock interaction (G is the shear modulus)
rectangular leaf elements [14]. Approximation of the lune-shaped slot by division into triangular leaf elements, as shown in Figure 12 [15], has also been carried out in order to check the errors inherent in the rough approximation by rectangular division. Figure 13 shows the calculated aperture distribution over^ lune-shaped slot. In this case, part of the lune-shaped boundary corresponding to the area of contact with the flatjack is subjected to internal pressurization. Figure 14 is the calculated relationship between the flatjack/slot area ratio and the ratio of expansion by internal fluid pressure to convergence by external rock stress. Thefigureshows that the expansion/convergence ratio is about 0.9 if the stress compensation process is monitored by the displacement at a point 7.5 cm from the slot. Figure 14 also shows that the superior model using triangular leaf elements gives results very close to those obtained from the model using rectangular leaf elements. 23.5.5 Coupled FSM-DDM Modeling of the Behavior of Rock around a Pressurized Sleeve Coupled FSM-DDM analyses were carried out in order to investigate the double-fracturing mechanism. As Figure 15 shows, in the coupled FSM-DDM model the borehole boundary is
664
Predictive Calculation and Monitoring of Rock Stress X
Figure 11 DDM model, using rectangular leaf elements, for calculation of the interaction between a recoverable, luneshaped flatjack and the rock around it
Figure 12 DDM division of the lune-shaped slot and flatjack into triangular leaf elements
Figure 13 Calculated aperture distribution over a lune-shaped slot subjected to internal pressurization
divided into FSM elements and the primary fracture, which extends as the internal pressure increases, is divided into DDM elements. However, the DDM boundary in the calculation is represented by a straight line passing through the borehole. We need such a model to get a reliable result because no displacement discontinuity can be obtained from an FSM element on the borehole boundary, even though the maximum fracture aperture must appear in the DDM element connected to the borehole boundary. The details of the calculated results are presented in the literature [4].
665
Back Analysis Monitoring i.Or-
£ 2
d: distance between the measuring point and the slot
0.2
o Rectangular leaf element model a
J
0.2
Triangular leaf element model
I
L
0.4 0.6 0.8 Flatjack /slot area ratio
J
1.0
Figure 14 Calculated relationship between the ratio offlatjackarea to slot area and the ratio of expansion by internal fluid pressure to convergence by external rock stress
illlJIIII λ = /_//?
FSM boundary DDM boundary
Figure 15 Coupled FSM-DDM model for investigation of the double-fracturing mechanism
23.6 MEASURED RESULTS The initial stress state measured in the Yanahara mine is shown in the first row in Table 4. Stress changes measured by hydraulic fracturing at points H2 and H3 are shown in the second and third rows, respectively, in the table. The roof sinkages measured by a series of connected vessels set in the drift along R24 on the level of SL26 are shown in Figure 16. The vertical stress component on the wall, measured by a lune-shaped flatjack, is shown in the bottom row in Table 4. The pressure variations measured by capsules buried with cement mortar in a drill hole are shown in Figure 17(a). Figure 17(b) shows the pressure variations measured by recoverable pressure capsules. The initial stress state measured by the double-fracturing method in the Kokura mine is shown in the left-hand column in Table 5. An example of the stress compensation process shown by wall stress measurements using a luneshaped flatjack is shown in Figure 18. The vertical and horizontal components of the surface stress at F in the side wall of the main drift are shown in the right-hand column in Table 5. Figure 19 shows the pressure variations measured by the pressure capsules simply put in the shallow drill holes. The capsules at PH and PV1 in the pillar are put 0.5 m deep, and the capsules at PV2 in the pillar and P V3 in the side wall of the main drift are put 1.5 m deep.
666
Predictive Calculation and Monitoring of Rock Stress Table 4 Stress State Variations at Various Sites as Mining Progresses (Yanahara Mine) February 1984
November 1984
13 (25°/64°)a 7.9 (89°/H°) 6.6 (177°/22°) 11 (130°/67°) 7.7 ( - 65°/21°) 7.6 ( - 118°/9°)
11 (17°/69°) 8.2 ( - 90°/6°) 6.3 (178°/20°) 12 ( - 85°/77°) 9.5 ( - 189°/1°) 7.4 (90°/13°)
ffj (MPa) HI σ2 (MPa) σ3 (MPa) σχ (MPa) H2 σ2 (MPa) σ3 (MPa) aj (MPa) H3 σ2 (MPa) σ3 (MPa) F σ ν (MPa)
April 1985
October 1985
March 1986
9.7 ( - 145°/85°) 2.0 (17°/5°) -0.1 (113°/2°)
12 (-6°/48°) 10 ( - 163738°) 6.8 (100°/15°) 7.7 ( - 145°/85°) 1.9 ( - 5°/8°) -1.1 (85°/7°)
12 (17°/20°) 10 ( - 127°/66°) 8.4 (112°/13°) 3.2 ( - 60°/21°) 2.6 ( - 153°/9°) 2.1 (96°/67°)
July 1986
24 (77°/90°)
a
Azimuth/dip in all cases.
1977
30
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
L
Figure 16 Roof sinkages measured by a series of connected vessel systems set in the drift of R24 and SL26
The vertical and horizontal components of the movement of the rock surface at SB on thefloorof the main drift, measured by LASDIS, were stored in a personal computer. The cabin in which the computer was situated was outside the mine and the computer was connected by cable to the LASDIS system underground. The air temperature and humidity around the LASDIS system were also observed and recorded in order to investigate how they affected the rock displacement measurements. Figure 20 shows the variations with time of the rock displacement, air temperature and humidity, printed out from the personal computer. As thefigureshows, the data for the three months from August to November 1989 are not recorded. This was because limestone dust contaminated thefloppydisk drives and caused a malfunction. Furthermore, the ventilation shaft shown as SH in Figure 3 was excavated and 17 blastings were carried out during the period from 18 October 1989 to 23 April 1990 in order to complete it. These blastings might explain the large jump in horizontal displacement that has apparently occurred during this period. However, since the magnitude of the displacement is too large to be considered as a real rock movement, it is possible that the jump is due to a displacement of the laser source during the shaft excavation, especially during the first six blastings carried out before 6 November 1990. 23.7 PREDICTED ROOF SINKAGE Roof sinkages above the centers of the segments shown in Figure 2 were calculated by a quasithree-dimensional elastic analysis using the displacement discontinuity method [16]. Figure 21 shows the predicted roof sinkage variation with time along a series of measuring points located at the connected vessel systems set in the drift of row 24 on sublevel 26 of the Yanahara mine.
667
Back Analysis Monitoring
I I I I 1 I
3
6 9 1986
12
I
3
I
6 9 1987
I I I I I I I 1 I I I I I I I I I I I
3
12
6 9 1988
12
3
6 9 1989
12
3
6 9 1990
12
Time (month) _
^
8
Q-
(b)
- ^ ^ \ R 3
\-
L-
Έ
4|
S
*·k V \-,
Λ-'' - ^ ^
"'""A-,
-. '«.-"'"'
V
R
Temperature
χ^"
—
1 1 11 9 1986
1 1 1
]y/\
1 1 1 1 1 6 1987
1
1 . I
1
I
LI
R2 H
1 1 L. 1 6 1988
Time (month)
Figure 17 (a) Pressure variations measured by pressure capsules buried in drill holes with cement mortar in and around the orebody. (b) Pressure variations measured by recoverable pressure capsules in the R21 pillar. The temperature variation of the air near the capsules is shown by the dashed line
Table 5
The Initial Stress State and Rock Surface Stress Components, Measured around the Main Drift of the Kokura Mine
Initial stress components in the vertical plane parallel to the main drift
Tangential stress components in the wall of the main drift
σ, = 7.0 MPa (72°/57°)a
σ ν = 14 MPa (72°/0°)
σ2 = 5.4 MPa (72°/ - 23°)
σΗ = 15 MPa (72790°)
a
In all cases, the first angle is the azimuth of the main drift and the other angle is the inclination from the vertical.
The magnitude of the calculated sinkage based on the measured initial stress is inversely proportional to the Young's modulus used. If the elastic constants are assumed to be the values shown in Table 7, the calculated sinkages agree reasonably well with the measured sinkages shown in Figure 16. Hence, the predicted sinkage curve based on such elastic constants can be used as a scale for the elastic deformation of rock. For other series of connected vessel systems in addition to this series, the measured sinkages in the current extraction can be monitored and compared with the predicted sinkages and used as a scale of safety. Unfortunately, all measurements of roof sinkage
Predictive Calculation and Monitoring of Rock Stress
668
E
=1
E loo
150"
100
Pressure (bar)
Figure 18 Stress compensation process recorded in stress measurements using a lune-shaped flatjack
2
3
6 1990
Time (month)
Figure 19 Variations in fluid pressure in recoverable pressure capsules in a pillar by the main drift, and temperature variations in the air near the pillar
Table 6
Overcoring Stresses Measured by Kyoto University in 1978
Measuring point
*1
Stress* (MPa)
σν
HI H2 H3 H4
33 101 91 27
17 51 80 12
15 45 78 12
1 30 37 7
a
The σ, are principal stresses and σν is the vertical stress component.
were discontinued after 1987 because of frequent breakdowns in the connected vessel systems and also because of a recent decline in the mining industry. 23.8 DISCUSSION In relation to the Yanahara mine, as the third row in Table 4 shows, the stresses in the center pillar along R21 have gradually decreased as the extraction has progressed, while the stresses around the orebody have not significantly changed. As shown in Figures 17(a) and 17(b), most of the pressures in the capsules in the center pillar have also gradually decreased while the pressure in the capsule at P4, which is relatively distant from the extracted blocks, has slightly increased.
669
Back Analysis Monitoring jcement jJ^. Horizontal displacement
9>
, io
S
-10
fl)
VerticalI displacement
50
t: 100 •a
Ë
Humidity Temperature
Time (month)
Figure 20 Variation in displacement components measured by LASDIS, and temperature and humidity variations in the air near the LASDIS system
o
V5 —o—
V3 -o—
VI
V9 —o—
V7 —o—
VII —o-
r
20
40
60r-
100
12/1989 Under extraction Coming extraction
L
v Measured during 1/1979-12/1980 o Measured during I /1981 - 8 /1983 • Measured during 9 /1983 - 6 /1987
Figure 21 The predicted variation in the distribution of roof sinkage and comparison with the measured sinkages Table 7 Elastic Constants Determined for the Predictive Calculation of Roof Sinkage Material Ore Rock
Young's modulus (GPa) 15 20
Poissoiïs ratio 0.2 0.25
In 1978, Kyoto University carried out stress measurements by the overcoring method [17] at four measuring points in the Yanahara mine. The locations of three of those points are very close to the points HI and H3 where hydraulic fracturing stress measurements were made. Another point of overcoring stress measurement is shown in Figure 1(a) as H4. The measured results are shown in Table 6. Comparing Table 4 with Table 6 we see that the stresses at H2 (in the center pillar along R21) and H3 (very close to the end of the pillar) have decreased down to about one tenth during the period from 1978 to 1985, while the stresses at HI (distant from the orebody) have decreased down to about one third. The difference at HI may be due to an overestimation of the stresses by the overcoring method and an underestimation by the hydraulic fracturing method, but the differences at H2 and H 3 may be from an actual unloading as the mining progressed.
670
Predictive Calculation and Monitoring of Rock Stress
These facts show that the stresses in the rock increase with distance from the area around the extracted blocks. With regard to the magnitude of the stress/pressure change, of course, the capsules in places near the extracted blocks may be sensitive, but the pressure changes in the capsules at P2 are large in spite of the distance. The reason for this is that P2 is along the well-ventilated main drift. The air temperature at point R was monitored in order to evaluate the effect on the pressure change detected, and it was confirmed that a 1 °C rise in temperature causes a pressure increase of 0.1 MPa. The pressures in the capsules at P2 fluctuate over a range of 3.5 MPa and it seems that the seasonal variation in air temperature is involved, even though the temperature change at P2 is always less than 35 °C. In any case, the stresses at P2 decrease overall as they fluctuate. The experimental site in the Kokura mine is not far below the surface entrance. Therefore, the pressures measured by pressure capsules inserted 0.5 m or 1.5 m deep in the rock are significantly affected by changes in air temperature and stress changes are masked by this effect. However, it can be seen from Figure 19 that the vertical and horizontal components of the pillar stress decrease slightly with time, and this is in agreement with the predicted trend shown in Table 3. 23.9 REFERENCES 1. Mizuta Y., Sano O., Ogino S. and Katoh H. Three dimensional stress determination by hydraulic fracturing for underground excavation design. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 24, 15-29 (1987). 2. Kuriyagawa M., Kobayashi H., Matsunaga I., Yamaguchi T. and Hibiya K. Application of hydraulic fracturing to three dimensional in situ stress measurement. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 26, 587-593 (1989). 3. Serata S. and Kikuchi S. A diametral deformation method for in situ stress and rock property measurement. Int. J. Min. Geol. Eng. 4, 15-38 (1986). 4. Sakuma S., Kikuchi S., Mizuta Y. and Serata S. In situ stress measurement by double fracturing. Proc. Jpn. Soc. Civ. Eng. 406/III-11, 87-96 (1989). 5. Mizuta Y., Sakuma S., Katoh H. and Kikuchi S. Stress and stress change measurements by hydraulic fracturing and double fracturing for safe underground excavation. In Proc. 2nd Int. Workshop Hydraulic Fracturing Stress Measurement, Minneapolis, vol. 1, pp. 205-244 (1988). 6. Ogino S., Araki H., Mizuta Y. and Sano O. Measurement and monitoring of rock stress by means of flatjack, pressure capsule and hydraulic fracturing techniques at the Yanahara mine. In Proc. 2nd Int. Symp. Field Measurements in Geomechanics, Kobe, vol. 1, pp. 339-343 (1987). 7. Sakuma S., Kikuchi S., Nakamura T. and Mizuta Y. Three dimensional stress determination from conditions of primary fracture initiation by internal pressurization of the borehole boundary. In Proc. 22nd Symp. Rock Mech. JSCE, Tokyo, pp. 436-440 (1990). 8. Sugawara K., Kaneko K., Obara Y. and Okamura H. Determination of the state of stress in rock by the measurement of strains on the hemispherical borehole bottom. In Proc. Int. Symp. Large Rock Caverns '86, Helsinki, vol. 2, pp. 1039-1050 (1986). 9. Kanagawa T., Hibino S. and Ishida T. In-situ stress measurements by the over-coring method - Development of an 8element gauge for 3-dimensional estimation. Central Research Institute of the Electric Power Industry, Report No. 385033 (1986). 10. Kanagawa T., Kitahara Y. and Hayashi M. Determination of geo-stress in rock samples using the Kaiser effect of acoustic emission - uni-axial tests and applications. Central Research Institute of the Electric Power Industry, Report No. 381004 (1981). 11. Lee H. K. and Mizuta Y. Analysis of rock behavior for safe excavation of mine pillars. In Proc. Int. Symp. Large Rock Caverns '86, Helsinki, vol. 1, pp. 681-692 (1986). 12. Crouch S. L. and Starfield A. M. Boundary Element Methods in Solid Mechanics, pp. 266-276. Allen & Unwin, London (1974). 13. Nomi T., Ono H. and Mizuta Y. Proc. MMIJ Spring Meet. 5-6 (1987). 14. Mizuta Y. and Lee H. K. A study on the applicability of the three dimensional elastic analysis by the displacement discontinuity method. Suiyo kwai-shi 20, 146-154 (1984). 15. Kuriyama K. and Mizuta Y. Three dimensional elastic analysis by displacement discontinuity method with boundary division into triangular leaf elements. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. in press. 16. Lee H. K., Mizuta Y. and Chou W. Analysis of rock behavior for the determination of the in situ Young's modulus and safe excavation of mine pillars. In Proc. Int. Symp. Eng. Complex Rock Formation, Beijing, pp. 318-324 (1986). 17. Saito T., Araki H., Kameoka Y. and Hiramatsu Y. Increasing the extraction ratio at Yanahara Mine and an associated programme of field measurements. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 23 (1), 77-83 (1986).
24 A Method for Monitoring Rib and Lining Pressure PETER FRITZ and KALMAN KOVARI Swiss Federal Institute of Technology, Zurich, Switzerland
24.1
INTRODUCTION
24.1.1 24.1.2 24.1.3 24.2
671
Stress Measurements by Stress Relief Stress Measurements on Elements Acting Together with the Lining integrated Measuring Technique
INTEGRATED MEASURING TECHNIQUE WITHOUT CONSIDERATION OF SHEAR STRAINS
24.2.1 24.2.2 24.2.3 24.2.4 24.2.5 24.2.6
Measured Quantities State of Deformation Determination of Cross Sectional Forces Determination of External Load Error Investigation Application: Steel Ribs in the Gotthard Road Tunnel
INTEGRATED MEASURING TECHNIQUE WITH CONSIDERATION OF SHEAR STRAINS AND LARGE PLASTIC DEFORMATIONS 24.3.1 Loadings in the Elastic Range 24.3.1.1 Accuracy investigation 24.3.2 Loadings in the Plastic Range 24.3.2.1 Yield condition, flow rule 24.3.2.2 Back-calculation in the plastic state 24.3.2.3 Limitations of plastic back-calculations 24.3.3 Illustrative Example
671 672 674 675 675 676 677 678 679 680
24.3
684 684 685 686 686 687 688 689
24.4
SUMMARY AND CONCLUSIONS
691
24.5
REFERENCES
693
24.1 INTRODUCTION In tunneling, determination of the stresses in the lining and of the rock pressure acting upon it are of great interest. However, from rock mechanics practice it is well known that measuring stresses in situ is a very difficult task. Stresses have to be measured indirectly, whereby three basic techniques are at hand: (i) measurement of deformations caused by stress relief (overcoring, slot cutting), eventually followed by applying a pressure until compensating the deformations (the flat jack method); (ii) monitoring of measuring elements which act together with the lining; and (iii) monitoring of the lining itself. Because the first two techniques are well known and have been extensively documented in recent years, they will be referred to only briefly here, placing the main emphasis on the technique where the lining is observed as a whole. 24.1.1 Stress Measurements by Stress Relief For measuring just one component of normal stress at the surface theflatjack method may be used [1]. One or more sets of measuring marks (bolts) arefixedat the surface (Figure 1) and the distance 671
672
Back Analysis Monitoring
I Rock surface Flat jack
Figure 1 Flat jack method applied for measuring tangential stress
Figure 2 Overcoring technique to measure a plane state of stress
/1>2 between each set is measured accurately. Then, a slot is created and the distances between the bolts are measured again. A thinflatjack is introduced into the slot and pressure is exerted until the deformation is compensated. It is assumed that this pressure corresponds to the normal stress component searched for. Furthermore, the deformability of the lining may be determined from the deformation measurements. An advantage of this procedure is that it may involve a relatively large area for determining an average normal stress. The disadvantages are the great effort required and the destructive nature of the procedure, not to mention the problems arising with greater stress gradients over the thickness of the lining. In addition, it must be assumed that the starting and endpoints of the unloading/loading-deformation cycle coincide, which means in practice a restriction to the ideal elastic behavior of the tunnel lining. The overcoring technique enables the determination of three components of the plane state of stress at the surface. Measuring bolts are fixed at the lining surface to form a 45° rosette (Figure 2). After taking a reading of the distances between the bolts, this arrangement is totally stress relieved by overcoring. A second reading, together with the stress-strain law of the lining, allows the backcalculation of the stresses prior to overcoring [2]. The advantage of this method with regard to theflatjack technique is that all three components of the stress at the surface may be determined. The disadvantages are the same as mentioned for the flat jack method; furthermore, the stress-strain law of the lining for this particular loading path should be known. 24.1.2 Stress Measurements on Elements Acting Together with the Lining Attempts are frequently made to determine the forces on a measuring element built in as part of the lining indirectly. The kind of elements varies within vast limits: one of the first approaches was to place wooden timbers between steel ribs and the rock. The forces acting could then be estimated purely on a visual basis [3]. Strain gauges are rarely applied [4] due to problems stemming from water, humidity, long-term stability and the small size of the strain gauges, compared with the inhomogeneities of the rock. Nowadays, mostly hydraulic pressure cells are used [5]. They are built-in radially within a concrete lining or tangentially at the interface between lining and rock. To overcome initial shrinkage of the concrete, a compensating tube may be installed, which allows the exertion of
A Method for Monitoring Rib and Lining Pressure
673
a positive pressure in the cell. When considerable temperature variations are expected it may be advantageous to use vibrating-wire load cells instead of hydraulic pressure cells. Pressure cells represent a straightforward way of measuring the loads acting in and on concrete linings and steel ribs. A large number of measurements have been executed in the last few years with quite satisfactory results. However, one should be aware of the problems inherent to this technique. The measured contact stresses between lining and rock are reliable only up to a point because the radial forces vary very much from one location to another. Furthermore the influence of a stress gradient over the thickness of the lining (bending moments) may alter the results of the measured normal stresses considerably. And last but not least the stiffness of the cells and the manner of the contact between cell and concrete (or rock) may yield erroneous readings. A more recent development in the monitoring of measuring elements, which is built in as part of the lining, is the so-called 'Philipp Holzmann (PH) measuring device' [6]. This device consists of two longitudinal girders, usually of 1 m length, being interconnected by three steel posts (Figure 3). The steel posts are dimensioned to have the same normal stiffness and flexural rigidity as the lining, thereby not affecting its behavior. Measuring the strains in these posts by means of vibrating-wire or strain gauges yields, thanks to the known material properties of steel, the sectional forces. This method has been applied with success in several tunnels, even for measurements in shotcrete linings (cf. Figure 4). An advantage of the PH method is definitely that in the longitudinal direction of the tunnel a length of 1 m or even 2 m is involved, as opposed to the pointwise measurements of pressure cells. However, in the cross section only local information may be obtained here also. Another limitation of the method stems from the requirement that the PH measuring device should not alter the behavior of the lining. Even for an ideal layout and ideal placing of the device, this can be fulfilled for an ideal elastic behavior of the lining only: creep or fissures in the concrete due to tension may falsify the results substantially.
rrrfWi^^ Lengths of advance
Direction of drivage
1
Shotcrete lining
/?*30cm mdPH
md PH
_ ^ | | Measuring device PH
N,M
ujjjj^jja^j-Ltjj
Figure 3 Layout of the PH measuring device (after Baumann [6]) - Drivage direction ■ Im ■
H-
■Im-
fly* [md2] Px-<èOY.H m £>2 = 6 0 k N m" 2
Δρ = 3 0 kN m" Δ/ν -Δρ Rm = 30-3.3 = 99 kN m"
23 43
22 21 42 41
[md~4l
Vj TA/V : 99 kN m '
Imdl | I 13 12 II post no. 33 32 31 post no.
^ΔΛ/ = 99 l·
9τ
Figure 4 Variation of forces in the shotcrete lining of the Munich subway due to a change of the compressed air pressure (after Baumann [6])
674
Back Analysis Monitoring
Summarizing, it may be said that the elements mentioned above provide a practicable means for a straightforward determination of the pressure in linings. However, all of them have the drawback of forming an additional part of the construction, whose behavior and interaction with the original structure may be difficult to estimate and whose location must be determined beforehand. Furthermore, it is not possible to recalibrate the measuring elements once built in. Therefore long-time measurements are susceptible to errors introduced by undetectable zero shifts of the measuring elements. Additionally, these elements furnish local information at discrete points only. Measuring practice demonstrates that strains in concrete linings may often be subjected to large variations within small distances. Therefore, a measuring procedure is sought without the introduction of additional parts. Furthermore, it should be recalibratable and it should yield global deformations of the structure. In the next section a measuring technique that aims to fulfill these requirements is presented.
24.1.3 Integrated Measuring Technique The integrated measuring technique (IMT) [7] uses the lining itself as the measuring element. The basic idea consists in reversing the procedure that is well established in structural analysis. There, beginning with a given load, the stress resultants at a given section are determined mathematically by integration, and further integrations yield the deformations. In the method dealt with here, the deformations are measured firstly, and subsequently, based on the stress-strain behavior of the lining or the steel ribs, the stress resultants are determined. From the latter, the loading is obtained by numerical differentiation. Both procedures are illustrated in Figure 5 for the case of a simply supported beam with elastic properties, expressed by the flexural rigidity El, carrying a distributed load p. The fundamental characteristic of the IMT is that its state of deformation is obtained by measuring the strains and curvatures in consecutive points along the intrados of the lining. With such 'linewise observations' [8] the distribution of the measured quantities along the line is obtained without the possibility of passing over important information. No additional parts must be built in which could affect the behavior of the lining. The measuring instruments are portable and therefore recalibratable. The locations of measurement may be chosen even after installation of the lining. A potential problem inherent to this method arises from the necessity of knowing the stress-strain relation of the prototype, i.e. the lining, for determining the stress resultants from the measured values. This is usually no problem for steel ribs, even in the case of (large) plastic deformations. However, for concrete linings and even more for shötcrete, this requirement may limit the accuracy or even the applicability of the method to a varying degree. In the following, the theory and applications of the IMT will be presented in more detail.
p (x )
Load
|
Mix)
Bending moment
Displacement
M"--p
u
ix)
Structural analysis
p -► M -► u
(integration)
Integrated measuring technique :
u -► M -► p
(differentiation)
Figure 5 Computational procedure in the case of a simply supported elastic beam
A Method for Monitoring Rib and Lining Pressure
675
24.2 INTEGRATED MEASURING TECHNIQUE WITHOUT CONSIDERATION OF SHEAR STRAINS 24.2.1 Measured Quantities In Figure 6 a lining segment with the depth h and the radius R at mid-height of the cross section is shown. The lengths to be measured are defined by the measuring marks A, B and C with the eccentricity e. The distance between A and B is denoted by L, the rise to C by F. The actual measured values are the change in length / of the chord L and the change in length/of the rise F (the sign is defined as positive for shortening). For control purposes or error compensation of these values, the relative displacements d between individual points of the lining may also be measured. For measuring the three values/, / and d, based on well-known principles, three instruments have been developed [7]. Due to the unfavorable error propagation in the back-calculation (numerical differentiation!), these instruments must fulfill high accuracy requirements. The curvometer (Figure 7), which serves to measure the change of the rise, exhibits under field conditions a mean error of mf = 2 μηι only; the deformeter (Figure 8) for measuring the change of the chord exhibits a mean error of mt = 5 μιη. Both instruments have a base length of 50 cm. They are available in a purely mechanical form with dial gauges or with electronic displacement transducers, which facilitate the recording of a great number of values. The relative displacement d between points in a greater distance D is measured with the distometer (Figure 9). Its accuracy depends on the distance and is about + 5xlO~ 6 D.
Figure 6 Segment of the lining with the measuring lengths F and L
Figure 7 Curvometer with calibration gauge for measuring the change in rise
676
Back Analysis Monitoring
Figure 8 Deformeter with calibration gauge for measuring the change in the chord
Figure 9 Distometer to measure the convergency on the basis of the tensioned invar wire
24.2.2 State of Deformation In the following derivations the deformations are supposed to be sufficiently small to allow the use of thefirst-ordertheory of structural engineering. Furthermore, loads and displacements outside the plane of the cross section are neglected. The lining is assumed to be circular or made up of parts of circular shape. To simplify the computation by using the classical beam theory, the lining is idealized as a slender, straight beam. Because in tunneling practice slender linings are usually used, only the condition R
-h>l° must be paid attention to. With this the angle ocL of a lining segment (Figure 6) is rarely greater than « 10°, thus allowing the simplifications sinaL « aL cosaL « 1 The change of the state of deformation according to the measured changes in length of the rise and the chord may be expressed by the curvature κ, the axial strain ε and the shear strain y. Neglecting y for the moment, κ and ε are determined with the principle of virtual work from / =
KMf as +
eNf as
I =
KMtas +
eNtds
(1)
A Method for Monitoring Rib and Lining Pressure ( a )
677
True strain system
1
1 5
\-e/R
"
Curvature
©
Axial strain
Θ ( b )
Virtual load system
Normal force N Figure 10 (a) True strain system and (b) virtual load system for determining κ and ε from the measured values of / and /
with
\-e/R
The virtual load system used is shown in Figure 10. For an arch segment loaded by uniformly distributed loads, ε varies linearly along a segment and κ parabolically. However, for the sake of simplicity, constant values are assumed for both ε and κ. Using the corresponding true strain system indicated in Figure 10 leads to K = Kff+
K X\
ε = eff + εζ/ where for small eccentricities e <ζ R the above parameters become 8
1 RL 1
Se
Bf
=
T2'
(3)
ε» = ■
For neglectable shear strains the state of deformation of the lining is defined uniquely by κ and ε along the axis. The relative displacements d between arbitrary points of the lining may be determined purely kinematically, i.e. without involving material properties. If such displacements d are additionally measured directly with the distometer, they may serve to check the accuracy of the readings and/or to compensate for errors. 24.2.3 Determination of Cross Sectional Forces With the assumption of cross sections remaining plane, from the values κ and ε the axial strain εΛ at a distance yh from the axis is given by (cf. Figure 11) Bh = ε + Kyh
678
Back Analysis Monitoring V/////////////À Y//S//SSS/V/VSSA
.IE«
Figure 11 Cross section composed of individual materials
Figure 12 General form of a nonlinear stress-strain relation
The cross section may be inhomogeneous, composed of a number of different materials, each of them characterized by a stress-strain relation σ = σ(ε)
Different characteristics for loading, unloading and reloading may be taken into account (Figure 12). The moment M and the normal force N are obtained by integrating over the area A, viz. M =
yha(e)aA
N =
σ(ε)άΑ
(4)
As a simple example, an elastic steel rib with the area A, the moment of inertia 1 and the modulus of elasticity E may serve, for which Hooke's law σ = Εε is valid. Substitution in equations (4) yields the well-known expressions M = κΕΙ N = εΕΑ
(5)
where ε applies at the centerline of the rib. 24.2.4 Determination of External Load The loads acting are found by equilibrium considerations in an infinitesimal arch element. In Figure 13 such an element with the arch length as and the cross sectional forces M, N and Q is shown. The external loads are pds in radial and ids in tangential directions, and a distributed moment mds. Formulating force equilibrium conditions in radial and tangential directions and moment equilibrium at the center yields three equations for the four unknown quantities p, i, m and Q; i.e. for solving the equation system a further condition is needed. This is provided by considering that in reality a support is loaded by radial and tangential loads only, but rarely by external moments. Thus setting m = 0 leads to the simple differential equations for the external loads N
p = —c
\ds
ά2Μ , as2 1 dM\ R ds )
(6)
A Method for Monitoring Rib and Lining Pressure
679
Figure 13 Infinitesimal arch element Ä + Ä/2
with and the relation for Q is _ iàM
dM
dNh\
(7)
" V"d7 ~ "d72/ C * ~d7
Q
The differentiation of equation (6) is carried out numerically according to the trapezoidal rule. Thus the required load quantities are given by the discrete values M, and Nt in n segments as Pi
for
_ Ni
" *C
M,.,, - 2Mt + Mi+l
?
c
*-£(*...-*.-,+*^) 1< i< n
24.2.5 Error Investigation The accuracy of the rock load components determined by the above formulae depends on three main factors. These are the accuracy of the measurements, the constitutive laws for the tunnel lining and the deformations of the lining out of its plane. First of all the error propagation due to the mean errors mf and mi in the measured quantities is considered. If a value W is defined as a function of n measured quantities Vh the mean error mw in W may be determined according to the error propagation rule of Gauss mw
- wJmy'+
+
w.
I
m
vn
as a function of the mean errors mVi. Applying this expression to equations (S) and (2) for M and N (for the elastic case), the mean errors in the stress resultants M and N become
■'-(TTKI)'-^-'] Together with the expressions for pt and th the required mean errors in the load components p and t can be obtained as
m
i = h[h<+ {if <]
2 '
i Γi 2L2IR2
2 M
21 N
]
i 2Ü-
N
680
Back Analysis Monitoring c==
rr==3_i
100 200 300 400 Depth of beam (mm)
100 200 300 400 Depth of beam (mm)
Figure 14 Mean errors in section forces and external loading due to inaccuracies of measurements
Of special interest is the ratio mt/mp. If the absolute values of Table 2 are used, a ratio of » 3 for a HEB 240 rib and « 1/2 for a quadratic cross section are obtained. This means that for arches with a relatively small A: I ratio (e.g. for steel ribs) the back-calculated tangential loads will be less accurate than the radial loads. In order to illustrate the influence of the accuracy of the measurements in a practical example we refer to Figure 14, in which the mean errors in M and N as well as in the external loads p and t are determined for an I-shaped beam of HEB type as a function of the depth of the beam. It follows from the equations above that only the radial force component depends on the radius R of the arch. This dependence, however, is so small that it is neglected in Figure 14(b). Let us now take one example from the diagram and consider a beam of depth 300 mm (HEB 300). It may be seen that the normal force can be determined with an accuracy of about ± 50 kN. If the yield stress in the steel is 360 N m m - 2 the section is able to support a normal force of 5360 kN, which gives an error of less than 1% in relation to the maximal value. For smaller loads the relative error increases correspondingly. Especially in such cases it has proved itself to be advantageous to work with spline interpolations to smooth the measured values [7]. 24.2.6 Application: Steel Ribs in the Gotthard Road Tunnel The integrated measuring technique found its first application in the Gotthard Road Tunnel north heading, at the contractor's request. Due to very unfavorable rock conditions, the 'side drift' tunneling method was employed with subsequent excavation of the crown, leaving a central core [7]. In the side drifts the portions of the permanent concrete lining were constructed, which served as a support for the I-section steel ribs in the crown. For the excavation of the crown (top heading), lances were used, supported by the steel ribs. In addition to a trial section, a further four measuring sections were installed, between 4.0 and 4.5 km along the tunnel length. All the measurements reported here were carried out in the top heading according to the layout in Figure 15. The steel ribs were subjected to loading immediately after installation at the face of the top heading (Figure 16), since they served as a support for the lances. However, at a distance of about 5 m from the face a complete unloading occurred under the skin-tail of the lances. Under this condition the initial reading was carried out in order to determine the rock loads transmitted to the ribs by liner plates. To give an example of a series of readings, we consider Figure 17, in which the measured values /and / are given in terms of the arch length. The scatter and some stray points (point P) are typical. The full line is the smooth curve obtained by spline interpolation. The determination of moments and normal forces from these values is best done with the aid of a diagram even on site. Such a diagram is shown in Figure 18 for the beam type HEB 240. The two families of curves define bending moments and normal forces of constant intensity. The full line defines the boundary of the
681
A Method for Monitoring Rib and Lining Pressure
Figure 15 Layout of the measuring bolts (points) and that of the distometer control lengths
Figure 16 Steel arch with measuring bolts at installation close to the tunnel face (Gotthard Road Tunnel)
o Change in F + Change in L E E
a>
-io
-30,
2.50
5.00
7.50
10.00
12.50
Arc length (m)
Figure 17 Distribution of the measured values of / and / along the arch after 46 days (at 4.330 km)
zone of pure elastic deformation. A different way of presenting the results is shown in Figure 19. This figure, which was obtained from Figure 18, shows how the stress resultants M and N contribute to the stressing of the cross section. The boundaries of the zones of elastic behavior are shown as straight lines.
Back Analysis Monitoring
682
2 4 0 mm 40
£ E
240 mm
(M
30 Ό .Yield limit Proportional limit
M (klMm ) N
(kN)
•
Left side-wall
*
Roof
o
Right side-wall
Figure 18 Plot of the measured deformations / and / after 46 days, indicating the corresponding values of M and N (at 4.330 km)
300
200 h
Έ 2
i+^ c
E -lOOh
-300r-
o Right side-wall
Figure 19 Interaction diagram plot of moments and normal forces after 46 days (at 4.330 km)
The corresponding distributions of the stress resultants M and N along the axis of the arch are shown in Figure 20. The derived rock pressure distribution at tunnel length 4.330 km and its development with time are shown in Figure 21. The maximum value of the radial load was determined for the next-to-last reading and found to be around 200 kN m 2. The I-section used here has a load capacity of about 500 kNm" 2 for a uniformly distributed load. To check the accuracy of the measured values off and /, measurements were carried out using the distometer to determine the changes in the distances given in Figure 15. Using the formulas presented in [7], these quantities could also be computed from the values obtained using the deformeter and curvometer. Table 1 shows a comparison of the results for the case of the cross section at 4.254 km over a period of 38 days.
A Method for Monitoring Rib and Lining Pressure
683
Figure 20 Distribution of moments and normal forces along the steel support, checked by special load cells at points A and B (asterisks) after 46 days (at 4.330 km)
Figure 21 Distribution of rock pressure along the steel support for different readings (at 4.330 km)
Table 1 Comparison of Deformations Obtained with Different Measuring Devices at 4.254 km Measurement
Left side-wall Left roof zone Roof Right roof zone Right side-wall
Computed from f and I (mm)
Measured with distometer (mm)
-0.95 0.62 0.41 0.02 -1.24
-1.07 0.67 0.56 0.12 -1.30
684
Back Analysis Monitoring
24.3 INTEGRATED MEASURING TECHNIQUE WITH CONSIDERATION OF SHEAR STRAINS AND LARGE PLASTIC DEFORMATIONS The original version of the integrated measuring technique as described above disregarded the influence of shear strains. However, extensive tests showed, that for cross sections with a small area effective for shear compared to the moment of inertia, this influence may not be neglected. Therefore the following considerations are of importance for such cross sections only, e.g. for steel ribs.
24.3.1
Loadings in the Elastic Range
The influence of shear strains y on the deformations of the arch, i.e. the measured values of/ and /, may be found by extending the equations of virtual work (equations 1). In view of the variable shear stresses over the cross section, the corresponding shear strains will also be variable; the stress distribution does not remain plane. However, because of its small influence, this will be neglected as is usual in structural engineering. The following derivations hold for a mean value y, which is given with the web area A' (which is effective for shear) and the shear modulus G by 7
with
Q_ GA'
=
G =
(8)
2(1 + v)
where v is the Poisson's ratio. Equations (1) may then be extended by the term
ί
yQds
The true strain system and the virtual load system used are basically the same as in Figure 10, but are extended for the true shear strain y and the virtual shear force. Since a constant shear strain does not have any influence on the measured quantities, a variation of y along a measuring segment must be permitted. Therefore, constant strain distributions over half of the segment are assumed, with a step at the middle of the segment. Thus the measured values κ and ε may be expressed analogously to equations (2) by K=
+ κ,Ι + KvAy
Kff f
y
ε = eff + etl + eyAy
(9)
The constants Kf,Kh ef and st are the same as before. The constants κΊ and ey due to y are (for small eccentricities e k R) 2 1 1_ Ky
—
L
sy =
+
2e
SR2
Ä
Z
L
~ ÏR
It should be observed that the change in shear strain Δγ is not determined by direct measurements. It is determined analytically by substitution of the shear force according to equation (7) into equation (8) for y, which leads for a constant cross section to s/2 fa2M hd2N\ L/2d2M Ay = —'—[ — ; « ~ (10) Y cGA' \ ds2 2 ds2 J G A' ds2 Substituting Ay into the expressions for κ and ε (equations 9) leads to recursive formulas for the computation of these values. The derivations for the cross sectional forces and the external loads remain the same as above, i.e. without taking into account shear strains.
A Method for Monitoring Rib and Lining Pressure (a ) Load
685
( b ) Geometry
x
240 mm
+^?
h = 240 mm
' HEB 240
0.59 m :0.75 ml · R
f^ri
2.25 m
Figure 22 Steel rib HEB 240 subjected to concentrated loads P = 150 kN, E = 190 x 106 kNm" 2 , G = 70 x 106 kNm" H = 357 kN, K = 4.65 m, s = 51.6 cm Segment
I
Outside
(a )
(b) Inside Computed from 'measured values' Correct values
Scale strains : I 1 0.2%o
Figure 23 Strain distribution over the cross section for the example of Figure 22: (a) consecutive procedure and (b) overlapping procedure
For the measuring procedure presented in the foregoing, the individual measuring segments have been aligned one after the other in a consecutive manner. With the designations of Figure 6, this means that the end point B of the actual segment coincides with the starting point A of the following segment. Investigations revealed poor results for back-calculations based on consecutive measurements when concentrated changes of shear strains occurred [9]. The reason for this is that these concentrated changes of shear strains are only partly caught with consecutive measurements. Imagine a concentrated load, which acts exactly at the boundary between two measurements, provoking a sharp bend of the deflection line. In this case the shear change is not detected or taken into account at all with the measurements. Therefore the consecutive measuring procedure is not suited to account for concentrated changes of shear strains, as they are encountered in steel ribs with a small web area and concentrated loads. Therefore we use an overlapping measurement procedure if concentrated changes of shear strains are to be expected: corresponding to the designations in Figure 6, point C in the middle of the current segment then coincides with the starting point A of the following segment. To compare the overlapping procedure with the consecutive one, the symmetric arch presented in Figure 22 will be discussed. With the aid of a finite element program [10], the displacements were calculated for the prescribed symmetric loads. From these the 'measured quantities' / and / were computed and hence the strains back-calculated. The resulting axial strains in the cross sections along the arch are shown in Figure 23. Whereas consecutive measurements lead to the unsatisfactory results of Figure 23(a), the accuracy of the overlapping procedure as shown in Figure 23(b) is of the order of the precision of drawing. 24.3.1.1
Accuracy investigation
The accuracy of the values κ, ε and y depends on systematic and random errors. Systematic errors are caused by the system, i.e. by the method of measurement or evaluation. Examples of these are the neglect of deformations outside the plane of the measuring section, ignorance of the local deformation pattern within a measuring segment, and errors introduced by the numerical differentiation. Only the last item will be discussed here.
686
Back Analysis Monitoring Table 2
Related Mean Errors on Curvature and Strain due to Errors in Measurements (in %) Without considering y
With considering y
mK
mt
HEB 240 mK mE
240 x 240 mK mE
100 100
100 100
65 45
90 75
Procedure Consecutive Overlapping
85 80
98 90
The numerical differentiation of the moments or curvatures by means of the trapezoidal rule yields correct results for a parabolic and a linear distribution of these values only. For a smooth distribution of the curvature, as is usual for loads in tunneling practice with several concentrated or evenly distributed loads, the deviation from the correct values is small. For a single concentrated load with a pronounced peak of the curvature, the deviation may be slightly larger. It is advantageous that this influence is limited locally to the immediate vicinity of the load application. Random errors are understood to be the actual measuring errors. The mean errors on κ and ε may be determined according to the error propagation rule of Gauss and then related to the corresponding values mK and mE of the original procedure. The results are listed in Table 2, where the values of the original procedure have been assumed to be 100%. Two different cross sections have been investigated: a steel rib and a full cross section of similar dimensions. The two rows of data in Table 2 give the results for the consecutive and overlapping measuring procedures as presented in this chapter. It may be seen that both procedures lead to smaller mean errors with consideration of shear strains than without. This is caused by the involvement of more measuring segments for the determination of one value, or, expressed in other words, by the involvement of more but smaller coefficients. The summation of their squares leads then to a smaller value for the mean error. However, it is pointed out that Table 2 refers to the random errors only. When including the systematic error mentioned above (differentiation across the point of discontinuity for concentrated loads), this may cause deleterious effects when considering y. Comparing the first and the second row leads to the statement that the overlapping procedure is superior to the consecutive one also with respect to the random errors. However, it must be borne in mind that it requires double the number of measurements.
24.3.2 24.3.2.1
Loadings in the Plastic Range Yield condition, flow rule
In the web of a steel rib subjected to shear, a two-dimensional state of stress prevails, the principal axis of this being no longer parallel to the axis of the rib. Therefore the yield condition and the flow rule, which define the direction of the plastic strain increments, should also be formulated accordingly. On the other hand, the flanges are nearly free of shear stresses, i.e. a one-dimensional approach is adequate there. It can be shown [9] that the state of deformation is governed mainly by the uniaxially stressed flanges. If the area of the web were greater, the shear strain would be smaller and the state of deformation of the whole cross section would be uniaxial. Therefore in the present measuring technique, the yield condition will always be formulated uniaxially for the axial stresses. The influence of shear stresses is taken into account by a flow rule. In the plastic state, equation (10) for the change of shear strain Ay does not apply any more because the shear stresses over the cross section are no longer constant and because the relation is no longer linear. Equation (10) is now replaced by assuming a flow rule of the form s/2 (à2M Δνrρ = —-— cGfA'\ as2
hd2N\ L/2 d 2 M %— 2 2 as ) GfA' as2
(11) V '
where Gf designates a fictitious shear modulus, which accounts for a partial yielding of the web according to a nonlinear stress-strain law τ =
τ(γ)
A Method for Monitoring Rib and Lining Pressure
687
5°/c Correct values
Axial strain, εΛ
Shear modulus (10 kN m" )
Figure 24 Secant shear moduli of a partly plastified cross section (σ = 160 Nmm - 2 )
Because in the plastic state the moments are no longer proportional to the curvature (i.e. equation (5) no longer applies), Ayp of equation (11) must be determined iteratively together with equation (9). As an illustration, the 'correct' solution from an FE computation will be discussed first, before outlining the determination of the fictitious shear modulus Gf. For that purpose an arch with the same geometry as in Figure 22 is used, but supported by two hinges, one at each end. It is subjected to a loading that could also occur in tunneling practice; an equally distributed load of p = 500 k N m " 2 over a length of 1.50 m at the top. In Figure 24 the resulting strain distribution and the computed secant-shear moduli τ(Λ) GsecW =
—
in three integration points are shown for the plane of symmetry. Because, for the measuring technique presented here, a correct determination of these shear moduli is out of the question, a simplified approach must be used. For this, a pragmatic assumption for the shear modulus in the fiber h of the cross section is used in the'form G„M with
= 2(1 + v)
E^(h) =
As may be seen from Figure 24, the shear stresses determined in this way aproach the 'correct' values sufficiently accurately. The desired value Gf is then obtained by an integration over the cross section from =
as
24.3.2.2
τ άΑ' == yy τάΑ'
*-*L
G, Gsec άΑ' = yA'Gf
GsecdA'
Back-calculation in the plastic state
For steel ribs with large plastic deformations the back-calculation of strains leads to unsatisfactory results even for distributed loads. For illustration purposes the same example as above is used but with a distributed radial load of p = 500 k N m - 1 over a length of 1.50 m at the top. In the plane of symmetry this load produces strains at the outside fiber which are about seven times greater than the elastic strains. The strains and cross sectional forces, back-calculated with the overlapping procedure, are subject to large errors (Figure 25a), and the resulting external load is too inaccurate to be useful. The reason for this is to be found in the sensitivity of the cross sectional forces of steel ribs to the strain distribution. In the plane of symmetry of Figure 25, theoretically the axial strain at the inside fiber should vanish, i.e. the inner flange contributes hardly anything to the cross sectional forces. However, because of the great area of the flange, an inaccuracy of these strains of about 10% of the strains at the outside fiber may change the cross sectional forces by 50%. Because the strains are computed iteratively from the second derivative of the moments and normal forces, the errors on the strains also become relatively large.
Back Analysis Monitoring
688
E-
r
-~UR-
Figure 25 Back-calculation for an arch with great plastic deformations: , 'correct' values; , back-calculated values, (a) Shear strains based on discrete values of M; (b) shear strains based on mean values of M
Because the distribution of moments along the arch and their derivatives are apparently of outstanding significance for the accuracy of the integrated measuring technique, they have to be discussed in somewhat more detail. Until now, the discrete values for the moments have been computed directly from the strains determined at the corresponding points. The derivatives of these moments then lead to the unsatisfactory results mentioned above. In the sense of a pragmatic approach, these discrete values are no longer considered as decisive; instead, a mean value is used, also taking into account the data of the neighboring segments. In Figure 26 this mean value at point i would be just one-half of the discrete value mentioned above. This procedure corresponds arithmetically to the assumption of constant moments, which is also adopted in the following for the normal forces and the loads. In Figure 25(b) the corresponding results for the example of the distributed load are shown. The accuracy is now satisfactory. 24.3.2.3 Limitations of plastic back-calculations In spite of the great plastic deformations, the back-calculation for the distributed load of Figure 25(b) leads to good results. The reason for this is the smooth line of deflection (cf. Figure 27b). On the
A Method for Monitoring Rib and Lining Pressure
689
M, Discrete
Mj Average
Arch length
Figure 26 Decisive moments for determination of shear strains
HEB 240
Figure 27 Computed vertical displacements for an arch with great shear stiffness (Gsec = 5£sec): (a) for concentrated loads; (b) for a distributed load
left-hand side of Figure 27, the line of deflection for the same statical system as on the right-hand side - but now subjected to two concentrated loads - is shown. For the elastic case the line is still smooth. As soon as the yield strength is reached, very big plastic deformations occur near the points where the loads are applied, leading to a discontinuous line of deflection at these points. To practically eliminate the influence of shear, the arch was assigned numerically a big shear stiffness (Gsec = 5£sec). Thus, the discontinuity of the line of deflection is not produced by shear, but by plastic deformations only. Measurements across this discontinuity point must be avoided, because they would lead to irregular results. For such a case the measurements and back-calculations must be carried out separately at both sides of the discontinuity points. If in practice these points have not already been detected visually in advance, they may be recognized from the large changes in rise with opposite signs measured at either side of these points. 24.3.3 Illustrative Example To demonstrate the applicability of the technique presented, a steel arch was subjected in the laboratory to prescribed loads and the changes in length of the chords and the rises were measured in an overlapping manner. From these values the cross sectional forces in the arch could be backcalculated and from these the external loads. Comparison of the latter with the known values allowed the accuracy of the technique to be checked.
690
Back Analysis Monitoring
5xP HEB 240
/?=4.90m 2.42 m
Figure 28 Test arch subjected to prescribed loading conditions: (a) test arrangement; (b) detail of the top; (c) system sketch
In Figure 28 the test arrangement is represented. The steel rib used was supported at both ends by (nonideal) joints. The load was induced by seven pistons at the top, acting in a radial direction: five of them exerted the same pressure; the two at the boundaries exerted 50% pressure. The load in each piston was increased gradually up to P = 50 kN, then it was decreased to 20 kN and finally increased again up to P = 190 kN. For this value the measured data were no longer stable. After the test it was recognized that, in the critical cross section at the top, the yield strength had been reached for a load of about P = 105 kN. On prismatic samples extracted from the flanges of the same steel rib, the material properties were determined in advance Modulus of elasticity
E = 210-240 x 106 kNm" 2
Yield limit
ay = 270-290 Nmm" 2
Deformation modulus EH = 0-2.5 x 10 6 kNm~ 2 The deformation modulus applies after reaching the yield limit sy in the range of interest ey < ε < 10%o. In Figure 29, the interpretation of the readings is shown for changes in loads of P = 20-50 kN (elastic case). Whereas for unloading and reloading practically the same values were measured, the corresponding data for the first loading were slightly greater. This could be a consequence of a small slip in the construction. The power of the measuring technique may be checked by a comparison of the back-calculated and the directly measured ('correct') radial forces. The corresponding mean error on the force P is ±3.5 kN. This agrees with the accuracy of the instruments for the original measuring technique discussed at the beginning, i.e. without consideration of shear strains. Therefore, the influence of shear could be taken into account without prejudicing the absolute accuracy. Finally, in Figure 30 the evaluation for the load P = 180 kN is represented - the state shortly before the collapse of the system. In this case the displacement of the top was nearly four times the
691
A Method for Monitoring Rib and Lining Pressure 'Correct' values
+1
~-u—^__^_
E Ξ
_^
—
-t_ _
200
I 1
5
I
I 8
1 II
15
.* Correct' values
Figure 29 Back-calculation for the test arch: elastic state. E = 210 x 106 k N m , reloading
2
;
, first loading;
, unloading;
displacement when reaching the yield limit. In spite of this heavy plastification over a length of about 2 m, the results are still reliable, as may be confirmed again by comparison with the 'correct' radial forces. 24.4 SUMMARY AND CONCLUSIONS Determination of the stresses acting in a tunnel lining or in steel ribs is a very difficult task. Three basic techniques are at hand: (i) measurement of deformations caused by stress relief, (ii) observation of elements which act together with the lining, and (iii) observation of the lining itself. Each one of these techniques has been briefly discussed with its advantages and shortcomings. The most important limitations are based on the facts that for the first two (standard) techniques, only local values in some selected points are obtained, and that the interaction of the measuring elements with the construction is not always determinable. Therefore, the main emphasis here is put on the so-called integrated measuring technique, which belongs to the third group mentioned above and circumvents the problems of the other measuring techniques at the expense of introducing some other limitations, as outlined below. The integrated measuring technique enables the determination of the cross sectional forces in, and the loads acting on, a tunnel lining based on deformations measured at its intrados. The basic idea of the technique relies on inverting the procedure of beam analysis, well known in structural engineering. Here, in successive positions along the intrados of a lining, the changes in length of the chord
692
Back Analysis Monitoring
Correct values
_ <Λ_~
0)
50
1
t
5
-50 -100
1 8
1
15
II
| 1
1500
σ ■*
E
"
1000 500
Figure 30 Back-calculation for the test arch: plastic state. E = 210 x 106kNm E„ = 2.5 x 10 6 kNm- 2
2
; v = 1/3; σ
'
280 N mm"
and the rise for a base length of 50 cm are measured. From these values the state of deformation, expressed by axial strain, curvature and shear strain, may be determined. In the elastic case without shear forces, purely kinematic considerations are sufficient; otherwise the stress-strain relation of the lining must also be taken into account. Based on the state of deformation and the constitutive equations, the cross sectional forces may be found and, together with equilibrium conditions, the external load may be back-calculated. The original version of the integrated measuring technique disregarded the influence of shear strains. However, extensive tests on steel ribs have led to the conclusion that for cross sections with a small area effective for shear, compared to the moment of inertia and concentrated loads this influence cannot be disregarded. Furthermore, numerical simulations have shown that for single concentrated loads, after reaching the yield limit, the curve of deflection is no longer continuous, thus forcing the measuring technique to be applied separately at both sides of such points. Distributed loads, as are usual in tunneling practice, may be handled without any problem even for large shear and plastic deformations, provided the measurements are carried out in an overlapping manner. It should be remembered that, when applying the technique on precast concrete dowels [9] or even on shotcrete, the uncertainties and the scatter of the material properties and/or of the geometry reduce the accuracy of the results dramatically or even rule out any back-calculation. On the other hand, the state of deformation may be determined (without considering shear strains) purely kinematically from the measured quantities / and / only. Because the actual technique allows measurements on a linewise basis and not only locally at individual points, these deformations may provide valuable hints regarding the behavior of the lining under loads.
A Method for Monitoring Rib and Lining Pressure
693
24.5 REFERENCES 1. Rocha M, Lopes J. J. B. and da Silva J. N. A new technique for applying the method of theflatjack in the determination of stresses inside rock masses. In Proc. 1st Congr. Int. Soc. Rock. Mech., Lisbon (Edited by M. Rocha), vol. 2, pp. 57-65. Bertrand, Lisbon (1966). 2. Barla G. and Rossi P. P. Stress measurements in tunnel linings. In Proc. Field Measurements in Geomechanics, Zürich (Edited by K. Kovari), vol. 2, pp. 987-998. Balkema, Rotterdam (1983). 3. Proctor R. V. and White T. L. Rock Tunneling with Steel Support. Commercial Shearing Company, Youngstown, OH (1946). 4. Kohlbeck F. and Scheidegger A. E. Application of strain gages on rock and concrete. In Proc. Field Measurements in Geomechanics, Zürich (Edited by K. Kovari), vol. 1, pp. 197-207. Balkema, Rotterdam (1983). 5. Londe P. Field measurements in tunnels. In Proc. Int. Symp. Field Measurements in Rock Mechanics, Zürich (Edited by K. Kovari), pp. 619-638. Balkema, Rotterdam (1977). 6. Baumann T. Numerical analysis and reality in tunneling — verification by measurements? In Proc. 6th Int. Conf. Numerical Methods in Geomechanics, Innsbruck (Edited by G. Swoboda), pp. 1457-1464. Balkema, Rotterdam (1988). 7. Kovari K., Amstad Ch. and Fritz P. Integrated measuring technique for rock pressure determination. In Proc. Int. Symp. Field Measurements in Rock Mechanics, Zürich (Edited by K. Kovari), pp. 289-316. Balkema, Rotterdam (1977). 8. Kovari K. and Peter G. Continuous strain monitoring in the rock foundation of a large gravity dam. Rock Mech. Rock Eng. 16, 157-171 (1983). 9. Fritz P. Measuring the forces in a tunnel lining due to large shear and plastic deformations. In Proc. The Role of Rock Mechanics in Excavations for Mining and Civil Works, Zacatecas, Mexico (Edited by R. Sanchez-Trejo and R. de la Llata), pp. 441-452 (1985). 10. Rossi M. and Bazzi G. Two simple reinforced concrete beam elements for static and dynamic analysis. IA BSE Colloquium on Advanced Mechanics for Reinforced Concrete, Delft, pp. 397-412 (1981).
25 Dynamic Indications of Rock Mass Failure THOMAS VLADUT Hydro Environmental
Research Group, Calgary, Alberta,
25.1
INTRODUCTION
25.2
ENERGY RELEASE ASSOCIATED WITH ROCK FAILURE
25.2.1 25.2.2
Canada 695
Background on Microseismic Emission Microseismic Monitoring as a Geophysical Method
696 698 701
25.3 TECHNOLOGICAL ELEMENTS FOR MONITORING OF DYNAMIC PHENOMENA 25.3.1 Instruments and Equipment (Principles) 25.3.1.1 Geophones 25.3.1.2 Preamplifiers 25.3.1.3 Cables 25.3.1.4 Data processors 25.3.1.5 Output 25.3.2 Microseismic Monitoring Systems
704 704 704 704 705 705 705 705
25.4
707
EXAMPLES OF DYNAMIC MONITORING
707 708 708 711
25.4.1 Applications of Dynamic Monitoring 25.4.2 Examples of Field Observations 25.4.2.1 Underground mining 25.4.2.2 Surface mining 25.5
RETROSPECT ON DYNAMIC MONITORING OF ROCK MASSES
713
25.6
REFERENCES
714
25.1
INTRODUCTION
Engineering interest in rock mass failure is usually related to the desire and the need to predict such phenomena as a means of early warning. Energy release has been linked to rock failure under various conditions ranging from laboratory specimen testing to microseismic activities, commonly related to deep mining and earthquake activities. To the engineers and geologists concerned with the behavior of geological materials, microseismic monitoring provides a distinctive method for understanding ground deformation and failures. The phenomenon of microseismic emission has been employed to evaluate and predict impending failures of rock, both surface and underground. It has been employed to locate highly stressed regions, and to estimate the effectiveness of the practical measures taken to mitigate geological hazards like rock bursts, outbursts of gas and other failure-associated phenomena. In rock and other ground materials the origin of microseismic emission is not yet fully understood. It appears to be related to processes of deformation and failure, and to be associated with a sudden release of strain energy. The sudden release of stored elastic strain energy generates an elastic stress wave which travels from the point of origin to a boundary where it is observed as a microseismic/seismic event. Underground miners have long recognized this sudden anomalous change or 'rock talk' and the relationship with popping or cracking, and the potentially unstable conditions of the surrounding opening. 695
696
Back Analysis Monitoring
The term 'microseism' derives from the Greek 'seismos' meaning earthquake and 'micro' meaning small, for 'small earthquake'. The microseismic energy release may be an audible or, more frequently, a subaudible noise, resulting from fracturing and slipping of rock materials. In the area of geomechanics, microseismic monitoring has been done with field measurements with very different domains of applications: rock testing, hard and soft rock (coal, potash) mines, gas storage reservoirs, mined caverns for fluid storage, slope monitoring, dam foundations, hydrofracturing, earthquake prediction, etc. Difficult field conditions, uncertainties of equipment performance and reliability, and uncertainties associated with the origin of microseismic activity, have resulted in a number of unrewarding field tests. However, it should be noted that many associated laboratory measurements provide useful input in the process of utilization of the microseismic technology for engineering and practical risk identification applications. Monitoring of dynamic phenomena of rock mass failures concerns a large domain of energy release and the observational procedures have many common elements, but with significant differences generated by different levels of energy and associated specific technologies. Rock engineering in practice, even in a safe ground environment, often deals with stress concentrations and associated failures. Engineering practice requires the evaluation of the pattern of failure development which, when intensive, may affect the purpose of the project. Estimation of hazardous releases of energy confronts a number of unknowns in which neither the state of stress nor the properties of rock materials are fully known. The monitoring of rock masses is an observational method in which the dynamic elements cannot be disassociated from the traditional rock engineering assessment. This chapter presents some of the microseismic contributions in relation to ground control for engineering practice both on rock and in rock (surface and underground).
25.2
ENERGY RELEASE ASSOCIATED WITH ROCK FAILURE
Mining and some civil engineering activities alter the stress equilibrium of the surrounding rocks, thereby sometimes triggering undesirable failures, mostly of a local nature. The process of failure is always related to the release of energy, which, if not fully absorbed by the surrounding material (often soft rocks with light dumping ability), can be monitored. The monitoring ability is related to the released energy which may be sensed by specific instrumentation. Microseismic monitoring is based on the principle that rock failure is preceded by gradual microscale energy emission development and early detection of these emissions can be used to locate any build-up of critical stress. The emission of energy is divided in two groups: seismic and microseismic. The two differ from each other by the size of the energy emission: large energy emissions are measured by their magnitude on the Richter scale; microseismic and low energy level emissions are measured in terms of the frequency (number count) of the released energy. The physical significance of both emissions are the same and are related to the intensity of the fracture development. Considerations of the direct relationship between energy release and fracture development relate to several objectives of rock mechanics; in particular, the brittle failure of rocks (see Jaeger and Cook [1], updated by Hoek [2]) and the dynamic parameters of rock materials. Understanding the dynamic response of rock masses [3,4] is of paramount importance in the interpretation of elements collected by instrumental monitoring. To utilize seismic and microseismic measurements as an early warning of rock failure, the interpretation of the physical meaning of such energy releases is useful. An initial estimation can be made using the Richter approach in defining the energy release associated with the seismic activities. OF=10M"3
(in mm)
where OF represents the induced offset of a failure fracture of 1 km length and M represents the strength or the magnitude of a seismic event on the Richter scale. Fractures of 1 cm correspond to a magnitude of four events, 1 mm to a magnitude of three events, 0.01 mm to a magnitude of one event (Figure 1). The above simplified definition is in agreement with regard to the order of magnitude with the results of seismological observations made on a large scale in which fractures at the crust level of the earth are considered. A more detailed assessment of the size of fractures in the rock mass should consider better approximations provided by seismological observations in which the fracture size may be estimated
697
Dynamic Indications of Rock Mass Failure 10 times
V-
Λ/ = 6 I m - 100 cm — 1000 mm
10 cm
100 mm Λ/ = 5
M =5
V
I mm
b—
100 t i m e s
^
100 m m 10 cm
+*
pJlO mrT P I cm
*·-
/tf=4
M=7 10 m -
1 m $
Μ'β
10 m
I time
--
100 m ■
1000 times
Λ/=8
Figure 1 Fractures associated with seismic events on the Richter scale
by 1.75 log OF = M — 3.35 [5] or similar evaluations, instead of the expression of log OF = M — 3 given by the above approximation represented in Figure 1. Qualitative interpretation of the intensity of the fracture development from seismic data may not be fully applicable to a lower range of energy emissions. In this case, it is essential to take into consideration the nature of the material. Although the actual interpretation öf fracture intensity is related to seismic data, it should be used only as a reference to evaluate the order of magnitude of the induced fracturation. Using data on fracturation, the energy emission associated with fracture development may be evaluated from the same approximation and the magnitude of the event is expressed in relation to the size of the induced fracture for a dislocation of 1 km length: M = 3 + log OF (OF in mm). An alternative estimation of the energy release associated with seismological observations may be given by the relation M = 3.35 + 1.75 log OF, or other seismological data [5]. Certainly such a seismic energy assessment does not relate to the energy release of a particular event, but to the total energy released which may occur in several patterns. This assessment is valuable if the release is a singular event (which is seldom the case) but is of little use for the early warning of failure events. If the release consists of uniform energy emissions, such as the microseismic activity, a general frequency of emission can be estimated. Because of the unknown manner of the pattern of energy release, identification of a failure mode is important. Eventually if the failure mode is estimated, which is sometimes possible, simple rock mechanics (such as laboratory testing) can supplement information on the expected pattern of the energy release. An evaluation can be made by monitoring events in an unstable area and by further estimating the energy per event. Such interpretations and knowledge of the fracture mechanism are critical in developing a reliable, early warning procedure for any microseismic monitoring. Most of the microseismic energy released in fracturation is elastic energy (as in an overstressed spring) and the frequency of these emissions is significant. It has been established that seismic events induced by mining activity radiate energy of a magnitude ranging from microseismic motions to large bumps and rock bursts corresponding to a seismic magnitude of 5 on the Richter scale. The strongest mining-related seismic activity recorded took place in March 1989 at a potash mine near the Werra potash-mining region where activities of magnitude 5.7 on the Richter scale were associated with significant ground control instabilities. Seismic activities are also connected with hard rock mines, gold operations and coal mines. It is, however, difficult to separate natural seismic activities from man-induced seismic activities near active mining areas. One of the problems is that developments are often located in seismic regions; also, blasting is always associated with mining operations. Recent significant microseismic and seismic activities are known to take place in deep hard rock mines in Ontario (Canada). The technological approach of improving utilization of microseismic emissions is related to improvement of the early warning through the detection of signals, their frequency and source of location [6].
698
Back Analysis Monitoring
The earth is a permanent source of microseismic activity, e.g. temperature variations, ice formation in ground materials, groundwater flow, and industrial activities. In analyzing the variation of emissions it is assumed that there is increased microseismicity with increasing energy release. This concept is not new and research into the prediction of larger seismic activities, such as earthquakes, is in progress. Historically, the impoundment of large reservoirs of hydro power plants represents the first type of engineering activity that significantly altered the stress in the local state of the earth's crust. The stress change in such reservoirs is sometimes related to reservoir-induced seismicity (RIS). The highest magnitude of any man-made activity recorded on the Richter scale was 6.5 at the Koyna Dam in India. From an analysis of stress development, at any specific site, it is possible to identify the factors leading to such triggering and it is possible to predict potential seismic risks; some aspects of this are detailed in the chapter on 'Man-made Induced Seismicity' (see Volume 5, Chapter 23). A review of experience gained in other engineeringfieldsis useful in outlining safety measures, and to understand microseismic and seismic activity. The evaluation of triggering mechanisms in mining could be rendered more precise through the identification of stress conditions in mines having similar rock burst problems. Other man-made seismic activity can also be reviewed for the purpose of evaluating critical threshold conditions. Audible warnings, 5-10 s prior to failure, have been reported in the vicinity of mine outbursts. The typical time lag of an audible warning is insufficient for the implementation of effective mine safety mitigative procedures. Detection of higher frequency noises developing from small cracks has been suggested as an alternative to extend the warning time. Field observations suggest that the warning time may be increased to at least 15 s prior to failure. 25.2.1 Background on Microseismic Emission Stress build-ups lead to different scales of fracturing (micro to macro) of the rock material, and cracks develop and propagate along preferential paths; energy emission (micro to seismic) is released, part of which could be monitored. Energy changes brought about by making an underground excavation as the strain energy associated with crack developments and the dynamic nature of stress are part of fundamental rock mechanics assessment [2]. Microseismic emissions are characterized by some basic properties such as wavelength, cycle, frequency, velocity and intensity. Microseismic terminology is often similar to the terminology associated with acoustic emissions and standard definition terms are often used [7]. Some special features of the acoustic properties of rocks and the relationship to mechanical properties should be considered and are detailed elsewhere [8-10]. Propagation of seismic waves in rocks is mainly an elastic phenomenon determined by rock density and elastic constants. These properties, together with the nature of the seismic wave (longitudinal, transverse, etc.), determine the velocity of propagation. As the wave emanates from a point source, there will be a real decrease in the total energy of the wave due to 'friction' or 'damaging' the genetic properties of the rock. The spectral nature of the source determines the frequency of waves produced and thus propagated. According to frequency, elastic waves are divided into infrasonic (under 20 Hz), audio waves (20 to 20 000 Hz), ultrasonic (more than 20000 Hz) and hypersonic (more than 109 Hz). Microseismic or acoustic emission is the name given to high-frequency elastic energy radiation generated in rock material under stress. Infieldstudies the usual frequency of the radiation that can be detected is in the range of 20 kHz to 10 MHz. In the earth, elastic radiation generated by an earthquake has a frequency of 10 Hz. It is similar to microseismic emissions except that the frequency is much lower, and the amplitude of signals is much greater. Depending on the type of strains, since the elastic waves represent strain propagation, waves of different kinds may arise: P-waves longitudinal (compression) or S-waves transverse (shear strains). The velocity of elastic waves is determined by the elastic constants and the density of the rock in which they are propagated. In practice, velocity does not depend on frequency, so any frequency of vibration could be used to determine velocity for rock materials. However, attenuation is dependent on frequency; low frequency waves are quickly damped (seismic waves). Thus absorption (friction, dumping) and dissipation are responsible for this loss of intensity. The loss of elastic energy in the rock is related to the fraction between the vibrating particles, to heat transformation and to dissipation related to heterogeneities (fractures, pores, etc.). The amplitude of waves involved with
699
Dynamic Indications of Rock Mass Failure
such rigid behavior is related exponentially to the distance traveled by the waves. The absorption coefficient of elastic waves depends on the properties of the rock (elastic, thermal and internal friction) and on the frequency of the vibrations. Experimental data on the absorption of elastic waves in rocks show a linear dependence on frequency, as opposed to the quadratic dependence on frequency in metals and other materials. In clay rocks, absorption is proportional to the logarithm of frequency. There is evidence that absorption in rocks is mainly due to diffusive dissipation. The acoustic absorption coefficient is always greatest in rocks in which the velocity is the lowest. The monitoring of microseismic emissions is also affected by the capacity of rocks to reflect and refract elastic waves. The factors governing these properties are known as the 'specific acoustic impedance', which is the ratio of wave pressure to the velocity of the vibrating particles. Often the acoustic impedance is expressed as the product of density and velocity. Reflection and refraction take place either at the interface between rocks of different properties or when waves pass into (or out of) rock from an external medium. As a first approximation, the laws of geometrical optics are applied. Most of the acoustic properties in rocks depend on elastic characteristics. The velocity of longitudinal waves increase as Young's modulus and Poisson's ratio increase (up to 45%). The maximum velocity in gabbro or basalt will not exceed 6000 to 7000 m s - 1 . Velocity is also influenced by the grain size of the rock (it is generally greater in fine-grained rocks than in coarse-grained rocks). The selection of proper instrumentation to monitor seismic emission is related to the expected range of frequency. It is difficult to have instrumentation that covers a broad range of frequency and detects all emissions caused by fracture and microcracking. Most microseismic events have low levels of energy release. Table 1 shows the range of frequency connected with different dynamic monitoring procedures. An example of the recorded number of microseismic events from a field monitoring program is presented in Figure 2 and detailed in Section 25.4.2.2. Energy release is that part of the (stored) strain energy of the body that is released by a given crack propagation event. Much of this energy is consumed non-elastically, e.g. by the crack propagation process, if the rock is of a soft nature or is less brittle. The stored strain energy available for release is an accumulation of static elastic energy, and a portion is converted to waves of dynamic elastic energy. Microseismic monitoring thus yields information about the elastic state of the rock only. It is difficult to evaluate the state of a particular rock by monitoring the dynamic energy release which is only a part of the balance of energy involved in the failure. Generally, it should be recognized that the relationship between the component of deformation, and the corresponding energetic expression of deformation of a microseismic release is not fully known. Improved definition of this relationship allows the utilization of microseismic measurements as a reliable means of evaluating the process of failure. Utilization of the energy approach on rock failure requires knowledge of the specific details of waveform emissions, particularly the peak amplitude and a complete evaluation of the duration of Table 1 Frequency Range and Type of Data for Dynamic Monitoring Principal range of frequency
Applications
Particular range of frequency
Special applications
Type of measurements
Parameters
10-M00
Earthquake studies
—
—
Seismology
Magnitude Number of events
10°-7 x 101
Exploration seismology
5-10 2
Microearthquake studies
Seismology
Energy Number of events
10 2 -4 x 104
Microseismic 4 x 102—5 x 103 laboratory and field studies
5 x 104-5 x 105
Limited field microseismic
studies
8 x 10 4 -3 x 106
Early microNumber of seismic events laboratory and Microseismic field studies amplitude Microseismic energy
Rate of emission Acumulated activity Energy Energy release rate Energy per event
Acoustic emission for metal studies
Different frequency specific to materials
700
Back Analysis Monitoring Cardinal
River
Coal Mine 1809
|!9. 3
1000 800
£
600
142 0.98
400
X
0.88 0.4
I J2_ pr=: . I L 0.55 0.51 0.74
40 ,_ c
I I 21
M
12
27
21
D : distance (events-origin) origin is location of geophone 7 nr: number of microseismic event.
2 0.45
2 0.47
200 154 0.96 0 0
Number of similar events Relation to other sources a
15
16 2J
145
I 2 A Q5Î °: 6
I 54 63
I 82
Z.
80
20
33 °' 5 Ô9Î
J_
88 95
Events with identical coordinates
96 108
Events with 10ft distance
4
-Number of close events
Events with 3 0 f t distance —
160 21 10
|
! i
20
120
c
0 4 7
DM : day of m o n i t o r i n g 1 - 2 8 NS : number of similar events (scale 0 - 2 0 0 and 0 - 2 0 )
i
Events with 5 0 f t distance
DM
TT 16
nr a + nr b + nr c + nr d nr<7+ nrÄ + nr c + nrd
13 a a a a a
= = = = =
I to 0.4 I unique event 0.8 up to 10ft 0.6 up to 3 0 f t 0.4 up to 5 0 f t
Figure 2 Events recorded during pit slope instability monitoring (1 foot = 0.3048 m)
the events. The duration is generally well assessed by actual routine microseismic measurements which monitor event rates (the number of events per unit of time) or accumulated activity by detecting the number of events during a specific time interval. The relationship between emission rate and loading cycle from fatigue tests (Figure 3) may indicate failures without warning or gradual microcracking. It should be underlined that most of the actual measurements are only focused on microseismic rates or on accumulated activity. Measurements of the amplitudes of emission, and other parameters of waves have recently been used in more sophisticated monitoring systems. Great variations in energy and frequency of emission can be expected, depending on the volume of rock and the mechanism of failure. A typical microseismic emission indicates that individual events are made up of a wide range of frequency components. The usualfieldtests refer to frequency ranges between zero and 5000 Hz. Frequencies over 5000 to 10 000 Hz are present but increased travel time or distance alters the detection of high frequency components. The volume and the properties of rocks (particularly brittleness) relate to the release of stored stress energy. Energy released can vary from the strong events felt over large areas, to the very small events. In general, the larger the event, the lower the frequency. For all events most of the energy is contained in the lower frequency components of the microseismic wave. Most of the energy associated with microseismic modification of mine geometry is expected to be in the shear portion of waves. The microseismic emission from a rock subjected to an increasing restress indicates a recollection of stress that existed prior to the new increased state of stress. This recollection phenomenon, known as the Kaiser effect or 'rock memory', is associated with the existence of little or no emission until previously applied stress levels are exceeded. Once the level of initial stress (often of a historical nature) is exceeded, a significant increase in the rate of emission occurs. Exceptions have been noted in tests at very high stress levels and for cyclical applications. The Kaiser effect does not necessarily occur at the exact previous maximum stress, and variations may be related to the duration of the stress and the relation between stress levels and ultimate stress. The Kaiser effect, i.e. the memory held in a sample, has the important advantage of estimating former stresses by monitoring microseismié emissions and the effect is only dependent on stress. Independence of stress evaluation from strain measurements implies that the evaluation does not require detailed knowledge of rock mass particularities or constitutive models of behavior. These elements provide a cost-effective new potential for the estimation of existing stress conditions with the opportunity of considering the influence of confining stress [12] and effectively enhancing means to design underground structures such as caverns, where knowledge of the initial stress conditions
Dynamic Indications of Rock Mass Failure
701
Figure 3 Microseismic emission from cylindrical specimens of Tennessee sandstone under uniaxial fatigue compressive stress tests (AE = acoustic emission) (from [11])
is of paramount importance [13]. A note of caution, considering that mostly uniaxial loading observations prevail, no unique solutions may be associated to triaxial stress conditions. Similar caution must be addressed in association with the effect of water, temperature and retention time (days up to a month) on the stress level identified by the Kaiser effect. As a final note, the nature of emissions from rock materials is complex and it may be necessary to consider petrophysical elements, e.g. electroseismic effects [14], etc. As petrophysical aspects must be considered, the nature of the enclosed liquids are often of significance, in parallel with other more complex physical phenomena. 25.2.2 Microseismic Monitoring as a Geophysical Method Microseismic emissions resulting from rock fractures and rock movements are detectable signals in rock bodies near the energy source. Monitoring microseismic emissions has the same benefits and disadvantages as other geophysical methods. Geophysical methods monitor modifications of natural or induced fields. As a whole, geophysical methods measure the modifications of specific fields to evaluate the properties or state of the rock body subjected to a natural or inducedfield.To refer to the properties or rock body shape, size or location, suppositions are made relating to selected parameters. It should be underlined that geophysical measurements focus on a singular type of information and are supposed to provide several characteristics of the disturbant body. The most useful information in prospective geophysics is related to defining the shape of a disturbed rock body and information about the meaning of disturbance effect as a reference about the body properties. Because of the uncertainties involved in the interpretation of this type of information (geophysical) while there are several characteristics, hypotheses are proposed to obtain a solution to the objective of the investigation. One of the common methods used to solve uncertainties relating to geophysical information is to use several different types of geophysical measurements. For example, in mining applications there is a parallel use of seismic and electric measurements to estimate the excavatability of ground material, in the prospecting stages. Two other examples are mineral body identification and physical property assessment. Quite often, the utilization of different types of geophysical methods may encompass the utilization of both natural and induced field measurements. It should be noted that geophysical measurements generally observe the geological rock body under steady conditions. Microseismic monitoring is used alongside the mining process and is related to a more unsteady condition phase. Geological observations of formations undergoing structural changes may be evaluated according to the initial conditions and assessed by the initial measurements. Structural development may be of a mining or extractive nature. A typical situation isfluidextraction, such as in oil production, where the modifications are generated by the progress of steady depletion.
702
Back Analysis Monitoring
Solving the problem by utilizing a single geophysical method or using a multitude of methods requires correct interpretation of data. Definition of the geological body is related to: (i) the contrast of physical properties that are measured; (ii) the volume and orientation of the observed body in relation to the measurement; and (iii) the distance between the perturbing body and the point of observation. The following three suppositions should be made when utilizing geophysical measurements and should be kept in mind as a developing interpretation of microseismic monitoring data: (i) the monitoring disturbance has a significant value pertinent to the geophysical method that is used for investigation; (ii) the interpretation of the physical meaning of a measurement is related to the knowledge of the geological setting of the observed body; (iii) for bodies subjected to continuous modification, as generated by mining extraction, interpretation of changes in geophysical parameters is related to the knowledge of the induced variations by the mining operation. Several other aspects of geophysical measurements are of importance from different view of practice or specific methods and are detailed in Volume 3, Chapters 24 to 27. Methods for estimation (prediction) of the hazard of imminent rock failure may be assessed by mining and geophysical procedures. To note that because of significant differences of the physical elements associated with the particular method, stability estimation would differ and must be correlated for practical applications. Mining assessment methods are mostly comparative and it is necessary to have a good knowledge of the mining and existing geological conditions, with reference to spatial development, record of mining work cut in the rock mass, their influence on the stress concentration and implication of extraction technology. From these complex contributory factors and their relationships, and by analogy with previous cases of failures, e.g. rock mass bursts, a degree of hazard could be determined by comparison. The energy balance of the mining development may give a synthetic representation of the relations existing between the mechanical properties of the surrounding rocks and evaluate a relative proportion of energy types present in the process. The principal energy types involved are: total energy imparted during the process, elastic energy accumulated in the rock material, energy lost in permanent strain, thermal losses, and energy associated with the loosening process and dissipation. An evaluation of the stability can be derived with the strain energy concept (Figure 4) [14]. A simplified model of the strain energy distribution in rock structures indicates the extent to which a certain area is influenced or controlled by nonuniformities in the rock and the geometry of the structure and the loading system. This type of model emphasizes the tridimensional nature of a dynamic source of energy release. The quantity of the strain energy that can be stored safely within elasticity limits, and the material that forms an individual zone of influence, is limited. When the energy reaches the limit, the material in a specific zone of influence will fail. This may happen either because of load change, or because of modifications of the material properties, or both. When the material that affects a zone of influence fails, the unbalanced part of the energy in the loading system is transferred to other influence zones.
£" A - stored energy Ec- energy limit 5 -stability S=
(EC-EA)/EC
Is -instability Is-
1-5
Strain
Figure 4 Strain energy and its relation to the stability-instability definition (from [15])
Dynamic Indications of Rock Mass Failure
703
This leads to chain series of failures when the amount of transferred energy causes the energy of another appropriate zone to reach the limit and the amount of released energy increases proportionally. Each energy transfer translates to a rapid deformation (elastic response) of the parts of the system that acquire the energy. This elastic response constitutes a source of seismic waves as a part of the phenomena of microseismic activity. Consequently, seismic waves generated by each failure carry the information regarding the response of the components of the strain energy system surrounding the failure, and the loading status of these components. Seismic waves travel through the structure, away from the source. The velocity of the waves and the attenuation along the transmission paths provide information about conditions between the source and the wave detection point. Microseismic activity provides a direct indicator of the instability development, identifying conditions at the location where a particular failure occurred, and the characteristics of the rock along the wave path between the failure (seismic source) and the point of detection. The reliable analysis of microseismic activity requires a precise and fast location technique. The location mode is based on the first peak amplitude arrival time which best represents the nature of the source or the traditional first arrival signature. The precision of the location is routinely the best inside an array, and can be improved by considering the distribution of nonuniformities in the structure, given by particular geological structures which may interfere with refracted waves dominating the initial parts of observed waveforms. The location of individual sources (associated with a series of individual failures) and the spectral distribution of their energies allows the determination of potential sources and directions of an instability development. The rate of seismic emissions and the rate of energy release in specific frequency ranges allow the evaluation of the tendency for a chain reaction, or triggering of more developed failure, to occur. Results of this evaluation should be compared against the rate of energy interference in an effort to extrapolate a time evaluation of an expected next major strain energy release. Motion picture type presentations of mining development, facilitated by means of computer-aided design (CAD), parallel with the microseismic activity location may identify paths of the activity trend development, and add a dynamic means to the safety evaluation procedure. An average propagation velocity from an individual source to each point of observation could be found as a by-product in the source location procedure. This allows the mapping of velocity distribution and the monitoring of its changes with time. Microseismic core logging associated with a fragmentation procedure indicates a corresponding deterioration of the rock structure [16]. Similar interpretations can be applied to the attenuation spectrum to estimate the strain energy transfer and its concentration outside the area of microseismic activity. A more evolved monitoring trend is associated with a full, dynamic, extended observational procedure of geotomographic imaging. Geotomography is a technique that allows seismic waves to be used to gain information about the internal structure of rock masses, and is similar to medical scanning procedures. Imaging of rock mass through time may be carried out on wave velocity (compressional and shear) as well as attenuation and systematic imaging may reflect, in situ modification sensed by remote means. Particularities of seismic tomography are related to variable velocities and are not straight ray path associated. Iterative procedures are used from straight ray solutions and data inversions corresponding to slowness imaging or imaging of attenuation parameters instead of velocity, in which travel times are observed. Remote sensing of rock masses is extensively used in association with hydrocarbon exploitation procedures (migration, enhanced oil recovery, etc.) but is a relatively new tool in mining [17]. Minewide seismic networks will provide source location and three-dimensional images of velocity distribution and variation. The energy source would be naturally microseismic energy releases and energy from artificial investigative sources in use, to create geotomographic images of velocity attenuation ahead of advancing mine fronts. Development of larger scale geotomography three-dimensional surveys requires expert concentration of human and hardware (triaxial accelerometers, pre-trigger memory, etc.) of extraordinary size for today's mine technology. An example of today's analysis of rock properties and structures by cross well measurements is presented in Volume 3, Chapters 26 and 27. Earlier applications associated with the spatial characterization of rock masses by cross hole analysis were used in petroleum and dam engineering applications, but often involved tremendous computations which were difficult to process without today's computer technology.
704
Back Analysis Monitoring
25.3 TECHNOLOGICAL ELEMENTS FOR MONITORING OF DYNAMIC PHENOMENA 25.3.1
Instruments and Equipment (Principles)
Seismic, microseismic and acoustic emission monitoring systems perform essentially identical functions. They all detect and record ground motions produced by failure or displacement within a rock mass. In recent years, monitoring equipment (seismic and microseismic) has undergone improvements with significant technological innovations. The basic components of a seismic or microseismic system include three major components: the transducer (geophone or accelerometer) which may include a transducer and a preamplifier, a signal conditioning unit, and a recording output. Updated information on equipment and associated procedures is a must when selecting microseismic monitoring equipment. This section presents some of the basic elements of instrumentation without detailing state-of-the-art or specific criteria for selection for which reference should be made to manufacturers' specifications and specific literature [8-10]. A team consisting of a mining professional and a high-tech professional geophysicist should select the equipment for any application. The two elements to be considered are: (i) contact with the equipment manufacturer to evaluate system performance; (ii) contact with other users who have experience in the utilization of the particular system. Support by high-tech professionals is essential when developing a seismic or microseismic system because there are specific problems related to calibration, operation and interpretation. Other problems encompass trade-offs between sensitivity, frequency response, gain distortion, stability and drift. The role of the professional user is to decipher the variability of data and to correlate the events with the failure process. Some elements of monitoring equipment are of significance and are presented briefly below.
25.3.1.1 Geophones A point in a rock mass moves as a seismic wave passes through the point. These ground motions can be measured with respect to rock particle displacement, velocity and acceleration with time. Velocity gauges and accelerometers are used as tranducers for geophones. The geophone case, attached to the rock mass, moves with respect to its contents when the rock mass is vibrated by a passing seismic wave. Velocity gauges consist of a case containing a suspended coil and a fixed magnet. Ground vibrations cause the magnet to move with respect to the coil in a way that generates a voltage proportional to the particle velocity, or rate of particle displacement. Velocity gauges are constructed to respond to specific vibrations, therefore, the gauge orientation must correspond to the direction of motion to be sensed. Accelerometers usually consist of a piezoelectric material attached to a suspended mass such that the motion between the case and the mass generates a voltage in the piezoelectric element that is proportional to rock particle acceleration or rate of particle velocity. Some problems associated with the selection of types of transducers are of importance. Velocity gauges and accelerometers are most sensitive to vibrations in the 1-2500 Hz and 10-50 000+ Hz frequency ranges respectively. For this reason, velocity gauges are used for low to moderate frequency sources such as seismic exploration and microearthquake studies, and accelerometers are used in higher frequency applications, such as impact or crack detection studies. Because of the broad frequency-spectrum (10-50+ kHz range) of microseismic events, transducers in microseismic geophones are usually velocity gauges or accelerometers. Transducers can be procured with a wide selection of sensitivity and frequency responses. The choice between geophones as velocity transducers or accelerometers entails a trade-off between accuracy and cost. Accelerometers are preferred in relation to broader frequency response, greater dynamic range and more solid construction. 25.3.1.2
Preamplifiers
Virtually all microseismic installations use a preamplifier located at the transducer. For velocity transducers the preamplifier is necessary because the output is of a very low level, and the signal might be lost in extraneous noise picked up by the transducer cable and connections. For
Dynamic Indications of Rock Mass Failure
705
accelerometers the situation is worse because its impedance is very high and without preamplification any appreciable cable length will attenuate its output below usable levels.
25.3.1.3
Cables
In most dynamic monitoring applications, the geophone is placed at some distance from the processing unit; hence the geophone output signals must be transmitted. A number of transmission methods, including wireless and fiber optic, are in use for microseismic systems but presently the metallic electrical cable is the most reliable, economical and widely used. When running a long length, one should select a cable with as low a capacitance and resistance per length as is economically possible, considering size, weight and installation factors.
25.3.1.4
Data processors
Dynamic transducers are converted to electrical signals, preamplified and transmitted to a data processing unit. The data processor can be simple, consisting of only a post-amplifier and an output stage for some single-channel portable system. In complex systems, the data processor as in a rock burst monitoring system is usually very sophisticated. It may contain a signal conditioning section, a timing and control section, an energy section, a waveform analysis section, a computing section and an output section. Whereas the timing and control sections of earlier data processor units were made up of hardware components, recent trends are towards microprocessor-based systems. Where complex or lengthy calculations are necessary, such as in a three-dimensional source location system, additional capabilities are required.
25.3.1.5
Output
The output stage of a dynamic monitoring system can include a wide variety of display or storage options, such as audible, visual, numerical, graphical, storage, digital or connection to a data link. Additional computer capability is required for meaningful analysis of large volumes of data generated at a high rate; on-line minicomputers are used for most applications or for batch processing on an interrupt and priority-task schedule.
25.3.2
Microseismic Monitoring Systems
Early monitoring systems were developed and constructed in-house, but with a widening usage commercial units are now available. Now a number of manufacturers sell both components and complete microseismic monitoring systems. The type of equipment selected is influenced by economic limitations and the geological and mining complexity of the field situation. Such elements control the scale of instrumentation necessary for useful monitoring. It should be mentioned that the cost of instruments represents only about 30-60% of the total installation cost for complex underground instrumentation. Significant expenditures are related to materials such as cables, junction boxes, and other electronics which cost up to 10-20% of the total cost. Labor for installation (20-30%) and initial operation (10-25%) should be considered in a realistic cost estimate. The figures are general estimates and should be considered only as a guideline, and are in relation to the degree of sophistication and extent of computer support for fast data analysis. Examples of the analysis and comparison of two microseismic systems are included in Table 2. Single-channel systems. One of the earliest commercial microseismic monitoring systems consists of a geophone probe containing a crystal transducer and an impedance matching transformer, an amplifier, a headset and a counter. A small meter on the amplifier deflects when an event is detected and is also used as a battery and amplifier operating-level check. An additional output jack is available to record from or display the amplified geophone output. This type of single-channel system detects signals in the audiofrequency range of 20-15 000 Hz but is most sensitive to signals in the 1000 Hz range. A number of other single-channel microseismic or acoustic-emission monitoring units are available. Most of these later units are more sophisticated,
706
Table 2
Information on Equipment
Electro-Lab 250 MP
DyneerISprengnether Q-Log-AE Disadvantages
Advantages
Disadvantages
Advantages
Provides event time, arrival times, locations and some size estimate
No discrimination between P and S trigger waves
Not yet used in a mining environment; some usage in similar work
Provides virtually continuous monitoring
Sensors are limited to arrival time use only
Packages unit with or without a computer Programs for data evaluation have been developed by different mines 3-4 years of experience with the system at different Canadian mines Simple system with respect to internal operation
Sensors can easily be saturated Size estimate system under review
Provides events, times, locations with statistics, P and S wave properties, stresses, amount of energy released and could be extended if required Continuous system, originally designed for battery operation Internal component processors that are known Detailed technical information on events for further interpretation
Event acceptability criteria are possible for the sensors which record both P and S waves separately Multiple styles of source location programs available
At present, for a 15-channel system, standard storage is about 100 events for full output Built-in computer expected to have graphics, printerplotter, and increased memory Some program development may be required because more event data are provided
Location solution has many mathematically sensitive parameters which may lead to poor locations Technical backup for electronic problems
Requires clean power supply (60 Hz) and environment Cable costs may be high for a large numbers of sensors
Manufacturer will provide technical support, including sensor location if necessary Manufacturer is willing to modify to better suit mining needs
Input buffer holds one event and the rest are lost Event analysis time is 40 s for 15-channel system with full output
Back Analysis Monitoring
Lower installation cost than Q-Log-AE unit
Simple location analysis methods by limited input data
Expensive
Dynamic Indications of Rock Mass Failure
707
having variable frequency response filtering capabilities, data processing capabilities, and usually some form of printed or hard copy output. Multi-geophone systems. Portable and fixed systems are available for dynamic emission monitoring. These systems are basically identical to the single-channel systems except that 8 to 64 channels of microseismic data from one or more monitoring locations are obtained simultaneously. Source location systems, involve more sophisticated equipment including a arrival-time sourcelocation system. Short-period seismograph systems (SPS) will determine first arrival times, source location and energy number for large seismic events occurring on a site. However, the amplitude and duration of these events quickly saturate such units causing any number of echo events until the amplitude tails off and finally drops below thresholds. In addition, during mine blasting when both blasting events and microseismic events occur at high data rates, it is not uncommon to miss small events or to get scrambled arrival times for large events. Hence, a short-period seismograph that provides hard copy of all large mine-related seismic events, as well as local and distant earthquakes, is useful to adjust to traditional SPS installation. The short-period system can be very helpful in estimating energy levels of rock bursts. Seismic and microseismic networks. Microseismic stations are some distance (kilometers) apart. This explains the difference between microseismic network and seismic net working on a more regional scale. Microseismic stations, to monitor low energy events, are sited near probable focuses and therefore are close to each other, e.g. in the same mine or mine entry. Theoretically, to identify the concerned elements (the three coordinates of the source and the 'time of origin'), four stations would suffice; in practice there are seldom less than six accepted for reliability.
25.4 EXAMPLES OF DYNAMIC MONITORING 25.4.1 Applications of Dynamic Monitoring The discovery of microseismic emission was accidental and, at first, the significance of the discovery was not recognized. The discovery was made in a lead-zinc mine in northeastern Oklahoma in 1938 during experiments to determine whether or not seismic velocity in mine pillars was dependent on the pillar load. Awareness of the fact that the microseismic signals were caused by stress, even in perfectly stable pillars, was later discovered by tests in a northern Michigan copper mine [18] by Obert and Duvall of the US Bureau of Mines. The significance of microseismic monitoring is related to the fact that microseismic emissions are generated by high stresses and the detection of the emissions may not require knowledge of the mechanical properties of the rock or the state of stresses in the rock body. During the last decade, further developments in microseismic monitoring were best reflected at four International Conferences organized by Professor H. Reginald Hardy, Jr. and F. W. Leighton at Pennsylvania State University [8]. A better understanding has been achieved because of the variety of applications in other fields such as petroleum, gas, underground waste disposal, civil engineering and avalanche control. In mining, microseismic monitoring applications are increasing. However, in this respect, ground control for improved safety appears to be the principal area. The effectiveness of mechanical mining is highly influenced by geomechanical disturbances and their advance detection is expected to improve through microseismic monitoring. Many of these applications are designed to reduce the hazard of rock bursts, which is often related to seismic activities in mines [19]. Microseismic activities are also monitored in other domains where the risk of rock failure is of concern, such as slope stability in surface mining, underground operations and civil construction developments (grouting, flow) where the non-destructive testing (NDT) aspect is of significance. Several large scale microseismic monitoring systems operate around the world and several applications are described in the above-mentioned symposia or state-of-the-art reviews on mining applications [8-10, 19]. Resolution of such complex rock mechanics relationships is mainly developed through field testing and specific to observational methods applied in other rock and geotechnical practice should be considered as general principles for utilization of complex monitoring procedures [20]. In monitoring dynamic phenomena the number of events, microseismic or seismic, the expression to a convenient time rate (days or hours) is of fundamental significance (see Figures 2 and 5). The understanding of the significance of the time rates on the process related to the physical failure of the
708
Back Analysis Monitoring
rock body, like the stress-strain development, should be considered and is mostly used in association with the preliminary failure as a reference element. Usually such relationships are obtained during smaller scale observations in the laboratory. An example of such a relationship obtained in the laboratory is exemplified in Figure 2, for rock fatigue and stress variations on a sandstone specimen. Laboratory measurements complement field testing with the view to support and clarify aspects obtained through field monitoring procedures. Some studies have been conducted to investigate the fundamental aspects of microseismic emissions. Laboratory scale investigations have used microseismic emissions to define the point of unstable fracture in specimens under uniaxial and triaxial states of stress. It is important that laboratory studies are not considered to be fundamental studies. The controlled conditions of testing, environment or materials allow significant contributions to basic and applied studies. Laboratory measurements of dynamic processes differ from field studies mainly with regard to the frequency range at which emissions are monitored; also, the sophistication of the instrumentation employed is usually more developed than in field conditions. Some of the aspects of rock properties analysis by laboratory geophysical testing are presented in Volume 3, Chapter 24. Most important is the fact that the complexity of geological materials and the complexity of behavior related to stress, thermal effects, etc., can be simulated in simple laboratory approaches. Significant contributions should be mentioned by the expansion of the use of man-made material to simulate rock-like materials and the tendency to reproduce by simulation of field phenomena, such as coal outbursts or others of a more complex nature. It should be noted that laboratory scale experiments are most significant in the improvement of the rate of success of field applications and other developments on critical issues such as signal processing, hardware enhancement and several specific concerns (e.g. processing low-level signals), etc. One of the most used procedures is associated with the principle of counting dynamic emission and is related to the sensitivity of the system to discriminate high frequency emissions. Dynamic emissions, like other failure-related phenomena, occur in discrete increments irregularly spaced in time. An important feature of any dynamic monitoring is to discriminate between rock failure and background noise. Background noise is generated from other acoustic sources associated with mining operations, laboratory testing machines, or the weather. Because of this interference, high resolution observation equipment is required to detect rock noises and stress-related emissions. Most microseismic events are of low level with respect to energy release. Table 1 shows the range of frequency associated with microseismic and seismic emissions. It is difficult to evaluate the state of a particular rock by monitoring the energy release which is only a part of the balance of energy involved in the failure.
25.4.2
Examples of Field Observations
A systematic review of microseismic monitoring practice refers to field applications as the predominant direction of such developments. Underground mining, specifically hard and soft rock mining, rock bursting in hard mine and coal developments and a few surface mining applications would be noted as the prevailing direction of development. The included examples are biased by the author's own experience and the significant increase in mine-related operating monitoring systems. The examples presented illustrate the different degrees of complexity of such undertakings.
25.4.2.1
Underground mining
(i) Hard rock mining INCO's Creighton Mine (Ontario, Canada) has a history of rock bursts dating back to the mid19308. During 1986, 12 rock bursts at the mine were recorded on the Eastern Canada Seismic Network [6]. The mine's microseismic system located most of these events in the crown pillars of mechanized cut-and-fiU stopes between the 5400 and 6800 levels. In a major departure from previous mining practice, a distress slot is being mined in the center of the crown pillar of the 6800 level slopes.
Dynamic Indications of Rock Mass Failure
709
The seismic activity associated with this distressed slot is being closely monitored to evaluate zones of fracture around the slot. A fiber optics shaft cable, to transmit the microseismic signals to surface, has been installed. From the distribution of micro- and macroseismic events it can be inferred that three conditions exist in the rock mass: stable unfailed, characterized by a few small events: failing, where high stresses or significant deviatoric stresses cause both frequent and large events; and failed (post-failure materials), showing a relative absence of dynamic activity. During 1986, INCO Ltd also installed two additional Electrolab microseismic systems at the Garson and Copper Cliff North Mines. At Garson Mine seismic activity appears to be associated with dykes and auxiliary shear structures in the footwall. This is similar to the seismic activity in the adjacent Falconbridge Mine. Copper Cliff North Mine experienced rock bursts in 1986, especially after large pillar blasts. Seismic activity is significant (up to 2000 events in a weekend) after these blasts and is fairly widespread.
(ii) Mining supports A microseismic roof-fall warning system was field tested in coal mines in the eastern and western United States to better define the capabilities and limitations of the microseismic method in predicting roof fall. Microseismic event and energy type data were obtained and documented for 23 coal mine roof falls. Fields test results indicate that both the microseismic event and energy methods, as applied, have a high probability for successful application in a commercially practical roof-fall warning system [21]. The ideal way to achieve increased safety in underground mining environments is to remove the hazards by reducing the need for human involvement in hazardous activities. Acoustic emission based instruments situated underground provide means and become more the sensory elements of robotics instead of the humans working in dangerous work sites. The availability of powerful microprocessors makes possible the packaging and marriage of robotic systems and artificial intelligence into highly functional tools. Mining process control development is often based on microseismic intelligent sensors obtained through integration of the field monitoring process. Acoustic signals induced by a transducer pressed against bolts are transmitted through the roof bolts, which have different degrees of bonding (some are well bonded and others are poorly bonded), and signal variables known as discriminants are generated. These discriminants represent the signature of the bolt and its bonding. Once known, these discriminants, which contain differences generally too subtle to be classified by the human eye, are remembered by the computer and used to predict the bonding quality of other unknown bolts. In this manner, an intelligent roof-bolt tester may be possible. Resin roof bolts are widely used in underground mining because, in general, they exhibit dependent roof control properties. However, they present a serious hazard because, once in place, there is no reliable way to test them. Because of improper installation or aging of the resin, some bolts loosen and fall. Prior research has failed to develop a device that can reliably measure the degree of bonding. At the Pittsburgh Research Center of the US Bureau of Mines, this problem is being tackled by the use of computer-based signal processing and analysis systems known as adaptive networks, similar to the technique developed to help cutters avoid cutting hard rock elements. The group of instruments where acoustic signals are of use could be exemplified by the Ultrasonic Spot Coming Roof Failure instrumentation (US Bureau of Mines, Spokane Research Center). An ultrasonic instrument has been used to measure roof-to-floor closures and rates of closure in several coal mines, as well as in mine roofs up to 8 m in height in salt mines. Thus far, distance measurements of 0.3 to 10.5 m where changes in distance of at least 12 mm are expected can be readily made. The instrument is effective primarily in measuring convergence during longwall mining of coal, pillar robbing and other mining operations resulting in large ground movements. Other possible uses, as in volumetric surveys and high roof measurements make the instrument a versatile tool for any mine. The small ultrasonic transducer, similar to an audio speaker diaphragm, is mounted to a roof bolt or other suitable location. The transducer is triggered by the readout device every 6 s and thus induced to send out an ultrafrequency wave through the air. The wave is reflected back to the transducer off any surface at which the transducer is pointed. The amount of time needed by the ultrasonic wave to make the round trip is measured by a precise oscillator/timer in the readout unit and is converted
710
Back Analysis Monitoring
into a measure of distance. Changes in distance are also converted to a rate in length per minute. The two readings provide measurements with a resolution of 0.3 mm. Such instruments do not always monitor a natural emission (e.g. microseismic) but use induced wave measurements. The examples were included to refer to the wide range of applications of dynamic measurements in mine environments which often address more complex mining sequences such as sensing when continuous miners and longwall shearer are straying into strata other than coals. When the machine is cutting coal, it creates a strong seismic signal in the coal that propagates through the coal for great distances. By analyzing these signals, it can be determined when the cutter begins hitting rock. Such future monitoring systems would thus be used to manipulate the cutting head of the machine to avoid mining rock via a computer-controlled process. This means that computer output is transmitted to automatically control cutter motions of the mining machine, which would include the process to teach the computer equipped with some adaptive learning process capability, a technological mean under actual development. Enhancement of robotic systems and artificial intelligence will together provide new solutions to machine-control problems that previously were difficult, if not impossible, to solve. An example in this class of future technologies could be mentioned as being in development, for machines that will install permanent roof support ahead of the extraction of the supporting coal. Studies indicate that a roof support ahead of mining, will provide advantages over computercontrolled conventional roof bolters or intelligently-controlled miner-bolters systems. (Hi) Possibilities associated with more sophisticated equipment For some years the plotting and counting of located microseisms has provided a general idea of the relative stability of specific areas within a mine. A transducer, which may be either a velocity gauge or an accelerometer, senses the ground motion caused by a microseismic event and outputs a voltage that varies with the amplitude and frequency of the motion. An electronic circuit detects and times the initial rise of this voltage above background noise. When several (a minimum of four) transducers detect the same microseism, the event arrival time at each transducer can be used to locate the event. More complex information in the microseismic waveform, such as the spectral energy distribution, could provide material for further analysis and perhaps enhance the ability of mine personnel to evaluate microseismic activity. Improvements relate to both the reliability and accessibility of the microseismic data and to the ease with which mine personnel may install and operate the monitoring system. Field trials of a portable microseismic processor recorder (MPR) system constructed by Science Applications Inc., La Jolla, CA, based on Bureau of Mines design recommendations, were done at Galena Mine, USA [22]. The portable MPR monitored the Galena East end in parallel with the well-tested Denver Research Center fixed rock burst monitor (RBM) system. In the second set of trial runs, the portable MPR independently monitored a stope that had suddenly begun showing intense audible activity. The activity was recorded on both the portable MPR and the RBM. The broad features of the two records are the same, but the portable MPR counted more events in the same time than did the fixed RBM. The events were triggered by bolting and drilling for a 1.8 m backstop round. The fixed RBM had only single-event storage capability and had to pass these data on to a host computer before continuing with data acquisition. The count surplus is related to the differing storage capacities of the two devices. The portable MPR, if the host computer or the on-board cassette is currently occupied, stores up to 50 events in the on-board memory. The emplacement of the portable MPR at the new stope was part of a prompt response to a sudden increase in activity around the stope. Three events were identified as relatively large by a software routine that checks for triggering in a unique low-gain-system channel. The largest of these three events was a rockburst that registered on a seismograph on the surface some 1800 m away. In 16 work hours, a seven-phone network was on-line collecting data. The portable MPR, together with a set of variable-gain amplifiers and an oscilloscope, was installed in a wooden box some 180 m from the raise. The temperature outside the box was 32 °C, and high humidity. The equipment successfully logged in data for 60 days, until the low grade of mined ore and bursting problems dictated a stope shutdown. Several experimental units are in use in United States mines. The microseismic processor recorder (MPR) was designed to capture and process data from random microseismic events within a mine structure. The unit stores data in a nonvolatile read/write RAM cartridge and transmits the information to a remote data terminal from further processing.
Dynamic Indications of Rock Mass Failure
711
The unit will accept up to 32 acoustic sensor signals. A split system capability (1-16 or 17-32) is included. The unit is designed to operate in the mine environment and is housed in a rugged waterresistant suitcase enclosure. The use of state-of-the-art low-power CMOS logic minimizes power consumption. It operates from commercial power but has an internal battery which allows 2 to 3 h of recording when AC power is lost. The sensors that are generally used are crystal accelerometers. Sensors are placed in drill holes located in the vicinity of the mine workings. The sensor may include an internal amplifier that will generate a high-level signal that is suitable for transmission via cable to the centrally located microprocessor processor recorder unit. Time-of-arrival data, along with the sensor location coordinates, plus the velocity of sound in the media that are instrumented, are the data required for determination of the origin of the microseismic emission. Essentially, all that is required is AC power, and should this power be interrupted, the MPR will maintain the data display and the time of day clock for up to 72 h. The portable MPR provides data to a host computer through a standard RS-232 interface; alternatively, the MPR will run without a host computer and store data in an on-board cassette. Communication between the MPR and the computer is two-way and may be initiated by either the computer or the MPR. In the former case, the computer commands the MPR to transmit a block of data for each event stored in the internal memory. In the latter case, the MPR requests permission to begin sending data. If permission is not granted, the MPR returns to logging in data and stores the data in a circular buffer with a 50-event capacity. Data may be retrieved from the cassette and obtain the stored data through a cassette reader, which outputs either directly to a printer or to a computer. Through a system of automatic gain control, the portable MPR distinguishes drilling from other seismic events and eliminates it from further processing. Since each channel independently screens out drilling noise, one or two overly active channels will not cause the MPR to suppress genuine microseismic data.
25.4.2.2
Surface mining
Increased productivity in large open pit operations is related to maintaining good stripping ratios. In complex geological structures for open pit operations the maintenance of high stripping ratios is related to the condition of the stability of slopes. The maintenance of slope stability has an impact both on the economical input of each particular pit and on the risks related to production in quite hazardous conditions. Hazards related to pit slopes are usually related to complex geological structures. Risks are connected to hazards for man and equipment, both of which operate at the toe of unstable slopes. Risks are related to the whole economic output of each particular pit. Slope instability and failures develop as pits become deeper. In the application of microseismic methods for slope monitoring, concerns were expressed mainly in relation to the different sizes of instabilities. Expected application of the use of microseismic observations is to improve and optimize the unloading of potential failure areas in the removal of material from the top of the slope walls rather than the clean-up of slides from the bottom of the slide. Several microseismic applications are known for monitoring landslide activity. A second generation of microseismic instrumentation [23] developed for slope monitoring was employed for a large open pit operation in the foothills of Alberta. The field tests were a cooperative effort made by the operator and a joint USBM-CANMET team which evaluated the contribution of microseismic measurement as a supplementary monitoring system to the routine surveying system of electronic distance measurement (EDM) [24]. The system uses an array of eight geophones linked by cable to a radio transmitter connected with a receiver unit outside of the pit (by radio link) where the location of the seismic event is given on a printer, based on arrival at a minimum of four geophones. Approximately 12 000 events were identified as unique events and about 3000 as multiple events with source locations. The image of microseismic activity expressed by the number of events corresponds to the overall movement of the slope and the total displacement, as well as to the rate of displacement. The expected slope height was of about 144 m and an installation array of seven geophones was operating when the pit reached a depth of about 100 m. Significant movements were triggered as the operation developed from 100 to 140 m and the microseismic monitoring was in operation until the observation bench at 90 m depth underwent a significant displacement of 2.1 m. The dynamic aspect of the slope movement was important and reached a maximum of 12 m in the final stage, with strong distortion of the slope.
712
Back Analysis Monitoring
Implementation of microseismic monitoring of slope instabilities for an early warning system in open pit mines was demonstrated, but two supplementary steps should be mentioned prior to final recommendation for current operational use. Thefirstcomplementary completion step refers to the instrumentation itself, which needs better protection with regard to climatic interference, e.g. lightning, being well protected from the interference of common mining noises (blasting, transportation and excavation). The second complementary step is related to links to move powerful data processing system to provide the decision-making support in triggering the early warning signals in the pit area. This would require a more sophisticated software location and significant components
| \//\
| / A - t o t a l a c t i v i t y index (events h ') IQ- locatable a c t i v i t y index
7mm
7
Θ
9
10
II
Time (days)
Figure 5 Distribution of microseismic activity (IA, 7B) and the rate of displacement of the slope 100
■ ■ |
Unique channel events : 3179 ( 2 0 . 4 % ) |
Y/\
Multiple
channel events 12 3 6 7 ( 7 9 . 5 % )
T o t a l events: 15 5 4 6
V
Prisms
•
Geophones
uly 1984) Geophone No.
Ever ts No.
%
1
5600
2
2172
100 38.7
3 4
1706 4123
30.4 73.6
5 6
3871 5379 1247
69.1 96.0 22.0
7
July
1984)
Figure 6 Microseismic slope monitoring: distribution of multiple and unique types of events (Γ = 0.3048 m)
Dynamic Indications of Rock Mass Failure
713
related to the interpretation of the data arising from the microseismic source and the fracture development process. Monitoring instabilities of slopes allows a better evaluation and definition of instability developments. This is reflected by the variation of the safety factor of the slope during the mining operation and extensive interpretation of slope deformation by finite element method applications and the relationship to failure development, as indicated by the source of microseismic activities. The evaluation was supported by the survey data of the EDM and through this provides a realistic understanding of the risk development evaluation. A significant increase in microseismic activity preceded an important movement of the slope (Figure 5). The threshold of unsafe conditions for the slope movement was 7 mm h" 1 . Such risk evaluation and the impact of the microseismic methods are more difficult to develop for the estimation of the factor of safety for underground developments. In the Cardinal Rivers Mine slope monitoring it was possible to monitor a significant volume of unstable rock with a much bigger volume than in underground instabilities. During the observation period, unique types of events and multiple events were evaluated. The comparative analysis illustrates the rate of success that might be expected from multi-channel and single-channel monitors. Even though the number of singular events is relatively small (about one in four) in comparison to multiple events, the activity in the most active areas of the slope was still well defined by singular events (Figure 6). This may indicate that single-channel microseismic monitoring can be successful in a well-planned test in a very active area. If confirmed, such a conclusion would also support single-channel monitors as possible contributors to safety enhancement when mining in unstable areas. Such considerations are of significance for microseismic monitoring in both underground mining and surface operations when the tendency of increased points of observation in a large system is the most convenient practice. 25.5 RETROSPECT ON DYNAMIC MONITORING OF ROCK MASSES Backanalysis of the critical behavior of rock structures is a direct part of the observational method applied in geoscience and associated engineering procedures. Dynamic monitoring is mostly concerned with the energy release associated with failure phenomena, and sensing the rock mass response under dynamic loads induced to generate a response of the rock body for observation purposes, most often concerning evaluation of the state of the rock mass. Backanalysis provides an evaluation of the critical conditions (large deformations and failures) usually in order to decipher and enhance methods of designing and mining in use, and to foresee perceived or unperceived occurrences of such uncertain conditions. Usually the backanalysis develops after a certain time. From this standpoint, microseismic and dynamic monitoring are concerned with developments in the available time lapse of zero and the backanalysis procedure would focus on a real time evaluation. From this point of view, microseismic and dynamic observations are real time observations of the development of critical conditions. Even though no time lapse is available, the scope of analysis focuses on the same scope as any backanalysis procedure, to refer and decipher the mechanisms of development of critical conditions in the rock mass. Through specific observational means, mostly related to the gradual development of rock responses to stress-strain developments, it is intended that forecasts of behavior will be made possible by using energy emission particularities as a precursory signal of triggering critical conditions. The development of dynamic phenomena as a precursor of failure is a target not yet attained, in which sophisticated observations of all components of energy releases are to be used and synthesized for future enhancement of prediction-estimation procedures. Full utilization of highly computerized technologies could make microseismic means to be the intelligent sensors and support for utilization of observations of dynamic phenomena as an instant backanalysis means to generate safety assessment and early warning for evaluation of development of unstable critical conditions. The expected contribution of microseismic observations is enhanced by the possibility of development of critical conditions assessment without full knowledge of specific rock properties and of stress conditions. This expectation may be achieved only when identification of failure development patterns is fully combined with a background understanding of the evaluation of rock mass behavior by traditional rock engineering procedures. Lack of safe mining procedures could not be replaced with hopes for failure precursors to be discovered in the future. At actual stage, the microseismic observation technology should still be considered as a component of good rock engineering practice, mostly as a means for development of mitigative procedures to implement remedial or modifications of traditional engineering practices.
714
Back Analysis Monitoring
Dynamic monitoring of rock mass behavior represents a high technology domain and successful application is reliant on a multidisciplinary team contribution of practitioners with very different backgrounds. The extraordinary use of today's technology and the rate of development of such applications may reflect on the trend of ordinary future practice. Increased burden is placed on the rock mechanical engineer, who should use complex technological sources and several highly technological instrumentational means. His role could be remembered by the pertinent comment of one of the pioneers of observational methods, Dr R. B. Peck, 'An instrument too often overlooked in our technological world is a human eye connected to the brain of an intelligent human being' [20]. 25.6 REFERENCES 1. Jaeger J. C. and Cook N. G. W. Fundamentals of Rock Mechanics, p. 585. Chapman and Hall, London (1976). 2. Hoek E. Brittle failure of rock. In Rock Mechanics in Engineering Practice (Edited by K. G. Stagg and O. C. Zienkiewicz), pp. 99-124. Wiley, New York (1968). 3. Ambrasey N. N. and Hendron Jr. A. J. Dynamic behaviour of rock masses. In Rock Mechanics in Engineering Practice (Edited by K. G. Stagg and O. C. Zienkiewicz), pp. 203-231. Wiley, New York (1968). 4. Coates D. F. Rock Mechanics Principles, v. Mines Branch Monograph 874 (revised 1981 CANMET, Energy, Mines and Resources Canada, catalog No. M32-874/1981E). 5. Bonilla M. G. Surface faulting and related effects. In Earthquake Engineering (Edited by R.L. Wiegel), pp. 47-74. Prentice Hall, New Jersey (1970). 6. Brenhant C. H. and Hedley D. G. F. Annual Report of the Canada-Ontario-Industry Rockburst Project, CANMET (Energy, Mines and Resource Canada) publication SP87-7E (1988). 7. ASTM, Standard definition of terms relating to acoustic emission, E610-77. In Annual ASTM Standards, Part 11, pp. 603-605. ASTM, Philadelphia (1973). 8. Hardy Jr. H. R. and Leighton F. W. (Editors), Proceedings (1-4) Conference on Acoustic Emission/Microseismic Activity in Geological Structures and Materials, Trans Tech, Clausthal, Germany (1977, 1980, 1985, 1990). 9. Vladut T. I. State-of-the-art of microseismic applications for mining. CANMET Division Report ERP/CRL 86-l(TR), Micromedia, Toronto, #86-3506, pp. 1-92 (1986). 10. Blake W. Microseismic Applications for Mining. A Practical Guide. A Bureau of Mines Report, US Department of Interior, p. 206 (1982). 11. Khair A. W. A study of acoustic emission during laboratory fatigue tests on Tennessee sandstone. In Proc. 1st Conf Acoustic Emission/Microseismic Activity in Geological Structures and Materials (Edited by H. R. Hardy Jr. and F. W. Leighton), pp. 57-86. Trans Tech, Clausthal, Germany (1977). 12. Crawford A. M. and Hughson D. R. Kaiser effect gauging: The influence of confining stress on its response. In Proc. 6th Int. Congr. Rock Mech., Montreal (Edited by G. Herget and S. Vangpaisal), pp. 981-985. Balkema, Rotterdam (1987). 13. Hayasi B. M., Kanagawa T., Motozima M. and Kitahara Y. Detection of anisotropic geo-stresses by acoustic emission, and non-linear rock mechanics on large excavating caverns. In Proc. 4th Int. Congr. Rock Mech., Montreux, pp. 211-228. Balkema, Rotterdam (1979). 14. Antsyferov M. S. The electroseismic effects in rocks. In Seismo-Acoustic Methods in Mining (Translation Editor G. V. Keller), pp. 131-134. Plenum, New York (1966). 15. Descour J. Stability analysis and other applications of seismic waves. In Fred Leighton Memorial Workshop on Mining Induced Seismicity at the 6th Int. Congr. Rock Mech., Montreal, pp. 1-19 (1987). 16. Thill R. E. and D'Andrea D. V. Acoustic core logging in blast-damaged rock. Eng. Geol. (Amsterdam) 10, 13-36 (1987). 17. Young R. R., Hutchins D. A., McGaughey J., Towers J., Jansen D. and Bostock M. Geotomographic imaging in the study of mining induced seismicity. In Fred Leighton Memorial Workshop on Mining Induced Seismicity at the 6th Int. Congr. Rock Mech., Montreal, pp. 41-67. (Editor and Convenor for the Rockburst Sub-Committee of the Canadian Institute of Mining: Dr. R. Paul Young, 1987.) 18. Obert L. and Duvall W. I. Microseismic method of determining the stability of underground openings. In Bulletin # 573 of US Bureau of Mines (1957). 19. Gay N. C. and Wainwright E. H. (Editors) Rockburst and Seismicity in Mines. The South African Institute of Mining and Metallurgy, Symposium Series No. 6, pp. 1-363 (1984). 20. Dunnidiff J. and Derre D. U. (Editors) Judgment in Geotechnical Engineering, The Professional Legacy of Ralph B. Peck. Wiley, Toronto (1984). 21. Fisher Jr. C. and Thomson R. H. Microseismic Roof Fall Warning System Development, Bur. Mines O.R.163(l)-81, pp. 1-194 (1980). A US Department of Commerce Publication by National Technical Information Service. 22. Cauglin J. R. and Sines C. D. Field Trials of a Portable Microseismic Processor Recorder, US Bureau of Mines Information Circular 9022, p. 198 (1987). 23. Lepper C. M., Poland A. P. and Mullis C. T. A microseismic system for monitoring slope stability. In Bureau of Mines Report of Investigation R.I. 8641, pp. 1-64 (1982). 24. Vladut T. I. and Lepper C. M. Early warning of slope instabilities by microseismic monitoring. In Proc. 4th Conf. Acoustic Emission/Microseismic Activity in Geological Structures and Materials, Pennsylvania State University (Edited by H. R. Hardy Jr. and F. W. Leighton), p. 29. Trans Tech, Clausthal, Germany (1990).
26 Infrared Thermographie Observations of Rock Failure MINH PHONG LUONG Ecole Polytechnique, Palaiseau,
France
26.1
INTRODUCTION
715
26.2
MECHANICAL CHARACTERISTICS OF ROCK-LIKE MATERIALS
716
26.3 HEAT PRODUCTION MECHANISMS 26.3.1 Coupled Thermo-Visco-Elastic-Plastic Analysis 26.3.2 Thermal Conduction 26.3.3 Thermoelasticity 26.3.4 Intrinsic Dissipation 26.3.5 Heat Sources
717 718 719 719 719 720
26.4
720
INFRARED THERMOGRAPHY TECHNOLOGY
26.4.1 26.4.2 26.4.3 26.5
720 721 721
Infrared Radiation Infrared Thermography Infrared Scanner
INFRARED SCANNING AND ROCK ENGINEERING
26.5.1 Infrared Thermographie Observations of Rock Failure 26.5.2 Infrared Scanning of Rock Mass Thermal Properties 26.5.2.1 Aerial imagery 26.5.2.2 Ground-based imagery
722 722 724 725 726
26.6
CONCLUSIONS
729
26.7
REFERENCES
729
26.1
INTRODUCTION
Current technological developments tend towards increased exploitation of material strengths and towards tackling extreme loads and environmental actions. The tendency to extend the service life of structures by increasing maintenance, rather than replacements, increases the need for monitoring of structures and supports the need to perform global or local test loading. In addition to a knowledge of the ambient stress field, the in situ mechanical and thermal rock mass properties are particularly important in the case of radioactive waste disposal. If these properties vary with stress and/or temperature, the variations must be quantified. It must be noted that monitoring techniques have previously suffered from difficulties in the interpretation of recorded data. Geotechnical projects for tunnels or underground construction to provide storage sites, shelters for civil and military defence needs and others, employ rock and rock-like materials which exhibit cracking, softening and dilatancy in their ultimate behavior under compressive loading. The load sustained by the rock element begins to decrease if the deformation is further increased. Softening behavior is often considered unstable because the loading process is load-controlled. In the strain-controlled case, the descending load-displacement curve may be detected experimentally. This mechanical fact may be comprehensive and important in connection with design or regulations of geotechnical structures. Knowledge should therefore be obtained of the thermomechanical behavior of rock materials subject to both mechanical and thermal loadings. 715
716
Back Analysis Monitoring
Diverse damage analysis methodologies [1] for engineering materials have been developed in recent years which isolate the factors affecting crack initiation and growth, and enable the prediction of their cumulative effects on the fatigue performance of structural components. They are based on: (i) the formulation of analytical models for fatigue crack and growth; and (ii) the acquisition of supporting baseline data and the validation of such models by means of a comprehensive testing procedure. In cases where continuum mechanics can be applied, the concept of an effective stress 0"ef =
Infrared Thermographie Observations of Rock Failure
111
breakdown is accompanied by both loss in stiffness and accumulation of irrecoverable deformations. At the structural level, breakdown appears as microcracking and possibly as slippage at the aggregate-cement paste interfaces. Natural rock salt appears to be a discontinuous mass with a medium- to coarse-grained polycrystalline structure. It contains cracks, defects, joints, dispersed clay impurities and eventually bedding planes with varying degrees of cohesion along these discontinuities. It sometimes exhibits variable mineralogy and crystal sizes. Depending on its crystal structure, bond character, and the temperature, it may be completely brittle, semi-brittle or ductile [7]. Elucidation of the various mechanisms responsible for fracture is related to the plastic resistance associated with grain boundaries and the effects of plastic anisotropy. When it is loaded, the rock salt deforms also as a whole in spite of significant incompatibilities between the crystal aggregates. Stress concentrations occur and result in localized forces which are sufficient to promote plasticity and crack formation or both. Quite similar to the concrete case at the macroscopic level, breakdown is accompanied by both loss in stiffness and accumulation of irrecoverable deformations. At the structural level, breakdown appears as microcracking and possibly slippage at the crystal interfaces. Failure in plain concrete, rock salt or in any other rock-like material may be viewed as a microstructural process through the activation and the growth of one preexisting flaw or site of weakness, or through the coalescence of a system of interacting small flaws and growing microcracks. The stress level corresponding to the activation of the flaws is related to the flaw size and connected with the encompassing microstructure. Flaw-initiating concrete failure may be divided into two classes: (i) the intrinsic flaws develop during the processes of hydration and curing of the cement paste; and (ii) the extrinsic flaws result from significant incompatibilities between the aggregate and the matrix when the material as a whole deforms under applied loading. It may be said that fatigue of concrete is associated with the development of internal microcracks, probably both at the cement matrix/aggregate interface and in the matrix itself. For rock salt, it may occur at the crystal interface and in the crystal itself. The occurrence of microcracking and slippage leads to nonlinearity and softening in the stress-strain response of concrete with a marked dependence on the mean stress [8]. Under repetitive stresses, the fatigue mechanism is a progressive, permanent, internal structural change occurring in the concrete, rock salt or in any other rock-like material. These changes result in the progressive growth of cracks and complete fracture. The formation and propagation of microcracks have been detected by means of different well-known measuring methods. (i) The ultrasonic pulse velocity technique involves measurement of the transit time of an ultrasonic pulse through a path of known length in a specimen. The velocity of the ultrasonic pulse in a solid material will depend on the density and elastic properties of the material and it will therefore be affected by the presence of cracks. (ii) The acoustic emission method is based on the principle that the formation and propagation of the microcracks are associated with the release of energy. When a crack forms or spreads, part of the original strain energy is dissipated in the form of heat, mechanical vibrations and in the creation of new surfaces. The mechanical vibration component can be detected by acoustic methods and recorded, hence microcracking may be detected by studying sounds emitted from the rock-like material. It can be considered that the failure mechanism of rock-like materials consists primarily in the formation and propagation of microcracks. The formations of microcracks are often associated with points of stress concentration. The stress concentrations are located onflawspresent in the material, or on existing cracks and notches. Cracks initiate quite early at a site of weakness or stress concentration, then propagate through the plastic zone and into the elastic region. In other cases, flaws are inherent in the material owing to the process of fabrication of concrete or crystallization of rock salt. These defects exist prior to the application of any load. There may be some initiation period during which the material, at the tip of the flaw, undergoes dislocation pile-up, microvoid formation and coalescence, etc. prior to the onset of progressive growth with increasing plastic deformations. The work done in plastic deformation results in the generation of heat, readily detected and analyzed by infrared thermography. 26.3 HEAT PRODUCTION MECHANISMS Cases in which heat is produced in the materials are becoming increasingly important in technical applications. Heat may be produced by the passage of an electric current, dielectric or induction heating, radioactive decay, absorption from radiation, mechanical generation in viscous or plastic
718
Back Analysis Monitoring
flow, and chemical reaction, including in this such diverse matters as the hydration of cement. In this last case for example, the rate of heat production is roughly of the form k exp ( —αί), the heat generated in three days being of the order of 50-100 cal per gram of cement; this is sufficient to have important technical consequences, particularly in the design of large dams. Infrared thermography has been successfully used as an experimental method for the detection of plastic deformation during crack propagation of a steel plate under monotonie loading [9] or as a laboratory technique for investigating damage, fatigue and creep mechanisms occurring in engineering materials [10-12]. This experimental tool is used to detect the onset of unstable crack propagation and/or flaw coalescence due to the thermomechanical coupling, when increasing irreversible microcracking is induced by vibratory loading.
26.3.1 Coupled Thermo-Visco-Elastic-Plastic Analysis Traditionally, thermomechanical coupling effects have been neglected in thermal analyses. It is generally assumed that the inelastic deformation is rate(time)-independent at low homologous temperatures. The theory of plasticity is consequently formulated in a rate(time)-independent fashion and phenomena such as loading rate sensitivity, creep and relaxation are excluded. The temperature field and the deformation induced by thermal dilation and mechanical loads were solved separately. However, this effect could become noticeable if the material is significantly loaded beyond its reversible threshold. The development of the thermo-visco-elastic-plasticity equations requires three types of basic assumptions [13-17]. (i) The basic thermomechanical quantities describing thermodynamic processes: the motion x, the second Piola-Kirchhoff stress tensor S, the body force per unit massft,the Helmholtz free energy φ9 the specific entropy s, the heat supply r 0 , the absolute temperature Γ, the heat flux vector per unit area q, the inelastic strain tensor El and a set of internal state variables a(I) characterizing the material. (ii) The fundamental equations of mechanics postulating for the balance laws of linear momentum, angular momentum and energy, as well as the second law of thermodynamics expressed in the variables given in (i): (a) balance of linear momentum (1)
where F denotes the transformation gradient, (b) conservation of energy (2)
(c) second law of thermodynamics (3) 3
where p (kg m" ) is the mass density in the reference configuration, e the specific internal energy and E the Green-Lagrange strain tensor: E = (VxT.Vx — l)/2. The superposed dot represents the material time derivative. The continuity equation and the balance of angular momentum are implicitly satisfied in the fundamental equations. (iii) The constitutive assumptions describing the material response and abiding the compatibility of the constitutive equations with the fundamental equations of mechanics. When adopting the separability of the strain tensor (4)
where β is the coefficient of the thermal expansion matrix and r R the reference temperature, the requirement of inequality (4) yields the following. (a) The response functions S, φ and s are independent of the temperature gradient VT. (b) φ determines both the stress tensor and the specific entropy through
Infrared Thermographie Observations of Rock Failure
719
(c) ψ9 Ê1 and q obey the general inequality (S - pdij//dEl):El - q.VT/T>0
(5)
The above thermodynamic restrictions may now be applied to equation (2) yielding ρφψ/dE1 - Td2ij//dTdEl):Èl - T(dS/dT):È* - ρΤ@2ψ/άΤ2)Τ + S:ßT = S:Èl - di\q + r0
(6)
Assuming φ = ψο + (£ e :D:£ e )/2 - CvT\n(T/TR - 1)
(7)
and the Fourier heat conduction law q = -KgradT 4
(8)
4
where φ0, D, Cv and K are material constants. D stands for the fourth-order elasticity tensor. Cv (J kg" x K~ *: Joule per kilogram per degree Kelvin) is the specific heat at constant deformation and K (W m " 1 K " 1 : Watt per meter per degree Kelvin) is the thermal conductivity. Finally we get the coupled thermomechanical equation pCyt
= KV2T - {ß:D:Ee)T + S:Èl + r0
(9)
which shows the varied potential applications and uses of the infrared scanning technique in rock engineering. The volumetric heat capacity C = pC v of rock material is the energy required to raise the temperature of a unit volume by 1 °C (or degree Kelvin). 26.3.2
Thermal Conduction
The first term on the right hand side of the thermomechanical equation governs the transference of heat by thermal conduction in which the heat passes through the material to make the temperature uniform in the specimen. The second-order tensorial nature of the thermal conductivity K may sometimes be used for the detection of anisotropy of heavily loaded rock-like materials. Where an unsteady state exists, the thermal behavior of a rock is governed not only by its thermal conductivity but also by its heat capacity. The ratio of these two properties is termed the thermal diffusivity a = K/C (m2 s~ *), which becomes the governing parameter in such a state. A high value of the thermal diffusivity implies a capability for rapid and considerable changes in temperature. It is important to bear in mind that two materials may have very dissimilar thermal conductivities (e.g. rock and insulator) but, at the same time, they may have very similar diffusivities.
26.3.3
Thermoelasticity
The second term illustrates the thermoelastic effect, investigated by Lord Kelvin in 1853. Within the elastic range, a material, subjected to tensile or compressive stresses, experiences a reversible conversion between mechanical and thermal energy causing it to change temperature. Provided adiabatic conditions are maintained, the relationship between the change in the sum of the principal stresses and the corresponding change in temperature is linear and independent of loading frequency. It is the reversible portion of the mechanical energy generated; this thermoelastic coupling term may be significant in cases of isentropic loading.
26.3.4
Intrinsic Dissipation
The third term is the energy dissipation generated by viscosity and/or plasticity. The work done in plastic deformation per unit volume can be evaluated by integrating the material stress-strain curve. This internal dissipation term constitutes an important part of the nonlinear coupled thermomechanical effect. The quantification of this intrinsic dissipation for rock-like materials is an extremely difficult task if infrared thermography is not used. This chapter emphasizes the advantages of the infrared thermographie technique for the detection of this effect.
720
Back Analysis Monitoring
26.3.5 Heat Sources The last term shows the existence of sources or sinks of heat in the specimen. The surface heat patterns displayed on the scanned specimen may be established either by external heating, referred to in literature as 'passive heating', where local differences in thermal conductivity cause variations on isothermal patterns, or by internally generated heat, referred to as 'active heating', where isothermal patterns are established by the transformation of internal energy into heat. In the case of forced convection, if fluid at the temperature of the medium is forced rapidly past the surface of the solid, it is found experimentally that the rate of loss of heat from the surface is proportional to the surface conductance or coefficient of surface heat transfer. This fact has been used for the detection and location of heat, gas or fluid leakage through rock-like materials. 26.4 INFRARED THERMOGRAPHY TECHNOLOGY The infrared portion of the electromagnetic spectrum was discovered accidentally by Sir William Herschel, Royal Astronomer of England, in 1800 during the search for a new optical material. While testing different samples of colored glass, which gave similar reductions in brightness, he was intrigued to find that some of the samples passed very little of the sun's heat, while others passed so much heat that he risked eye damage after only a few seconds of observation. Looking for the heating effect, he moved the thermometer into the dark region beyond the red end of the electromagnetic spectrum, referred to by him as the 'thermometrical spectrum', and he confirmed that the heating effect reaches a maximum in the infrared spectrum. His son, Sir John Herschel, managed to obtain a primitive record of the thermal image on paper, which he called a 'thermograph'. 26.4.1 Infrared Radiation Knowledge of the basic physics of infrared radiation and related heat transfer [18] is necessary for effective thermography evaluations. Electromagnetic radiation is a form of energy characterized as waves or as particles called photons. The electromagnetic spectrum is a categorization by wavelength of electromagnetic energy (Figure 1). Visible light is the most familiar form of electromagnetic energy. The visible portion runs from 0.4 μηι to 0.75 μιη where infrared or thermal radiation begins. The range from 2 μπι to about 100 μιη is called 'thermal infrared'. All bodies at a temperature above absolute zero spontaneously radiate electromagnetic energy because they contain charged particles being accelerated (changing speeds or directions), the higher the temperature the greater the acceleration. The amount of energy radiated depends on the object's temperature and its ability to radiate. Since visible radiation is limited to such a small portion of the electromagnetic spectrum, we must rely on special instruments to detect and measure thermal infrared radiation. In general, the successful application of these instruments is dependent upon the radiation being emitted by the object of interest. An ideal subject for an infrared instrument would therefore emit the maximum amount possible at any given temperature. Such a subject would be considered to be an ideal radiator or blackbody and would also absorb all radiation that impinges on it. The energy emitted The electromagnetic spectrum Visible γ ■
\ i IÂ
X '
i UV \ ; Infrared i ' j ii ' ' i ' \00 & //j\ μ\ \ v Imm
Radio ' GHz
' Im
I km 3 MHz
λ 3 kHz
Figure 1 Region of the electromagnetic spectrum covered by infrared thermography
721
Infrared Thermographie Observations of Rock Failure Table 1 Emissivity Coefficients at 20 °C for some Common Materials Brick: common red Concrete Distilled water Glass: polished plate Human skin Iron sheet, heavily rusted Paper: white bond
0.93 0.92 0.97 0.94 0.98 0.69 0.93
Plaster: rough coat Sand Soil: dry Soil: saturated with water Stainless steel: buffed Water: distilled Wood: planed oak
0.91 0.90 0.92 0.95 0.16 0.96 0.90
or absorbed per unit surface area per unit increment of wavelength for a blackbody is given by Planck's law of spectral distribution. Its radiating and absorbing efficiency, called its emissivity ε, is said to be unity. A body at absolute temperature T surrounded by a blackbody at the temperature T0 will lose heat at the rate ce(T4 - TQ), where c is the Stefan-Boltzmann constant. Unfortunately, real objects almost never fulfill these conditions, although they may approach blackbody behavior in certain wavelength regions. Special blackbody radiators can be constructed and are commonly used as instrument calibration sources. When determining the temperature of a real object through the use of an infrared instrument, its 'apparent temperature' is observed. The real temperature of an object is dependent only upon its emitted radiation. Corrections can be made to the apparent temperature if the emissivity of the object is known. Real objects, influenced by absorbed, reflected and transmitted infrared, have emissivity values less than unity. It must be determined experimentally for each object. The emissivity ranges from zero for 'mirror-like' surfaces to nearly one for lamp black, zapon black, and such surfaces. Since the emissivity of real objects will vary with temperature and wavelength, average values are often given along with specific temperatures. Table 1 provides a list of emissivity values for a few common materials at 20 °C. 26.4.2 Infrared Thermography Infrared thermography is a technique for producing heat pictures from the invisible radiant energy emitted from stationary or moving objects at any distance and without surface contact, or in any way influencing the actual surface temperature of the objects viewed. A scanning camera is used which is analogous to a television camera. It utilizes an infrared detector in a sophisticated electronics system in order to detect radiated energy and to convert it into a detailed real-time thermal picture in a video system both color and monochromatic. Response times are shorter than a microsecond. Temperature differences in heat patterns as fine as 0.1 °C are discernible instantly and represented by several distinct hues. This technique is sensitive, nondestructive and noncontact, thus suited for records and observations in real time of heat patterns appearing on the surface of an object being scanned. No interaction at all with the specimen is required to monitor the thermal gradient. The quantity of energy W (W m~2 μιη"1) emitted as infrared radiation is a function of the temperature and emissivity of the specimen. The higher the temperature, the more important is the emitted energy. Differences of radiated energy correspond to differences of temperature (Figure 2). 26.4.3 Infrared Scanner For an infrared system viewing any source, the received power at the system aperture is given by H = Wœ/Q, where ω is the angular field of view of the viewing system defined by the optical system and the detector, and Ω is the total solid angle about the source. According to Lambert's cosine law, the radiant energy emanating in a given direction from any point on a surface is a function of the cosine of the angle between the normal to the surface at that point and the given direction. If the source is small compared with the field of view of the detector, the received radiation will vary with the distance between the source and detector but not with the angle about the source. Interestingly, if the source is large compared with thefieldof view, the received radiation varies with neither distance to the source nor the angle about it. As the increase in viewed area exactly matches the reduction in radiation energy, a detector looking at an emitting surface will always receive the same amount of energy no matter what the angle between the detector's line of sight and the radiating surface. As an example (Figure 3), the AGA 782 SW infrared scanner unit [19] comprises: (i) a set of infrared lenses which focuses the electromagnetic energy radiating from the object being scanned
722
Back Analysis Monitoring Wien's law 10*
Λ?
' ^ ^
/Planck's law
E W. Herschel
=*CM
E
2 ICT
I I Heat patterns
Figure 2 Radiation and temperature variations according to Planck's, Wien's and Stefan's laws
IR detector
Scanned field Figure 3 Arrangement for an infrared thermographie scanner
into the vertical prism; (ii) an electro-optical scanning mechanism which discriminates the field of view in 10 000 pixels by means of two rotating vertical (180 rpm) and horizontal (18 000 rpm) prisms with a scanning rate of 25fieldsper second; (iii) a set of relay optics containing a selectable aperture unit and a filter cassette unit which focuses the output from the horizontal prism onto a single element point detector, located in the wall of a Dewar chamber; (iv) a photovoltaic SW (short wave) infrared detector composed of indium antimonide, InSb, which produces an electronic signal output varying in proportion to the radiation from the object within the spectral response 3.5 μπι to 5.6 μπι; (v) a liquid nitrogen Dewar which maintains the InSb detector at a temperature of —196 °C allowing a very short response time of about one microsecond; and (vi) a control electronics with preamplifier which produces a video signal on the display screen. Since the received radiation has a nonlinear relationship to the object temperature, can be affected by atmospheric damping and includes reflected radiation from the object surroundings, calibration and correction procedures have to be applied. Knowing the temperature of the reference, the object temperature can then be calculated with a sensitivity of 0.1 °C at 30 °C. 26.5 INFRARED SCANNING AND ROCK ENGINEERING 26.5.1 Infrared Thermographie Observations of Rock Failure Rock-like materials present a low thermomechanical conversion under monotonie loading. Plastic deformation, whereby microcracking and slips occur creating permanent changes globally or
Infrared Thermographie Observations of Rock Failure
Figure 4
723
Different stages of heat dissipation describing the process of unstable crack propagation occurring in a natural rock salt specimen (0.2 °C for each color hue)
locally is, however, one of the most efficient heat production mechanisms. Most of the energy which is required to cause such plastic deformations is dissipated as heat. Such heat development is more easily observed when it is produced in a fixed location by reversed applied loads. These considerations define the use of vibrothermography as a nondestructive and noncontact method for observing the damage processes of rock-like materials [20]. In the laboratory, studies on geomaterial mechanical behavior use specific equipment to set up experiments. The high-frequency servohydraulic test machine provides a means of vibration and dynamic testing of engineering materials. Control of the machine is provided by a sophisticated closed-loop electronic control system. This utilizes feedback signals from the force and displacement transducers. The programming section comprises a digital function generator and a frequency sweep controller which enables resonant phenomena testing. The sample is observed in a nondestructive, noncontact manner by means of an infrared thermographie system. The thermal image is shown on the monitor screen. The parameter investigated in this test is heat generation due to the energy dissipated by the material which has been excited beyond the stable reversible domain. A vibratory loading at 100 Hz on the specimen subjected to a given static compression exhibits in a nondestructive manner the irreversible plastic strain concentrations around gaps or cracks generated by stresses exceeding locally the stability limit of the material. The contribution of the plasticity term is revealed by the rapid evolution of heat dissipation once the stable reversible domain has been exceeded, demonstrating the occurrence of an unstable crack propagation (Figure 4) or the coalescence of flaws existing in the natural rock salt specimen (Figure 5). It can be seen that: (i) with a vibratory excitation between 30 and 50% of the nominal uniaxial compression σΝ = F/S0, the heat dissipation detected for 2000 load cycles is small, even at the hottest location (Figures 4a and 4b); (ii) when 0.50 < σ/σΝ < 0.70, stress concentrations around cracks or defects are readily detected at the 1000th load cycle (Figures 5a and 5b); and (iii) for 0.70 < σ/σΝ < 0.90, cracking occurs increasingly in the reduced section part of the specimen. The different phases of heat dissipation, operating during an unstable failure are described in Figures 4(c) and 4(d). When defects or weakness zones are present on the specimen, Figures 5(c) and 5(d) evidence the progressive mechanism of defect coalescence.
Back Analysis Monitoring
724
(C)
(b)
(α)
ΔΓ(°0
0.2
(d)
0.6
1.0
1.2
(e)
1.8
Figure 5 Records of flaw coalescence mechanism in a natural rock salt specimen: the growth rate of heat at the warmest point enables the detection and the location of the onset of specimen failure (0.2 °C for each color hue) Rock salt
Vibratory loading
0.75
0 25
Number of load cycles
Figure 6 Stress-strain curves of a natural rock salt specimen under normalized unconfined compression (a/apeak) show three conventional different thresholds 7Ί, T2 and Γ3 respectively related to axial (ea), radial (εΓ) and volume change (εν) curves. On the right, the heat generation rate Tat the warmest location is shown as a function of the number of load cycles. The change of heat generation rate may be considered as a damage threshold TA of the tested material
The rate of heat generation at the hottest location may be used to detect the threshold of the failure process if compared with the traditional stress-strain curve (Figure 6). Thus the thermomechanical coupling offers a quantitative evaluation for the growth rate of thermal dissipation monitoring the damage evolution of the material. The damaged areas are located and highlighted by heat patterns. These results support and validate the assumptions to be taken into consideration in numerical procedures for stability assessment of rock structures. The phenomenological behavior of the rock materials in consideration is therefore the standard of reference, allowing the use of the methods and results of continuum mechanics for analyzing and modeling their engineering performance. 26.5.2
Infrared Scanning of Rock Mass Thermal Properties
The application of infrared thermographie technology to aerial thermal land mapping, geological survey, geotechnical instrumentation, leakage insulation and construction practices for rock-like
Infrared Thermographie Observations of Rock Failure
725
Horizon Ground
Scan line
Field of view
Figure 7 Airborne multispectral scanner system
and rock engineering [21, 22] has been extensive in recent years. Two types of infrared imaging systems can be distinguished. 26.5.2.1 Aerial imagery Aerial thermal imagery can be collected using two thermal bands simultaneously recorded in the 8.2-9.3 μηι and the 10.4-12.5 μιη wavelength ranges. The simplified diagram in Figure 7 illustrates the operation of a line scanner. The scanner is basically an optical telescope with its narrowfieldof view continuously redirected by a spinning flat mirror. The mirror causes the system to scan in a plane perpendicular to the direction of flight of the aircraft. Two cryogenically cooled radiation detectors in the focal plane of the telescope convert the focused beam of filtered radiation to electrical signals. The signal from each detector is digitized and recorded on magnetic tape for later playback. In operation the scanner's instantaneous field of view (ground resolution element) scans laterally across the aircraft ground tract through an opening in the bottom of the aircraft. Before making the next complete scan, the optics record two temperature reference plates internal to the scanner. These reference plates are maintained at carefully controlled temperatures for later temperature calibration of the thermal imagery. By the time the next scan begins, the aircraft has moved forward such that a continuous strip image of the terrain beneath the aircraft is recorded on tape. To produce the calibrated slices, the aerial data are quantized into discrete signal levels representing apparent temperature ranges. These are individually imaged and photographed such that all portions of the scene having the same temperature range are highlighted, while all surface temperatures outside that range appear black. By producing a series of these slices, each with a different increment, the full temperature range of the scene is represented. Surface water and vegetated areas generally are warmer than the rock mass surfaces. Rock mass temperatures are generally colder than air temperatures on clear nights due to the net transfer of heat by radiation to the cold sky. It must be noted that these are apparent temperatures. In other words, in producing these images it is assumed that the emissivities of all the different surfaces are constant and nearly unity. In nature, this is a fair assumption; most natural surfaces have emissivities of 0.8 or greater. But some surfaces, specifically metallic surfaces, have emissivities much less than unity, and temperatures of these
726
Back Analysis Monitoring
surfaces must be individually corrected. An analytical model must be developed to interpret the observed temperatures from the flight data. All calculations are made with environmental parameters determined by conditions as they occurred during the time of the flight. Apparent temperatures are obtained from the aerial thermal data as level slices. Each slice portrays all surfaces with temperatures which correspond to a given known radiance level. However, part of the observed radiance is a reflection of the cold sky. The radiance temperature of the cold sky was assumed to be — 30 °C for the clear sky conditions that prevailed at the time of the flight.
26.5.2.2
Ground-based imagery
Heat transfer in rock mass plays an important part in many types of thermal problems [23, 24]. An understanding of the thermal behavior of rock clarifies these problems: values of the rock's thermal properties are required for quantitative analysis. (i) Detecting heat transfer A laboratory test has been carried out on a rock salt cube of 0.50 m in dimension for analyzing the thermal diffusion through natural rock salt (Figure 8). A spherical heat source is installed off-centre so that the boundary influence can be estimated by thermistors and by heat patterns displayed on the different faces of the rock salt cube. The isotropic thermal diffusion assumption has been confirmed in spite of the layered structure of rock salt and the axisymmetrical geometry of the experimental layout. It has been found that infrared thermography is particularly useful in detecting infiltration problems, missing insulation and construction defects. It permits the survey of heating systems in industrial plants [25,26]. This qualitative survey has been found to be an excellent method to detect environmental heat losses and water or gas leakages in process equipment and auxiliary systems. In addition to the qualitative survey, quantitative data may be gathered by calibrating the temperatures of the 'hot spots' uncovered in the survey. These 'hot spots' included problems in the basic insulation design, flaws in the insulation and maintenance errors. However, many 'hot spots' indicated malfunctions in the equipment such as leaking valves, leaking electrical components, failures in materials and components, failures in materials and components in invisible areas of the equipment, overheated bearings, etc. This information was very useful in assigning priorities and estimating the magnitude of heat loss due to a given defect. In the construction of any complex facility or structure for radioactive waste disposal, construction defects can arise through human error. Detection is a problem for visual inspection, but not severe, if the defect is located on the surface. The detection of such defects poses greater problems after construction has covered up the work area as in, for example, defects in wall construction. Hand-held probes for examining the surface have been used with limited success. The size of the involved structure related to the size of the defect makes the job nearly impossible unless the area of defect location is reduced to a small region by a separate method. Ultrasonic techniques using computer enhancement were proposed to detect remotely not only the defect, but also the location by triangulation; they were found unworkable in this application because normal random noise would null or obscure the noise caused by a slight diffusion leak. In many cases, once a defect has been located with thermography, a retrofit measure can be implemented to correct it. But it cannot be assumed that because a retrofit has been performed, the defect is necessarily corrected. Infrared thermography is also useful in showing if a retrofit measure has corrected the defective area. (ii) Detecting leakage The infrared radiometric measurement is sometimes proposed as a means of not only locating the presence of the leak, but also the location. Figure 9 demonstrates the feasibility of detecting, in a laboratory, the location of gas leakage through a very small crack when a pressure difference of several kPa is applied. An infrared scanning system can provide a wide field of view and a full field of data in a short period of time with a good spot resolution. If the leak causes a change in thermal radiation, the defect will show up on the scanner display scene as a cooler or hotter area depending upon the technique of leak stimulation. The extent of the defect will be more difficult to pin down with infrared alone, but the speed at which the leak becomes visible at the surface is a function of flow and defect size. Subsequent measurements using probe techniques which are more directly related to flow can indicate the extent.
Infrared Thermographie Observations of Rock Failure
727
Figure 8 Thermal diffusion through a rock salt cube
When an abnormal thermal pattern appears, it is most often a hot or cold spot on the wall surface. Its origin is usually an area of accidental damage which can occur due to a fault, etc. The resultant thermal pattern is sharp, well defined and easily marked for repair. A less defined indication can be warmer areas which appear as bands or zones - these involve cracking or weakness zones and are due to diffuse leakage. If any abnormal areas are found they are marked for repair and occasionally further instrumented to allow continuous monitoring of the problem area while the test and scanning continue. After the problem area has been repaired the test is repeated to verify the adequacy of the repair. (Hi) Detecting wet insulation Thermal signatures of wet and dry insulation have been systematically documented with an infrared camera. Recorded temperature differences were sufficient to allow the wet area to be detected by the infrared camera.
Back Analysis Monitoring
728
Figure 9 Detection of gas leakage in a laboratory test
Temperature differences must exist on a rock wall surface above wet and dry insulation in order for the infrared camera to locate areas of wet wall insulation. It is easy to imagine how a thermal anomaly might be created by wet insulation during the winter as building heat is conducted more rapidly through wet insulation. This creates a warmer surface over the wet insulation. Fortunately, aeration by forced convection with dry and heated air driven by a large fan in the tunnel heats the wall surface. This heat is conducted away from the wall surface by any wet insulation and stored there. Dry insulation resists penetration of hot air in circulation and has little energy storage capacity. Because of these conduction and storage differences, the wall surface is hotter over dry insulation than over wet insulation when hot air is circulated. (iv) Detecting delaminations behind rock wall In the field, the thermal diffusion through rock mass may be analyzed for use in the detection of subsurface fracture planes, or delaminations, in underground rock works. This is necessary when determining the priorities of structures for repair and for the identification of areas of unsound rock prior to a repair contract. There is a need to identify subsurface fracture planes when assessing the condition of structures to determine priorities for repair. At present, delaminations are detected by manual methods which are tedious and require expensive underground work closures. Manual methods, such as a sounding rod, are commonly used for delamination detection, by virtue of the fact that a delaminated area emits a hollow sound when struck. These methods, however, are tedious, difficult to apply in a noisy location and dependent upon the skill of the operator. A portable instrument which electronically interprets acoustic signals generated by the instrument and reflected through the rock may also be used. This instrument, in common with the manual methods, has the disadvantage that parts of the structure must be closed while observations are made. A need for a nondestructive test procedure which will locate delaminated areas of rock wall quickly and inexpensively, and preferably without closing the structure, has been readily identified. An investigation may be initiated to determine if different surface temperatures are associated with areas of sound and unsound rock and whether these differences can be detected and recorded using infrared thermography. (v) Interpreting thermographie data The speed of thermal interpretation presents an opportunity for quicker evaluation of performance. Consequently, thermography can speed up recommendations for future design guidelines. Although quantitative temperature measurement is possible using thermography, its use has been
Infrared Thermographie Observations of Rock Failure
729
little developed due to inadequate knowledge of the various factors which can affect readings of thermal evaluation data. These factors are: condensation, material roughness, air pollutants, reflected radiation and angle of sight. All these five factors influence the actual value of a material's emittance and must be taken into account to obtain accurate readings of temperature variation. While there are other techniques for measuring temperature variation on a material's surface, thermography has been proven to have certain advantages. It allows large surfaces to be examined easily and quickly. Measurements can be taken at considerable distances from only one vantage point. In addition to qualitative temperature assessment, quantifiable measurements are achievable if certain controls are established. Because thermographie data can be interpreted to assess performance, the technique can be used to interpret the energy implications of different types of wall construction. Although the common types of construction defects (omitting insulation and air leakage) are not easily detected by visual inspection, they can be detected readily using infrared thermography. 26.6 CONCLUSIONS Owing to the thermomechanical coupling, infrared vibrothermography offers the possibility of a nondestructive, noncontact test of mechanical degradation on concrete, rock salt or any other rock-like materials. It allows a measure of the material damage and permits the detection of the limit of a progressive damaging process under load beyond which the material is destroyed. It is of particular interest that the method allows not only qualitative work such asflawdetection, but also quantitative analysis of the effects of flaws on strength and durability of structural components. This useful and promising technique allows accurate illustration of the onset of unstable crack propagation and/or flaw coalescence when increasing irreversible microcracking is generated by vibratory loading. The laboratory testing facilitated the development of procedures for accurate and consistent interpretation of equipment readings. The effects of reflected radiation, emissivity variations of scanned materials and their relations to each other, as well as the influence of water or interstitial condensation on the material, demonstrated that readings must be interpreted using standards of measurement. Field and aerial applications verified laboratory conclusions that thermographie analysis is capable of identifying thermal problems. Aerial tests showed that structures with thermal problems could be identified easily and put into a priority listing for subsequent ground evaluations. This approach is quicker and more efficient than any other approach now being used for assessment of structure performance. Infrared thermography is a useful tool for scanning industrial equipment and processes to detect excess heat losses and malfunctions which result in temperature changes. A quality improvement may be gained in the practices of builders and contractors through improved construction standards and better building codes.
26.7 REFERENCES 1. Bui H. D. and Stolz C. Damage theories for brittle and ductile materials. In Fracture of Non-Metallic Materials (Edited by K. P. Herrmann and L. H. Larsson), pp. 33-46. Brussels (1987). 2. Dougill J. W. Constitutive relations for concrete and rock: applications and extensions of elasticity and plasticity theory. In Preprints William Prager Symp. Mechanics of Geomaterials: Rocks, Concretes, Soils, Northwestern University, Evanston, IL (Edited by Z. P. Bazant), pp. 17-54 (1983). 3. Zaitsev Y. V. Inelastic properties of solids with random cracks. In Preprints William Prager Symp. Mechanics of Geomaterials: Rocks, Concretes, Soils, Northwestern University, Evanston, IL (Edited by Z. P. Bazant), pp. 76-148 (1983). 4. Tepfers R., Friden C. and Georgsson L. A study of the applicability to the fatigue of concrete of the Palmgren-Miner partial damage hypothesis. Magazine of Concrete Research, 29, 100, 23-130 (1977). 5. Tepfers R., Hedberg B. and Szczekocki G. Absorption of energy in fatigue loading of plain concrete. Matériaux & Construction, 17, 97, 59-64 (1984). 6. Hardy R. H. Jr. and Langer M. The mechanics of salt. Proc. 1st Conf. Pennsylvania, Trans Tech, Rockport, MA (1981). 7. Stokes R. J. Fracture of ceramics. In Proc. 4th Symp. Fundamental Phenomena in the Materials Sciences, Boston pp. 151-175 (1966). 8. Kotsovos M. D. and Newman J. B. Generalized stress-strain relationships for concrete. J. Eng. Mech. Div., Am. Soc. Civ. Eng., 194, EM4, 845-856 (1978). 9. Bui H. D., Ehrlacher A. and Nguyen Q. S. Etude expérimentale de la dissipation dans la propagation de fissure par thermorgraphie infrarouge. C. R. Acad. Sei., 293, II, 1015-1017 (1981).
730
Back Analysis Monitoring
10. Reifsnider K. L., Henneke E. G. and Stinchcomb W. W. The mechanics of vibrothermography. In Mechanics of Nondestructive Testing (Edited by W. W. Stinchcomb), pp. 249-276 (1980). 11. Luong M. P. Characteristic threshold and infrared thermography of sand. Geotech. Test. J., GTJODJ, 9, 80-86 (1986). 12. Luong M. P. Infrared observations of damage and fracture on concrete. In Thermomechanical Couplings in Solids, Paris (Edited by H. D. Bui and Q. S. Nguyen), pp. 85-94. Elsevier, Amsterdam (1987). 13. Taylor G. I. and Farren W. S. The heat developed during plastic extension of metals. Proc. R. Soc. (London), 107, 422 (1925). 14. Dillon O. W. Jr. Coupled thermoplasticity. J. Mech. Phys. Solids, 11, 21-23 (1963). 15. Kratochvil J. and Dillon O. W. Thermodynamics of elastic-plastic materials as a theory with internal state variables. J. Appl. Phys., 40, 3207-3218 (1969). 16. Kratochvil J. and Dillon O. W. Thermodynamics of crystalline elasticviscoplastic materials. J. Appl. Phys., 41, 1470-1479 (1970). 17. Allen D. H. A prediction of heat generation in a thermoviscoplastic uniaxial bar. Int. J. Solids Struct., 21, 4, 325-342 (1985). 18. Gaussorgues G. La Thermographie Infrarouge. Technique et Documentation Lavoisier, Paris (1984). 19. AGA Infrared Systems AB Thermovision 782. Operating manual (1984). 20. Luong M. P. Infrared Vibrothermography of Plain Concrete. Magnetic sound 16 mm film, Videotape UMATIC VHS PAL SECAM Systems. Edited by IMAGICIEL, Ecole Polytechnique, Palaiseau, France (1984). 21. Charrier J. and Marucic J. A. Possibilités d'utilisation des méthodes thermiques à des fins d'essais non destructifs en génie civil. Synthèse bibliographique, Rap. de rech. LPC-MULT-LCPC, 11, Paris (1982). 22. Clement B., Leveque P. Ch., Majourau S. and Sirieix C. Utilisation des données aérospatiales dans les études de sites de grands ouvrages. Mémoires de la Société Géologique de France, Nouvelle Série, 157, 263-277 (1990). 23. Carslaw H. S. and Jaeger J. C. Conduction of Heat in Solids, 2nd edn. Clarendon, Oxford (1986). 24. Bonnet G. and Jouanna P. Vers une approche énergétique des mécanismes de ruine des massifs rocheux fissurés. Revue de rindustrie Minérale, 7, 1-7 (1975). 25. Bérest P. Phénomènes thermiques en géotechnique. In La Thermomécanique des Roches, Manuels & Méthodes, Aies (Edited by P. Bérèst and Ph. Weber), 16, pp. 13-67. BRGM Editions (1988). 26. Sirieix C , Monnet R. and Castanier G. Mise en oeuvre de diverses méthodes de reconnaissance d'un environnement géologique original. Application à la géométrie et à la conception du voile d'injection du barrage de Vieux-Pré. Mémoires de la Société Géologique de France, Nouvelle Série, 157, 355-363 (1990).
27 In Situ Testing and Monitoring of a Test Drive in an Underground Coal Mine MICHAEL J. PENDER University of Auckland, New Zealand and KENNETH W. MILLS Strata Control Technology Pty. Ltd, Wollongong, Australia; formerly of University of Auckland, New Zealand
27.1 INTRODUCTION 27.1.1 General 27.1.2 Mine Location 27.1.3 Sequence of Investigation
732 732 732 732
27.2
733
COAL PROPERTIES FROM LABORATORY TESTS ON NX CORE
27.2.1 27.2.2 27.3
733 734
Strength Properties of NX Coal Core Stiffness Properties of NX Coal Core
TEST PANEL
734 734 735 736 Ί3Ί 738
27.3.1 Test Panel Layout 27.3.2 Test Panel Instrumentation 27.3.2.1 Extensometer system 27.3.2.2 Manometer system 27.3.2.3 Stress change system 27.4
MONITORING DURING AND AFTER THE TEST DRIVE EXCAVATION
27.4.1 27.4.2 27.4.3
739 739 740 742
Deformations Stress Changes General Observations
27.5 IN SITU PRESSUREMETER TESTING OF THE COAL AND FIRECLAY STIFFNESS 27.5.1 Pressuremeter 27.5.2 Results
743 743 743
27.6 MEASUREMENT OF IN SITU STRESSES 27.6.1 Technique 27.6.2 Results
744 744 744
27.7 FINITE ELEMENT MODELING 27.7.1 Input Data 27.7.2 Results 27.7.2.1 Displacement distribution 27.7.2.2 Stress distribution 27.7.2.3 Failure zones 27.7.2.4 Other factors
745 745 746 746 Ί4Ί 749 749
27.8
SUMMARY AND CONCLUSIONS
750
27.9
REFERENCES
750
731
732 27.1 27.1.1
Back Analysis Monitoring INTRODUCTION General
In the design of an underground mine assumptions have to be made about rock mass properties and in situ conditions. When an underground excavation is made, monitoring behavior of the surrounding rock mass provides a means of checking the validity of assumptions necessary at the design stage. In this chapter a case study is described involving a comprehensive program of measurement of rock mass properties and in situ conditions with observations of the response, during and after excavation, of material surrounding an underground coal mine. The objective of the work was to perform a detailed investigation at a specific location in a coal seam to provide information to assist in the selection of mining methods for future development of the mine. A greatly increased coal recovery rate was the requirement for these future plans. An underground mine is a difficult environment for measurement of rock properties and for monitoring of deformations and stress changes in a rock mass. The in situ tests and monitoring techniques described in this chapter are effective but simple. The most sophisticated instrumentation used was a portable strain indicator, for reading strain gauges, and a battery operated portable digital micrometer to read extensometers. As will be seen the system worked well for the application described. A possible disadvantage of the techniques employed is the labor intensive aspect of the data gathering. The work described involved measurement of the properties of coal core, in situ measurement of coal stiffness, measurement of in situ stresses in the coal, monitoring the behavior of coal during and subsequent to the excavation of a test drive. The final step involved finite element analysis to check if the displacements observed during the excavation of the test drive were consistent with measured in situ stresses and stiffness of the coal mass. The project is perhaps unusual not so much in the details of the various measurement techniques, although a new technique was developed to determine the in situ stresses in the coal seam, but in the comprehensive manner in which all the steps contributed to the final result. More commonly underground monitoring is carried out with limited objectives in mind. The outcome of the project justified the time and effort required to follow through the full program because the major conclusions were then cross-checked and confirmed. These were that (i) the in situ horizontal stress in the coal seam was considerably less than the vertical stress; (ii) apart from small localized regions, the coal responds elastically to the changes caused by excavation; and (iii) any anisotropy inherent in the coal, because of the cleat, has a much less significant effect on the behavior of the drive than the in situ stresses.
27.1.2
Mine Location
The mine is located at Huntly in the Waikato Basin about 80 km south of Auckland in the North Island of New Zealand. There are two coal mines at Huntly; the Huntly West mine is discussed in this chapter. The ground surface above the mine is approximately horizontal and used mainly for agricultural purposes. The coal seam in which the mine is located is about 10 m in thickness and is about 250 m beneath the ground surface.
27.1.3 Sequence Of Investigation Ideally the various phases of the investigation would have been in the following sequence: (i) test panel established; (ii) coal cores recovered from the test panel coal; (iii) laboratory tests on coal core; (iv) in situ tests on the test panel coal; (v) measurement of the in situ stress in the coal; (vi) excavation of the test drive and monitoring of the behavior of the adjacent coal; (vii) monitoring of the response of the coal adjacent to the test drive after the excavation; and (viii) finite element analysis to check if the results of the various components of the investigation are consistent. Because of other priorities in the operation of the Huntly West mine it was not possible to perform all the steps in the order set out above. Items (iv) and (v) did not in fact occur until after the test drive was excavated. The sequence in this chapter follows the actual order of operations rather than the ideal. Complete details of the project are given by Mills [1] and a summary by Mills et al [2].
In Situ Testing and Monitoring of a Test Drive in an underground Coal Mine
733
27.2 COAL PROPERTIES FROM LABORATORY TESTS ON NX CORE 27.2.1 Strength Properties of NX Coal Core Samples recovered from NX boreholes at the test location in the mine were tested for strength in a Hoek-Franklin triaxial cell. It was difficult to recover core in directions parallel to the cleat. Core recovery perpendicular to the cleat was easier. Core that could be tested was coated with a thin layer of epoxy cement and stored in a moist environment to prevent drying and the formation of shrinkage cracks. Specimens for triaxial testing were cut to length with a diamond saw and the ends ground smooth. Strain gauges were used to measure strains during the testing. Figure 1 shows the effect of confining pressure on the strength and stiffness of the core. In Figure 2 the failure envelope for all the triaxial specimens is plotted. Dots in the diagram represent the results of the separate tests. Some of the scatter evident in Figure 2 is due to variability of the individual specimens and some is due to the loading direction relative to the cleat direction in the core.
50
s. Έ
40
-
//
8
\
/ /
//
-
3a =
IC
y//4
in
jjj 30 to Έ "x < 20 Perpendicular to cleat Deformation rate 5 0 mm h~ 10
L
i
I
1
1 | 2 3 Axial deformation (mm)
Figure 1 Effect of confining pressure on triaxial stress-strain curves for NX specimens of the coal
Note : axes are drawn to different scales
φ = 38° (k =4.2)
^Failure envelope determined from 2-IOMPa confining pressures (considered to be representative of intact coal ) Bilinear failure envelope Failure envelope determined from 0-1 MPa confining pressures (thought to be representative of heavily jointed coal) v
= 55° (k =10)
5 6 7 Confining pressures (MPa)
Figure 2 Failure envelope for Huntly coal derived from triaxial tests on NX core
734
Back Analysis Monitoring
27.2.2 Stiffness Properties of NX Coal Core Young's modulus of the triaxial specimens of coal was dependent on the confining pressure of the test. When tested unconfined, the modulus was about 1 GPa, rising to 2.5 to 3 GPa at a confining pressure of 3 M Pa. For further increase in confining pressure there was no apparent increase in modulus. Poisson's ratio was found to be about 0.40. Statistical analysis of all the stiffness results shows no clear tendency for dependence on loading direction relative to the cleat direction. 27.3 TEST PANEL 27.3.1 Test Panel Layout A section of the mine, which in the long term was intended to act as a sump, was set aside for the investigation of the coal behavior. Herein this is referred to as the test panel. The plan and long sections of the test panel are shown in Figure 3. Instruments were installed in the test panel so that the response of the coal could be monitored during and after excavation of the test drive. Bends at D and L are a consequence of the unexpected basement high to the west of the test panel. It was initially intended that B-G and M-H would be straight and perpendicular to the face cleat direction, thus allowing several test drives of various cross section to be excavated. However, first the rising ground and second the small fault encountered at G and H limited the sites for test drive excavation to the one location between E and J. The monitoring drive, F-I, was located as shown to avoid the poor ground associated with the fault. The cleat direction is shown in Figure 3. Over sections B-D, cleat direction is approximately perpendicular to the drive but as the fault is approached its orientation changes to align with the fault, a swing of approximately 20°. The test drive was excavated using a roadheader, and the dimensions were carefully controlled during excavation to a rectangular profile 2.5 m high by 5.0 m wide. Subsequent overbreak occurred as a result of rib spalling leaving a final tunnel profile of slightly larger dimensions. The reinforcement used in the test drive was minimal. The ribs were not meshed. Two bolting techniques were employed. Over thefirst20 m, the bolts were vertical in the center of the opening and inclined nearer the rib. Over the second 20 m, all the bolts were kept vertical. This was not planned but happened by coincidence with a change in mine policy. No difference was apparent in the performance of the two techniques although it did take considerably longer to install the vertical bolts. Underground work began in October 1981 when access to the site was gained. The test drive was excavated in March 1983, and field testing was completed in September 1983 with the final pressuremeter tests and in situ stress measurement tests conducted from within the test drive. Excavation of the test drive occurred within afiveday period, working on two shifts per day and maintaining an approximately uniform rate of progress per day. The excavation was done by mine personnel using standard mining techniques and equipment to represent as nearly as possible a typical mine roadway. Initially the instruments were monitored by seven personnel for 24 hours per day. From day 3 to day 5, they were monitored by four personnel for 18 hours per day. Thereafter
Figure 3
Plan of the test panel
In Situ Testing and Monitoring of a Test Drive in an underground Coal Mine
735
readings were taken at increasing intervals for about six weeks. Changes noted during this latter period were generally not significant and the monitoring was discontinued. The timescale of the test drive excavation is shown in Figure 4. Although the average rate of advance was maintained at the required rate, it can be seen that progress over short distances was sporadic. The roadheader was able to advance 0.3 m and cut out the entire face in 3-5 min. However, only two or three such advances were possible in rapid succession. The load then had to be cleared by the relatively slow conveyor system. The time involved in these processes meant a fastest rate of advance of 3.5 m h ' 1 . Other delays that reduced this rate of advance were roof bolting (1 h every 4 m), survey adjustments, time spent setting up ventilation ducting, mealtimes and mechanical breakdowns. The average rate of advance of 3.5 m in a 6 h shift was close to the maximum that could have been sustained over the 10 shifts. 27.3.2 Test Panel Instrumentation The coal was monitored for stress changes and displacements. The instruments were installed two tofivemonths before the excavation, providing a rare opportunity to measure all the displacements and stress changes that occured. During this period, readings were taken intermittently to establish a zero reading, to assess the stability of the instruments and to check reproducibility. Displacements were measured in a horizontal direction using wire extensometers and in a vertical direction using manometers. A special system was designed to meet the demands of this project. As Hole through
|
8
20
c
1
8 io
0
U^
J
20
Actual progress
I
40
I
60
I
80
I
100
Hours from midnight Sunday 13/3/83
Figure 4 Progress of the test drive excavation
Figure 5 Instrument installations around the test drive
L
120
736
Back Analysis Monitoring
Figure 6
The extensometer system
many as eight extensometer points (of which six also had manometers) could be placed in a single borehole. The greatest distance between an extensometer point and the measuring head was 20 m. The extensometers were placed in 25 NX sized boreholes drilled into the immediate vicinity of the test drive. As the boreholes were generally not horizontal and the deformations within the rock were not necessarily either horizontal or vertical, it was only possible to determine the displacement vector at an anchor point if both an extensometer reading, assuming the direction of the borehole to be known, and a manometer reading were available for that point. Stress changes that occurred as the excavation advanced were monitored using a modification of the stresscell technique described in Section 27.6.1. Twenty one stresscells were ganged in strings of two or three per hole. The holes were drilled from the opposite side of the test drive to the extensometers holes to minimize the risk of unintentional interference between the different instruments. The layout of the instrumentation is shown in Figure 5. 27.3.2.1 Extensometer system The aim of the deformation monitoring system was to measure deformation in a concentrated zone around the proposed test drive. Unlike most extensometer applications, where the reference head, installed from within the excavation of interest, moves relative to the measuring point, the reference head in this case remained fixed while the furthest extensometer anchors moved. To maximize information from each hole, multiple extensometers were used to measure horizontal displacements in the rib and shoulders of the excavation. The extensometer wires were tensioned individually for each reading. The anchor points reacted against this force and had to resist it without movement. In addition, extensometer wires and manometer tubes from anchor points further along the hole had to pass through an anchor point without hindrance. A solution to these problems was provided by 22 gauge spring steel clips cemented to the borehole wall. The clips were formed to a diameter of 100 mm. The borehole was 75 mm in diameter. When released within the borehole, the clips pressedfirmlyagainst the wall. Before placement, the outside of the clip was coated with a layer of epoxy cement, of a type which cures in wet conditions. When the clip was released the epoxy cement bonded to the surface of the borehole and after 24 h the clip was firmly anchored in place. Extensometer wires and manometer tubes, riveted to the clip before placement, were thus anchored in the borehole. At the reading end of the wire an eyelet was formed. This loop hooked over the end of the reading device and acted as the point of reference for the extensometer readings. Each reading head consisted of a steel tube, 73 mm diameter and 4 mm wall thickness, cemented into the mouth of the borehole for a length of approximately 1 m with 300 mm extending clear of the rib. The protruding end had been machined to accept the reading instrument. A tension of 200 N was applied to the extensometer wire each time it was read. A digital micrometer was used to read the position of the extensometer clip relative to the reading head. Details of the movement system are shown in Figure 6. In practice, the most consistent readings were obtained by overtensioning, undertensioning, overtensioning and finally correct tensioning of the extensometer wire. The better consistency was thought to be due to more even tensioning of the wire throughout the length of the hole. It took one man approximately 3 h to read 100 extensometers. A check on the accuracy of the extensometers was made in the period before excavation began. All the extensometers were read and reread on the same day so any difference was due entirely to reading error. A distribution of differences between these readings is shown in Figure 7. An average
In Situ Testing and Monitoring of a Test Drive in an underground Coal Mine
i
Til
20
τ5 oa> 'S io
S E 3
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8 0.9
1.0
I.I
1.2
1.3
1.4
1.5
1.6
Difference between two sets of readings (mm)
Figure 7 Random errors associated with extensometer readings
Manometer tube
Figure 8 The manometer system
difference of 0.2 mm was determined with 90% of the readings having a difference of less than 0.9 mm. These are acceptable and indicate that the displacements can be measured to better than 1 mm. During excavation of the test drive, many of the extensometer boreholes were intersected. In these cases it was possible to watch the extensometer wire at the anchor point during the tensioning process. In all cases correct tensioning was confirmed.
27.3.2.2
Manometer system
The manometer system used to measure vertical displacements was less accurate than the extensometer system. The smallest division on the measuring board was 1 mm. Owing to the various practical difficulties in taking the readings, the reproducibility of the manometers is of the order of 2-3 mm. It was fortunate that the largest displacements were in a vertical direction in the roof and of the order of 10 mm. The manometer consisted of an open ended U tube system filled with water and had one end higher than the other. As the water flowed out of the lowest end in the borehole the water level was drawn down in the other end. Eventually the water stopped flowing and an equilibrium was established where the water level at both ends was at the same elevation. By measuring the elevation of the water column in the end outside the borehole, the elevation of the other end was established. A schematic diagram of the system is shown in Figure 8. The only constraint was that all air had to be cleared from the line. This was done by thoroughly flushing the 6.5 mm diameter lines as part of the overfilling process. In upholes, the excess manometer water flowed out of the holes and presented no problems. In downholes, however, the excess water from the manometers and seepage from the country tended to fill the borehole. To avoid this problem, a suction system was devised. Operated by a mine pump through a network of 15 mm PVC tubes leading to the bottom of each hole, the system allowed the boreholes to be kept dry. Each borehole had its own tube which was left in place.
Back Analysis Monitoring
738
Four measuring boards were established in the monitoring gallery to cover the range of elevations from 8.5 m to 16.0 m. Each manometer board consisted of a clear acrylic sight tube forming the measuring end of the U tube, and a measuring scale fixed to the board behind the sight tube. The manometer boards were rigidly fixed to either the rib or the brattice work props. The measurement scales were periodically leveled using optical surveying techniques and related back to a distant bench mark. It took one man approximately 6 h to read all the manometers.
27.3.23
Stress change system
The mode of operation, the casting details and strain gauges were the same as those for the stresscell described in Section 27.6. As many as three stresscells could be connected in series in one borehole. Wires and tubes from the more remote stresscells ran through the hollow center of the stresscells nearer the surface. The only constraint against more than three was the physical size of the wires and tubes. To equalize the line resistances a dummy gauge connection was made at the instrument rather than at the strain indicator. A laboratory test with the long leads indicated accurate functioning of the strain gauges and no problems with inability to balance or insensitivity were encountered in the field. An 11 pin quick-connect coupling on the end of the strain gauge wires allowed the 21 stresscells to be monitored sequentially using a single, 10 channel strain indicator. The stresscells were constructed and assembled in Auckland. For transport, the 30 m long placing rods, with the wires still inside them, were folded concertina fashion into a 6 m long bundle. At Huntly the rods were reassembled and glued together. Stresscells were installed in EX diamond drilled boreholes. To increase drilling speed and accuracy a 110 mm diameter borehole 30 m long was first drilled into the region of interest. The 38 mm diameter EX borehole was then drilled slowly, to minimize damage to the borehole surface, for a further distance of approximately 5 m. For monitoring stress changes the stresscells were installed at a much greater depth than was required for the measurement of in situ stress by overcoring. This made it more difficult to clean the boreholes prior to installation. The EX hole was washed clean with a special tool that sprayed high velocity jets of water onto the borehole surface. No attempt was made to clean the 110 mm diameter borehole. Instead, a 40 mm PVC liner tube was used over this section. The stresscell string was inserted through this liner, so keeping the instruments clean and helping to line them up with the small EX hole. The position of the drill bit was determined periodically during drilling. The distance into the rib was determined by the length of the drill string and the vertical elevation was determined using a manometer system with the drill rods making one arm of the U tube and the water hose the other. With this information, the positions of the long term stresscells, shown in Figure 9, were estimated. Pressure tests were conducted on all the stresscells prior to tunnel excavation (refer to Section 27.6). This allowed confirmation that gauges were functioning correctly. The horizontal cross-hole component of the borehole position was not easily determined. There was a tendency for the drill bit —*--^^
Fireclay Boundary material IOL
3L
ΙΟΜ
IOS 6L
3M
3S Δ | 2L " 2M 4L · 4M · 4 5
6M
6S
2S
8L
9Land9S (Mm) |8M
8S
7L
Test drive
Coal Fireclay Figure 9
Locations of the long term stresscells
7M
7S
In Situ Testing and Monitoring of a Test Drive in an underground Coal Mine
739
to wind itself sideways, towards the right, looking into the hole. Stabilizer rods reduced the effect but did not eliminate it. Lack of control over measuring the cross-hole component precluded using holes drilled parallel to the test drive for stress monitoring, as accurate location in this direction is essential for interpretation of the results. The cross-hole position of the stresscell was also of interest in holes drilled perpendicular to the test drive. However, the exact location was determined only if the borehole daylighted in the test drive or another borehole. S7 and S8 were extended to intersect the test drive. S8 was encountered during drivage but no sign of S7 was found. S2 and S3 fortuitously intersected an exploration hole running parallel to the test drive, which gave a fix on their positions. S10 was located purely by chance when a piece of stresscell was recovered in the core of a vertical pressuremeter hole. It was also possible to identify the exact position along the string from whence the piece had come. Thus the three stresscells in S2, S3, S8 and S10 were located reasonably accurately. Positions of the remaining stresscells, those in S4, S6, S7 and S9, were estimated. 27.4 MONITORING DURING AND AFTER THE TEST DRIVE EXCAVATION 27.4.1 Deformations At cross sections 4 and 5, the first to be passed, the readings were taken many times and a clear pattern was observed. At later cross sections, the readings were taken less frequently owing to the length of time to read all the points. In these cases it was not always possible to determine clearly the average reading either immediately before or just after the excavation had passed. Most of the displacements occurred gradually as the face passed the measurement point. The results for extensometer 5A3 are shown in Figure 10. As with all the boreholes that did not daylight in the test drive, the exact location along the length of the drive of the extensometer 5A3 in borehole 5A is not certain. Therefore the exact time at which the excavation passed the measurement point cannot be determined. The best estimate of this time based on the positions of the other boreholes in the cross section is shown in Figure 10. Displacement vectors computed from the measured displacements from all 24 boreholes are presented in Figure 11. Anchor point positions are drawn relative to the opening. The displacement of the anchor point in two dimensions could only be determined if both the extensometer and manometer readings were available. These are represented in Figure 11 by solid lines. The displacements determined from either an extensometer or a manometer reading alone are represented in Figure 11 by dotted lines. The arrowheads are not drawn when the displacement is less than 2 mm. It can be seen that the downward displacements occurring in the roof are consistently greater in magnitude than the horizontal displacements in the rib. The maximum vertical displacements in the roof above the opening are of the order of 10 mm while those in the rib are negative (away from the
Excavation period
Ί
2 E E
w
Uj
~
Face passes
l 0 6
.
*&>.
U.-Î
i0.9 mm
"Γ
2
I °
6 mm
i -2 -4
-5 h -IOOO
0
J
20
I
40
I
60
I 80
L 100 Time (h)
_L 200
Figure 10 Displacement record for extensometer 5A3
J_ 400
I
600
740
Back Analysis Monitoring
», '
I· I ,
· . *· I
Test drive
Displacement exaggeration 5 0 x Displacements < 2 mm — Displacements > 2 mm - ♦ One component only ■··♦ Inferred displacement
♦ J -f
Figure 11 Measured displacement vectors
opening) and are less than 1 mm. The relative magnitudes of the displacements are discussed further below in conjunction with the finite element analysis. 27.4.2 Stress Changes Locations of the individual long term stresscells relative to the test drive are shown in Figure 9. One was in line with the opening and was actually excavated. Stresscells in boreholes 2, 3 and 4 were located just above the line of the roof. The change in grade midway through the test drive was made so as to accommodate these instruments. Stresscells in 6 and 10 were located 1.5-2.5 m above the opening and spanned from one rib to 2.5 m into the other. Two controls were placed in borehole 9 at a distance of 21 m (4 x width) from the tunnel centerline. The position of thefireclayrelative to the roof and floor varies slightly over the length of the test drive. Stresscell locations have been reduced to a single plane and are drawn relative to the opening. Therefore the seam boundaries shown in Figure 9 are approximate only, corresponding to a section about 18 m along the test drive. The plot of strain versus time for stresscell 10S (Figure 12) is the best example of the elastic behavior of the coal. Until the test drive excavation advanced to within 6 m of the stresscell location no significant strains were registered. Unfortunately the next reliable reading was not taken until the excavation was directly beneath the stresscell location (at 36 h). Therefore it is not certain exactly when the change began to occur, other than to say that it began within 6 m (approximately one tunnel width) of the face of the excavation. This is consistent with the stresscell overcoring plots which do not register the advance of the overcore hole until it is within one diameter of the strain gauges. The next reading of the stresscells was taken at 65 h when the tunnel had advanced 2 m beyond the stresscell location. By this time, all the stress changes had occurred. The readings taken over the next four weeks, when the monitoring was discontinued, showed no significant change from the state of stress that had been established 2 m after the face had passed. At this stage it is interesting to consider further the effect of the approaching drive. The record of 2L is shown in Figure 13. This instrument was sited 0.5 m above the opening. A reading was taken while the excavation was still 2 m distant. It can be seen that the stresscell had registered the approach of the excavation. The strains are mainly in the opposite direction to thefinalstrains but of lower magnitude. This is similar to the behavior of the in situ stress measurement cells during overcoring and is consistent with a stress concentration ahead of the advancing excavation. As the excavation passed under 2L some large tensile strains consistent with stress relief occurred, and these remained approximately constant throughout the remainder of the monitoring period. Similar behavior is noted in 2M and for all of the stresscells in borehole 4. Eight stresscells placed in the rib do not show strain changes as large as those observed in the roof. Proximity of the excavation was registered when it approached to within 4 m of 8L and 8M. When 8L was excavated, the wires to 8M were damaged and further readings were precluded. 8S did
In Situ Testing and Monitoring of a Test Drive in an underground Coal Mine
741
4000
c
2000
Ë «A
0
Time (h) Figure 12
0
Record for stresscell 10S
100
Figure 13
Time (h )
200
Record for stresscell 2L
not give a reliable result. Strains of 7L are steady once the excavation has passed. Slightly compressive circumferential strains indicated a small increase in stress. The two stresscells used as controls also show small strains although three of the gauges on 9L became irregular in the later stages of monitoring. This is thought to be a consequence of gauge malfunction rather than load redistribution. Readings for the long term stresscells were analyzed to determine the magnitude and direction of the stress change. The stress changes are three-dimensional but it is more convenient to reduce them to two dimensions in a plane perpendicular to the line of the opening. For the most part, the calculated stress changes align with the opening (10-20°) and only a small adjustment to the calculated magnitude would be necessary. For simplicity, the out of plane shear stresses were ignored and the in-plane stresses were considered to be principal stresses with the directions shown in Figure 14. It can be seen that the largest magnitude stress changes are observed in the roof. In stresscells 10S, 10M and 10L, the maximum stress changes are 3MPa, 5.5 MPa and 4MPa respectively in a direction toward the center of the opening. The fact that they are approximately symmetrical about the opening is further evidence that the instruments are functioning correctly. Also shown in Figure 14 are in situ stresses determined after the test drive was excavated.
Back Analysis Monitoring
742
5.5 MPa
1.4 MPa In situ stress
7L
7M
Figure 14 Stress changes measured using the long term stresscells
The result of 3 M suggests that the maximum principal stress at that location after excavation is approximately 1.8 MPa higher than the in situ stress and therefore is approximately 7 MPa. This compares favorably with the value of 6.6 MPa determined in the finite element analysis discussed below. In the rib, 7L and 7M are the only stresscells that bear analysis. In both cases, stress changes measured are compressive in a near vertical direction and tensile in a direction toward the drive. Vertical stress changes were very small (0.13 MPa and 0.16 MPa) and somewhat lower than predicted by the finite element analysis. Where changes in stress measured along the direction of the drive are in the range 0—1.2 MPa, these relatively low stress changes were of the same order as those predicted by the finite element analysis. 27.4.3 General Observations Stresscell and extensometer results suggest that the behavior of a single opening is predominantly elastic and prefailure. This is thought to be the general case in the test drive area apart from the small scale rib buckling failures and tension cracks in the roof and floor. As the excavation was driven and for several days afterwards the ribs emitted audible creaks, groans and thumps which are normally associated with compression or shear failure. This suggests that there were some regions of failure in the coal. Underground noises are very deceptive and it is possible that the noises were near surface effects associated with rib buckling. In several instances, though, it sounded as if the noises were emanating from deep in the rib, thus suggesting failures within the pillar and an associated redistribution of stress. These noises were not heard coming from the roof. Although monitoring was discontinued after four weeks, the area was visited regularly for a further period of several months. Apart from small-scale rib spalling, there was no apparent change during this time, and no further noises. It is considered that all the activity had ceased within a few days of excavation. The line painted at midheight along the ribs near to the test drive was intended to act as an indicator of areas of rib failure where excess spalling occurred. In general, the rate of spalling did not alter noticeably in the periods before, during or after excavation. A point at which large-scale spalling was expected was in the diminishing pillar created ahead of the test drive just before holing through. In fact, this remained intact right up until the roadheader punched through. At this stage it was 0.3 m thick. The cleat was running in the most favorable direction which undoubtedly increased its stability. It is still interesting that it remained intact for so long. Loose roof coal was observed in the test drive area. In several instances, tension cracks could be observed running along the length of the drives, particularly in those drives parallel to the cleat. These are discussed further below. As well as the visible cracks, there was also an observable looseness in the roof coal. When the hydraulic legs of the drilling rig were jacked up between roof and floor, the roof could be seen to move upward a distance of the order of 2-5 mm over an area of 0.5 m. When a vertical hole was drilled into roof coal in the monitoring drive (F-I in Figure 3), water began dripping from the roof up to 3.0 m from the 1.1m deep hole. It dripped from cleats and from
In Situ Testing and Monitoring of a Test Drive in an underground Coal Mine Endcap
Flexible membrane
/
/
Internal pressure cavity
Spring steel cantilever strip
Strain gauge Figure 15
743
Body
Inturned self sealing ends
Feeler arm
Pivot
Counterbalance cantilever
Details of the N X pressuremeter
bolt holes. Drilling was stopped immediately and the hole was abandoned. The experience illustrated that the roof coal was very loose and had possibly separated in some places. During excavation of the test drive, an extensometer was installed from within the drive and a row of reference pins was set up across the roof. After an initial settling down period, where movement attributed to anchor creep was observed, no movement was observed in either the extensometer or across the reference pins. 27.5 IN SITU PRESSUREMETER TESTING OF THE COAL AND FIRECLAY STIFFNESS 27.5.1 Pressuremeter A pressuremeter was constructed for measuring the stiffness of coal andfireclaysurrounding NX boreholes. The concept of the instrument is shown in Figure 15. The membrane is manufactured from polyurethane and is cast in such a way that the ends are self sealing. Changes in diameter are measured in orthogonal directions with an internal feeler mechanism. Strain outputs from the instrument are calibrated to changes in diameter with a small correction being made to allow for changes in thickness of the membrane as the diameter is increased. The membrane was pressurized hydraulically with a hand operated pump. 27.5.2 Results Figure 16 shows the result of a typical pressuremeter test in coal. The first time that the pressure is applied deformations are greater than for the subsequent cycles. Reasons for this behavior are thought to be nonrecoverable closure of joints, bedding in of the membrane and, at higher pressures, formation of radial tension cracks. These cracks reduce the apparent modulus of the coal by about one half. The average value obtained for the modulus of the coal surrounding the borehole was 1.7 GPa. Young's modulus for coal obtained from the pressuremeter phase of the measurement of in situ stress (refer to Section 27.6) was 2.8 GPa (the pressure used was less than that required to form radial cracks).
Back Analysis Monitoring
744 10 H
8 H
//l
8?
a.
,
/
200
/
//i ^
400 600 Strain (/is)
800
Figure 16 Results of a pressuremeter test in an NX borehole in coal
Fireclay is a term used to describe all materials other than coal, i.e. coal measure rocks above and below the seam. The pressuremeter modulus prior to formation of radial cracks is 5-6 GPa. Poisson's ratio for fireclay, obtained from tests on core, was in the range 0.17-0.2. 27.6 MEASUREMENT OF IN SITU STRESSES 27.6.1 Technique A common method of measuring in situ stresses in rock is based on overcoring of a strain gauged probe, or stresscell, which has been fixed in an EX borehole. In low modulus materials special techniques are required because the stresscell itself must not restrain the borehole. Stresscells of the CSIRO type (Worotnicki and Walton [4,5] ) were tried in Huntly coal but were found to be too stiff. After this experience, a stresscell for measuring in situ stresses in coal was especially developed by the authors - details can be found in Mills [1] and in summary form in Mills and Pender [3]. The instrument attaches three rosette strain gauges to the surface of an EX borehole. This is achieved by mounting the gauges on the outer surface of aflexiblepolyurethane membrane which forms part of the stresscell assembly. Before the cell is inserted into the borehole its outer surface is coated with epoxy cement. On placement in the borehole the membrane is inflated and when the epoxy cement sets the membrane and strain gauges are bonded to the coal surface. The ability to apply pressure to the membrane means that the stresscell can be used as a small pressuremeter. Pressure is applied and the response of the strain gauges noted. In this way correct operation of the strain gauges can be confirmed before the start of overcoring. If they are operating correctly the data can be used to infer the modulus of the coal that will be overcored. Thus separate determination of the modulus of the overcored material, using, say, a biaxial test, is not a prerequisite for reducing measured strains to stresses. 27.6.2 Results The measured in situ stressfieldis shown in Figure 17. Of considerable interest is the fact that the in situ stresses in a horizontal plane are less than the vertical in situ stresses. It is apparent that the test drive is oriented very close to the direction of the intermediate principal stress. In situ stresses in the coal were measured at a distance of 3-5 m from the opening and the measurements took place after the test drive had been excavated. At this distance the presence of the opening will influence the virgin in situ stress field. The effect can be seen by referring to Figure 14 in which the stresses determined by overcoring are plotted along with the stress changes observed during the excavation of the test drive. Stresscells 7L and 7M registered a very small change in vertical stress during the excavation so the vertical stress of 5.5 MPa, measured during the overcoring, is a component of the virgin in situ stress state. It is apparent that there was a decrease in horizontal stress at instruments
In Situ Testing and Monitoring of a Test Drive in an underground Coal Mine
745
N
* 4.5 ΜΡα
Test drive
1.5 MPû U.H.
Test * T307 + T307 X T307 Θ Τ308
(3800) (3100) (3580) (3830)
• T3 (best) ΟΤ3 (all) ■ T2,T3(best) DT2,T3(all)
Figure 17
6.1 6.0 6.0 6.6
σ2 4.5 3.8 4.2 4.0
1.5 1.2 1.3 0.4
6.1 5.8 6.5 5.6
4.5 5.3 3.5 4.8
1.2 1.5 0.9 1.2
σ-3
Measured state of in situ stress in the coal
7L and 7M during the excavation. Thus the virgin minor principal in situ stress will be slightly greater than the 1.5 M Pa shown in Figure 17. 27.7 FINITE ELEMENT MODELING 27.7.1 Input Data The aim of this section is to model the test drive excavation process using the finite element method and thereby investigate if the measured coal properties and in situ stresses are consistent with observed behavior. Since there is a degree of uncertainty in some of the input parameters, due to scale effects and sampling bias, these were individually varied about the measured value to see what effects each one had on stress and displacement distributions. Horizontal stress, elastic modulus offireclay,Poisson's ratio of coal and properties of the boundary layer were varied. A failure condition was not included in the model so stress and displacement distributions are the result of linear elastic behavior only. This produces zones of tension which would manifest themselves as cracks in reality. It also tends to produce higher stresses near the opening. Failure in zones of high stress and tension would cause redistribution of loads to other areas. Therefore the elastic analysis with equal properties in tension and compression is somewhat simplified. While sophisticated modeling techniques are available to simulate failure, roof bolts, variations of modulus with confining pressure, dilatancy and time dependence, it was first necessary to investigate whether or not the observed behavior could be approximated by a simple elastic model. As most of the
746
Back Analysis Monitoring
0 I
5 (m) I
Figure 18 Finite element mesh used for test drive simulation
displacement and stress change results suggested that the test panel was within the elastic range, a linear elastic model is considered appropriate. To simulate the test drive, the following assumptions were made, (i) Geometry: the coal seam is 9.5 m thick and is level; the opening is rectangular, 2.5 m high by 5.0 m wide; there are 2.5 m of coal on the floor and 4.5 m on the roof, (ii) Material properties: linear elastic behavior with no failure and the same properties in tension and compression; for the coal Young's modulus = 2.5 GPa and Poisson's ratio = 0.4; for the material above and below the coal (fireclay) Young's modulus = 5 GPa and Poisson's ratio = 0.2. (iii) In situ stressfield:the horizontal stresses are less than half the vertical; in situ stresses in the coal seam and fireclay are the same; the vertical and horizontal stresses in the plane of the problem are principal stresses, (iv) Boundary conditions: the test drive is symmetrical about a central vertical axis so that a half mesh can he used; the boundaries of the mesh arefixedin the normal direction; nearby openings were assumed not to influence the stress and displacement distributions in the region of the test drive. The mesh used in the analysis is shown in Figure 18. 27.7.2 Results 27.7.2.1 Displacement distribution The distribution shown in Figure 19(a) was determined using the material properties, in situ stresses and geometry discussed above. The measured displacement distribution from Figure 11 is reproduced again in Figure 19(b). There is good agreement between measured and calculated distributions. Horizontal rib displacements in both cases are small in comparison to the vertical displacements. Maximum vertical roof displacements are of approximately equal magnitude. This confirms the modulus of 2.5 GPa used in the analysis and is further confirmation that the horizontal in situ stress acting across the opening is much lower than the vertical stress. Figure 20 shows a comparison of the measured and calculated rib displacements near the opening. Immediately adjacent to the opening the measured displacements were negative (i.e. away from the opening), ranging in magnitude from — 1.13 mm to + 0.2 mm with an average of — 0.7 mm at the surface. This distribution is not reflected in the calculated values. Even when horizontal in situ stress is zero, displacements are positive at the surface. For the measured stress ratio positive rib
In Situ Testing and Monitoring of a Test Drive in an underground Coal Mine
747
(a)
»
*
\ \
Coal
i
\
M
\
I
i
i
* l
'
k \ \ i i *
11 * * J
Displacement exaggeration 5 0 x Fireclay Calculated displacements
(b)
Test drive Displacement, exaggeration 5 0 x Displacement < 2 m m -~ Displacement > 2 m m " * One component only ···* Inferred displacement Measured displacement
Figure 19 Comparison of measured and calculated displacement distributions
displacements occur at all depths. While no explanation can be suggested for the discrepancy between measured and calculated rib displacement distributions, the fact that measured rib displacements are small in comparison to those in the roof is consistent with a low horizontal stress. Rib displacement increases rapidly as the horizontal in situ stress increases, rising to 6.0 mm for hydrostatic conditions. 27.7.2.2 Stress distribution The stress distribution generated by the finite element analysis for the test drive excavation is shown in Figure 21. A most notable feature of this stress distribution is the zone of tension in the roof, floor and to a lesser extent in the rib. In reality, the coal is unable to sustain tensile stresses, so cracks would appear in areas where tensile stresses are calculated. At midspan in the roof beam, this would cause cracks running longitudinally along the drive. Such cracks were indeed observed. In addition, a general loosening of the roof coal was also observed. This is consistent with the low state of stress calculated in the roof.
748
Back Analysis Monitoring __ Model rib displacement for a range of ah /σ ν ratios — · —
Measured displacement
L· -1.0 h 4.5
Distance from centerline (m) Figure 20 Measured and calculated rib displacements Test drive simulation σν = 5.0 MPa
ah = 1.5 MPa
Ef /Ec = 2.0
+ + +
3(m)
10 MPa Figure 21
Scale
J
Stress distribution about the test drive
There is a zone of horizontal tension in the rib immediately adjacent to the opening at midheight. This combined with the higher vertical stresses would tend to cause the buckling behavior that was observed in the full-scale excavation. It should be noted that the propensity for such a failure is strongly influenced by the joint orientation. In the tunnel monitoring program, stresscells registered a large reduction in the roof stresses as the tunnel was excavated. The instruments in borehole 10 indicated a reduction of approximately the same magnitude as the in situ stress. This is consistent with the calculated stress distribution. If the
In Situ Testing and Monitoring of a Test Drive in an underground Coal Mine
\
I
x x xi
I
XX X x/< • I X X X XX X X ^ X , / |X X X X X X X IX X X X X X X X
*3'
x
Tensile failure
•
Compressive failure (σβ = 5 Μ Ρ α )
749
x γ IX X X X X X X X _ ^
X XX XX X X i · x N s X XX X X X XNX \ X X X X X X X \ X ·Ν X X X X XX \ · / X X X
X X
X
X
Figure 22 Zones of failure around the test drive
horizontal in situ stress was greater than the vertical, the horizontal stresses in the roof would increase as a result of the excavation. The calculated stress field is therefore consistent with observation and with long term stress measurements made during the monitoring program. 27.7.2.3 Failure zones Regions of failure surrounding the test drive are shown in Figure 22. Regions at which tensile failure would occur are indicated by a cross. Regions in which compressive failure would occur because the stresses exceed the strength given by the failure envelope are indicated by a dot. Unconfined compressive strength is used to represent rock strength if the minor principal stress is tensile. The unconfined compressive strength was assumed to be 5 MPa. In general, failure occurs in tension in the roof, floor and to a small extent in the rib. This results in cracks at these sites, but does not always endanger the opening stability. It is, of course, undesirable to have large zones of tension in the roof since some form of artificial support is then required to reinforce it. Figure 22 shows that compressive rib failure is expected near the opening. This was observed in the test drive. Formation of this failure mode is thought to be assisted by joint orientation. At the top and bottom corners of the rib, particularly when they are sharp, very high stresses occur. For the test drive, compressive failure is not calculated in these areas nor was any observed. However, the coal is very near to failure and in many areas of the mine this type of failure can be observed. A reentrant forms if some form of confinement is not provided. This sometimes extends by 0.5-1.0 m into the rib or roof. Such failure is undesirable because it increases the effective width of an opening and so weakens the overall structure. 27.7.2.4 Other factors Stress and displacement distributions calculated using the elastic finite element model compare well with observations made in the tunnel monitoring program. This gives confidence in the model, measured stresses and measured material properties.
750
Back Analysis Monitoring
The effects of increasing horizontal in situ stress, changing modulus of the coal, changing modulus and Poisson's ratio of the fireclay, using properties for the boundary layer different from those of the fireclay were all investigated with the finite element analysis. Horizontal in situ stress was found to be the most significant parameter. Only when the in situ horizontal stress was about half the vertical stress were the computed and observed displacement fields compatible.
27.8
SUMMARY AND CONCLUSIONS
This chapter has described a case study of in situ testing and monitoring about a test drive in an underground coal mine. In situ testing was supplemented with laboratory tests on core from the coal seam. Behavior observed in the test drive was compared with an elastic finite element analysis of the excavation process. The project demonstrates the advantages of a comprehensive program of rock testing and monitoring. Confidence in the in situ stress measurements and the stiffness values determined for the coal mass was greatly enhanced by monitoring of the test drive and the finite element analysis. The fact that there was reasonable agreement between the observed and computed displacement fields provides confirmation that measured in situ stresses were of the right order and that the properties assigned to the coal mass were realistic. Determination of a horizontal in situ stress considerably less than the vertical stress is of great interest. This measurement also confounds the common assumption that, in a 'soft' material like coal, the in situ stress state will be hydrostatic. Rather simple techniques were used to monitor displacements and they worked adequately. A stresscell for the measurement of in situ stress in coal was especially developed for the project. It was also used to monitor stress changes. From the point of view of new innovations this aspect of the project is very important as previous in situ stress measurement techniques have not been successful in coal. Another aspect of the stress measurement technique that is notable is the pressuremeter test prior to overcoring. This serves the dual functions of verifying correct operation of strain gauges and providing the modulus of the surrounding rock for interpretation of strain changes. In this way the need to determine the modulus of an overcored section of rock can be circumvented in difficult conditions.
ACKNOWLEDGEMENTS Financial support for the project was provided by the New Zealand Energy Research and Development Committee. The work described would not have been possible without considerable assistance from the staff of State Coal at Huntly. Both of these sources of assistance are gratefully acknowledged.
27.9
REFERENCES
1. Mills K. W. In situ mechanical behavior of Huntly coal, Report 406, Civil Engineering Department, University of Auckland (1986). 2. Mills K. W., Pender M. J. and DePledge D. Measurement of m situ stress in coal. In Proc. Int. Symp. Rock Stress and Rock Stress Measurement, Stockholm (Edited by O. Stephansson), pp. 543-549. Centek, Lulea (1986). 3. Mills K. W. and Pender M. J. A soft inclusion instrument for in situ stress measurement in coal. In Proc. Int. Symp. Rock Stress and Rock Stress Measurement, Stockholm (Edited by O. Stephansson), pp. 247-251. Centek, Lulea (1986). 4. Worotnicki G. CSIRO triaxial stress measurement cell. In Comprehensive Rock Engineering (Edited by J. A. Hudson), vol. 3, chap. 13, pp. 329-394 Pergamon, Oxford (1993). 5. Worotnicki G. and Walton R. J. Triaxial hollow inclusion gauges for the determination of rock stresses in situ. In Proc. ISRM Symp. Investigation of Stress in Rock and Advances in Stress Measurement, Sydney, suppl., pp. 1-8 (1976).
28 Subsidence Behavior of Rock Structures BARRY N. WHITTAKER Formerly of University of Leeds, UK and DAVID J. REDDISH University of Nottingham, UK
28.1
INTRODUCTION
28.1.1 28.1.2 28.1.3 28.2
752
Subsidence Occurrence and Influence of the Natural Environment Role Played by Time in Relation to Geological and Mining Extraction Features Importance of Subsidence Engineering in Relation to Design of Structures Overlying Rock Formations
GENERAL CHARACTERISTICS OF SUBSIDENCE
28.2.1 28.2.2
754 754 756
Natural Subsidence Phenomena Mining Induced Subsidence
28.3 PRINCIPAL 28.3.1 Observed 28.3.2 Observed 28.3.3 Observed
752 752 754
BEHAVIORAL CHARACTERISTICS OF MINING SUBSIDENCE Behavior of Subsidence: UK Behavior of Subsidence: Germany Behavior of Subsidence: Europe and other Countries
758 758 759 760
28.4 BASIS FOR PREDICTION OF LONGWALL SUBSIDENCE 28.4.1 Development of Empirical Prediction Method 28.4.2 Influence of Goaf Treatment on Maximum Subsidence 28.4.3 Subsidence Development Curve 28.4.4 Extent of Influence of Subsidence 28.4.5 Ground Tilt
760 760 762 763 763 764
28.5 LONGWALL MINING SUBSIDENCE PREDICTION: THE UK EMPIRICAL MODEL 28.5.1 Subsidence Prediction 28.5.2 Prediction of Strain due to Subsidence
764 765 765
28.6
GEOLOGICAL FACTORS INFLUENCING LONGWALL MINING SUBSIDENCE
28.6.1 28.6.2 28.6.3 28.6.4 28.6.5 28.6.6
Site Conditions and Effect of Local Geology Bedrock Conditions and Subsidence Hydrogeological Effects in Respect of Mining Influence of Cover Rock Type on Longwall Subsidence Influence of Geological Faults on Surface Subsidence Surface Topography and Influence of Seam Inclination
28.7 PREDICTION OF MINING SUBSIDENCE: GENERAL PERSPECTIVE OF THE MAIN METHODS 28.7.1 Profile Functions 28.7.2 Influence Functions 28.7.3 Numerical Models 28.7.4 Physical Models
767 767 767 768 769 769 770 771 772 772 772 113
28.8 PREDICTION OF SINK HOLE DEVELOPMENT ASSOCIATED WITH ROOM AND PILLAR MINING 773 28.8.1 Basic Considerations 773 28.8.2 Prediction of Caving Height 773 28.8.3 Caving Height Above Collapsed Junction 775 28.8.4 Effect of Water Gaining Access to Collapse Chimney 775
751
Back Analysis Monitoring
752 28.9
SUBSIDENCE ASPECTS IN RELATION TO ABANDONED MINES
28.9.1 Nature of Problems Associated with Abandoned Mines 28.9.2 Examples of Collapses Associated with Abandoned Metal Mines 28.9.3 Sink Hole Occurrences above Coal Mine Workings: US 28.9.4 Sink Hole Development: UK
776 776 111 111 111
28.10
CONCLUSION
779
28.11
REFERENCES
779
28.1 INTRODUCTION 28.1.1 Subsidence Occurrence and Influence of the Natural Environment Subsidence is a natural phenomenon which has been occurring on localized, regional and continental scales since the formation of the Earth's Crust. Subsidence is also associated with the extraction of minerals and natural resources such as oil, gas and water. The natural geological setting plays an important role in the character of the resulting subsidence. For example, in many mining situations, the depth of the mineral deposit and the overall competence of the overlying'rock structures may result in insignificant subsidence of the surface even though substantial underground extraction of minerals may have occurred. In the latter situation there is however potential for subsidence to occur in the future as is encountered above some forms of abandoned mines many years after cessation of the mining operations. Subsidence can take the form of a general lowering of the surface and occur as a saucer shaped depression. Such subsidence features can be very extensive involving multiples of kilometers where natural regional subsidence is involved. This contrasts markedly with that of the occurrence of a localized surface collapse above a solution cavity formed naturally in limestone or evaporite rock types. In the latter situation the subsidence depression can have distinct edges and suddenly appear at the surface following collapse of the diminishing cap rock thickness. These forms of subsidence occur naturally and are associated with tectonic processes and the weathering and degradation by solution processes. Where solution degradation and weathering play major roles in the occurrence of subsidence, the rock types which are generally susceptible to such processes, e.g. limestones and evaporites, play major roles as does the nature of subterranean water channels. Figure 1 shows examples of natural subsidence features associated with limestone country. Figure 1(a) shows a large conical depression formed as a result of the action of water. Most subsidence features in limestone country adopt a conical depression form of this nature. The magnitude of such depressions depends upon the significance of the original subterranean water course and resulting cavity. Many such depressions are relatively insignificant but can be appreciably deep as demonstrated by this photograph. Figure 1(b) contrasts markedly with that of 1(a). The large subsidence depression illustrated in Figure 1(b) shows a collapsed form of underground solution cavity which has resulted in clearly defined edges at the bedrock horizon together with sloping sides in the material forming the overlying unconsolidated deposits. Figure 1(c) shows detail of the subsidence depression illustrated in Figure 1(b). 28.1.2
Role Played by Time in Relation to Geological and Mining Extraction Features
Time plays an important role in the occurrence of subsidence at the surface. Where subsidence occurs in association with earthquakes, it may happen almost immediately or shortly thereafter in the form of slope instability caused by significant change to the immediate topographical character. Subsidence occurrence in association with the formation of a major sedimentary basin as part of the weathering process, namely the depositional stage, can take thousands of years to develop and consequently, the impact of such subsidence of the surface is hardly recognizable over time scales involving just a few years. Surface collapses above solution cavities may manifest themselves suddenly by failure of the cap rock covering or they may appear as a small conical depression which gradually increases in size but does not give rise to a sudden collapse form of failure. The latter situation is very much governed by the nature and thickness of the unconsolidated materials overlying the bedrock. Time also plays an important role in the occurrence of subsidence at the surface above underground mining extractions. Where extensive underground caving is practised as with the
Subsidence Behavior of Rock Structures
753
Figure 1 Natural subsidence features in limestone country: (a) large conical subsidence depression caused by action of water and subsequent collapse of sides, (b) subsidence of surface created by collapse of large underground solution cavity, and (c) detail of (b) illustrating scale and steepness of sides.
754
Back Analysis Monitoring
longwall mining method then there is a correlation between the development of mining and the progressive development of the surface subsidence trough. This is not generally the case, however, where the mining method results in the formation of surface support pillars which remain as permanent features as is often the case with room and pillar mining. In such situations some subsidence of the surface may occur during the mining process but is more likely to occur at a later date and, in some cases, many years after completion of mining owing to collapse of mine room intersections and the progressive collapse towards the surface of a cavity which can ultimately reach the surface in the form of a sink hole or crown hole collapse. The general rock types forming the cover above such mining operations play an important role in such a process, for example, if competent sandstones or other such rocks formed the rock cover then no such sink holes may occur providing the depth below surface is not shallow to the extent that natural surface weathering processes had not significantly weakened the cover rocks. 28.1.3 Importance of Subsidence Engineering in Relation to Design of Structures Overlying Rock Formations Knowledge of subsidence engineering is of considerable importance to the planning and development of the surface in those areas which may be deemed to be prone to significant subsidence phenomena. Firstly, it is vital to be aware of the risk of possible surface collapses above underground openings whether they be natural solution cavities or mining excavations. In such situations, knowledge should be established, as far as practicable, as to the nature, state and dimensions of such openings in addition to their depth below the surface. The nature of the cover rocks, and unconsolidated materials, should be established as these form an important influencing factor on the nature of any resulting subsidence. Prediction methods exist for determining the likely magnitude and extent of subsidence development at the surface above different types of underground opening. Subsidence and accompanying changes in ground strain and tilt can be determined for the purpose of ascertaining how such effects can influence existing surface structures. New surface structures can be designed to respond favorably to subsidence effects due to longwall mining operations, particularly in respect of foundation design. In other situations where there is a marked degree of uncertainty concerning the possible occurrence of subsidence following completion of mining operations, surface planning needs to take this into account and, in some cases, such areas need to be avoided or systematic underground filling of voids needs to be adopted in order to be assured of surface stability. Much can be done in terms of field investigation, instrumentation and monitoring in order to establish a clearer picture of the risk of surface subsidence and its likely effect at the surface. Subsidence engineering principles can be applied to good effect in order to gain an improved appreciation of surface stability in those areas deemed at risk to the subsidence process either from mining or from localized natural occurrences. 28.2 28.2.1
GENERAL CHARACTERISTICS OF SUBSIDENCE Natural Subsidence Phenomena
Subsidence on a regional scale involving multiples of square kilometers occurs as a natural geological feature as part of sedimentology. Additionally, tectonic activity within the Earth's Crust involving continental plates results in wide scale subsidence. The formation of rift valleys in association with major faulting also results in widespread subsidence. These forms of subsidence can take place over long periods of time involving up to thousands of years or more. Earthquake activities can trigger subsidence either in the form of slope instability or increased settlement of sediments to occur at the time of earth tremors. Localized natural subsidence can occur due to changes in drainage patterns and lowering of the water table generally. The shrinkage process can cause surface subsidence in addition to the generation of significant lateral strains in some cases. The principal form of localized natural subsidence is that of the occurrence of a sink hole at the surface. This can occur due to collapse of a naturally formed underground cavity. Such naturally formed cavities can collapse progressively until reaching the surface where they can form a well defined hole with steep or even overhanging sides. Such holes can also take the form of a conical shape. Sink holes generally occur suddenly and often without warning. They consequently represent a special hazard to the surface. Underground cavities can be formed in lava due to volcanic action or
Subsidence Behavior of Rock Structures
755
occur as a result of solution in evaporites or limestone formations. Water plays a significant role in their formation and frequently encourages their upward propagation to form a sink hole at the surface. Lowering of the water table in unconsolidated materials lying over karst formations can trigger the occurrence of extensive sink hole developments due to the change in the subterranean drainage pattern at the bedrock horizon. This is a special problem in South Africa above gold mining operations in the Transvaal. (α)
(b)
Sinkhole
Figure 2 Example of natural subsidence associated with a salt dome (a) illustrating effect of water percolating in contact with salt dome, which creates solution cavity, and (b) illustrating collapse of strata above solution cavity to form a subsidence depression at the surface; continuing percolation of water in contact with salt dome can allow further episodes of subsidence to take place
(α)
Figure 3 Landslide aspects in relation to subsidence: (a) landslides are a form of natural subsidence with pronounced lateral displacement, and this illustrates a common form of occurrence, (b) illustrating Assuring at the crest of a localized slope instability, which is subsiding
756
Back Analysis Monitoring
Figure 2 illustrates a particular natural subsidence problem associated with salt domes. The salt plug is being thrust through a sequence of sediments. It is common for the salt dome to come into contact with groundwater regimes which allow the salt to become dissolved with the formation of a natural cavity. Should the solution cavity become sufficiently large then it is possible for the overlying rocks to collapse into the cavity and give rise to an upward collapse process which can eventually cause a sink hole to form at the surface. The collapse process is frequently affected by water draining into the collapse chimney and thus allows the opportunity for further dissolution of the salt and accompanying acceleration of the collapse process. A further common form of natural subsidence is that of slope instability and landslides generally. The subsidence is localized and it causes damage to surface structures by differential movement. Figure 3(a) illustrates a common form of subsidence associated with slope instability, whilst Figure 3(b) illustrates the character of tensile cracking in a road in mountainous country. 28.2.2 Mining Induced Subsidence The most common form of subsidence is probably that associated with the longwall mining method. This involves systematic extraction of a tabular mineral deposit using a working front which can be commonly around 200-300 m length but can be as little as 30 m or as great as 500 m. It is usual to allow the immediate rocks overlying the mineral deposit to cave and collapse into the void left behind the longwall extraction. The caving process allows the upper beds to subside and cause such movements to propagate to the surface to result in formation of a subsidence trough which is generally more widespread than the underlying extraction. Figure 4 shows a representation of the general characteristics of subsidence development associated with a longwall extraction. This form of subsidence is predictable in terms of magnitude, extent and duration. Room and pillar mining is a common form of mineral extraction which is applied to tabular deposits especially mineral seams. The pillars left to form support to the surface may be withdrawn as part of the mining process but this is generally rare since the operation is specialized and hazardous. Surface subsidence above room and pillar extraction areas can take the form of a gradual lowering of the overburden, which can arise should the pillars fail, either completely or partially, or that the pillars give rise to failure of the floor or roof beds in immediate contact with the pillars. A more common feature of subsidence encountered with room and pillar mining is that of sink hole formation. Collapsed intersections of mining rooms can result in the collapse process progressing upwards until reaching the surface and in which case a hole can suddenly occur or an appreciable area collapse. Figure 5 illustrates general features associated with room and pillar subsidence. Depth below surface is an important factor as is that of the extracted seam height and the dimensions of the rooms and pillars. The rock type overlying room and pillar extractions is also an important factor as weak cover rocks readily allow formation of a collapse chimney which propagates to the surface and results in formation of a sink hole. Examples of sink hole characteristics are illustrated in Figure 6. A common form of surface depression above abandoned room and pillar workings is shown in Figure 6(a), where the surface has suddenly lowered by about 1.5 m. Such depressions can gradually increase in size due to Tension
Figure 4
Compression
Tension
Longwall mining subsidence features; the principal terms and general characteristics of trough subsidence for a longwall cross-section
Subsidence Behavior of Rock Structures
757
(a)
Stable
Partial collapse
Crownhole
Depression
(b)
(C)
Figure 5 Illustrating general subsidence features associated with room and pillar mining operations: (a) localized failure of mining room intersections can allow collapse chimneys to propagate through shallow depths of cover to give rise to sink hole development, which can have clearly defined and steep sides or adopt the form of a simple conical or saucer-shaped depression, (b) showing failure of pillars as a cause of surface subsidence, and (c) where strong support pillars punch into the immediate roof and floor consisting of weaker rock types, then failed rock which displaces into the mining rooms and intersections can result in general subsidence of the surface; under such circumstances, rock over the pillars is sheared and encourages general lowering of the immediate cover rocks
subsequent movement induced by further collapse or inflow of the column of broken debris into the mine. Figure 6(b) shows a sink hole which has been infilled from the surface and has subsequently experienced further activity. The infilled chimney has tended to flow into the mine in this particular case. Removal of fill material by the action of subterranean water is a common cause of such sink holes continuing to exhibit subsidence. Sink holes can display steep, and even overhanging, sides when they first appear at the surface. Figures 6(c) and (d) show examples of sink hole developments over room and pillar mine workings. The steepness of the sides generally decreases due to weathering effects. These subsidence holes can represent a significant hazard both to the surface and the mine. Sink hole occurrences can cause surface waters to flow into an operating mine. Conversely, sink holes can cause surface structures to suddenly collapse into appreciably deep holes. Knowledge of the risk of this form of surface collapse occurrence is vital in those areas affected by mining subsidence.
758
Back Analysis Monitoring
Figure 6a, b
28.3 PRINCIPAL BEHAVIORAL CHARACTERISTICS OF MINING SUBSIDENCE 28.3.1 Observed Behavior of Subsidence: UK Principal sources of data on the observed behavior of mining subsidence exist in various countries. The UK Coal Industry's Subsidence Engineers' Handbook [1] is probably the best known publication. This presents observed data on longwall mining subsidence relating to coal mining operations in the UK, and gives comprehensive data, graphs and prediction charts for determining surface subsidence and ground strain profiles. The effects of subsidence on various types of surface structures are discussed and general guidelines for planning future developments are also presented. The prediction methods given in this publication are empirical, being based entirely on UK subsidence data. However, these charts have been used by various investigators in other countries for predicting subsidence, and a measure of compatibility has been demonstrated in respect of the profile for coal mining conditions although the maximum subsidence value does vary owing to the geological character of the overburden. 28.3.2 Observed Behavior of Subsidence: Germany Kratzsch [2] has published a comprehensive discussion of mining subsidence engineering principles based mainly on observations made in coalfields in Germany. This publication presents a
Subsidence Behavior of Rock Structures
759
Figure 6 Examples of sink hole features over iron ore room and pillar workings: (a) recent sink hole occurrence ( < 1 week old); the thickness of the soils above the bedrock has influenced its general character, (b) continued subsidence of previously filled sink hole, (c) recently formed large sink hole (20 m maximum diameter), and (d) sink hole development showing collapse chimney emerging from underground workings
detailed treatment of prediction methods especially as related to longwall coal mining. The response of surface structures to mining subsidence is given special attention, and forms an important reference source. This book gives an excellent introduction to the prediction of mining subsidence for longwall operations, especially in conditions similar to those encountered in German coalfields.
760
Back Analysis Monitoring
28.3.3 Observed Behavior of Subsidence: Europe and other Countries The publication of Whittaker and Reddish [3] on subsidence gives an overview of the response of various geological settings to the subsidence phenomenon. This includes those effects whether caused naturally or resulting from man-made extraction processes such as underground mining, water and oil extraction, and various other forms of underground activity. Discussion is focused on how various types of underground opening collapse can result in surface subsidence. Prediction methods are discussed, and several case histories are given in order to demonstrate particular influences of geological, structural and mining factors. Other notable publications exist which deal with particular aspects of subsidence. Peng has organized several successful conferences on ground control in mining which have dealt mainly with subsidence aspects relating to North American coalfields [4-6]. Attention has been focused on correlating observed subsidence with various methods of predicting ground movements. Substantial data have been presented on the effects of abandoned mines and the associated patterns of the resulting surface subsidence. Brauner [7, 8] has presented comprehensive reports on subsidence prediction methods. His work gives useful comparisons of European case histories. The reports also give critical comments on the interrelationships of observed data with predicted values. Subsidence aspects with special reference to Australian experiences have been discussed and presented in a concise volume which formed the proceedings of an important conference on wide ranging aspects of subsidence and its control [9]. The proceedings devote particular attention to subsidence prediction and its comparison with case histories. Several excellent papers have been published and exist in a variety of journals and conference proceedings. Some of these papers give detailed accounts of mining subsidence observations and examine their treatment and interpretation in relation to achieving improved prediction of this phenomenon. The main factors governing subsidence have been thoroughly researched and reported upon. A number of papers give accounts of observed behavior of ground movement as observed in particular countries, for example, in the USA [10, 11], former USSR [12, 13], France [14], Canada [15], South Africa [16-18] and Japan [19]. An overview of observed subsidence behavior in the main countries which experience mining subsidence problems has been presented by Whittaker and Reddish [3]. The behavior of surface structures when affected by ground movements has occupied the attention of many investigators. Important conferences have devoted most of their attention to ground movement and structures whereby several case histories have been discussed and comparisons made with predicted behavior [20-22]. Several of the papers presented in the proceedings discuss observed data and interpretations based on detailed analyses of related measurements. 28.4 BASIS FOR PREDICTION OF LONGWALL SUBSIDENCE 28.4.1 Development of Empirical Prediction Method The UK approach to subsidence prediction has been based on reliable data observed at several field sites which are considered representative of the situations likely to be encountered by coal mining operations in different UK coalfields. It was recognized during the late 1940s and early 1950s in the UK that a sound basis for subsidence prediction was required and, consequently, investigations were conducted throughout the UK. The collation of a substantial body of subsidence data was judged to be the best approach at this time. Extensive research programs were launched in order to examine the interrelationship of the various factors governing the occurrence, characteristics and overall influence of subsidence due to mining operations. The UK empirical subsidence model is based mainly on the early investigations of Orchard [23, 24] and Wardell [25, 26] and their coworkers. Their main findings and the results of many other investigations and measurements relating to longwall mining subsidence have been embodied in the UK Coal Industry's Subsidence Engineers' Handbook [1]. The investigations were principally an extensive series of field measurements of subsidence and lateral displacements from which ground strains were deduced. As a consequence of the subsidence measurements, ground tilt changes were also deduced. Lines of survey stations were formed from which successive levels and linear measurements were made. Generally, the aim was to establish two main survey line directions, one preferably along the center line of the longwall and one or more across the entire extraction width. In practice, however, the subsidence measurement lines needed to be positioned to suit surface constraints which frequently resulted in their positioning at oblique lines across the extraction area.
Subsidence Behavior of Rock Structures
761
However, substantial results were obtained from investigation sites at the surface overlying many longwall extractions, and these results allowed rationalization in respect of subsidence behavior resulting from longwall mining. The subsidence measurements and deduced lateral strains were rationalized. An example of the form of representation adopted for the subsidence results is shown in Figure 7. This shows subsidence expressed in terms of extracted seam height (S/M) and the results have been plotted against the ratio of extraction width over depth below surface (w/h). An important feature demonstrated by these results is that for w/h ratios greater than around 1.4, the S/M value tends towards a constant of 0.8-0.9 for the data presented. These were early results established by Orchard and whilst the trend has been demonstrated to be consistent, later results have indicated that the maximum value of S/M can attain 0.9 in most supercritical cases. This relationship between subsidence and the w/h ratio has formed the basis for predicting subsidence in the UK coalfields since the mid-1950s. The relationship readily demonstrates that in order to decrease significantly the magnitude of surface subsidence, then reducing the w/h ratio represents a highly effective control parameter. This method of limiting subsidence to prescribed values has been the main method of achieving effective subsidence control in UK coalfields rather than that of adopting such control measures as stowing of the voids created by the longwall extraction. The measured lateral displacements across a subsidence trough have been used to deduce the associated strains due to formation of the depression over the longwall. The strain values have been plotted in the form illustrated by Figure 8. Positive strain values represent tension whilst negative strains indicate compression.
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 W/h
Figure 7 Early observations of longwall subsidence expressed in terms of extracted seam height and plotted against the longwall width to depth ratio (w/h) (after Forrester and Whittaker [28], based on Orchard's [23] early observations)
/w^
£*= 3.25 mm m '
-E - 2.85 mm m~ 5 =1 m M- 1.34 m
600 m
/tf=l.34 m
Figure 8 Representation of subsidence and strain results for a longwall cross-section; illustrating the influence of depth on surface strain (after Forrester and Whittaker [28], based on Orchard's [24] early observations)
762
Back Analysis Monitoring
The magnitudes of the strain values have been shown to be a function of k(S/h), whereby the constant k is governed by the w/h ratio. Figure 8 also demonstrates the influence of depth. In these two particular situations which correspond to the same ratios of w/h and the same extracted seam height, then the resulting maximum subsidence is the same. However, owing to the influence of depth, the strain magnitudes are markedly different with the shallower situation giving rise to maximum tensile strain values around three times those of the deeper situation used in this comparison. The shallow depth situation shown in Figure 8 demonstrates greater concentration of the subsidence strains, although they are spread over an appreciably decreased surface area. 28.4.2 Influence of Goaf Treatment on Maximum Subsidence Figure 9 shows maximum subsidence expressed in terms of extracted seam height and plotted against the longwall w/h ratio for three conditions of goaf treatment. Strip packing is referred to on this illustration and this corresponds to the provision of localized support in the form of loose pack structures which run parallel to the face center line; their main purpose was to steady the caving effect of the immediate roof behind the longwall and thus reduce the risk of the relatively lightweight supports, used at that time, being disturbed by the caving action behind the longwall. Strip packs are rarely used today as they do not conveniently lend themselves to application with modern longwall supports. Additionally, modern longwall supports effectively control the roof thus eliminating the need for strip packs. The materials used for the strip packs were usually obtained from caved material within the goaf. As a consequence, the provision of strip packs, although providing some form of improved local support at the face, did not make any significant difference to the resulting subsidence observed at the surface above the longwall. For strip packing and caving, the maximum subsidence can achieve 90% of the extracted seam height for supercritical widths of extraction involving w/h ratios of 1.4 or greater. The influence of stowing can achieve a reduction in the maximum value of subsidence experienced at the surface. This is shown to be around 45% for pneumatic stowing and 15-20% for hydraulic stowing expressed in terms of the resulting maximum subsidence experienced above the center of the longwall extraction. These subsidence values are based on the assumption that the longwall has advanced sufficient distance as to allow full development of the subsidence profile. Experience with stowing behind longwall faces has indicated that the w/h ratio should generally be greater than 0.6 in order for stowing to achieve a significant reduction in subsidence effects at the surface.
o 10
-
\\
20
\\ \ \ \ \\
-—
30 ~
40 —
^
50
X
\
* ^ s
—
CO
\o
\o
60
70
\%
— -
\1
-
V%
^31
3
80 ^M
90 100 0
Figure 9
1 0.2
1 0.4
1 0.6
1 0.8
w/h
1 1.0
1 1.2
J 1.4
1 1.6
Relationship between width/depth ratio and subsidence for various goaf treatments; these results are based on European coalfields observations (after Forrester and Whittaker [28])
763
Subsidence Behavior of Rock Structures 28.4.3 Subsidence Development Curve
The fact that subsidence of the surface extends outside of the plan area of the longwall extraction means that subsidence develops ahead of an advancing longwall face. Figure 10 demonstrates remarkable consistency for the several subsidence development curves even though different types of cover rocks existed with these case histories. The graphs show that about 15% of the total subsidence has occurred by the time that the longwall face is directly underneath the reference point (P) at the surface. Conversely, subsidence is almost complete at a distance behind the longwall equal to about the depth below the surface. Figure 10 implies that all the subsidence is likely to be complete at a distance equal to the depth behind the longwall. This is not necessarily the case in all situations as a measure of time dependence can be observed in particular physical conditions. Time-dependent behavior is more pronounced in the cover rocks immediately above the rib edge of the longwall extraction whilst time-dependent effects are not so well pronounced at the extremities of the subsidence depression. Residual subsidence is observed to occur in some circumstances but is generally complete within 3-6 months of cessation of longwall working. The amount of residual subsidence is usually less than 10% of the total at the place where time-dependent subsidence occurs, i.e. generally over rib edges. Certain types of surface geology are more prone to exhibiting residual subsidence effects [27]. 28.4.4 Extent of Influence of Subsidence Figure 8 shows the subsidence limit angle to be 35° (measured to the vertical) and this represents the limit of discernible movement at the edge of the subsidence zone. This angle is referred to as the angle of draw or subsidence limit angle. The average value for the angle of draw as observed in European coalfields is generally quoted as 35° although Beevers and Wardell [26] have reported values which lie within the limits of 33-38°. Values of the angle of draw as reported in earlier publications tended to quote appreciably smaller values owing to less sensitive instruments and less suitable measuring techniques being used to observe subsidence at that time. Surveying instruments have been developed which allow subsidence limits to be accurately established, especially since the 1950s. The angle of draw observed in UK coalfields has been generally consistent at around 35° for Coal Measures sedimentary rocks. The same value for the angle of draw has also been observed where the cover rocks have consisted of Permian limestones, sandstones and marls, and also where Triassic sandstones have comprised of cover rocks. In other countries, however, where markedly stronger rock types have been encountered to form the cover rocks, then the angle of draw has been generally less than 35°.
P
A B
h M SP Sp/AfTr Pe (m) (m) (m) (%)(m) (m) 275 1.30 0.90 69 0 0 660 1.83 1.27 69 0 45
CM(m) 275 615
Distance x in terms of depth h
Figure 10 Development of subsidence in relation to face position (after Forrester and Whittaker [28], based on revision of Wardell's original results [25])
764
Back Analysis Monitoring
28.4.5 Ground Tilt The generation of a subsidence trough over a longwall extraction gives rise to surface tilt which can have an appreciable influence on certain types of surface structures, particularly main sewers, tall buildings, etc. The maximum tilt has been shown to be a function of k(S/h). This is similar to the relationship to ground strain except that different values are used in respect of the constant k. The maximum tilt occurs at the transition point, namely where the tensile strain zone meets the compressive strain zone and its maximum value (G) for a critical width of extraction is G = 2.15S/h. 28.5 LONGWALL MINING SUBSIDENCE PREDICTION: THE UK EMPIRICAL MODEL The UK coal industry has developed an empirical subsidence prediction model based on field measurements. The model is described in the Subsidence Engineers' Handbook [1]. It is intended to indicate the scope of this particular empirical model and to demonstrate its advantages and attractiveness to engineers and surveyors involved with subsidence prediction work. Figure 11 illustrates three particular features of the UK empirical model. The basic approach is to firstly establish subsidence profiles similar to the example shown. The charts given in the subsidence Subsidence Profile
Strain Profile
Displacement Profile
Figure 11 (a) Subsidence, (b) strain and (c) displacement profiles for a transverse section across a longwall extraction, based on European coal mining observations and related predictions
765
Subsidence Behavior of Rock Structures
handbook enable strain profiles to be established again as illustrated in Figure 11. Both compressive and tensile strain areas are identified by the empirical model. A further important aspect is that of the nature of the lateral displacements which occur across the subsidence trough. A displacement profile for a longwall extraction transverse section is also illustrated in Figure 11. 28.5.1 Subsidence Prediction Figure 12 shows the basic subsidence prediction chart for a caved condition. The axes of the chart are those of longwall face width (w) and depth below surface (h) respectively. Determination of the corresponding subsidence for a given combination of w and h simply involves reading off the corresponding value of S/M at the intersection point of lines drawn from the axes. The accuracy of predicting subsidence appears to be better satisfied using this graphical approach than that of plotting S/M against w/h as used in the first edition of the Subsidence Engineers' Handbook. The graph does not consider data for depths of less than 50 m mainly since there are insufficient case histories in this region on the graph to allow accurate prediction. Figure 12 allows the maximum value of subsidence to be determined with a fair degree of accuracy (better than 10% of the maximum subsidence). In order to allow the subsidence profile to be drawn across a longwall extraction, then it is necessary to resort to using Figure 13. The w/h ratio is established and a horizontal line drawn to intersect the subsidence contours and at which point the position of each contour is established by observing its intersection by simply reading off its distance from the panel center which is expressed in terms of depth below surface, i.e. d/h. Having established the data points from either full subsidence or 0.95 of the full subsidence (S) to the limit of subsidence then a half profile can be drawn and its mirror image correspondingly produced where symmetry exists. 28.5.2 Prediction of Strain due to Subsidence Figure 14 shows a strain prediction chart with axes of w/h and d/h respectively. This chart allows both the maximum tensile and maximum compressive strain values to be indicated in terms of their position relative to the center of the panel. Proportions of the maximum tensile and compressive strain values are also indicated as contours in order to allow sufficient points to be established for the purpose of drawing the strain profile to sufficient accuracy. It is necessary, however, that the maximum tensile strain ( + £) and the maximum compressive strain ( - £ ) be established and this is achieved by using Figure 15. This graph shows a plot of the
0
200
400
600
800
Depth, Λ(ΓΤΙ)
Figure 12 Relationship between longwall face width (w) and depth below surface (h) showing various subsidence (S/M) contours for prediction of maximum subsidence (after Subsidence Engineers' Handbook [1])
766
Back Analysis Monitoring
o o o o o o o o o SO* i 0.4
o o &
f
_!_
0.8
_L 1.2
_L 1.6
2.0
Distance from center of panel in terms of depth, d/h
Figure 13 Relationship between longwall face width and the magnitude of the subsidence profile (after Subsidence Engineers' Handbook [1])
04
0.8
1.2
Distance from center of panel in terms of depth,
Figure 14
1.6
2.0
d/h
Relationship between longwall face width and the magnitude of the ground strain profile (after Subsidence Engineers' Handbook [1])
multiplication factor k(S/h) against w/h for the longwall extraction. Consequently, + E and - E can be readily determined by simply reading off the corresponding multiplication factor from knowledge of the w/h ratio. Figure 14 readily allows strain profiles to be determined across a longwall extraction. Multiple profiles can be determined in order to allow a plan view to be established of the nature of the strains associated with a particular extraction. It should be noted that although Figure 15 only shows two graphs for the maximum values of the compressive and tensile strains, a third graph is usually included which allows the maximum tilt (G) to be read off the graph from knowledge of w/h. The Subsidence Engineers9 Handbook gives details of the determination of tilt in respect of the subsidence profile.
Subsidence Behavior of Rock Structures
767
Width/depth, w/h Figure 15 Multiplication factor k(S/h) for determination of maximum values of strain in relation to the width/depth ratio (after Subsidence Engineers' Handbook [1])
28.6 GEOLOGICAL FACTORS INFLUENCING LONGWALL MINING SUBSIDENCE 28.6.1 Site Conditions and Effect of Local Geology Site condition aspects such as surface soils in respect of thickness and type can appreciably influence the response of the surface to mining subsidence. Structural discontinuities, hydrogeology and the effects of weathering can play important roles. These various aspects have been commented upon by Forrester and Whittaker [28, 29]. In UK coal mining conditions, the nature of the site can result in a degree of abnormality being encountered in the prediction of subsidence by comparison to the observed profile. This deviation from the predicted behavior using standard procedures can generally be explained by careful consideration of the conditions affecting the site. Groundwater plays an important role in most conditions. 28.6.2 Bedrock Conditions and Subsidence The nature of the bedrock plays an important role especially where major fissures exist as these can encourage the concentration of tensile strain and consequently gives rise to opening up of such fissures. An example is shown in Figure 16 of afissurein Bunter sandstone being opened by mining subsidence and this has resulted in increased localized erosion of the surface soils. The thickness of the surface soils and the degree of opening of the fissure in the bedrock influence the scale of this problem. Figure 17 illustrates diagrammatically the sequence of events from the opening of a bedrock fissure to the occurrence of major trenching caused by natural erosion into the fissure opened by subsidence effects. The illustration shows the localized removal of surface soils into the fissure and this process is aggravated by the presence of water which encourages increased erosion particularly in allowing such material to be washed deeper into the fissure. The occurrence of such subsidence induced effects are dependent mainly on the nature of the bedrock and consequently knowledge of the existence of major fissures is important in respect of predicting the likelihood of disturbance of the surface by this phenomenon. The problem can affect agriculture especially in respect of causing localized drainage pattern changes. Major fissures in fields have occurred some appreciable time after mining has ceased and this has been explained to be due mainly to the time lag involved in localized erosion manifesting itself at the surface. Where such fissures are opened up appreciably then their filling with appropriate materials needs to be carefully considered, since there will be continual drainage through the fissure and the likelihood of further erosion taking place with the recurrence of more surface holes. Where bedrockfissureshave occurred under highways, then theirfillingwith suitable engineering materials needs to take account of allowing drainage to occur but without loss of stability of either the bedrock or the fill material. In UK coalfields, Assuring arising from bedrock conditions mainly occurs with sandstones and limestones. The line of the Assuring can often be predicted following careful site investigation. In addition, the occurrence of Assuring is generally confined to the region over the ribside, i.e. in the tensile strain zone. Case histories have been reported which relate the occurrence of major Assuring to the ribside and also the magnitude of the maximum tensile strain [3].
768
Figure 16
Back Analysis Monitoring
Exposed joint in bedrock opened by mining subsidence; the top soils have been washed into the opened joint and, as a consequence, given rise to localized subsidence by an erosion process
(α)
(b)
(c)
Figure 17 Diagrammatic representation of the subsidence process shown in Figure 16; the occurrence of subsidence is linked with the drainage of water through the bedrock
28.6.3 Hydrogeological Effects in Respect of Mining Groundwater plays an important role in respect of surface subsidence. Localized drainage giving rise to significant lowering of the water table can result in appreciable subsidence which can be caused by natural processes or mining induced effects. Groundwater levels can fluctuate appreciably from season to season or from year to year as a natural phenomenon. This can cause marked subsidence effects. Where mining subsidence gives rise to the opening or indeed closing of bedrock fissures then this can cause a change in the drainage gradients and patterns and groundwater flow generally. As discussed above, localized erosion features can occur at the surface above such bedrock fissures.
Subsidence Behavior of Rock Structures
769
Hydrogeological factors need to be taken into account where room and pillar mining exists since collapse chimneys from mining rooms can readily give rise to a significant drainage hole which can appreciably affect surface drainage. This is especially important if the subsidence hole occurs within proximity of a significant surface water course. 28.6.4 Influence of Cover Rock Type on Longwall Subsidence Attention has been drawn earlier in this chapter to the observed amount of subsidence attaining 90% of the extracted seam height in several European coalfields and especially in the UK. This amount of subsidence represents the maximum observed and it only occurs with supercritical extractions namely particularly shallow situations or with very wide extractions at moderate depths. Observations of mining subsidence reported in several countries have shown that the maximum subsidence above a longwall extraction can be appreciably influenced by the nature of the cover rocks. Figure 18 presents observed subsidence data from different countries [3]. An important feature of these subsidence results is the similarity of the subsidence profile irrespective of the country of origin or the cover rocks involved. What is of special significance, however, is the difference in magnitude of the maximum observed subsidence expressed in terms of the extracted seam height. These results have been collated by Shadbolt [30]. The nature of the cover rocks influences the character of their caving properties and also their ability to span with little or no support across significant extracted areas. Where the cover rocks comprise predominantly of sandstones, then the resulting maximum subsidence is observed to be much less than in those situations where the cover rocks consist virtually of weak mudstones. 28.6.5 Influence of Geological Faults on Surface Subsidence Where geological faults allow a major discontinuity to extend from the currently worked seam to the surface, then the surface subsidence profile is likely to experience an anomaly. Figure 19 shows how geological faulting can give rise to anomalous subsidence profiles occurring at the surface. Underhade workings are likely to be more affected by virtue of the strain relieving process encouraging increased movement along the fault plane with subsequent stepping at the surface. Overhade workings tend to discourage significant surface stepping by virtue of the fault orientation tending to discourage the type of movement encountered with underhade situations. Lee [31] has reported the findings of a study he carried out on the occurrence of stepping associated with subsidence profiles due to the presence of geological faults. His main findings
w/h Figure 18 Relationship between the magnitude of subsidence expressed in terms of extracted seam height and the longwall width/depth ratio for coal mining operations (after Whittaker and Reddish [3], based on Shadbolt [30])
770
Back Analysis Monitoring (a)
(c)
(b)
(d)
Figure 19 Illustrating the effect of anomalous steps occurring in longwall subsidence profiles due to the presence of major geological faults. Subsidence trough stepping due to (a) fault with underhade workings, (b) vertical fault with workings, (c) fault with overhade workings and (d) fault with underhade workings and fault pillar
indicated that the occurrence of a surface step in the subsidence profile could be related to the ribside edge, i.e. between extracted area and solid coal which clearly showed that the fault steps occurred within a region of 0.7(depth) on each side of the ribside. Most of the fault stepping occurred over the extracted area. Where faults extend from the mining horizon to the surface, they can exhibit marked movement at the surface well ahead of the advancing longwall. Faults tend to cause appreciable concentration of subsidence strain at the surface. They can also exhibit time-dependent behavior which tends to be aggravated by mining subsidence. Where geological faults intercept the surface, the thickness of surface soils becomes of special importance. Thick top soil covering can result in masking the presence of such faults but, at the same time, encourage more even spreading of the effect of the fault, i.e. it is less likely that a pronounced step will occur. The occurrence of geological faults between the mining horizon and the seabed is of special significance to undersea mining [32]. The planning of undersea longwall mining operations requires special consideration to be given to the thickness, nature and degree of disturbance of the cover rocks between the mine and the seabed. Geological faults require special consideration particularly in respect of avoiding their disturbance which could result in a change in their water transmitting potential.
28.6.6
Surface Topography and Influence of Seam Inclination
The surface topography can influence the nature of the resulting subsidence above longwall or, indeed, other forms of mining operations. Where steep hillsides exist at the surface overlying mining operations, significant Assuring can occur as shown in Figure 20 which illustrates Assuring above mining extraction of a 40 m thick seam pitching at around 75°. Substantial Assuring of this nature can introduce slope stability problems especially where the fissures connect up with surface water courses. These fissures can allow water to gain access to rocks at some depth below the surface and result in decreasing the natural slope stability. Tensile strain concentrations play a major role in this Assuring phenomenon. These zones of tensile strain can be predicted using specialized methods [2, 3]. Steeply pitching seams which are extracted and the overburden allowed to cave generally result in a concentration of the subsidence effects at the surface [2, 3]. An example of the surface subsidence profile above a steeply pitching seam (75°) and of 40 m thickness is illustrated in Figure 21. The dip side has given rise to a step characteristic as seen on the left hand side of this photograph whereas distinct Assuring can be seen on the rise side. Specialized computer-based prediction methods need to be employed in order to determine the nature of the subsidence, displacement and strain proAles over such steeply pitching seams which are extracted by longwall operation and subsequently caved [3].
Subsidence Behavior of Rock Structures
Figure 20
28.7
771
Subsidence fissures above coal mining extractions (Northern Spain): (a) occurrence of wide fissure with a differential subsidence step, and (b) illustrating occurrence of deep tensile fissures
PREDICTION OF MINING SUBSIDENCE: GENERAL PERSPECTIVE OF THE MAIN METHODS
Mining subsidence can be predicted using various methods. There are many different methods ranging from site specific to general applications and they depend upon the geological conditions and type of mining involved. Some of the prediction methods have been designed for routine use, whereas others have been established as research techniques. Profile functions and influence functions are the most commonly
Back Analysis Monitoring
772
Figure 21
Surface subsidence trough above steeply inclined thick seam extraction
used prediction techniques. Within these broad categories, however, there is a wide range of choice and application to particular situations. Subsidence prediction methods can be placed in the following broad categories. 28.7.1 Profile Functions These methods are based on predicting the magnitude and shape of the subsidence trough from mining dimensions, using some form of equation or graph. The equations may be simple or complex but are generally based on results obtained from case studies. The use of graphs instead of equations allows more parameters to be taken into account and most of the more commonly used methods are of this form. UK experience obtained from around 200 or more case histories has been embodied in the Subsidence Engineers' Handbook [1] and probably represents the most widely used profile function application to mining subsidence prediction work. This publication presents its profile function prediction method in the form of comprehensive prediction charts which allow scope for modification to a wide range of coal mining subsidence situations. 28.7.2 Influence Functions Influence functions are based upon a surface point having a circular area of influence underneath it. This area is defined by the angle of influence and the depth of the seam. Any mineral extracted within this area of influence will cause a corresponding lowering of that surface point. Equally sized extractions directly under this point and at the extremities of the area will result in different amounts of subsidence. The resulting difference is defined by the influence function which is expressed as an equation relating the influence to radial position. This approach lends itself to application by computer program and a number of sophisticated packages have been developed based on the influence function approach. This method has a major advantage over others in that it possesses the ability to predict subsidence, displacements and strains over various irregularly shaped extractions whether inclined or otherwise [3]. 28.7.3 Numerical Models Surface subsidence has been predicted using a number of theoretical models based predominantly on simplified elastic assumptions. Many of these methods appear to make assumptions that are invalid in the context of weak Coal Measures rocks and, consequently, give results which oversimplify subsidence prediction. With the development of nonlinear programing techniques in
Subsidence Behavior of Rock Structures
773
the context of elastic finite element modeling, the ability to model weak rock behavior has been considerably enhanced. Several research institutions have developed programs with appreciable ability to cope with the significant size of typical longwall subsidence problems. This approach is limited, however, in the context of surface subsidence prediction owing to difficulties in the yielding of sufficient detail and, consequently, it is more often used for subsurface prediction of rock movements between the mining horizon and the surface.
28.7.4
Physical Models
The use of small scale physical models in subsidence prediction studies has several advantages. The rock cover details can be incorporated in the model in addition to various types of surface topography. Seam inclination can be taken into account in the physical modeling process. The physical model can then be extracted and the resulting movements between the extraction horizon and the surface observed. Detailed measurements can be made to establish the nature of the subsurface strains whilst photographs are used to identify the nature and distribution of cracks generating from the mining horizon and propagating towards the surface. Such models readily allow identification of potential flow paths for water from aquifers between the mining horizon and the surface and, consequently, the modeling process can be repeated with modifications to the mining technique in order to observe their influence on changing the pattern and magnitude of the cracks generated due to the mining process. Physical models, which are designed to collapse under their own weight, are generally constructed from sand and plaster mixed with water and cast in a purposely designed frame which embodies the mining extraction in the form of blocks which are removed successively after the physical model has been constructed and allowed sufficient time to cure and achieve its desired strength and deformation properties. The physical modeling process, using sand and plaster materials, has been demonstrated to produce consistent subsidence results which correlate closely with the UK empirical subsidence model [3].
28.8
PREDICTION OF SINK HOLE DEVELOPMENT ASSOCIATED WITH ROOM AND PILLAR MINING
Discussion has focused earlier in this chapter on the problems associated with surface subsidence above room and pillar mined out areas. The occurrence of sink holes, due to mining subsidence, has been extensively researched [3]. Comments are given here on the basic principles which are commonly used to predict their possible extent in relation to the main mining and geological factors.
28.8.1
Basic Considerations
Figure 22 illustrates the basic considerations used in the development of caving above a four-way junction. As indicated by this illustration, the collapse chimney is shown as circular which is consistent with underground observations involving hundreds of such collapses. The question arises as to the assumed diameter of the chimney; two sizes are shown in Figure 22, but the chimney diameter does appear to vary. These two assumed dimensions appear to represent the limits in situations observed by the authors where such collapses have propagated to the surface and resulted in a sink hole. Figure 22 shows the development of caving above the collapsed junction. The collapsed material spills out into the adjoining rooms until reaching the roof level. At this point the material can no longer spill out into the rooms. Subsequently caved rock then fills the remainder of the chimney during a self-choking process. This theoretical explanation of the caving process above a junction in a room and pillar layout appears to be consistent with observations in practice. 28.8.2
Prediction of Caving Height
With reference to Figure 22, the volume of caved material can be represented by equation (1) Kcaved = kxnd2/4
(1)
774
Back Analysis Monitoring Basic symbols relating to mine junction
-J
S
f' ^
^^ -
T A ■MO'
Chimney diameter assumptions (a)
d-w
(b)
d^w/2
Development of caving above junction
I. Initial collapse condition of mine junction
2. Progressive development of roof collapse with caved rock spilling into adjoining rooms to roof level.
3. Caved rock fills remainder of chimney during self choking process.
Figure 22 Basic considerations in development of caving and formation of a collapse chimney above four-way junction in a room and pillar mine (after Whittaker [38])
Equation (2) assumes the condition that the space available to receive the caved material corresponds to that shown in Figure 22 = 4(0.5wH2cot) + Hw2 + xnd2/4
(2)
Equating (1) and (2) allows the collapse height (x) to be deduced as given by equation (3) x = 4(2wtf2cottf> + Hw2)/l(k - \)nd2~]
(3)
where Fcaved = volume of caved rock from collapse chimney, Kspace = volume of available space to receive caved roof rock, k = bulking factor considered to be in the range 1.33 to 1.5, x = height of the collapse chimney, d = diameter of the collapse chimney, w = width of mine rooms, H = excavated height of mine rooms, φ — angle of repose of caved rock within mine rooms adjoining the collapsed area. Various assumptions are made regarding equations (1) and (2). Firstly, the rock density in its solid state is assumed to remain uniform. Secondly, the bulking characteristics are also assumed to be consistent and remain unaffected by loading created during the extension in height of the collapse chimney. Thirdly, it is assumed that the increased loading on the broken material at the base of the chimney will produce insignificant change in volume during the process of caving up to the full height of the chimney.
Subsidence Behavior of Rock Structures
775
28.8.3 Caving Height Above Collapsed Junction Figure 23 has been established for a practical range of mining room dimensions in respect of equation (3). Room widths have been considered to be in the range 4-8 m which corresponds to UK iron ore room and pillar mine workings. The maximum height (x) of the collapse chimney is indicated by Figure 23 to be in the range 3-9 H (extraction height). The extraction height assumed in this example has been 3 m. Figure 23 implies that narrow rooms could result in an increased height of caving, but due consideration needs to be given to the fact that narrower rooms are less likely to collapse. The most common room width for coal and stratified iron ore mine workings is that of 6 m, and Figure 23 indicates a height of 3.5-7 H. Figure 23 indicates that, when assessing the caving height, due account needs to be taken of the rooms, in addition to the geological aspects and bulking characteristics of the immediate roof strata. From the information presented in Figure 23, it would appear that a caving height of 4-10 H could occur although there is increased likelihood of a height of up to 7 H being more probable. It should be noted, however, that this assessment does not take into account the possibility of water gaining access to the chimney and causing the caved rock to flow into the mine. 28.8.4 Effect of Water Gaining Access to Collapse Chimney A common occurrence is that of water gaining access to a collapse chimney which is developing towards the surface. This can occur by virtue of the chimney intercepting some water course or aquifer in the process of its upward collapse. The chimney essentially represents a significant drainage feature which can readily draw water to it on intercepting an aquifer. The fact that the chimney is partially filled with caved material can result in the water draining through the broken material and gaining access to the mine. This process alone can encourage the caved material to become unstable and assist its movement into the mine by adopting a much shallower angle of repose. Figure 24 represents an observed phenomenon in British iron ore mines where the immediate stratum overlying the iron ore bed consists of clays which readily break down in the presence of water. Consequently, should a chimney tap an overlying aquifer which allows water to drain into the cavity, this can cause the caved clay material to further break down and tend to restrict the drainage of water through it. Water can become impounded by the clay within the collapse chimney. During wet periods, the build up of water in the collapse chimney can be significant. The plug of caved material in the chimney can suddenly discharge into the mine as an inflow, or indeed an inrush, of wet rocks and mud. This has been observed by the authors to occur both suddenly and, in some circumstances, as a slow creep process. I7.5r
d = l¥y/2
2
4
6
8
10
12
Room width, w (m)
Figure 23
Relationship between caving height and room width for a four-way junction in a room and pillar mine (after Whittaker [38])
776
Back Analysis Monitoring
Sinkhole development stages 1. Early stage in development. 2. Chimney taps aquifer horizon and allows water to drain into cavity. 3. Build-up of water pressure by ponding can promote instability of plug of caved rock and result in flowing into mine 4. Caved mass flows info adjoining mine rooms. 5. Chimney continues caving due to increased available space for broken rock. 6. Sinkhole can emerge at surface from considerable depths (greater than \OH ).
Figure 24 Influence of collapse chimney intercepting an overlying aquifer and changing the potential caving height above a collapsed mine junction (after Whittaker [38])
Should a chimney of caved materials be discharged into adjoining mine rooms at its base, then, this can create increased space within the chimney for further development towards the surface. The authors have observed situations in these rock conditions for a sink hole to emerge at the surface from depths greater than 10 H. Clearly, local hydrogeological conditions can play a significant role. Sink holes have been observed to appear at the surface after 3-10 years since the original junctions collapsed in an operating mine. However, there was a direct association with water introduced by tapping of an overlying aquifer which prevented natural choking by bulking of the caved roof rocks. In respect of operational mines where collapsed junctions have been judged to pose a significant risk of a sink hole appearing at the surface in the event of it intercepting an overlying aquifer, then this has called for immediate action. Effective dams have usually been required to be constructed on all sides of the collapsed area. This has served to prevent possible discharge of wet materials into the mine as a result of the collapse chimney progressing to the surface. Where the construction of dam walls around a collapsed area has been impracticable and the possibility of a sink hole reaching the surface and affecting some important surface structure has been a significant risk, then such action as drilling from the surface and filling with a suitable material has needed to be resorted to in some situations. 28.9 SUBSIDENCE ASPECTS IN RELATION TO ABANDONED MINES 28.9.1 Nature of Problems Associated with Abandoned Mines Abandoned mines which give rise to subsidence problems some significant period after cessation of underground working, are those which clearly possess the potential for some form of subsidence to occur at a future date. Room and pillar mining operations are designed so that the cover rocks are generally supported by the pillars (assuming that they are not extracted) and consequently, the rooms are left in the same condition as they were during mining. The voids left by forming the mining rooms represent potential subsidence risks in respect of the surface in the future. This does depend, however, on the mining dimensions and depth below surface. Many other forms of mining operations such as open stopes where large voids are left can form a significant risk in respect of future subsidence. Some such mines have collapsed after periods of 50-100 years or more have elapsed.
Subsidence Behavior of Rock Structures
777
Abandoned mine shafts which were not filled but were simply capped have collapsed many years later. Where no records exist of such abandoned shafts, then this can represent a significant subsidence hazard.
28.9.2 Examples of Collapses Associated with Abandoned Metal Mines Vein deposits associated with some forms of metal mines frequently pitch steeply and outcrop at the surface. Figure 25(a) illustrates a lead zinc quartzitic vein which has been worked and subsidence has occurred in the form of collapses into the mine or as a result of sink holes. The subsidence effects at the surface are confined to a relatively narrow area immediately above the bedrock associated with the vein workings. It is common to leave crown pillars at the surface where the mineral vein outcrops. The pillars usually form part of the mineral deposit. Subsequent loss of strength of the pillars can result in the rock material falling into the voids below the crown pillar. As a result, subsidence holes of the form illustrated by Figures 25(b) and (c) can occur. Clearly, such subsidence holes represent a significant hazard at the surface. The form of protection adopted in respect of this hazard has been that of fencing, although in other localities, filling of the voids is also adopted.
28.9.3 Sink Hole Occurrences above Coal Mine Workings: US The extensive use of room and pillar mining, for the extraction of coal seams in the US, has been accompanied by subsidence incidents. The forms of subsidence have been mainly that of the trough type and that of sink holes. The main problems associated with subsidence have mainly arisen with sink holes. A detailed study has been carried out by Bruhn and coworkers [33] and their principal findings are listed as follows. (i) Sink holes were observed to occur predominantly where the overburden thickness was less than 50 m. Of special significance was the fact that most of the sink holes developed at depths of less than 15 m. (ii) The majority of the sink holes had a mean diameter of less than 3 m. However, some had diameters of up to 10 m. In respect of the depth at the centre of the sink holes, this was observed to be generally not more than 6 m measured from surface level. (iii) Sink hole development was observed to occur mainly along the line of the seam outcrop on hillsides. More than 70% of the recorded sink holes occurred within 150 m of the outcrop. (iv) Whilst some sink holes developed within 10 years after mining, a few other sink holes were reported as occurring a hundred years or so since cessation of mining. However, some 60% of the sink holes appear to have occurred within 50 years after mining. (v) Comparing the occurrence of sink holes with subsidence troughs, the former type outnumbered the latter type by around 30 to 1. These authors commented that this was probably due to the majority of the case histories employed in the study as having overburden thicknesses of 25 m or less. (vi) In respect of subsidence troughs overlying abandoned room and pillar workings, these have generally exhibited a dish-shaped profile. Their diameters have generally been in excess of 10 m with depths of not more than 1 m measured at the center. Other investigations have reported on abandoned mine subsidence aspects in US coalfields [34-36]. Their findings have tended to agree with the observations made by Bruhn et al. [33].
28.9.4 Sink Hole Development: UK Piggott and Eynon [37] and Whittaker [38] have reported upon UK experiences in connection with sink hole developments. Their findings have tended to support the discussion above, particularly in respect of the basic principles influencing the occurrence of sink holes. It would appear that the bulking principle of caved rock has a valid application in respect of predicting the likely extent of collapse of underground room and pillar workings to form sink holes at the surface. These authors draw attention to the significant problems of sink hole occurrence and the need to take such subsidence hazard risks into account.
778
Back Analysis Monitoring
Figure 25 Surface subsidence features associated with abandoned mineral vein mines: (a) general appearance of subsidence holes running along the line of the outcrop of the mineral vein on the mountain side, (b) illustrating the nature of crown pillar failure at the surface due to instability of the pillar and its subsequent slumping into abandoned stope workings below, (c) illustrating the general nature of crown pillar erosion/deterioration which has led to subsidence of the surface
Subsidence Behavior of Rock Structures
779
28.10 CONCLUSION Subsidence occurs naturally on widely differing scales. It can affect areas involving hundreds of square kilometers or involve occurrence of localized holes due to the effect of drainage changes through rocks and soils. In respect of mining operations, subsidence can be localized or exceed the plan dimensions of the mined-out area. Subsidence in connection with mining has been carefully monitored and studied in several countries. A vast wealth of information exists on subsidence behavior and engineering principles have been established relating to the prediction of subsidence, displacements and surface ground strains. Geological and hydrogeological factors play a major role in connection with the nature and magnitude of subsidence occurrences whether due to natural causes or induced by mining. Consequently, when conducting an assessment of the likely occurrence or predicting the magnitude of subsidence using engineering principles and established subsidence relationships, geology and hydrogeology require to be taken into account and given appropriate assessment.
28.11 REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32.
Subsidence Engineers' Handbook, p. 111. National Coal Board, London (1965, revised 1975). Kratzsch H. Mining Subsidence Engineering, p. 543. Springer Verlag, Berlin (1983). Whittaker B. N. and Reddish D. J. Subsidence: Occurrence, Prediction and Control, p. 528. Elsevier, Amsterdam (1989). Peng S. S. (Ed.) 1st Annual Conf. Ground Control in Mining, Morgantown, WV (1981). Peng S. S. (Ed.) 6th Int. Conf. Ground Control in Mining, Morgantown, WV (1987). Peng S. S. and Hanthill M. (Ed.) Workshop on Surface Subsidence due to Underground Mining, Morgantown, WV (1982). Brauner G. Calculation of ground movement in European Coalfields. In Proc. Symp. Subsidence in Mines, Wollongong, pp. 10.1-10.8. Australasian Inst. Min. Met., Melbourne (1973). Brauner G. Subsidence due to Underground Mining. Part 1 Theory and Practices in Predicting Surface Deformation, p. 56. US Department of the Interior, Bureau of Mines (1973). Hargraves A. J. (Ed.) Proc. Symp. Subsidence in Mines, Wollongong. Australasian Inst. Min. Met., Melbourne (1973). Adamek V. and Jeran P. W. Evaluation of existing predictive methods for mine subsidence in the US. In Proc. 1st Annual Conf. Ground Control in Mining, Morgantown, WV (Edited by S. S. Peng), pp. 209-219 (1981). Bauer R. A. and Hunt S. R. Profile, strain and time characteristics of subsidence from coal mining in Illinois. In Proc. Workshop on Surface Subsidence due to Underground Mining, Morgantown, WV (Edited by S. S. Peng and M. Harthill), pp. 207-219 (1981). Akimov, A. G. On methods of precalculating ground surface movements. Ugol Ukr. 2, 20-23 (1958). The Movement of the Rock Masses and of the Surface in the Main Coalfields of the USSR, p. 250. General Institute of Mine Surveying, Ugletekhizdat, Moscow (1958). Arcamone J., Schroeter P. and Dejean M. J. P. State of the art of mining subsidence in France. In Proc. 88th Annual General Meeting ofCIM, Montreal, Paper no. 84, p. 17 (1986). Bawden W. F. and Mottahed P. Comparison of three subsidence prediction techniques applied to Saskatchewan potash mining, In Proc. 88th Annual General Meeting ofCIM, Montreal, Paper no. 89, p. 34 (1986). Galvin J. M. Total Extraction of Coal Seams: The Significance and Behaviour of Massive Dolerite Sills. Chamber of Mines of South Africa, Research Report 19/82, p. 80 (1982). MacCourt L., Madden B. J. and Schumann E. H. R. Case studies of surface subsidence over collapsed bord and pillar workings in South Africa. In ISRM Symp. SANGORM, Sandton, South Africa, pp. 25-32 (1986). MacCourt L., Madden B. J. and Schumann E. H. R. The effect of underground mining on surface. ISRM Symp. SANGORM, Sandton, South Africa, p. 155 (1986). Hiramatsu Y., Okamura H. and Sugawara A. Surface subsidence and horizontal displacement caused by mining inclined coal seams. In Proc. 4th Congr. ISRM, Montreux, vol. 1, pp. 665-670 (1979). Geddes J. D. Large Ground Movements and Structures, p. 1064. Pentech, London (1977). Geddes J. D. Ground Movements and Structures, p. 964. Pentech, London (1980). Geddes J. D. Ground Movements and Structures, p. 876. Pentech, vol. 3, London (1984). Orchard R. J. Recent developments in predicting the amplitude of mining subsidence. J. R. Inst. Chart. Surv. 33, 864 (1954). Orchard R. J. Surface effects of mining - the main factors. Trans. Inst. Min. Eng. 116, 942-955 (1957). Wardell K. Some observations on the relationship between time and mining subsidence. Trans. Inst. Min. Eng. 113, 471-483 (1954). Beevers C. and Wardell K. Recent research in mining subsidence. Trans. Inst. Min. Eng. 114, 223-244 (1955). Orchard R. J. and Allen W. S. Time-dependence in mining subsidence. In Proc. Symp. Minerals in the Environment, pp. 643-659. Institution of Mining and Metallurgy, London. (1975). Forrester D. J. and Whittaker B. N. Effects of mining subsidence on colliery spoil heaps - 1 . Int. J. Rock Mech. Min. Sei. ά Geomech. Abstr. 13, 113-120 (1976). Forrester D. J. and Whittaker B. N. Effects of mining subsidence on colliery spoil heaps - II. Int. J. Rock Mech. Min. Sei. & Geomech. Abstr. 13, 121-133 (1976). Shadbolt C. H. A Study of the Effects of Geology on Mining Subsidence in the East Pennine Coalfield. PhD Thesis, University of Nottingham, p. 570 (1987). Lee A. J. The effect of faulting on mining subsidence. Min. Eng. (London) 71, 735-743 (1966). Aston T. R. C. and Whittaker B. N. Undersea longwall mining subsidence with special reference to geological and water occurrence criteria in the North East of England Coalfield. Min. Sei. Technol. 2, 105-130 (1985).
780 33. 34. 35. 36. 37. 38.
Back Analysis Monitoring Bruhn R. W., Magnuson M. O. and Gray R. E. Subsidence over the mined-out Pittsburgh Coal Bed. ASCE Convention Coal Mine Subsidence Session, Pittsburgh, PA, pp. 26-55 (1978). Karfakis M. G. Mechanisms of chimney subsidence above abandoned coal mines. In Proc. 6th Int. Conf. Ground Control in Mining, Morgantown, WV (Edited by S. S. Peng), pp. 195-202 (1987). Marino G. G. Mine subsidence damage from room and pillar mining in Illinois. Int. J. Min. Geol. Eng. 4,129-150 (1986). Hunt S. R. Surface Subsidence due to Coal Mining in Illinois. PhD Thesis, University of Illinois Urbana, IL, p. 129 (1980). Piggott R. and Eynon P. Ground movements arising from the presence of shallow abandoned mine workings. In Large Ground Movements and Structures. (Edited by J. D. Geddes), pp. 749-780. Pentech, London. (1977). Whittaker B. N. Surface subsidence aspects of room and pillar mining, Mining Department Magazine, University of Nottingham 37, 59-67 (1985).
29 Ground Surface Movements Due to Underground Excavation in the People's Republic of China LIU BAOCHEN Changsha Research Institute of Mining and Metallurgy, People's Republic of China
29.1
INTRODUCTION
29.2
THE BASIC EQUATION FOR AN ELEMENTARY SUBSIDENCE BASIN
29.2.1 29.2.2
Hunan,
782 783 783 786
Elementary Subsidence Elementary Horizontal Displacement
29.3
GROUND SURFACE MOVEMENTS AND DEFORMATIONS DUE TO NEAR-SURFACE TUNNELING 29.3.1 Precalculation of Ground Surface Movements Due to Tunneling with an Arbitrary Cross Section 29.3.2 The Process of Ground Surface Subsidence Produced by Near-surface Tunneling 29.3.3 Analysis of Case Histories
29.4
GROUND SURFACE MOVEMENTS AND DEFORMATIONS DUE TO UNDERGROUND MINING OF A HORIZONTAL COAL SEAM
29.4.1 29.4.2 29.4.3 29.5
797 797 799
Ground Surface Movements and Deformations Analysis of Case Histories
29.6
GROUND SURFACE MOVEMENTS AND DEFORMATIONS DUE TO UNDERGROUND MINING OF A VERY THICK COAL SEAM 29.6.1 Ground Surface Displacements and Deformations 29.6.2 Analysis of Case Histories
29.7
791 791 795 796
Two-dimensional, Time-independent Problem Three-dimensional, Time-independent Problem Two-dimensional, Time-dependent Problem
GROUND SURFACE MOVEMENTS AND DEFORMATIONS DUE TO UNDERGROUND MINING OF AN INCLINED COAL SEAM
29.5.1 29.5.2
786 786 789 790
BASIC PARAMETERS FOR GROUND SURFACE MOVEMENT
29.7.1 Parameters Dependent on Geological Conditions 29.7.1.1 The tangent of the main influence angle, ß 29.7.1.2 The extraction influence transmission angle, Θ 29.7.1.3 The horizontal displacement coefficient, b 29.7.1.4 The subsidence time coefficient, C 29.7.2 Parameters Dependent on Mining Method 29.7.2.1 The subsidence coefficient, η 29.7.2.2 The parameters S0, Sv and SL 29.7.3 Parameters Dependent on Extraction Geometry 29.7.3.1 The extraction thickness M of the coal seam 29.7.3.2 The minimum extraction depth, h 29.7.3.3 The extraction width, L 29.8
802 802 804 805 805 805 805 806 806 806 807 807 807 807 807 807
DESIGN PRINCIPLES FOR THE EXTRACTION OF COAL SEAMS UNDER SURFACE-PROTECTING OBJECTS 808 29.8.1 Introduction 808
781
Back Analysis Monitoring
782 29.8.2
Methods for Reducing Surface Movement and Deformation
29.8.2.1
Compact
filling
29.8.2.2 Partial mining 29.8.2.3 Sublayer extraction 29.8.2.4 Time delay extraction 29.8.2.5 Extraction harmonizing 29.8.2.6 Speeding the extraction 29.8.3 Protection Classification 29.8.3.1 Classification of the ground foundation 29.8.3.2 Classification of the protection level 29.9
COMPUTER PROGRAMS FOR MINING SUBSIDENCE CALCULATIONS
29.9.1 SFCMOV 29.9.1.1 Input parameters 29.9.1.2 Results 29.9.2 TUNNEL 29.9.2.1 Input parameters 29.9.2.2 Results 29.9.2.3 Example 29.9.3 BUILD 29.9.3.1 Input parameters 29.9.3.2 Results 29.9.3.3 Example
808 808 808 808 808 809 809 809 809 810 810 811 811 811 811 811 811 812 812 812 813 814
29.10
CONCLUSION
815
29.11
REFERENCES
816
29.1 INTRODUCTION This chapter introduces the basic theory of stochastic media which was developed in China and has been widely used there since the 1960s. The application of this theory for the prediction of ground surface movements and deformations due to near-surface tunneling and extraction of coal seams is described here. A large number of cases and design principles are described. Since the 1950s an extensive research program has been conducted in China, aimed at predicting the ground surface movements and deformations produced by underground mining and nearsurface excavations. The aim is to minimize damage to surface structures such as buildings, railways and rivers from mining and the excavation of underground railways, storage caverns and other large spaces. Recently, an increase in the need for improved transportation and a drive towards the conservation of surface areas and other ecological aspects have led to increased use of underground space for railways, storage and shopping centers. These underground services are now regarded as an essential part of life in modern, large cities and are placed close to the surface for convenience of use and to keep excavation costs low. However, caving into these spaces is liable to damage existing surface structures and services. In some cases the potential surface damage was estimated to be so great that the planned underground project was either changed or canceled. With the accumulation of engineering experience, a theoretical approach has been developed for the prediction of ground surface movements, deformations and surface damage due to underground excavation. This theoretical approach, which is based on a stochastic method, has been used since the 1960s by design institutes and companies to design the extraction of coal seams beneath buildings, railways and rivers. It can also be used to design underground railways in cities. For a quantitative approach to any mechanical phenomenon and its effects, it is necessary to understand the physicomechanical nature of the stressed body - but the intrinsic properties of natural masses can be very complex. However, when making a detailed analysis of the mechanical behavior of a body, it is necessary to idealize the actual body as if it is composed of a certain medium. Since the 1890s several kinds of idealized medium have been used in rock mechanics to model rock masses. Among those frequently used are continuous media such as elastic, elastoplastic and rheological media and discontinuous media such as loose media, geomedia and stochastic media. Because of jointing, the rock mass can be considered as a structure composed of a large number of rock blocks, which are different in size and shape but closely locked together. The degrees of freedom for a single block are too many for classical mechanics to be able to define precisely the motion trajectories of the rock particles. In the mid-1950s, taking into account the fact that the movement of a fractured rock mass is governed by a great number of known and unknown factors, Litwiniszyn
Ground Surface Movements Due to Underground Excavation in China
783
suggested a new method of computing rock mass motion [1]. According to Litwimszyn a rock mass can be considered as a stochastic medium, and the problem of calculating its motion can be solved by a stochastic method [2]. Since the 1960s this method has undergone continual improvement through experiment and has been widely applied to coal mining in Poland and China [3]. Based on the stochastic medium concept, a number of solutions for the calculation of rock motion in different geological and extraction conditions have been obtained. The solutions have been used in mining practice and underground space construction to solve the excavation problems under buildings, railways and rivers. Satisfactory comparisons have been made between theory and practice. 29.2 THE BASIC EQUATION FOR AN ELEMENTARY SUBSIDENCE BASIN 29.2.1 Elementary Subsidence According to statistics, an underground excavation can be divided into infinitesimal excavation elements The effect due to the total excavation is then equal to the sum of the effects caused by the infinitesimal excavations. An excavation with an infinitesimal unit width, length and thickness (άξ, άζ and άη, respectively) is called the elementary excavation. The subsidence and horizontal displacement of any point in the elementary basin are called, respectively, the elementary subsidence We and the elementary horizontal displacement Ue. A rectangular coordinate system is chosen with the vertical axis Z directed upward from the elementary excavation. Based on a probability analysis, the motion of a rock mass element over the elementary excavation may be considered as a random event which takes place with a certain probability If the rock mass is isotropic in the horizontal plane, then the probability density function will be continuous and symmetrical about the axis Z. The occurrence of subsidence in an infinitesimal area dS ( = dXd Y) at the horizon Z with the point A(X, Y Z) at itscenter is equivalent to the occurrence of simultaneous subsidences in the horizontal strips dX and άΥ through A (Figure Mathematically, we can write the probabilities for these two events &sf(X2)AX and/(7 2 )dr, where/is the density function. The probability of the simultaneous occurrence of these two events is P(dS) = f(X2)dX f(Y2)dY
=
f(X2)f(Y2)dS
(1)
Through the origin (O), new rectangular coordinates (Χ', O, T) are chosen such that the coordinates of point A are (Xu Ylt Z). Using the new coordinates, the probability of simultaneous occurrence will be P(dS,) = f(X2)dXlf(Y2)dYi
=f(Xi)f(Y2)dSi
Figure 1 Influence of the elementary excavation
(2)
784
Back Analysis Monitoring
Based on the fact that the probability P(dS) does not change with the selection of the coordinate system, if the excavation elementary area dS = dSx and point A does not change, then f(X2)f(Y2)
=f(X?)f(Y2)
(3)
X 2 = X 2 + Y 2, ^ = 0
(4)
If the axis OX passes through the point A, then
and inserting equation (4) into equation (3) gives f(x2)f(Y2)
= f(x2
+ r 2 )/(0) = cf(x2 + Y2)
(5)
Differentiation of equation (5) yields J[
}
df(x2 + Y2) d(x2 + Y2) = df(x2) = 2 d(X ) d(X2 + Y2) d(X2) d/ir 2 ) _ ^df(x2 + Y2) d(x2 + Y2) _ d(Y2)
d(X2 + Y2)
d(Y2)
δ / ( χ 2 + Y2) d{X2 + Y2) df(x2 + Y2) d(X2 + Y2)
then 1
d/(* 2 ) d(A:2)
=
2 fry2,d/(r ) n 2
' d(Y )
and finally l
df(x2)
/ ( X 2 ) d(^ 2 )
l
d/(r 2 )
f(Y2)
d(Y2)
(6)
The left-hand side of equation (6) is the function for X2, while theright-handside is the function for Y2. Both sides must equal a constant K, thus df(X2)
2 d(X 22Γ ) = K/(* )
df(Y2) = d(Y2)
Kf(Y2)
W
Solving the differential equations (7) and considering the condition that as X approaches ± oo then P(dS) = 0 give f(X2)
=
f(Y2)
= <ϊ(Ζ)εχρ[-π7 2 /Γ 2 (Ζ)]
and hence P(dS) = q2(Z)exp\_-n(X2 + y2)/r2(Z)]dA-dy
(8)
The three-dimensional density function can now be written as f{X,Y,Z)
= q2(Z)expl-n(X2
+ r 2 )/r 2 (Z)]
(9)
where q(Z) and r(Z) are coefficients dependent on Z. The above density function governs the geometric distribution of the subsidence of rock particles in the elementary excavation. As the elementary excavation is a component of the total excavation, which is large enough to cause the rock mass above the goaf to move, then the elementary subsidence must be an actual occurrence. Equating the subsidence distribution of the elementary basin to the density function gives We(X,Y,Z,t) = [9(Ζ)]2βχρ[-π(Α-2 + y2)/r2(Z)]d^dÇd^
(10)
Ground Surface Movements Due to Underground Excavation in China
785
From equation (10), the volume of the elementary basin will be Λ + οο
Λ + οο
q2(Z)cxpl-n(X2
Υ2)/ν2(Ζ)]άξάζάηάΧάΥ
+
J - oo J - oo
(H)
The subsidence develops with time as the overlying rock mass presses onto the goaf at a rate governed by UK dr
(12)
C(l - K)
where C is a constant. Under the condition t = 0, Ve = 0 and t -► oo, Ve = άξ άζ άη and the solution to equation (12) is (13)
Ve(t) = [1 - exp(-Ct)]d{dCd>j Substituting equation (11) into equation (13) gives q2(Z)
=
1 r2(Z)
(14)
[1 - e x p ( - Q ) ]
Substituting equation (14) into equation (10) produces We(X, y, Z, t) =
1 [1 - e x p ( - a ) ] e x p [ - 7 r ( X 2 + r2(Z)
Υ2)/ν2(Ζ)-]άξάζάη
(15)
For the two-dimensional problem, the length of the elementary excavation is infinite in the 7 direction. Integration of equation (15) gives Wt (X,Z9t)
= i + 0 °r- 4 ^ [ l - e x p ( - a ) ] e x p { - [ Z 2 + (Y J -co
r(Z)
\IJ)
ζ)2Μτ2(Ζ)}άξάζάη
[1 - exp(-a)]exp[-7iX 2 /r 2 (2:)]d£drç
(16)
After Knothe [4], the coefficients R and β are called the main influence range and the main influence angle, respectively, and are related by r(H) = R =
H
tanß
where H is the excavation depth.
w <£,ζ>
z.-n Figure 2 Rectangular excavation ( V is the horizontal displacement in the Y direction)
(17)
786
Back Analysis Monitoring
Finally, the elementary subsidence in the elementary basin on the ground surface is given by 1 W9(X, Y,t) = - ^ [ 1 - exp(-a)]exp[-7r(X 2 + Υ2)/Κ2^άξάζάη
(18)
and WJX,t) = L·1 -exp(-Ct)lexp(-nX2/R2WdV K
(19)
This is the basic equation for the analysis of ground surface movements. By integration, the ground surface subsidence caused by a rectangular excavation will be (see Figure 2) W(X9Y9t)
= P r ^ - ^ C l - e x p ( - a ) ] e x p { - 7 r [ ( X - ξ)2 + ( 7 - ζ) 2 ]/Κ 2 }α£αζ
(20)
where 1¥(ξ9ζ) is the roof subsidence.
29.2.2
Elementary Horizontal Displacement
Like the ground surface subsidence, the horizontal displacement of the ground surface is produced by the underground elementary excavation. In the study of horizontal displacement it is assumed that the rock mass is incompressible at any moment, and for a plane strain state sx + εγ + εζ = 0, εγ = 0
(21)
ôl/e dWe —- + — - = 0
(22)
thus
where εχ, εγ, and ε ζ represent the strains in the rock mass in the X, Y and Z directions, respectively. Solving equation (22) yields Ue(X) = ί^ d * ÖZ
+
C
(23,
Inserting equation (16) into equation (23) and assuming the boundary condition X -+ + oo and Ue = 0 produce the expression for the horizontal displacement of the ground surface in the elementary subsidence basin Ut(X,t) = - ^ ? [ 1 - exp(-a)]exp(-7rX 2 AR 2 )d£d,y
(24)
where b is the horizontal displacement coefficient
29.3
GROUND SURFACE MOVEMENTS AND DEFORMATIONS DUE TO NEAR-SURFACE TUNNELING
29.3.1 Precalculation of Ground Surface Movements Due to Tunneling with an Arbitrary Cross Section An example is shown in Figure 3 of a long horizontal tunnel of depth H with arbitrary cross section. It is clear that the problem can be reduced to that of a plane strain state. Consider an extreme case where the underground tunnel has totally collapsed. The maximum subsidence will be reached after an infinite time (t -+ oo ) following the total collapse. A sufficiently large area of excavation can be treated as being composed of an infinite number of infinitesimal elementary excavations άξ άζ which have the same influence on the surface. The equation for the
Ground Surface Movements Due to Underground Excavation in China
787
Figure 3 Tunnel excavation
final surface subsidence produced by the elementary excavation can be obtained from equation (19) [5, 13] for t-> oo W.(X9H) = —exp[-7rX 2 /r 2 (>/)]did>7
(26)
The ground surface subsidence due to the total collapse (area Ω) can be obtained using the principle of superposition and considering that r(rç) = rçcot /?, that is W(X) =
2
-Titan 22o(X-t) 0
^-exp
άξάη
(27)
The total collapse of the tunnel represents the worst case, producing subsidence outside the acceptable limits. It also provides the upper estimate of the maximum ground surface subsidence. Experience in underground engineering has shown that when the tunnel is excavated and supported correctly only small movements occur in the surrounding rock. Hence, ground surface movement depends on the nature and extent of the convergence over the cross section of the working. After excavation, the area of cross section Ω converges to ω and the amount of surface subsidence is equal to the subsidence difference between the areas of cross section Ω and ω. The difference takes the form W(X) = WQ{X)
-
Ψω(Χ)
-ί
ί
tanß -exp
-t)2~
X
J -7ütan22/?
άξάη
(28)
J(0-Û))J
If a tunnel of circular cross section with initial radius A converges to radius B after a long period of time, then the surface subsidence can be obtained from equation (28) and -exp -7rtan 2 )g ( *
W(X) = tanß j
2®
άξάη
-ΓΓΗ--"'^]«*}
where a = H
c = -JA2
b = H +A - (H -
ηγ
d = JA2
-(H-
η)2
b' = H + B
a' = H - B
c' = -JB2
(29)
-{H-
η)2
d' = V ß 2 - (if - n)2
Back Analysis Monitoring
788
The performance of the above equation can be compared with the observations from a case history described in the literature [5]. In this case the tunnel is 17.4 m below the surface with a cross section 4.5 m high by 5.7 m wide. The area of convergence is about 0.76 m 2 . The parameter ß which characterizes the behavior of the rock mass is determined by comparison of the cross sectional area of the subsidence basin with that of the convergence. The assumption is that these areas are equal, and for this case tan j? = 1.37. A comparison between the calculated curve and the measured values is shown in Figure 4. Owing to the complexity of the integration in equation (29), it is necessary to use a microcomputer for the calculation. The program 'TUNNEL' has been developed for the prediction of the ground surface subsidence W(X\ the horizontal displacement U(X) and the horizontal deformation E(X) produced by near-surface tunneling [5, 13]. A more practical method is derived from simplified boundary geometries. For example, in the Cartesian coordinate system shown in Figure 5 a vertical, uniformly distributed subsidence S occurs across the roof of the tunnel (5 <^ H\ and the tunnel is completed at time t = 0. The subsidence equation for this case is obtained from equation (19) (noting άη = S) mx,t)
= ^ [ 1 - exp(-Ci)] P exp[-7T(X - ξ)2/Ι121άξ = 0.55[1 - e x p ( - C 0 ] | y ( y ^ ^ ^ ) -
erf^^-^j
(30)
where L is half the width of the tunnel. If the parameters S, L and R are known then values from equation (30) can be computed with a small calculator or mathematical tables. The final surface subsidence can be obtained from equation (30) by assuming that, as t -► oo W(X) = 0 . 5 s [ e r f ( y ; ^ ) -
erf(Vi^)]
X(m) 6
14
8 1
10
12 ^ 1
1
6
4 1
1
2 1
H 20 E E
^J 80
Figure 4 Comparison between calculated and measured subsidences
7
0
H
1
'
1
'♦m '♦m ' L
r !
L
Figure 5 Simplified calculation model
(31)
Ground Surface Movements Due to Underground Excavation in China
789
Equation (31) can be used to determine the maximum surface subsidence which occurs above the center (X = 0) of the tunnel roof as =
Wm
SatiJÏL/R)
(32)
29.3.2 The Process of Ground Surface Subsidence Produced by Near-surface Tunneling Let us assume that the tunnel is excavated at a constant velocity V (m year -1 ) and that the distance D at time t is given by D = Vt and ζ = VT (Figure 6). Then, the ground surface subsidence after advancing a distance άζ can be obtained from equation (19) W(X, Y,t) = Γ Γ Α j i _ expTcft - l Y | J « p [ - ( X · - ξγ/V = 0.25s[erf(^^) -
-
{Υ-ζγ/Κ^άξάζ
e r f ( ^ ^ ) ]
X {e rf ^(K + K () ]-e rf (^,)-e Xp (g!-^-C t )
*Κχ
-j-fê'-iâ)]}
R
ινφι
IVyß)
(33)
Assuming that the working face advances in the negative Y direction and at time t = 0 the face is a horizontal distance R from the point under consideration, then from equation (33) W(X,t) = o . 2 5 s [ e r f ( y ^ ^ 4 ^ ) " "*(>/* S ^ ) ] H ] l ( K t ~ R)]
+ 1
(34)
— exp 4nV2
If all the parameters are known, the excavation depth is small and the excavation velocity is large, then as V -> oo equation (34) can be simplified to - c r f ^ ^ — ^ ) 1 [ 1 - exp (-Ci)]
W(X,t) = 025s\cd(ji^^\
(35)
The above equation can also be used to determine the horizontal displacement U(X9 i), the inclination T{X, t\ the curvature K(X, t) and the horizontal strain E(X, t).
7
Y
1
1
0
vt dC
.1 \,Vr
L
Ί Γ 'iiUUililiHUi • IA·—
1 S = dT7
\
t
Figure 6 Tunnel driven ahead with velocity V
790
Back Analysis Monitoring
29.3.3 Analysis of Case Histories Many near-surface underground excavations in the north-western part of China were driven through soil strata containing gray-yellow, uniform, compact soils. Figure 7 shows the excavation geometry at underground engineering number 1 where a tunnel 80 m long was driven 19.5 m (H) below the ground surface. The tunnel cross section is 5.3 m high (h) by 6.7 m wide (2L) with a clearance height and width of 4.5 m and 6.0 m, respectively. A full-face driving technique advanced the tunnel at a rate of 1 m day" 1 in loose soil which was supported using a three-pieced arched structure formed from precast concrete plates. A survey line of seven survey pegs was established on the ground in the direction normal to the center line of the workings. The surface subsidence was measured 17 times over a period of 180 days. The radial displacement of the tunnel walls was measured directly below the survey line. The results illustrated in Figure 8 show a maximum tunnel roof subsidence of 16 mm. Assuming that the subsidence area of the tunnel roof is equal to the profile area of the surface subsidence basin, using the condition S = 16 mm and substituting H and L into equation (31) produce the equation for the profile of the subsidence basin on the ground surface W(X) = 8[erf(0.0636 X + 0.213) - erf (0.0636 X - 0.213)]
(in mm)
(36)
where tan ß = 0.7 is derived from the observed data. The comparison between the calculated results from equation (36) and the observed values is shown in Figure 9. Substituting H, L, S, tan/? and C = 13 year"1 into equation (35) produces the equation for the ground surface subsidence process as a function of time W(X,t) = 8[erf(0.0636X + 0.213) - erf(0.0636X - 0.213)] [1 - exp(-13t)]
(in mm)
(37)
The comparison between the calculated results from equation (37) and the observed values is shown in Figure 10. Five other tunnels in this region were monitored in the same way. The
1
2 1
1
3
4
5 1
1
6 1
7
1
19.5 m
0 L_
5 1
10 m 1
S
1
\
6 / 7m
5.3 m
,
Figure 7 Profile of underground engineering number 1
Figure 8 Cross section convergence of underground engineering number 1
Ground Surface Movements Due to Underground Excavation in China
791
Figure 9 Comparison between calculated and observed surface subsidences
Figure 10 Subsidence as a function of time: comparison between calculated curve and results from observation
Table 1 Parameters of Tunnels Tunnel number
Depth (m)
Width (m)
W max (mm)
tanß
1 2 3 4 5 6
19.50 41.00 45.11 27.70 29.60 26.67
6.70 6.50 10.80 7.04 8.10 10.50
4.04 3.40 8.20 4.75 4.95 13.50
0.7 1.5 1.2 1.2 1.5 0.7
geometries of those tunnels are summarized in Table 1. Comparisons between the observed results and the theoretical calculations for each of these five other tunnels are shown in Figure 11. The comparisons in Figure 11 show that the subsidence theory is suitable for near-surface tunneling in loose strata. Values that could be used for a predictive calculation are tan ß =1.0 and C = 13 year"1. 29.4 GROUND SURFACE MOVEMENTS AND DEFORMATIONS DUE TO UNDERGROUND MINING OF A HORIZONTAL COAL SEAM 29.4.1 Two-dimensional, Time-independent Problem Consider a horizontal coal seam of thickness M with roof subsidence \¥(ξ, η) after extraction (Figure 12). The maximum possible subsidence Wmax in this case is given by W& η) = WmA%n
(38)
where η < 1 and is the subsidence coefficient which is dependent on the extraction system and strata control method.
792
Back Analysis Monitoring X (m) 0
20
10
30
o > ^ ~ °l
1 h
H = 41.0m ZL = 6.5 m
4 hr
1
1
>"
-
/
L \
^
/
H = 45.1 m ZL = 10.8 m
^r
// = 27.7 m ZL = 7.0 m
4
1
\^^r\
2 4
o /
H = 29.6 m 2Z. =8.1 m
/ / = 26.7m 2Z. = 10.5 m
Figure 11 Comparison between calculated and observed surface subsidences forfivetunnels
Figure 12 Finite excavation of a coal seam
When t -* oo the extraction has been completed and the subsidence basin fully formed. The ground surface subsidence due to an extraction of finite width S (calculation width) and infinite length (c -► oo, d -> oo) will be (Figure 12) [5] W(X) = 0 . 5 ^ m a x p ( y i £ ) -
e r f ^ ^ ^ l
(39)
Ground Surface Movements Due to Underground Excavation in China
793
The horizontal displacement of the ground surface is obtained by substituting equation (39) into equation (21) and considering that as t -► oo, c -> oo and d -► oo, and for X -► + oo, U = 0; hence U(X) = bWmilx{exp(-nX2/R2)
- exp[-7c(X - S)2/R2]}
(40)
where b is the horizontal displacement coefficient and is given by b =
1 dr(Z) Umax —- = ——
(41)
Figure 13 compares the subsidence data from one mine with the curves obtained from equations (39) and (40) [2]. Under conditions of semi-infinite extraction (S -> oo, practically 5 > 2R)9 the rock mass is sufficiently disturbed and the maximum possible subsidence appears on the surface. Equations (39) and (40) now become W(X) = iWmaxll - erf(v^*/Ä)]
(42)
U(X) = b\Vmaxexp(-nX2/R2)
(43)
Equations (42) and (43) are of basic importance both for the prediction of surface movements and deformations and for the design of extractive processes that are to take place under surfaceprotective objects. Since the 1960s numerous field observations of subsidence have been collected and used to calibrate the theory. Some of these observations are shown in Figure 14 in dimensionless coordinates (X/R9 W/WmdiX) [2] where they are compared with the results calculated from equation (42), shown as a continuous curve. Calculated horizontal displacements and the corresponding field observations are compared in Figure 15, which uses the dimensionless coordinates X/R and U/Umax [2, 3]. Differentiation of equations (39) and (40) produces the following parameters which define the deformations of the ground surface due to an extraction of finite width S. (i) Inclination of the ground surface = ^{exp(-nX2/R2)
T(X) = ^ ^
- exp[-7r(X - S)2/R2^}
(44)
(ii) Curvature of the ground surface K
(X)
= —777- =
=
Av2
+ [ ( * - S)/R^xpl-n(X
{-(X/R)exp{-nX2/R2)
p2
- S)2/R2V
(45)
(iii) Horizontal strain of the ground surface Ι7/Λ^
E{X) =
dU(X) άχ
=
2nbWmax
+ l(X - S)/R1 exp L-n(X
V2,r>2^
{-{X/R)exp(-nX2/R2) - S)2/R21}
U (mm)
J^(mm)
Figure 13 Comparison between observed and calculated subsidences
(46)
794
Back Analysis Monitoring
Figure 14 Observed subsidence and theoretical curve
Figure 15 Horizontal displacement and theoretical curve
where S is the width of the extraction used in the calculation and 5 = L - 2S0
(47)
where L is the extraction width of the coal seam and S0 is the width of the uncompacted zone. From equation (47), when L <> 2S0, W{X) = 0; thus, there is no subsidence under these conditions. Differentiation of equations (42) and (43) produces the following parameters which define the deformation of the ground surface due to a semi-infinite extraction. (i) Inclination of the ground surface T(X) =
w
-βχρ(-πΧ2/Κ2)
(48)
The inclination distribution follows a Gaussian curve with the maximum inclination at the point X = 0, that is
w
(49)
2nWm&xX exp(-nX2/R2) R2
(50)
(ii) Curvature of the ground surface K(X) =
The maximum curvature occurs at the point X, where X = ±
R
JÏ*
= ±0AR
(51)
and K± m .x =
W ±1.52-
(52)
Ground Surface Movements Due to Underground Excavation in China
795
(iii) Horizontal strain of the ground surface w
E[X)=
x -2nb^-exp(-nX2/R2) R
(53)
The maximum horizontal strain occurs at the points Xe, where Xe = ±—7=
=
±0AR
(54)
±L52bTm
(55)
and bW £±max = ± 1 . 5 2 - j p =
29.4.2
Three-dimensional, Time-independent Problem
Consider the extraction of a coal seam where both the width and the length of the extraction area are less than 2R, then the extraction volume is given by 2Sx2qxM (see Figure 16). The ground surface subsidence obtained from equation (20) is given by [7, 8]
Γ 1- e x p [ - 7 t ( X -
W(X, Y)
S) 2 /R 2 ]dS
J-sR W
R
J-f
exp[-7r(r -
q)/R2ldq (56)
W°(X)W°(Y)
where W°(X) and W°(Y) are the subsidences due to finite extractions in the X and Y directions, respectively. The expressions for horizontal displacement in the present case are [7, 8] UX(X, Y) =
Wmax{exp(nX2/R2)
w rr
- ε χ ρ [ - π ( Χ - S) 2 /« 2 ]}
m
J-, «
cxp[-n(Y-q)2/R2ldq
=
W
U$(X,Y)W°(Y)
(57)
and Uy(X, Y) =
w
■U?{X, Y)W°(X)
(58)
where ϋχ(Χ, Y) and ϋγ(Χ, Y) are the horizontal displacements due to finite extractions in the X and Y directions, respectively. Some results from theory and field observations are compared in Figure 17.
/ Y
,
Q
H:
/
s / . {
+1 /
.
A
/
,y
Figure 16 Two-dimensional extraction
796
Back Analysis Monitoring 240 h h E
160
Γ
80
VMK
l· D
S-^
80 \- 200 ~
E
240 -
600
Λ-
oV
160 U 400 ~
"
>
/ ^
*
0
L
20
1
40
1
60
1
80 m
1
Figure 17 Comparison between observed and calculated horizontal displacements
29.4.3 Two-dimensional, Time-dependent Problem A longwall face in a flat coal seam at depth H advances uniformly with constant velocity V (Figure 18). By adapting equation (18) to the two-dimensional case and substituting in the corresponding parameters we get the subsidence as a function of time Wt(X,t)
=
W
:{1 - exp[-C(i - τ)]}βχρ[-π(Α- - S)2/i?2]dS
(59)
which after integration becomes
W(X,t) = 0.5 ^m„ ίβΓΐΓγ(* + Κί)1 - ^Lfi^j x [erfjV^
+
^ f
-
ex
p ( ^ I - %X -
Ct
- -£*,) - erf(^f - -i*=)]}
(60)
Similarly, the horizontal displacement is given by r7/v
x
bRWmax \C
fC2R2
C
\Γ
/ rX
rVt
CR \
(61)
Equations (60) and (61) can be applied to the solution of important problems such as: (i) surface deformation with time; (ii) displacement rate and strain rate; and (iii) the subsidence process of a surface point and its trajectory. The subsidence and displacement of a surface point distant from the stable edge of the extraction are given by [2, 6]
W(t) = 0.5 ^ Jerfl^Vt - K)l - lj 0.5 W^ exp
p
+ c R _ C l ) H ^,_ R ) __^]_ l }
(62)
and i/(t) =
bCRWm IV
(63)
In Figures 19 and 20, 35 observations from seven mines are compared with the subsidence curves computed from equation (62).
Ground Surface Movements Due to Underground Excavation in China z
X
r 1
NS
0 1
VT
Vt
S~*
v
H
—
1
j ^ g
^
797
-H
h-dSf =0 5 =0
Figure 18 Longwall face advancing with velocity V
Figure 19 Time-dependence of subsidence: a comparison between results from observation and theory
S 0.4
¥
5
Figure 20 Time-dependence of subsidence: a comparison between results from observation and theory
Figure 21 illustrates the trajectories of surface points from three mines and compares them with curves computed from equations (62) and (63). 29.5 GROUND SURFACE MOVEMENTS AND DEFORMATIONS DUE TO UNDERGROUND MINING OF AN INCLINED COAL SEAM [2, 3, 9, 10] 29.5.1 Ground Surface Movements and Deformations Practice and model experiments [2] have shown that when mining inclined coal seams the center of the elementary subsidence basin shifts from a point above the center of the excavation (B; Figure 22) to a new point (A). The shift of the subsidence basin does not change the symmetry of the basin shape. The angle between the horizontal and the line passing through the maximum subsidence point and the center of the excavation is called the angle of extraction influence transmission, and is
798
Back Analysis Monitoring
Figure 21 Comparison between calculated and observed trajectories of surface points
Figure 22 Excavation of an inclined coal seam
denoted by Θ =
90° -
(64)
OLK(Z)
where K(Z) is a coefficient dependent on the properties of the rock mass with values in the range 0
tan0
(65)
tan[90° -
Applying equation (20) to the two-dimensional case and substituting X — p(Z) for X we get the elementary subsidence basin for inclined strata as t -* oo W.(X)
=
1
■exp r
r(z - Z)
{1
■\X - S - p(z - Z)V
Λζ-. r2(z-Zy
(66)
From equation (23), the elementary horizontal displacement will be
,„„, Vt\A)
r(z - Z)dr(z - Z)dfVt{X) =
2π
r-^ 9Ζ
r-r; SX
9^-Z)^,^ — 6Z
M^(A)
(67)
Ground Surface Movements Due to Underground Excavation in China
799
Integrating equation (66) gives the surface subsidence due to the extraction of an inclined coal seam with extraction width L and minimum extraction depth h [2, 10] W(X
fft 1 f IX - S - (h + Stana)cot0] 2 , J Jo = Wmaxt3inß\ e xr p ^ - — y —-^ — π tan 2 0>dS * J e Λ + Stana \ (h + Stana) 2 *J
68
where a = Sv cos a and b = (L — SL) cos a. Combining equations (67) and (68) gives the surface horizontal displacement as U(X) = -bWmax\
f
*b
27ttan2jg _ [AT - S - (6 + Stana)cotö] (h + Stana) 2
/f t c^
iX x ez*P\-ntzn>ß \
{h St -S~ Z*)C°2 ter\*S (h + t5 tana)
j
" * ( * > « « * (69)
The maximum possible subsidence is ^max = We max sin Θ = Me>/sin0
(70)
where W^max and M 0 are the maximum ground surface displacement and the effective coal seam thickness, respectively, in the direction of Θ. If M
* =
then max
M
yv cos(l — K) n
(71)
M»cos(JCa) cos[a(l - K)2
K
The inclination of the ground surface is given by τ χ
()
C aW(X) _ w Γb*In 2;rtan3ff = —dX 7^ ~ = ~ ~* I /L ο , Λ 33 ^ -S-(h + S tana) cot0] " " L (Λ +, Stana) x ex
P i - ,, "*!" , i [ * - S - (Λ + S tan a) cotö] 2 IdS _| (h + Stana) 2 j The ground surface curvature is given by
K{X) =ffi)= _ „ _ p 2«^/» f dX
a
J a (h + S tana) 3 [
W/>_
(h + Stana) 2 L
v
(73)
Stana)cotö] 4 J
π x exp< ** η P [ x _ s - (fc + S tan a) cot 0] 2 >dS [ (h + Stana) 2 J
(74)
The horizontal strain of the ground surface is given by E(X) = d-^P dX = -bW„ x exp<{ -
29.5.2
2β β Λ 2 - (h *Τ' «ν2 L ]\\*? + Stana) a {h + Stana) (
7Γ tan 2 ß
2
-
S
- <* v + ^ a n « ) c o tJ0 ] l ' J
[X - S - (fc + Stana)cot0] 2 }>dS jdS - Γ(Χ)α*0
(75)
Analysis of Case Histories
Because of their complexity, equations (68) to (75) are normally solved using a microcomputer. The program 'SFCMOV for use on a PC has been developed by the author [10]. Figure 23 shows an example from a mine where the theoretical curves (solid lines) for subsidence W(X) and horizontal displacement U(X), calculated from equations (68) and (69), respectively, are compared with the results obtained from in situ observations.
'
800
1
400
1
Back Analysis Monitoring
200
^250
0 1
60 1
E
^ ^ ^Τ&' 750
120 m 1 37
^Ά
*
"
Figure 23 Comparison between calculated curves and results from observation for subsidence and horizontal displacement
Consider the case of the Jixi coalfield in China, which uses the following parameters h = 43 m
L = 69 m
K = 0.7
a = 13°
tanß = 1 . 9
η = 0.73
θ = 8Γ
b = 0.36
M = 1500 mm 5L = 14 m
5t; = 6m
Substitution of these parameters into equations (68), (69), (73), (74) and (75) yields the following expressions. (i) Vertical subsidence
W(X) = 2059
f 53 · 59
1 Γ (X - 1.0368 S - 6.88ΥΊ —— exp - 11.34 —— d5 Js.85 43 + 0.235 Fl V 43 + 0.235 ) \
(76)
v
;
(ii) Horizontal displacement of the ground surface
U{X)
-
-
no„ f 5359 * - 1.03685 - 6.88
8852
J,85
(-53.59
"
326
(43 + 0.23S)* !
Γ
J , 8 5 43-Tä235
11 34
«PL" · !
r eXP
/ * - 1.03685 - 6.88\2~| IdS
t4
/χ _ 11 34
L' - (
L 0 386S
43 + 0.23S
- 6.88\ 2 ~|
43 + 0.23S
)J
JO
) ]«"
(77)
(iii) Inclination of the ground surface (in mm m * ) Ç53.95 x _ 1>()3685 - 6.88 Γ (X - 1.30685 - 6.88V1 T(X) = -46720 — „„ox, exp -11.34 —— d5 v F ' J5 85 (43+0.235 3 L V 43+0.235 ) \ 3
(iv) Curvature of the ground surface (in 10 f 53 · 95
y
l
\
1
m
l
(78)
) Γ
43 + 0.23S
(X - 1.03685 - 6.88V1
/ J
(79)
Ground Surface Movements Due to Underground Excavation in China
801
(v) Horizontal strain of the ground surface (in mmm"1) E(X)
= -8852
f53·95
i
3
Γ_
J5.85 (43+0.23 5) |_
X- 103{ 1.03685 - 6.88 \ 2 226J V 43 + 0.235 5395
x exp
5 . 85
X - 1.03685 - 6.8 (43 + 0.23 5) 3
Γ (X - 1.03685 - 6.88 V "I jj x«p[-11.34( 43+023S
(80)
The calculated curves are compared with the observed results in Figure 24.
WMmm)
(b)
50 h
T ( mm m" ' )
P 50 U
Figure 24 Comparison between results from observation and calculated curves for (a) vertical subsidence, (b) inclination, (c) curvature, (d) horizontal displacement and (e) horizontal strain
802
Back Analysis Monitoring
29.6 GROUND SURFACE MOVEMENTS AND DEFORMATIONS DUE TO UNDERGROUND MINING OF A VERY THICK COAL SEAM 29.6.1 Ground Surface Displacements and Deformations The coal seams of the Fushun and Fuxin coalfields in north-east China are inclined with thicknesses reaching several tens of meters. For a rhomboidal extraction area the slope of the seam is relatively small over the extraction depth and length (Figure 25). The subsidence basin resulting from such an extraction is called a 'Fushun'-type subsidence basin. From ref. 3, the subsidence will be W(X,Y,t)
2 - S - p)2 + (Y - q)2 \dSdqdm
-
= q2{Z)Gxp\^Zï^(X ( " ^
(81)
and at time i, the volume of the basin will be Λ + οο Λ + οο
V(t) =
W(X,Y,t)dXdY
=
q2(Z)r2(Z)dSdqdm
(82)
When the extraction rate is constant, the relationship between the volume extracted (Figure 26) and time for 0 < t < T is given by
SM<4-H
dV(t)
(83)
where V0 is the extraction volume. The solution of equation (83) under the initial condition of t = 0, V(0) = 0, is V(t) = j£lCt
+ e x p ( - C t ) - 1]
(84)
Substituting equations (82), (84) and Wmix = ±dSdm
(85)
into equation (81) gives, for 0 < t ^ T W W(X,t) = -^ICt
+ exp(-Ct)-
llexp(-nX2/R2)
(86)
and for t > T W(X,t) = W(X,T) + \_W(X) - W(X, Γ)]{1 - exp[-C(t - Γ)]}
(87)
where lim W(X,t) = W(X) =
W^expi-nXi/R2)
Figure 25 Extraction geometry
(88)
Ground Surface Movements Due to Underground Excavation in China
803
vit)
t
T
Volume of extraction with time
Figure 26
Finally W(X,t)
= Wm
{1 + ^
\_2CT - e x p ( - C r ) - 1 ] [ 1 - exp(-Ci +
CTmexp(-nX2/R2) (89)
The horizontal displacement of the ground surface is obtained from equation (21) as [3] U{X) = - WmJlnb^
+cot0jexp(-7rX 2 /K 2 )
(90)
The equations for W(X) and U(X) (equations 88 and 90) are used to obtain the following parameters which define the ground surface deformations, (i) The inclination distribution of the ground surface is given by T(X)
=
iffiQ . „„I».,,-.*./«")
(91)
with the maximum inclination occurring at the point X =
+
2π
=
+0.4K
(92)
W R
(93)
given by
(ii) The curvature distribution of the ground surface is given by ^ _
dT(X) dT(X)
2nWmAX(lnX2
\
(94)
The maximum concave curvature occurs at X = 0 and is given by K - « . = 6.28
^
(95)
±0.7/?
(96)
The maximum convex curvature occurs at X = ± ,/(3/2π) R = and has the volume
w
K + max — 2.8·
(97)
804
Back Analysis Monitoring
(iii) The horizontal strain distribution of the ground surface is given by E(X) = ^ ^ ^ 2 π | ^ - l ) + | ο ο ΐ ψ χ ρ ( - π Χ 2 / * :
(98)
The maximum values of horizontal strain are calculated from the distribution of the horizontal strain. 29.6.2 Analysis of Case Histories The results of ground surface subsidence observations from both coalfields are compared in Figure 27 with the theoretical curve calculated from equation (88). Observations from both coal-
Figure 27 Comparison between observed subsidence and theoretical curve
Figure 28 Time-dependence of subsidence: a comparison between results from observation and theory
Figure 29 Comparison between observed horizontal displacements and theoretical curves
Ground Surface Movements Due to Underground Excavation in China
805
fields showing the time dependence of the subsidence process are compared in Figure 28 with the theoretical curve calculated from equations (86) and (87). Both of these figures use dimensionless coordinates {X/R and W{X)/Wm^ in Figure 27; Ct and W(t)/WK in Figure 28). The horizontal displacements from eight observation stations are shown in Figure 29. The theoretical prediction curves based on equation (90) for b = 0.1,0.2 and 0.3 are included in the figure. 29.7 BASIC PARAMETERS FOR GROUND SURFACE MOVEMENT The basic parameters for the prediction of ground surface movements and deformations due to underground extraction are divided into three groups which are dependent upon geological conditions, mining method and extraction geometry. 29.7.1 Parameters Dependent on Geological Conditions The subsidence parameters which are dependent on geological and stratigraphie conditions include tan j3, 0, band C [2]. 29.7.1.1 The tangent of the main influence angle, ß The parameter tan ß is mainly dependent upon the mechanical properties of the overburden strata. For Chinese coalfields tanj? varies between 0.7 and 3.5; thus, ß varies between 35° and 74°. For most cases, tan ß = 1.2-3.0 and ß = 50-71.6°. Tan ß increases as the stiffness and strength of the overburden decrease. For an overburden consisting of clay or sand, the value of tan ß is close to 1.0. Figure 30 shows the values of tanjS gathered from most of the Chinese coalfields. The figure illustrates the tendency for tan ß to increase with extraction depth. The line of best fit is (H in meters) tanß = 1.4 + 0.00225 H
(99)
29.7.1.2 The extraction influence transmission angle, Θ The extraction influence transmission angle Θ is defined as Θ = 90° - KOL
(100)
where K is a coefficient with values in the range 0 < K < 1 and a is the dip angle of the coal seam. For coalfields where the overburden has not been heavily disturbed by tectonic activities and the rocks are strong, the value of K is close to 1. Table 2 lists data gathered from eight coalfields in China. The relationship between the value of K and the extraction depth is shown in Figure 31 with the best fit estimate given by (H in meters) K = 0.5 + 0.0005 H
Figure 30 The relationship between tan ß and extraction depth
(101)
806
Back Analysis Monitoring Table 2 Parameters of Coalfields Coalfield
Depth (m)
Fushun
383 342 525 530 534 385 379 347 550 33-105 108-208 35-105 108-208 208-328 126-156 82-125 13-87 43-55
Huainan Fuxin Beipiao Shuanyashan Jiawang Benxi Jixi
Dip angle (°)
K
71 71.5 69 64 70 71 66 62 60 77 62 74 55 50 81 80 74 81
28 34 34 32.5 29 27 31 32 46 65 60 32 26 63 15 15 30 13
200
0(°)
400
0.680 0.540 0.620 0.805 0.690 0.736 0.772 0.594 0.605 0.20 0.47 0.50 0.96 0.93 0.60 0.65 0.59 0.69
600
H (m)
Figure 31 The relationship between K and H
29.7.1.3 The horizontal displacement coefficient, b The parameter b is determined from the observation results when a = 0° and the extraction area is large b =
Um
w
(102)
For Chinese coalfields the coefficient b lies in the range 0.2-0.4. A suitable initial estimate in most cases is b = 0.4 [2]. 29.7.1.4 The subsidence time coefficient, C The parameter C mainly depends on the behavior of the overburden. The value of C for an overburden consisting of massive, strong sandstone is lower than it is for an overburden consisting of layered, weak rock. The results from observations in coalfields show that the value of C decreases as the extraction depth increases. 29.7.2 Parameters Dependent on Mining Method The values of the parameters η, S09 Sv and SL depend on the extraction and strata control methods used.
Ground Surface Movements Due to Underground Excavation in China
807
Table 3 Subsidence Coefficients for Different Mining Methods
29.7.2.1
Mining and strata control methods
Range ο/η
Value ofr\ used
Longwall face with roof caving, without pillars Room pillar extraction without recovery pillars Partial dry-filling by zones with input materials Full dry-filling with input materials Hydraulic filling with waste oil shale Hydraulic filling with small hard rocks Hydraulic filling with sand Shortwall face with roof caving and long pillars (recovery 50-60%) Shortwall face with hydraulic sand filling and long pillars (recovery 50-60%) Low-strength concrete hydraulic filling
0.60-0.80 0.40-0.60 0.55-0.70 0.30-0.40 0.20-0.30 0.15-0,25 0.06-0.20 0.03-0.10
0.70 0.50 0.60 0.40 0.25 0.25 0.15 0.08
0.02
0.02
0.02-0.05
0.02
The subsidence coefficient, η
For different mining schemes, the value of the subsidence coefficient η varies widely from 0.02 for a shortwall face with hydraulic sand filling and long pillars (recovery 50-60%) to 0.8 for a longwall face without chain pillars where the roof collapses. Values of the coefficient that have been gathered from coal and metal mines in China are listed in Table 3. The values given in Table 3 are used to predict mining subsidence in a new coalfield or a new area of a mine. In situ observations show that the coefficient is only weakly related to the extraction depth and rock properties.
29.7.2.2
The parameters S0, Sv and SL
The horizontal distance in the strike direction between the inflection point of the subsidence basin and the goaf edge of the longwall face is defined by the parameter S0. Similar definitions apply to the parameters Sv and S L , except that the distance element is measured in the dip direction. These parameters are mainly dependent on the strata control method and are determined separately for each mine.
29.7.3 Parameters Dependent on Extraction Geometry 29.7.3.1
The extraction thickness M of the coal seam
The extraction thickness M of the coal seam is measured in the direction normal to the coal seam in units of millimeters. In most cases the extraction thickness is equal to or less than the thickness of the coal seam. If the extraction thickness in the extraction panel is not uniform, then the average thickness is used in any relevant calculations.
29.7.3.2
The minimum extraction depth, h
The minimum extraction depth h is measured, in meters, at the upper edge of the extraction panel.
29.7.3.3
The extraction width, L
The width of the extraction is measured, in meters, as the distance between the two edges of the extraction area in either the dip direction or the strike direction of the coal seam.
808
Back Analysis Monitoring
29.8 DESIGN PRINCIPLES FOR THE EXTRACTION OF COAL SEAMS UNDER SURFACE-PROTECTING OBJECTS 29.8.1 Introduction The underground extraction of coal seams causes a variety of phenomena such as ground surface movement, deformation, and so on. The objects (e.g. construction, river or railway) on the ground surface within the region affected by mining will be to some extent damaged as a result of the extraction process. The damage owing to the extraction results from an accumulation of a variety of factors, such as the extraction thickness of the coal seam, distance between the coal seam and the ground surface, mining method, strata control method, rock properties, underground water and geological conditions. To protect surface buildings from damage requires mining procedures other than those normally employed. The principles of extraction under surface structures have to be specially considered, and these should include special extraction methods and strata control methods based on the importance of the surface building. With these considerations in mind the protection level and the related tolerant surface movement magnitude index can be estimated, back-analysis from which will facilitate the choice of a better mining method. In the following discussion, which is based on the author's experience and his early statistical data, some generally useful ways of reducing the surface movement and deformation resulting from underground extraction will be described. The building protection level and its relevant index of tolerant surface deformation, a concept recently developed, will also be discussed. 29.8.2 Methods for Reducing Surface Movement and Deformation 29.8.2.1 Compact filling Data from a large number of observations show that the maximum subsidence of the ground surface is closely related to the extracted thickness of the coal seam and the strata control method. The maximum surface subsidence is about 60-80% of the extracted thickness of the coal seam when the roof-caving method is employed, but for compact filling, such as hydraulic sand backfill, the maximum subsidence is only 6-20% of the extracted thickness of the coal seam. The extent of compact filling required depends both on the backfilling method and the filling material. Values of the subsidence coefficient η for different mining and strata control methods are shown in Table 3.
29.8.2.2 Partial mining Partial mining, or strap mining, is used in coal mines and is successful in reducing subsidence. The strap method involves extraction of alternate straps in a circular extraction area. The recovery ratio is in the range of 50-60%. From the existing data, where strap mining is combined with hydraulic filling the maximum possible subsidence is only 2% of the extracted thickness of the coal seam, and if combined with the roof-caving method the maximum possible subsidence will be in the region of 3-10%. Strap mining is mainly used when there are some important buildings on the ground surface.
29.8.2.3 Sublayer extraction Sublayer extraction is suitable for a very thick coal seam. The basic idea is to divide the whole thickness of the coal seam into several layers. In this way the extraction can be spread over a long enough period of time to guarantee that the subsidence caused by the early extractions has ceased or diminished to an acceptable level before the next sublayer is extracted.
29.8.2.4 Time delay extraction Time delay extraction is important in the case of strap mining. After extraction of part of the strap of coal the subsidence is allowed to cease or diminish; the remainder of the strap is then extracted. In this way the total surface movement appears at certain time intervals. This is a superior method for
809
Ground Surface Movements Due to Underground Excavation in China
the protection of surface buildings. As a matter of fact, time delay extraction and sublayer extraction are two of the best methods at reducing the accumulation of the various effects of excavation. 29.8.2.5 Extraction harmonizing Extraction harmonizing is a very powerful and more-skilled way of protecting surface buildings from damage owing to underground extraction. The method is particularly effective when several coal seams or sublayers in a thick coal seam are to be mined. The idea is that, for example, tensile deformation regions caused by surface movements in the later extraction steps will be counterbalanced by compressive deformation regions caused by surface movements in the earlier mining steps. In this way the total effect on the surface can be significantly reduced. 29.8.2.6 Speeding the extraction Accelerating the extraction is a better way of protecting the ground surface during the extraction period. Normally, the relative reduction in surface deformation will be large for fairly hard rock or small extraction depths, and for soft rock or large extraction depths the effect will be small. 29.8.3 Protection Classification In practice, no matter what kind of mining or roof control method is applied, the safety of buildings on the ground surface will still be to some extent affected. The effects on surface structures on the one hand depend on the ground base movements and on the other hand depend on the building itself, such as its ability to resist deformation and shearing. Apart from these factors, the importance of the building is the main point that has to be considered. The level of protection required both for the ground base and the building and the relevant allowable magnitude of surface deformation, the assessment of which is based on statistical observation and experimental data, will presently be discussed. The index described below has been successfully used for many years in the coal-mining industry of China [12]. 29.8.3.1 Classification of the ground foundation The classification is applied to surface buildings specifically to determine the extent of the damage caused to the buildings as a result of ground deformation. Table 4 classifies ground foundations of constructions and their deformation tolerances into five levels. (i) Class I means that no resources have been mined and no extraction is planned for the future. (ii) Class II applies when excavation activities have been stopped for 10 years or more and there will be no future extractions. However, if at least three separate observations show that the average Table 4
Ground Base Classification
Ground base classification
Construction protection level
I
Building area needs no protection, only very small cracks in walls Building area needs simple protection, small cracks in walls, easily repaired Building area needs careful protection, fairly large cracks in walls, still easily repaired Building area needs to be specially protected, but damage still repairable Building impossible in this area: large cracks in walls, caving of structures
II III IV V
Allowable deformation T E K -1 (mmm-1) (xl0~6m_1) (mmm ) <2.5
< 1.5
<50
<5.0
<3.0
<83
< 10
<6.0
<166
<15
<9.0
<250
>15
>9.0
>250
810
Back Analysis Monitoring Table 5 Ground Base Classification Ground base classification
Horizontal deformation (mmm-1)
I II III
< 1.5 <3.0 <6.0
Minimum depth of coal seam, Hmin (m) 1000 500 250
Table 6 Protection level and Deformation Magnitude Index
Protection level
Condition of the construction
I
No major damage, small cracks in walls Some breaks in the building, easily repaired Major damage to the building, but no danger of collapse Special treatment required to prevent collapse of the building
II III IV
Inclination (mmm - 1 )
Allowable movement Deformation Curvature (mmm - 1 ) (10_6m-1)
<2.5
<1.5
<50
<5.0
<3.0
<83
<10.0
<6.0
<166
<15.0
<9.0
<250
subsidence velocity is less than 3 mm month" 1 for an extraction depth greater than 70 m, then the ground foundation can be considered as class I, while for an extraction depth less than 70 m the ground foundation belongs to class II. (iii) In the case where extraction has been stopped for less than 10 years and the average subsidence velocity is more than 3 mm month -1 , the ground foundation can only be classified as class IV. (iv) Table 5 can be used in the estimation of the ground foundation classification in cases where a coal seam is being extracted or will be extracted in the near future. 29.8.3.2 Classification of the protection level There are four classes of protection level for buildings, based on the importance of the building and its sensitivity to the horizontal deformation of the ground. Class I is for the more important buildings, the damage of which may lead to disastrous consequences. Class I buildings include those that contain very important industrial equipment, factories which are very sensitive to surface movements and deformations and other constructions the damage of which may have life-threatening consequences. Recently, some cultural remains were considered as class I because of their value. Class II buildings are not very sensitive to surface movements and deformations. Ordinary factories, railways and tunnels, large rivers, reservoirs, schools and dormitories can all be considered as class II structures. Class III structures include highways, bridges, small rivers and wood houses. Class IV structures include sports grounds, unimportant highways and other unimportant constructions. Table 6 shows the relationship between the protection level and the tolerant surface deformation magnitude index. 29.9 COMPUTER PROGRAMS FOR MINING SUBSIDENCE CALCULATIONS Computer programs for the calculation of ground surface subsidence and deformation are very important for design engineers. The main aspects of predicting surface movement and deformation with a computer program are as follows.
Ground Surface Movements Due to Underground Excavation in China
811
(i) Some buildings on the surface are directly above the coal seam to be mined or are within the area affected by surface movement caused by the underground extraction. To meet the condition of both efficient extraction of coal and proper protection of buildings from damage it is necessary to predict the allowable ground deformation and hence choose the appropriate mining method. (ii) As industrial requirements for resources increase, some safety pillars will be mined, such as coal pillars remaining from the strap-mining method. In these cases the surface movement must be calculated and the safety of the buildings must be guaranteed.
29.9.1
SFCMOV
SFCMOV is a computer program for calculating the ground surface movements and deformations due to the underground extraction of a coal seam. The program is based on a plane strain analysis and can be applied to the extraction of a flat or an inclined coal seam [10, 11, 14].
29.9.1.1
Input parameters
Tangent of the main influence angle, ß Dip angle of the coal seam, a Propagation angle, Θ Subsidence coefficient, η Thickness of the coal seam, M Minimum extraction depth, h Up inflection point movement, Sv Down inflection point movement, SL Horizontal movement coefficient, B
29.9.1.2
(dimensionless) (degrees) (degrees) (dimensionless) (mm) (m) (m) (m) (dimensionless)
Results
The following distributions can be calculated and the results presented in tabular and graphical form. Ground surface subsidence, W(X) Ground surface horizontal displacement, U(X) Inclination of the ground surface, T(X) Curvature of the ground surface, K(X) Horizontal strain in the ground surface, E(X)
29.9.2
TUNNEL
TUNNEL is a computer program for calculating the ground surface movements and horizontal strains due to near-surface tunneling [5, 13, 14].
29.9.2.1
Input parameters
Excavation depth H, the vertical distance between the ground surface and the center of the tunnel Tangent of the main influence angle, β Initial radius of the tunnel, A Reduced radius of the tunnel after deformation, B Horizontal movement coefficient, Bx
29.9.2.2
(m) (dimensionless) (m) (m) (dimensionless)
Results
The following distributions can be calculated for a cross section normal to the tunnel axis. The results can be presented in tabular and graphical form.
812
Back Analysis Monitoring
Ground surface subsidence,. W(X) Ground surface horizontal displacement, U(X) Horizontal strain in the ground surface, E(X) 29.9.2.3 Example The determination of ground surface movements and horizontal strains caused by tunneling at 20 m below the ground surface has the following input parameters H = 20m
A = 3.0m
tanß = 1.2
Βχ = 0.40
B = 2.9 m
The resulting curves of ground surface subsidence, horizontal displacement and horizontal strain are shown in Figure 32. 29.9.3 BUILD BUILD is a computer program for the determination of the optimum pillar size in a single coal seam such that a building on the ground surface is adequately protected from damage. When the deformation of the ground surface due to a proposed underground extraction is likely to be greater than that allowable for the buildings on the ground surface, two solutions are available: (i) select a new mining method which decreases the deformation of the ground surface; or (ii) leave a safety pillar of optimum size [11, 12]. 29.9.3.1 Input parameters Tangent of the main influence angle, ß Dip angle of the coal seam, a Influence propagation angle, Θ Subsidence coefficient, η Thickness of the extracted coal seam, M Minimum extraction depth, h
(c) O
18
16
14
12
(dimensionless) (degrees) (degrees) (dimensionless) (mm) (m)
10
8
6
4
2
C
9.94 4.97
Γ*
° i
E
-
-4.97 — Uj -9.94 -14.91
Figure 32 Computer-calculated curves for (a) subsidence, (b) horizontal displacement and (c) horizontal strain
Ground Surface Movements Due to Underground Excavation in China
813
Begin I Protection category I
i
i
'
Γ Input | 1 Initial geometry display """1 r-H Calculation 2 h—i I Calculation 3 h Calculation I Is deformation in direction of upper strike permissible? f
I
=
Is deformation in direction of dip permissible? £"<£max Y/N -N 1 —v '
Is deformation in direction of lower strike permissible?! £
| Function keysj | Restart |
\WiX)\
[ Printing [
T (X) I Figure 33
Deformation distribution
\UiX)\
Pillar | profile
\KiX)\
Pillar plane
End
\^m
Algorithm of the computer program BUILD
The coefficient of the inflection point location, ß, is given by the inclined distance in the dip direction between the inflection point of the subsidence basin and the goaf edge divided by the main influence range of the goaf edge (dimensionless) Horizontal distance between the origin and the nearest side of the building, ΧΛ (m) Horizontal length of the building in the dip direction, AB (m) Horizontal length of the building in the strike direction, BA (m) Horizontal displacement coefficient, B (dimensionless) Maximum allowable horizontal strain in the ground surface, £max (mmm" 1 ) Maximum allowable inclination of the ground surface, T (mmm - 1 ) ■*■ m a x The function block of the program BUILD is shown in Figure 33. 29.9.3.2 Results Ground surface movements and deformations are presented in tables and graphs. The following distributions can be calculated. Ground surface subsidence, W(X) Ground surface horizontal displacement, U(X) Ground surface inclination, T(X) Ground surface curvature, K(X) Ground surface horizontal strain, E(X) The geometry of the cross section in the dip direction can also be determined. The relative locations of the building on the ground surface and the safety pillar in the coal seam can be shown on a graph. The location plan of the building and the pillar geometry can also be plotted. The following parameters pertaining to the geometry of the safety pillar can also be determined. Left part extraction length, Lx Right part extraction length, L2
814
Back Analysis Monitoring
Length of the safety pillar in the dip direction, A Length of the safety pillar in the upper horizontal direction, Ax Length of the safety pillar in the lower horizontal direction, A2 29.9.3.3
Example
Consider the determination of the optimum pillar size in a single, inclined coal seam for an ordinary factory on the ground surface. An ordinary factory has a class II protection classification and the tolerant surface deformation magnitude indexes are (Table 6) TInax = 5mmm - 1
£max = 3mmm~1
The input parameters are tanß = 2.5 M = 4000 mm ΧΛ = 150 m
a = 25°
η = 0.70
Θ = 80e
B = 0.40
h = 50 m
Q = 0.1
AB = 50 m
BA = 100 m
The following plots and parameters are obtained from the above input data. (i) The ground surface movements and deformations are shown in Figures 34 and 35. (ii) Figure 36 shows the geometry of the cross section in the dip direction along with the locations of the building on the ground surface and the safety pillar in the coal seam.
Figure 34 Computer-calculated curves for (a) subsidence and (b) horizontal displacement
E E E
ξ
Figure 35 Computer-calculated curves for (a) inclination, (b) curvature and (c) horizontal strain
Ground Surface Movements Due to Underground Excavation in China
815
(iii) The plan sheet of the location of the building and the pillar geometry is shown in Figure 37. (iv) The following parameters relate to the geometry of the safety pillar L,
= 104.58 m
L2 Ax
= 206.46 m
= 114.48 m
A = A2
156.36 m
= 289.84 m
29.10 CONCLUSION Mining and underground engineering shows that the theory of stochastic media can be used in practice for solving the problems encountered in the prediction of ground surface movements and deformations. X
Figure 36 Table 7
(m)
Cross section in the dip direction
Comparison between the Theory of Stochastic Media, the FEM and the DDM
Theory of stochastic media
Numerical analysis (FEM, DDM)
Based on the probability of rock block motion No constitutive equation for the medium Prediction of displacements and deformations only Can solve four-dimensional problems (X, Y, Z, t) Maximum number of parameters depending on rock properties is four (β, Θ, B, C) Easy to use for mining engineers Calculations require portable computers or microcomputers
Based on the mechanics of the phenomenon Uses various kinds of constitutive equation Prediction of displacements, deformations and stresses Can solve four-dimensional problems (X, Y, Z, t) Number of engineering properties for each stratum is at least two Engineers must be trained in methods of numerical analysis Calculations require microcomputers, minicomputers or mainframe computers Accuracy of prediction is low
Accuracy of prediction is quite high
Figure 37
Plan sheet of the location of the building and the pillar geometry
816
Back Analysis Monitoring
^f^DDM
Figure 38 Comparison between the results from observation and those calculated from the stochastic method, the FEM and the DDM X (m) -73
-37
0
\ 201
\\
108 1
•
\\ \\
402
503
DDM
X^ 6 0 3 ~
1
144
180
1
216
..•H··"
252 1
r loi
302
X
36 72 1 1 Shoemaker II mine
/
1 Λ
1
Observation
•3
704 ~ \ 805
Î5
905
FEM
.'\ » ·\
Stochastic.^· method ^*""'·*
-V \\ i / - X Λ ; ··'
1006
Figure 39 Comparison between the results from observation and those calculated from the stochastic method, the FEM and the DDM
Table 7 shows the comparison between the stochastic medium theory, the finite element method (FEM) and the displacement discontinuity method (DDM) in the prediction of ground surface movements. Figures 38 and 39 show examples in which the results from observations are compared with the results calculated from the stochastic medium theory, the FEM and the DDM. ACKNOWLEDGEMENTS I am grateful to Professor Liao Guohua of Beijing University of Science and Technology for his cooperation during the last 30 years in the research of mining subsidence. I would also like to thank Dr Lin Dezhang, Dr Cai Yuejun, Dr Shen Huiqun and Dr Zhang Jiasheng for their help in computer programming. 29.11
REFERENCES
1. Litwiniszyn J. The theories and model research of movements of ground masses. In Proc. Eur. Congr. Ground Movement, Leeds, UK, pp. 203-209 (1957). 2. Liu Baochen and Liao Guohua. The Basic Laws of Ground Movement in Coal Mining, p. 191. The Industrial Press of China, Beijing (in Chinese) (1965).
Ground Surface Movements Due to Underground Excavation in China 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14.
817
Liu Baochen, Liao Guohua and Yan Roungui. Research in the ground surface movement due to mining. In Proc. 4th Congr. Int. Soc. Rock Mech., Montreux, pp. 408-413. Balkema, Rotterdam (1979). Knothe S. The displacements of the surface under the influence of mining extraction and their theoretical interpretations. In Proc. Eur. Congr. Ground Movement, Leeds, UK, pp. 3-12 (1957). Liu Baochen and Lin Dezhang. Surface movements and deformations due to near-surface tunneling. Undergr. Eng. 7, 1-7 (in Chinese) (1983). Liu Baochen. Motion of rock masses due to advancing exploitation butt in the light of the theory of stochastic media. Bull Acad. Pol. Sei. Sér. Sei. Tech. 10(4), 243-252 (1962). Liao Guohua and Liu Baochen. The time-space problem in ground movements. J. Chin. Coal Soc. 1(3), 3-14 (in Chinese) (1964). Yan Roungui and Liu Baochen. The space problem in ground movements due to mining. J. Chin. Coal Soc. part 1, 30-42 (in Chinese) (1979). Liu Baochen. Application of the theory of stochastic media to the determination of the profile of the subsidence trough on the ground surface due to the exploitation of an inclined deposit. Bull. Acad. Pol. Sei. Sér. Sei. Tech. 9(9), 541-546 (1961). Shen Huiqun and Liu Baochen. Calculation of ground movement and deformation due to the extraction of an inclined coal seam. J. Chin. Coal Soc. part 4, 44-50 (in Chinese) (1987). Liu Baochen and Shen Huiqun. The mathematical model and solutions for the optimization of coal pillars in the prevention of damage to buildings. J. Chin. Coal Soc. part 3, 18-25 (in Chinese) (1988). Liu Baochen and Cai Yuejun. Computer analysis for the determination of the optimum size of safety pillars. In Proc. 33rd Annual Reports of CRI MM, pp. 4-7 (in Chinese) (1988). Liu Baochen. Theory of stochastic media and its application in surface subsidence due to excavation. Trans. Nonferrous Metals Soc. China 2(3) (1992). Liu Baochen. Stochastic method for subsidence due to excavation. In Proc. Int. Symp. Application of Computer Methods in Rock Mechanics and Engineering, Xian, China (1993).
Subject Index abrasion wear drag tools, 168 abrasion value tunnel boring machines performance, 278, 282 abrasive jet drills excavation, 242 abrasive water jets excavation, 241 abrasive wear cutting tools tunnel boring machines, 280 abrasivity drag tools wear, 172 number Cerchar, 174 accelerometers blast modeling, 49 monitoring, 98 microseismic activities monitoring, 704,710 accuracy deformation measurement, 685 acoustic emission crack detection, 717 mining supports monitoring, 709 rock failure, 698 adapters extension drilling percussive drilling, 151 adhesion rock reinforcement, 462 aerial imagery thermal, 725 AFNO blasting, 41 airblast analysis blast monitoring, 102 blasting, 96 stress wave energy, 41 control, 107 measurement instruments, 120 microphones, 98 monitoring regulations, 103 production, 97 air decks high-speed photography blasting, 67 air overpressures
blast-induced, 114, 119 blasting, 96 regulatory compliance blast effects, 134 air temperature inversion blast effects, 120 aluminum modeling of blasting mechanisms, 53 analytic definition, 4 anchoring systems Pfaender Tunnel field measurements in decision making, 590 anchors definitions, 414 heads, 427 construction, 436 definitions, 415 friction grips, 431 length definitions, 415 Anderton shearers * coal mining, 158 angle of draw subsidence, 763 anhydrite mining stability, 392-393 tunneling decision making, 591 water field measurement tunneling, 573 anisotropy caverns stability, 374-388 tunnel boring machines penetration rate, 279 antichaos definition, 15 arches natural supports, 351 stability, 365 stiffness tunnels, 365 arching safety margins, 612 tunnel supports, 361 argillaceous materials mining stability, 393-393 audits construction, 34 Australia coal mining
819
820
subsidence, 760 axial cutting roadheaders theoretical model, 200
back analysis definition, 544 deformation jointed rock, 552 direct approach, 547 displacements, 550 inverse approach, 547 mathematical formulation, 548 procedures, 547 rock engineering, 543-569 rock mass failures, 713 stochastic, 548 back filling linings, 28 background noise microseismic emissions monitoring, 708 barrier branching blasting, 45 basalt tunnel boring machines penetration rate, 273 performance, 278 beam element method support design, 335 bearing plates rock anchorages, 427, 432 bedding horizontal stresses, 389 mining stability, 390-391 oblique stability, 389 vertical galleries and stability, 389 bedding planes blasting, 45 visual monitoring, 619 bedding sets blasting compressive strength, 43 bedrock conditions subsidence, 767 Belchen Tunnel stability, 392 swelling decision making, 590 Berea sandstone pick-tool interactions modeling, 180 bifurcation theory failure boreholes, 379-380 bit wear reduction jet-assisted cutting, 254 blast disturbance environmental aspects, 84 ground vibration, 88 blast size, 91 coupling factor, 90 delay timing, 94 single hole response, 90 super-position of waveforms, 88
Subject Index
vibration attenuation curves, 90 vibration enhancement, 92 overpressure, 85 face vibration, 86 reduction, 87 seed waveform model, 88 venting explosion gases, 85 blast effects control, 129 environmental effects, 132 blast excitation character, 114 blasting common problems, 74 damage vibration, 79 damage of hanging-wall Luossavaara mine, 509 energy input rates, 21 environmental effects, 96 excavation, 19 fracture control, 62 fracture controlled technique, 65 fragmentation mechanisms, 39-68 frequency effects structural response, 124 hanging-wall drift Luossavaara mine, 508 human response, 114 improvements, 71-93 Luossavaara mine, 498,499 measurement instruments, 120 mechanisms, 67 open stope mining rock mass response, 485-510 optimization, 72 performance monitoring, 75 postsplit, 20 presplit method, 19 productivity, 108 rock loosening, 354 smooth wall, 20 stress wave controlled, 62 structural response, 124 blast monitoring blasthole velocity of detonation, 75 measurement, 76 blast vibration records, 74 control, 105 control techniques, 95-109 design optimization, 74 event monitoring, 72 limitations, 73 explosive performance, 75 extensometer measurements, 82 gas penetration, 81 instrumentation analysis, 99 procedure, 99 instruments, 121 deployment, 124 placement, 98 methods, 95-109 regulations, 95-109 blast motions transient nature, 117 blasts air overpressures, 114 blast vibration records blast monitoring, 74
Subject Index
blast vibrations effects, 96 monitoring, 111-134 peak level determination, 99 peak levels monitoring, 100 sinusoidal, 116 block cave mining rock mass rating system, 321 blocks description stability, 390 displacement stability, 391 in situ size distribution of excavation, 15 instability rock reinforcement, 472 reinforcement, 315 rotation rock reinforcement, 472 toppling rock reinforcement, 475 block shear tests rock reinforcement, 467 block size support design, 326 block theory reinforcement design, 315 blowouts airblast control, 107 body waves blast excitation, 115 bolting portable microseismic processor recorders, 710 bolts fully grouted convergence measurements, 621 monitoring, 709 bonds rock reinforcement, 462 boom cutters tunnels excavation, 354 boom force excavating machines P.C.MAP program, 214 boreholes breakout, 372 collapse mechanisms, 369-409 dilatometers deformation measurements, 579 field tests rock anchorages, 416 geometry crater blasting, 58 notching fracture control blasting, 63 ovalization, 372 failure, 381 pressurization blasting, 40 stability, 372 faults, 407 supports faults, 407 television cameras, 641 thermoelastic analysis stability, 395
wall stability anisotropy, 374 boundary conditions computer modeling percussive drilling, 140 boundary element analyses modeling stress measurement, 662 boundary integral method rock reinforcement, 474 Brazil test tunnel boring machines, 270 breakout wedge blasting, 51 bridges long-span suspension rock anchorage, 441 stone Cayeli, 3 brittle cracking tunnel boring machines, 267 brittle failure tunnel supports, 363, 365 brittle rock indentation tunnel boring machines, 269 buckling monitoring Huntly West mine*s test drive, 742 Buffalo subway tunnels tunnel boring machines penetration rate, 273 BUILD computer program mining subsidence, 813 bulk modulus back analysis, 547 burden blasting, 40 effect on blasting, 51 button bits extension drilling percussive drilling, 152 button cutters, 20
cable bolts crown pillar Luossavaara mine, 498 design, 341 cables microseismic activities monitoring, 705 caking drilling fluids, 394 calcrete tunnel supports, 356 calibration measurement instruments blast effects, 123 cascading system definition, 7 casting airblast control, 107 cathedrals construction, 3 caverns arched concrete linings design, 648 excavation
821
822
rock mass behavior, 631-651 failure, 373-403 natural stability, 372 rock anchorage, 441 shape excavation, 647 shapes rock mass behavior, 632 solution-mined stability, 372, 399-402 thermal effects and stability, 395 stability anisotropy, 374-388 back analysis, 556 storage stability, 371,400,401-402 supports, 360 caving falls failures, 381 irregular sandstone roof beds in coal mines, 540 mining method, 18 roof pillar extraction in coal mines, 515 subsidence, 762 caving height collapsed junction, 775 prediction, 773 cavitating water jets excavation, 239 cavities elliptical supports, 351 tangential stress, 365 ceiling rocks subsidence cavern excavation, 635, 647 cellular automata construction, 9-11 cements grouts compressive strength, 425 rock reinforcement, 458 rock anchorages grouts, 425 center cutters tunnel boring machines, 263 CERCHAR test tunnel boring machines, 270 performance, 282 chalk weathering rock anchorages, 423 Channel Tunnel construction, 23 chaotic behavior engineering perturbations construction, 15 charges linear shaped fracture controlled blasting, 67 chipping P.C.DRUM program, 212 picks modeling, 182 tunnel boring machines, 267 water jets picks, 222 churn drilling percussive drilling
Subject Index
method, 137 circumferential grooves crater blasting, 57 civil engineering structures support design, 314 clamping pressure fragmentation blasting, 49 classification scheme rock reinforcement, 471 clay tunnels ground support, 363 water tunneling and field measurements, 573 clay minerals mining stability, 393 clay shales tunnel boring machines, 264 clearance drum shearers evaluation, 214 clearance design excavating machines P.C.DRUM program, 211 clearance picks drum shearers modeling, 195 clearance rings cylindrical drums design, 187 laboratory scale tests, 204 drum-type heads modeling, 193 excavating machines models, 205 cliffs stabilisation rock bolt reinforcement, 441 coal deposits, 514 origin, 514 geology, 514 inclined seams subsidence, 797 mechanized mining, 159 properties laboratory tests, 733 stiffness in situ measurement, 732 pressuremeters, 743 underground monitoring, 734 strength, 733 very thick seam mining subsidence, 802 Young's modulus, 734,743 coal cutting machines, 20 coal industry UK face support loading, 538 coal measures carboniferous, 514 coal mines design, 515 in situ testing, 731-750 layout, 515-518 methods of working, 515-518 sink holes United States of America, 778 supports, 513-541 tunnel boring machines
Subject Index
penetration rate, 273 tunnels supports, 530-536 coal mining blast effects response spectrum analysis, 127 excavation machines, 156 explosions vibrations, 118 extraction processes, 514 longwall panel stress concentrations, 26 People's Republic of China, 791 rock mass rating system, 321 supports, 514—515 coal ploughs coalmining, 157 development, 178 coal seams extraction thickness subsidence, 807 surface-protecting objects extraction and subsidence, 808 coatings rock anchorages corrosion control, 429 cobalt tungsten carbide drag tools, 167 coefficient of friction rock reinforcement, 462 coefficient of rock strength tunnel boring machines, 270 collapse mechanisms, 369-409,614 roof solution-mined caverns, 400 tunnels People's Republic of China, 786 collapse chimneys height subsidence, 774 water, 775 compact filling subsidence, 808 complaints blast monitoring, 104 compliance blast monitoring, 104 compressed air adjacent tunnels, 602 storage cavern stability, 401 tunneling field measurements in decision making, 604 compressive strength assessment of cuttability machine performance, 172 blasting strain levels, 41 rock mass rock anchorages, 416, 421 rocks tunneling costs, 304 compressive stress cage controlled fracturing, 60 compressive stresses rock reinforcement, 471 computer graphics percussive drilling, 148 computer modeling percussive drilling, 137-153
computer programs mining subsidence, 810 computer simulation cylindrical drums, 199 pick cutting machines, 211 concrete blast effects, 132 mechanical characteristics, 716 precast segment linings tunneling costs, 301 rheology, 716 supports coal mines, 531 temperature arched concrete linings, 638 tunnel supports, 354 concrete linings design, 337 conical depressions limestone, 752 connected vessel leveling roof sinkage measurement, 654 constitutive equations phenomenological approach, 716 construction blast effects response spectrum analysis, 127,128 blasting, 67 dominant frequencies, 118 process rock engineering, 1-35 construction industry tendering costing methods, 304 Construction Industry Research and Information Association tunnel boring machines rock properties, 283 constructions foundation classification subsidence, 809 protection classification subsidence, 810 continuous miners coalmines, 515 coal mining machine, 158 cutting head, 188 cylindrical drums kinematic analysis, 187 introduction, 178 models laboratory scale tests, 204 number of picks modeling, 217 performance determination, 210 picks modeling, 194 continuous rock reinforcement response, 467-470 control system definition, 7 convergence back analysis, 551 caverns measurement, 634 cavern walls arched linings, 640 field measurement in tunneling decision making, 581
823
824
longwall systems, 541 measurement cavern excavation, 645 deformation monitoring, 619 tunnel supports, 353 recording devices, 610 rock mass behavior Luossavaara mine, 508 support design, 334 supports performance, 621 tunnels People's Republic of China, 787 tunnel supports, 358, 365 convergence curves ground pressure, 611 tunnel supports, 363 conveyors tunnel boring machines, 289 core logging ground support, 356 Luossavaara mine, 486-487 corrensite swelling mining stability, 393 corrosion rock anchorages, 428-432 rock reinforcement, 462 costs site erection allowance tunneling, 298 tunnel driving, 294 tunneling rock properties, 303 COSTUN tunnel boring machines performance prediction, 287 coupled thermo-visco-elastic-plastic analysis, 718 coupling sleeves extension drilling percussive drilling, 152 crack detection gauges blasting modeling, 49 cracking blast effects, 113 mechanism, 716 residential structures blast effects and control, 129 blast-induced, 130 multiple origins, 131 natural, 130 probability, 129 rock mass blast effects, 134 structural response blasting, 124 crack propagation direction blasting, 48 disks modeling, 186 tunnel boring machines, 267 cracks gas flow into blasting, 65 tension monitoring Huntly West mine's test drive, 742 unstable propagation and detection, 718 crack velocity measurements
Subject Index
blasting, 49 crater blasting increasing efficiency, 57 mechanisms, 41,51 crater fractures blasting, 56 creep rock anchorages, 439 rock reinforcement testing, 466 tunnel supports, 363 Creighton Mine microseismic emissions monitoring, 708 critical thrust tunnel boring machines, 268 cross sectional forces determination stress at linings, 677 crown Luossavaara mine rock mass behavior, 504-508 crown hole collapse room and pillar mining, 754 crown pillars design Luossavaara mine, 493-498 displacement instrumentation Luossavaara mine, 502-503 instrumentation Luossavaara mine, 502 stability Luossavaara mine, 493 crushed rock jet-assisted cutting excavation, 248 crushing radial stress blasting, 43 tunnel boring machines, 267 water jets picks, 223 CSIR Geomechanics Classification rock reinforcement, 471 curvometers stress measurement linings, 675 cushioned charges smooth blasting, 63 cut slopes deformation back analysis, 558 constitutive equation, 559 mechanical constants back analysis, 562 stability back analysis, 558 cuttability machine performance, 172 cutter downtime tunnel boring machines rock properties, 283 cutterheads tunnel boring machines development, 288 layout, 304 cutter life index tunnel boring machines performance, 278,282 cutter/rock members percussive drilling modeling, 146
Subject Index
cutters changing tunnel boring machine costs, 306 failure rates tunnel boring machines, 281 linear tests tunnel boring machines, 278 replacement rates tunnel boring machines, 281 cutting inclined picks, 184 interactive deepened picks, 184 linear laboratory tests tunnel boring machines, 268 linear tests tunnel boring machines, 271 rock water jets, 21 single unrelieved modeling, 181 specific energy cutting efficiency, 268 tunnel boring machines, 277 cutting curves continuous miners model validation, 207 cylindrical drums computer simulation, 199 laboratory scale tests, 204 excavating machines determination, 209 modeling, 197, 198, 199 P.C.MAP program, 214 roadheaders modeling, 218 models, 205 cutting heads drag tools design, 165 cutting rate tunnel boring machines, 267 cutting tools adaptation modeling, 220 mechanical high pressure water jets, 242 speed wear, 168 tunnel boring machines failure, 280 fragmentation, 267 replacement rates, 267 rock properties, 280,283 cybernetic feedback control system rock mass construction, 33 cylindrical charges cratering, 51 cylindrical drums computer simulation, 199 kinematic analysis, 187 models laboratory scale tests, 204
damage blasting, 79 incipient blasting, 80
observable blasting, 80 powder factor, 84 smooth blasting, 80 air decking, 83 lateral decoupling, 83 loading density, 82 minimum standoff distance, 84 damage analysis crack initiation, 716 damage evolution thermal dissipation, 724 damage probability blast monitoring, 101 damping blasting dynamic response, 125 structural response, 125 dams construction, 4 rock anchorage, 441 data processors microseismic activities monitoring, 705 decoupling blast monitoring instruments, 98 defect coalescence mechanism, 723 deflagration propellants, 60 deflectometers deformation monitoring, 610 deformability rock mass rock anchorages, 417 deformation arches shear strains, 683 coal mining People's Republic of China, 791 cut slopes back analysis, 558 discontinuous jointed rock back analysis, 553 jointed rock back analysis, 552 magnitude monitoring, 619 measurement accuracy, 624 linings, 676 mining inclined coal seams, 797 very thick coal seam, 802 modeling back analysis, 558 modulus tunneling field measurement, 579 monitoring, 618 Huntly West mine's test drive, 739 safety margin, 611 stability of underground openings, 607-629 P-waves blasting, 44 rates extent of yield zones, 623 monitoring, 619,622 monitoring guidelines, 612 safety margin, 623 support effectiveness, 623 reduction
825
826
mining, 808 shear waves blasting, 44 spatial distribution monitoring, 619 stress cavern excavation, 640 tunneling field measurements, 579 People's Republic of China, 786 underground mining People's Republic of China, 782 walls, 636 wells, 380 deformeters stress measurement linings, 675 degradation blast effects, 113 delays vibration control, 105 depillaring coalmining, 515 depreciation tunneling costs, 295 depth of burial blasting, 53 design methods supports for underground excavations, 314 rational methods supports, 333 detonation blasthole velocity blast monitoring, 75 explosive performance, 75 blast monitoring, 72 blast vibrations, 96 initial pulse blasting, 51 shock energy blasting, 76 Devonport Dockyard Submarine Refit Complex rock anchorages, 443 dewatering tunneling decision making, 601 diabase stresses cavern excavation, 640 diamond bits drilling mechanically assisted cutting, 245 diamond drilling Luossavaara mine, 486-487 diamond picks excavating machines, 224 tungsten carbide picks comparison, 224 diamond-tipped picks assessment, 220 differential displacement structural response blasting, 125 difficult to relax rocks displacement cavern excavation, 647 dimensionless numbers percussive drilling modeling, 147 diorite
Subject Index
stresses cavern excavation, 640 tunnel boring machines performance, 282 Direct Strain Evaluation Technique back analysis, 548 disc cutters, 20 application, 169 diameter, 170 introduction, 178 jet-assisted cutting force reductions, 257 penetration, 169 spacing, 171 speed, 170 tire edge wedge angle, 169 discontinuities cuttability, 175 reinforcement, 460 rock anchorages, 416 rock behavior, 616 rock reinforcement, 475 responses, 470 testing, 467 sliding back analysis, 555 stability, 406-408 tunnel boring machines penetration rate, 279 tunneling costs, 308 discontinuous rock reinforcement response, 470 discrete mechanical elements rock reinforcement, 457-458 disc rolling cutters tunnel boring machines, 263 discs excavating machines laboratory scale tests, 204 modeling, 185 tunnel boring machines modeling, 192 displacement back analysis, 550 blast effects, 113 blasting, 108 caverns measurement, 634 coal mines underground monitoring, 735 damage blasting, 81 distribution in situ monitoring in coal mines, 746 field measurement in tunneling decision making, 581 horizontal basic equations, 786 instrumentation Luossavaara mine, 502-503 monitoring, 624 Hundy West mine's test drive, 739 layout, 624 opening cavern excavation, 647 ore extraction monitoring, 653-670 rock mass blast effects, 134 roof rock monitoring, 653
Subject Index
strain, 647 stress change measurement, 626 stress measurement, 654 displacement discontinuity method modeling lune-shaped flatjack, 662 pressure capsule, 662 multiple-layer mining, 662 parallel plane segments rectangular leaf elements, 662 stress measurement, 657 dissipated energy percussive drilling modeling, 144 distinct element method rock reinforcement, 474,475 stability, 391 distometers prestressed concrete linings, 585 stress measurement linings, 675 dolomite tunnel boring machines penetration rate, 273 dolostone tunnel boring machines penetration rate, 273 dominant frequency blast effects estimation, 117 doming tunnel supports, 361 double embedment axial tension test rock reinforcement, 466 double fracturing stress measurement, 656, 658, 659 down-the-hole drilling efficiency modeling, 148 percussive drilling method, 137 drag bits drilling mechanically assisted cutting, 245 jet-assisted cutting force reductions, 249 rock cutting, 230 drag picks, 20 mechanism, 20 rock cutting, 155, 161 drag tools clearance angle, 165 coalmining, 157 cutting speed, 164 drums design, 165 materials, 166 rake angle, 164 spacing, 165 water jet-assisted cutting, 166 wear, 166 drainage subsidence, 754 drifter drilling percussive drilling method, 137 drilling blast damage of hanging-wall Luossavaara mine, 509 deep stability, 372
drill bit position measurement coal mines, 738 mechanically assisted cutting large holes, 247 medium holes, 254 small holes, 243 portable microseismic processor recorders, 710 rock anchorages, 433 tunnel supports, 356 drilling bits temperature wear, 254 drilling rate index tunnel boring machines, 270 performance, 278 drum shearers cylindrical drums kinematic analysis, 187 drum design modeling, 195 introduction, 178 modeling evaluation, 214 P.C.DRUM program, 211 performance determination, 209 picks modeling, 194 water jets economics, 224 drum-type heads modeling, 193 ductile rock indentation tunnel boring machines, 269 ducts corrugated rock anchorages and corrosion control, 430 dust jet-assisted cutting excavation, 254, 259 suppression tunnel boring machines, 264 dust shields design tunnel boring machines, 288 dynamic monitoring examples microseismic activities, 707 failure, 695-714 rock masses, 713 dynamic photoelasticity fringe pattern blasting, 42 modeling blasting mechanisms, 40 dynamic response estimation blasting, 125 dynamic stress patterns measurement blasting, 42
earthquakes structural response natural frequency, 125 subsidence, 752, 754 easy to relax rocks displacement cavern excavation, 647
827
828
economics excavating machines, 223 tunnel boring machines rock properties, 293-311 tunnel driving, 294 economy tunneling field measurements, 574 Edinburgh Castle rock rock bolt reinforcement, 441 Edmonton Convention Centre excavation safety margin, 613 efficiency percussive drilling modeling, 148 elastic deformation cut slopes back analysis, 559 percussive drilling modeling, 144 elastic strain energy percussive drilling computer modeling, 140 modeling, 144 elastic waves amplitude, 698 frequency, 698 rods percussive drilling, 138 reflection coefficients, 140 transmission coefficients, 140 velocity cavern excavation, 641 El Teniente mine dimensions, 18 emissivity values, 721 empirical design supports rock mass classification systems, 317 Enassan tunnel convergence monitoring, 622 energy dissipation joints percussive drilling, 150 microcrack detection, 719 energy interaction intensity construction, 15 energy release rock failure, 696 energy transfer coordinates construction, 13 energy transfer rates engineering perturbations construction, 15 engineering perturbation construction, 7-9 entropy bar construction, 15 environmental aspects blasting, 84 environmental effects blast effects, 132 errors random deformation and measurement, 686 stress measurement linings, 679 systematic deformation and measurement, 685 Europe
Subject Index
blast monitoring standards, 104 coalmines, 514 subsidence, 760 excavating machines, 156 assessment, 178 history, 178 laboratory scale tests, 204 models validation, 204,206 performance, 178 determination, 208 specification modeling, 219 vibration, 178 excavation, 15-22 anisotropic stresses, 382 caverns rock mass behavior, 631-651 Huntly West mine test drive monitoring, 734,739 mechanical advantages, 156 future prospects, 225 harsh environment, 219 mechanized system, 262 pore pressure failures, 387 postexcavation rock reinforcement, 454 preexcavation rock reinforcement, 454 rate deformation, 613 reinforced design, 470-474 rock anchorages, 417-419 rock mass structure failures, 404-408 rock reinforcement, 452 stability rock reinforcement, 474 supports rational design methods, 333 stability, 4 0 8 ^ 0 9 swelling rocks support design, 344 types, 19-22 use of water jets, 229-259 excavation heads kinematic analysis, 187 expansion shell bolts discrete frictionally coupled devices, 458 expansion shells performance monitoring, 481 exploding wires blasting modeling, 48 explosive fracturing oil wells stimulation, 59 explosives blasting mechanisms, 39 detonation crater blasting, 58 excavation, 19 performance blasthole velocity of detonation, 75 blast monitoring, 75 energy partitioning, 78 unconfined tests, 76
Subject Index
extension drilling percussive drilling efficiency, 151 percussive drilling method, 137 extension rods drilling percussive drilling, 152 extensometers deformation monitoring, 610 hanging-wall Luossavaara mine, 508 in situ measurement coalmines, 736 length displacement monitoring, 625 measurements back analysis, 551 rock mass behavior Luossavaara mine, 508 external load determination stress at linings, 678 extraction acceleration subsidence, 809 coal seams surface-protecting objects and subsidence, 808 geometry mining subsidence, 807 harmonizing subsidence, 809 minimum depth subsidence, 807 sublayer subsidence, 808 width subsidence, 807 extraction influence transmission angle subsidence, 805
face cutters tunnel boring machines, 263 failure anisotropic tight material, 382-384 brittle, 617 caverns, 373-403 detection monitoring, 617 ductile, 617 monitoring, 617 energy release, 696 infrared thermography, 715-730 initiation mechanisms, 614 mechanisms, 614,716 rock reinforcement, 470 propagation mechanisms, 614 rock mass dynamic monitoring, 695-714 rock mass structure, 404-408 weaknesses, 615 failure modes shear, 381 tunnels, 364 failure zones in situ monitoring coalmines, 749 fatigue blast effects, 133
faults instability, 406-408 subsidence, 769 visual monitoring, 619 feed member percussive drilling modeling, 144 fictitious stress boundary element method back analysis, 547 fictitious stress method boundary element method modeling of excavation boundary, 661 field axial tension testing rock reinforcement, 468 field measurement instrumentation tunneling, 578 purpose tunneling, 578 tunneling decision making, 571-606 field sampling rock anchorages, 416 field shear testing rock reinforcement, 468 finite difference method rock reinforcement, 474 finite element analysis back analysis underground openings, 549 finite element method back analysis, 547 rock reinforcement, 474,475 finite element modeling blasting modeling, 49 in situ monitoring coal mines, 745 fireclay monitoring Huntly West mine's test drive, 740 Poisson's ratio in situ monitoring, 750 pressuremeters Poisson's ratio, 744 stiffness pressuremeters, 743 firing times blasting, 108 fissures subsidence, 770 flank abutments longwall mining, 526 flat jack loading plates deformation measurement, 579 flat jack method stress measurement, 671 flaw coalescence thermomechanical coupling, 718 flaws crater blasting initiation, 57 effect on blasting, 44 fracture control blasting, 63 fragmentation blasting, 42 large blasting, 46 natural length fracture control blasting, 64 small
829
830
blasting, 45 flexural rupture blasting, 51 flexural rupture theory blasting, 48 floors blast excitation ground motion, 115 vibration blasting, 126 flow rule loading steel rib stress measurement, 686 fluid pressure stability, 391 flushing drilling rock anchorages, 433 flying buttresses cathedrals construction, 3 fly rock blast effects, 113 blasting stress wave energy, 41 foot-wall support Luossavaara mine, 502 foundations construction subsidence, 809 Fourier spectra blast effects ground motion, 127 blast monitoring, 100 fracture control use of dynamic photoelasticity measurement, 63 fracture control plane blasting, 65 fracture initiation blasting recording, 42 fracture propagation picks modeling, 182 rock failure, 181 tunnel boring machines, 267 fractures dilation blasting, 81 joint friction and blasting, 82 mining stability, 390-391 oil wells stimulation, 59 rock behavior, 616 stem induced propagation, 61 tensile propagation Serrouville iron ore, 183 toughness blasting, 63 P.C.DRUM program, 212 tunnel boring machines performance, 277 fracture spacing cuttability, 175 fracturing controlled blasting, 59 nonradial
Subject Index
blasting, 46 water jets picks, 222 fragmentation blasting, 72, 108 in situ block size, 75 mechanisms, 39-68 blast monitoring, 75 dynamic stress waves, 40 measurement blasting, 75 tunnel boring machines, 267 fragments size distribution excavation, 15 France blast monitoring standards, 104 coal mines rock bolting, 536 free surface instabilities deformations wells, 380 Freudenstein Tunnel yielding support, 596 friction bonds rock reinforcement, 462 percussive drilling modeling, 144 rock reinforcement testing, 467 frictionally coupled devices continuous, 463 rock reinforcement, 463 frictionally coupled elements continuous rock reinforcement, 457 rock reinforcement, 457-458 frictional sparking jet-assisted cutting excavation, 254, 259 friction angle P.C.DRUM program, 212 friction bolts design, 343 friction coefficient excavating machines determination, 209 P.C.DRUM program, 211 front abutment strata pressure longwall mining, 527 FSMDDM modeling pressurized sleeve, 663 full face tunnel boring machine, 21 full face tunneling machines, 20
galleries mined storage and stability, 371 square rectangular or straight failures, 387-388 gap member percussive drilling modeling, 144 gas confinement high-speed photography blasting, 67
Subject Index
gases storage cavern stability, 401 gas leakage location infrared thermography, 726 gas penetration rates measurements blasting, 49 gas pressure blasting explosives, 39 caverns stability, 402 gas pressures crater blasting, 58 gas pressurization blasting, 40 mechanism, 47 gassy ground tunnel boring machines, 287 gas venting blast effects, 119 gas wells stimulation blasting, 59 controlled fracturing, 59 gate roadways closure, 530 longwall mining, 525 formation longwall advance mining, 528-529 supports longwall advance mining, 529 resistance, 529 gate roadway stability longwall advance mining, 528-530 gauge cutters tunnel boring machines, 263 geohydrology tunnel supports, 355, 357 geological grading tunnel supports, 358 geology longwall mining subsidence, 767 tunnel supports, 353 engineering interpretation, 356 geophones blast monitoring placement, 98 microseismic activities monitoring, 704 monitoring systems, 705 geophysical measurements, 701 geotechnical factors tunnel boring machines, 284 geotechnology tunnel supports, 353 geothermal project Cornwall, UK monitoring, 30 geotomography rock mass structure, 703 Germany blast monitoring standards, 104 coal mines rock bolting, 536 coal mining subsidence, 759 goafs
subsidence, 762 gold mining methods, 18 subsidence, 755 Goodrich wear number tunnel boring machines performance, 281 Gotthard Road Tunnel steel ribs stress measurement, 680 Gotthard tunnel supports, 353 granite fragmentation modeling, 52 stresses cavern excavation, 640 granodiorite stresses cavern excavation, 640, 641 greases anchor heads corrosion control, 431 rock anchorages corrosion control, 429 Great Pyramid Giza size, 3 Green's function blast monitoring, 101 Grimsel-Oberaar hydroelectric power scheme prestressed concrete linings, 585 grinding Serrouville iron ore picks, 183 ground anchors, 459 capacity safety margins, 612 ground-based imagery thermal, 726 ground characteristic line rock reinforcement, 473 ground control coalmines, 515 ground difficulty index tunnel boring machines penetration rate, 280 ground displacements swelling rock tunneling and decision making, 588 ground falls mechanisms, 617 ground motion blast effects measurement instruments, 120 blast excitation, 115 kinematic relationships, 116 ground reaction curve support design, 334 ground response curve supports, 27-29,314 ground settlements adjacent tunnels prediction, 602 ground support, 28 geology engineering interpretation, 356 tunnel boring machines automation, 289 tunnels clay, 363
831
832
Subject Index
ground surface movements basic parameters, 805 coal mining People's Republic of China, 791 mining inclined coal seams, 797 very thick coal seam, 802 reduction mining, 808 tunneling People's Republic of China, 786 underground excavations People's Republic of China, 781-817 ground tilt subsidence, 764 ground vibration blasting, 88,96 control, 105 measurement transducers, 97 monitoring regulations, 102 production, 96 groundwater adjacent tunnels compressed air, 602 Luossavaara mine, 491 rock anchorages, 416 subsidence, 768 longwall mining, 767 tunnel boring machines, 286 tunneling costs, 309 grouted bolts convergence design, 334 grouting rock anchorages construction, 433-^36 rock reinforcement laboratory testing, 466 grouts resinous rock anchorages, 425 types rock reinforcement, 458 gypsum mining stability, 390,392-393
hammers extension drilling percussive drilling, 151 hanging-wall displacement instrumentation Luossavaara mine, 502-503 drift Luossavaara mine, 500 Luossavaara mine, blasting, 508 instrumentation Luossavaara mine, 502 stability instrumentation at Luossavaara mine, 508-510 support Luossavaara mine, 500-502 hardness tunnel boring machines laboratory tests, 269 hard rock mining microseismic emissions monitoring, 708
haulage force drum shearers evaluation, 215 excavating machines determination, 209 P.C.DRUM program, 211 heads excavating machines specification, 219 heat generation plastic deformation, 717 production mechanisms, 717 heat sources detection, 720 heat transfer detection, 726 crack propagation, 716 heave tunneling field measurements, 573 heave energy blasting, 76 kinetic energy, 77 burden movement velocity, 77 average burden velocity, 78 face velocities, 78 high speed photography, 77 diggability, 77 heterogeneous rock mass excavations failures, 388-391 high speed photography heave energy burden movement velocity, 77 modeling blasting mechanisms, 40 highwall control high-speed photography blasting, 67 holes large mechanically assisted cutting, 247 medium mechanically assisted cutting, 254 small mechanically assisted cutting, 243 Homolite 100 modeling of blasting mechanisms, 46 horizontal displacement coefficient subsidence, 806 Huntly West mine underground monitoring, 732 hydraulic conductivity tunnel supports, 357 hydraulic fracturing oil wells stimulation, 59 stress measurement, 654, 656, 658 hydrocarbons storage cavern stability, 401 hydroelectric power plant underground caverns stability and back analysis, 556 hydroelectric schemes underground excavation, 23 hydroengineering construction, 4 hydrology
Subject Index
tunneling costs, 309 Hydrostone fragmentation modeling, 52
igneous rocks stresses, 640 Imigrantes Highway tunnels, 582 impact tests tunnel boring machines, 270 indentation cutting tools tunnel boring machines, 267 tunnel boring machines, 267 brittle rock, 269 laboratory tests, 269 index tests tunnel boring machines, 269 performance, 276 India coalmines, 514 Indiana limestone kerfs cavitating water jets, 239 induration tunnel supports, 356 infrared radiation infrared thermography, 720 infrared scanners infrared thermography, 721 infrared scanning, 722 thermal properties, 724 infrared thermography failure, 715-730 infiltration, 726 plastic deformation crack propagation, 718 rock failure, 722 technology, 720,721 initiation instantaneous blast monitoring, 74 sympathetic blast monitoring, 74 initiators blasting, 108 electronic blasting, 108 in situ stress ground support, 356 rock mass Luossavaara mine, 488-491 supports, 24 tunnel boring machines penetration rate, 280 tunneling costs, 310 in situ testing underground coal mines, 731-750 instability mechanisms rock reinforcement, 470 instrumentation blast monitoring, 97 deformation monitoring, 621 drilling supports, 354 monitoring rock reinforcement, 480 performance, 480
833
rock mass behavior Luossavaara mine, 502-508 support design observational methods, 336 test panel underground monitoring at Huntly West mine, 735 instruments blast effects number, 123 deployment blast monitoring, 124 integrated measuring technique Gotthard Road Tunnel steel rib stress, 680 linings stress, 674 shear strains, 675, 683 interaction matrix construction, 5 energy flux construction, 12-15 internal friction angle P.CDRUM program, 212 invert arches yielding supports tunneling and swelling rock, 594
jet-assisted cutting excavation, 248,259 tool force reductions bit velocity, 251 jet position, 251 jointed rock deformation back analysis, 552 discontinuous deformation and back analysis, 553 reinforcement, 460 response, 470 shear strength tunnel supports, 361 joint friction fracture dilation blasting, 82 jointing tunneling costs, 308 tunnel supports, 357 joint initiation fracturing blasting, 47 joints blasting compressive strength, 43 density, 487 Luossavaara mine, 486 energy dissipation percussive drilling, 150 extension drilling percussive drilling, 151 flaws effect on blasting, 45 frequency tunnel boring machines, 279 Juossavaara mine, 491 mining stability, 408 opening cavern excavation, 644 percussive drilling modeling, 144
834
spacing tunneling costs, 308 subsidence People's Republic of China, 782 tunnel supports, 356 visual monitoring, 619 jointy rock displacement cavern excavation, 647 stresses cavern excavation, 640
Kaiser effect, 700 Kaiman filter parameter identification method back analysis, 548 karsts formation, 755 kerf cutting high pressure jets, 230 kerfing cavitating water jets, 239 tunnel boring machines, 263 kerogen removal oil shale, 51 kinetic energy percussive drilling computer modeling, 140 modeling, 144 Kokura mine initial stress state measurement, 665 mining plan, 654 multiple-layer mining DDM modeling, 662 stress measurement, 669 monitoring, 654
labor tunneling costs, 299 incentive payments, 300 requirements, 300 laboratory testing rock anchorages, 416-417 laboratory tests tunnel boring machines, 269 abrasivity, 270 drillability, 270 landslides subsidence, 756 LASDIS stress monitoring, 654 laser beam and solar battery system floor displacement measurement, 654 Laubscher system support design, 321 limestone subsidence, 752 tunnel boring machines penetration rate, 274 linings arched design, 649 arched concrete cavern excavation, 648 design, caverns, 648
Subject Index
role, 648 stresses, 638,640 concrete coalmines, 531 stress, measurement, 674 curvature caverns, 648 pressure monitoring, 671-693 stress measurement, 672 support stiffness design, 334 tunnel supports, 354 Litani gallery stability, 391 load measuring rock reinforcement, 480 relaxation test rock reinforcement, 466 loading anisotropic pore pressure, failures, 386 elastic range shear strain measurement, 683 face supports coal mines, United Kingdom, 538 plastic range steel rib stress measurement, 686 plastic state back-calculation limitations, 688 steel rib back-calculations, 687 rates rock reinforcement and testing, 466 loading level Luossavaara mine rock mass behavior, 508 stability Luossavaara mine, 493 support Luossavaara mine, 499 load sharing rock support, 363 load transfer rock reinforcement, 453-454 load transfer mechanisms rock reinforcement, 462 longwall advance mining coal, 516 gate roadway stability, 528-530 longwall faces supports coalmines, 536-541 design, 540 longwall mining chain pillars design, 521-523 coal, 514 coalmining, 157 critical zone, 527-528 gate roadway closures, 525 layouts, 516-518 strata stresses coalmines, 526 subsidence, 752,756 cover rock type, 769 geology, 767 prediction, 760,764 longwall retreat mining coal, 518 Lorenz waterwheel, 15
Subject Index
Lugeon test permeability cavern excavation, 646 lune-shaped flatjack DDM modeling, 662 stress compensation measurement, 654 monitoring, 654 Luossajärvi Lake Luossavaara mine, 491 Luossavaara mine instrumentation rock mass behavior, 502 layout design, 491-502 open stope mining blasting and rock mass response, 485 orebody, 486
machine boring rock loosening, 354 machine excavation types, 20 machine vibration jet-assisted cutting excavation, 254, 259 magnetite Luossavaara mine, 486 manometers in situ measurements coalmines, 737 displacements, 735 Marietta Borer mining machine, 159 marlstones water field measurement in tunneling, 573 masonry tunnel supports, 354 masonry arches tunnel supports, 351 massive rock reinforcement, 460 maximum shear stress measurement blasting, 42 mechanical excavation theory and practice, 177-226 mechanically assisted cutting excavation, 242, 258 mechanically coupled elements continuous rock reinforcement, 457 mechanism concatenations system behavior construction, 11 mechanized excavation energy input rates, 21 metallic coatings sacrificial rock anchorages and corrosion control, 430 microcracking vibratory loading, 718 microcracks detection, 717 percussion drilling crater blasting, 58 micro joints cavern excavation, 645
microphones airblast measurement, 98 blast monitoring placement, 98 microphonics blast monitoring, 99 microprocessors control tunnel boring machines, 288 microseismic activities monitoring, 696,701 monitoring systems, 705 rock failure, 695 microseismic emissions, 698 monitoring, 707 microseismic monitoring stability mining, 403 microseismic processor recorders, 710 milling roadheaders modeling, 191 mineralogy tunnel supports, 357 mines abandoned subsidence, 776 collapses rock mass structure, 404-406 metal abandoned and collapses, 777 minimum rock bolting density hard rock mining, 328 mining coal People's Republic of China, 791 extractions subsidence, 752 inclined coal seams subsidence, 797 methods, 18 Luossavaara mine, 491 subsidence, 806 multiple-layer DDM modeling, 662 naturally supported methods, 18 rock mass behavior Luossavaara mine, 504-510 sequence Luossavaara mine, 492 subsidence, 756 characteristics, 758 computer programs, 810 prediction, 771 support design, 314 very thick coal seam subsidence, 802 mining rock mass rating support design, 321 misfires blast monitoring, 74 Mobile Miner Robbins Co. disc cutter, 169 hard rock applications, 175 hard rock mining, 160 Mode II fracture loading blasting, 47 modeling back analysis, 546 bottom-up synthetic approach construction, 4
835
836
boundary element analyses stress measurement, 662 centrifuge rock reinforcement, 478 DDM lune-shaped flatjack, 662 displacement discontinuity method pressure capsule, 662 equivalent material rock reinforcement, 478 fracture contours blasthole, 80 fragmentation blasting, 75 FSM-DDM pressurized sleeve, 663 ground vibration seed waveform, 88 percussive drilling, 141 physical rock reinforcement, 478 rock cutting heads, 186 rock mass, 546 rock reinforcement, 476 stability tunnel supports, 353 tool-rock interactions, 180 top-down analytical approach construction, 4 models excavating machines laboratory scale tests, 204 subsidence longwall mining, United Kingdom, 764 tunnel supports, 363 validation excavating machines, 204 modulus of elasticity back analysis, 551 plate bearing test back analysis, 544 uniaxial compressive tests back analysis, 544 Mohr diagram mining stability, 396 Mohr's hardness scale tunnel boring machines, 270 monitoring blast vibration, 111-134 construction, 29-32 types, 32 cost stability assessment, 609 deformation, 618 safety margin, 611 displacement, 624 dynamic phenomena technology, 704 failure detection, 617 guidelines deformation rates, 612 Huntly West mine test drive and excavation, 739 instrumentation rock reinforcement and performance, 480 microseismic activities, 704,705 qualitative deformation, 619 rib pressure, 0671-693 rock anchorages, 440-441,443 rock reinforcement
Subject Index
instrumentation, 480 stability assessment, 608 support design observational methods, 336 tunnels failure, 364 types stability assessment, 610 underground coalmines, 732 underground coal mines, 731-750 visual deformation, 619 Mont Cenis Tunnel supports, 353 morphological system definition, 7 multiple spark gap camera blasting mechanisms, 42 Munich subway system field measurements decision making, 600
New Austrian Tunneling Method supports, 314,336,360,362 NGI Tunneling Classification Scheme rock reinforcement, 471 Niagara Falls rock bolt reinforcement, 441 nitroglycerine explosive fracturing oil well stimulation, 59 noise blast effects, 114 North America coal mines supports, 536 coal mining subsidence, 760 Norwegian Institute of Technology tunnel boring machines performance, 277 notching fracture controlled blasting, 67 nuclear waste storage stability, 395 thermal effects, 395 nucléon decay stress measurements, 661 numerical modeling support design, 335 numerical models subsidence prediction, 772
observational design approach stability of underground openings deformation monitoring, 608 oedometer tests swelling rule tunneling and decision making, 592 oil fields pore pressure failures, 387 oil shale kerogen removal, 51 oil wells
Subject Index
stability faults, 406 stimulation blasting, 59 controlled fracturing, 59 thermal effects stability, 396 Opalinus clay tunneling decision making, 591 open joints cavern excavation, 645 open stope mining blasting rock mass response, 485-510 open stopes subsidence, 776 orebody extraction DDM modeling, 662 ore extraction stress monitoring, 653-670 ovalization boreholes failure, 381 overcoring in situ stress measurement coalmines, 744 stress measurement, 672 overpressure blast disturbance, 85 fracture control blasting, 63
pack supports longwall advance mining, 529 parameter interactive intensity construction, 33 partial mining subsidence, 808 particle velocity appropriate measurement blast effects, 121 structural strains blast effects, 121 P.C.DRUM program pick cutting machines, 211 P.C.MAP program excavating machines, 213 penetrating cone fracturing crater blasting, 58 penetration rate tunnel boring machines rock properties, 273 tunneling economics, 304 Penmaenbach tunnel rock anchorages, 448 People's Republic of China ground surface movements underground excavations, 781-817 percussive drilling, 20 computer modeling, 137-153 components, 141 efficiency, 146 end members, 146 energy, 139 hammers, 141 impact, 141 impact velocity, 141 impulse, 139
initial conditions, 141 interface members, 144 momentum, 139 normal force, 139 particle velocity, 139 segment members, 142 validation, 138 computer simulation programs, 148 crater blasting, 58 down-the-hole efficiency, 149 simulation, 137-153 permeability measurement cavern excavation, 646 supports tunnels in clay, 363 tests boreholes, 641 tunneling costs, 309 petroleum drilling stability, 394 Pfaender Tunnel field measurement decision making, 589 Philipp Holzmann measuring device stress, 673 photoelasticity rock reinforcement, 478 physical models subsidence prediction, 773 pick cutting machines computer simulation, 211 picks rock interactions modeling, 180, 195 pillar effect adjacent tunnels compressed air, 604 pillars barrier legal requirements of coal mines, 524 longwall mining, 523-526 room and pillar mining, 520-521 stability of coal mines, 525-526 chain coal mines, 518 loading, 523 longwall mining, 521-523 coal height to width ratio, 519-521 loading of square, 523 strength, 519 strength of rectangular, 521 stresses, 522 coal mines design, 515 rectangular loading, 523 room and pillar mining design, 519-521 strength, 519-521 square longwall mining, 521-523 strength, 18 stresses, 18 pipelines blast effects, 132 tunnel boring machines, 289
837
838
plastic deformation creep tunnel supports, 363 plastic yield tunnel supports, 365 plate bearing tests modulus of elasticity back analysis, 544 PMMA fragmentation modeling, 52 point loads linings, 28 point load tests rock anchorages, 416 f>olycrystalline diamond picks mechanical excavation, 219 polycrystalline diamond compact cylindrical bits, 220 polygonal precast reinforced concrete segments linings caverns, 650 pore pressure excavation failures, 387 failures caverns, 384 ground support tunnels in clay, 363 poroelasticity failures caverns, 384 porosity tunnel boring machines, 271 porous materials failures caverns, 384-387 mining stability, 395 thermal effects on stability, 397-399 porous rock indentation tunnel boring machines, 269 portable microseismic processor recorders field trials, 710 postSplitting controlled fracturing, 59 fracture control, 62 potash mechanized mining, 159 power plants underground stability heterogeneous rock mass, 390 preamplifiers microseismic activities monitoring, 704 precast concrete lining segments supports, 28 presplitting controlled fracturing, 59 fracture control, 62 pressure capsule borehole flatjack, 654 displacement discontinuity method modeling, 662 stress change measurement, 656 stress measurement, 657 pressure cells hydraulic stress measurement, 672
Subject Index
pressure containment device fracture control blasting, 65 pressure gauges blasting modeling, 49 pressuremeters coal stiffness, 743 ground support, 356 pressurized sleeve modeling FSMDDM, 663 prestressed concrete linings pressure tunnels Grimsel-Oberaar's hydroelectric power scheme, 585 primary state variable evolution construction, 11-15 process-response system definition, 7 propagation blast-induced air overpressures, 119 ground motion blast effects, 118 propellant crater blasting, 58 propellants blasting, 40 explosive fracturing, 60 proppant controlled fracturing, 59 protection classification subsidence, 809 pseudo velocity response spectrum ground motion, 126 PS strands linings caverns, 650 P-waves blasting, 42 pyrites swelling mining stability, 394-395
Q classification tunnel boring machines, 272 Q system support design, 322 quality control construction, 29-32 quarries collapses rock mass structure, 404-406 roof collapsing limestone, 405-406 quarry blasting fragmentation, 49 mechanisms, 51, 67 quartz drag tools wear, 168 tunneling costs, 305 quartzite kerfs abrasive water jets, 242 tunnel boring machines penetration rate, 273 performance, 282
Subject Index
radial cracks blasting, 51 system, 43 radial fractures blasting, 55, 56 radial strain compressive blasting, 42 radioactive waste disposal underground, 23 repositories construction, 3 raise boring definition, 161 rock cutting, 155 ravelling rock supports, 362 Rayleigh waves blast excitation, 115 recording blast effects, 122 digital, 122 blast monitoring, 98 regulatory controls blast effects, 129 reinforcement devices axial testing, 466 reinforcing elements combined response mode, 465 large-scale, 459 relaxed zones ceiling rocks cavern excavation, 635 remote sensing rock masses, 703 resin-anchored deformed bars performance monitoring, 481 resins grouts rock anchorages, 425^26 rock anchorages corrosion control, 430 rock reinforcement, 458 response spectrum analysis case histories, 127 blast monitoring, 101 dominant frequency estimation, 118 pseudovelocity ground motion, 126 rhyolite stresses cavern excavation, 640 rib pillars longwall mining coal, 524 rib ratio steel arches design, 340 ribs pressure monitoring, 671-693 rib spalling monitoring Huntly West mine's test drive, 742 ripple marks fracture mechanics blasting, 56 RMR classification tunnel boring machines, 272 penetration rate, 280 roadheaders, 20
axial phase modeling, 190 boom tunneling machine, 159 development, 178 dust levels excavation, 254 heads laboratory scale tests, 205 theoretical model, 200 Huntly West mine test drive, 734 jet-assisted cutting, 250 modeling, 218 models validation, 207 P.C.MAP program, 213 performance determination, 210 roadways coal mines rectangular steel supports, 533-535 rock bolting supports, 535-536 steel supports coal mines, 531-533 roadway supports longwall advance mining, 530 robotic miners small-charge crater blasting, 58 Rocha classification supports design, 326 rock anchorages, 0413^49 cement grouted bond values, 421 cementitious groups, 425 construction, 432-437 definitions, 414 design, 417-428 dynamic responses, 446-448 grout/tendon interfaces, 424-425 installation, 417 length, 458 monitoring, 440-441 on-site acceptance tests, 437-439 proof load-time data, 439 rock/grout interface, 421^23 short term service behavior, 439-440 slenderness ratios, 420 static performance, 443^46 support requirements, 419 tendon corrosion, 428 tensioned, 419 testing, 437^141 untensioned, 419 uplift capacity, 419-421 rock arch spacing support, 316 rock behavior characteristics cavern excavation, 647 rock bolting longwall mining, 516 supports coal mines and roadways, 535-536 tunnel supports, 365 rock bolts cement grouted design, 341 definitions, 415 dimensioning
839
840
stochastic modeling, 316 installation tunnel boring machines, 264 mechanically anchored design, 340 performance monitoring, 481 resin grouted design, 341 supports, 28 tunnel supports, 354 rock boreability index tunnel boring machines, 277 rock bursts, 403 rock chips jet-assisted cutting excavation, 248 rock composition tunneling costs, 305 rock cores curation tunnel supports, 356 rock cutting mechanics, 155-176 specific energy, 163 rock cutting heads modeling, 186 rock decomposition tunneling costs, 307 rock engineering history, 3 rock mass total systems behavior, 9 rock headers tunnels excavation, 354 rock impact hardness tunnel boring machines, 270 rock load estimation early method, 351 support design, 322 rock mass modeling, 546 stiffness tunnel boring machines, 264 total systems behavior construction, 9 rock mass automaton, 10 rock mass behavior cavern excavation, 631-651 mining Luossavaara mine, 504—510 rock mass classification ground support, 356 tunnel boring machines, 272 rock mass classification systems comparison, 328 empirical design supports, 317 rock mass properties cuttability, 175 tunnel boring machines, 271 rock mass rating classification systems excavation cuttability, 175 rock mass rating system mining applications, 321 support design, 320 rock mass structure failures, 404-408 rock mass system support design, 326 rock memory, 700
Subject Index
rock pressure determination strain in tunnel linings, 588 tunneling field measurement, 579 rock properties tunnel boring machines, 267 economics, 294 penetration rate, 273 tunneling costs, 303 rock reinforcement action, 459-465 auxiliary, 458-459 design, 451-481 evaluation, 451-481 hardware, 457-459 instability mechanisms, 470 interaction, 473-474 mechanical interlocks, 462 overstressed zones, 471 performance evaluation, 479-481 monitoring programs, 480-481 permanent, 457 posttensioned, 454-456 pretensioned, 454-456 semipermanent, 457 techniques, 452-457 technology, 451-481 terminology, 452 testing, 451-481 rock salt mechanical characteristics, 717 thermal diffusion, 726 rock stability damage blasting, 79 rock strength models laboratory scale tests, 204 rock structure rating rock support, 324 rock support interaction analysis support design, 334 rock zoning systems tunnel supports, 359 rods elastic waves percussive drilling, 138 extension drilling percussive drilling, 151 percussive drilling modeling, 143 roller cones drilling mechanically assisted cutting, 245 roller cutters excavation, 230 rolling force tunnel boring machines, 268 roof blocks height coalmines, 537 roof falls microseismic warning system, 709 roofs coal mines control and supports, 540-541 roof sinkage measurement, 654 predicted, 666 roof supports
Subject Index
tunneling costs, 308 room and pillar method mining, 18 room and pillar mining abandoned mines subsidence, 776 coal, 157,158,514,515 collapses, 404-406 layouts barrier pillars, 520-521 pillars strength and design, 519-521 sinkholes, 773 subsidence, 754,756 RossivaTs hardness scale tunnel boring machines, 270 rotary disc cutters rock cutting, 155 rotational slip analysis rock reinforcement, 472 roughness rock reinforcement, 462 RQD classification tunnel boring machines, 272 penetration rate, 280 RQD method support design, 326 RSR classification tunnel boring machines, 272 penetration rate, 280 rupture rock reinforcement, 465
safety check field measurement in tunneling, 578 control field measurement in tunneling, 578 tunneling field measurement, 574 underground openings, 610 safety factors crown pillars Luossavaara mine, 496 rock anchorages, 427-428 tunnel supports definition, 360 safety margin deformation monitoring, 611 deformation rates, 623 salt drilling fluids, 394 mechanized mining, 159 salt domes subsidence, 756 sandstone tunnel boring machines, 285 penetration rate, 274 scaled distance ground vibration, 105 scaled volume blasting, 53 schist tunnel boring machines penetration rate, 273 schistose rock tunnel supports, 356 Schmidt hammer rebound hardness tunnel boring machines, 270 Schmidt hammer tests
rock anchorages, 416 Scleroscope hardness tunnel boring machines, 270 Scleroscope hardness test tunnel boring machines performance, 281 Seabrook tunnel tunnel boring machines penetration rate, 273 seam inclination subsidence, 770 seams level in coal mines strata loading, 538 steep strata loading, 538-540 steeply pitching subsidence, 770 sedimentary rocks bedding tunneling costs, 309 stresses, 640 sedimentology subsidence, 754 seismic monitoring system blast damage to hanging-wall Luossavaara mine, 508 seismographs blast monitoring, 99 Selby mines tunnel boring machines penetration rate, 273 settlement tunnel supports, 365 SFCMOV computer program mining subsidence, 811 shafts tunnel supports, 356 shaft sinking mechanical, 161 rock cutting, 156 shales mining stability, 393 tunnel boring machines penetration rate, 274 shear response modes rock reinforcement, 463-465 shear bands deformation wells, 380 shear compression rock failure picks, 181 shearers dust levels excavation, 254 thrust limited evaluation, 216 shear fracture picks modeling, 182 shearing drum shearers evaluation, 215 excavating machines determination, 210 roadheaders modeling, 191 Serrouville iron ore
841
842
picks, 183 stability, 406-408 shear keys, 459 shear modulus back analysis, 547 reduction with increasing strain, 363 shear strains deformation arches, 683 integrated measuring technique, 675,683 shear strength jointed rock tunnel supports, 361 rock mass rock anchorages, 416 tunnel boring machines performance, 281 shear testing laboratory rock reinforcement, 467 rock reinforcement, 465 shielding tunneling costs, 310 shield supports longwall faces coal mines, 540 Shintakase underground power station elastic waves velocity, 641 shock energy blasting, 76 borehole strain, 79 peak particle velocity, 79 charge weight scaling blasting, 80 induced blasting, 78 shock waves blasting, 39 Shore hardness tunnel boring machines, 270 short-period seismograph systems, 707 shotcrete supports tunnel boring machines, 264 tunnel support costs, 307 tunnel supports, 354, 365 shotcrete linings caverns, 650 design, 337 stress measurement, 673, 674 Sievers J value tunnel boring machines performance, 278, 282 silcrete tunnel supports, 356 SIMSUPER5 tunnel boring machines performance prediction, 287 SIMTUN tunnel boring machines performance prediction, 287 simulation percussive drilling, 137-153 progams, 148 single-channel microseismic monitoring, 713 single degree of freedom model blasting structural response, 125 single embedment axial tension test rock reinforcement, 466
Subject Index
sink holes characteristics, 756 collapse solution-mined caverns, 400 occurrence, 754 coal mines, United States of America, 777 room and pillar mining, 754, 773 United Kingdom, 778 site investigation rock anchorages, 415-417 tunnel supports, 355 Skempton's coefficient pore pressure, 385 sleeve fracturing stress measurement, 654 monitoring, 654 sliding deformation cut slopes and back analysis, 559 discontinuities back analysis, 555 slip planes back analysis, 552 sliding micrometers continuous strain measurements swelling and tunneling, 599 hanging-wall Luossavaara mine, 508 measurements back analysis, 551 strain profiles tunneling and decision making, 603 water pressure, 601 swelling tunneling and decision making, 589 slip percussive drilling modeling, 144 rock reinforcement, 465 slip planes sliding back analysis, 552 slope failures, 23 safety margins, 612 slopes instability subsidence, 756 monitoring microseismic activities, 711 stability Imigrantes Highway, 583 rock reinforcement, 472 smectite mining stability, 393 smooth blasting cushioned charges, 63 fracture control, 62 loading density damage, 82 soil densification blast effects, 113 soluble materials mining stability, 392-395 solution cavities subsidence, 752 surface collapses, 752 solution mining stability, 399
Subject Index
sonic velocity tests rock anchorages, 416 sound waves stability mining, 403 spall blasting, 47 fragmentation mechanism, 41 spalling discontinuous plane back analysis, 555 joints back analysis, 552 specific kerfing energy water jet excavation, 233 spectral analysis stability mining, 403 spherical charges cratering, 51 Split set bolts design, 343 squeezing ground excavations support design, 344 stability coal mines barrier pillars, 525-526 strata pressure abutments and longwall mining, 526-528 crown pillar Luossavaara mine, 499 deformation monitoring underground openings, 607-629 evaluation strain energy, 702 gate roadways longwall advance mining, 528-530 Luossavaara mine, 492-493 monitoring, 608 rock mass, 390 rock reinforcement, 470 underground openings back analysis, 548 standards blast monitoring European, 104 stand-up time tunneling costs, 310 steel arches design, 339 H-section supports in coal mines, 532 tunnel supports, 351 V-section yielding supports in coal mines, 533 steel ribs Gotthard Road Tunnel stress measurement, 680 stress measurement, 674 steel supports coalmines, 531-532 rectangular roadways, 533-535 yielding roadways in coal mines, 532-533 stemming retention high-speed photography blasting, 67 stiffness coal
in situ measurement, 732 pressuremeters, 743 underground monitoring, 734 shear rock reinforcement, 463 structural response blasting, 125 stiffness springs rock reinforcement, 476 stochastic modeling rock bolts dimensioning, 316 stowing subsidence, 762 strain coal mines underground monitoring, 733 continuous measurement swelling in tunnel, 599 measuring rock reinforcement, 480 subsidence, 765 tunnel linings rock pressure determination, 588 tunnels back analysis, 547 underground openings stability and back analysis, 548 strain gauges percussive drilling efficiency, 148,150 strain pulses measurement blasting, 41 strap mining subsidence, 808 strata loading face supports coal mines, 537-538 level seams coal mines, 538 steep seams coal mines, 538-540 strata pressure abutments longwall mining stability, 526-528 strata pressure abutment zone longwall retreat mining, 518 stratified deposits reinforcement, 315 stratified rock reinforcement, 460 strength coal, 733 point-load support design, 326 stress anisotropic excavation, 381 lateral, 380-382 arched concrete linings, 638 cavern excavation, 640 barrier pillars coal mines, 526 change displacement measurement, 626 determination reliability, 658 discontinuous rock reinforcement, 470 distribution in situ monitoring in coal mines, 747
843
844
dynamic mechanisms, 402 excavation rock mass structure and failures, 404—408 hydraulic fracturing measurement, 654 initial back analysis, 551 cut slopes and back analysis, 562 ground field measurement, 573 in situ measurement coal mines, 734,736,738,744 linings monitoring, 671 measurement hydraulic fracturing, 656 linings, 672 methods, 654 stress relief, 671 monitoring Hundy West mine's test drive, 740 ore extraction monitoring, 653-670 redistribution, 616 reinforcement caverns and measurement, 634 safety margin monitoring, 611 tunnel face supports, 363 underground excavations supports, 314 stresscells in situ measurement coalmines, 738 in situ stress measurement coalmines, 744 monitoring Hundy West mine's test drive, 740 stressing rock anchorages, 436-437 stressing head rock anchorages, 427 stress patterns blasting, 42 stress relief stress measurement, 671 stress/strain models tunnel supports, 358 stress waves blasting, 40, 56 mechanisms, 41 strip packing subsidence, 762 structural mapping density, 487 structural response blast effects, 113 blasting, 114, 124 frequency effects, 124 structural strains particle velocity blast effects, 121 subsidence abandoned mines, 776 bedrock conditions, 767 caving, 18 ceiling rocks cavern excavation, 635, 647 characteristics, 754 development curve, 763 elementary basin
Subject Index
basic equation, 783 empirical prediction methods, 760 extent of influence, 763 faults, 769 goafs, 762 limit angle, 763 longwall mining cover rock type, 769 geology, 767 prediction, 760,764 mining, 522 characteristics, 758 computer programs, 810 prediction, 771 natural environment, 752 occurrence, 752 People's Republic of China, 782 prediction, 754,765 rock structures, 751-780 strain, 765 time coefficient, 806 tunneling People's Republic of China, 789 subsidence basin Fushun-type, 802 subsidence coefficient, 807 subsidence engineering, 754 Subsidence Engineers' Handbook UK Coal Industry, 760,764 subway tunneling decision making, 600 sumping continuous miners model validation, 207 excavating machines determination, 210 roadheaders modeling, 191 superstructure natural frequency dynamic response to blasting, 125 supports artificial safety margins, 612 coalmines, 513-541 coalmining, 514-515 coal mining tunnels, 530-536 construction, 22-29 definition formal, 350 incremental, 350 informal, 350 primary, 350 secondary, 350 design geological data, 354 permanent, 315 rational methods, 333 stand-uptime, 326 structurally controlled environments, 315 swelling excavations, 344 temporary, 315 underground excavations, 313-345 development, 354 effectiveness deformation rates, 623 excavation stability, 4 0 8 ^ 0 9 gate roadways longwall advance mining, 529 resistance, 530 loading
Subject Index
coalmines, 538 longwall faces coalmines, 536-541 mining microseismic emissions and monitoring, 709 'natural', 23-27 objectives, 22 rock bolting coal mine and roadways, 535-536 rock reinforcement, 452 roof control coalmines, 540-541 soft, 27 steel coal mine roadways, 532-533 coalmines, 531-532 stiff, 27 strength safety margin and monitoring, 611 tunneling field measurement, 575 tunnels, 349-367 costs, 301 yielding, 27 support systems Austrian, 351 Belgian, 350 design, 337 English, 351 German, 350 traditional, 350 surface exposures tunnel supports, 356 surface mine blast response spectrum analysis, 127 surface mines monitoring, 99 surface mining microseismic activities monitoring, 711 surface strain gauges blasting modeling, 49 surface topography subsidence, 770 S-waves blasting, 44 Sweden blast monitoring standards, 104 Swedish brittleness test tunnel boring machines, 270 performance, 278 Swellex bolts design, 343 swelling continuous strain measurements sliding micrometers, 599 mining stability, 392-395 rock mass rock anchorages, 417 tunnels decision making, 588 swelling ground excavations support design, 344 swelling pressure tunneling field measurement, 579 swelling rock characteristic line
845
tunneling and decision making, 592 tunneling constructive countermeasures, 594 decision making for ground displacements monitoring, 588 synthetic definition, 4 system behavior mechanism concatenations construction, 11 system response analyses construction, 9-15 systems approach construction, 5
T8 tunnel yielding supports, 595 Taber abrasion hardness tunnel boring machines, 270 performance, 281 tailored pulse loading blasting, 40 controlled fracturing, 60 tangential strain tensile blasting, 42 tangent of main influence angle subsidence, 805 tape recorders blast effects monitoring, 123 tectonics activity subsidence, 754 tunnel supports, 353 temperature blasting stress wave energy, 41 tendons anchorages definitions, 415 apparent free length rock anchorages, 439 rock anchorages, 426-427 construction, 433-434 Tennessee marble pick-tool interactions, 180 tensile failure rock reinforcement, 471 tensile strength blasting strain levels, 41 tunnel boring machines performance, 281 tensile stress blasting, 49 tension rock reinforcement laboratory axial testing, 466 test drive Huntly West mine excavation monitoring, 739 test panel Huntly West mine underground monitoring, 734 instrumentation underground monitoring at Huntly West mine, 735 thermal conduction
846
microcrack detection, 719 thermal dissipation damage evolution, 724 thermal effects solution-mined caverns stability, 395-399 thermal properties infrared scanning, 724 thermoelasticity microcrack detection, 719 thermomechanical coupling crack propagation, 716 thermoporoelastic effects mining stability, 397 time subsidence, 752 time delay extraction subsidence, 808 time dependence ground support tunnels in clay, 363 time-dependent behavior tunnel supports, 363 TNT blasting, 41 tool flexion modeling, 183 tool-rock interactions modeling, 180 tool wear excavation rate, 230 toppling deformation cut slopes and back analysis, 559 transducers blast effects attachment, 122 measurement, 121 blast monitoring, 97 microseismic activities monitoring, 704 transport tunneling costs, 298 transversal cutting modeling, 202 transverse pillars displacement instrumentation Luossavaara mine, 502-503 instrumentation Luossavaara mine, 502 trepanner coalmining, 157 trial adits tunnel supports, 356 tungsten carbide drag picks, 161 drag tools, 167 tungsten carbide cutters tunnel boring machines, 263 tungsten carbide picks diamond picks comparison, 224 excavating machines, 224 TUNNEL computer program mining subsidence, 811 Tunnel and Reservoir Plan tunnel boring machines penetration rate, 273 tunnel blasting
Subject Index
current procedure, 59 tunnel boring machines advance rate, 266 capabilities, 288 coalmining, 156 costs mucking systems, 309 performance, 294 rock properties, 303 sealing, 307 shielding, 308 cutterhead, 262 fatigue, 262 stiffness, 262 vibration, 262 cutting mechanics, 279 cutting coefficient, 268 cutting cycle, 160 discontinuities, 273 economics advance rate, 294 cutterspacing, 294 rock properties, 293-311 support systems, 294 tunnel driving, 294 gas detection, 264 introduction, 178 laboratory tests, 269 mechanically assisted cutting, 247 modeling, 192 new rock surfaces, 18 penetration rate, 266 rock mass characterics, 279 rock properties, 273 side wall gripper system, 275 torque, 275 performance, 261-291 intact rock, 275 performance parameters, 266 performance prediction, 287 rock cutting, 155 roof shield, 307 shielded, 262,264 shields theoretical model, 202 stand-up time, 272 supports installation, 264 system description, 262 systems developments, 288 thrust systems, 263 torque system, 263 utilization, 266 rock properties, 283 tunnel face stress, 360 tunneling constructional method field measurement, 580 cost effective, 303 costs depth of cover, 310 in situ stress, 310 support materials, 301 decision making, 577 field measurements, 571-606 fuel costs, 298 geology, 354 ground surface movements People's Republic of China, 786
Subject Index
holistic approach, 354 labor costs shifts, 300 machines P.C.MAP program, 213 methods field measurements, 575 plant availability, 298 utilization, 298 plant costs, 295 spare parts, 298 plant hire, 295 rock conditions field measurements, 572 site management costs, 301 subsidence People's Republic of China, 789 tunnel boring machines performance, 294 tunnels clay ground support, 363 coal mines supports, 530-536 collapse mechanisms, 369-409 linings curvature, rock pressure determination, 588 design, 336 lining segments costs, 301 shallow stability, back analysis, 556 stability strain, back analysis, 552 strain distribution back analysis, 547 swelling decision making, 588 yielding supports design, 595 tunnel supports design, 357 geology, 353 multiple drifts, 354 organizational aspects, 366 philosophy, 349-367 practical aspects, 365 procedural aspects, 366 stress-strain compatibility, 365
ultrasonic pulse velocity crack detection, 717 Ultrasonic Spot Coming Roof Failure instrumentation, 709 underground excavations collapse mechanisms, 369-409 ground surface movements People's Republic of China, 781-817 support design, 313-345 underground mining microseismic emissions monitoring, 708 underground openings dimensions field measurement, 574 safety, 610 shapes
847
field measurement, 574 stability back analysis, 548 deformation monitoring, 607-629 structural behavior, 572 yielding safety margins, 612 underground workings monitoring, 99 undersea mining subsidence, 770 undertunneling old houses field measurement decision making, 602 uniaxial compressive strength P.C.DRUM program, 212 rock mass Luossavaara mine, 488 tunnel boring machines, 270 uniaxial compressive test modulus of elasticity back analysis, 544 United Kingdom coal mines face supports and loading, 538 coal mining subsidence, 758 sinkholes, 778 subsidence in longwall mining empirical model, 764 United States of America National Committee on Tunneling Technology tunnel boring machines and rock properties, 283 sink holes coalmines, 778 US Bureau of Mines blast monitoring regulations, 102
validation models excavating machines, 204 percussive drilling modeling, 148 vanes cylindrical drums design, 187 vane sections drum shearers evaluation, 214 P.CDRUM program, 211 vane spirals drum shearers modeling, 195 vapor bubbles cavitating water jets excavation, 239 velocity exposure level blast monitoring, 102 velocity gauges microseismic activities monitoring, 704,710 velocity of detonation gauges blasting modeling, 49 velocity transducers blast monitoring, 97 venting airblast control, 107
848
vibration blasting, 20 continuous miners model validation, 206 damage blasting, 79 drum shearers evaluation, 214 excavating machines determination, 209 economics, 223 modeling, 198, 199, 217 P.C.DRUM program, 213 specification, 219 mechanical excavation harsh environment, 219 pick cutting machines computer simulation, 211 picks modeling, 218 roadheaders model validation, 207 vibration control use of delays, 105 vibration gauges blast monitoring, 72 vibration impulses blast monitoring, 72 vibration monitoring system blast damage of hanging wall Luossavaara mine, 508 vibration time history recording blast effects, 123 vibratory densification soil blast effects, 113 vibrothermography microcrack detection, 723 video recordings heave energy burden movement velocity, 77
walls blast excitation ground motion, 115 deformation, 636 failures solution-mined caverns, 401 stability rock anchorages, 443 stress monitoring, 654 stress measurement, 654 structural response blasting, 124 vibration blasting, 126 water collapse chimneys, 775 tunneling field measurement, 573 water cannons excavation, 237 water conductivity tunneling decision making, 589 water jet assistance tunnel boring machines, 288 water jet cutting
Subject Index
fracture controlled blasting, 67 water jet picks assessment, 221 water jets excavating machines economics, 224 excavation abrasive, 241,257 cavitating, 239,257 cavitation bubbles, 239 continuous, 230,257 discontinuous, 234,257 flow rate modulators, 237 high pressure, 233 interrupted continuous, 237 high pressure mechanical cutting tools, 242 picks mechanical excavation, 219 rock cutting, 21 use in rock excavation, 229-259 waveforms measurements blast monitoring, 100 wave propagation theory blasting, 41 weaknesses failure, 615 wear drag tools gross fractures, 169 mechanism, 167 drum shearers evaluation, 215,216 excavating machines determination, 211 economics, 223 P.C.DRUM program, 212,213 specification, 219 picks water jets, 222 weathering rock mass rock anchorages, 417,423 subsidence, 752 tunnel supports, 356 wedge bolts discrete frictionally coupled devices, 458 wedge failure safety margins, 612 tunnels, 365 wedges stability, 315 wells stability, 372 wind blast effects, 119 wire extensometers displacement in situ measurement, 735 Luossavaara mine, 502
Yanahara mine initial stress state measurement, 665 stress measurement, 668 monitoring, 653 yield condition
Subject Index
loading steel rib stress measurement, 686 yielding supports invert arches tunneling in swelling rock, 594 swelling
tunneling, 600 tunnels design, 595 yield zones deformation rates, 623
849
