Use of Timber on Multi-Storey Building

Structural Engineering Documents

13 Use of Timber in Tall Multi-Storey Buildings

International Association for Bridge and Structural Engineering (IABSE)

About the Authors:

Dr. Ian Smith is Lifetime Professor Emeritus of Structural Engineering at the University of New Brunswick in Canada, where he leads a research group in hybrid construction. He holds Doctor of Philosophy and Doctor of Science Degrees from London South Bank University in the United Kingdom.

Dr. Andrea Frangi is Professor for Structural Engineering at the Department of Civil, Environmental and Geomatic Engineering at ETH Zurich, where he leads the research group of Timber Structures. He received his diploma in civil engineering and his doctoral degree from ETH Zurich.

With Contributions From: G.C. Foliente, R.H. Leicester, S. Gagnon, M.A.H. Mohammad, C. Ni, M. Popovski, A. Asiz, A. Ceccotti, A. Polastri, S. Rivest, B. Kasal, D. Kruse, E. Serrano, J. Vessby, J. Bonomo, H. Professner.

Structural Engineering Documents

13 Use of Timber in Tall Multi-Storey Buildings

International Association for Bridge and Structural Engineering (IABSE)

Copyright © 2014 by International Association for Bridge and Structural Engineering All rights reserved. No part of this book may be reproduced in any form or by any means, electronic or mechanical, including photocopying, recording, or by any information storage and retrieval system, without permission in writing from the publisher. ISBN 978-3-85748-133-8 Publisher: IABSE c/o ETH Zürich CH-8093 Zürich, Switzerland Phone: Fax: E-mail: Web:

Int. + 41-44-633 2647 Int. + 41-44-633 1241 [email protected] www.iabse.org

Table of Contents

1

2

3

Introduction

1

1.1 Historical use of timber for construction 1.2 Modern renaissance of timber as a construction material 1.3 Other chapters

1 5 6

Structural Design Issues

9

2.1 2.2

Introduction Design practices and assumptions 2.2.1 Load combinations, load factors, and resistance factors 2.2.2 Achieving an elastic response and allowance of damage 2.2.2.1 Factors to consider 2.2.2.2 Recommended design practices 2.2.3 Analysis methods 2.3 Effect of superstructure shape and height 2.4 Importance of horizontal diaphragms 2.5 Acceptable risk levels and avoidance of disproportionate damage 2.5.1 Risk 2.5.2 Mitigating damage potential 2.6 Podium and other constructed systems with articulated dynamic responses 2.7 Additional comments

9 11 12 13 13 15 18 18 20 21 21 21 23 24

Fire Design Concepts

25

3.1. 3.2. 3.3. 3.4. 3.5. 3.6.

25 26 27 28 30 36 38 38 39 40 40

Introduction Fire action Fire safety objectives and strategy Fire resistance of structural timber elements Design model for the verification of the separating f unction Fire design concept for tall timber buildings 3.6.1 Main differences between mid-rise and tall buildings with regard to fire safety 3.6.2 Is it still possible to design a tall building using timber as structural material? 3.7. Example of tall building project 3.8. Experimental studies 3.8.1 Fire performance of timber structures under natural fire conditions

4

5

6

7

3.8.2 Resu Result ltss of sprinklere prinklered d fire tes tests 3.8. .8.3 3 Re Resu sult ltss of non-s non-sprinklere prinklered d fire tes tests 3.9. Add Additional itional comm mment entss

41 41 44

Durability Design Concepts

45

4.1 Intro Introduc duction tion 4.2 State-of-the-art 4.3 4. 3 Attac Attack mechani hanisms sms 4.3 4. 3.1 Mou Mould 4.3 4. 3.2 Dec Decay 4.3 4. 3.3 Ter erm mite itess 4.3 4. 3.4 Corros Corrosion 4.4 Des Design strategie trategiess 4.4.1 Non-s Non-str truc ucttural elem element entss 4.4.2 Non-c Non-criti riticcal str truc ucttural elem element entss 4.4.3 4.4. 3 Critic Critical str truc ucttural elem element entss 4.5 4. 5 Cal Calcu culation lation of engineered engineered servi ervicce life 4.5 4. 5.1 Aus ustralian tralian approac approach 4.5 4. 5.2 Example cal alcu culation lation 4.6 Add Additional itional comm mment entss Acknowle knowled dge gem ment entss

45 46 48 48 48 50 50 51 51 52 52 53 53 54 55 56

Timber Frameworks with Rigid Diaphragms: Special Considerations

57

5.1 5.2 5.3 5.4 5.5

57 58 61 64 69 69

Introduc Intro duction tion Usef ul less lesson onss fro from m low-ri low-risse tim timber con onsstr truc uction tion (c (cir ircca less less than than 20 m tall) Mod Mo dern renaiss renaissan ancce tall tim ti mber fram frame sys syste tems ms ( (ccir ircca 20–80 m tall) Effective conne Effec onnecction metho ethods ds Add dditional itional comm mment entss Acknowle knowled dge gem ment entss

Steel or Reinforced Concrete Frameworks with Timber Diaphragms: Special Considerations

71

6.1 Intro Introduc duction tion 6.2 Mass Massive ive ti tim mber diaphrag iaphragms ms for for compo possite hy hybri brid d sys syste tems ms 6.3 6. 3 Twenty Twenty-fo -fou urr-sstore torey y case stud udie iess 6.3 6. 3.1 Scope and and metho ethods ds 6.3 6. 3.2 Cas Case study udy re resu sult ltss 6.3 6. 3.2.1 Struc Structtural steel fram framework sys syste tems ms 6.3 6. 3.2.2 RC fra fram mework sys syste tems ms 6.4 General impli pliccation ationss of us using ing CL CLT T slab labss 6.5 6. 5 Add dditional itional comm mment entss Acknowle knowled dge gem ment entss

71 72 72 72 76 76 78 81 82 83

Platform Construction Using Timber Plates: Special Considerations

85

7.1 7.2

85 86 86 87 88 89 92

Introduc Intro duction tion CLT CL T as str truc ucttural material 7.2.1 General chara haraccteri terissti tics cs 7.2.2 Ty Typi piccal design properties properties 7.3 7. 3 Platform Platform con onsstr truc uction tion con onccept 7.4 Connec Connection metho ethods ds 7.5 Str 7.5 Struc ucttural analys analysiis an and d design

7.5 7. 5.1

7.6

7.7

8

92 92 92 94 95 97 98 100 100 100 101 106 107

Example Project 1: Six-Storey Hybrid Building in Quebec City, Canada

109

8.1 8.2

109 110 110 112 114 116 116 117 118 119 120 120 121 122 123 12 3

8.3 8. 3

8.4 8.5 8. 5

8.6

9

General aspe peccts 7.5 7. 5.1.1 Bas Basis of analys analysiis an and d design 7.5 7. 5.1.2 Load Load path pathss an and d rob robus ustne tness ss 7.5 7. 5.1. .1.3 3 De Dessign of floors floors 7.5 7. 5.1.4 Des Design of walls walls 7.5 7. 5.1. .1.5 5 De Dessign of conne onnecction tionss 7.5 7. 5.2 Expe peccte ted d perfor perform man ancce du during ring sei eism smiic event eventss 7.5 7. 5.3 De Dessign man anu ual alss Example of sei Ex eism smiic design prac practi ticces 7.6.1 Bac Backgro kgrou und 7.6.2 Seven-s Seven-store torey y case study Additional Add itional comm mment entss Acknowle knowled dge gem ment entss

Bacckgro Ba kgrou und Super Su persstr truc uctture sys syste tem m 8.2.1 Desc Description ription and and con onsstr truc uction tion 8.2.2 Glu Glula lam m fra fram mework and and diaphrag iaphragms ms 8.2.3 8.2. 3 Ti Tim mber conne onnecction metho ethods ds Struc Str ucttural Des Design 8.3 8. 3.1 General aspe peccts 8.3 8. 3.2 Projec Project spe peccifi ificc con onssideration erationss 8.3 8. 3.3 Analys Analysiis metho ethod d an and d design resu result ltss Fire design Measu Mea sure rem ment of the bu buil ild ding res respon ponsse 8.5 8. 5.1 Differential move ovem ment entss 8.5 8. 5.2 Vibration re resspon ponsse Additional Add itional comm mment entss Acknowle knowled dge gem ment entss

Example Project 2: Fire Fire Design of a Seven-Store Seven-Storey y Hybrid Building in Berlin, Germany 9.1 9.2 9.3 9. 3 9.4

9.5 9. 5

Bacckgro Ba kgrou und Description Desc ription of the bu buil ild ding su super persstr truc uctture Fire compart partm mentalization of the bu buil ild ding Detailed a Detailed asspe peccts of the design 9.4.1 Floor slab labss 9.4.2 Critic Critical elem element ju junction tionss 9.4.3 9.4. 3 Gravity Gravity loa load d sys syste tem m 9.4.4 Cavity Cavity fire firess an and d tran transm smiiss ssion ion of hot gas gases Add dditional itional comm mment entss Acknowle knowled dge gem ment entss

10 Example Project 3: Limnologen Limnologen—Block —Block of Four Eight-Storey Residential Buildings in Växjö, Sweden 10.1 Ba Bacckgro kgrou und 10.2 Arc Archite hitecctural design 10.3 10. 3 Str Struc ucttural design 10.3 10. 3.1 Wall ele elem ment entss 10.3 10. 3.2 Floor elem element entss

125 12 5 125 125 125 12 5 127 129 129 131 132 133 133 134

135 135 136 137 137 138

10.4 10.5 10. 5 10.6

10.7

10.8

10.3.3 Lateral loa 10.3 load d design Fire design Acous usti ticcal design Protection of elem Protec element entss an and d con onsstr truc uction tion of bu buil ild ding ingss 10.6.1 Mois Moisture and and weather protec protection 10.6.2 Cons Constr truc uction tion of bu buil ild ding ingss Resear Res earcch stud udie iess 10.7.1 Measu Measure rem ment entss of vertic vertical settle ettlem ment 10.7.2 Tim Time study udy on on ins installation of load load-bearing elem element entss Additional Add itional comm mment entss Acknowle knowled dge gem ment entss

11 Example Project Project 4: Björkbacken, Björkbacken, a 10-storey 10-storey hybrid building building in Stockholm, Sweden 11.1 11.2 11.3 11. 3 11.4 11.5 11. 5 11.6 11.7

Bacckgro Ba kgrou und Super Su persstr truc uctture con onccept Fire compart partm mentalization Vertical load Vertic load re ressisting sys syste tem m Lateral loa load d re ressisting sys syste tem m Constr Cons truc uction tion of bu buil ild ding Additional Add itional comm mment entss Acknowle knowled dge gem ment entss

12 Looking to the Future 12.1 Likel Likely y li lim mit itss on heights heights of mu multilti-sstore torey y su super persstr truc uctture sys syste tems ms 12.1.1 Lightweight ti tim mber plate ass asseemblie bliess 12.1.2 Mass Massive ive tim timber plate ass asseemblie bliess 12.1.3 12.1. 3 Heavy Heavyweight tim timber-fra ber-fram med a ass sseemblie bliess 12.1.4 Hy Hybri brid d / compo possite ass asseemblie bliess 12.2 Ex Example of propos proposed sys syste tems ms:: LifeCycle Tower con onccept 12.3 12. 3 Refocus Refocusing ing design codes 12.3 12. 3.1 General req requ uire irem ment entss 12.3 12. 3.2 Tim Timber str truc ucttural design 12.3 12. 3.3 Ti Tim mber fire design 12.4 Final comm mment entss Acknowle knowled dge gem ment entss

13 References

139 139 140 140 140 140 142 143 14 3 145 14 5 146 146

147 147 148 150 150 151 151 153 153

155 155 156 158 159 161 163 16 3 167 167 168 169 169 169

171

Preface

Much has been written in the last few decades about the relative merits of alternative materials for building construction. As part of such efforts, this Structural Engineering Document (SED) provides guidance to engineers on how to properly design multi-storey buildings that incorporate timber and timber-based products as superstructure elements. The scope encompasses traditional systems for buildings up to 10 storeys made from conventional timber products and innovative systems that employ modern timber-based composites, as well as emerging possibilities for using timber elements in very tall buildings. Poor building performance is usually accompanied by a failure to integrate design across all aspects of a project; or a failure to link design concepts with the realities of local construction and maintenance practices. For example, if timber elements are not properly protected from wetting (i.e. more than occasionally wetted at rates that exceed ambient drying rates), they are unlikely to be durable. However, if they are protected adequately, timber elements are likely to retain their initial properties for centuries. This document emphasises attainment of Total Performance Goals on a cradle to grave basis, taking account of structural and non-structural considerations. In the contemporary parlance, structural design decisions must support attainment of Total Performance Goals from cradle to grave. Even though the lifespan of most buildings are indeterminate at the time of their conception, their design and construction must address issues like capability of the fabric to retain integrity up to and beyond the likely lifespan and eventual dismantling. The intended audience for this SED is structural engineering practitioners, construction professionals, academic researchers, code drafting bodies, and students. However it is hoped that there will be ancillary audiences amongst architects, property developers, town planners, and governmental policy makers. Ian Smith Andrea Frangi

1

Chapter

1

Introduction

Summary: Since the dawn of civilization, timber has been a primary material for achieving great structural engineering feats. Yet during the late 19th century and most of the 20th century it lost currency as a preferred material for construction of large and tall multi-storey building superstructures. This Structural Engineering Document (SED) addresses a reawakening of interest in timber and timber-based products as primary construction materials for relatively tall, multi-storey buildings. Emphasis throughout is on the holistic addressing of various issues related to performance-based design of completed systems, reflecting that major gaps in knowhow relate to design concepts rather than technical information about timber as a material. Special consideration is given to structural form, fire vulnerability, and durability aspects for attaining desired building performance over lifespans that can be centuries long. This chapter discusses the historical use of timber as a high-performance construction material and lays the groundwork for detailed discussion of modern practices and possibilities in other chapters.

1.1

Historical use of timber for construction

Evidence has been found that in Neolithic China the pre-human species “Peking Man” constructed “nest residences” from branches and thatch. Earth was compacted around thick timber struts, and it is speculated that this was to prevent them from catching fire [1]. Although the practices were crude, this arguably means that timber engineering (structural use of timber) and fire engineering (control of fire risk) were born between 300 000 and 1 million years ago and predate humans. Similarly, carpentry skills that are the basis of modern ability to interconnect timber members have ancient origins. Stone Age people created load-bearing building systems that interconnected timbers using mortise-and-tenon joints that are the direct forerunner of traditional Chinese architecture [1]. From antiquity onwards, urban utilization of construction materials has been shaped by their fire performance when assembled into buildings. City-wide or district-wide conflagrations were the impetus for prescriptive building regulations that date back to the Roman Empire [2]. More modern catastrophes like The Great Fire of London in 1666 and The Boston Fire in 1872 have reinforced fear of urban fires, and many specific building code restrictions created between 17th and 19th centuries are recognizably alive today in some jurisdictions (e.g. not allowing timber buildings to have more than four storeys above ground). Historical building regulations

2

CHAPTER 1. INTRODUCTION

and material prohibitions were enactments to mitigate the extent of destruction that could be wrought by invading enemies, as well as safeguards against disproportional consequences of accidental events. Modern knowledge in fire-proofing (as well as modern firefighting techniques) precludes the likelihood of the uncontrollable fires of previous millennia. However, until the latter part of the 20th century societies have been slow to adapt to changes in fire technologies and threat levels, but now that they have adapted, a new era in timber construction has commenced. During the last 50 years, fire performance and safety of buildings has become the subject of systematic study, and fire engineering has developed as a science that can be applied robustly in practice. Contemporaneously, fire detection and suppression methods have been developed to an extent where it is no longer necessary to simply contain localized fires within non-combustible compartments long enough for building evacuation or arrival of the firefighters. Technical measures available today can detect and extinguish fires quickly using a wide range of devices. In practical terms, modern fire engineering and available technical measures result in mitigation of fire damage to the extent that large buildings can be easily repaired after fires, and minimise the possibility of fire spreading from one building to another. Building codes and governmental authorities are starting to retreat from blanket prohibitions of timber materials for certain uses. Additionally, prescriptive building regulations have been replaced by explicit Building Performance Outcome (BPO) requirements (e.g. in Australia, Canada, New Zealand, and Switzerland). BPO codes recognise that satisfactory fire performance can be achieved in circumstance-specific ways, and that previous “one size fits all needs” approaches are not necessarily ideal. Such codes embody principles, objectives, and quantitative performance-related technical limitations that are independent of the construction materials employed. In Canada for instance, there is now no material-specific limitation on building heights as long as a building’s design meets required objectives in accordance with applicable regulations. Although the BPO approach extends beyond fire aspects of building design, it has so far been the most effective mechanism to overcome outdated and overly restrictive codes and practices related to timber. Despite nature designing trees to decompose once they are dead, structures like the Yingxian Wooden Pagoda in China and the Gümmenen Bridge in Switzerland demonstrate that decomposition can be delayed indefinitely, if structures are appropriately designed, constructed, and maintained (Fig. 1.1). One of the most prestigious examples of this is the pagoda of the Horyu ji Buddhist temple in Japan, arguably the world’s oldest timber building. One of its Japanese cypress posts is thought to have been felled in 594 AD, thus that structure might actually be around 1400 years old. Its longevity reflects that if timbers are cut and dried soon after trees are felled and then kept dry in service, they are unlikely to experience extensive damage from biological attacks by decay fungi or other destructive agents (Chapter 4). Ancient timber buildings owe their longevity to people who constructed them understanding the characteristics of timber and employing system level design concepts that protected the primary structural elements from durability threats. Similarly, if longevity is a goal of modern design then timber should be shielded from direct or indirect wetting, especially if the timber is a portion of a critical structural load path. Simple measures like large roof overhangs, draining water away from foundations, and providing barriers to capillary wetting are highly effective [2,3]. In relatively modern times it has been a practice in some countries to treat exposed timbers using creosote or synthetic chemicals. However, although durability is improved by introducing these toxic materials, the serviceable lifetimes of wetted timbers remain finite. Furthermore, the

3

1.1 HISTORICAL USE OF TIMBER FOR CONSTRUCTION

(a)

(b)

(c)

Fig. 1.1: Durable historical timber structures: (a) Yingxian Wooden Pagoda (courte sy of Dr. Lin’an Wang, China National Institute of Cultural Property) From Lam & He SEI 2, 2009; (b & c) Gümmenen bridge over the river Saane close to Mühlenberg, Switzerland (length about 100 m, built in 1555 and still used)

leaching of these chemicals from timbers can poison groundwater. Consequently this approach is being increasingly banned. The philosophy espoused in t his SED is that engineers should not use timber for structural purposes if it cannot be kept dry. The meaning of dry can be ambiguous. As used here the meaning corresponds to Service Class 1 of Eurocode 5 or similar definitions, i.e. “... characterized by a moisture content in the material corresponding to a temperature of 20°C and the relative humidity of the surrounding air only exceeding 65 percent for a few weeks per year” [4]. As discussed in detail in Chapter 4, there can be valid durability design reasons for inoculating structural timbers with low dosages of synthetic preservatives, with combating termite attacks being the most common one. In the 1940s Howard Hughes and his engineers designed, constructed, and flew the H-4 Hercules single hull flying boat constructed from laminated birch. The erroneously named “Spruce Goose” was designed to carry 750 troops, is 24.18 m high, 66.65 m long, and has a wingspan of 97.54 m [5], making it approximately the same size as the Airbus A380 (the world’s largest airliner at the beginning of the 21st century). That the H-4 Hercules could fly and resist enormous internal force flows generated during takeoff and landing is irrefutable. Furthermore, aeroplanes were not the first large high-performance timber structures. More than 5000 years before around 3500 BC enormous timber sailing ships were common sights in Egypt and Mesopotamia [2]. The most famous was the three-mast Syrakosia of Alexandria, thought to have been about 70 m

4


Year of construction

Name and location

Number of storeys

Total height (m)

1934 (destroyed 1945)

Muhlacker Radio Tower, Germany

N/A

190

1935

Gliwice Radio Tower, Poland

N/A

118

2003

Sa˘pânt¸a-Peri Church, Romania

N/A

75

1056

Sakyamuni Pagoda, Yingzian, China

9

67

1720 (moved 1802)

Bârsana Monastery, Romania

N/A

62

1942–1943

Tillamook Hanger, OR, USA

N/A

58.5

1942–1943

Hangers 2&3, Moffit Field, CA, USA

N/A

52

1906

Claremont Hotel, Oakland, CA, USA (Tower)

10 + Cupola

49

1709

Daibutsu-den, Todaiji Temple, Japan

N/A

48.6

1992 (demolished 2008) Sutyagin House, Russia

13

44

1890

St. Georges Anglican Catholic Church, Guyana

N/A

43.5

1893 (destroyed by fire 1995)

St. Paulus Church, San Francisco, USA

N/A

43.5

1921

Lattice frame industrial building, Cardona, Spain

N/A

32.4

2008

Stadthaus, London, UK

9

30

Started 1992

Tennessee Tree House, USA

10

29.5

Table 1.1: Tallest man-made timber structures

long, which was commissioned by the tyrant Hieron and superintended and launched by the mathematician Archimedes [6]. There are also many examples of historic wood structures that towered over the landscape. Built in 1056, China’s 67.13 m tall Sakyamuni Pagoda was the world’s tallest building for several centuries and has withstood many earthquakes [7]. Currently the tallest timber structure is the 118 m high Gliwice Radio Tower in Poland that was constructed in 1935 ( Fig. 1.2). Table 1.1 summarises the tallest known historical and large man-made timber structures. For further information, Langenbach [8] provides additional details of many of the li sted structures, and Foliente [9] provides an engineering overview of the historical development of human use of timber as a construction material. In summary, timber is a material that can be used highly effectively to create large and tall structural systems capable of withstanding intense external actions. However, as with any other structural material, good performance is intrinsic to the skill of the designers and builders and not only an attribute of the material itself. Timber buildings that are well designed and properly maintained, on the basis of educated know-how, typically exhibit exemplary performance under normal and abnormal circumstances associated with human usage and natural events.

1.2 MODERN RENAISSANCE OF TIMBER AS A CONSTRUCTION MATERIAL

1.2

5

Modern renaissance of timber as a construction material

Although humanity has a long and rich history of using timber as a construction material, little of this was seen in the 20th century when usage was mostly restricted to low-rise construction. This reflected 19th century bans on the use of timber buildings of more than a few storeys (e.g. maximum of two storeys in Switzerland, maximum of four storeys in most of North America) as reaction to a rash of large urban fires. However, fire performance concerns were not the only factors in the declining use of timber as a preferred structural engineering material. In the early years of the 20th century the rise of architectural modernism, the wide availability of structural steels, the rapid development of reinforced concrete technologies, and the urbanisation of populations following the Second World War, especially in the USA, were all strong drivers in the diminution of the role of timber as a tall building construction material. One result of the changed preferences was that during the 20th century the technological underpinnings necessary for architectural and engineering design in timber and the practical skills for construction of other than low-rise timber buildings, fell largely into obscurity. This included widespread loss of timber instructions in training curriculums for architects, structural engineers, and construction personnel in countries that formerly Fig. 1.2: World’s tallest timber structure: had strong backgrounds in the subject matter. Gliwice Radio Tower, Poland (courtesy of Also, with the exception of glued laminated Andrzej Jarczewski) timber (glulam) products, during the middle of the 20th century structural design codes did not place a strong emphasis on high-performance structural applications of timber. Particularly important in this context was that first generation of what is thought to be modern timber design codes (created circa 1940–1950) were strongly oriented towards the design of lightweight framework superstructures constructed from small dimension lumber framing materials and sawn lumber boards. In some countries that still remains the orientation of such documents, which strongly emphasize the design properties for lightweight timber products (e.g. sawn lumber, plywood, and oriented strand board) and the design practices for joining such materials.

6


Despite broad loss of interest during the 20th century in using timber as a fully fledged structural material, the know-how necessary to use it did not die. This was largely because of the dedicated efforts of a cadre of aficionados who collectively maintained existing knowledge and further developed it in readiness for a reawakening of broader interest in timber as a construction material for large and tall building superstructures. Instrumental in this context have been expert panels like Working Commission W18 – Timber Structures who operate under the umbrella of the International Council for Research and Innovation in Building and Construction; and the IABSE Working Commission 2 – Steel, Timber and Composite Structures. Their efforts were not in vain, and during the last three decades there has been renewed technical interest in timber by researchers and practitioners around the world. Consequently, new ideas have emerged for design and construction of timber structures in ways consistent with modern needs and practices. Recent renewed interest in timber coincides with an almost total reversal of the factors that led to suppression of interest in it approximately one century earlier. As already indicated, modern BPO-based codes, circa 1990 onwards, permit timber usage in relatively tall multi-storey buildings provided it can be demonstrated that fire performance objectives will be met by design solutions. In addition, the availability of new high-performance timber-based composites (e.g. Laminated Strand Lumber), innovative building element manufacturing technologies (e.g. Computer Numerical Controlled cutting machines), novel architectural styles and vernaculars; and the search for “greener” construction methods have fuelled a renewal of interest in timber. In response, timber design codes in various parts of the world have begun to be reoriented towards provision of generalized structural engineering information, instead of only focussing on information appropriate for low-rise building applications [4,10]. Currently educational and professional training mechanisms related to modern applications of timber in construction are not at a level comparable with what exists in relation to materials like structural steel and concrete, but significant initiatives addressing this are underway in Austria, Canada, China, France, Germany, Italy, New Zealand, Sweden, Switzerland, United Kingdom, and elsewhere.

1.3

Other chapters

This SED is timely and appropriate for above mentioned reasons. Whether the 21st century will see full awakening of timber as a structural material for construction of tall multi-storey buildings will depend largely on whether new building systems are developed in harmony with evolved needs of societies in various nation states. Transitioning from what has been done to what can be done requires creation of design and construction techniques that result in buildings that can accommodate modern lifestyles, are healthy to occupy, are green to construct and operate, are durable, and are economic to construct. Some already well developed and ef ficient traditional timber construction methods will surely continue to be used, but those already in common usage may not address future challenges. Principally what are required are the following: (1) engineered systems that fill existing or emerging construction niches and (2) dissemination of design and construction knowledge to practicing professionals. Other chapters of this document are intended to contribute towards fulfilling those requirements. The remainder of this SED is arranged to present information in four blocks: •

General engineering concepts related to structural, fire, and durability design of large and tall building superstructures (Chapters 2–4).

1.3 OTHER CHAPTERS

7

•

System-specific concepts applicable to engineering design of building superstructures, with an emphasis on those systems most likely to be applied in the construction of mid-rise and high-rise superstructures (Chapters 5–7).

•

Examples of modern applications of timber in the construction of mid-rise and high-rise superstructures (Chapters 8–11).

•

Thoughts on future possibilities, with an emphasis on hybrid construction systems where timber structural and non-structural elements are used in combination with elements made from other materials (Chapter 12).

Throughout this document, there is an emphasis placed on international practices, as well as an intentional avoidance of proprietary interests related to marketing of specific construction products.

9 Chapter

2

Structural Design Issues

Summary: This chapter relates to the application of engineering methods where provisions of written loading and material design codes are used as explicit justification of the acceptability of structural system designs. In this context, it is assumed that all elements within engineered load paths will be sized sufficiently that it is highly unlikely that they will sustain damage during their functional lives, irrespective of events. Even then, it is necessary to design large and com plex structural systems in ways that will accommodate any localized damage that does occur and limit it to an extent admissible for structural systems only constructed from elements that fail by ductile processes. Such provisions are not expected to adversely impact costs because the sizing of structural elements in relatively tall buildings is controlled by the magnitude of lateral drift under wind and seismic loading scenarios. Recommendations are made which when applied together facilitate proper design of any type of large and tall timber superstructure. The recommendations concern safety, serviceability, and avoidance of unacceptable damage in the event of overload.

2.1 Introduction Just like other engineering materials, timber has specific characteristics that need to be considered to maximize strength and stiffness capabilities of structural elements, substructures, and structural systems. This chapter discusses key issues related to the design of mechanically ef ficient building superstructures. The most important engineering characteristic of some, but not all, timber structural elements is that mechanical properties are directionally dependent. Glued laminated timber (glulam) elements are illustrative of this as the mechanical properties depend upon the direction relative to the axes of regular prismatic members. Table 2.1 shows typical direction of loading design properties of glulam members, based on values in the Canadian timber design code. Quoted specified strength properties in the table apply to glulam made from softwood species, and apply to what is termed Standard Term Loading conditions, which means that those properties are directly applicable to most loading combinations associated with self-weight of building superstructures, plus roof snow loading and/or normal building floor occupancy loads [10]. The Canadian specified strength values are 5-percentile strengths, multiplied by 0.8 to adjust from capacities observed in short-term laboratory tests to those appropriate to Standard Term Loading. Relative to discussion in this Structural Engineering Document (SED), loading combinations

10

CHAPTER 2. STRUCTURAL DESIGN ISSUES

Strength properties (MPa) Longitudinal bending

Longitudinal tension

Longitudinal compression

Longitudinal Longitudi- Transverse Transverse modulus of nal shear tension compression elasticity (GPa)

Douglas Fir-Larch, 24f-EX grade

30.6

20.4

30.2

2.0

0.83

7.0

12.8

2.0

0.83

7.0

12.4

0.51

5.8

10.3

0.83

7.0

13.1

Douglas Fir-Larch, 16c-E grade

14.0

20.4

30.2

Spruce-Lodgepole Pine-Jack Pine, 22f-EX grade

25.5

12.7

25.2

1.75

Hem-Fir and Douglas Fir-Larch, 24EX grade

30.6

15.3

25.2

1.75

Table 2.1: Selected specified strength and longitudinal modulus of elasticity values of softwood glulam, based on Ref. [10]

most likely to govern design of structural glulam elements in superstructure systems would be appropriately sized using such properties. As tabulated values show, glulam members are much stiffer and stronger when loaded parallel to their axes in tension or compression or bending, than when loaded in directions transverse to their axes. This reflects the orientation in which nature manufactured tree stems to be ef ficient in resisting effects of self-weight and environmental loads. Care needs to be taken to avoid construction details that create high stresses transverse to grain or in shear. As Table 2.1 indicates, manipulation of the wood species, quality of lamellas, and layering arrangements of lamellas all enable the creation of a broad range of engineering property combinations. As such, particular element capabilities can be accurately matched with internal force demands on elements. Thus, although the starting point of glulam manufacture is a tree stem with variable characteristics, the end product is an ef ficient structural product. The same applies to other engineered timber composite products, like cross laminated timber (CLT), as discussed in Chapters 6 and 7. Contemporary timber design codes and most proprietary timber-product information are based on representing strength and stiffness properties as apparent properties corresponding to linear elastic material response. This makes the calculation of member properties quite simple. Usually, the greatest complexity is in the design of the connections and that is the aspect that separates the skilled from novice timber designers. For that reason, there is great emphasis in Chapters 5–9 on connection selection and design. Within this SED, the term “structural elements” refers to linear members, plates, connections, and any other parts within engineered load paths of a superstructure. There is often a great emphasis on three characteristics of structural timber: sensitivity of mechanical properties to moisture content, proneness of timber to static fatigue (sometimes called creep rupture), and size of member effects on apparent strength properties. However, for the applications in this SED, these are largely unimportant. The sensitivity to moisture is only a concern where timber is exposed to moisture such as at an outdoor/indoor interface

2.2 DESIGN PRACTICES AND ASSUMPTIONS

11

or by capillary action where timber bears on wet materials. These are neither typical nor appropriate uses of timber elements within the superstructure of a large building. This topic is discussed further in Chapter 4 with respect to long-term wetting and subsequent durability problems. Some timber design codes include what appear to be stringent requirements that discount strength properties of timber elements when their volumes increase beyond reference dimensions (i.e. sizes associated with indexing of tabulated design values). However, such adjustments are largely the result of the already mentioned employment of apparent elastic strength properties, which results in inclusion of post-elastic behaviour. Dependencies of apparent strength properties of elements on their sizes are not highly relevant to the discussion herein because adjustments generally relate to unsuitable products such as small dimension lumber.

2.2

Design practices and assumptions

Under most contemporary, regulated, building design regimes there are th ree pathways by which structural systems for buildings can be justified: 1. Prescriptive Design of Acceptable Solutions 2. Engineered Design of Acceptable Solutions 3. Alternative Solutions equivalent to Acceptable Solutions Pathway 1 – Prescriptive Design is the approach traditionally used to design superstructures of small buildings and, therefore, not relevant to the current d iscourse. Pathway 2 – Engineered Design is the normal approach by which engineers design structural systems and is based on the application of provisions in written design codes that specify loadings and how structural elements manufactured using various materials [e.g. timber, structural steel, reinforced concrete (RC)] should be sized. In general, this pathway applies to the use of generic construction materials and elements manufactured to specific product specifications. This is the default approach discussed in this chapter, and the SED in general. Pathway 3 – Alternative Solutions is the route by which engineers can apply technical information that is beyond the scope of written design codes to demonstrate that designed solutions are at least technically equivalent to Acceptable Solutions based on pathways 1 or 2. The Alternative Solutions pathway is discussed most extensively in this SED in the context of fire design (Chapters 3 and 8–10), and in respect of structural design of systems using timber-based engineered composites like CLT (Chapters 5–10). Discussion in this SED is based on the assumption that the structural design will be undertaken via the Load and Resistance Factor Design (LRFD) approach, also known as Partial Coef ficients Design (PCD) approach. Essential characteristics of LRFD/PCD are as follows: (i) Consideration of the effects possible loading combinations have on stability and deformation of the structural systems. (ii) Explicit separate consideration of various potential limiting states related to stability and functionality of the structural systems.

12


(iii) Accounting for systemic and statistical variability in properties of elements that are parts of the structural systems (i.e. variability associated with or resulting from material manufacturing, placement, and service environment). (iv) Consistency in requirements for the design of structural elements made from various materials (e.g. timber, structural steel, structural glass, RC, reinforced masonry). (v) Consideration of the deformation compatibility of various elements in the structural systems (i.e. employing structural analysis techniques that account rationally for boundary and other applicable deformation constraints and for the interaction of elements in systems). Notably, there is no explicit discussion in this SED about foundation substructures beyond a brief reporting on systems employed in example projects and case studies of Chapters 8–11. Nevertheless, it is intended that information given will be used in combination with well-established principles in foundation design applied by engineers experienced in such matters. Additionally, it is presumed that engineers designing building superstructures and their foundations will be fully cognizant of the principles and details of complete building design.

2.2.1

Load combinations, load factors, and resistance factors

LRFD equation formats applicable within various nations, regions, and other building construction regulatory jurisdictions are variable with regard to specific details but conform in principle to the requirements described below. Worst-case Effects of Factored Design Loads ≤ Factored Resistance of System/Element

(2.1)

The fundamental principle embedded in Eq. (2.1) is that the maximum effect of all possible loading combinations must be resisted in respect of effects related to loss of stability or unacceptable deformation. This principle is applicable with respect to the behaviours of the complete superstructures, substructures, and individual structural elements. Loading codes account for the geographic locations of buildings and define all necessary information related to magnitudes and combinations of loads to be considered by engineers designing superstructure systems. In most instances, the partial load factors are independent of the type(s) of material used to construct building superstructures. However, there are some important exceptions, with the most typical being cases where engineers are allowed to perform equivalent static force design of relatively simple buildings, as in the example project of Chapter 8, and cases where temporally varying processes associated with material responses (e.g. shrinkage and creep) can generate significant stresses in the structural systems. Material specific codes define characteristic material properties necessary for sizing structural elements and making deformation predictions and specify partial material coefficients. Those codes include guidance on necessary adjustments of characteristic material properties to transform them from reference values to values appropriate to particular service situations. Technically, resistance factors in LRFD (commonly termed phi f values) and material partial factors in PCD (commonly termed g values) are the inverse of one another: M

f = 1/ g M

(2.2)

13


In the context of this document, it is presumed that engineers will employ LRFD/PCD design practices deemed appropriate to the design of the structural systems in particular geographic locations.

2.2.2

Achieving an elastic response and allowance of damage

2.2.2.1 Factors to consider Under pathway 2 to Acceptable Solutions, there is an implicit requirement that all elements in a structural system be able to distort suf ficiently to achieve the levels of deformation associated with their characteristic design strength properties. However, the only reliable way to ensure that this presumption is not violated is to design and construct structural systems such that they are statically determinate. In all cases where systems are statically indeterminate, adoption of codified design strengths of elements (i.e. element design capacities) presupposes that substructures can accommodate deformations required to achieve full strengths of critical parts. Similarly, adoption of codified strength of elements presupposes d a o that complete structural systems can L accommodate deformations required PY to achieve full strengths of substrucP = yield load attained d tures. Figure 2.1 conceptually illusP = achievable capacity constrained by d trates the conundrum that elements PC within statically indeterminate systems can have reserve capacity that will not be utilized unless other parts of the system have sustained damage. Deformation d C d Y Such situations may or may not result Fig. 2.1: Ductile element design strength RY (= PY ) in unstable damage propagation affectbased on unconstrained deformation, versus ing substructures or even complete achievable capacity RC (= PC ) attainable under structural systems. within system constrained displacement d C Questions related to what would constitute acceptable damage, if a structural system of any building were to be overloaded must address three perspectives. First is concern for life safety of occupants or rescue workers (if a building collapses in part or in its entirety), which is the mandatory aspect of code requirements. However, satisfying this perspective does not imply any requirement to avoid damage, if that damage would not imperil human lives. The second is concern that damage to any one building will not result in damage to adjacent buildings. Contemporary regulatory requirements address this issue, with avoidance of damage to neighbouring building being an explicit requirement of many building codes. However, again the objective does not equate to inadmissibility of damage to critically loaded buildings. Thus, in an overall sense regulatory requirements do permit build superstructures of any type and sized to sustain damage under unusual circumstances like design-level wind storms or earthquakes provided that secondary consequences are avoided. The third aspect to consider is that of building owners and occupiers, who in general wish to have buildings designed to levels that preclude the possibility of other than superficial damage under anticipated scenarios. Y

C

Y

C

14


In the context of designing building superstructures, it is always necessary to recognize that structural design codes are best practice guidelines that supplement the know-how of structural engineers; and that such codes are written to provide the minimum information for adequate reasonable protection of life and property. Loading codes applicable to design of normal buildings are written based on the presumption that engineers will use linear elastic analysis concepts to transform design loadings into effects on the superstructure elements, with those effects being internal forces and deformations. However, in many instances material codes for steel, RC, timber and other materials provide building/structural element sizing rules that often reflect postelastic response capacities of elements. The extent of variance between force levels associated with the onset of post-elastic responses and force levels associated with the determination of design capacities for elements is highly dependent on the types of elements and failure mechanisms involved. On the other hand, as all engineers know, characteristic design strengths in material design codes are always based on low exclusion level estimates of strength (usually the 5th percentile strength), and commonly elements are oversized because of the discrete sizes in which elements are available. Consequently, especially in systems where elements arranged in parallel act together to resist applied forces, actual strengths will exceed presumed design strengths. Relevant to this SED, it is therefore acknowledged that in practice there is no exactness in the protocol by which engineers combine information in loading and material codes to design structural systems. Except when loaded in compression, structural timber elements fail by pseudo-brittle processes (i.e. exhibit limited post-elastic response prior to loss of capacity) [11]. Therefore, one should not presume that large and complex structural systems that employ structural timber elements will be capable of accommodating localized damage of an extent that may be judged admissible for structural systems only constructed from elements that fail by ductile processes. Engineers perform analyses of structural systems that transform code load i nformation into the effects of loads on structural elements. Irrespective of what design analysis methods and tools are employed, materials’ information contained in codes alone is not sufficient because some important stiffness property information is missing. In particular the information about stiffness of connections is minimal or non-existent. Most connections joining structural timber elements to each other and to other structural materials are comprised of mechanical fasters or carpentry joints [12]. Connection response is not simple to estimate with exactitude. Therefore, either it is assumed that timber end connections are pinned or experimental research information is used to characterize their responses. Often it is reasonable to suppose that: •

When building superstructures are low-rise and not slender in shape, primary influences on structural element sizing are satisfaction of strength demands associated with resisting gravity forces on walls, columns, slabs, and girders. In the case of lightweight elevated floors, avoiding vibration serviceability problems will often control elements sizes.

•

For medium-rise and moderately slender buildings, superstructure element sizing is often related to provision of adequate strength against the combined effects of gravity and lateral loads (e.g. wind and seismic).

•

When building superstructures are tall and slender, the dominant influence on structural element sizing is the need to control lateral motions associated with wind and seismic loads.

15

2.2 DESIGN PRACTICES AND ASSUMPTIONS 1.0

Sawn timber

0.9 s e i t 0.8 r ) s e i p s 0.7 a o r b p s 0.6 c s i a f i m 0.5 c t e i p n 0.4 s u e r v 0.3 i t e p a ( l 0.2 e R 0.1

Steel Plain conc.

0 Tension

Compression

MOE

Fig. 2.2: Approximate relative specific mechanical properties of common construction materials

As discussed further for example projects in Chapters 8–11, experience to date is that sizing of elements in relatively tall superstructure systems employing timber elements can be governed by strength considerations. However, more usually control of sway is the governing considerations with emphasis on the provision of horizontal diaphragms having sufficient rigidity to maintain the building shape under design-level wind or seismic loads. This reflects the relative high strength to mass and stiffness to mass ratios for timber ( Fig. 2.2). Timber structural systems are often claimed to be highly compliant and able to absorb energy well in elastic and post-elastic regimes during extreme loading events [13]. This is true of some highly statically indeterminate traditional and modern structural systems employing timber which can accommodate very high levels of local deformation and absorb large quantities of kinetic energy in their elements without more than localized damage. Figure 2.3 illustrates such systems. In the case of the traditional timber-laced and timber infill-frame buildings shown, they accommodate very high distortion, because all of the building’s substructures are flexible. They dissipate significant energy through frictional contact with the constrained masonry pieces. Their survival from major earthquakes has proven that traditional timber-laced and timber infill-frame buildings have remarkable capabilities to recover elastically once unloaded [14]. Modern CLT buildings can also accommodate large deformations and absorb energy because the intra-wall segments connectors and those between the walls and horizontal floor diaphragms are metal and thus flexible and can distort plastically [15]. Additionally, special multi-storey timber construction systems employing post-tensioning as a means of absorbing potentially damaging kinetic energy flows through connections during earthquakes are being developed in New Zealand [16]. However, despite it clearly being possible to avoid damage to structural timber elements by utilization of special techniques, it is not envisioned that engineers will often want to avail themselves of exotic design options. 2.2.2.2 Recommended design practices The following recommendations relate to design of large multi-storey superstructures employing timber as a primary, but not necessarily the only, construction material:

16


(a)

(b)

(c)

Fig. 2.3: Old and modern highly statically indeterminate superstructure systems known to accommodate very high levels of local deformation in elements: (a) timber-laced bearing wall house (courtesy of R. Langenbach); (b) timber infill-frame house (courtesy of R. Langenbach); (c) modern massive timber CLT assemblies (courtesy of A. Ceccotti)

Recommendation 1: Designers should aim to achieve strict compliance with loading codes and material specific design codes and provide clients with reasonable surety that superstructure systems will not be damaged by loading scenarios anticipated by design codes. Recommendation 2: For buildings that do not contain special measures to absorb kinetic energy during design-level earthquakes or other unusual design events, the Engineered Design approach is an appropriate method for demonstrating acceptability of superstructure system designs. Recommendation 3: Engineered designs should focus on ensuring that primary elements in superstructure systems do not exceed the elastic response range. Recommendation 4: For buildings that contain special measures to absorb kinetic energy during design-level earthquakes or other unusual design events the Alternative Solutions approach is an appropriate method for demonstrating acceptability of superstructure system designs.

17


1. Assign initial element size & stiffness: i= 1, I 2. Designate elements as brittle (Type B) or ductile (Type D); i = 1, I

START Reset: k = k + 1 k = 1 Load case j ( j = 1, J )

Elastic analysis of internal element forces and system deformations

j = j + 1

Reset: k = k + 1 Is j = J ?

NO

YES Resize elements (selectively or globally to achieve target displacement responses)

Resize Elements:

OUTPUT iteration k: F i,j,k values system deflections k

Ri,k +1 = ( Ri,k )2 / IFR Max_i,k (i = 1 , I )

Note: element stiffness resized as a result

Calculate Maximum Internal Force Ratios, iteration k: IFR Max_i,k = Max F i,j,k / Ri,k ( j = 1 , J )

Find Global Maximum Internal Force Ratio for Type B elements: GIFR Max_k = Max. IFR Max_i,k (i = 1 , I B: number of Type B elements I B I )

Is GIFR Max_k = 1.0 ± tol.

NO

? YES

NO

Are k acceptable

YES

END

?

Fig. 2.4: Recommended iterative process to achieve an elastic system response Figure 2.4 shows a suitable design methodology for implementing “normal” design in a way consistent with recommendations 1–3. To follow this approach designers need to classify structural elements by failure mechanism: brittle (Type B) or ductile (Type D). Table 2.2 suggests a simplified method of classifying elements. Alternatively numerical analysis or experiment test information can be used for this classification.

Notably, the suggestion to employ distinctions between brittle and ductile elements within engineered load paths of superstructures is not an explicit suggestion that engineers adopt Capacity Based Design (CBD) practices that are permitted options under some design codes [17,18]. Adopting CBD practices can provide an added layer of surety against unstable damage propagation. However, they are not sufficiently reliable to be used as the first layer of protection that load paths containing timber members will behave as desired, which is why retention of an elastic response is advised herein. It is suggested that those engineers interested in following the Alternative Solutions approach to proving the acceptability of proposed designs refer to the comments in Subsection 2.5.2 that

18


Type B – brittle elements

Type D – ductile elements

Sawn timber

CLT

Laminated veneer lumber

Timber connections with small diameter fasteners

Glulam

Reinforced mechanical timber connections

Parallel strand lumber

Laminated strand lumber

Welded steel connections

Structural steel

Cast iron

Bolted steel connections

Unreinforced concrete

RC

Unreinforced masonry Structural glass Table 2.2: Approximate classification of common structural element as brittle or ductile

address robustness in the context of need to block the possibility of damaging kinetic energy flows which can lead to disproportionate damage of large structural systems.

2.2.3

Analysis methods

Engineers embarking on the design of complete or partial timber superstructure systems for relatively tall multi-storey buildings must be prepared to employ relatively sophisticated analysis methods/techniques and tools. However, except when the objective is to design special systems based on the Alternative Solutions approach, those methods and tools will be ones with which engineers who design other types of tall multi-storey superstructures are familiar and, therefore, the requirement is a normal one. Most general purpose commercial structural analysis software packages based on finite element analysis methods have full capabilities for handling the determination of design forces in timber structural components and their connections. Structural analysis of timber construction is essentially no different from design of other systems, if the structural skeleton is well defined, even though timber frameworks will typically contain less moment transferring connections and more bracing diagonals and/or supplementary stabilizing components (as discussed further in Chapter 5).

2.3

Effect of superstructure shape and height

Architectural form (height, slenderness, and shape) affects the structural form and, thus, how building superstructures will respond to various external and internal loadings. Therefore, consideration of heights of building superstructures alone is not a direct indicator of whether or not they will be difficult to design and construct. Adding geometric proportioning and structural form does, however, enable some generalizations to statements about design and construction complexity. As noted in Subsection 2.2.2.1, there exist approximate relationships between slenderness and which limiting state will govern lateral load responses of building superstructures. The simplest superstructures to design and construct are those having squat shapes (i.e. a modest ratio of

2.3 EFFECT OF SUPERSTRUCTURE SHAPE AND HEIGHT

19

height to footprint dimensions) and containing many internal divisions. This is true irrespective of the absolute height. In such circumstances, satisfying ultimate limiting state design criteria is mostly a case of being able to provide sufficient resistance to gravity forces and sufficient shear walls or bracing to resist effects of horizontal shear forces during extreme wind or seismic events. From a strength point of view, principally what limits the height of squat buildings is the point at which columns or walls occupy an unacceptably large proportion of the floor plans at the lowest levels in the system. Figure 2.5 illustrates internal force flows that can be expected to be of primary importance for low-rise squat and relatively tall and slender buildings, based on the illustrative case of platform timber construction and seismic ground shaking. Essentially, in low-rise buildings the key internal actions within superstructures are the horizontal forces between layers through

Inertial forces

Reaction forces

Low-rise platform construction • Storeys are like stacked shoeboxes • Key structural issue is resisting horizontal shear flows • Horizontal motions are usually not problematic • Modest overturning forces to be resist between layers • Shallow foundations are adequate • Mode shapes are quite simple • Structural detailing requirements are not similar for wind and seismic design

Tall and slender platform construction • Storeys are like stacked shoeboxes • Key structural issues are: -Resisting horizontal shear flows, and -Resisting uplift in and compression in walls • Horizontal and possibly vertical motions problematic • Overturning forces are large • Shallow foundations are not adequate • Mode shapes can be complex • Structural detailing not dissimilar for wind and seismic design

Fig. 2.5: Effect of building height and shape on internal force flows: seismic actions

20


the system’s depth. In taller and more slender buildings, the important internal actions within superstructures are the horizontal shear forces and axial forces because of gravity and the bending forces. Even without specific variables or calculations, it is readily apparent that for slender buildings controlling horizontal drift will be problematic. Resisting the total applied shear forces and bending moments near to the base is also likely to be a concern. Controlling of drift levels for slender, all-timber buildings beyond 8 or 10 storeys is likely to be impossible under normal economic constraints. Thus, the most viable option for using timber in relatively tall and slender superstructures is hybrid construction that combines timber substructures with other substructure types [19,20]. As exemplified by the behaviour of slender steel and RC superstructure frameworks that work in conjunction with RC building cores and massive timber floor and roof diaphragms (Chapter 6), torsion must be explicitly considered for superstructure deformation under wind and seismic design events. This includes the need to consider demands on substructures and structural elements that can result from irregularities in the architectural or structural shapes.

2.4

Importance of horizontal diaphragms

As has been recognized for many centuries by craftsmen builders of slender masonry towers, ensuring that horizontal diaphragms have high in-plane rigidity and high resistance to out-ofplane buckling enables walls to function in structurally efficient ways. If the horizontal diaphragms maintain their shapes, are not spaced at excessively large vertical intervals, and are securely tied to walls, even severe events like strong earthquakes can be sufficiently withstood to keep brittle unreinforced walls undamaged. Technically, the explanation is that perimeter walls can resist cantilever and torsion distortions and stresses through in-plane actions of the walls provided that their arrangement is not distorted out-of-plane; the high self-weight of the masonry counteracts overturning; high self-weight and friction also counteract interlayer sliding across horizontal planes. Keeping the vertical spacing between horizontal diaphragms modest compared with their plan dimensions avoids the need for excessively thick walls, because their buckling lengths are constrained. Historically timber was commonly the material of choice for construction of the horizontal diaphragms of slender masonry towers, with such diaphragms being constructed in ways that made them highly rigid. In the modern context, the essential principles of ensuring efficient structural actions of slender structures are the same in respect of the importance of the diaphragms. Building shapes may be more complex and the construction less massive, but the use of horizontal diaphragms as substructures to maintain the general shape of systems remains critical to efficient design of the vertical substructure members. Multi-storey building superstructures of types discussed in this SED employ large timber horizontal diaphragm elements as critical parts of fully three-dimensional, self-bracing arrangements. As discussed in detail in Chapter 6, it is possible to construct massive CLT diaphragms that are as rigid and as strong as RC diaphragms at near equal thickness. The ability to do this is important to the utilization of timber in very tall superstructures, because the rigidity of those substructures is sufficient to ensure that framework substructures will deflect laterally in unison with RC or other massive shear walls. Making parts of superstructures so that they move in unison dynamically is important, because it minimizes the potential that damaging forces will flow through connections (e.g. between framework and shear wall substructures) during seismic or other events that shake the buildings.

2.5 ACCEPTABLE RISK LEVELS AND AVOIDANCE OF DISPROPORTIONATE DAMAGE

2.5

Acceptable risk levels and avoidance of disproportionate damage

2.5.1

Risk

21

The acceptable risk level under predictable low frequency events like severe storms, earthquakes, and accidental explosions relates to occupancy levels for individual buildings and by how many buildings might be affected by one event. This reflects lower societal tolerance and diminished rescue response capabilities when the number of people at risk is large. The possible magnitude of loss of building functionality and of finances to owners or insurers, as well as disruption to lives and businesses may also diminish the acceptability of risks possibly leading to building failures. Given that tall building constructed using timber (in part or whole) are most likely to be located in urban environments, acceptance of other than localized repairable damage is low. Setting minimum acceptable standards for protection of life and property are national or regional government responsibilities. However, target risk levels for tall multi-storey buildings having timber structural superstructures or substructures can be expected to be considerably more stringent than for low-rise timber buildings. In practice, the simple way to keep structural systems from being damaged is to not stress their constituent parts beyond the elastic range. That combined with selection of structural systems that are inherently robust is the strategy recommended in this SED ( Subsection 2.2.2.2).

2.5.2

Mitigating damage potential

Robustness is defined as “ability to survive unforeseen circumstances without undue damage or loss of function” and “provides a measure of structural safety beyond traditional codified design rules” [21]. Much debate, and seemingly some confusion, has surrounded questions of what constitutes acceptable damage, what constitutes system robustness in various contexts, and how to formulate design provisions leading to robustness. Massive timber plate systems with up to 15 (or possibly more) storeys (as discussed in Chapter 7) are honeycombed and highly structural redundant. They can be made to be very robust and able to arrest the spread of damage to the systemic level. In such systems, connections are highly dispersed and ductile, and toughening in their responses, if overloaded. Ensuring system robustness is, therefore, usually straightforward with respect to massive timber plate systems. Robustness of tall buildings containing frameworks of large timbers (as discussed in Chapter 5) depend s on proper selection of the overall structural system, with the goal being that frameworks are laterally stabilized by massive shear walls or building cores constructed from RC o r other massive materials. Also for robustness, building cross-section shapes need to be maintained by horizontal diaphragms, as outlined in Section 2.4. Avoiding damage and ensuring robustness depend on using framework connections and other construction details that make the timbers function in ways that exploit their strength (i.e. do not subject them to forces that develop less reliable responses). Practices discussed in this SED are modernized applications of framing practices proven in Victorian and Edwardian eras in the context of large cotton mills and other large multi-storey buildings. Tall hybrid buildings with massive timber diaphragm substructures are discussed in Chapter 6 with an emphasis on systems where steel or RC frameworks function in combination with RC building cores. Robustness considerations for such buildings directly parallel practices applicable to other buildings with moment resisting steel or RC frameworks.

22


Apart from philosophical and subjective issues of relative security of tall versus low-rise buildings, there are many compelling fracture mechanics-based reasons why the design of large building systems should be based on elastic analyses. Irrespective of whether damage occurs in a small object like a fracture toughness test specimen or in a large object like a building superstructure, damage will continue to propagate if the conditions in Eq. (2.3) occur [22,23]:

d We > d U + W f + d V + d W p + d Wd + d L

(2.3)

where d We is the increase in the work to the system since the last increment of damage, d U is the increase in strain energy in the complete system because of the last increment of damage, W f is the work required to create the next increment in damage, d V is the increase in kinetic energy in the system because of the last increment of damage, d W p is the plastic work to distort “other” components of the system, d Wd is the work to overcome damping in the system, and d L is other energy loses in the system, because of phenomena like creep and generation of heat. If only the first three terms in Eq. (2.3) are considered, which corresponds to standard static analysis, it will be concluded that the capacities of structural components and complete structural systems are unique. However, when the equation is considered in its entirety, it becomes clear that capacities of components are always conditional quantities. Therefore, the capacities of two identical components situated within two different structural systems (or two different locations within the same system) will fail at different levels of internal force [22]. In practical terms, the implication of Eq. (2.3) is that as systems get larger, they tend to become more brittle, because d U becomes larger with respect to W f . For buildings constructed with non-timber structural materials, Eq. (2.3) has been applied in conjunction with search tree algorithms to predict which amongst a large range of the possible failure mechanisms and sequences is likely to govern the failure of a large structure under various loading scenarios [23]. Such an application is illustrated in Fig. 2.6 . The approach could be carried out to predict the minimum failure loading for the system or to identify system modification strategies that mitigate against occurrence

One branch of a 2D Search Tree with twigs: Branch ends

Need to know which branch and which twigs control the process.

Damage arrested

Possibly many stems exist, # branches and twigs depends on complexity of the system. Damage evolves

Undamaged structure

Multiply by # t ype events (e.g.) • Seismic loading • Gas explosion • Bomb

Damage event

Damage evolves

Fig. 2.6: Example of systemic analysis

Dynamic roots

Multiply by 3-D

= # Stems

2.6 PODIUM AND OTHER CONSTRUCTED SYSTEMS WITH ARTICULATED DYNAMIC RESPONSES 23

of disproportional collapse under various damage scenarios (e.g. overloading and explosions). Although the details of any outcome would depend on the structural form and construction details for a building, it is often the d V term that is responsible for occurrence of disproportional damage. Thus, fracturing is a phenomenon for consideration that emphasizes the need to maximize material damping in the deformation range before there are unrecoverable localized distortions in elements. As a sidebar to the present discussion, it is worth mentioning that Eq. (2.3) leads to the unerring conclusion that exact designs for strength are only possible if the structural analyses of systems are based on energy methods. Of course, the implication is that failure criteria for components should be related to how much energy it takes to destroy them, but it can be anticipated that most structural designers will not wish to embrace this idea because it requires abandoning a long history of designing structural systems according to apparent strength criteria, in which they mostly have great faith. For this reason, the discussion in this subsection should be regarded by design practitioners as simply conceptual reasoning supporting recommendations in this chapter.

2.6

Podium and other constructed systems with articulated dynamic responses

A concept attracting interest in North America is construction of four to six lightweight timber superstructure storeys on top of several RC superstructure storeys (Fig. 2.7 ). Such systems that are commonly referred to as podium construction raise interesting design questions related to how the timber and RC storeys should be designed. Feedback from design engineers indicates that the principle question they have concerning how to design such systems relates to the application of equivalent static force methods for determining the possible effects of

Fig. 2.7: Modern podium construction with light-frame timber construction above two RC plinth storeys (courtesy of FPInnovations)

24


earthquakes. Reportedly, the practice normally being adopted is to design RC plinth and timber upper storeys as fully RC superstructures and fully timber superstructures, respectively. The construction of podium or other superstructure systems that couples substructures with different uncoupled dynamic response characteristics has the potential to create articulation in the mode shapes and to result in unexpected force flows at points of connectivity/articulation. Practices such as analysing substructures as though they are separate structural systems are unreliable and can lead to inconsistencies in values assigned to ductility-related force modification factor ( Rd ) and the overstrength-related force modification factor ( Ro). The product 1/ Rd Ro is used to adjust equivalent static design forces applicable to systems with brittle–elastic responses to equivalent design forces applicable to the expected “actual” response of the system. Technically Rd Ro is equivalent to what in Europe is called the Behaviour Factor q, defined as the “factor used for design purposes to reduce the seismic actions in a linear static or modal analysis in order to account for the non-linear response of a structure, associated with the material, the structural system and the design procedures” [18]. In principle, the issue related to seismic design of podium construction is the same as issues addressed in design of the hybrid ti mber and RC superstructure of Example Project 1, as presented in Chapter 8. The issue is addressed in this chapter because it is important to be clear about what this SED recommends in cases where articulated dynamic responses of substructures must be evaluated. The approaches recommended are:

Recommendation 5: In cases where it is clear that a hybrid superstructure containing coupled substructures meets conditions associated with use of simplified equivalent static seismic design forces (as stipulated in the applicable structural design codes), the most conservative possible interpretation of the behaviour factor q (or equivalent factors Rd Ro) should be adopted.

Recommendation 6: In cases that fall outside the scope of recommendation 5 the effects of seismic forces on superstructure systems should be evaluated on the basis of a full dynamic analysis.

2.7

Additional comments

In the past, with few exceptions, design and construction of timber buildings has been based on a combination of experience and simplified engineering analysis. However, if the objective is to design timber building superstructures that have in the order of six or more storeys, then modern computer-based engineering analysis approaches are required. Attempts to sidestep this will undoubtedly lead to the creation of unsuccessful designs. Design engineers need to have available as wide a range of choices as possible. The discussion and commentary in this chapter is aimed at facilitating that with respect to traditional timber products (e.g. sawn timber and glulam) and modern timber-based, engineered composites as high-performance structural materials. Chapters 5–7 place what is said in this chapter in context relative to various types of structural systems for buildings.

25 Chapter

3

Fire Design Concepts

Summary: Combustible building materials like timber burn on their surface, release energy and thereby contribute to fire propagation and the development of smoke during building fires. Largely because of these characteristics, building codes have historically prescriptively limited the use of timber as a building material. In particular, codes have restricted the number of storeys and heights of timber building superstructures. However, timber elements are very predictable in terms of how they burn, and if properly designed, they will perform excellently during building fires. Effective fire design and firefighting is essential for very large, low-rise timber buildings, as well as for tall buildings constructed of other materials. There is no reason that this knowledge cannot be applied to high-rise timber construction. Notably, research has already verified the creation of appropriate construction details and produced models of how tall timber buildings should perform in fires. Modern performance-based codes recognize this, thereby making it possible to rationally account for how buildings containing timber will behave as constructed systems in the event of fires. Such codes recognize that timber elements can be designed to perform well by either being large enough that they only char on their surfaces during fires, or being shielded from fire by being encapsulated in layers of non-combustible materials. Fire compartmentalization is crucial to good fire performance of any large building and that is emphasized in the discussion herein, as a primary mechanism for ensuring that tall multi-storey building superstructures containing timber will perform as desired. Specifically included herein are discussions that outline the historical development and current state of international building fire design regulations. These explain how the performances of individual timber parts of buildings can be predicted; discuss building system design concepts; and illustrate how tall buildings with desirable fire performance can be constructed. As the example of the 120 m tall Dock Tower project concept illustrates, tall buildings incorporating timber as a primary construction material can be expected to have fire performance at least equal to, and potentially better, than typical traditional non-combustible tall buildings of today. As with all other aspects of design and construction, finding acceptable solutions depends on equipping engineers with knowledge and tools with which to exploit the potential of fire engineering. Discussion in this chapter presents the relevant know-how.

3.1. Introduction Many building codes have historically limited use of timber as a building material because of its combustible nature. Such limitations manifest themselves as prescriptive limits on the number

26

CHAPTER 3. FIRE DESIGN CONCEPTS

of storeys and heights of timber buildings [24]. For example until 2005, new timber structures in Switzerland were mostly limited to low-rise buildings of no more than two storeys. As discussed in Chapter 1, the basis of this was societal memory of past disastrous urban fires that had occurred in various places over centuries. Fire safety is always a vital factor for hu mans to feel comfortable and an important criterion for the choice of material for buildings, especially in residential buildings where people sleep and can be slow to react to fire events. Since the 1990s, many research projects around the world have focused on the fire behaviour of timber structures [25–27], with some specifically aimed at supplying basic data and information on the safe use of timber in multi-storey buildings. This led to novel fire design concepts and models, based on extensive element and full scale testing [28,29] and a large statistical database on fires in timber and concrete/brick buildings [30]. This improved knowledge, combined with technical measures, especially sprinkler and smoke detection systems and well-equipped fire brigades, allow for the safe use of timber in a wide field of applications. Again taking the Swiss example, research has supported changes to the building code regulations that allow construction of mid-rise residential buildings with up to six storeys. Such code alterations have been made in other countries, similarly based on improved knowledge of how completed buildings perform during fires. Consequently, there is now widespread international convergence of codified fire performance-related regulations founded on requirements that buildings once constructed will meet performance-based design objectives (Sections 3.3 and 3.5). While, there are only a few examples of applying modern fire design practices to genuinely very tall timber buildings, the practices are well proven in other contexts. For example, it is well known how to handle design and fight fires in very large low-rise timber buildings and how to do the same for tall buildings constructed of other materials. Research has verified the creation of appropriate construction details and produced models of how tall timber buildings should perform in fires. This reflects that fire safety engineering has developed rapidly as a discipline that integrates all aspects of fire safety (structural, technical, and organizational) and combines knowledge into the design of buildings. However, it is perhaps unrealistic to suppose that building regulatory authorities will rapidly permit unlimited use o f timber in tall buildings worldwide given the historical aversion to that. This chapter presents a generic fire safety concept for future tall buildings in which a large proportion of the volume and mass of the structure (wall and floors) is constructed using timber or wood-based composite products.

3.2. Fire action Figure 3.1 shows the development of a fire in a typical room [32]. After ignition, fires can grow very rapidly, very slowly as in a smouldering fire, or self-extinguish; depending on the arrangement of combustible materials in the vicinity of the ignition source, the type, geometry and dimensions of combustible materials, and ventilation of the room. Fires not controlled by firefighting actions may lead to rapidly rising temperatures and flashover where all unprotected combustible materials burn. The combustion process releases heat energy, gases, and smoke, with the gases and smoke most often being the cause of fatalities. While approximately 80% of fire fatalities are because of smoke [30,32], heat is the primary cause of damage to building superstructures. Because mechanical and thermal properties of building materials change with increasing temperatures, knowledge of the time–temperature development during a building fire is crucial for structural fire analysis. For the purposes of design, fire action is represented using simplified models, and several “nominal fire

3.3. FIRE SAFETY OBJECTIVES AND STRATEGY

27

curves” have been proposed. The most frequently applied fire curves are the ISO Growth Burning Decay 834 [33] curve and the ASTM E119 [34] curve, which are almost identical. For special fire situations, other curves like the hydrocarbon and external fire curves have been developed [35]. Nominal fire curves provide a simple relationship between the temperature of the gases in Time Ignition Flashover a compartment and time, and represent fully developed fires. Such curves neglect Fig. 3.1: Typical stages of fire development [32] the amount of time that elapses between the beginning and full development of a fire, which in some instances can be considerable. Similarly, the cooling phase of a fire is not considered. More realistic models of fires are given by parametric fire curves, which take into account important parameters related to temperature development such as the following: e r u t a r Pre e p ignition m e T

•

Fire load, which is the amount, type, and arrangement of combustible materials.

•

Ventilation conditions in the room of a fire’s origin.

•

Thermal properties of the enclosures, that is, walls, ceiling, and floor.

•

Firefighting actions.

Parametrical fire curves are calculated with simplified formulas developed for limited boundary conditions [35]. For a more comprehensive and more detailed analysis, computer simulations may be used like multiroom zone models or computational fluid dynamics models. Using probabilistic approaches to fire simulation, it is possible to incorporate the input parameters as variables and, thereby, account for uncertainties relevant to an analysis [36]. However, because of the complexity and expense, such an approach is restricted to special situations and buildings where simple calculations cannot be applied with confidence.

3.3. Fire safety objectives and strategy The knowledge of the basic behaviour of fire, buildings, and building occupants during fires is a precondition for the development of successful fire safety strategies. Fire safety must be regarded as an engineering design requirement of equal importance to structural design under normal (non-fire) conditions. The most efficient way to control the effects of fire is to establish a comprehensive fire safety strategy with an adequate combination of measures that fulfil fire safety objectives [37,38]. Depending on the type of structure, various combinations of technical and organizational measures are necessary besides the traditional structural fire safety measures. The starting points of any efficient fire safety concept are the following objectives: •

Safety of occupants and firefighters.

•

Safety of occupants of neighbouring buildings and their property.

•

Limitation of financial loss associated with buildings and their contents.

•

Protection of the environment against effects of fires like harmful gasses and pollution of ground water.

28


As achieving absolute safety is impossible, the level of acceptable risk is in general quantified by building regulatory authorities, or by building owners and insurance companies in th e case of financial losses. Objectives can be reached with different fire safety concepts taking into account the type of structure and occupancy. The most efficient ones should be selected through comparison of options based on criteria like the total cost of fire safety measures, flexibility of application, and architectural constraints on acceptability of alternative fire safety concepts. Depending on the main focus of the measures the following generic concepts are distinguishable: – Structural concept : Buildings are divided into compartments with fire-resistant floors and walls that limit the spread of fires beyond where they started, with compartment sizing depending on building type and usage. Floors are usually the primary fire separating elements in multi-storey buildings. It is accepted that contents inside the building will be totally destroyed and that building compartments will be damaged. This favours construction of buildings with cellular structural forms. However, small rooms may limit use and flexibility, especially for occupancies like shopping centres and industrial facilities. – Surveillance concept : Automatic fire detection systems can locate fires and alert nearby fire brigades, thereby allowing adoption of less stringent fire protection measures. Sometimes this even leads to acceptability of unprotected steel or timber, especially in low-rise buildings with small fire loads and slow fire development (e.g. offices, schools, and gymnasiums). However, in such instances fire detection systems must be properly maintained and regularly checked for functionality. – Extinguishing concept : A sprinkler system can extinguish fires at an early stage or keep their intensity low, until firefighters arrive. In some jurisdictions, deployment of a sprinkler system is considered to greatly reduce the necessity of the passive fire resistances of structures and fire compartment separations. At least yearly checks are important to achieving high reliability of such sprinkler systems. Prescriptive, also known as “deemed to comply” fire regulations give detailed rules leading to standard concepts. Standard concepts are considered to give an acceptable level of fire safety. If prescriptive fire safety concepts are employed within the scope of building regulations, no quantification of the fire risk is required. The scope of such approaches is usually limited to relatively small low-rise buildings of types known to perform satisfactorily in fires. For large and uncommon buildings, detailed probabilistic and engineering-based analysis is necessary and appropriate to design fire safety measures [29,39].

3.4. Fire resistance of structural timber elements When sufficient heat is applied to timber or wood-based composites, a process of thermal degradation (pyrolysis) takes place, thereby, producing combustible gases accompanied by a loss of mass. A charred layer is then formed on the fire-exposed surfaces and the char layer grows in thickness as the fire progresses, thus reducing the cross-sectional dimensions of various elements [32]. The char layer protects the remaining uncharred residual cross section against heat. In order to calculate the resistance of structural timber members exposed to fire, the loss in cross section because of charring, as well as the reduction in strength and stiffness near the charred layer because of elevated temperature must be considered. For timber surfaces unprotected throughout the time of fire exposure, the residual cross section can be calculated by assuming a constant charring rate [40]. The one-dimensional charring rate b 0 is usually taken as

29

3.4. FIRE RESISTANCE OF STRUCTURAL TIMBER ELEMENTS dchar,n dchar,0

Key: 1. Border of residual cross-section (real shape) 2. Border of equivalent rectangular residual cross-section

2

with t = time of fire exposure

1

Fig. 3.2: Charring depth d char,0 for one-dimensional charring and notional charring depth d char,n [40] ) m 40 m ( n , r a h c

d

r o

0 , r a h c

d

30

20

h t p e d g 10 n i r r a h C 0

Key: 1. Relationship for initially unprotected members for charring rate 0 and n

2b 1

d char,0 =25

mm

or d char,n = 25 mm

2a

t f

2. Relationship for initially protected members where charring t ch starts at the same time as the failure of protection t f : 2a. After protection has fallen off charring increased at double rate 0 and n 2b. After char depth exceeds 25mm charring rate reduces to 0 and n

t a

Time t

Fig. 3.3: General description of charring for initially protected timber surfaces according to EN 1995-1-2 when start of charring t ch occurs at the same time as the failure (i.e. fall off) of the fire protective cladding t f (line 2). Line 1 is for initially unprotected timber surfaces

the value observed for one-dimensional heat transfer under ISO-fire exposure of a semi-infinite timber slab. EN 1995-1-2 [41] suggests the value b 0 = 0.65 mm per minute for softwood, as confirmed by experimental studies [42,43]. To take into account the effects of corner rounding and fissures and to simplify the calculation of cross-sectional properties (area, section modulus, and second moment of area) by assuming an equivalent rectangular residual cross section, design codes generally define charring rates greater than the one-dimensional charring rate. These are called notional charring rates b n [41]. Figure 3.2 shows the definition of the charring depth d char,0 for a one-dimensional charring versus a notional charring depth d char,n. For protected timber surfaces, specific charring rates should be applied during different phases of fire exposure [44]. Figure 3.3 gives the simplified model adopted by EN 1995-1-2 wherein the start of timber charring t ch occurs at the same time as the failure t f of the covering cladding. Phase 2a describes the increased charring of timber after cladding has fallen off. The approach assumes that timber charring takes place at double the rate for an initially unprotected surface. The main physical reason for the increased charring rate is that at the time of failure of the cladding the fire temperature will already be high and that there will be no protective char layer to reduce the rate at which the timber continues to get heated [45]. The protection provided by

30


1 2

Key: 1. Initial surface of member 2. Border of residual cross-section 3. Border of effective cross-section

3 d char,n k 0 d 0 d ef

d char, n d 0 k 0 t d ef

notional charring depth zero strength layer: d 0 = 7mm k 0 = 1.0 for t ≥ 20 minutes; /20 for t < 20 minutes k 0 = t time of fire exposure effective charring depth

Fig. 3.4: Definition of residual cross section and effective cross section [40]

the char layer is assumed to grow progressively until its thickness has reached 25 mm. After that, the charring rate decreases to the value for initially unprotected surfaces. For simplicity, a 25 mm criterion is adopted for calculations based on either one-dimensional or notional charring rates. The simplified model can be used for protective claddings made of wood-based panels or wood panelling and normal gypsum plasterboards [40]. Fire reduces the cross section, stiffness and strength of heated timber close to a burning surface, with stiffness and strength decreasing significantly when temperature increases [46,47]. At a temperature of about 200°C wood begins to undergo rapid thermal decomposition. The pyrolysis zone has temperatures between 200°C and 300°C, with the char front being at about 300°C [40,42]. Because of the efficient insulating behaviour of the char layer and the timber, typical temperature profiles exhibit steep temperature gradients through the burning timber members. Temperaturedependent reductions in strength and stiffness near the charred layer can be considered in different ways. EN 1995-1-2 for example gives two methods: the “Reduced cross-section method” and the “Reduced properties method” [40]. The reduced cross-section method is the simplest and accounts for strength and stiffness reductions near the charred layer by adding an additional depth k 0.d 0 (called zero strength layer) to the charred layer depth d char,n (Fig. 3.4). This zero strength layer is assumed to build up linearly with time during the first 20 minutes of fire exposure. This method permits designers to use strength and stiffness properties for ambient temperature for the resulting effective cross section. Thus, the temperature-dependent reduction factor is taken as k mod,fi = 1.0 for the effective cross section. The reduced properties method, by contrast, downwardly adjusts mechanical properties of materials and uses those in combination with realistic estimates of residual cross-section dimensions, making the method relatively complex [40]. On the basis of extensive experimental and numerical analysis novel fire resistance models were developed for load-bearing structural elements. These include timber-concrete composite slabs, cross-laminated timber panels, timber slabs made of hollow core elements, and multiple steelto-timber dowelled connections with slotted-in steel plates. Details of the design models can b e found in reference documents [48–51].

3.5. Design model for the verification of the separating function In order to limit fire spread by guaranteeing adequate fire compartmentalization, elements forming the boundaries of fire compartments are designed and constructed to maintain their separating function throughout the anticipated fire exposure, which places requirements on their integrity

3.5. DESIGN MODEL FOR THE VERIFICATION OF THE SEPARATING FUNCTION

31

( E ) and insulation ( I ). The required period of time is normally expressed in terms of fire resistance using the standard fire exposure and is specified in building regulations. While fire tests are still used extensively for the verification of the separating function of timber substructures like light-frame wall-and-floor assemblies, design models are becoming increasingly common. For the ISO-fire exposure criterion, I (insulation requirement) may be assumed to be satisfied, if the average temperature rise over the whole of the non-exposed surface is limited to 140°C, and the maximum temperature rise at any point of that surface does not exceed 180°C. This prevents ignition of objects in neighbouring compartments. The criterion E (integrity requirement) may be assumed to be satisfied, if no flames or hot gases on the non-exposed side of an assembly can be observed. As I is clearly defined, verification can be made by heat transfer calculations instead of testing. The E, on the other hand, is mostly determined by observations, because calculations are too complex and involve prediction of when cracks will form, hot gas dynamics, and other factors commonly beyond analytical capabilities. Premature integrity failure may occur because of sudden failure of claddings or opening of gaps, which often is dependent on how the material layers are fastened together. Extensive experience with full-scale testing of wall-and-floor assemblies supports rules about the construction detail ing of some wall-and-floor assemblies included in EN 1995-1-2, for example. The integrity criterion may be assumed to be satisfied, if the insulation criterion has been satisfied and panel materials remain fastened to timber frameworks on non-exposed sides of assemblies. In timber buildings, walls and floors are mostly built up by adding different layers to form an assembly. For the verification of the separating function of timber assemblies, additive methods are common for determining combined behaviour of components. With additive models, the fire resistance of layered constructions is obtained by simply summing the contribution of the individual layers. Calculation models for verifying the contributions of the different layers to the separating function of light timber frame wall-and-floor assemblies are used in the UK [52], Canada [17] and Sweden [53]. The current design method according to EN 1995-1-2 (Annex E) is based on the Swedish component additive method. As an enhancement of the Canadian method, the Swedish approach considers the influence of adjacent materials on the fire performance of each layer and, therefore, more realistically describes the fire performance. However, it should be cautioned that the method is based on data from only a limited number of fire tests on wall assemblies and, therefore, only covers a limited range of timber structures. König et al. [54] give a comprehensive review of the related international practices. A comprehensive research project on the separating function of light timber frame wall-andfloor assemblies with cladding made of gypsum plasterboards and wood-based panels was carried out at ETH Zurich, in collaboration with Empa Dubendorf [55,56]. The objective was the development of an improved design model for the verification of the separating function (E and I criteria) of light timber frame wall-and-floor assemblies. A large number of small-scale fire tests permitted the analysis of the influences of material type, thickness, position, and number of the layers on the thermal behaviour of protective cladding made of gypsum plasterboards and wood-based panels. Results of the fire tests allowed the verification and calibration of thermal properties used in thermal finite element (FE) analysis. An extensive FE parametric study enabled calibration of coefficients of a model used for the verification of the separating function of light timber frame wall-and-floor assemblies. The resulting model is capable of considering timber assemblies with an unlimited number of layers of gypsum plasterboards, wood panels, or combinations thereof, with wall cavities that may be eith er empty or filled with rock or glass

32

CHAPTER 3. FIRE DESIGN CONCEPTS 1

4

5

6 Key 1. Timber frame member 2. Panel 3. Void cavity 4. Cavity insulation 5. Panel joint 6. Position of services a–e Heat transfer paths

5

6 c

a

2

e

3 b

d

Fig. 3.5: Possible heat transfer paths through separating multiple layer construction Timber frame member Layer n Layer n–1

Last layer with insulating function

Layers with protective function

Layer i Layer 1

Fig. 3.6: Timber frame wall-and-floor assemblies: numbering and function of the different layers

fibre insulation. The model is based on the additive component method given in EN 1995-1-2. Thus, the fire resistance t ins of the timber assembly is taken as the sum of the contributions from the different layers for the worst possible path of heat transfer (Fig. 3.5), and according to their function and interaction (Fig. 3.6 ): i

t ins = i

= n −1

= n −1

∑ t prot, + t ins,n i

i

(3.1)

=1

with ∑ t prot,i = Sum of the protection values t prot ,i of the layers (in the direction of the heat flux) i =1 preceding the last layer of the assembly on the fire-unexposed side [minute] t ins,n = Insulation value t ins,n of the last layer of the assembly on the non-exposed side [minute]

Protection and insulation values of the layers can be calculated according to Eqs. (3.2) and (3.3), taken into account the values of the layers: t prot ,i = (t prot ,0,i · k pos,exp,i · k pos,unexp,i + ∆t i) · k j,i

(3.2)

t ins,n = (t ins,0,n · k pos,exp,n, + ∆t n) · k j,n

(3.3)

with t prot,0,i = basic protection value [minute] of layer i, see Fig. 3.6


33

t ins,0,n = basic insulation value [minute] of the last layer n of the assembly on the non-unexposed side, see Fig. 3.6 ∆t i, ∆t n = correction time [min] for layers protected by gypsum plasterboards of type F or type × as well as gypsum fibreboards (GF) k pos,exp,i, k pos,exp,n = position coefficient that takes into account the influence of layers preceding the layer considered k pos,unexp,i = position coefficient that takes into account the influence of layers backing the layer considered k j,i, k j,n = joint coefficient

The basic insulation value t ins,0 corresponds to the fire resistance of a single layer without the influence of adjacent materials and joints (i.e. the average temperature rise over the whole of the non-exposed surface is limited to 140°C, and the maximum temperature rise at any point of that surface does not exceed 180°C). The basic insulation value can be assessed by tests or FE thermal analysis. For FE thermal analysis, only the temperature criterion of 140°C is used. The temperature of the layer at the beginning of the analysis on the exposed side and the nonexposed side is assumed to be 20°C. The definition of the basic insulation value t ins,0 is illustrated in Fig. 3.7. Wall-and-floor assemblies with only one layer are rarely used in buildings. Most assemblies consist of two or more layers. The contribution to the separating function of the construction of each layer, except the last layer of the assembly on the exposed side, is mainly protection of the layer(s) below (see Fig. 3.6 ). Therefore, it is appropriate to introduce a basic protection value t prot,0 defined as the time until failure of the protective function. Analogous to the calculat ion for fire protective claddings of load-bearing timber constructions, according to EN 13501-2 [57], the definition of the basic protection value t prot,0 is as illustrated in Fig. 3.8. The testing method for fire protective claddings is performed with a particleboard with a thickness of 19 mm backing the layer studied. The contribution of the cladding to the fire protection of the particleboard may be assumed to be satisfied, if the average temperature rise over the whole exposed surface of the particleboard is limited to 250°C and the maximum temperature rise at any point on that surface does not exceed 270°C. For FE thermal analysis, only the temperature criterion of 250°C is used. The temperature of the layers on the exposed and non-exposed sides at the beginning of an analysis is assumed to be 20°C (Fig. 3.8). Analytical equations for the calculation of the basic insulation value t ins,0 and the basic protection value t prot,0 for different materials have been calculated by FE simulations and verified by fire tests. The position coefficient considers the position of a layer within an assembly, in the direction of the heat flux, because the layers preceding and backing a layer under consideration have an influence on its fire behaviour. The physical meaning of the position coefficient is illustrated in

Considered layer

e n r o u i t t a r u e b p i r t s m i e d T

20°C 20°C

tins,0

20°C + 140°C = 160°C (average temperature rise)

Basic insulation value according to EN 13501-2, 2003

Fig. 3.7: Definition of the basic insulation value t ins,0 using FE thermal analysis

34


19 mm particleboard

Considered layer

n o i t u b i r t s i d e r u t a r e p m e T

20°C 20°C

20°C + 250°C = 270°C (average temperature rise)

tprot,0 Basic insulation value according to EN 13501-2, 2003

Fig. 3.8: Definition of the basic protection value t prot,0 using FE thermal analysis

h h h

3. layer 2. layer 1. layer

(a) Start of fire: t=0

n o i t u b i r t s i d e r u t a r e p m e T

20°C h

3. layer

h

2. layer

20°C 20°C (1. layer falls off) 20°C

n ≥20°C o i t u b i r t ≥20°C s i d e r u t a r e 270°C p m e T

(b) Second layer exposed to fire: t = t prot,1

h

3. layer (2. layer falls off)

n ≥20°C o i t u b i r t s i d 270°C e r u t a r e p m e T

(c) Third layer exposed to fire: t = t prot,1 + t prot,2

Fig. 3.9: Temperature distribution at different times of timber assembly with three layers

Fig. 3.9 for a three-layer assembly. For simplicity, it is assumed that each layer has the same thickness and density and that the influence of joints between layers is neglected. In this case, the basic protection value for each layer is the same (t prot,0,1 = t prot,0,2 = t prot,0,3). The first layer is directly exposed to fire and backed by the second layer. The temperature of all layers at the beginning of the fire on the exposed side and the non-exposed side is assumed to be 20°C (Fig. 3.9a). The contribution of the first layer to the total fire resistance is defined as t prot,1. The position coefficient k pos,1 of the first layer can be described as the ratio t prot,1 to t prot,0,1 and depends on the layer backing the first layer. The second layer is protected by the first layer. It is conservatively assumed that, after failure of the protection provided by the first layer that the second layer is directly exposed to fire, with failure occurring when temperature at the interface between the first and second layers reaches 270°C time t = t prot,1. The main differences in comparison with the initially unprotected first layer is that the temperature in the fire compartment is already at a high level and that no protective char layer exists [45]. Further, the temperature of the second layer on the fire-exposed side is 270°C, as defined previously, while the temperature on the non-exposed side is equal or greater than 20°C depending on the thickness of the second layer and the material preceding and backing the layer (Fig. 3.9b). For these reasons, the contribution of the second layer to the total fire resistance is lower than the contribution of the first layer, that is, t prot,2 < t prot,1. The position coefficient k pos,2 of the second layer can be described as the ratio t prot,2 to t prot,0,2 and is < 1.0. For the same reasons, EN 1995-1-2 assumes that after failure of a cladding, charring occurs at the increased rate of initially unprotected surfaces [41]. The third layer is protected by the second layer. After failure of the protection by the second layer at 270°C time t = t prot,1 + t prot,2, the third layer is directly exposed to fire.


35

The third layer is the last layer in the assembly and takes an insulating function (Fig. 3.6 ). Therefore, for that layer, the temperature criterion of 140°C is applicable, and an insulation value of t ins,3 must be calculated. Because of the further increased temperature in the fire compartment and the missing protective char layer, as well as the temperature rise criterion of 140°C, instead of 250°C, the contribution of the third layer to the total fire resistance is lower than the contributions of the first and second layers, that is, t ins,3 < t prot,2 < t prot,1. The position coefficient k pos,3 of the third layer can be described as the ratio t ins,3 to t ins,0,3 and is < 1.0. The fire resistance of the entire timber assembly is the sum of the contributions from the different layers, that is, t ins = t prot,1 + t prot,2 + t ins,3. The influence of the layers preceding and backing the layer considered is analysed separately. In that case, the position coefficient k pos,exp accounts for the influence of the layers preceding a layer considered, whereas the influence of the layer backing the layer studied is considered 1.0 by k pos,unexp . The position coefficients calculated using FE simulations showed that the position 0.8 coefficient k pos,exp is mainly influenced by the fire k temperature at the time when a layer being con t n 0.6 e sidered is exposed directly to the fire, as opposed i c i f f to the thickness of the layer itself. The influence e o c 0.4 of the preheating was shown to be small. The n o 25 mm i design model for the verification of the separat t i s 18 mm o ing function of timber assemblies was developed P 0.2 15 mm 12.5 mm for the ISO fire exposure and is a function of the 10 mm sum of the protection values of the preceding 0 0 10 20 30 40 50 60 layers (i.e.∑ t prot,i − 1). The thickness of the layer Protection time Σtprot,i-1 (min) considered is expressed as the basic protection value t prot,0,i or basic insulation value t ins,0,n. As an Fig. 3.10: Position coefficient k pos,exp of example, Fig. 3.10 shows the position coefficient gypsum boards with different thickness k pos,exp of gypsum boards with different thickas a function of the sum of the protection nesses, as a function of the sum of the protection values of the layers preceding the gypvalues of the layers preceding the gypsum board sum board (i.e. ∑ t prot,i − 1) (i.e.∑ t prot,i − 1). The position coefficient k pos,exp decreases with increasing ∑ t prot,i − 1, because the gypsum board is protected longer and at higher exposure temperatures. The position coefficient k pos,exp decreases with reducing thicknesses of the gypsum board. Similar effects are observed for wood-based panels and batt cavity insulation, making it possible to determine the position coefficient k pos,exp as a function of the sum of the protection values of the layers preceding a layer (∑ t prot,i − 1) and the basic values of a layer considered (t prot,0,i and t ins,0,n). Calculation of position coefficient k pos,exp is therefore simple. p x e , s o p

The influence of a layer backing the layer of interest is defined by the position coefficient k pos,unexp. Results of fire tests, Fig. 3.11, supported by FE simulations show that the influence of a backing layer is small, if the backing layer is made of gypsum or timber. Also, batt insulation backing layers cause other layers to heat up more rapidly reducing the protection time of the layer, and therefore the position coefficient k pos,unexp should be evaluated for the case of batt insulation backing. For timber or gypsum backing assuming k pos,unexp,i = 1.0 is conservative.

36

CHAPTER 3. FIRE DESIGN CONCEPTS 25 to 26 min

28 min

Insulation

Cavity

30 to 32 min

30 to 33 min

Timber GF

GF

Air GF

GF

Fig. 3.11: Time taken to reach the temperature rise of 250°C (average) and 270°C (at any point) on the fire-unexposed side of 15 mm thick GF tested as a single board or backed with different materials

The methods described in this chapter extend the range of application of EN 1995-1-2 approaches significantly, with further details given elsewhere [55,56].

3.6. Fire design concept for tall timber buildings This section uses Swiss building regulations to illustrate application of prescriptive fire safety concepts for load-bearing and separating elements in residential, office, and school buildings. Requirements depend on the number of above-ground storeys in a building superstructure, as shown in Table 3.1. In practice, as shown by the requirements in the table, timber superstructures are allowed in residential buildings up to six storeys. The requirement R60/EI30(nbb) or EI60/EI30(nbb) means that the load-bearing and separating building elements shall have fire protective, non-combustible cladding, with a fire resistance ( R value) of the cladding itself of 30 minutes. If sprinklers are present, the requirements are relaxed. For light-frame multi-storey structures, all floors and nearly all walls are load-bearing elements. Therefore, timber framing materials are not allowed to be consumed at all on the fire-side of an assembly until the R value is achieved. Consequently, the presence of non-combustible material on interior surfaces is a universal requirement for buildings with more than four storeys. Figure 3.12 shows examples of cross sections with fire resistance REI60/EI30(nbb). In Fig. 3.12a, the claddings on either side of the timber framing are EI60. This means that the interior of the timber structure need not be designed for a specific fire resistance rating, as attainment of the required performance is guaranteed by the claddings alone. The solution in

Storeys

1

Load-bearing elements

–

Separating elements

2

3

Design for normal R30 temperature (cold design)

EI30 EI30

4 R60

5–6

7–8

R60/ R60(nbb) b EI30( nbb )

Tall buildingsa R90(nbb)

EI30 EI60 EI60/ EI60(nbb) EI90(nbb) EI30(nbb)

a

Tall buildings (high-rise) are defined as buildings with a total height of more than 25 m or with the top floor located at the height of more than 22 m above the level of the terrain used by firefighters. b

(nbb) means non-combustible material.

Table 3.1: Fire requirements for load-bearing and separating elements in residential, office, and school buildings designed based on the structural concept depending on the number of storeys [31]

37

3.6. FIRE DESIGN CONCEPT FOR TALL TIMBER BUILDINGS Cladding EI30(nbb)

Cladding EI60

R60/EI30(nbb)

(a)

EI60/EI30(nbb) REI60/EI30(nbb) Cladding EI30(nbb)

Cladding EI60

Cladding EI30(nbb)

Insulation, melt point ≥ 1000°C

R60/EI30(nbb)

(b)

EI60/EI30(nbb) REI60/EI30(nbb) Cladding EI30(nbb) Cladding EI30(nbb) R60/EI30(nbb)

(c)

EI60/EI30(nbb) REI60/EI30(nbb) Cladding EI30(nbb)

Fig. 3.12: Examples of cross sections with fire resistance REI60/EI30(nbb) Fig. 3.12a will therefore, ensure a fire resistance longer than 60 minutes. In Figs. 3.12b and 3.12c, the claddings on either sides of the timber framing are EI30, in which case the fire resistance of the timber structure shall be at least R30. Notably, the presence of insulation with a melting point ≥ 1000°C inside wall cavities ( Fig. 3.12b) can improve the fire resistance, but only if the insulation remains in place after the failure of the cladding on the side exposed to fire. Besides requirements related to use of combustible material and fire resistance of building elements, the fire regulations include mandatory rules for the design of escape routes consisting of corridors and staircases, emergency exits, and necessary organizational and technical measures such as smoke detectors and alarm systems, sprinkler systems, and smoke exhaust systems. All such additional measures are required, irrespective of the type of construction materials used.

Basic safety objectives for prescriptive requirements are summarized in Table 3.2 for mid-rise (five or six storeys) residential buildings, with those objectives reflecting the assumption that occupants can leave buildings or be evacuated by fire brigades in case of fire. The limited nature

Type of building

Evacuation of people during fire

Fire spread to other parts of building

Building collapse

Mid-rise Feasible buildings

Acceptable after a Acceptable after a defined period of time defined period of time

Tall Not reliably feasible – people buildings should stay in safe places until burn-out

Not acceptable

Not acceptable

Table 3.2: Main differences in acceptable behaviour between mid-rise buildings and tall buildings in case of fire

38


of the number of storeys and building heights is also crucial in respect of firefighting actions, with it being assumed that a fire will usually be fought from the outside of a building. The Swiss regulations recognize that the fire safety objectives adopted for mid-rise residential buildings can be achieved via prescriptive requirements, even when combustible structural materials are used. The same applies in most other countries where use of timber is common.

3.6.1

Main differences between mid-rise and tall buildings with regard to fire safety

Because of the height of tall buildings, occupants located on upper levels need considerable time to leave the building, in the event of fire. Additionally, firefighters must be able to fight fires from within buildings, which means that it can also take considerable time for them to reach the fires. Additionally, some escape routes may become blocked and evacuation of building occupants may be by circuitous routes, or in extreme cases evacuation rescues may not be possible at all. For these reasons, fire safety concepts for tall buildings are based on the scenario that occupants located in upper parts of buildings cannot leave during fires. Additionally, it is assumed that fires cannot be extinguished and will continue until all combustible material in any affected fire compartment(s) has burned (i.e. full burn-out). On the basis of this scenario tall buildings must comply with fire requirements that are more rigorous than for mid-rise buildings (Table 3.2). Tall composite buildings of types discussed in Chapters 7, 9 and 11 must be treated for fire in the same way as conventional tall buildings. In such cases requirements for building elements are as follows: •

Separating building elements shall be designed in ways that sustain a full burn-out, thereby preventing uncontrolled spread of fire to other parts of buildings throughout the duration of a fire.

•

Load-bearing building elements shall be designed in ways that prevent their structural collapse during full burn-out without intervention of the fire brigade.

Consequently, tall buildings in which timber is a primary construction material should be designed in ways that occupants will survive full burn-out of the fire compartment(s) in which a fire starts, whereas other parts of the building remain undamaged. On the basis of this scenario and assuming that timber will not necessarily self-extinguish, most building codes do not permit the use of combustible materials in tall buildings either for the structure or for room linings.

3.6.2

Is it still possible to design a tall building using timber as structural material?

One way to fulfil the requirements for separating and load-bearing building elements is to protect combustible structural elements like timber members by non-combustible materials that remain in place throughout a fire’s duration. This is the simplest approach by which rooms of fire origin can complete burn-out without the structural and separating timber elements charring. Sufficient protection can be achieved using non-combustible claddings like layers of gypsum plasterboard. This is widely known as the “building encapsulation” approach. Building elements with fire resistance R60/EI30(nbb) or EI60/EI30(nbb) (as required in Switzerland for residential

39

3.7. EXAMPLE OF TALL BUILDING PROJECT

timber buildings of five and six storeys) can be considered as a partial building encapsulation method (i.e. timber elements have fire protective non-combustible claddings with a fire resistance of 30 minutes, Fig. 3.12). Here the classification of “partial” reflects that in many cases an R-value of 30 minutes is insufficient to prevent the start of charring of timber structures during a complete burn-out. Another way of guaranteeing effective compartmentalization is the use of composite elements. For example, timber-concrete composite slabs can be designed in ways that enable them to efficiently carry design loads applicable to normal and fire design situations, while at the same time being effective fire barriers. Use of technical measures like sprinklers is possible, as a way of reaching design objectives by preventing fires from developing. However, many building regulatory authorities are reluctant to accept sprinkler concepts alone for fire protection of large and tall buildings. In some instances, approaches acceptable to regulatory authorities combine the passive ability of buildings to compartmentalize fires along with sprinkler systems. Such solutions are commonly known as “Special Solutions” or “Alternative Solutions” and require specialist fire engineering as their basis.

3.7. Example of tall building project The “Dock Tower” project investigated the technical feasibility of a tall timber residential building having an above-ground height of 120 m [58]. That tall building has a mixed/multi-material construction, as shown in Fig. 3.13. The fire safety concept is based on the creation of primary fire compartments around a central reinforced concrete (RC) core, along with four external

Staircase with controlled ventilation

Multi-use room

Water mist system

Staircase with natural ventilation Shaft

Elevator Staircase Corridor

Water mist system

Multi-use room Projecting concrete slab

Staircase with controlled ventilation

Definition of use of rooms

Definition of fire safety measures

Fig. 3.13: Study on technical feasibility of a tall timber building of 120 m

40


staircases encapsulated in RC combined with projecting RC fire floors every three storeys. The projecting concrete slabs have the function of preventing fire propagation across the building facade. Residential apartments form secondary fire compartments with timber-concrete composite floor slabs and timber walls designed according to the requirement of burn-out. The building has five staircases in total, each placed as far away from each other as possible. In addition, every apartment has direct access to at least two escape routes. Furthermore, two of the exterior staircases are open to the outside environment, whereas the other two are pressurized to avoid the entry of smoke. In order to control and extinguish a fire in an early stage, all building rooms are equipped with a water mist sprinkler system. Activation of the water mist system is temperature-actuated or controlled by a fire alarm system. In such a system, the nozzles break the water down into very small drops creating a cooling and smothering water fog, which does not allow a fire to persist. An additional measure is that the building has two high-pressure water mist fire hydrants on each floor of the central core. Organizational measures were also designed to enable safe evacuation and firefighting. The project shows the feasibility of using structural measures in combination with the technical ones (water mist system, alarm signal, etc.) and organizational measures to enable construction of tall timber building that have fire safety levels higher than the standard measures for high-rise buildings. Essential to this is that measures taken implement an approach of providing redundancy in each aspect of the fire design.

3.8. Experimental studies The fire performance of timber structures under natural fire conditions has been studied experimentally in full-scale [59–61]. The objectives of such tests are to verify the efficiency of different fire safety concepts for multi-storey timber buildings and to find possible vulnerabilities.

3.8.1

Fire performance of timber structures under natural fire conditions

In the Swiss project by Frangi and Fontana [60] efficiencies of technical measures, especially automatic fire detection and fast response sprinkler systems, were studied (series BE ). A second series (series BÜ ) of full-scale tests was carried out to look at the ability of structural fire safety measures to limit fire spread until full burn-out. In that second series, the sprinkler system was turned off, and the window was opened so that the fire was able to grow quickly, as it was supported by a large air supply. Special attention was given to the fire propagation across the facade and the influence of combustible surfaces on the severity of the fire. As part of that testing, four room modules (H1, H2 and G1, G2) were prefabricated off-site with light-frame timber construction. Each module was 6.6 m long, 3.1 m wide, and 2.8 m high and had a window (opening 1.5 m × 1.7 m) with a standard double layer of insulation glass. The modules were identical apart from the wall and ceiling linings. For modules H1 and H2, combustible wood-based panels (oriented strand board) were used as wall and ceiling li nings. In contrast, modules G1 and G2 had one to three layers of non-combustible gypsum plasterboard wall and ceiling linings. The combustible floor was made of a light timber frame construction/ timber hollow core elements and was covered by a thin layer of linoleum. All modules were equipped with an automatic fire detection system having four different sensors, as well as two sprinkler systems. During the test fires, the contents (mobile fire load), as well as the combustible construction materials contributed to the total fire load. Each module contained a mattress made

3.8. EXPERIMENTAL STUDIES

41

of polyurethane foam material and 11 wooden pallets, as additional mobile fire loads. The total fire load density (calculated over the floor area) for the modules with non-combustible wall and ceiling linings varied between 363 and 366 MJ/m2, and for the modules with combustible wall and ceiling linings the total fire load density was approximately 855 MJ/m2.

3.8.2

Results of sprinklered fire tests

The series BE with activated detection and sprinkler systems showed that the sprinkler system was able to control the fire within the shortest time, even though the mattress was ignited from below. Measured activation times for the sprinkler systems varied between 2 and 3 minutes after ignition. Those tests also showed that the ventilation conditions (window opened or closed) did not substantially influence the sprinkler activation. All detection algorithms of the automatic fire detection system discovered the fire within 2 minutes (i.e. about 1 minute earlier than the sprinkler system deployed). At the time of sprinkler activation, room temperatures at different locations varied between 50°C and 200°C. As flashover can only occur at higher temperatures, the presence of combustible room linings had no influence on the severity of fires during its early stages. This reflected that in all experiments, the sprinkler system extinguished the fire before it could spread within the compartment. Figure 3.14 shows damage to a mattress, wall, and floor lining. The damage was limited because of the fast sprinkler system activation. The series BE tests confirmed that if fast response sprinkler systems are deployed, the influence of having combustible compartment linings can be offset adequately to achieve fire safety objectives. As mentioned in Section 3.7 , particular building regulatory authorities may or may not deem the use of sprinklers satisfactory for large and tall buildings.

3.8.3

Results of non-sprinklered fire tests

The series BÜ tests with deactivated sprinkler systems were performed using two stacked modules (i.e. one placed above the other and designated “lower” and “upper” herein). Table 3.3 summarizes the most important results of those tests. After ignition, fire grew very rapidly in all tests, and the temperatures rose to flashover condition within only a few minutes. For the modules with combustible wall and ceiling linings, flashover occurred after about 4 minutes. For modules with non-combustible wall and ceiling

Fig. 3.14: Damages on the mattress (left) and on the wall and the floor (right)

42 Fire test Modules Window


BÜ nbb

BÜ bb

BÜ nbb demo

Lower: G1 Upper: H2 Lower: H1 Upper: H2 Lower: G2 Upper: H2 Opened

Ignition time ca.01´30˝ of mattress Flashover ca.06´00˝ Failure time of – exterior glass layer of the window in the upper module Failure time of — interior glass layer of the window in the upper module Sprinkler acti- 02´15˝ (air) vation time on the ceiling Sprinkler 02´20˝ (air) activation time on the wall End of fire 44´15˝ test (sprinkler activation)

Closed

Opened

Closed

Opened

Closed

–

ca.01´40˝

–

ca.01´40˝

–

— 13´57˝

04´27´ –

— 06´09˝

06´58´ –

— 14´25˝

42´35˝

—

07´28˝

—

40´16˝

42´40˝

03´20˝ (air)

—

02´35˝ (air) 42´30˝ (air)

42´41˝

03´27˝ (air)

07´30˝

02´44˝ (air) 41´21˝ (air)

42´40˝

18´53˝

07´30˝

59´01˝

59´37˝

Table 3.3: Main results of the fire tests BÜ with sprinkler system turned off

linings, flashover occurred after 6 to 7 minutes. In full-scale fire tests by the Technical Research Centre of Finland where different timber compartments with and without gypsum plasterboard protection of timber substructures were tested, flashover occurred between 4 and 6 minutes [62], thereby confirming the results of the series BÜ tests. For the module with combustible wall and ceiling linings, the fire spread through the window to the exterior. Following this, flashover was much more severe than for the modules with non-combustible wall and ceiling linings (Fig. 3.15). This was confirmed by observations of the facade with an infrared camera. In the BÜ nbb demonstration test, the interior glass layer of the window of the upper module failed after about 40 minutes. In the BÜ bb test with combustible linings, the heat flux from the flames emerging from windows higher on the facade was much greater. Consequently, fire spread could not be significantly delayed by merely placing a 250 mm protruding ledge made from 1 mm thick steel sheet placed on the facade between the lower and the upper modules (Fig. 3.15). In the bb test, the interior glass layer of the window of the upper module had already failed within 7 minutes of ignition. Owing to excessive flaming in the BÜ bb test with combustible linings fire could not be stopped for about 20 minutes.

3.8. EXPERIMENTAL STUDIES

43

Fig. 3.15: Fire development 7 minutes after fire ignition; left: lower module with combustible linings, right: lower module with non-combustible linings

The BÜ nbb test with non-combustible linings made of one layer gypsum plasterboard, was stopped after about 45 minutes and only after the interior glass layer of the window of the upper module had failed. By that time, all mobile combustible material was burnt out. However, the fire continued because of charring of the wood-based fibreboards placed behind the gypsum plasterboard, which started to fall off after about 30 minutes. The BÜ nbb demo test with noncombustible linings made of two or three layers of gypsum plasterboards was stopped after about 60 minutes when no more flame emerged from the window opening and the fire inside the lower module was in the later stages of decay. Because of the fire protection provided by the gypsum plasterboard, the light timber framing in the walls and ceiling was not damaged, even when complete burn-out was achieved. Figure 3.16 shows the room temperatures measured on the ceilings of modules at the front (near the window) and rear. Temperatures at the rear of the module were lower than at the front because of the limited amount of oxygen away from the open window. Also, no significant differences were observed in the early stage and peak temperatures for the modules with and without combustible wall and ceiling linings. This confirms that pyrolysis gases released by combustible wall and ceiling linings were only partially burnt. As unburned pyrolysis gases flowed out of the window, they caused intense combustion outside where oxygen was freely available. Other tests have yielded similar results [62].

Full-scale tests confirmed that by using purely passive structural measures, fire may be contained to a single room within a timber structure. The BÜ nbb demo test in which timber framing was protected by three layers of gypsum plasterboards on the ceiling and two layers of gypsum plasterboards on the walls is explicit proof that in multi-storey buildings, complete burn-out of a lower room/module can occur without significant damage to the timber structure

44

CHAPTER 3. FIRE DESIGN CONCEPTS 1400 1200

) C ° ( 1000 e r u t 800 a r e p m 600 e r m o 400 o R

Test BÜ nbb front Test BÜ nbb rear Test BÜ bb front Test BÜ bb rear

200 0

0

10

20

30

40

50

Time (min)

Fig. 3.16: Room temperatures measured on the ceiling in the front as well as in the rear of the lower modules for the fire test BÜ nbb with non-combustible linings as well as the fire test BÜ bb with combustible linings (the temperature of the test BÜ bb front are not complete because of a loss of electrical power occurred during the fire test after about 10 minutes)

or fire propagation to an upper room. Even smoke damage to upper rooms was limited until the window broke in the room with the fire.

3.9. Additional comments For tall buildings, fire safety is a very important yet controllable aspect of design, even when combustible materials are used. In that context, it is fundamentally important to address differences between how high-rise buildings can be evacuated compared with low or medium-rise buildings and how fires in buildings with different heights can be fought. No assumption should be made that people in tall buildings can be speedily evacuated or that fire brigades can use external firefighting equipment. The most appropriate design concept for tall buildings, irrespective of the construction materials employed, is to create fire compartments in which burn-out will occur without infringing on the structural stability of adjacent compartments or the complete system. Such practice is analogous to the provision of structural robustness as discussed in Chapter 2. Special measures related to using timber and wood-based composites in construction of superstructures relate to the need to encapsulate such materials within protective layers of non-combustible material, and/or to use composite mixed material construction. Studies show that hybrid tall buildings that use timber in combination with RC, or potentially fire protected frameworks of steel and other materials, are entirely feasible in terms of technical and organizational measures. As the example of the Dock Tower project discussed in Section 3.7 illustrates, tall buildings incorporating timber as a primary construction material can be expected to have at least equal to, and potentially better, fire performance than typical traditional non-combustible tall buildings today. As with all other aspects of design and construction, finding acceptable solutions depends on equipping engineers with knowledge and tools with which to exploit the potential of fire engineering. This discussion is aimed at alerting them that the necessary knowledge and tools exist.

45 Chapter

4

Durability Design Concepts

Summary: Durability problems in buildings are always preventable through proper design, construction, and maintenance practices. Timber construction elements in buildings can have indefinite lifespans if they are properly protected against hazards to the durability. Primary threats are biotic decay or insect damage to timber parts and corrosion of metal parts. The exact natures of the threats can vary across and within geographic regions, as a function of climate and human interventions within ecosystems. Critical elements within structural systems should never be put at risk of attack from mechanisms that degrade timber or corrode metal parts, because primary structural systems of buildings must always remain fully functional. However, it may be an acceptable design decision to expose replaceable, non-critical structural and nonstructural timber elements, as part of life cycle cost optimization or architectural decisions. Designing building elements so that they have appropriate durability is crucial to the implementation of construction systems that contain timber. The durability design of multi-storey buildings discussed in this Structural Engineering Document (SED) is as important as their having proper structural and fire design, and must be similarly addressed at both the system and the element levels. Attack mechanisms, risk mitigation methods, and the estimation of lifespan expectations for non-critical structural and non-structural timber elements are discussed in this chapter.

4.1

Introduction

This chapter addresses durability design of timber components in building superstructures from the perspective of risk of material degradation over certain lifespans. This is required as a tool for making initial material selection, planning building maintenance, designing replacement programmes, and optimizing life cycle costs. Even though material selection decisions are ultimately made at the building element/component level, degradation hazards are determined primarily at the building system level. Thus, durability design depends on establishing building element performance requirements based on their functional purposes. Herein, the discussion of durability requirements is based on linking assessment of risk to element utilization classes: •

non-structural elements,

•

non-critical structural elements, and

•

critical structural elements.

46

CHAPTER 4. DURABILITY DESIGN CONCEPTS

From an engineering perspective, the supporting logic to the ut ilization classes is that buildings can be designed such that non-structural and non-critical structural components are repairable or replaceable. In such cases, designers should be allowed to make design decisions that trade repair and replacement costs against the expected durability risk. However, critical structural elements should not deteriorate and must last until a building is decommissioned. Crucial to this is the recognition that, in the context of durability, the applicable performance requirement is integrity of the primary structural system throughout a building’s actual lifespan. Specifically this means that critical structural parts of buildings should not decay, corrode, or be otherwise structurally compromised during normal service with respect to what the structural codes assume as their available capacities. Within this concept, the question arises as to what constitutes a critical structural element. There will rarely be precise answers to this. However, a suitable definition may be one whose failure could lead to the loss of life, blockage of emergency escape and access routes, or loss of critical equipment (e.g. emergency electrical generators). Such a definition can be translated for implementation into requirements such as “a critical structural element is one whose failure would cause the collapse of an area of building wall, floor or roof of more than 50 m 2, or would sever access to any portion of a building having a floor area greater than 50 m 2”. Herein adoption of 50 m2 is purely illustrative and may differ significantly from applicable requirements in specific regulatory jurisdictions. The remainder of this chapter discusses quantitative approaches by which engineers can estimate the likely service life of a timber or wood-based composite and affiliated fasteners and hardware that connect such material to other structural parts. Although the presented approach is applicable to only common, naturally occurring deterioration hazards, the conceptual principles can be extended to rarer or human-induced hazards. With such tools, engineers can rationally select materials for non-structural, non-critical structural, and critical structural parts of particular buildings.

4.2

State-of-the-art

Much scientific and design information exists on durability of generic and proprietary timber and wood-based construction materials, including thermo-mechanically and chemically modified wood, and products created by fractionating and reconstituting wood in combination with natural and synthetic adhesives. In Australia, for example, where a vast range of timber species are used in construction, timber species and preservation treatments are grouped into classes according to durability expectations [63]. However, such lists reflect how materials perform under standardized assessment conditions, rather than providing direct indications of deterioration risks that can exist in specific construction situations. Consequently there is still a need for more direct performance criteria for durability of building elements, as addressed by international and national standards and best practice guidelines [64–66]. Standards and guidelines that exist set a framework for establishing target performance durability levels for building elements but do not give specific guidance relevant to design practice. The discussion below aims to fill that void. The ability of engineers to apply concepts to practice depends on access to robust and consistent approaches for making building element service life predictions under defined service

47

4.2 STATE-OF-THE-ART (a)

(b)

D

DARWIN

D

DARWIN

B

Cairns Broome

Broome

Cairns

Townsville

Townsville

Mount Isa

Port Hedland

A

Alice Springs

Mount Isa

Port Hedland

B

C

Alice Springs

A

Bundaberg

Bundaberg

Roma

Roma BRISBANE

BRISBANE

Kalgoorlie

Kalgoorlie Coffs Harbour

Coffs Harbour PERTH

PERTH

Dubbo

ADELAIDE

ADELAIDE

CANBERRA ZONE A

ZONE C

ZONE B

ZONE D

Dubbo

C

SYDNEY

MELBOURNE

ZONE A

ZONE C

ZONE B

ZONE D

SYDNEY CANBERRA

MELBOURNE

HOBART

HOBART

(c)

(d) DARWIN

DARWIN

F Cairns

Cairns Broome

Broome

D

Townsville

Townsville Mount Isa

Port Hedland

Mount Isa

Port Hedland

Alice Springs

Alice Springs

Bundaberg

Bundaberg

C

Roma Roma

Roma BRISBANE

BRISBANE Kalgoorlie Kalgoorlie

Kalgoorlie

PERTH ADELAIDE

Dubbo Dubbo

PERTH

CANBERRA ZONE A

ZONE C

ZONE B

ZONE D

Coffs Harbour

Coffs Harbour

B

SYDNEY

C

D

ADELAIDE

Dubbo SYDNEY

E

CANBERRA MELBOURNE

MELBOURNE

B A

A HOBART

HOBART

(e)

(f)

A

DARWIN

C

DARWIN

Cairns

Cairns

Broome

Broome

B


Port Hedland


Port Hedland

D

Alice Springs

Alice Springs Bundaberg

A

Roma

B

Bundaberg

Roma BRISBANE

BRISBANE

Kalgoorlie

Kalgoorlie Coffs Harbour

PERTH

G

E

ADELAIDE

Coffs Harbour PERTH

Dubbo

ADELAIDE SYDNEY CANBERRA MELBOURNE

ZONE A

ZONE C

Dubbo

C

SYDNEY CANBERRA

MELBOURNE

ZONE B

HOBART

HOBART

Fig. 4.1: Hazard zones for Australian durability attack mechanisms [68]: (a) in-ground decay; (b) above-ground decay; (c) termite attack; (d) marine borer attack; (e) atmospheric corrosion; (f) embedded corrosion (courtesy of CSIRO)

48


situations. More specifically, the requirement is for models that enable calculation of the strength and stiffness that structural elements are expected to have after a certain time in service, or how long non-structural elements will remain functional for intended purposes under defined threats and particular applications. Such models have been developed for timber construction in Australia [67,68]. Those models were developed to predict decay from fungi, marine borers and termites, and corrosion of steel parts. Using the resulting attack models, a draft engineering design code has been developed [69–71]. The draft Australian design code relates structural load capacity to target design life and gives design procedures intended to generate design solutions with comparable levels of structural reliability under all hazard situations. Figure 4.1 shows the hazard zones of Australia associated with various attack mechanisms. As structural engineers will recognize, the approach is in principle similar to mapping of expected design wind speeds or seismic ground accelerations. Obviously, the principles developed in Australia could be applicable to other parts of the world or to non-timber materials, but so far that has not been done. However, by analogy engineers should be able to make extrapolated judgements applicable elsewhere. Design lives of non-structural and non-critical structural elements reflect how often they are intended to be repaired or replaced, and there is no reason to match lifespans of such elements to one another or to design lives for critical structural elements. The distinction made here speaks to providing designers with options for life cycle cost optimization and not to circumstances under which they might avail themselves of such an option. As already indicated, critical structural elements must have design lifetimes associated with actual expectations of building lifespans based on reasonable expectations of lifelong maintenance and evolution in building occupancies. There is of course uncertainty in durability design, especially whether proper maintenance will be done, but the situation here is in principle no different than for structural and fire design which presumes that buildings will not be altered in ways that compromise the original design intent.

4.3

Attack mechanisms

4.3.1

Mould

Moulds do not affect the structural performance of timber and wood-based composites, but their occurrence in indoor environments may pose health threats. Mould on timber usually occurs if the surrounding air temperature exceeds 20°C combined with relative humidity in excess of 80% for more than brief periods. At lower air temperatures, only very high humidity levels will cause mould on timber.

4.3.2

Decay

At temperatures between 0°C and 60°C decay fungi attacks can progress rapidly when the timber has a moisture content above the fibre saturation point (≥ 30%) [72]. Therefore, such conditions should never be deliberately created internally or externally. The existence of moisture conditions in timber that sustain decay fungi are always the result of bad initial design, poor workmanship, inappropriate building renovation methods, unintended uses of buildings, and/or improper building maintenance. Excessive moisture in timber is often associated with

49

4.3 ATTACK MECHANISMS

(a)

(b)

(c) Rain

Rain Wood based Material

Gap

Rain

Check

Decay pocket Decay pocket Decay pocket

Impermeable skin

(d)

(e)

Wood sucks up water from soil via footing

Masonry wet by rain

Dry air Sandstone foundation

Moist soil

Wood absorbs moisture from masonry

Fig. 4.2: Avoidable mechanisms for moisture accumulation in traditional timber low-rise construction: (a) direct wetting by rain; (b) water ingress via surface check; (c) water ingress via construction joint; (d) moisture wicking via stone foundation; (e) moisture wicking via masonry contact (courtesy of CSIRO)

capillary action (also known as wicking) and/or water entrapment with water coming from the ground, rain, or leaks within buildings. Thus, decay prevention must focus on practices th at keep timber dry. Figure 4.2 illustrates mechanisms by which moisture typically enters structural and non-structural elements of traditional timber structures. These examples illustrate that in all cases the problems result from poor or shoddy practices. Site investigations of buildings always reveal that best practices were not adopted during at least one stage in either design or maintenance. Under proper practices, wetting of timbers should never occur except under rare or accidental events such as damage to glazing, spillage, or leaks.

This SED does not recommend impregnation of timber with chemical preservatives as a primary means of mitigating risk of decay in critical structural elements, because such treatments have not been proven reliable over the long term. Instead, the recommended practice is to employ durable design strategies at system and element levels that eliminate the hazards. In practice, this recommendation translates into designing the building superstructure (and foundation, where applicable) systems so that water is drained away from timber members and to ensure that intentional spaces are ventilated. Elements should be designed to have geometries that do not entrap moisture. This does not mean that exposing non-structural or non-critical structural elements to risk of decay is not an acceptable design decision. However, any replaceable

50


elements that are allowed to decay should not be directly attached to critical structural elements, should be separated from them by impermeable barriers (e.g. vapour membranes), and should be intentionally selected as sacrificial. Models exist for estimating service lifespans of elements based on measures such as the time taken for non-structural elements to develop a certain depth of surface decay under a particular external service use, or the time for a certain depth of surface decay of non-critical structural elements in contact with the ground [73]. These should be used where appropriate.

4.3.3

Termites

Termite attacks are possible in all ecosystems where the mean annual temperature is greater than 10°C, with speed of travel and ferocity of attacks tending to accelerate with any increase in temperature. Boundaries of termite infestations can advance at around 1 or 2 km per year. Therefore, maps showing areas of infestation are only rough indicators of whether a termite hazard might exist within the design life of a specific building (Fig. 4.1c). Accurate estimates of possible termite arrival dates can be made using termite attack models and up-to-date i nfestation data as discussed by Leicester et al. [74]. Although unlikely, termite attack may exist where the mean external temperature is below 10°C. Termites have spread northward in both North America and Europe, because they are sometimes transported along with traded commodities (e.g. in packing cases). For example, viable colonies exist in the core areas of several Canadian cities because metro systems and other underground spaces such as pedestrian walkways provide year-round conduits for termite movement under sufficiently warm conditions [75]. A rarer possibility also exists that termites get blown by wind onto the roofs of high-rise buildings during their alatory phase. They then establish colonies where rain water has penetrated building envelopes, as has been observed in Hawaii and elsewhere [76].

4.3.4

Corrosion

Corrosion attacks unprotected metal fasteners inserted into timber in two ways (Fig. 4.3). In the first, airborne salt attacks exposed parts of fasteners. In the second, the acidity of any impregnated chemicals for biological or fire protection and natural acidity of timber itself become agents of attack on the embedded parts of fasteners. In the case of joints with bolts fitted in tolerance holes, the corrosion rate of shanks is enhanced by any salt and water that accumulate within oversized holes (Fig. 4.3c). This type of attack may be mitigated by protection measures such as covering a shank with a plastic sleeve, or using tight-fitting dowels instead of loose-fitting bolts. Notably, using tight-fitting dowels, instead of bolts in tolerance holes also has structural benefits in terms of greater stiffness and strength. Practical models are available for estimation of rates at which exposed and embedded parts of steel fasteners in timber will corrode [69,77,78]. The corrosion rate of embedded fasteners tends to be quite low, unless the timber has been treated with a mineral-based chemical, such as alkaline copper quaternary (ACQ), or if the timber has high moisture content [79]. For the case of atmospheric corrosion, the rate for mild steel can be very high for buildings located within 1 km of saltwater (Fig. 4.4). Corrosion rates

51

4.4 DESIGN STRATEGIES

(a) Atmospheric corrosion

Embedded corrosion

(b) (N2) Corrosion on bolt shank cbolt

(N1) Oversized hole

NAIL

Fig. 4.3: Types of fastener corrosion: (a) nail corrosion; (b) bolt shank corrosion (courtesy of CSIRO)

typically attenuate with distance inland. Attention needs to be given in some locations to corrosion agents created locally or transported over long distances by prevailing weather patterns (e.g. industrial process emissions). Local experience is often highly indicative of such threats.

4.4

Design strategies

4.4.1

Non-structural elements

250 ) m µ 200 ( h t p e d n 150 o i s o r r o 100 c r a e y t 50 s r i F

Partially closed bay Open surf

0 0 2 4 6 Failures of non-structural elements are Distance to coast (km) by definition not immediately critical to the continued fitness of elements Fig. 4.4: First-year corrosion depth for mild in other usage and risk categories for steel fasteners used in coastal locations near their intended purposes. However, over Melbourne, Australia (courtesy of CSIRO) time deterioration of non-structural elements often exposes structural elements to deterioration risks. Hence, design of non-structural elements normally does not involve substantial engineering input, but their selection and design should balance direct initial and potential maintenance and replacement costs with their potential to become agents for longterm deterioration of structural elements. Important aspects of design decisions relate to the definition of service life criteria for non-structural elements and the methods by which to predict when those criteria will be reached. Some such criteria and models have been developed for specific locales [74,80].

52

4.4.2


Non-critical structural elements

Non-critical structural elements are efficiently designed based on the application of conventional structural engineering codes, subject to the governing premise that their dimensioning will be sufficient to ensure target stiffness and load capacity capabilities over the full extent of their design lives. More specifically, design practice should avoid increased long-term risks of component failure between the commissioning of a building and its first scheduled repair or replacement intervention. This concept is embodied in an Australian model design code [77] based on assumed relationships between load capacities of structural elements and their service lives. Empirical durability design methods for double-skinned wall, roofs, and floor subsystems are available from organizations such as Canada Housing and Mortgage Corporation [81]. Close attention should be paid when designing structural subsystems with cavities so that provisions for drainage and airflow to cavities do not conflict with or negate fire and heating, ventilation, and air conditioning (HVAC) strategies for the building. Situation-specific computational analysis is always preferable to rule of thumb approaches to check whether designed systems will remain adequately dry [82,83]. Attack analyses should be undertaken for all non-critical structural elements in contact with porous components made from materials such as masonry or concrete that may accumulate moisture from rain or soil contact. For buildings located within a few kilometres of salt water, the potential for airborne salinity should be computed to determine if the concentration poses a threat to exposed metal parts and fasteners interconnecting or embedded in non-critical timber elements [84]. If such computations imply corrosion, material specifications should require zinc coating or stainless steel.

4.4.3

Critical structural elements

Given that failure of critical structural elements leads to highly undesirable consequences, durability hazard risks undertaken should be negligible, p recluding special temporary building situations such as flooding because of failure of river protection dykes. This can only be achieved by judicious combinations of best design and construction practices. Situations exist where the prohibition of timber and wood-based composites as critical structural elements of large and tall buildings is appropriate. This occurs, for example, when the risk of fungal or insect attack cannot be acceptably mitigated via technical measures. Otherwise, risk suppression against temporary consequences of rare events (e.g. cyclonic and seismic event damage, floods) that partially destroy normal protection of timber critical structural elements can sometimes be attained by impregnating such members with low dosages of preservative treatment chemicals. Such products are always proprietary in nature and manufacturers provide data on water resistance and durability capabilities of each type and brand; in the case of some modern engineered wood-based composites, the materials have sufficient inherent capability to repel water and preserve themselves. Irrespective of other measures, when buildings are constructed in locations where risk of termite attack exists low-level preservative treatments should be applied to critical structural elements. The H2 preservative loading described in the Australian Standard AS 1604.1 [63], or similar, will typically be a suitable treatment. Use of timber classified as naturally durable is insufficient, because of the potential presence of sapwood, which is never reliably durable.

4.5 CALCULATION OF ENGINEERED SERVICE LIFE

53

Notably, low-level preservative treatments against termites are never an adequate practice on its own. Buildings must also be equipped with a whole-building protection system that Concrete Building will protect elements in all utilization and slab risk categories. One suitable whole-building protection system is illustrated in Fig. 4.5. Barrier The essential feature of that system is that the building is fully encircled by a physical or soil–chemical barrier extending 1 or 2 m into the ground and capped by a continuous conSoil crete slab. Capping concrete slabs need not be Fig. 4.5: Whole-building protection system obtrusive and may also function, for example, against termites (courtesy of CSIRO) as pathways. The intent of whole-building protection systems is to force termites to build above-ground galleries in locations that make invasion attempts easy to detect. Once any invasion is detected the remedy is to eradicate the termite nest(s). Using low dosage preservative timber treatments are simply a backup measure, in case an invasion is not detected and promptly addressed. Whole-building strategies that do not acknowledge the capabilities of termites to find ways of circumventing man-made barriers tend to be unsuccessful. Finally, where a termite threat is known to exist, a site should be cleared of all termites prior to the commencement of any construction operations. The realizable benefits from taking great care during design and construction of the structurally critical parts of tall timber structures are easily demonstrated by the longevity of monumental timber structures (Fig. 1.1) that have remained serviceable for many centuries with only repair to their non-structurally essential fabrics. A constant factor in longevity is that apart from occasional temporary wetting timbers in structures have remained dry since installation [2]. As discussed in Sections 4.3.4 and 4.4.2, embedded corrosion of steel fasteners can be rapid if timber is inadvertently wetted. Therefore, given the potential consequences it is recommended that stainless steel fasteners, and if appropriate other stainless steel parts, be used in conjunction with timber and wood-based composite elements that are critical to the structural integrity of buildings.

4.5

Calculation of engineered service life

4.5.1

Australian approach

The draft Australian code for service life design [78] provides approaches for predicting the effective residual cross section of structural timbers that remains at the end of a proposed service life. The scope of that document covers: •

materials: timber, treatments, and metal

•

environments: outdoor and indoor

•

attack agents: decay fungi, marine borer, and corrosion agents

•

element configuration

54


•

degradation attack patterns

•

degradation rates

In that document, more than 100 Australian timber species are considered based on groups classified according to their resistance to in-ground decay, above-ground decay, marine borer attack, and wood acidity. Chemical treatment may also be considered based on standard classifications given in Australian Standard 1604.1 [63]. Local climate effects are accounted for based on the hazard zones presented in Fig. 4.1 and consideration of other locality-specific parameters. For example, in assessing the degradation of metal fasteners because of atmospheric corrosion, the design hazard also takes into account the distance from the ocean coast, the coastal exposure, the type of terrain in which the building is embedded, and the proximity of any industrial plants producing potentially corrosive emissions. Not all provisions of the draft Australian service life design code will be appropriate for initial material choices and selection of maintenance and repair strategies for non-structural and noncritical structural timber and wood-based composite elements. However, those provisions are the most appropriate existing framework for bringing such decisions into the quantitative arena of engineering practice.

4.5.2

Example calculation

The following is an example to illustrate the application of the draft Australian code for service life design, for considering the bending strength at ground level of an embedded round pole [70]. Such elements are unlikely to be widely used in large and tall buildings but are convenient to demonstrate the nature of the recommended code provisions. Consider a 400 mm diameter pole of untreated hardwood from durability Class 2 (second best of four classes) from which the sapwood has been removed and which is located in a region of Hazard Class B (Fig. 4.1a). For this case, the attack is assumed to comprise a simple perimeter and core decay pattern, as shown in Fig. 4.6a. Also shown in that figure are the computed mean values of perimeter and core decay. The mean residual strength ratio of the pole after any chosen service

(a)

(b) 60

1.5

Perimeter

Core

) m m40 ( h t p e d y a 20 c e D

y t i c a p 1 a c d a o l e v 0.5 i t a l e R

Core

Perimeter

0

Mean prediction

Design value

0 0

10

20

30

Time (years)

40

50

0

10

20

30

40

50

Time (years)

Fig. 4.6: Decay and bending load capacity of 400 mm diameter hardwood pole, Class 2, located in Hazard Zone B: (a) mean decay dep ths (b) relative load capacity (courtesy of CSIRO)

55

4.5 CALCULATION OF ENGINEERED SERVICE LIFE

Add 20 mm of treated sapwoodb

Hazard zonec

Maintenanced

Design service lifee (years)

2

–

B

–

25

2

–

B

Yes

35

2

Yes

B

–

75

2

–

D

–

15

1

–

A

–

150

Durability classa

a

Four classes, Class 1 is most durable. Treated to H4 level [63]. c Four hazard zones, Zone A is least aggressive (Fig. 4.1a). d Application of external diffusing paste at 10 and 20 years. e For a choice of 40% reduction in initial design bending capacity. b

Table 4.1 Design service life for 400 mm diameter hardwood pole (courtesy of CSIRO)

life is then calculated as the ratio of the elastic section modulus of the residual cross section to the initial elastic section modulus, resulting in the relationship shown in Fig. 4.6b. Computation of the design bending strength capacity proceeds in a similar manner, except that the mean depth of decay d is replaced by an effective depth d eff that takes into account uncertainties associated with the prediction of the loss of cross section. This effective decay depth is given by Eq. 4.1: d eff = d (1 + 0.8 V d )

(4.1)

where V d is the uncertainty expressed as a coefficient of variation. Using the recommended value of V d = 1.5, leads to the design value relationship shown in Fig. 4.6b. On the basis of that relationship, the design bending capacity of a 400 mm diameter pole that has been in service for 25 years is predicted to be about 60% of the design bending capacity of a new pole, if no maintenance is undertaken after installation. Table 4.1 summarizes the impacts that selected combinations of timber durability class, service situation, initial pole sizing, and pole maintenance have on the predicted service lifespan of a 400 mm diameter pole (i.e. predicted times to 40% loss of initial capacity). As comparisons in the table illustrate, it is possible to control the initial specification of timber elements and select the maintenance strategies for a wide range of service lifetimes, with a concomitant ability to integrate that into life cycle cost optimization decisions. To be consistent with the control of structural risk strategy outlined in Section 4.4.2 for non-critical structural elements, practical application of the method just described would involve initially oversized elements to allow for the possible reduction in dimensions due to decay. In the context of the illustrative example of a pole with a circular cross section, 40% lo ss of design bending capacity corresponds to a 12% reduction in the diameter (i.e. a new pole is required to be 456 mm in diameter to guarantee a residual design capacity equal to that of a new 400 mm diameter pole). Similar example calculations exist for estimation of the service lives of metal fasteners [85].

4.6

Additional comments

This chapter is primarily to address legitimate concerns about life expectations for structural and non-structural timber elements in buildings. This chapter also highlights that if designed

56


and used properly timber elements can have design life expectancies consistent with needs of building owners and occupiers. The technologies are in many cases well known and have been applied successfully over many centuries, as testified by surviving buildings from various ages and eras. Additional contemporary measures add to the available tools but do not alter the validity underlying applicable philosophy that building elements must be shielded directly from durability risks such as termite attack, corrosion, and fungi. Emerging modern durability design practices, such as those now existing in New Zealand and Australia, for timber construction are formalizations of good practices. Although quantitative engineering calculations are at their core, it is important to be cognizant during design that desired service lives of non-critical structural and non-structural elements may not be achieved through passive resistance alone to attacks by various agents. Provision of passive resistance to some types of attacks on individual elements must be inextricably married to the selection of an appropriate whole-building design concept. Whole-building concept practices need to be ones that do not expose critical structural elements to aggressive service environments. In cases where termite attack is possible, active whole-building protection is required for all parts of the building fabric, rather than passive resistance termite attack. As readers who have consulted Chapter 3 will realize, strong similarities exist between fire and durability design in the sense that both are fundamentally about risk mitigation by protecting vulnerable elements from undesired exposure, or where that is not possible making sure that sufficient sacrificial material exists such that critical elements will retain functionality during and after any “attack”.

Acknowledgements This chapter incorporates background information, figures, and diagrams provided by Dr. Robert H. Leicester and Dr. Greg C. Foliente of the Commonwealth Scientific and Industrial Research Organization (CSIRO). This does not mean that what is discussed herein matches practices adopted in Australia or mirrors opinions of those contributors.

57 Chapter

5

Timber Frameworks with Rigid Diaphragms: Special Considerations

Summary: This chapter discusses the structural performance issues associated with design and construction of multi-storey building superstructures having frameworks constructed from relatively large linear timber elements and “rigid” floor and roof diaphragm slabs. Emphasis is on situations where superstructures have six or more storeys, with the primary function of the timber frameworks being to directly resist the effects of gravity loads. The rigid diaphragms trans fer forces associated with wind or seismic loadings to supplementary shear walls/building cores made from material such as reinforced concrete (RC). Irrespective of other factors, composite action always exists between skeletons and “other parts” of buildings. All parts must be aligned together to make the total system work structurally and to prevent non-structural parts from becoming detached under various loading scenarios. Therefore, although timber may form the bulk of framework systems, it is in fact the connections that mostly define the structural performance characteristics of what are commonly called timber-framed buildings. Much discussion herein concerns selection and design of connection methods. From the structural engineering perspective, connections are arguably more important when using timber than when using other framing materials to sympathetically meld selection of the system level concept with the capabilities of the building elements. Hence this chapter concentrates on utilizing timber elements in ways that maximize the overall structural efficiency of systems in normal and overloaded situations. This recognizes that although timber design codes and other technical support tools provide appropriate element sizing information, the skill of individual engineers is what results in elegant solutions that transcend mere functionality.

5.1

Introduction

What is presented in this chapter is consistent with the structural design principle outlined in Chapter 2. This means, for example, that it is intended that high-rise superstructure systems having timber frameworks that act compositely with rigid diaphragms (of timber or other material) should be designed and constructed in ways that ensure that primary elements within them will not exceed their elastic response ranges under design level events. Although important lessons have been learned from the behaviours of low-rise timber frameworks and high-rise frameworks constructed from other materials, it is crucial to recognize that unique circumstances apply to design of high-rise timber frameworks. In particular, special attention needs to be paid to selection of structural connection methods and marrying that to choice of structural

58

CHAPTER 5. TIMBER FRAMEWORKS WITH RIGID DIAPHRAGMS: SPECIAL CONSIDERATIONS

arrangements. Focus herein also places emphasis on the proper use of information contained within contemporary timber design codes.

5.2

Useful lessons from low-rise timber construction (circa less than 20 m tall)

While many structural forms have been adopted for construction of low-rise timber building superstructures, the types illustrated in Fig. 5.1 can be considered as common. Notably, while the good safety performance of low-rise ancient to modern timber buildings might be thought to have little direct relevance to tall timber building performance expectations when overloaded, this is not the case. The exploitation of the inherent high strength-to-mass ratio of timber and modern timber-based composites, the ability to develop alternative load paths and arrest the propagation of damage, and the capability to absorb energy associated with inertial forces when close to collapse via mechanisms other than material damping are all characteristics for which superstructures of any height should be designed. Also, need for tight control of component manufacturing and site construction practices is apparent from post-mortem field observation of collapsed low-rise buildings. This implies a preference for prefabricated systems for the construction of tall timber-framed buildings. Other lessons learned relate to the fact that solutions for providing adequate protection against disproportionate structural damage can be integrated with strategies for fire containment [20]. In addition, designing low-rise superstructure systems against infringing upon limiting states related to dynamic motions and static deflections is often the crucial factor. Furthermore, in tall buildings both vertical and horizontal motions can be problematic from a serviceability perspective [86]. In general, static deformations in well-designed and constructed timber building superstructures are modest in magnitude. Experience suggests that it is very rare for such buildings that are

Traditional buildings • Many types of occupancies • Composite walls • Walls resist all types of loads • Timber upper floors • Timber roof framing • Two or three storeys most typical • Five or six storeys examples exist • Constructed based on experience

Modern heavy-frame buildings • Mainly non-residential occupancies • Heavy timber columns and beams • Diagonal bracing added to resist lateral forces • Connections designed for axial and shear forces • Timber joisted or composite upper floors • Timber or composite roof, typically joisted • More than two storeys is unusual • Engineering design is always mandatory

Modern platform buildings • Mainly residential occupancies • All timber or composite walls • Walls resist all types of loads • Timber joisted, timber plate or composite floors • Storeys are like stacked shoeboxes tied together • More than four storeys is unusual • Six storey examples exist in several countries • Engineering design is not always mandatory

Fig. 5.1: Common structural forms for low-rise timber buildings

5.2 USEFUL LESSONS FROM LOW-RISE TIMBER CONSTRUCTION (CIRCA LESS THAN 20 M TALL) 59 Floors and roof plans

Column

Exterior walls

Girder

Wall bracing

Horizontal diaphragm

Fig. 5.2: Essential elements of low-rise multi-storey timber frameworks

serviceable with respect to vibration response to be unserviceable with respect to deformations under static loadings. Part of this assertion is the presumption that ensuring acceptable vibration serviceability is based on a realistic three-dimensional vibration analysis. Such an analysis combines the effects of the structural form and the construction detailing to avoid vibrationinduced serviceability problems and sound transmission problems. These need to be considered by structural designers, even if they fall outside the traditional scope of responsibility. The best solutions are those that combine isolation of propagation sites from receptor sites with a concentration of the mass at selected locations. In low-rise construction, this is often not done, but in taller construction it is frequently seen. For tall timber buildings, the potential for vibration and sound transmission must be designed against based on the recognition that isolation, damping provisions, and placement of relatively massive elements are key components for good solutions. Traditional timber framed buildings are relatively flexible with respect to global response. Also flexibility exists at subsystem levels. Furthermore, such systems have the ability to adjust the geometry of their substructures during seismic events [14,87]. Relative to other structural systems there can be high levels of embodied structural damping, reflecting that the building fabric is often not monolithic and can consume energy via frictional and other contacts. However, the ability to achieve this is not always realized, because it requires careful and deliberate attention to construction detailing. Lateral bracing of low-rise frameworks is normally achievable without the need to make frameworks act compositely with shear walls. This is in contrast to tall timber structures. However, various factors related to the intensities of design loads usually combine so that structural systems can normally be very simple, with well-defined flow paths for resisting gravity and lateral forces (Fig. 5.2). In modern times, glued laminated timber (glulam) elements have been the most common type of framing element, reflecting the lack of reliable supplies of new, large, sawn timbers in most parts of the world. Length restrictions for glulam members depend mostly on transportation methods and on-site lifting capabilities, which means that limits on possible dimensions match restrictions for prefabricated steel and RC elements. As illustrated in Fig. 5.3, employing continuous girders and continuous columns in various configurations is technically possible and often done. Importantly, continuity of primary framing members is employed to reduce moment peaks and deflections within the girders and column members, but for reasons explained below not to transfer moment forces between beams and columns.

60


Grider to column pin connection

Column to foundation pin connection

Wall bracing

Fig. 5.3: Illustration of the use on continuous members in low-rise frameworks (a)

(b)

Metal bearing element

Metal link element

Laminate/grain direction

Laminate/grain direction

Element internal force

Element internal force

Fig. 5.4: Logic of how timber/glulam framework elements should interface with connections: (a) bearing transfers of gravity forces; (b) axial transfers of tensile forces resulting from wind and seismic forces

In the design and construction of timber heavy-frame systems, special attention must be given to the connections between framework elements and for those between the superstructure framework and the foundations. Fundamentally, choices for interconnecting system parts are carpentry joints, connections made using metal parts and metal fasteners (with and without local reinforcement of timber members), hybrid carpentry-metal part connections, glued joints, and hybrid glue-metal part connections. Figure 5.4 shows preferred types for timber frameworks. Decisions about connection methods are always intrinsically linked to the overall structural design strategy. Three-dimensional timber frameworks are not usually intended to work in an analogous manner to most three-dimensional RC and structural steel frameworks. As shown in Figs. 5.2 and 5.3 , arrangements of timber columns and girders are not usually self-stabilizing (i.e. they need to be supported by temporary bracing during erection and then permanently braced). An arrangement of floor and roof diaphragms in conjunction with diagonal bracing in wall planes is the normal approach to stabilizing systems. Wall bracing is normally widely dispersed across bay locations in timber superstructure frameworks. Bays that are not directly stabilized must be anchored to those that are. Modern multi-storey buildings of any height nearly always have elevators and stair shaft walls and/or fire walls with significant shear wall capabilities. In contrast, timber low-rise superstructures are normally not designed to work compositely with shear walls (i.e. the two types of substructures are not structurally interconnected).

5.3 MODERN RENAISSANCE TALL TIMBER FRAME SYSTEMS (CIRCA 20–80 M TALL)

61

Low-rise timber buildings that are large in plan are often designed so t hat each fire compartment is an independent structure, with each fire compartment deriving its three-dimensional stability from geometry and an independent bracing system. Such practices yield buildings that are inherently robust and not prone to disproportionate spread of damage whatever the cause of the damage. However, when large building volumes derive from them having significant height, the same strategy cannot be followed. Present timber design codes are written predicated on the design of traditional low-rise buildings. Therefore as usage of timber transitions to also include construction of high-rise buildings, it becomes prudent to consider whether codes still adequately reflect needs of designers. When this Structural Engineering Document (SED) was written Canadian code committees, for example, had commenced the task of reviewing relevance of provisions in their national building and timber design codes.

5.3

Modern renaissance tall timber frame systems (circa 20–80 m tall)

Construction of timber frame systems up to about 80 m tall requires the discriminating application of know-how. Analysis of structural framing techniques and design concepts applicable to similar height multi-storey building frameworks constructed from other materials (i.e. steel and RC) clarifies that considerations and demands on systems alter with building dimensions. Tall timber frameworks for buildings must employ approaches specific to their circumstances and not simply be thought of as bigger versions of low-rise timber frame systems. The two primary concepts that underpin functionally efficient approaches are: 1. Timber frameworks must employ individual elements in a manner that sensibly approximates ways in which timber is designed by nature to perform efficiently. 2. Tall timber frameworks function most efficiently when working compositely with other substructures to jointly resist effects of lateral loads. The need to respect these concepts was recognized by architects and structural engineers in the Victorian and Edwardian eras, and many of their structures are still in use today [14,87,88]. What is discussed further herein is based on the concept of using timber frameworks to resist the primary effects of gravity loads, whereas systems constructed from RC (or other relatively massive material) resist the primary effects of lateral loads. Tree stems (and therefore sawn timbers and substitutes such as glulam) are inherently strong and able to resist very large compressive, tensile, and bending forces that stress the material parallel to the grain (i.e. parallel to axes of tree stems). Yet such products are quite weak in situations that stress the material perpendicular to the grain. Utilizing large timber members efficiently revolve on using them as simple beams, simple columns, simple ties, beam-columns, and beamties. Maximizing height capabilities of timber gravity load-resisting frameworks depends on maximizing capacities of vertical members in compression (because they are weaker in compression than tension). That requires loading vertical members as closely to concentrically as possible. Horizontal members in gravity load-resisting frameworks must resist bending and associated shear forces. For overall efficiency of the total system, horizontal members will also have to (at least partially) participate in the transfer of forces associated with lateral loads on

62


the buildings to the lateral load-resisting system. Normally horizontal members act as beamties or beam-columns, but not because of any intent to create moment framework action as in RC or steel frameworks. Of course, this does not mean that vertical, or in some instances also horizontal, members cannot be continuous or semi-continuous, as illustrated in Fig. 5.3. However, framework connections need not be complex, and well-established traditional and modern mechanical connection technologies can be employed. In Victorian and Edwardian eras gravity load-resisting frameworks were stabilized primarily against effects of lateral loads by massive exterior masonry walls (i.e. perimeter shear walls; Fig. 5.5). The principle remains valid, but three considerations militate against direct replication of past practices. First, elevator shafts, staircases, and service conduits are commonly bundled together in tall buildings and encased in massive walls made of RC. This creates a tower-like building core that can be designed to function Fig. 5.5: The McLennan and McFeely Buildefficiently as a very rigid shear wall ing, Vancouver constructed in 1906 with masonry system fully capable of handling latexterior and heavy timber framework (courtesy of eral loads. Second, especially when FPInnovations) building shapes are irregular, having massive perimeter walls can create undesirable modal mass distributions with respect to responses to earthquake loadings. Third, modern architectural and constructional practices do not favour the placement of large exterior structural walls (i.e. modern preference is for glazing or other relatively lightweight panels having limited structural capabilities). With respect to suitable overall structural arrangements, buildings with gravity load-resisting timber frameworks must have one or more functionally separate but physically attached and relatively rigid building cores. Also, to fulfil essential requirements for three-dimensional structural stability the completed system must have rigid floor and roof diaphragms. The diaphragms are necessary to make the building cores and gravity load systems move synchronously in horizontal directions and to resist warping of plan shapes at floor and roof levels. Suitable diaphragms must have approximately isotropic responses in their planes to perform properly [e.g. massive timber plate elements such as cross laminated timber (CLT) slabs]. CLT slabs can be rapidly installed and their lightweight is highly beneficial from a structural engineering perspective to reduce structural demand on the superstructure and foundation systems. Costs and construction times are optimized by the use of CLT slabs [89]. Figure 5.6 shows the logic of an acceptable primary structural scheme employing a rigid RC building core, a timber gravity-resisting framework, and rigid floor and roof diaphragms. The principles outlined in this chapter are extendable to most building shapes and floor plans. Anchoring of timber frameworks to building cores, and when appropriate direct anchoring of massive timber floor and roof diaphragms to RC or other structural building cores, is a critical consideration. As with all connections involving timber, careful attention should be paid to

5.3 MODERN RENAISSANCE RENAISSANCE TALL TALL TIMBER FRAME SYSTEMS (CIRCA (CIRCA 20–80 M TALL)

(a)

(b)

(c)

(d)

63

Fig. 5.6: Illustration of a suitable overall structural scheme for a modern renaissance tall timber frame system: (a) foundation and RC building core; (b) timber framework; (c) horizontal diaphragms; (d) assembled primary structural system

avoid connections that transfer loads other than thrusts and shear forces. Depending on construction practices and sequencing, attention may also need to be directed towards accommodation of differential movements between timber and other materials. Shrinkage related to moisture movements and drying in timber and curing of concrete can be significant in some instances, as can temperature-related movements. Thus, these will need to be accommodated. As in other structural engineering situations the “golden rules” are to design connections so that they accommodate, rather than resist differential movements related to contraction or dilation of components and to avoid solutions that grip members too close to their ends or edges. This should, however, not be taken to imply that the difficulties associated with effectively tying parts together are any greater with timber than any other materials. Consistent with general principles of structural engineering, large and tall buildings employing timber as a primary structural material should be designed to handle effects of even the most extreme loads in the elastic range ( Recommendation Recommendati on 3 ). Doing that is the first line of defence in achieving solutions that are robust and reliable. This need not entail loss of material or economic efficiency. The consequence is in fact mostly that architects and engineers will have to apply their skills in new ways (i.e. differently from how they conceptualize the design of low-rise timber superstructures). Designing systems to respond elastically should not be the only defence against damaging events. Avoiding the possibility of disproportionate damage or catastrophic collapse of systems such as those illustrated in Fig. 5.6 requires requires taking deliberate measures. Column failure presents the greatest danger of widespread damage. There exist a number of viable ways of ensuring systems have the robustness to contain

64


damage, if for instance a column was removed because of a vehicular impact or because of an explosion. Use of continuous horizontal beam/girder elements in the gravity load system and continuity in floor and rood diaphragms can provide the possibility of alternative load paths, if such damage were to occur occur.. Practical implementation of this approach can be based on ensuring that potentially critical elements would not exceed their ultimate capacities were an adjacent parallel component(s) removed. Such calculations can be based on non-partial, factored design loads for structural systems to avoid design solutions being unreasonably conservative. servati ve. The determination of how many columns could be b e removed and survival be maintained should depend on considerations such as the class of building occupancy and physical proximity to hazard sources. Another illustrative robustness and damage containment strategy is to provide supplementary lateral bracing systems (located within timber frameworks) that are located in regions of the building judged most potentially critical under disaster scenarios. Secondary triangulated bracing that is permanent or would be activated when required could also be installed (e.g. at at lower storey corners). corners). The approaches suggested in this chapter essentially replicate what has been followed for many centuries [14,87]. However, in situations where timber skeletons braced in the vertical plane act compositely with other substructures to resist lateral wind or seismic loads, full dynamic analyses of the entire hybrid system is the preferred approach.

5.4

Effective connection methods

As summarized by Madsen [90,91], certain general considerations pervade design and construction decisions related to structural timber connections: •

Strength efficiency: Connections must have the ability to resist potentially damaging effects of individual or simultaneous flows of shear, axial, and moment forces from ends or surfaces of elements (e.g. (e.g. framework framework elements and diaphragms) to other parts of the system. system.

• Stiffness/flexibility: Connections should be designed and constructed to have the ability to enable or limit deformations in ways that are consistent with the intended structural concepts for the behaviour of the completed building. •

Failure mechanisms: It is always essential to be able to identify how connections might fail, if structural systems are overloaded. Some types of timber connections are prone to exhibiting brittle failure mechanisms and are, thus, inappropriate for construction of frameworks of large and tall buildings. Therefore, it is essential that engineers be able to reliably identify how connections would fail under various scenarios, what residual capabilities connections would have after reaching their ultimate capacities, and what system level design loads would be associated with connection failure. Modern Load and Resistance Factor Design/Partial Design/Partial Coefficients Design (LRFD/PCD) (LRFD/PCD) timber design codes and proprietary information belonging to connection hardware manufacturers support these needs.

•

Force reversa reversals: ls: Reversals of force flows through connections are common in practice under the effects of various normal and abnormal loads. Heights of multi-storey superstructures and their slenderness often mean that their total gravitational masses are insufficient to prevent lateral forces associated with wind and seismic loads from creating tension situations in vertical members (i.e. superstructures respond like giant cantilevers). Thus, even connections that rely on simple bearing action should have some capability to resist the effects of force reversals.

65

5.4 EFFECTIVE CONNECTION METHODS (a)

(b)

Fig. 5.7: Installation of glulam superstructure framework segments Fig. segments (further details in Chapter 8): (a) two-store two-storeyy segment; (b) interconnection in terconnection of segments •

Simplicity and transparency of behaviour: It is essential that the functions of individual structural connections be transparent, not only to the original designers but also to others who might have to perform structural assessments (possibly decades after buildings were constructed). Load paths and intended behaviour b ehaviour of each connection within any superstructure must be fully consistent with the intent of the design strategy. strategy. If, for example, connections are assumed to be pinned, then that is what should be implemented. It is vital that the addition of secondary structural and non-structural elements during initial construction or renovation does not transform the way in which the primary structural system works under various design load scenarios.

Traditional connection methods that have proven successful all adhere to these simple principles (Fig. 5.4). To be effective, modern connection methods also need to do so. In large superstructure systems one must consider how elements will be manufactured and erected. As illustrated by example projects in Chapters 8 and 10, timber/glulam framework elements of large multi-storey superstructures are nearly always manufactured off-site in factory conditions. This includes the exact, final element shaping and the cutting and drilling associated with connections. Similarly Similarly,, metal connections are nearly always manufactured off-site. Therefore, for building superstructures of types discussed in this chapter, the elements are received as parts intended to fit together exactly. The situation is in no way akin to traditional timber construction practices, where elements were typically manufactured from raw materials such as pieces of lumber. In modern practice, large ti mber frameworks frameworks are often partially preassembled before being lifted into place, as illustrated in Fig. 5.7 . In principle, on-site assembly practices are similar to those that apply to structural steel and precast RC element frameworks. Joints made using generic mechanical fasteners are pervasive in modern timber construction. The term mechanical fastener encompasses all devices that are inserted into two abutting members in ways that result in transferred forces flowing across joint interfaces via those devices. Most generic mechanical fasteners have been mass-produced since the 19th or early 2 0th centuries, with wire nails, steel bolts, and plain steel dowels being the most common. The popularity of joints using generic g eneric fasteners reflects several several factors including that the fasteners are typically inexpensive and easy to install, and that joint strength can be reliably predicted using easily accessed information (i.e. in codes and handbooks). In the context of large frameworks, joints

66


(a)

Wood

Forces on gripping element

Resistance from wood (bearing + skin friction) Gripping element

(b)

(i)

(ii) Wood Steel dowel (embedded)

Steel bolt

Steel plates

Steel link element (embedded)

Connection interface

Sacrificial wood plug Wood Wood

Fig. 5.8: Joints with laterally loaded fasteners: (a) principle of gripping action; (b) classification of mechanical joints with metal parts: (i) exposed metal, (ii) concealed metal

in connections are typically made using laterally loaded steel dowels and bolts. As indicated in Fig. 5.8, joints made with generic dowel fasteners often interconnect timber elements with metal link elements that bridge connections, with the metal parts being at the hub. The slenderness of metal dowel fasteners normally determines the flexibility and degree of ductility of joints and connections employing them. Reliance on skin friction around dowel fasteners for primary force resistance is something structural engineers usually try to avoid because of its relative lack of reliability after timber members have dried or dried and then been rewetted. Therefore, with the exception of bolts, plain dowels, and some types of screws, emphasis tends to be on using dowel fasteners in situations where force transfer does not involve the possibility of pulling them out of or through joined materials. In the context of structural frameworks formed from large timbers, all mentioned types of generic fasteners have been employed. However, experience suggests that using plain metal dowels and metal bolts are most appropriate and economical when large forces need to be transferred. There exists extensive literature on design and construction of joints employing generic dowel fasteners. Nearly all general textbooks and design aids on timber engineering devote substantial attention to the appropriate use of various fasteners and provide sample design solutions [92,93]. This stated, designers who intend system responses to remain in the elastic range should verify the response state associated with design capacities for recommended dowel fastener connections [93].

67

5.4 EFFECTIVE CONNECTION METHODS

Glued-in rods Steel connecting assemblies Glulam members

Fig. 5.9: Illustration of glued-in rod connections suitable for multi-storey framework applications

Many types of proprietary joint/connection hardware are highly suitable for making connections in timber framework elements, for attaching diaphragms to frameworks, and for attaching frameworks to shear walls and foundations. In all cases, necessary design information about joint/connection stiffness, strength, and failure mechanisms is outside the scope of design codes and must be obtained from the specific manufacturer. Glued-in, rod joint systems involve the embedding of metal or other types of reinforcing bars into structural elements. This creates local continuity akin to connections in RC frameworks, or diffusely transfers forces into or from ends of members to other parts of connections (e.g. rollers, pins, and swivels). Reinforcing rods are inserted into predrilled oversized holes in glulam or other large timber members with gap-filling adhesives or grouts employed to bond the rods. For economy, rods are normally of the types employed in RC construction processes. Considerable research has been devoted to developing methods in Scandinavia, Russia, New Zealand, and Canada. Most of the research has focused on the behaviour of joints/connections under static and seismic loads for situations where glulam columns and girders intersect or where columns connect to foundations. As expected, certain connection configurations, details, and fabrication methods are superior to others. Generally, it is preferable that reinforcing bars be installed into timber members off-site under controlled factory conditions and then fastened to hub-like connection devices by on-site bolting to create arrangements such as that illustrated in Fig. 5.9. The illustrated arrangement would achieve a self-bracing frame action and steel connecting assemblies could be designed to yield in desired ways, if the system was overloaded. Using glued-in rod approaches can create force flows consistent with almost any desired structural system behaviour [91]. A discussion of standard timber jointing methods would not be complete without mentioning glued joints. Rigid adhesives are employed under factory conditions to create intra-component joints in glulam members, multilayered timber-based composites such as laminated veneer lumber and CLT, box-girders, and stressed-skin panels. On-site gluing of large timbers has been practiced on a limited basis to create long members for low-rise construction and conceptually such technology could be applied to construction of relatively tall multi-storey framed superstructures. However, the technology has not been proven suitable for construction of superstructures of types that are the focus of this chapter, and hence, is not recommended at this point of time.

68


(a)

(b) Surface or embedded energy dissipation elements (absorb energy via plastic work during load cycling)

Glulam beam

Glulam column

Elastic response

Unbonded post-tensioned tendon (compressing connection to enforce selfcentering on release of external loading)

Plastic response

(c) (i)

(ii)

M

M

(iii)

Elastic M

Plastic θ

θ

θ

Elastic Plastic

Fig. 5.10: Logic of self-centring frame connections, based on Buchanan et al [16]: (a) components of a frame connection; (b) deformed connection during seismic action; (c) logic of flagshaped hysteresis loops (moment M versus rotation q : (i) effect of post-tensioning, (ii) effect of energy dissipation elements, (iii) combined response: flag-shaped hysteresis loop

On a final note, researchers from New Zealand and Italy have been performing experiments and undertaking numerical modelling to define self-centring connection technologies suitable for timber heavy-framed building superstructures. Focus is on creating frameworks that efficiently resist the effects of seismic actions [16], based on adaptation of what have become fairly common practices for post-tensioned prefabricated RC frameworks [94]. The principle is that by suitably arranging post-tensioning tendons and sacrificial metal energy absorbing elements (that deform plastically, if overloaded) it is possible to create a self position-restoring system response without sustaining more than minimal damage (Fig. 5.10). Essentially the post-tensioning forces that do not exceed the elastic range of the tendons pull the system back into its original position when external forces are removed. The energy dissipation elements can undergo plastic deformation while absorbing considerable amounts of system-level inertial work. If properly designed, the dissipation elements can be easily replaced during post-disaster repair work. Buchanan et al. [16] have also developed post-tensioning connection systems suitable for minimizing damage in timber buildings containing massive wall panels. They claim that their connection methods are suitable for constructing building superstructures with 10 or more storeys in high seismic risk locations. No explicit proof exists yet of satisfactory field performance of post-tensioning

5.5 ADDITIONAL COMMENTS

69

connection systems, but it can reasonably be presumed that they will work well, because what is done replicates in principle the approaches to seismic design employed in traditional bearing wall and timber infill-frame buildings (Figs. 2.3a and 2.3b). The modern approach is to achieve via post-tensioning and energy absorbing elements what was formerly achieved by overburden pressures, frictional damping, and carpentry details.

5.5

Additional comments

This chapter is not intended to be read alone. Readers are strongly encouraged to pay close attention to Chapters 3 and 4 that respectively address fire and durability performance and design of large and tall building superstructures employing timber as a primary construction material. It is also advisable to pay close attention to the discussion of the example project of a six-storey hybrid building in Quebec City in Chapter 8, the case stud ies in Chapter 10, and future possibilities discussed in Chapter 12.

Acknowledgements This chapter draws on input from Drs. Mohammad A.H. Mohammad, Marjan Popovski, and Chun Ni and Mr. Sylvain Gagnon of FPInnovations, Canada.

71 Chapter

6

Steel or Reinforced Concrete Frameworks with Timber Diaphragms: Special Considerations

Summary: This chapter discusses the structural performance issues associated with employing relatively lightweight timber floors and roof slabs in high-rise buildings, as substitutes for reinforced concrete (RC) slabs. The focus is specifically on hybrid construction systems where structural steel or RC framework substructures are the primary means of resisting effects of gravity forces associated with the occupied building space. RC building core substructures are the primary means of resisting effects of lateral forces associated with wind or seismic loading. Use of massive engineered timber composites known generically as cross laminated timber (CLT) is discussed as a substitution for RC slabs. CLT is approximately mechanically equivalent to RC slabs of equal thickness, but with the advantage of being only around one third of the weight. The lightness of CLT reduces demands on the structural framework and foundation substructures with respect to both gravitational and lateral loads. These advantages are the greatest when CLT is used in combination with structural steel frameworks, because the proportional reduction in total mass is much greater than with RC frameworks. In practical terms, the advantage of using CLT instead of RC slabs is that either the structural framework and foundation materials can be significantly reduced or the structural performance of the entire superstructure system can be improved without extra cost. The design per formance of a 24-storey building with alternative steel and RC framework substructures is used to illustrate the typical magnitude of structural gains associated with substituting CLT for RC slabs.

6.1

Introduction

The discussion below addresses the design of superstructure systems having massive timber floor and roof slabs, supported by steel or RC frameworks that act compositely with RC building cores. Specifically, such systems employ massive timber slabs as horizontal diaphragms at all floor levels and at the roof level or combine them with some RC fire floors. By employing a 24-storey (96 m) building, a comparison is made between using massive timber floor slabs and using normal weight RC slabs, so that readers can obtain a sense of the relative structural performance of alternative systems under various design load scenarios.

72

CHAPTER 6. STEEL OR REINFORCED CONCRETE FRAMEWORKS WITH TIMBER DIAPHRAGMS

(a)

(b)

Fig. 6.1: Massive timber plate elements: (a) PLT; (b) CLT

6.2

Massive timber diaphragms for composite hybrid systems

There are several methods for constructing massive timber floor and roof slabs, with the two primary methods being a parallel laminated timber (PLT) or a CLT plate (Fig. 6.1). The timber laminates can be either mechanically laminated using semi-rigid fasteners such as metal nails, or rigidly laminated using structural adhesives. Both PLT and CLT products are prefabricated under factory conditions and shipped to a site as large plates in thicknesses up to about 600 mm. Use of diaphragms made from CLT plate slabs is discussed in this chapter, because they have highly efficient two-way slab and diaphragm capabilities similar to RC slabs, and are simple to attach to structural steel or RC frameworks or RC building cores. Traditional timber floor diaphragm construction methods do not possess sufficient in-plane rigidity to maintain floor shapes in tall, framed buildings in a manner mechanically equivalent to using RC slabs and are, therefore, not discussed in this chapter.

6.3

Twenty-four-storey case studies

6.3.1

Scope and methods

Figure 6.2 shows the building superstructure configuration adopted for the case studies underpinning this chapter. The structural frameworks in Fig. 6.2a have either steel I-sections or RC rectangular-sections. Connections in either type of framework are rigid. Therefore, the framework acts compositely with the RC tower (which constitutes the building core) and the horizontal diaphragms to resist the effects of various loads. Floor and roof diaphragms of the hybrid buildings in this chapter derive their in-plane and out-of-plane rigidities and strengths from the composite action between the horizontal framework members and the slabs attached to their top surfaces, as illustrated in Figs. 6.3 and 6.4 for steel frameworks. These figures show the construction arrangements for alternatives of semi-rigidly attached CLT plate slabs and monolithic RC slabs attached to steel members using rigid studs; with the slabs in each case having approximately equal mechanical capabilities. Slab thicknesses in the case studies presented in this chapter were as follows: roofs 112/110 mm and floors 190/150 mm (CLT/RC), Fig. 6.3. Although CLT slab thicknesses are greater than those of the RC slabs, these CLT were selected from currently available products, rather than ones

73

6.3 TWENTY-FOUR-STOREY CASE STUDIES

(a)

Sub beams

Columns

Main beams

Shear walls m 2 . 3 × 8

Y X

6 × 6.4 m

(b) Framework sections Steel framework*

Columns

1st-8th: W14 × 500 9th-16th: W14 × 342 17th-24th: W14 × 159

m 6 9 = m 4 × 4 2

RC framework**

X m 6 . 5 2

W24 × 94

Secondary beams

W16 × 49

Columns

1st-12th: 800 mm × 1200 mm 13th-24th: 600 mm × 800 mm

Z

3 8 . 4 m

Main beams

Main beams

350 mm × 650 mm

Secondary beams

200 mm × 400 mm

* AISC sections [95] ** Dimensioned based on ACI [96]

Y

Fig. 6.2: Twent y-four storey building: (a) typical floor layout; (b) three-dimensional system model

optimized for the application. Here the notations CLTf and RCf are used to distinguish systems embodying CLT and RC floor/roof slabs, respectively. Firef is used to denote mixed CLT–RC slab situations where most floors have CLT slabs but every fourth floor has a RC slab. This arrangement achieves a total floor area between adjacent fire floors consistent with accepted sizes of fire compartments within low-rise timber buildings (Chapter 3). As discussed by Smith and Frangi [20], this approach applies when regulatory authorities will not permit fully CLTf systems.

74


(a)

(b) Cast in-situ concrete

t =150 mm (floor),

CLT plate

t =190 mm (floor),

110 mm (roof)

112 mm (roof)

Shear studs Screw fasteners Corrugated steel permanent formwork

Steel joist

Steel joist

Fig. 6.3: Floor systems used with steel frameworks: (a) RCf (RC floor) system; (b) CLTf (CLT floor) system

(a)

(b) Steel column

Steel girders Steel floor joists

m 8 . 2 1

1.6 × 6.4 m slab elements with screwed lap joints

6.4 m

Strong axis

Fig. 6.4: Typical CLT slab arrangement and connection methods: (a) CLT panel layout over supporting steel floor framing; (b) schematic CLT connections

The load combinations considered included the effects of gravity (self-weight and imposed floor and roof) loads, seismic excitation (peak ground acceleration 0.5 g), and wind forces (basic wind speed 240 km/hour). Details of the load data and factor combination are drawn from contemporary US Load and Resistance Factor Design (LRFD) practices, with details given in Ref. [89]. The load magnitudes and combinations used are not for a particular site. Instead, they approximate a worst-case wind and seismic design scenario for cont inental North America. This approach gives an indication of how buildings of the type studied herein are expected to perform in general, rather than at a specific construction site, for the alternative CLTf , RCf, and Firef systems. Members in steel and RC system frameworks were the same irrespective of floor slab

75


Steel

Concrete

CLT

Isotropic

Isotropic

Orthotropic

Density (kg/m3)

7200

2400

400

Elastic modulus (GPa)

200

25

E1 = 9 E2 = 4.5 G12 = 0.5

Poisson’s ratio

0.30

0.25

ν12 = 0.3

Strength (MPa)

250

27.5

ft–1 = 20, f t–2 = 15 f c–1 = 30, f c-2 = 25 f shear = 5

Property

E = modulus of elasticity; G = modulus of rigidity; 1 = CLT major direction; 2 = CLT minor direction; t = tension; c = compression. Table 6.1: Material properties used in case studies

type and corresponded to the necessary minimum dimensions for systems with only RC slabs ( RCf systems, Fig. 6.2). Thus, the results show only the effects of altering the slab type, and in the case of CLTf systems introducing flexibility to the slab connections. Fully three-dimensional finite element (FE) models using SAP2000 [97] predicted structural response. Modelling approaches were directly equivalent, except with respect to the physical and mechanical properties for CLTf and RCf systems. Floor and roof slab components were modelled using four-node shell elements. Framework members were modelled with twonode frame elements having six degrees of freedom per node (i.e. three translations and three rotations). Table 6.1 summarizes the material properties used in the analyses. Link elements represented flexibilities of inter-element connections between consecutive CLT slab elements and between them and other parts. A linear dynamic analysis based on the response spectrum method was used to apply seismic displacements at the base of the buildings ( Recommendations 2–4 and 6 from Chapter 2 ). RC building cores were modelled as four-node shell elements having thickness from 250 to 300 mm. In the analysis, it was assumed that CLT slabs were connected to frames with translational link elements spaced at 800 mm and having stiffness of 8 kN/mm for forces parallel to the slab planes and 10 kN/mm for forces normal to the slab planes. The stiffness values were estimated based on connection tests done in the work of Ref. [98]. Abutting edges of the CLT panels interlock via lap joints fastened using long wood screws to ensure slab continuity as shown in Fig. 6.4b. Both for the RCf and CLTf systems, the connections between the steel skeleton and shear wall cores were modelled as rigid [95]. Pinned joints were used to simulate the connections between the RC or CLT slabs and the RC cores. In the case of the CLT slabs, this corresponded to fastening them to the RC using light steel angles, whereas for RC slabs it corresponded to continuous steel bar links that run from the slabs to the cores (no heavy reinforcing anchorages). The column frames were modelled as rigidly fixed at the foundation level. The foundations were excluded from the analyses. All parts of superstructures were assumed to respond as crack-free bodies.

76


The most deformed frames under critical seismic disturbance

Fig. 6.5: The most deformed frame in steel or RC systems during seismic event

6.3.2

Case study results

6.3.2.1 Structural steel framework systems With the steel frameworks, the total weight of the RCf system was 115.9 MN, and the total weight of the all CLTf system was about 50% of that (57.2 MN). The maximum gravity load floor diaphragm deflections (i.e. effects of dead, superimposed plus live loads) were 23.1 mm and 19.0 mm, respectively for the RCf and CLTf systems (i.e. less than span/300). Peak stress levels in bending were far below tolerable levels, thereby indicating acceptability of floor diaphragms as distribution elements for floor loads. The superior performance of the CLTf system was a direct consequence of the comparative lightweight of the CLT slab relative to that of the RC slab (Table 6.1). Significant benefits also accrued with respect to reduced impacts of the gravity force demands on frameworks, cores, and foundations for the CLTf systems. However, the lightness of the CLT slabs yielded the largest benefits when the effects of the lateral loads were analysed. From that analysis, the critical lateral load for the RCf and CLTf systems was as a result of the combination of gravity and earthquake loads. Figure 6.5 shows the typical deformed shape of the frame under the critical earthquake load (30%-X + 100%-Y), and Fig. 6.6 shows the predicted peak lateral displacements (drifts) at each storey, along with the associated inter-storey drifts [30%-X indicates 30% of the peak ground acceleration applied at the building base in the X-direction (parallel to the long plan axis), and 100%-Y indicates 100% of peak ground acceleration applied in the Y-direction (normal to the long plan axis)]. At the roof level, the predicted drift of the RCf system was 240 mm, and 149 mm (about 40% less) for the CLTf

77


(a)

(b) 24

24

18 r e b m u n 12 y e r o t S

CLTf

18 r e b m u n 12 y e r o t S

Firef

6

RCf

6

0

RCf CLTf Firef

0 0

50

100

150

Drift (mm)

200

250

300

0

3

6

9

12

15

Interstorey drift (mm)

Fig. 6.6: Peak drift and inter-storey drift in steel frameworks during seismic event: (a) lateral displacement; (b) inter-storey drift

system. The maximum inter-storey drift was 12.0 mm for the RCf system, and 7.9 mm for the CLTf system, both occurring in the 17th storey. In both cases the inter-storey drift was less than 0.5% of the storey height, thereby achieving a performance level typically required of a tall building intended to be fully functional after a major earthquake [17,99]. Using CLT as floor slabs components in the tall building in lieu of RC slabs was advantageous in terms of total system response and mitigation of the potential for damage. As would be expected, the peak lateral and inter-storey drift of the Firef system was intermediate to the matching responses of the CLTf and RCf systems (Fig. 6.6 ). However, it should be noted that the vertical mass irregularity between floors in the Firef system causes irregularities in mode shapes. Given that the analysed case represents a building with simple overall geometry and construction, the results shown here exemplify why simplified assumptions about design loadings and structural responses (e.g. as embedded in equivalent static load analyses of superstructures) are not accurate or reliable in general, and can be the source of unexpected damage or other manifestations of poor building performance. Peak horizontal shearing forces carried by fasteners around the perimeters of CLT slab panels were about 15 kN/m. This equates to using screws of around 10 to 12 mm diameter at a spacing of 200 mm. The largest forces on the fasteners occurred under load combinations involving seismic excitation of the fifth floor slab, which is where the predicted torsional accelerations were the greatest. Analyses compared the response of a fully CLTf building having realistic, semiflexible connections with the response of an otherwise similar building having slabs rigidly connected together and rigidly connected to the framework. The maximum difference in shear flows in the slab to the framework connections was less than 4%, indicating that neglecting connection flexibilities does not necessarily result in serious errors. Reductions in computing effort can be significant by assuming rigid connections (as is often done when designing steel framed buildings), particularly when the buildings are large and h ave complex shapes and framing layouts. Notably, the slab connection force demands need to be calculated by accounting for forces developed at nodal points linking specific CLT shell and steel frame elements together.

78


Structure

First mode (seconds)

Second mode (seconds)

Third mode (seconds)

CLTf

1.96

1.69

1.20

Firef

2.29

2.03

1.42

RCf

2.90

2.41

1.74

Fig. 6.7: Mode shapes and periods for steel frameworks (illustrated mode shapes are for CLTf case)

In the dynamic analysis underpinning this discussion of buildings with steel frameworks, 24 mode shapes were extracted to ensure that modal participation factors reached 90% (i.e. to ensure they were valid for modal combination analysis) [20]. Figure 6.7 illustrates the first three mode shapes and associated structural periods for RCf , CLTf, and Firef systems. As the modal frequencies in the table embedded in the figure show, the lower order frequencies were not widely separated, thereby implying the strong possibility of motion amplification during normal service events such as low-intensity winds and abnormal service events such as hurricanes. No significant differences in mode shapes were predicted for the three systems. However, this should not be construed as implying that the time history responses will match in shape or that the peak drift profiles will be similar in “roughness”. As expected, given that modal stiffness values were derived essentially from the RC building core and steel framework, the lateral predicted structural periods were distinctly higher for the heavier systems, ( t 1 = 1.96, 2.29, and 2.90 seconds for CLTf , Firef , and RCf systems, respectively). The first mode shape was similar to a simple cantilever deflection, whereas the second and third modes had strong torsion components. The mode shapes of the quite simple case study buildings and the possibility of modal clustering are further indications that even relatively simple, tall, hybrid structures can exhibit complexities in responses and need to be analysed with care. Further analysis not reported in this chapter shows that structural demands on the steel skeleton are always much less in CLTf systems than in comparable RCf systems, irrespective of the design load combinations. On the basis of simple maximum sums of stress ratio criteria (i.e. Σ design stress to design strength ratios), “sum values” are 20% to 30% higher for framework members in RCf systems than in CLTf systems [89]. Although also not presented in this chapter, steel reinforcement demands in the RC building cores and foundations will tend to be minimized by CLTf systems. 6.3.2.2 RC framework systems In a broad sense, much of the discussion of RC framework systems parallels the discussion related to steel framework systems. The discussion in this chapter is, therefore, focussed on the consequences of different construction details and the heavier weight of RC frameworks.

79


(a)

(b) t

Cast in-situ concrete

= 150 mm (floor), 110 mm (roof)

t

CLT plate

Steel angle and fasteners

= 190 mm (floor), 112 mm (roof)

RC beam

RC beam

Fig. 6.8: Floor system used with RC framework: (a) RCf; (b) CLTf

(a)

(b) 24

24

20

20

r 16 e b m u n 12 y e r o t S 8

r 16 e b m u n 12 y e r o t S 8

RCf CLTf

RCf CLTf Firef

Firef

4

4

0

0 0

50

100

150

Drift (mm)

200

250

300

0

3

6

9

12

15

Interstorey drift (mm)

Fig. 6.9: Peak drift and inter-storey drift in RC frameworks during seismic event: (a) lateral displacement; (b) inter-storey drift Figure 6.8 gives detailed schematics of the construction for RCf and CLTf systems involving RC frameworks and RC cores. Dimensions of framework members are given within the information box in Fig. 6.2. RCf floors slabs are cast monolithically with horizontal framework members ( Fig. 6.8a), whereas CLT slab panels are mechanically attached using flexible fasteners in ways closely paralleling practices for steel frameworks (Fig. 6.8b). In practice, it is impossible to make completely rigid connections/to rigidly bond timber members to other materials using mechanical devices, even with adhesives because of the shear lag through the member depths. Also, long-standing experience teaches that structural timber substructures do not function well if the local ability of parts to “adjust themselves” during service is constrained excessively. The FE modelling details for the CLT slabs and associated connections were the same as in the steel framework system analyses.

The total weight of the RCf system was 166.1 MN, whereas the total weight of the CLTf system was about 35% less (108.4 MN). Maximum gravity load floor diaphragm deflections

80


System

First mode (seconds)

Second mode (seconds)

Third mode (seconds)

CLTf

2.67

2.47

1.74

Firef

2.87

2.59

1.86

RCf

3.25

2.87

2.10

Table 6.2: First three periods for RC framed systems

(i.e. effects of dead, superimposed plus live loads) were 20.1 mm and 18.0 mm, respectively for RCf and CLTf systems. Notably, those values are slightly lower than for the steel framed systems reflecting mostly the greater torsional rigidities of the RC framework members. Conclusions match those from prior discussion regarding acceptability of CLT slabs for tall hybrid construction. Figure 6.9 shows predicted peak lateral displacements and inter-storey drifts due to the most critical load combination involving seismic excitation. Critical locations are once again along the outer plan edges (as shown in Fig. 6.5). Predicted drift at the roof level for the RCf system is 274 mm, and 228 mm (i.e. 17% less) for the CLTf system. The maximum inter-storey drift was 13.5 mm for the RCf system occurring in the 13th storey, and 11.5 mm occurring in the 16th storey for the CLTf system. In both instances, the levels of inter-storey drift were less than that typically permitted (around 0.5% of the height). As in the case of steel framework systems, using CLT slabs as floor components in tall building in lieu of RC slabs was advantageous in terms of the total system response, as a direct reflection of lower modal masses. Magnitudes of predicted movements were not problematic in this particular instance, irrespective of slab type. However, that the storey level that will govern any aspect of a system’s design cannot be known in advance is another indicator of the pitfalls that could beset those who attempt to oversimplify calculations that support design. As with steel framework systems, demands on the RC frameworks are always lower under other loading scenarios too, for the same reason. The difference between the situations is, however, the gains are proportionally smaller when frameworks themselves are relatively heavy. Arguably, design choices of framing and other construction materials should be market driven. However, the greater the amount of vibrating mass involved, the harder it is to achieve economy without risking the possibility of damage during an extreme event. Approximately 10% more reinforcing steel is estimated to be required in the superstructure employing the only RCf solution. The Firef system, as expected, had a seismic response lying between the other systems with matching practical design implications. Delving more deeply into the dynamic responses of RC framed systems’ is informative. The first three mode shapes are similar to those illustrated in Fig. 6.7, with the associated natural periods given in Table 6.2. Cantilever action dominates the fundamental mode shape for all of the RCf , CLTf, and Firef systems, but their responses can be classified differently. According to practices in the Canadian National Building Code [17] only the RCf system can be considered to be responding like a very slender structure. The predicted fundamental period of 3.25 seconds nearly matches the upper limit of 3.45 seconds in that code, for cases that can be considered to be RC moment frames in simplified analysis. This indicates that it is highly feasible that hybrid design situations (involving the use of timber as a primary structural material) will lie on the margins of where simplified design practices are allowed by codes.

81

6.4 GENERAL IMPLICATIONS OF USING CLT SLABS

Quite high stresses in CLT components are predicted at the 23rd floor level for the deadweight, live load plus earthquake load combination. Even then, the stresses generated were much less than typical CLT strength properties (Table 6.1). Peak forces in the CLT slab to framework connections occurred in the shear around the slab perimeter of the fifth floor. However, simple construction details like the metal angle connection in Fig. 6.8b were found adequate based on data in the literature [100,101]. As when modelling steel framework systems, modelling CLTf system connections as rigid did not significantly alter the total structural response (i.e. drift, mode shapes, and modal periods). However, care should once again be taken regarding extrapolation of this observation.

6.4

General implications of using CLT slabs

As the systems analysed are not symmetric about their x-axes, a combined earthquake motion in two orthogonal directions (i.e. 100%-X + 30%-Y and 30%-X + 100%-Y were analysed) produces torsional responses causing horizontal diaphragms to deform as deep beams ( Fig. 6.10). The systems would also twist about their vertical geometrical axes under wind loading, except when the wind direction is normal to the wide faces of a building. As would be expected, the critical in-plane deformation occurred near the upper surface of the buildings at the 23rd storey level. On the basis of the steel framework systems, the predicted deformed shapes of the 23rd storey floor diaphragms indicates that the CLTf system has a flexibility of 2.5, whereas the RCf system has a slightly lower flexibility of 2.3 for the “long span” direction. (When envisioning a diaphragm as a beam, the flexibility is the ratio of the lateral drift at mid-span of the diaphragm to its average lateral drift at the ends [102].) Using the same location and definition, flexibility values for the RC frameworks are predicted to be 2.7 for the CLTf system and 2.4 for the RCf system. For both types of framework the flexibility is around 10% higher for the selected CLT slab construction than the selected RC slab construction. Clearly, these are degrees of difference that are easily adjustable via detailed design decisions and not indicative of any systemic relationship. The flexibility of CLT slab floors could, for example, be manipulated, by altering the thickness or specification of the slab material or adjusting the fastening schedule. Global control

d 1

d mid

d 2

100% EQ- x

y x 30% EQ- y

Fig. 6.10: Deformed shape of a horizontal diaphragm: Note: d avg is the average storey drift or (d 1 + d 2)/2, and d m is the mid-span deflection

82


of the general level of flexibility of diaphragms of the types discussed is handled typically as part of the overall structural concept. For example, flexibility can be controlled through provision of adequate frames/chords at the building perimeter to resist the flexural action and provision of collector frames capable of transmitting lateral forces to vertical lateral force resisting systems. Mention of diaphragm flexibility here is simply to indicate that it need not be radically different from familiar RC slab approaches. In the “short span” direction, diaphragm flexibilities were in the order of 1.1 for both CLTf and RCf systems (i.e. quite rigid). Analyses of case study systems were performed to determine the impact of simply assuming diaphragms are rigid in-plane, instead of accounting for their actual flexibility. In such simplified analysis, there was no need to account for details of either the diaphragms or framing components to which they are directly attached. All that was required was to consider two translational and one rotational degrees of freedom of a master node at each storey. Weight (mass) of a particular storey was lumped at its master node. Other elements at each level were modelled to deform horizontally with the same magnitude and direction as the master degrees of freedom. When this approach was employed, results indicated significant differences in predictions of drift and periods (some larger than 100%) between flexible (actual properties) and rigid models. This was especially pronounced for RC framework systems. The CLTf system with a RC framework exhibited largest differences, relative to the corresponding RCf system. Notably, this was in part because of the inability of the simplified analysis to incorporate the flexibility of connections between CLT plates and the RC building core. As a broad finding of the particular comparative analyses, rigid diaphragm response assumptions yield results similar to more detailed modelling in terms of the overall structural response when the structural frameworks act alone to provide the rigidity of a relatively tall building via moment frame action. However, when such frameworks act compositely with RC tower-like building cores and shear walls, the simplified approach can be unreliable. Although far from being specific to the instances discussed here [103], it is prudent to reemphasize that making simplifying assumptions about behaviours of timber diaphragms may or may not yield accurate results. Thus, extrapolation should always be undertaken with all possible care. The only robustly reliable approach is detailed analysis of a system, combined when appropriate with full dynamic analysis. This Structural Engineering Document (SED) intentionally omits any discussion of Load-Delta (i.e. P-D) or other secondary effects, because these are applicable broadly to any structural analysis. The premise herein is that engineers will be cognizant of the generic issues that dominate the practices of structural engineering.

6.5 Additional comments Opportunities to design and construct large and possibly complex hybrid buildings employing timber as a primary structural material bring with them much scope for innovation. The design and analysis practices for tall buildings were developed during the late 19th and 20th centuries. The history of structural design is one of many successes, and occasionally failures of which some were catastrophic. The resulting accepted practices combine hard theories, practical experiences, statistical data, intuition, and professional judgement. In the 21st century, realization of new possibilities (such as using timber in new ways or rediscovering its lost applications) depends on the ability to prove that what is proposed will work without a long learning process. From the outset, new solutions must match or exceed contemporary competencies for


83

the construction of tall hybrid buildings. This is th e reason why it is repeated in this chapter and emphasized elsewhere in this SED that analysis, design, and construction practices that eliminate the possibility of doubts about the reliability of proposed designs must be adopted. This is essential as engineers shun the adoption of structural analysis and situation-specific design code rules of unproven robustness. Obviously, they should apply only fundamental structural engineering concepts whenever in doubt. Here dynamic response analysis of the case study systems is a metaphor illustrating the need for this prudent approach. Modern engineers have readily available analysis tools and know-how gained over more than a century of practical learning with other materials and computational developments of the last about half century to support them in this. Or more precisely, what they need is to apply these skills and tools to explore the use of timber. The case study analyses presented demonstrate very clearly that timber, and potentially other lightweight materials, can occupy niche applications that exploit their mechanical characteristics and ancillary features with high efficiency. From a construction viewpoint, the exploited attributes of CLT are essentially the same as those widely taken advantage of to construct large buildings from ancient to Edwardian times (Chapters 1 and 5). CLT panels are lightweight, mechanically efficient, “hold” fasteners well, and at the time of writing are becoming widely available in world markets as commodity or specialty products. Hence, their discussion is an example, as opposed to being prescriptive or exclusionary.

Acknowledgements The analysis on which this chapter is based was done by Professor Andi Asiz, Prince Mohammad Bin Fahd University, Kingdom of Saudi Arabia.

85 Chapter

7

Platform Construction Using Timber Plates: Special Considerations

Summary: Since around the beginning of the 21st century there has been strong and growing interest in using timber plates as substitutes for reinforced concrete (RC) and masonry elements in medium-rise building superstructures in European countries like Austria, Germany, Italy, Sweden, and the UK. The concept has also begun to spread to regions like North America and The Antipodes. The generally preferred types of timber plates are known collectively as cross laminated timber (CLT or XLAM). They consist of layers of relatively small rectangular lumber arranged such that the axes of these pieces “cross-reinforce” pieces in other layers. There are many proprietary CLT products available in thicknesses that range from around 50 to 500 mm. Their layers are bonded together using rigid adhesives or aluminium nails. Some products have stiffness and strength properties comparable with those of normal-weight RC slabs of equal thickness but at only one quarter of the RC mass. Apart from low mass to mechanical property ratios, the growing popularity of CLT products reflects the ease in cutting and shaping to form complex geometries and that prefabricated elements can be rapidly and easily joined together using simple methods. When encapsulated in non-combustible materials, CLT elements perform excellently and predictably during building fires. This chapter emphasizes the use of CLT in residential construction based on the so-called platform construction method. In such buildings, CLT panels are interconnected using simple mechanical fasteners like long slender screws and simple anchor brackets and ties. This chapter also discusses the key aspects of structural analysis and design of 40 to 50 m high buildings.

7.1

Introduction

Since the millennium, there has been strong and growing European interest in CLT or XLAM as superstructure elements in low-rise and medium-rise buildings. More recently interest has spread to places as diverse as North America and The Antipodes. Such products are manufactured as large plates/panels having three or more layers of finger-jointed structural grade softwood lumber, with layer thicknesses ranging from 17 to 38 mm (Fig. 7.1). Lumber pieces in some layers are arranged orthogonally to pieces in other layers, such that when layers are bonded together using rigid adhesives or aluminium nails, the composite arrangement is cross-reinforced in all directions in the same manner as plywood. Thus, when loaded in-plane, CLT panels are toughened against splitting in a manner that eliminates structural design problems associated with more traditional timber products. Depending on the layering employed, CLT plates can be

86

CHAPTER 7. PLATFORM CONSTRUCTION USING TIMBER PLATES: SPECIAL CONSIDERATIONS

made to have similar or dissimilar inplane or out-of-plane structural properties in orthogonal directions. To date, CLT products have been exclusively proprietary types manufactured under third-party quality assurance regimes. Most of these products have adhesively bonded layers. Plates can be manufactured to be flat or curved in a manner analogous to the production of glued laminated timber (glulam). The denFig. 7.1: Five-layer CLT product sity of CLT is typically around 500 to 600 kg/m3, and the mechanical properties are similar to those of normal-weight RC slabs of equal thickness. This makes CLT a highly viable substitute for RC and masonry, especially for load-bearing walls and floors. The relatively low mass is highly advantageous for building superstructures and foundations and makes CLT an ideal material for construction of tall buildings in seismically active regions. This chapter discusses design and performance of building superstructures constructed according to the so-called platform method wherein wall, floor, and sometimes roof elements are CLT panels assembled in a manner analogous to giant doll houses made from plywood. Although the focus is on CLT products, the approaches discussed could be adapted to the use of other plate elements made completely or partially from solid timber or reconstituted plant-fibre (e.g. bamboo) materials. Most of the examples included in this chapter are from R&D projects undertaken by the Italian National Research Council Trees and Timber Institute (CNR-IVALSA).

7.2

CLT as structural material

7.2.1

General characteristics

Although layers of CLT plates can be laminated using either mechanical fasteners or adhesive, for large structural systems like mid-rise buildings, adhesively bonded types are nearly always used because only such types attain suitable in-plane and out-of-plane stiffness. In this chapter, references to CLT are exclusively to products manufactured by gluing layers together using rigid structural adhesives. Pieces of lumber in layers are always face-bonded together (i.e. at interfaces between layers). Depending on the manufacturing method and manufacturer choices, pieces in some or all layers may or may not be edge-glued to adjacent pieces. Mostly melamineurea-formaldehyde (MUF) or polyurethane (PUR) adhesives are used and cured in vacuum or mechanical presses, with the vacuum presses suitable for products where pieces in layers are not edge-glued together. Glue-line thicknesses are 0.5 mm or less. When pieces are edge-glued, often that is only in face layers that will be exposed after buildings are finished. Unless moisture contents and distributions of CLT plates will remain constant from the time of manufacture until the end of their service lives as building elements, drying shrinkage of the lumber in across-grain directions is highly likely to cause cracking that makes edge-gluing of pieces at least only partially effective. Therefore, assuming that strength and stiffness properties will exceed those associated with only face-bonding between layers is unrealistic.

7.2 CLT AS STRUCTURAL MATERIAL

87

Also, gaps initially existent between pieces of lumber in layers, or those that occur because of shrinkage, result in some degree of permeability. This permeability has important implications for fire performance with respect to the integrity requirement that hot gases could not escape from the non-exposed side of an assembly during a fire resistance rating period ( Section 3.5). Consequently, encapsulation of CLT plates that form fire separation elements using non-combustible material is necessary. Similarly, layers of other materials are often added for thermal and sound insulation, such that CLT is the structural layer of composite wall and floor plates. Adhesively laminated CLT products are not intended for use in wet service conditions and should only be used in service environments such as Service Classes 1 or 2 as defined by Eurocode 5 [4]: •

Service Class 1 is characterized by a moisture content in the materials corresponding to a temperature of 20°C and relative air humidity only exceeding 65% for a few weeks per year.

•

Service Class 2 is characterized by a moisture content in the materials corresponding to a temperature of 20°C and relative air humidity exceeding 85% for a few weeks per year.

Service Class 1 corresponds to average moisture contents in most softwood not exceeding 12%, and Service Class 2 corresponds to average moisture contents in most softwood not exceeding 20%. For most multi-storey buildings Service Class 1 is applicable. To date, there is no generic production standard for CLT manufactured in Europe and all products manufactured there are proprietary. However, a production standard has been developed in North America [104], which acknowledges that products can be generic or proprietary. To date, the only design properties that exist are ones recommended by manufacturers or third-party technical organizations (e.g. Canadian industry led R&D organization FPInnovations, USAbased APA – The Engineered Wood Association). In European Union countries, design properties are published by third-party notified European authorities, in accordance with procedures that apply to all member states. From a structural engineering perspective, the crucial characteristic of CLT is that the crosslamination of the lumber in layers reinforces it against splitting. This means that at any point of time, material in some of the layers is resisting the stiff and strong in-plane force flows in the parallel to grain orientation. Therefore, unlike most other structural products made from timber, it is possible to make reliable tension and shear connections using mechanical fasteners installed through the thickness, irrespective of the in-plane loading direction. This does not negate the need to select connection methods and fasteners carefully, but it does translate into possibilities of using timber in novel ways (Section 7.5).

7.2.2

Typical design properties

Design properties of CLT are equivalent (also termed apparent) to elastic properties that ignore the true nature of those materials. Such practice mirrors those employed for structural design of other timber- and wood-based composites (e.g. lumber, glulam, and plywood) and is the approach on which contemporary design codes are predicated ( Section 2.1). Because products actually vary in respect of the number, thicknesses, and grades of material in layers (and in some cases have facing layers of materials other than timber), design properties

88


Fig. 7.2: Examples of arrangements of layers/plies in CLT products

vary depending on the orientations that plates will have when installed in buildings (Fig. 7.2). Consistent with materials like plywood and oriented strand board (OSB), properties are specified according to the planes in which stresses are expected to be applied. Table 7.1 shows 5th-percentile characteristic strength and stiffness properties for representative European CLT products. Strength classes C24 and C16 denoted in that table correspond to grades of lumber from which European CLT is often manufactured [105]. Properties in Table 7.1 are intended to be used in conjunction with the Eurocode 5 timber design code [4]. This means that characteristic strength properties are adjusted during design for combined effects of load duration and service class according to Eurocode 5. The importance of this is that the values in the table are not directly equivalent to CLT design properties adopted elsewhere. For example, it can be assumed roughly that reference strength properties applicable in North America are 80% of those in Table 7.1. Irrespective of that, it is however valid to assume that similar thicknesses of CLT are required in similar situations in various countries when sizing is controlled by similar design event levels (e.g. similar peak seismic ground acceleration and similar peak wind speed).

7.3

Platform construction concept

The most common approach for constructing multi-storey buildings from CLT panels is to use them as wall and floor elements assembled following the platform construction method. In this method once the panels that form the ceiling of one storey and the floor of another are installed, they become the working platform for the next storey (Figs. 7.3 and 7.4). In some cases, the roof is also constructed from CLT panels. This approach has been used extensively in Europe for construction of mid-rise buildings having up to eight CLT storeys placed

89

7.4 CONNECTION METHODS

Strength class of timber in laminates

Property (MPa)

C16

C24

Strength properties (5-percentile characteristic values based on short-term load)

Bending

f m,k

16

24

Tension parallel to grain in face layers

f t,0 ,k

10

14

Tension perpendicular to grain in face layers

f t,90 ,k

Compression parallel to grain in face layers

f c,0 ,k

Compression perpendicular to grain in face layers

f c,90,k

2.2

2.5

Shear in the plane of the plate

f v,k

1.8

2.5

Shear through thickness of the plate

f r,k

0.7

0.7

0.4 17

0.4 21

Stiffness properties (mean values)

Modulus of elasticity in bending parallel to grain in face layers E 0,mean

8000

11,000

E 90,mean

270

370

Shear in the plane of the plate

Gv,mean

500

690

Shear through thickness of the plate

Gr ,mean

50

50

Modulus of elasticity in bending perpendicular to grain in face layers

Table 7.1 Example of strength and stiffness properties for proprietary European CLT products

directly on a RC foundation or on top of one or more RC plinth storeys. At the time of writing, the tallest building was reported in Victoria Harbour, Melbourne, Australia: the 10-storey Forte Apartments building, of which nine storeys are made of CLT. The approach is generally economically and technically suitable in situations where room sizes are quite small, which means that walls are numerous and floor spans are limited. In such cases, plate thicknesses need not be large. Most of these buildings are for residential or mixed commercial and residential occupancies. Wall and floor elements arrive on-site, accurately precut to their final dimensions, with the cutting normally done by computer numerical control (CNC) machines. Sometimes windows, doors, and other non-structural items are preinstalled. Total construction periods are often very short once the foundation is ready, and site equipment required is not very specialized. In cases where rooms are relatively large, panels sometimes have preinstalled timber stiffening ribs and double-skins, to economically increase feasible spans/lengths (Fig. 7.5). When doubleskin wall or floor panels are used, it is necessary to adopt construction details that prevent undetected fires to spread through the void areas (Section 10.4).

7.4

Connection methods

Typically CLT panels are interconnected to create three-dimensional plate superstructures and to connect the panels to the foundations using simple metal fasteners, connectors, and anchor ties (Figs. 7.3 and 7.6 ). Such approaches are structurally efficient because:

90


Connection to the base with holddown anchors at openings and steel angles

Vertical joints between panels with LVL and screws

Vertical joints may also be overlapped with screws

Screws at corner joints Screws at joints between walls

1. Connection to the base with hold-down anchors at corners and at openings and steel angles

2. ...the floor panels are connected with screws to the underlying walls

Vertical joints between panels with LVL and screws

3.

4. ...the second floor is connected in the same way

...the wall panels of the second floor are constructed over the first floor and connected again with steel connectors and screws

5.

6. ...the construction is then completed very quickly

7.

Fig. 7.3: Assembly of a typical timber plate superstructure employing the platform construction method (LVL = laminated veener lumber)

1. High rigidity of the panels prevents development of significant in-plane warping at abutting panel edges, which means that most panels move relative to the joint interfaces through rigid body translation and rotation. 2. Line contacts that exist where plates abut are long, making it possible to create sufficiently rigid and strong connections without concentrating the internal force flows that might fracture CLT or foundation elements.

7.4 CONNECTION METHODS

91

Fig. 7.4: Installation of precut CLT plate elements

3. The general rigidities of each storey are sufficient such that even when hold-down connections between walls in adjacent storeys cannot be directly aligned or when wall openings are not the same at each storey, the superstructures respond as stacked rigid layers. This means that such systems are normally very robust and not typically prone to so-called soft-storey problems. (a)

(b)

Fig. 7.5: Long-span floor plates: (a) rib stiffened plate; (b) double-skin plate

Failures of connections made using slender self-tapping screws (and some other slender fasteners) happen in a non-brittle elastic manner when subjected to forces that cause shear or tension force flows through the fasteners. This means that plate superstructures constructed from CLT panels usually will not disintegrate, even if severely overloaded.

Experience to date is that slender screws (usually proprietary self-tapping types) having diameters 8 to 12 mm are suitable for line connections between CLT panels, with such screws being often available in lengths up to at least 600 mm. Screw installation can be achieved without predrilling in most softwood CLT products. Such screws efficiently transfer forces either laterally or parallel to their axes. Sometimes self-tapping screws are used to locally strengthen panels in compression, by inserting them as reinforcement akin to steel reinforcement in concrete. Such

92


B

A

C

D

A

Fig. 7.6: Examples of metal fasteners, ties, and connectors (A = shear connector; B = anchor tie, anchor bolt not shown; C = fasteners for attaching shear connectors/ties to CLT; D = selftapping metal screws)

reinforcement can be used to locally strengthen floor plates against crushing under pressures beneath supported walls, as an economical alternative to increasing wall thicknesses.

7.5

Structural analysis and design

This section presents an overview of how multi-storey CLT plate building superstructures constructed according to the platform method (Fig. 7.3) are intended to function structurally.

7.5.1

General aspects

7.5.1.1 Basis of analysis and design The assumption here is that such buildings are designed according to contemporary Partial Coefficient Limiting States Design practices, such as those specified in the model codes of the European Committee for Standardization [4,18,106–108]. Consistency with such codes requires that linear elastic response analysis be used to estimate internal forces in components that are the resulting effects of individual or combined design load cases; and that characteristic strength and stiffness properties of the CLT are apparent values used in conjunction with geometric properties determined from gross-dimensions of elements (Section 2.1). For example, the second moment of area to be used in combination with E 0,mean or E 90,mean from Table 7.1 is the element width multiplied by h3 /12, and the section modulus used in combination with f m,k from the same table is the element width multiplied by h2 /6 (where h is the panel thickness). 7.5.1.2 Load paths and robustness Load paths in timber plate superstructures are quite transparent. Vertical force transfers occur directly from floors and roof plates to plates that form walls, with force in walls accumulating

7.5 STRUCTURAL ANALYSIS AND DESIGN

93

Fig. 7.7: Conceptualization of force flows resisted by CLT walls

through the heights of superstructures. According to the platform construction method, horizontal force transfers also occur cumulatively from level to level through the heights of superstructures. Two important aspects to consider with respect to the load flows between storeys are the rigidity of those storeys in plan and in elevation. Once assembled, unless plan shapes are irregular or very elongated, storeys within superstructures like those in Figs. 7.3 and 7.4 are quite rigid both in plan and in elevation. This means that floor platforms tend to behave as rigid, rather than as flexible diaphragms. The simplified approaches outlined in Subsections 7.5.1.3 and 7.5.1.4 for designing floors and walls are based on not making specific presumptions about the rigidity of horizontal diaphragms and adopts the worst interpretation of how forces may flow from floor platforms to walls. The question of how to consider the behaviour of walls and the effect of their behaviour on rigidities of complete storeys is more complicated. In-plane rigidities of individual wall segments, and therefore complete walls and assembles of walls, depend on the following: •

The extent of wall openings

•

How wall segments are interconnected

•

How wall segments are connected to horizontal elements that are floors, the roof, and the top of the foundation

•

The extent to which walls that are directly within a load path are stiffened and strengthened by transverse walls

Often, it is conservatively assumed that walls or segments within them are not stiffened or strengthened by transverse walls. Two extreme cases are considered. In the first, walls are cut from single pieces of CLT (Fig. 7.7 ). In the second, they are assembled from CLT plate segments (Fig. 7.3). These two scenarios can result in distinctly different behaviour mechanisms in terms of distortion, and therefore flexibility and internal force flows within walls subjected to racking forces that result from lateral design loads on superstructures. Figure 7.8 illustrates the two extreme cases. Intermediate cases will also exist depending on factors such as whether or not edges of abutting CLT panels in segmented walls are mechanically interconnected. Therefore, when performing wall design (as discussed in Subsection 7.5.1.4 ) the influences of construction details on the racking deformation mechanism and strength of CLT panels and their perimeter connections must be considered. Structural robustness of multi-storey building construction needs to be considered. In superstructures like those discussed in this chapter, the honeycombed nature of their arrangements

94


(a)

Base CLT connection

Floor platform

CLT wall panel

Top CLT connection

Floor platform

(b)

Base CLT connection

Floor platform

CLT wall panel

CLT wall panel

Top CLT connection

Floor platform

Fig. 7.8: Racking behaviour of CLT walls: (a) one-piece wall; (b) segmented wall

formed by floor and wall plates promotes even redistribution of forces flowing from one storey to another. Therefore most, if not all, walls in the various storeys participate in resisting effects of other than locally applied design loads. Superstructures like those in Figs. 7.3 and 7.4 contain much structural redundancy and forces would be redistributed relatively benignly were particular structural elements to fail during any type of loading event. Consequently, the general likelihood of progressive collapse is very limited, and timber plate systems such as the ones discussed in this chapter are highly robust, if properly designed and constructed. The most likely system-level vulnerabilities are connections that link storeys together, such as the roof to the walls or the superstructure to the foundation. Therefore, close attention should be given to selection and design of shear and hold-down connections resisting sliding and uplift due to wind and seismic loads. Connections used at level-to-level interfaces should contain redundancy and ductility. In the case of hold-down anchoring between elevated storeys, the flow of forces should be whenever possible directly from wall to wall, rather than from walls to floor platforms sandwiched between them. Discussion here applies directly to situations where elevator and staircase shafts are constructed from CLT or comparable timber panels. In such instances, no specific distinctions need to be made for the purposes of structural analysis between wall elements forming those shafts and other wall elements. 7.5.1.3 Design of floors Floor plates are designed considering them to act as one-way spanning elements (i.e. ignoring that panels are interconnected at abutting side edges, as seen in Fig. 7.9). This consideration reflects that the edge-to-edge joints are made using simple carpentry lap joints fastened by screws or just screws. Such joints are intended to transfer shear forces but not moment forces that result from vertical loads on floors. This approach tends to be conservative. Most often, serviceability performance-related bending deflection criteria control the required plate thickness. The deflection criteria take the form of the maximum acceptable ratio of deflection to span and are selected to avoid damage to non-structural elements of buildings (e.g. 1/500). In some cases

95


Fig. 7.9: One-way span analogy for design of CLT floors Distribution of normal stresses due to bending within the cross-section

Distribution of shear stresses within the cross-section

Compression Rolling-shear M

M

Rolling-shear V

V

Traction

Fig. 7.10: Analogies used to represent internal stress distributions due to moment and shear forces

such criteria are intended as indirect control of floor motions resulting from building use, like effects of footfall impacts. However, neither static deflection nor simplified dynamic analyses are robustly reliable ways of ensuring satisfactory vibration serviceability of CLT floors [109]. For the types of buildings discussed, installation of floating floors over CLT slabs is common and is an effective solution to vibration serviceability problems. When designing floor plates for strength it is necessary to consider both bending and shear strength. Design strengths such as those in Table 7.1 take into account the layered nature of plates (Fig. 7.10). Bending strengths are based on the assumption that only CLT laminations with lumber oriented parallel to the plane in which bending moments occur resist forces. However, such complexity is ignored during normal design, because it is integrated into the apparent design properties (Table 7.1). The explicit design consideration required with respect to ensuring floors behave adequately as horizontal diaphragms depends on factors like floor layouts and plate element connections as discussed in Subsection 7.5.1.1. 7.5.1.4 Design of walls Using CLT panels as walls provides interesting architectural options, because their in-plane stiffness and strength allow them to resist gravity, uplift, and racking design forces; to span gaps in facades; and to create overhanging storeys (Fig. 7.7 ). These applications have often been beyond the capabilities of more traditional timber construction methods (e.g. light-frame, post and beam) because of the requirement of large dimensions of elements, difficulties connecting elements, and affiliated high costs. CLT walls can consist of panels with or without stiffening ribs or as double-skin box elements with multiple glulam webs ( Fig. 7.5). The use of stiffened

96


plates usually permits economical construction of walls that can resist high vertical compression forces at the lower storeys of tall and slender superstructures. When superstructures are not structurally slender, it is often possible to use a number of quite simple approaches to estimate the flow of the horizontal forces (i.e. wall racking forces) that are caused by lateral design loads (i.e. equivalent static wind or seismic forces) to the walls. In cases where floor plans and wall openings are replicated between storeys, and the extent of openings is limited, buildings are essentially symmetrical about a vertical plane passing through the centroid of the building’s footprint. In such cases, it is often adequate to: •

ignore the resistance of walls that are not parallel to the vertical plane in which the superstructure’s response is being evaluated and

•

assume that the floor platforms and roof behave as completely flexible diaphragms (pro jected load area method) or as perfectly rigid diaphragms (relative wall stiffness method), and to take the worst outcome from those extreme case assumptions as the design force.

In other instances, like when systems are not essentially symmetric on plan but storeys are replicated, the same approach can be taken to estimate components of horizontal force flows to the walls. However account has to be taken of interactions between those forces and ones that are associated with bending and shear distortion in planes coincident with a vertical plane passing through the centroid of the building’s footprint. In such instances it is also necessary to add components of horizontal force flows that occur because of torsional distortion about the superstructure’s vertical axis (i.e. effects of torsion force due to plan eccentricities). The approaches outlined above, or similar ones, will usually provide a sufficient basis for deciding whether a building design concept is feasible, or if a more refined follow-up structural analysis is necessary. When further investigation is necessary, a finite element analysis may be required to represent CLT panels in walls and floors, the roof substructure and connections between elements, and any other substructures in the superstructure and foundation. A refined analysis will, for example, account for interactions between walls that do not lie in the same plane (Fig. 7.11). In many instances a refined follow-up analysis will not be required, but it is unwise to assume that is the case. The combined effects of vertical and horizontal force flows for various design load combinations (e.g. effects of factored dead and seismic loads) will determine the minimum required dimensions of wall panels and connections. Unless a specific rule is mandated by locally applicable design codes, use of the summation of the ratios of the factored force effects to the factored resistances approach is suggested: T f R f ___ Tension + Racking: + ___ ≤ 1.0 T r Rr

( )

C f 2 R f ___ Compression + Racking: + ___ ≤ 1.0 Rr C r

(7.1)

(7.2)

where T f is the factored tension force, T r is the factored tension resistance, C f is the factored compressive force, C r is the factored compressive resistance (accounting for either crushing or buckling), R f is the factored racking force, and Rr is the factored racking resistance.

97

7.5 STRUCTURAL ANALYSIS AND DESIGN Seismic actions:

Realistic analogue: Effective tube

Realistic analogue: Effective tube

Foundation Seismic or wind and heavy roof

Wind and light roof

Fig. 7.11: Interaction of wall panels directly in load paths with other walls to resist effects of lateral loads Screw connections:

1 wall panel-to-panel 4

9

3

2 floor panel-to-panel

1

3 wall corner 4 floor-to-wall junction

6

9

2

4

4

7

Hold-down connections:

5 wall-to-foundation 6 wall-to-floor-to-wall (using anchors) 3 5

8

7 wall-to-wall (using tie strap) Shear connections:

8 wall-to-foundation 9 wall-to-floor

Fig. 7.12: Example locations and types of connections requiring structural design

The foregoing does not apply without modification to hybrid superstructures in which multistorey assemblies of CLT plates work in combination with substructures of other types. This exclusion applies, for example, when a building has primary lat eral load-resisting systems constructed from RC or reinforced masonry. 7.5.1.5 Design of connections Figure 7.12 shows an illustrative example of connection types and locations in a timber plate superstructure that require structural design.

Because suitable metal fasteners, connectors, and anchors are mostly proprietary products, their design properties must usually be acquired from manufacturers or third-party technical organizations that conducted tests on behalf of manufacturers. Consequently, engineers must

98


(a)

(b)

A

B

C

D

E

F

G

H

I

J

Fig. 7.13: Selection of connections joining CLT plates: (a) inappropriate and appropriate wall panel to wall panel connections (methods A and B are unsuitable, method C is viable but not recommended, methods D and E are recommended); (b) inappropriate and appropriate floor panel to wall panel connections (methods F and G are unsuitable, method H is viable but not recommended, methods I and J are recommended)

assure themselves that the available information is consistent with the needs of specific design projects and applicable design codes. This necessity will, for example, ensure that available design information is based on the appropriate definition of capacity (in terms of the associated failure mechanism), deformation at failure, ductility ratio, and indexing effects of factors like duration and loading-type. Data in the public domain mostly indicate that connections in CLT made using slender self-tapping screws have high ductility when used to resist shear flows between panels [110], but it is prudent to check that such behaviour apply to particular product brands. When installed in systems of the type discussed in this chapter, screws and other dowel-type fasteners usually have design capacities that are higher than for products like glulam that is manufactured from the same grade and species of lumber, because of the already mentioned toughening that CLT possesses from cross-laminating of lumber. Figure 7.13a shows the suitability of sample connectors for joining wall panels at building corners, and Fig. 7.13b shows the same for attaching floor or roof plates to tops of walls. In each application, desirable connections minimize the likelihood of causing delamination type splitting of CLT (i.e. avoid wedging the layers apart).

7.5.2

Expected performance during seismic events

A universally adopted concept underpinning the seismic design of buildings is that superstructures occupied by humans should be designed to survive ground shaking during design level events (i.e. the maximum credible earthquake) without collapsing. Local regulatory restrictions, philosophies of the building designers, and building owner specified requirements will often place more onerous limits than codes for the purpose of damage minimization. These are cases


99

where simply satisfying the requirement of preventing building collapse is insufficient. The following paragraphs discuss how well-designed and properly constructed CLT plate superstructures can be expected to perform during seismic events. As kinetic energy is imparted to buildings during earthquakes, building foundations move either synchronously or out of phase with the supporting ground depending upon (1) the nature of the local geology and soils that surround them, (2) the construction system, and (3) the global stiffness and mass of each building. How superstructures move/oscillate during earthquakes depends on motions to which they are referenced (i.e. movements at interfaces between superstructures and foundations); the superstructure’s form, stiffness characteristics and mass distribution; the magnitude and distribution of any supported masses; and the state of damage that pre-existed or occurs during a seismic event. Similarly, individual storeys and substructures within superstructures respond depending upon the motions to which their equilibrium is referenced and their local characteristics and characteristics of whatever they support. System stiffness characteristics and mass distributions influence the frequencies at which superstructures and complete superstructures vibrate freely. Therefore, the proneness of particular systems to resonate as the ground and foundations shake beneath them and to amplify motions experienced by building elements and building occupants are variable but can be controlled. Frequency tuning is the most reliable way of ensuring that the most potentially energetic natural frequencies of any superstructure are unlikely to coincide with the likely ground shaking frequencies. Tuning building response frequencies such that they are higher than expected ground shaking frequencies is most desirable. Such “high-end” tuning requires that the superstructure’s stiffness to mass (including supported masses) ratio be maximized. Thus, making use of lightweight materials such as CLT is highly desirable (Section 7.1). Internal forces generated in elements of superstructures are proportional to their mass and acceleration relative to masses and accelerations of elements to which they are connected; proportional to their stiffness and stiffness of elements to which they connect; and proportional to energy sinks other than themselves that can dissipate kinetic energy flowing through them and the rest of the building to which they belong. Incorporating flexible (“soft”) elements or storeys in superstructures maximizes the chance that parts of superstructures will move out of sync. Such an approach is likely to maximize relative accelerations and therefore internal forces that interconnected elements must resist. If existence of soft elements or storeys coincides with the ne ed to support large masses, the possibility of force flows destructive to superstructure elements can increase dramatically. This also militates in favour of using materials like CLT that have high stiffness to mass characteristics and the use of connections that are stiff and thereby act as sinks for kinetic energy. Because of the way in which CLT plate superstructures are assembled, there are many frictional interactions between wall plates, and between plates in walls and plates in floor platforms. Those interactions dissipate kinetic energy flowing through such systems. The frictional damping potential of the wall to floor platform contacts increases towards the ground, because of the increasing overburden mass of the supported floors. Mechanical connections are also primary energy sink sites, but their potential cannot be mobilized without damage (e.g. plastic yielding of fasteners and/or crushing of CLT beneath them). Consequently the amount of damping superstructure mobilization would be considerably greater than that affiliated with material damping (which is widely reported to be in the region of 1% of viscous damping [15]). That CLT platform construction has a high inherent damping was confirmed by full-scale shake-table testing of a

100 CHAPTER 7. PLATFORM CONSTRUCTION USING TIMBER PLATES: SPECIAL CONSIDERATIONS three-storey superstructure [15], and later by tests on a seven-storey structure as mentioned in Section 7.6 . Consequently, seismic design strategies and methods for timber plate superstructures must recognize the potential influences that construction site characteristics, structural form, construction material, construction details, and building occupancies have on the overall responses of particular foundation and superstructure systems. An example of such a suitable design strategy and method is presented in Section 7.6 .

7.5.3

Design manuals

Technical organizations affiliated with wood industries have begun to produce design manuals that give geographic region based appropriate advice on design and construction of CLT plate building superstructures [111,112]. Such manuals contain information about structural and nonstructural aspects of design and construction. The structural advice relates to topics like the provision of generic characteristic design properties for CLT elements and information about how to design elements and substructures (e.g. buckling capacities of CLT wall elements). Suggested properties are lower bound estimates not specific to particular proprietary products and are, therefore, suitable for feasibility design studies and designs that do not necessitate the full capabilities of CLT products. What must be remembered is however that handbook information is essentially a starting point for practitioners unfamiliar with a specific material, and not necessarily the latest and best information. Engineers who invest time in seeking out information directly applicable to products in the marketplace and who use advanced analysis and design practices (elaborated in the technical literature) can reap substantial rewards in terms of project costs and building performance.

7.6

Example of seismic design practices

7.6.1

Background

This section illustrates seismic design practices based on the approach employed in Italy for the design of mid-rise CLT plate superstructures, based on research undertaken by the Italian National Research Council Trees and Timber Institute (CNR-IVALSA), with financial support from the Autonomous Province of Trento. What is done in Italy has been strongly influential in development of parallel practices elsewhere and can, therefore, be regarded as representing current best practice. In general, superstructure responses can be determined through linear dynamic analysis by applying seismic displacements at the base of a building according to the response spectrum method. In Italy, as elsewhere, provided some of the restrictions are met, design codes also allow the use of simpler equivalent static force methods to estimate factored force flows in the elements of a Seismic Force Resisting System (SFRS). In essence, the restrictions applicable to the simpler method are that buildings must not be structurally slender and must have regular geometry plans and elevations. To date, relatively tall multi-storey CLT plate superstructures constructed in Italy and other seismically active locations have conformed to these restrictions, thus permitting the use of equivalent static force methods of seismic design. This approach is discussed below.

7.6 EXAMPLE OF SEISMIC DESIGN PRACTICES

101

According to Eurocode 8 [18], the Factored Based Shear Force for a SFRS of a superstructure having total weight W is: F b(T 1) = g S (T 1)W I b

(7.3)

where S b(T 1) is the ordinate of the design spectrum at the building period T 1 (i.e. at the fundamental period of the SFRS) and taken to be: 2.5 S b(T 1) = agS ___ q

(7.4)

In these equations g is the importance factor for the building (1.0 for residential buildings and I 1.5 for strategic buildings), ag is the design ground acceleration assuming T 1 equals 0.2 seconds, S is the soil factor, and q is the behaviour factor. The assumption that T 1 is 0.2 seconds corresponds to the plateau of the European design spectrum [18]. Factor q is also known variously as the q-Factor , action factor, or force modification factor. It has the function of reducing design forces obtained from linear structural analyses to ones that account for capabilities o f the SFRS to dissipate energy (Section 2.6 ). Apart from q, the variables in Eq. (7.3) are specified independently of the materials and type of SFRS. Therefore, q is the quantity that requires to be quantified for CLT plate superstructures. This was done based on the calibration of q such that the results of simplified equivalent static force analyses match the results of detailed non-linear analysis studies and shake-table tests on isolated wall assemblies and complete building superstructures. That calibration exercise indicated that taking q as equal to 3.0 achieves acceptable design solutions, with values estimated during detailed background studies falling in the range of 2.5 to 4.6 [15,113].

7.6.2

Seven-storey case study

The example is a seven-storey superstructure which is an actual CLT plate building in the Italian Alpine city of Bolzano (Fig. 7.14). The SFRS of that building was tested full-scale on a shake-table in Miki, Japan (Fig. 7.15). During the shake-table tests, the SFRS was subjected to 12 historical earthquake records scaled to between 50% and 100% of the observed peak ground accelerations. The real earthquake records were interspersed with artificial, stepped earthquake

Fig. 7.14: Seven-storey building in Bolzano

102 CHAPTER 7. PLATFORM CONSTRUCTION USING TIMBER PLATES: SPECIAL CONSIDERATIONS Level South view

23.20 7

Storey 7 18.55 6 Storey 6 15.46 5 Storey 5

23.5 m

12.37 4 Storey 4 9.28 3 Storey 3 6.19 2 Storey 2 3.09 1

13.5 m 7.5 m

Storey 1 0.00 0

OST EG

1. OG - 6. OG

z

x

4 4 , 3 1

D 4 R 4 , O 3 1 N

0 6 , 1

D Ü S

0 6 , 1

7,68

y

7,68

WEST

Fig. 7.15: Seven-storey SFRS test

records of relatively low intensity to detect whether the system’s response changed during “real” earthquakes. The artificial tests indicated limited changes in the lowest order, natural frequencies of the system, throughout the complete series of earthquakes. This finding was taken as prima facie evidence of robustness and that any structural damage was localized. For the X-z and Y-z planes (as defined in Fig. 7.15) the lowest natural frequencies were 2.34 Hz and 3.32 Hz prior to application of earthquake records. After application of 10 earthquake records the fundamental natural frequencies had changed to 1.95 Hz and 2.93 Hz respectively. The system also remained self-centring throughout (i.e. free from residual distortions). Application of seismic events much stronger than those believed to be credible for Bolzano or other parts of Italy demonstrates the practicality of designing and constructing systems capable of sustaining no more than superficial damage during design level earthquakes (Recommendation 3 from Chapter 2). Combining Eqs. (7.3) and (7.4): 2.5 W F b = g a S ___ q I g

(7.5)

Application of Eq. (7.5) to the seven-storey SFRS is based on: g I = 1.5, reflecting that the size and usual height for a timber building, W = 290 tonne, ag = 0.82g = 0.82 × 9.81 = 8.04 m/s for X-axis direction (perpendicular to primary plan axis), = 0.60g = 0.60 × 9.81 = 5.89 m/s for Y-axis direction (parallel to primary plan axis),

103


S = 1.25, which corresponds to type B soil (very dense sand or very stiff clay), q = 3.0.

The values of ag used here correspond to peak values used during shake-table tests (i.e. not those applicable to design of the building in Bolzano). The mode shapes for lowest order beam modes are assumed to result in linear increases in horizontal displacement amplitude from the bottom to the top of the SFRS, thus resulting in the horizontal equivalent lumped-static-forces values as follows: F i = F b

zi mi 6

∑ z m j

j =1

j

(7.6)

i = 1, 6 (i.e. six lumped masses represent the SFRS)

where i signifies the level of a lumped mass, and zi is the height of a lumped mass of magnitude mi above the base of the SFRS. Figure 7.16 shows the lumped mass values, associated zi values, and resulting individual and accumulated F i values at various storeys. The accumulated F i values (∑F i values) at any level are the shear flows to be resisted by the SFRS below each lumped mass position. As can be deduced from horizontal shear flows (∑F i values) in Fig. 7.16 , the largest structural demand in terms of horizontal shear to be resisted occurs at the base of the SFRS, in the narrower plane direction (X-z plane). This is a consequence of both the relative slenderness of the building in that plane, which leads to the longest actual beam mode period and the relatively low lengths of CLT walls that are available to act as shear walls in the X direction. Practices for determining force flows to individual wall panels and connections in the lower and other storeys follow the principle outlined in Subsection 7.5.1.4. Because floor plans were quite simple and reasonably close to symmetric (Fig. 7.15), the design results were not very sensitive to the type of force flow analysis used to convert F i values into force flows in wall elements and their connections. Figure 7.17 shows the types of shear connectors and anchors used to resist sliding and overturning at elevated floor levels and at the base of the test system. The hardware used was a mixture of commercially available products and specialty anchors. The building in Bolzano has less heavy-duty connections, because the design peak ground acceleration is less. Detailed analysis of test information suggests that connections in the tested SFRS were overdesigned relative to actual demands on their capacities by around 30%. As is normal for timber building superstructures, the connections were primary in defining the behaviour of the SFRS, and their selection and detailed design was crucial to obtaining and exceeding the desired structural performance.

The maximum lateral drift observed during tests was in the order of H /80 for the X-z plane and H /130 for the Y-z plane, where H was the total height of 23.5 m. However, as noted previously, the earthquake records used in these tests included 100% of the ground accelerations observed during the actual earthquakes with one being the highly destructive Kobe earthquake on January 17, 1995. The maximum peak ground acceleration for Italy is only 0.35 g

104 CHAPTER 7. PLATFORM CONSTRUCTION USING TIMBER PLATES: SPECIAL CONSIDERATIONS

F i

F i F i

F i

F i

F i

F i

F i

F i

F i

F i

(a)

F i

(b) Level 7

Level 7

F i

F i Level 6

Level 6

Level 5

Level 5

F i

Level 4

Level 4

F i

Level 3

F i

Level 2

F i

Level 1

F i

Level 3

F i

Level 2

F i

F i

Level 1

F i

F i

Level 0

Level 0

Level

zi (m)

mi (tonne)

F i (kN)

6

18.55

20.1

488

5

15.46

48.1

4

12.37

3

F i (kN)

F i (kN)

Level

zi (m)

mi (tonne)

F i (kN)

488

6

18.55

20.1

357

357

974

1462

5

15.46

48.1

713

1070

51.3

831

2293

4

12.37

51.3

608

1678

9.28

53.5

650

2943

3

9.28

53.5

476

2154

2

6.19

55.4

449

3392

2

6.19

55.4

328

2482

1

3.01

61.2

247

3639

1

3.01

61.2

181

2663

Fig. 7.16: Lumped m i and associated zi values, and calculate F i and ∑ F i values: (a) X-z plane; (b) Y-z plane

[113], and drift levels that might occur in a building like the one in Bolzano are in the order of H /200 or less. Therefore, lateral drift levels would actually be within the range of target limits typically specified by design codes. Drift predictions for the tested system obtained via

105


(a)

(b)

(c)

IVALSA hold-down

2 4 1

(d)

IVALSA hold-down

Fig. 7.17: Connection hardware in SFRS test: (a) IVALSA hold-down anchor used at bottom storey; (b) Simpson HTT22-hold-down used at elevated storeys; (c) shear connector used at upper storeys; (d) shear connector used at bottom storey

relatively complex dynamic analysis were accurate, but a pushover analysis would also result in a satisfactory SFRS. Pushover analyses should account for flexural and shear deformation in CLT wall elements and any flexibility of connections that might lead to horizontal relative sliding or vertical opening (relative rotations of storeys because of uplift). Although not elucidated in detail here, the design of this seven-storey SFRS’ elements was otherwise quite straightforward. Finally, the admissibility of equivalent static force design depends upon the estimation of T 1. Therefore, engineers are required to estimate the fundamental natural periods associated with each design orientation (e.g. planes X-z and Y-z in the discussed building). Usually this cannot be done with high accuracy, and opinions differ on acceptable methods. Consequently, given that experience with design and performance of CLT plate superstructure is so far quite limited, the surest fall-back approach is to base estimates on dynamic analysis. However, even if that is done, the details of the SFRS will not be known at first, and a seed estimate of T 1 is required as the basis to begin iterative component sizing. For this, various empirical formulas have been suggested, with an example being: T 1 = 0.05 H 0.75

(7.7)

For the tested seven-storey system, this yields the estimate of T 1 = 0.05 × 23.50.75 = 0.53 seconds (fundamental frequency f 1 = 1.9Hz). This compares with the actual measured T 1 values of the system prior to the application of a series of earthquake records of 0.43 and 0.30 seconds for X-z and Y-z planes, respectively. Approximate formulas like Eq. (7.7) have limitations, and engineers need to satisfy themselves concerning what depth of analysis is appropriate for particular projects.

106 CHAPTER 7. PLATFORM CONSTRUCTION USING TIMBER PLATES: SPECIAL CONSIDERATIONS

7.7

Additional comments

As implied by the types of applications discussed in this chapter, the construction of taller, multi-storey building superstructures using only timber plates, such as CLT, as primary loadbearing elements can handle certain niche situations. Examples are cases where the load transfers from above are widely dispersed such that neither floor platforms nor wall elements will be subjected to highly concentrated force flows on their horizontal surfaces. This means that the viable number of storeys is finite, as are viable room dimensions. Building superstructure geometries will largely determine the limits on the numbers of storeys and height, because the geometry along with building occupancy and building location controls the intensities of design stresses in wall and floor platform elements. Roughly speaking, the more squat a superstructure is the greater the number of storeys that will be feasible. Most probably timber plate buildings will be multi-occupant dwellings, hotels or non-mercantile business premises. The type of occupancy will bring with it certain restrictions (e.g. related to building functionality or occupant preferences). Therefore, storey limits will likely flow from the maximum heights that engineers can achieve within an architecturally defined footprint and not on unfettered limits of engineering skill. What engineers will be able to achieve (with this particular type of construction system) will be governed by mechanical capabilities of timber panels and connection hardware and economics, once the architectural decisions are made. To make the engineering work beyond heights bounded in that way, structural systems will require fundamental alterations which mean solution are no longer classifiable as timber plate superstructures. The rough rule of thumb is that for residential and similar building occupancies the minimum wall thickness for three-storey systems is about 90 mm. Each additional storey requires a 10 mm increase in wall thickness per storey (except for the top three floors). Therefore, a 14-storey alltimber plate building superstructure would require 200 mm thick walls in the lowest storeys.

m 0 5

2 7 m

2 5 m

Vista 3D Strutt ura in legno X-LAM

Seizone Struttura in acciaio

Struttura in cemento armato

Fig. 7.18: Fifteen-storey CLT building concept designed for Northern Italy (Dante O. Benini & Partners, Milano)


107

The practical limit for the number of storeys or height for timber plate buildings cannot be stated exactly. To date, buildings up to nine CLT plate superstructure storeys have been successfully constructed in diverse locations like Berlin (Germany), London (UK), Milano (Italy), and Melbourne (Australia). A 15-storey CLT plate superstructure design concept has been created for Northern Italy (Fig. 7.18). However, as the cutaway schematic for that building design shows, the building’s superstructure consists of a 13 storey CLT plate assembly that wraps around a structural framework at the building’s core and sits on top of two above ground RC plinth storeys. This reflects that, as has already been conjectured, using only timber plates becomes impractical at around that height. Economic and other non-technical considerations suggest that the maximum number of storeys that will ever be constructed from timber plates alone lies in the range 12 to 15 (40 m and 50 m).

Acknowledgements This chapter incorporates ideas, information, diagrams, and photographs suppl ied by Professor Dr. Ario Ceccotti, who at the time of writing was the Director of the Italian National Research Council Trees and Timber Institute (CNR-IVALSA). Thanks are also due to Dr. Andrea Polastri of CNR-IVALSA who assisted with finalization of this chapter.

109 Chapter

8

Example Project 1: Six-Storey Hybrid Building in Quebec City, Canada

Summary: Prior to the relatively recent introduction of objective-based building codes in Canada, use of structural timber in new multi-storey building superstructures was commonly restricted to four above-ground storeys. However, the picture has begun to change since the publication of the 2005 edition of the National Building Code (NBC) of Canada, which permits what are termed “Alternative Solutions” as suitable ways of designing buildings, as an alternative to “Acceptable Solutions” that are automatically deemed to comply with design requirements. This chapter discusses the use of the Alternative Solutions pathway in fire design of a six-storey mostly timber hybrid office building in Quebec City, Canada. The structural system of the building was designed by standard engineering design methods that fall within the scope of Acceptable Solutions. The building has a glued laminated timber (glulam) structural framework and horizontal diaphragm substructures, and reinforced concrete (RC) shear walls and foundation. Post-construction measurements of the lateral vibration response and post-construction vertical settlement of the superstructure indicate that the building performs as intended.

8.1

Background

There are many fairly tall historical buildings in Canada that have primary timber-frame superstructure systems, with the tallest being the McLennan and McFeely Building in Vancouver (Fig. 5.5). However, during most of the last century, building regulatory provisions explicitly prohibited or limited the use of timber in multi-storey buildings other than very low-rise ones, with many jurisdictions applying a limitation equivalent to four storeys. Publication of the 2005 edition of the National Building Code (NBC) of Canada recognized that “Acceptable Solutions” and “Alternative Solutions” are suitable ways of designing buildings to achieve specific performance objectives [17]. The latter approach was introduced to facilitate and encourage the use of more technological innovations based on R&D, advanced engineering design tools, and other demonstrably reliable approaches. In the remainder of this chapter, mention of “accepted design” and “alternative design” are synonymous with the design classifications Acceptable and Alternative Solutions as defined by the NBC of Canada. Under the Canadian building regulatory system, the provisions of the NBC only become legally binding after they are adopted by provincial or municipal governments. In some instances there are deviations between the NBC model language and the actual building code requirements in a particular jurisdiction, but such deviations are relatively minor. Although the NBC was revised in 2010,

110

CHAPTER 8. EXAMPLE PROJECT 1: SIX-STOREY HYBRID BUILDING IN QUEBEC CITY

those revisions had not been adopted in Quebec, and the discussed building was designed according to the 2005 edition. When a timber building is not permitted under Acceptable Solutions provisions, most often it is because combustible materials are used as the load-bearing elements/assemblies within superstructure systems. Architects and engineers who propose Alternative Solutions must, as a minimum, demonstrate to building regulatory officials that the proposed solution will have a performance that is at least equivalent to prescriptively accepted solutions. Building regulatory authorities have the legal responsibility to analyse the affiliated technical documentation that underpins each proposed alternative design and decide whether it meets the NBC objectives. As discussed in Chapter 1, this is fostering a revival of interest in using timber and other new or modified products (e.g. structural glass and plastics) for diverse construction purposes from which Acceptable Solutions requirements had barred them previously. The remainder of this chapter discusses the design and construction of a six-storey building superstructure in Quebec City, Canada that was justified to the relevant regulatory authorities based on a special design concept and the use of timber construction products. As with many timber buildings, the main challenge was to address concerns of regulatory officials related to fire performance objectives. Consequently, advanced fire modelling and performance-based techniques became the main tools for demonstrating equivalency to accepted solutions. The structural system was designed via Acceptable Solutions methods.

8.2

Superstructure system

8.2.1

Description and construction

The six-storey superstructure is 22.1 m high, and the footprint is about 1000 m2 with the major plan axis being about 2.5 times the length of the minor axis. As shown in Fig. 8.1, the building superstructure has three flat facades, and one curved and sculpted facade with exposed steel columns that are sheathed in timber at the lowest above-ground storey. The roof is flat with a parapet wall and few minor obstacles such as pedestrian access stairheads and physical services. Primary superstructure elements are RC walls that act as shear walls and enclose fire escape and firefighting routes and glulam framework and horizontal diaphragm elements. The superstructure is anchored to a level RC slab that is monolithic with an underground three-level RC parking garage that forms the foundation. The building was constructed in stages, wherein the RC wall elements were cast in situ, prefabricated glulam framework and diaphragm elements were added, and floors were completed structurally. Although floors were added and completed approximately in harmony with the addition of the framework for different storeys, the construction sequence was coordinated such that two-storey framework segments could be preassembled and then lifted into place (Fig. 8.2). These segments are located in vertical planes parallel to the minor building plan axis. Therefore, within the superstructure there is continuity of column and girder elements across some connections. The framework functions such that main floor and roof girders are connected to prefabricated frame segments. Once the system was completed, floors functioned as continuous diaphragms. All glulam elements were cut precisely to length and shape, and holes for dowel fasteners were drilled using computer numerical controlled (CNC)

111

8.2 SUPERSTRUCTURE SYSTEM

(a)

(b)

m . 2 2 0

m 2 2

5 2 .6 m

Fig. 8.1: Six-storey office building in Quebec City, Canada: (a) completed building (courtesy of FPInnovations); (b) framework elements (courtesy of Nordic Engineered Wood)

machines. This resulted in less workmanship skill sets and was essential for ease and speed of construction. Precise cutting of glulam elements avoided the possibility of distorting or damaging the superstructure by force-fitting parts. Columns have one piece, and girders are made from either one piece or two pieces of glulam. In some cases, girders fit together as composite elements that wrap around the column, thereby resulting in desirable connection and framework actions. As seen in Fig. 8.2, secondary framework elements were connected to primary framework elements by brackets; floor slab elements were connected to the framework elements to achieve composite action; and in some locations special framework details were adopted (Fig. 8.2f ). After the structural parts were added to the superstructure, the roof was made watertight, and the exterior claddings were added. Finally the building was finished internally. The construction process is illustrated schematically in Fig. 8.3. In the finished state, the glulam structural elements are visible inside the building.

112


(a)

(b)

(c)

(d)

(e)

(f)

Fig. 8.2: Installation of glulam superstructure framework and slab elements: (a) two-storey segment; (b) interconnection of segments; (c) primary glulam framework elements; (d) glulam floor slab elements; (e) interior column supporting girders in and transverse to the planes of framework segments; (f) special framework of sixth storey on the curved and sculpted facade (courtesy of FPInnovations)

8.2.2

Glulam framework and diaphragms

From the structural performance design perspective, general points to note are that Quebec City is a location for which design roof snow loads are high; design wind pressures are moderate without threat of hurricanes and slight possibility of tornados; and the peak ground acceleration used in seismic design is moderate. Relevant to wind and snow loading is the building’s location which is the last in a row of abutting buildings of about equal height, in an urban setting at the bottom of a valley. Thus, the building facades are not highly exposed to wind. The roof has

113

8.2 SUPERSTRUCTURE SYSTEM

RC slab above below ground parking garage

First storey RC walls cast

First level of glulam framework installed

First elevated slab structurally complete and second level RC walls cast

Second level glulam framework installed

Second elevated slab structurally complete and third level RC walls cast

Roofing and facades added to make the building weather tight

Fig. 8.3: Selected stages of the supertstructure construction sequence (courtesy of Nordic Engineered Wood) low parapet walls and minor obstructions (Fig. 8.3, bottom-right diagram), leading to the possibility of snow accumulations, which is accounted for via standard provisions according to the NBC. Structural steel and RC framework buildings of similar general size and shape in Quebec

114


City can have superstructure elements sized for strength based on dominant effects of gravity or lateral loads within the design loading combinations. The overall superstructure shape, local topographical features, and the non-symmetric arrangement of superstructure elements of the discussed building indicate that it sways laterally and twists about its vertical axis under the effects of wind and would do so during seismic events. The framework of glulam columns and girders are the primary parts of the above-ground system for resisting effects of gravity forces associated with the self-weight of building elements, roof snow load, and occupancy floor loads (e.g. office and document storage). As already discussed in Subsection 8.2.1, the framework itself is structurally complex because of the in-plane continuity that exists in the prefabricated framework segments, and articulations/rotations that can occur where framework segments join together or are joined to RC substructures. Connections within and at the boundaries of framework segments are also complex in their design, construction, and behaviour. Elevated floors and the roof incorporate glulam elements that are arranged to make those substructures function as horizontal diaphragms capable of collecting the effects of external wind pressures or seismic ground accelerations (i.e. lateral load effects) and transferring them to RC shear walls.

8.2.3

Timber connection methods

As is discussed in Chapters 2 and 5, selection, design, and construction of connections are highly important aspects of the structural design of frameworks made from large glulam/timber elements. Framework and other connections were designed taking into account the implications for the stiffness and stability of the completed superstructure system. This led to adoption of framework connections made using steel dowels that fit tightly in pre-bored holes in glulam, steel plate linking elements with minimally oversized holes, and girder seating elements that tie glulam elements to adjoining superstructure components. Secondary connections were made using threaded steel bolts, self-drilling screws, and annularly threaded nails. All critical structural connections are embedded within structural members for reasons of aesthetics and fire protection (Section 8.4), apart from structural efficiency requirements. Care was taken to avoid eccentric loading of columns under governing design load conditions. This was achieved by arranging steel linking components such that they were symmetric relative to column axes. Figure 8.4 shows typical details of connections within the superstructure. The connections were designed such that glulam framework elements that are continuous through connections lock together (in the style of carpentry lap joints), and when glulam elements are not continuous through connections, their ends join precisely with steel plate linking and seating elements. Dowel fasteners keep elements in place, besides having force transfer functions. Where possible, transfer of forces is by direct bearing on glulam or steel plate elements rather than fasteners, which minimizes the possibility of splitting members. In all cases, the intimate nature of t he connection between elements means that force transfers are achieved by combination of glulamto-glulam, dowel fastener-to-glulam, dowel-fastener-to-steel linking element, and glulam-tosteel seating element bearing and friction. Close attention was paid to avoiding situations where moisture content-related dilations of glulam members might strain connections. Specifically, care was also taken to avoid placing fasteners that joined glulam to steel plates too far apart in the transverse direction, in which glulam could shrink and lead to splitting (Subsection 8.5.1). The Canadian timber design code [114] and similar international best practice specifications emphasize the need for such considerations.

115

8.2 SUPERSTRUCTURE SYSTEM (a)

Connection Trp. SOLD Connection Trp. SOLD

(b)

(c)

(d)

(e)

(f)

Fig. 8.4: Typical primary connection between glulam girder and column elements: (a) scheme for arrangement of interlocking glulam elements showing steel plate linking and girder seating elements at a location where an edge column is continuous between storeys (courtesy of Nordic Engineered Wood); (b) two-storey column element precut and drilled to receive dowel fasteners; (c) multi-bay girder element precut and drilled to receive dowels (one half of the complete built-up girder); (d) steel plate linking and seating elements ready to receive the preassembled framework segment for the next two storeys; (e) shear keys inserted in slots to create composite diaphragm action between glulam slabs elements in a floor; (f) attachment of glulam slab elements to RC shear wall using steel angle sections, self-tapping wood screws, and concrete anchor bolts (photographs (b) to (f) courtesy of FPInnovations)

116


8.3

Structural Design

8.3.1

General aspects

Applicable design codes when the building was designed were: •

Canadian NBC [17], that details Acceptable Solutions/Practices for engineering design of building superstructures (Chapter 4, of NBC). This includes definition of the load combinations to be considered according to a Load and Resistance Factor Design (LRFD) format, and incorporating the use of the so-called “companion loads approach” to account for the relative likelihood of simultaneous occurrence of different catastrophic events (Table 8.1). The LRFD loading combinations presume linear elastic analysis of the internal force effects that various components of the total load have on individual superstructure elements. The NBC specifies the minimum roof snow, floor occupancy loads, wind velocity, and peak ground acceleration applicable to the design of buildings in specific locations. Static analysis is applicable for determination of internal forces in superstructure elements that result from gravity loads. Dynamic response analysis approaches are the default methods for determining internal forces in superstructure elements that result from wind and seismic events. However, for building superstructures that do not exceed 60 m in height and conform to certain requirements (e.g. simple building shape and regularity of the structural arrangement), equivalent static load analysis may be performed for effects of wind and

A. Ultimate limiting state: load factors Load case

Primary loads

Companion loads

1

1.4 D

0.5S or 0.4W

2

1.25 D (or 0.9 D) + 1.5 L

0.5 L or 0.4W

3

1.25 D (or 0.9 D) + 1.5S

0.5 L or 0.5S

4

1.25 D (or 0.9 D) + 1.4W

0.5 L or 0.25S

5

1.0 D + 1.0 E B. Serviceability limiting state: load factors

Load case

Primary loads

Companion loads

1

1.0 D

2

1.0 D + 1.0 L

0.5S or 0.4W

3

1.0 D + 1.0S

0.5 L or 0.4W

4

1.0 D + 01.0W

0.5L or 0.5S

D = effect of dead load (self-weight of system) and other permanently applied forces. L = effect of live and imposed occupancy loads on floor or roof surfaces. S = effect of snow loads on roof surfaces. W = effect of wind forces on external and internal surfaces. E = effect of earthquake/seismic or other ground motions. Additional unmentioned provisions are applicable in certain cases.

Table 8.1: LRFD load combinations specified in the Canadian NBC [17]

8.3 STRUCTURAL DESIGN

117

seismic loads. In the case of equivalent static seismic load effects, it is permitted to reduce the magnitudes of calculated elastic internal element forces if elements in the Seismic Force Resisting System (SFRS – elements in the designed load paths to resist effects of seismic forces) belong to substructures that exhibit ductile behaviour and/or can redistribute forces to parallel substructures. The NBC provisions listed in Chapter 4 of that document apply to superstructure systems constructed from all types of materials for which the parallel Canadian material design code exists; the Code de construction du Québec applicable at the local level of the city of Quebec is equivalent to the NBC. •

Canadian Timber Design Code [114] gives detailed provisions related to the design of generic structural timber elements. This includes definition of design properties of glulam elements and detailed guidance on design of connections that join structural timber elements together, or join them to structural elements made from other materials. In the present context, provisions of the Canadian Canadia n Timber Design Code were employed to size glulam elements and an d for certain aspects of the connection design. This took into account all potentially applicable ultimate and serviceability limiting states for the load combinations in Table 8.1.

•

Canadian RC Design Code [115] gives detailed provisions related to the design of RC elements using steel reinforcement. In the present context, the provisions of this standard were used to design the foundation substructure and shear walls based on the need to ensure sufficient strength and control deflections. This took into account all potentially applicable ultimate and serviceability limiting states.

•

Canadian Structural Steel Design Code [116] gives detailed provisions related to the design of structural steel elements and connections. In the present context, the provisions of this standard were used to design steel plate linking and girder seating elements, and other steel elements in the superstructure; accounting for applicable ultimate and serviceability limiting states.

In the case of each of the timber, concrete, and steel design codes, it is required that elements be manufactured according to traceable specifications and standards under recognized quality assured schemes. When steel is employed, it is also necessary that any welding be done do ne under a specified quality control regime. Technical design provisions embedded in the Canadian material design codes are similar to provisions in comparable international material-specific design codes conforming to LRFD formats. The building is classified as belonging to Group D major occupancy according to the NBC (i.e. business and personal services occupancies). Design dead loads were 1.5 kPa for the roof and 2.45 kPa for the floors, including 1.0 kPa for effects of partitions. The minimum design roof snow load is 3.5 kPa, plus allowance for accumulation of snow behind the parapet wall and other obstructions. The design live loads level for floors associated with the type of occupancy is 3.6 kPa. Code-specified adjustments were made to loads that account for tributary areas supported by various columns and girders. The design wind pressure was calculated using a reference velocity pressure of 0.53 kPa (1 in 50-year return period), a composite external pressure and gust factor (C pC g) of 1.3, and a topography-related t opography-related exposure factor (C e) of 0.84 for building facades that face oncoming winds [17].

8.3.2

Project specific considerations

Owing to the relative lightweight of the glulam superstructure, element-specific considerations were given to vibration serviceability performance of the superstructure, with

118


emphasis on the behaviour of the glulam framed substructure under wind load. The concern was that as the total mass of the six-storey hybrid superstructure is about 1500 tonne, it was possible that wind-induced accelerations associated with lateral sway of the building could reach levels that disturb its human occupants. This is because the vibration period was expected to be much lower than for similar RC buildings (i.e. total superstructure mass would have been about 2800 tonne for an RC building). However, However, it is well known that timber structures typically embody relatively high modal damping because they often contain significant sources of damping beyond material damping (e.g. frictional damping) [20]. Therefore, close attention was given to prediction of the likely vibration period, which was also an important input for determining the admissibility of the equivalent static seismic load design approach in the NBC. Relative to the seismic design, it was determined that the building met requirements for application of the equivalent seismic load approach. Therefore, in the seismic design it was assumed that the d uctility-related force modification factor ( Rd ) and the overstrengthoverstrength-related related force modification factor ( R Ro) could be assigned values of 1.5 and 1.3 respectively, respectively, yielding an Rd Ro (= q Factor in Europe) factor of 1.95. This means that the estimated internal forces in elements associated with an elastic structural response of the system were divided by 1.95 to deter the earthquake load (E) components within the load combinations listed in Table 8.1. Adopted values of Rd and Ro are those assigned to RC shear walls by the NBC. As a comparison, Rd Ro of 2.25 is permitted for braced timber frameworks with limited ductility, and Rd Ro of 3.0 is permitted for braced timber frameworks framew orks with moderate ductil ity ity.. The practice adopted for seismic design was probably conservative.

8.3.3

Analysis method and design results

Static analysis was performed to determine the effects of truly static (gravity, occupancy, and snow) loads, and equivalent (wind and seismic) loads on forces within superstructure elements, and to estimate system deflections such as total and inter-storey lateral drift. Standard stiffness analysis techniques were used via the SAP2000 finite element software [117], in conjunction with element stiffness information in the material codes [114–116] and property information in the literature. Initial seismic analysis was conducted using an empirically estimated fundamental period of the superstructure ( T a) of 0.5 seconds, which is a primary variable entering the calculation of equivalent static seismic forces at each storey level. That enabled preliminary sizing of superstructure elements. The reasonableness of this estimate was later confirmed by a more refined finite element analysis [118]. Post-construction field measurements determined the fundamental period to be 0.37 seconds (2.72 Hz). The total maximum base shear for the superstructure was estimated to be 2.2 GN, which is not large and reflects the moderate nature of seismic design forces for Quebec City. All finite element structural analysis models accounted for framework element continuities and articulations in vertical and horizontal planes (Section 8.2.1 ). Design calculations determined that lateral accelerations due to wind should not be a serviceability performance problem. Main floor and roof girders are made from a grade of glulam intended for continuous beam applications (i.e. the bending moment capacity is the same for positive and negative bending). Girders are coincident with the planes of prefabricated framework segments and are parallel to

8.4 FIRE DESIGN

119

the minor building plan axis that is continuous over three spans approximately 6.0 m long, and have cross sections of 362 × 527 mm2. Girders spanning transversely to the main ones (parallel to the major building plan axis) are spaced 2 m apart and were designed as simple spans of 9 m, resulting in cross sections of 190 mm × 527 mm. For simplicity of fabrication, construction, and aesthetics, cross sections of columns are constant from the ground to the roof. Interior columns have cross-section dimensions of 362 mm × 480 mm, and perimeter columns have cross-section dimensions of 260 mm × 362 mm. The nominal dimensions of the glulam slab members are 89 mm × 450 mm with a maximum bending span of 2 m. These members span parallel to the major building plan axis and are the primary elements of horizontal diaphragms at each elevated floor and the roof. As shown in sub-diagrams (e) and (f) of Fig. 8.4, those slab elements are interconnected and attached to RC shear walls.

8.4

Fire design

To be classified as an acceptable solution according to Division B of the 2005 edition of the NBC, a six-storey office building without sprinklers and facing directly onto three streets (as the building discussed does) must have non-combustible construction capable of achieving a 1-hour fire resistance rating and have a maximum building area not exceeding 3600 m 2 per storey. Therefore, as the proposed hybrid timber–concrete building in Quebec City was using combustible material, it was necessary to demonstrate adequacy of its fire design by the Alternative Solutions approach method. A primary argument used to justify acceptability of the design as an alternative solution was that the total floor area of the office building is significantly less than the limit for acceptable solutions, that is, it is approximately 6000 m2 rather than 6 × 3600 = 21,600 m2. Also, automated fire detection and sprinkler fire suppression systems were installed to offset the use of combustible material in the form of internally exposed glulam columns, girders, and ceilings. This recognizes that using automated fire detection and sprinkler is a highly effective fire protection measure as discussed in Section 3.8.2 . R&D results were used to demonstrate to the building regulatory officials in Quebec City that installation of a fire detection sprinkler system ensures that the fire safety of the building is at least equivalent to the safety required for an equivalent building constructed from non-combustible materials. Also to note is that the installed sprinkler system exceeds the protection level required by the USA-based National Fire Protection Association. Load-bearing glulam elements were designed to have a fire resistance rating of at least 1 hour based on requirements in Appendix D of Division B of the 2005 edition of the NBC [17]. Additionally, the glulam slab elements in the floors and the roof were designed to achieve 1-hour rating using guidelines developed by the American Forest and Paper Association [119]. Essentially the solution adopted involves redundant fire safety measures in the sense that if the technical measure of sprinklers failed to extinguish a fire, then the passive fire resistance of the structure would attain the required level of fire performance. Structural connections were given adequate fire protection by embedding metal fasteners and steel plate parts within glulam elements or protecting them with sacrificial layers of timber or non-combustible material. Presently, no specific guidance is given in Canadian codes for ensuring the satisfactory fire performance of timber connections. Therefore, specifications in Eurocode 5: Part 1–2 [41] and relevant research information were taken into account while making design decisions.

120


(a) (b)

Fig. 8.5: Measurement of relative settlement of the glulam framework: (a) expected deformed shape of eight storey superstructure due to shrinkage of glulam; (b) attachment of potentiometer device at a column base

8.5

Measurement of the building response

Because the hybrid building was novel, studies were conducted during and after construction to ascertain that the performance was as expected, with respect to aspects of serviceability that could be investigated without disruption of the building function.

8.5.1

Differential movements

In hybrid constructed systems that embody multiple construction materials that can dilate, it is important to consider the possibility of differential movements between building elements that may distort or damage the superstructure. In buildings like the one discussed in this chapter, shrinkage and creep of RC and glulam elements can be of concern. Shrinkage of unreinforced concrete occurs because of the hydration process as cement hardens, with dimensional changes of about 0.03% to 0.04% being typical. However, owing to the constraint of the concrete by steel reinforcing bars, shrinkage of RC element lengths is quite small. Glulam as a hygroscopic material can shrink as it adjusts to be in equilibrium with the building climate, and subsequently can dilate in response to changes in the building climate. Moisture shrinkage/expansion coefficients for glulam elements are approximately 0.003% to 0.006% parallel to the axis, and 0.1% to 0.3% in transverse directions, for each 1% change in moisture content . Usually in timber construction, the main concerns are regarding potential shrinkage and distortion of an element at the cross sections (i.e. transverse to the grain) and lengthwise (i.e. parallel to the axis/grain) shrinkage that will distort the frameworks. The main concern during the design of the featured building was avoidance of excessive differential vertical movement of the glulam framework relative to RC wall substructures caused by post-installation, lengthwise shrinkage of the columns, with the greatest relative movement being at the roof level. It was estimated that the maximum possible roof settlement ( S ) relative to the upper end of the RC walls would be:

8.5 MEASUREMENT OF THE BUILDING RESPONSE

S = H × C × ∆ M = 22.1 × 0.00004 × 8 = 7.1 × 10−3 m = 7.1 mm

121

(8.1)

where H is the height of the superstructure (22.1 m), C is the shrinkage coefficient (estimated to be 0.004% per 1% change in moisture content), and ∆ M is the moisture content change of the columns between installation of the framework system (estimated to be 16%) and the in-service equilibrium moisture content (estimated to be 8%). Logic of this approximate calculation is that RC walls would shrink by negligible amounts; creep deformations of the superstructure would be negligible; all framework columns would shrink by the same amount; shake-down of the structural system (i.e. seating of structural elements into their permanent positions) would occur during construction because of self-weight of the building elements and floor and roof loading associated with construction processes. Figure 8.5a illustrates the expected deformed shape of the six-storey hybrid building due to timber shrinkage. Transverse shrinkage of glulam was not expected to influence the framework settlement because of the manner in which the connections were made. The likelihood of an about 7 mm relative settlement of the timber substructure at the roof was considered negligible for practical purposes of design and construction of the superstructure. Such a value would be far less than the typically assumed 25 mm relative settlement of column bases of RC or structural steel frameworks because of post-construction foundation movements. Nevertheless, the unusual nature of the hybrid construction method demanded that the estimate of the roof settlement be validated by field observations. Two potentiometer-type displacement measuring devices were installed by a leading organization (R&D institute FPInnovations) in the building after the superstructure was structurally complete in order to continuously measure changes at the level of the roof relative to the foundation level (Fig. 8.5b). A vertical line of columns located at the extremity of the building and close to the curtain wall was chosen for making observations. Measurements of relative roof level settlement indicated that actual movements during construction and after more than 1 year were about half the estimated possible deformation due to glulam column shrinkage. Most of the recorded settlement was thought to be associated with closure of gaps between column segments rather than shrinkage. The conclusion reached, therefore, was that differential movements in similar buildings can be expected to be negligible provided that careful attention is paid to the design and construction of connections and that glulam is installed dry and protected from getting wet during construction.

8.5.2

Vibration response

Ambient Vibration Tests (AVT) were conducted before non-structural components were installed to estimate building response characteristics such as modal frequencies, modal damping ratios, and mode shapes. Table 8.2 summarizes the frequencies and damping ratios of the first three vibration modes of the building, which are all associated with lateral motions that can be excited by wind or that would be excited by an earthquake. The first two are beam type modes, wherein the superstructure sways parallel to major and minor building plan axes respectively. The third is a torsion mode, wherein the superstructure twists around the vertical axis of the building. As the tabulated values show, two groups of researchers obtained similar results for modal frequencies, suggesting that the values are correct. Estimates of modal damping ratios differed between the groups of researchers. However, this is not surprising given that estimates of damping characteristics of built systems are influenced by and are

122


Organization performing tests McGill University Mode

Type of mode

FPInnovations

Frequency (Hz)

Damping ratio (%)

Frequency (Hz)

Damping ratio (%)

1

Sway east-west

2.72

2.1

2.72

1.2

2

Sway north–south

3.91

2.0

3.92

3.4

3

Torsion

6.92

2.3

7.06

5.3

Table 8. 2: Modal characteristics of six-storey building prior to installation of nonstructural elements

sensitive to many factors. Material damping of the primary structural materials is about 1% of the viscous damping, and the difference between that and observed modal damping ratios is mostly attributable to frictional contacts between superstructure elements. On the basis of observed values and considering that addition of non-structural components adds to the damping sources in the system, it would be reasonable to assume for design purposes that the system exhibits 2% of viscous damping.

Fig. 8.6: Fundamental mode shape, modal frequency is 2.72 Hz (courtesy of FPInnovations)

8.6

Figure 8.6 shows the shape of the first mode, where sway of the superstructure is in the east-west direction (i.e. parallel to the long plan axis of the building). Notably, the ratio of modal stiffness to modal mass is lowest parallel to the major plan axis, because that direction is parallel to vertical planes in which the substructure connections allow localized articulations (Section 8.2.2).

Additional comments

The described project proves the viability of using glulam elements as part of structural and fire performance systems of modern multi-storey buildings. For conditions applicable in Quebec City and other parts of Canada, the primary challenges are those associated with fire design and design of connections. Connections must maximize system stiffness and strength, and minimize the potential for post-installation settlement of glulam frameworks. Relative to the fire design strategy, the studied building was required to have a 1-hour fire resistance rating, because of its classification being for business and personal services occupancies and the three facades facing directly the streets into which occupants could escape in the event of a fire.


123

In Canada and elsewhere, lower fire resistance rating requirements are applicable to office buildings than to residential or other occupancies where building occupants sleep. Therefore, meeting Alternative Solutions fire performance requirements for residential buildings may be more challenging.

Acknowledgements This chapter is based on information, diagrams, and photographs supplied by Mr. Sylvain Gagnon and Dr. Mohammad A. H. Mohammad of FPInnovations, Canada, Mr. Stéphane Rivest of Bureau d’études spécialisées Inc., Canada, and Nordic Engineered Wood, Canada.

125 Chapter

9

Example Project 2: Fire Design of a Seven-Storey Hybrid Building in Berlin, Germany

Summary: The seven-storey superstructure discussed herein is for residential occupancy and utilizes massive timber and reinforced concrete (RC) wall segments, composite timber-andconcrete floor slabs, and steel framing members. Structural and acoustical performance design requirements were satisfied relatively easily, and the main technical challenges centred on the definition of a fire design concept, and selection of construction details and technical measures that result in an effective fire resistance rating of 90 minutes. The building demonstrates the feasibility of using timber as a modern, high-performance construction material based on hybrid construction methods, and employing a fire design that utilizes a combination of passive fire resistance of building elements and automated fire detection and suppression systems.

9.1

Background

Generally, the taller the buildings, the more stringent are the fire protection requirements. Until 2004, the above-ground height of top floors of timber building superstructures in Germany was prescriptively limited to a maximum of 7 m, which translated to a maximum of three storeys. This traditional limitation was associated with achieving buildings with 60-minute passive fire resistance and fires that could be fought externally with scaling ladders. Since 2004, the top storey height limit for timber buildings relying on passive fire resistance to achieve satisfactory fire performance has been increased to 13 m (i.e. maximum of five storeys; [120]). As illustrated in Fig. 9.1, buildings with upper floors more than 13 m above ground level must be designed based on the assumption that fires will be fought via internal access to upper floors. Also buildings with more than four storeys must have 90-minute fire resistance, which can be achieved prescriptively with non-combustible (class A) construction materials. Buildings with upper floor levels more than 13 m above ground must have 90-minute passive fire resistance and employ technical measures for fire detection and suppression [121].

9.2

Description of the building superstructure

The building considered has a 19.4 m high, seven-storey superstructure, and a one-storey ancillary superstructure constructed on top of a RC slab-on-grade foundation. Figure 9.2 shows the plan layout of primary vertical structural components and firewalls, and Fig. 9.3 shows the overall arrangement of structural and fire elements. Each level has elevator and fire escape staircase

126

CHAPTER 9. FIRE DESIGN OF A SEVEN-STOREY HYBRID BUILDING IN BERLIN Building classes 1

2

3

4

5

Middle height building

Other building

7 m > floor* ≤ 13 m

Floor* > 13 m

Buildings Low height floor* ≤ 7 m

Stand-alone

≤ 2 units

1 unit

≥ 3 units

Firefighter effort with scaling ladder

m 3 1

m 7

* Highest floor relevant

Fig. 9.1: Building fire classes [120] RC firewall REIM90-A

Auxiallary building

Main building

RC wall REI 90-A

RC shafts

Glass facade m 5 5 . 2 1

Solid-wood wall REI90 K60

RC columns REI 90-A

13.90 m

RC stairwall

Elevator RC firewall REIM90-A

Fig. 9.2: Plan arrangement showing the vertical structural components and firewalls

access that are physically separated from the main floor area by a distance of 3.1 m, as part of the architectural and fire design strategies. Although the auxiliary superstructure is physically adjacent to the main superstructure, the two are separated by a wall that meets th e 90-minute fire resistance requirements of DIN EN 13501-2 [121].

127

9.3 FIRE COMPARTMENTALIZATION OF THE BUILDING

RC fire wall REIM90-A RC Fire wall REIM90-A

Massive timber wall panel

Composite timber and concrete floor slab

RC shear wall

RC foundation slab and RC first storey

Fig. 9.3: Arrangement of primary structural and fire elements

The first storey is constructed from RC for reasons that include building security against accidental mechanical damage and vandalism, and separation of timber elements from the ground for durability. The overall structural concept combines massive timber and RC wall segments, composite timber-and-concrete floor slabs and steel framing members, with massive timber and concrete elements that also serve passive fire resistance functions. As shown in Figs. 9.2 and 9.3, the vertical structural elements were simple, relatively few, and designed to resist effects of gravitational forces associated with self-weight of the construction, occupancy floor loads, and roof snow load, as the main controls on dimensions of the columns. Composite timberand-concrete elevated floor slabs were designed to satisfy both structural and acoustical mass performance requirements, as well as achieving fire performance objectives. Those floor slabs are relatively heavy, compared with many traditional timber floor construction methods, but presented no special difficulties from a structural design perspective. Because peak seismic design ground accelerations and wind loads for Berlin are low, there were also no special problems associated with achieving satisfactory in-plane design resistances of shear walls and floor slabs. The most challenging design decisions were those associated with providing 90-minute fire resistance, which is why the remainder of this chapter is focused on those aspects.

9.3

Fire compartmentalization of the building

Attaining fire performance necessary for a class 5 building (i.e. highest floor level > 13 m), a compartmentalization strategy was implemented in conjunction with technical fire measures at each level that meet design requirements of class 4 buildings [122]. The footprint of elevated floors in the main area of the primary superstructure is 174 m² including public spaces, with apartment footprints being 148 m². According to the German building

128

0 6 K 0 6 I E R

CHAPTER 9. FIRE DESIGN OF A SEVEN-STOREY HYBRID BUILDING IN BERLIN

1600 m2

148 m2

1600 m2

148 m2

400 m2

1600 m2

148 m2

400 m2

1600 m2

148 m2

400 m2

1600 m2

148 m2

400 m2 REI 60 K60 400 m2

Class 4 building

A 0 9 I E R

1600 m2

1600 m2

0 6 K 0 9 I E R

Class 5 building

148 m2 REI90 K60 148 m2

Actual building

Fig. 9.4: Unlikely case of all floors burning to limits for prescriptive building classes

code, building class 4 can have apartment areas up to 400 m2. This translates to a total allowable fire load of: Q = q f,k × 400 m² = 1085 MJ/m² × 400 m² = 434 × 10 3 MJ

(9.1)

where q f,k is the characteristic fire load density of residential buildings [123]. Employing the actual apartment dimensions, the design fire load is 161 × 103 MJ. However, there is also the danger posed by the building construction itself fuelling a fire. Therefore, a smoke detection system was installed to reduce the level of risk, by guaranteeing early alerting of the fire brigade and mitigating the likelihood that the structural system of the building would burn (Chapter 3). The German code [120] permits a maximum fire compartment area of 1600 m2, with enclosing fire walls required to prevent fire spread for at least 90 minutes. This is achievable by a REIM90-A wall [121] with Euroclass A covering material. Fire walls must have the ability to bear additional loads caused by falling debris during fires as protection for firefighters. Figure 9.4 compares the potentially affected fire compartment areas for the actual sevenstorey building in the case of all apartments burning at once to compartment areas permitted for building classes 4 and 5. As shown, the very unlikely scenario of fires on all floors results in a total affected floor area of about 1000 m² (7 × 148 = 1036 m²). This contrasts with and lies between the respective limits for class 4 and 5 buildings of 60-minute fire containment in an area of 400 m² and 90-minute fire containment in an area of 1600 m². This effectively means that the total area of the worst envisaged case for the actual building is far less critical than the limit for any one compartment of a class 5 building. Taking into account all implemented measures, the fire compartmentalization strategy for the building is highly conservative.

129

9.4 DETAILED ASPECTS OF THE DESIGN

Design element affected

Building class 4

5

Implemented measures

A only (non- • 60 minutes for surface ignition protection of combustible walls, performance • ceiling protection by a transparent fire retarrequired) dant coating of class B, • composite timber–concrete floors, • encasing structural steel framework in incombustible material or sacrificial timber layers, • elimination of cavities in massive timber elements, • separated RC stairwell (safety stairwell), • automated fire detection and extinction systems, • separate water supply for fire equipment (dry pipe system)

Euroclass

B, C, D, E (timber allowed: B–D without technical fire protection measures)

Maximum apartment ² size (m )

400

1600

148

Maximum within storey escape routes

35 m

35 m

< 20 m

Table 9.1: Summary of building classes 4 and 5 fire requirements and implemented design measures

Local regulations (applicable in Berlin) require that the maximum distance between any location in a fire compartment and the wall of the escape staircase be 35 m. The actual maximum distance is about 20 m. Therefore, the time it should take for building occupants to reach a protected escape route is roughly half of what is allowed. The external location of the escape staircases further reduced the fire risk and minimizes the potential for hot flue gas contaminating the escape route. Other special measures were taken to ensure adequate fire performance of the building as summarized in Table 9.1. The table also compares those measures with requirements for building classes 4 and 5.

9.4

Detailed aspects of the design

9.4.1

Floor slabs

Design of the elevated composite timber-and-concrete floor slabs traded off benefits required for a high acoustical mass with structural benefits from minimization of self-mass of those slabs. The derived solution is a construction that has a 160 mm thick vertically laminated timber layer topped with 100 mm RC layer, with composite action between those layers achieved via slots in the timber that developed a mechanical key at the interface. Figure 9.5 shows the

130

CHAPTER 9. FIRE DESIGN OF A SEVEN-STOREY HYBRID BUILDING IN BERLIN (1) (2) (3) (4)

(5)

Fig. 9.5: Composite timber-and-concrete floor slabs; layers: (1) flooring, (2) 80 mm particleboard subflooring, (3) 70 mm insulation, (4) 100 mm concrete, (5) 160 mm laminated timber

Fig. 9.6: Installation of prefabricated laminated timber floor slab panels

completed slab construction that also includes non-structural layers of insulation, subflooring, and flooring. The timber layer was installed as prefabricated panels that subsequently acted as permanent formwork for the RC layer ( Fig. 9.6 ). Fire design calculations allowed for reduction of the depth of the timber layer, which would be exposed during an apartment fire below. The RC layer’s full thickness was used in the fire design calculations associated with fires above or below the slab, as is normal for fully RC elements. Table 9.2 summarizes the design parameters used for structural assessment of the complete floor assembly for normal (no fire) and after 90 minutes of fire exposure situations. The partial coefficients employed to assess the residual design load and deflection resistance capabilities of floor slabs after a fire are less stringent than the partial coefficients applicable

131

9.4 DETAILED ASPECTS OF THE DESIGN

Parameter

Deflection limit Maximum deflection rate Factored load combinations (based on the ultimate effects of loads partial coefficients design equation applicable in the European Union*)

Design situation Normal (no fire exposure) [124] l w ____ 300

Not applicable

After 90 minutes fire exposure [125,126] l2 w _____ 400d dw = ______ l2 ___ dt 9000h

g G = 1.35 g Q,1 = 1.5 y 0 = 0.7 g Q,2 = 1.5

g G = 1.0 g Q,1 = 1.0 y 0 = 0.2 g Q,2 = 1.0

k mod = 0.9 g m = 1.3

k mod = 1.15 g = 1.0 M, f 1

d = effective depth of the cross section (depth of the cracked section from the center of reinforcement to the top compressed layer) h = overall depth of the composite slab l = span t = time w = deflection of the horizontal member g G = partial coefficient for gravity load g Q,i = partial coefficient for live load y 0 = load combination coefficient k mod = modification factor accounting for duration of the loading and service (moisture) class = partial coefficient for material resistance g m, g M,f 1 *Applicable partial coefficients design equation:

Load effects

Table 9.2: Design parameters for floors

to normal situations. Nominally this implies allowance of larger deflections and stresses in buildings that have experienced fires. However, it should be borne in mind that buildings that have been damaged by fire are unlikely to be in normal use prior to their full repair. So in practice, safety levels for fire damaged buildings are not compromised. Undersides of floors are coated with a fire retardant and the building is fitted with fire detection and sprinkler systems. The fire retardant coating is transparent with a thickness of 150 µm and meets the requirements of the Euroclass B (hardly ignitable; [121]).

9.4.2

Critical element junctions

While the fire design of composite timber-and-concrete slabs was quite straightforward, designing connections and inter-element junctions represented significant design challenges. Applying two layers of 18 mm gypsum plasterboard to combustible elements results in approximately 60-minute passive fire resistance that is additive to any inherent resistance that protected elements have. Two layers of gypsum plasterboard alone are not capable of meeting building class 5 requirements. Shielded components such as connections and inter-element junctions must have 30-minute passive

132


8 , 7 0 3 / 6 1 r e d r i G

4 l o o w l a r e n i M

Reinforced concrete

0 1

6 1

6 l o o w l a r e n i M

Fig. 9.7: Inter-element junction of a floor slab and exterior massive timber wall (dimensions in cm)

fire resistance of their own that would be activated if the gypsum plasterboard layers are destroyed. In the case of the building discussed, an important critical inter-element junction situation occurs at locations where floor slabs met exterior massive timber wall panels (Fig. 9.7 ). In the design of those junctions, it was assumed that undersides of floor slabs will passively resist a full 90-minute fire. However, the massive timber walls are protected against interior fires by two layers of plasterboard. This means that they only require 30-minute passive resistance (i.e. total resistance = 60 + 30 = 90 minutes). Exterior surfaces of massive timber walls are protected by 12.5-mm gypsum plasterboard and 100 mm of mineral wool beneath an 8-mm surface layer of mineral stucco. Such external protection combats the possibility of inter-storey spread of fire via combustible building facades (Section 3.8.3).

9.4.3

Gravity load system

Columns were designed to have 60-minute passive fire resistance achieved by encasing steel cross sections in sacrificial timber. This was deemed as a sufficient measure, because the floor slabs and their junctions with exterior walls are designed to have 90-minute passive fire resistance; fire compartment areas are of limited dimension; escape routes are short; technical fire detection and suppression measures supplement the passive fire resistance; and timber burns very predictably. Regulations in Germany have a so-called sufficiency of fire safety requirement that permits designers latitude to make such judgments. Specifically, the decisions hinged on an assessment of how much time would elapse between the start of a fire and arrival of first fire engine. German codes do not specify expectations concerning what is termed “wirksame Löscharbeiten” (time before effective fire suppression begins; [120,127]). The

133

9.5 ADDITIONAL COMMENTS Electrical

Mechanical

Fire protection layer (2×18 mm gypsum board)

Fig. 9.8: Electrical and plumbing installations on a massive timber wall

design expectation was that fire engines in Berlin will reach any building within 15 minutes of being alerted. On that basis, selection of the 60-minute fire resistance is believed eminently reasonable. Additionally it is not to be forgotten that a number of special fire measures were employed (Table 9.1).

9.4.4

Cavity fires and transmission of hot gases

Prevention of the possibility of undetected spread on fire through concealed cavities is always a critical fire protection measure. Such possibility is mitigated in the seven-storey building by using massive timber elements, instead of hollow building elements. Building services within individual apartments are either encased within the incombustible RC layers of floor slabs or placed on the surfaces of (i.e. not embedded within) massive timber wall elements. As illustrated in Fig. 9.8, services on the surfaces of timber walls are covered by a layer of gypsum plasterboard that provides nominal fire protection (30 minutes), which is additional to the protection of the wall elements themselves. The possibility of between-storeys spread of fire in the complete building is guarded against by ensuring that electrical, plumbing, and other services for apartments only connect to vertical RC shafts that run the complete height of the building (Fig. 9.1). Those RC shafts are part of the overall structural system of the building (for resisting gravity forces), as well as being selfcapable of attaining a 90-minute passive fire resistance. Laminated timber slabs and panels are typically only partially effective as containment barriers for hot gases produced during building fires. Therefore, the composite floor slabs rely on the concrete layer rather than the timber, as a barrier to transmission of gases from an apartment(s) on fire.

9.5

Additional comments

The described project proves the viability of using massive timber elements as part of structural and fire performance systems of multi-storey buildings. The challenges associated with fire design of buildings in Berlin are fairly typical of those for other urban locations. Therefore, what was done can be a blueprint for solutions applicable elsewhere. Necessary solutions depended on establishing a comprehensive fire design strategy that married the technical capabilities of various construction materials and the adoption of technical fire detection and extinction

134


measures. Resolution of conflicting considerations, associated with structural, acoustical, and fire performance requirements was achieved without major impact on efficiency of the design or construction stages of the work. Such positive outcomes can be expected for many other projects.

Acknowledgements This chapter is based on input provided by Professor Dr. Dirk Kruse and Professor Dr. Ing. Bohumil Kasal of the Fraunhofer Wilhelm Klauditz Institute, Braunschweig, Germany. Kaden, Klingbeil Architects, Berlin provided the diagrams and the image in Fig. 9.6 ; and Dehne, Kruse Brandschutzingenieure, Gifhorn gave permission to describe the fire safety concepts.

135 Chapter

10

Example Project 3: Limnologen— Block of Four Eight-Storey Residential Buildings in Växjö, Sweden

Summary: Since 1994 performance-based design codes have been used in Sweden that allow use of timber to construct residential buildings having any number of storeys, provided general safety and serviceability requirements are met. At first this resulted in five- and six-storey residential buildings being constructed using timber, but in 2007 four eight-storey buildings having mostly cross laminated timber (CLT) load-bearing walls were constructed in Växjö. Floors of those buildings are of CLT plates with glued laminated timber (glulam)-stiffening ribs having spans up to 9.7 m. Apart from structural stability, the main performance concerns during design and construction were ensuring satisfactory sound and vibration serviceability. The discussion in this chapter addresses architectural considerations, wall and floor design, construction methods, and R&D studies based on the Limnologen buildings.

10.1 Background Four eight-storey residential buildings in Limnologen are the tallest, modern, completely timber building superstructures in Sweden. They are located on a narrow site on the western shore of Lake Trummen, between the city centre of Växjö and the campus of Linnaeus University, on a parcel of land known as the Wälle Broar. The site is a showcase for the use of timber as a high-performance construction material, as an initiative of the county of Småland and the city of Växjö. The local action is consistent with the national desire to increase timber usage in the building sector that is embodied by the Mer trä i byggandet (greater use of wood in the building sector ) programme launched by the Swedish parliament in 2004. The basic idea behind that national strategy is that the forestry-based sector, as the most important industrial sector in Sweden, should be given opportunities to create jobs, and increase net export income. Another consideration was redressing the effects of a ban on that industry, which had prohibited the use of timber to construct residential buildings with more than two storeys for 122 years. Technical enablement of construction of taller timber buildings flows from changes to t he Swedish building code in 1994 that made it possible to construct residential buildings having any number of storeys and using combustible materials, provided safety and serviceability performance requirements are met. The remainder of this chapter presents details about design, construction, and performance of the Limnologen project, that has the general characteristics summarized in Table 10.1.

136

CHAPTER 10. LIMNOLOGEN —BLOCK OF FOUR EIGHT-STOREY

Number of apartments

134 apartments: 6 with 1 room + kitchen, 40 with 2 rooms + kitchen, 44 with 3 rooms + kitchen, 28 with 4 rooms + kitchen, and 16 two-storey apartments with 3 to 5 rooms + kitchen

Apartment sizes

37–114 m2

Total floor area

10.700 m2

Type of ownership

Tenant-ownership

Production cost

320 MSek, including two-level timber parking structure for 140 vehicles, community buildings, and storage facilities

Builder

Midroc Property Development with Midroc Projects AB as main contractor

Architects

Ola Malm through Arkitektbolaget Kronoberg

Structural engineers

Martinsons Byggsystem (timber) and Tyréns (concrete)

Ventilation and sprinklers Martinsons Byggsystem Site management

NCC Construction until 13 September 2007, thereafter JSB.

Total building time

Approximately 17 months per stage (two stages)

Time to erect one storey

10 days

Table 10.1: General information for Limnologen project

The buildings are owned by the tenants, who form a tenant-ownership community responsible for the management and maintenance of the buildings.

10.2 Architectural design The builder acquired a land on the shore of Trummen in August 2005 and announced an architectural competition for designing a complex of buildings containing residential apartments of various sizes. Major design considerations were that all apartments should have a good view of the lake and that the narrow building site should be used efficiently. The competition was won by a firm that had significant experience in timber constructions. Instead of five buildings each having five storeys as initially envisioned, the architects suggested four buildings; each with seven full storeys and an eighth storey covering parts of the building footprint. The buildings are laid out as seen in Fig. 10.1a, with ancillary structures clustered around the four main apartment buildings. The shape of the main building plan enables all 134 apartments to have direct view of the lake. The ancillary structures include a small storage building, a two-level parking garage, and a communal building on the lake shore. The parking garage is constructed from glulam and CLT panels. A typical floor plan consisting of five apartments of different sizes, two staircases, and elevator shafts is shown in Fig. 10.1b. The three apartments closest to the lake have balconies on two facades, whereas the two furthest from the lake have only one balcony each. The shape of the building plan privacy for the building occupants. Another important aspect of the design is that in making the balconies on the southern side of the buildings contiguous, they could function

137

10.3 STRUCTURAL DESIGN (a)

(b) B305 B304

HISS B306

B307

HISS

B308

Hus B Plan 3

Fig. 10.1: Architectural layout: (a) site plan showing four residential and ancillary structures; (b) typical floor layout with five apartments

as an emergency escape route in the event of a building fire. The balconies could function as projections that prevent or mitigate vertical spread of fire on facades. Those measures in combination with the installation of automated fire detection and internal sprinkler systems enabled the use of timber boards as cladding on the facades. Portions of the facades that could not be reached from balconies for maintenance and repair are rendered.

10.3 Structural design As the ground conditions on the selected site were very poor, piles were used in the foundations of the eight-storey buildings. The first storeys were designed to be of reinforced concrete (RC), primarily because of the need for a counterweight to anchor and stabilize the shear walls against the uplift forces associated with wind. The majority of the load-bearing structure consists of CLT panels with three or five layers/plies that form wall and floor elements. Internal walls with separating functions are composed of conventional timber light-frame assemblies. Because of the special requirements associated with eight-storey buildings, the number of suppliers of suitable timber-based load-bearing system elements was limited. The supplier of the timber parts also took responsibility for their structural design. Foundation and RC substructure designs were carried out by the consulting firm Tyréns.

10.3.1 Wall elements Typically, each storey of any of the primary building has about 60 prefabricated timber wall elements. Three types of these elements are (1) exterior, three-layer CLT walls, (2) interior three- or five-layer CLT walls, or (3) interior light-frame walls ( Fig. 10.2). Most of the dead and live load from the flooring is transmitted through the exterior wall elements ( Fig. 10.2a) requiring them to perform structural functions associated with gravity and wind design loads.

138


(a)

(b)

(c)

Fig. 10.2: Example of (a) exterior CLT wall; (b) light-frame separating wall; and (c) interior CLT wall (covering layers are insulation and/or gypsum plasterboard)

Internal walls (Fig. 10.2b and c) also perform structural functions related to stabilization of the superstructures. External walls mostly have exterior insulation and timber cladding or rendering on the exterior (Section 10.2). Room linings are made of two layers of gypsum plasterboard in order to address fire performance requirements ( Section 3.6 ).

10.3.2 Floor elements There are about 30 prefabricated timber elements in every floor of each primary building, with those elements having various geometries but the same general construction features (Fig. 10.3). All storeys of a building have the same plan, except for the top storey. Floor elements have two load-bearing system layers, the upper one for supporting floor loads and the lower one (which is lighter) for resisting the deadweight of ceiling insulation and gypsum plasterboard and any attached non-structural fixtures. This approach was taken for acoustical performance purposes (Section 10.5). The load-bearing parts of the floors consist of a 73 mm (three layers) thick CLT slab strengthened by the addition of T-shaped glulam Fig. 10.3: Flooring assembly addressing structural, ribs that are glued to and fully acoustical, and fire performance requirements composite with the slab. The glulam ribs are spaced at 600 mm (centre to centre). When delivered to the site, insulation of floor elements was partially installed. The remaining insulation and gypsum plasterboard were installed on-site.

10.4 FIRE DESIGN

139

The largest floor elements are 11.7 m long and 3.7 m wide and continuous over three supports. With minor enhancements of the basic floor construction some elements have spans of 7.9 m and 9.7 m. Floor elements were joined together to create diaphragms that channel the force flows caused by wind pressures on the facades to the shear walls.

10.3.3 Lateral load design In Sweden, the most important lateral load case for tall buildings is wind loads. Unlike in certain other parts of the world, there is no requirement for residential buildings to be designed specifically to resist seismic force (i.e. Sweden’s populated areas have very low seismicity levels). As already indicated, wind-induced force flows are channelled from the facade surfaces to the floor and roof diaphragms, then to the shear walls, and finally through the foundation to the ground. The CLT walls and separator light-frame walls were all designed to carry part of the flows of the wind-induced forces through the system. Separator light-frame walls must resist design forces sufficiently high that it was necessary for them to be sheathed with high-density fibreboard mechanically fastened to the timber framing. CLT wall panels were designed to resist the effects of shear force flows and resist uplifting forces potentially resulting from overturning of parts of the superstructure or the superstructure in its entirety. Forty-eight hold-down rods are installed in each building to resist any uplift. Those tension rods are anchored to the RC first storey and extend to the top floor and are located inside the walls. This technique is intended to resist maximum uplift forces of 100 kN. Re-tightening of the rods periodically is a building maintenance requirement and intended to counteract any loosening that occurs because of factors such as timber shrinkage or creep.

10.4 Fire design Regulatory requirements in Sweden concerning fire safety are not dependent on the material used in the load-bearing structure. As the primary buildings in this example have more than two storeys they belong to European class BR1 (designated class BBR in the Swedish code), which is the class with the highest requirements. Each apartment was designed as a separate fire compartment and, therefore, had to achieve a 60-minute ( EI60) fire resistance rating. Only first floor rooms used as storage for items such as baby carriages were designed to achieve a lower (30 minutes – EI30) fire resistance rating. As discussed in Chapter 3, the EI rating signifies the number of minutes that room linings are required to retain their integrity and insulation capabilities. Timber vertical and lateral load-resisting elements are designed to achieve a 90-minute fire resisting rating, which is achieved by encapsulating them in two layers of g ypsum plasterboard. Besides the passive fire protection in the form of gypsum, the primary buildings are equipped with automated fire detection and sprinkler systems. According to Swedish building regulations, these additional technical measures are not required. However, adoption of them facilitated acceptability of the overall design concept to regulatory officials, building owners, and building occupants. As discussed in Section 3.8, the use of sprinklers minimizes the chance that fire will spread through windows and burn facia materials and was necessary because the south facade is clad entirely with wood. Other measures taken include that any spread of fire on the underside of balconies should be easily visible, and therefore is likely to be detected early. Each apartment has two exits and two fire escape routes, with one being a fire protected route inside a building and the other external to the building via the balconies.

140


10.5 Acoustical design Acoustical design of the primary buildings received special attention because problems are known to have occurred in cases where such attention was not given. Requirements for the Limnologen buildings included European class B level protection (designated class BBR in the Swedish code). Noise levels for that class are considerably lower than Swedish regulatory requirements for the building type, which is class C . Adoption of class B requirements led to two special actions. Firstly, floors at any level were not constructed to act monolithically. Secondly, flooring elements were seated on specially designed rubber strips known as Sylomer® strips. The first measure reduces direct sound transmission between apartments in the same storey, and the second reduces flanking sound transmissions between apartments. Besides those precautions, wall and flooring elements incorporate sound insulation materials that dampen direct transmissions between apartments and between master bedrooms and bathrooms. Also, polyurethane sealant (Sylomer® and Sylodyn®) was placed where wall and floor elements met to reduce flanking transmission, and apartments have high quality sound-rated doors. As shown in Fig. 10.3, there is an air gap that facilitates acoustical separation of apartments between the structural layer of floor assemblies and the ceiling layer and timber members that support it. After the project was completed, the builder hired a consultant who measured the sound level and concluded that the requirements for acoustical class B had been met.

10.6 Protection of elements and construction of buildings 10.6.1 Moisture and weather protection Owing to the hygroscopic nature of timber and most wood-based composites, such construction materials must be protected from becoming wet from the time they leave manufacturing plants in dry condition until the buildings in which they are installed are weather-tight. Keeping materials dry minimizes the likelihood that construction elements will change dimension or distort before or after installation. Appropriate moisture and weather protection measures are required at manufacturing plants, during transportation to construction sites, and at construction sites themselves. Wall and floor elements to be installed in the buildings in Limnologen were manufactured indoors and stored under cover before being transported to the construction site. The wall elements were wrapped in plastic film, covered by a tarpaulin, and oriented vertically during transportation to the site. Floor elements were covered by tarpaulins and stacked one on top of another in covered trucks (Fig. 10.4a). Unloading at the site was done using a forklift, and elements were temporarily stored under the tarpaulins, until being moved for installation into their final positions. Weather protection during construction consisted of storage beneath a large tent that covered the work area and the elements being moved into position by an overhead crane (Fig. 10.4b and c ).

10.6.2 Construction of buildings Apart from structural elements, the primary building had to have various building service systems installed, such as those for heating and ventilation, electricity, plumbing, and fire protection. In general, those installations run parallel to spans of stiffening ribs of the floor elements, and as far as possible were preinstalled at the manufacturing plant. Some parts of service

10.6 PROTECTION OF ELEMENTS AND CONSTRUCTION OF BUILDINGS

(a)

(b)

141

(c)

Fig. 10.4: Protection of elements: (a) during delivery; (b) during early stage construction; (c) late stage construction showing an elevator segment being lifted into position by an overhead crane

Fig. 10.5: Installation of under-floor heating system

installations running in across-the-ribs direction were installed on-site. Work sequences were organized to minimize the amount of time spent on-site, which minimized the overall construction period, and facilitated improved quality control, which is maximized when manufacturing and construction are done in factory settings. Installation of the und er-floor heating system took a considerable portion of the total construction time (Fig. 10.5). Grooves in floor elements into which tubes were placed were only partially prepared in the factory. Sheet metal to distribute heat laterally was installed on-site, as were the floor finishing layers. Each apartment has a centralized heating system control that enables occupants to regulate the temperature. The overall project was undertaken in two stages. Stage one involved construction of the two northern-most primary buildings, which were finished during spring and early summer 2008. The second stage comprised the two southern primary and ancillary structures and was finished during late spring 2009. The southern facade of the southern-most building is shown in Fig. 10.6a, whereas Fig. 10.6b shows the collection of buildings viewed from the north-west. Heat and water consumption is measured continuously in each apartment and is viewable by occupants/tenants through private web pages. It was expected that the energy consumption would not exceed 90 kWh/m2 per year and that the individual monitoring would result in consumption reductions by up to 30%. Measurements made during the first 2 to 3 years of occupancy show that excluding staircases and other communal areas the average heating energy was only 55 kWh/m2 per year. The combination of energy-efficient construction techniques and a consumption feedback system proved highly effective to limit energy usage.

142


(a)

(b)

Fig. 10.6: Completed buildings: (a) most southerly building; (b) collection of buildings viewed from north-west

10.7 Research studies The Limnologen project has been made available for a number of research studies aimed at independent documentation of aspects of building performance and construction practices. The following questions were identified as being of special interest: •

What technical aspects of the building process can be improved?

•

How building systems compete financially with alternative systems (timber-based and others)?

•

What are the attitudes of building occupants toward timber construction?

•

What are the environmental impacts of choosing timber as a construction material, and does that choice contribute to the creation of a sustainable society?

•

What are the main issues associated with managing and maintaining multi-storey timber buildings?

For practical tractability, answers to these questions have been developed based on study of the following: – Planning documentation associated with suggestions put forward and decisions taken during the early stages of the project. – Quality control documentation of errors made and problems encountered. – Inventories of the technical and environmental performance of the chosen solutions, including measurements taken by instruments installed in the buildings. – Economic and marketing documentation, including customer surveys of perceptions of timber construction. – Building process documentation, such as data from time studies and logistics of site operations. – General information documentation about the project.

The research projects are described in detail elsewhere [128–136]. Here, two examples are briefly presented that relate to the physical performance of the building and the efficiency of the construction practices employed.

143

10.7 RESEARCH STUDIES

10.7.1 Measurements of vertical settlement As with the six-storey building in Canada discussed in Chapter 8, it was of interest to determine how much vertical settlement would occur as the multi-storey timber building superstructures adjusted to moisture equilibrium with interior and exterior climates, and as the foundations and superstructures settled into place. The magnitude of settlement expected to occur because of in-place drying out of timber elements was estimated by simple calculations (Section 8.5.1), but how accurate those might be was unknown. Therefore, one building was equipped with measuring devices that recorded within-storey, vertical settlement movements of the second to seventh storeys over a period of more than 4 years following construction (Fig. 10.7 ). Measuring devices were installed after the building was erected, but before the insulation was completed and before the gypsum plasterboard room linings were added. Therefore, observations reflect movements that might have affected the building envelope and operations which occurred during construction and while the finished building was adjusting to its surroundings and service environment. Observations were made on the facade because that is where relatively large settlement displacement was likely to occur. The relative, within-storey displacements were measured using reference points located directly on an external wall. Actual displacement measurements were taken using a displacement

CLT-panel

m m 5 9 9 2 h t g n e l g 6 n y i r e r u o s a t S e M

Device for measuring temp. and R.H., storey 7

m m 5 9 9 2 h t 5 g , n 4 e , l g 3 , n 2 i r y u s e a r o e t M S

Cantilever used as support

m m 0 7 9 2 h t g n e l g 1 n y i r e r u s t o a S e M

~ 100

Channel in plastic filled with insulation after clompetion of installation of equipment. Bar of invar with potentiometers installed at the end placed on top of cantilevers and supported laterally by concentric bearings. 0 6 ~

Space for cables and devises measuring temperature and relative humidity.

1

2

3

4

5

6

Bar of the steel alloy invar Wood material

Displacement measurement potentiometer Device for measuring temp. and R.H., storey 2 Reinforced concrete slab and wall

Fig. 10.7: Displacement measuring devices on northern facade of the northern-most building

144

CHAPTER 10. LIMNOLOGEN —BLOCK OF FOUR EIGHT-STOREY 30

25 ) m m 20 ( n o i t a m r 15 o f e d l a c i 10 t r e V

2nd 3rd 4th 5th 6th 7th Total

5

0

24-Dec2007

11-Jul2008

27-Jan2009

15-Aug2009

03-Mar2010

19-Sep2010

07-Apr2011

24-Oct2011

Fig. 10.8: Vertical settlements per storey and total settlement of second to seventh storeys (September 2007 to December 2011)

potentiometer (Regal – WPL 50EFZ) placed at one end of a 20-mm diameter rod made of invar, a nickel – iron alloy notable for its unique low coefficient of thermal expansion, which minimizes extraneous thermal influences on observed displacements. As indicated in Fig. 10.7 the invar rods have rounded ends and connect to cantilever brackets via concentric bearings, which further minimized the possibility of extraneous influences on observations. The reference and measurement end brackets were attached to CLT panels using screws. Potentiometers were separately connected to a data acquisition system (Datataker DT85). Synchronized measurements were taken of the external surface temperature and relative humidity at the second and seventh storey ceiling levels using Vaisala model HMP50 devices. Figure 10.8 shows the displacement recorded between September 2007 and December 2011. The maximum observed cumulative settlement for second to seventh storeys was about 24 mm over a distance of 17.95 m and was recorded during the summer of 2011. This corresponded to 0.13% of the total measuring length. The major part of this displacement occurred during the first 9 months that data were collected. During the same period, the estimated average moisture content decreased by roughly 10%, which is normally expected to result in shrinkage of wood walls by about 0.04% (Section 8.5.1). There is no clear sequencing of displacement magnitudes according to the storey level. Results suggest that the main causes of settlement are shrinkage of floor layers and shake-down settlements, as the components bed themselves in and as buildings adjust to the surrounding climate and the influence of in-service building loads. The effect of cyclic seasonal variation of moisture contents of timber components was clearly apparent, and can be seen in Fig. 10.8. Extrapolating the observed trends, it is reasonable to conclude that a maximum settlement of about 25 mm would be reached around 6 years after construction. Although this value would not be replicated exactly in other circumstances, the results do show that with proper attention to delivery, storage, and installation, settlements in buildings of the type discussed herein can be expected to be of magnitudes that will not cause building performance problems, even in high-rainfall climate zones such as southern Sweden.

145

10.7 RESEARCH STUDIES

10.7.2 Time study on installation of load-bearing elements Time taken to install each wall and floor elements of two buildings were measured from when an element was hooked to the gantry crane until installation was complete and it was unhooked from the crane. The information was used in a Virtual Reality Modelling Language (VRML) model, in order to visualize the erection of the building. Figure 10.9 shows four stages of construction ranging from construction of a first storey concrete structure on 11 September 2007 to the completion of the full load-bearing structure. Analysis of the data determined that the average installation time for wall elements was about 22 minutes and that the average installation time for floor elements was about 26 minutes. This suggests that under normal circumstances the floor elements are slightly more time-consuming to install than the wall elements. However, this reflected some difficulties in fitting the flooring elements together that could be overcome by design and construction process modifications. Figure 10.10 provides a more detailed analysis of the average installation time as a function of which floor was involved. The highest and the lowest average duration per storey for installing a wall element were approximately 29 minutes and 18 minutes, with both being for building 1 (called house 1 in the diagram). The corresponding durations for the flooring elements were 30 minutes and 17 minutes for buildings 1 and 3, respectively. Such differences suggest that the installation process is very sensitive to disturbances and that installation times for both floor and wall elements could be shortened appreciably. Linear regression analysis showed that there was a tendency in both buildings studied for installation of wall elements to take a shorter time higher up in the building, despite the fact that the

11/9 -07

22/10 -07

22/11 -07

Complete house

Fig. 10.9: VRML model of construction stages

146


(a)

(b)

] s 35:00 e t u n i m30:00 [ e m i t 25:00 n o i t a l 20:00 l a t s n i e 15:00 g a r e v A10:00

] 35:00 s e t u n i m30:00 [ e m i t 25:00 n o i t a l l 20:00 a t s n i e 15:00 g a r e v A10:00

Walls house 1 Walls house 3 Walls house 1 linear regression Walls house 3 linear regression

1

2

3

4

5

6

7

Storey number

8

Flooring house 1 Flooring house 3 Flooring house 1 linear regression Flooring house 3 linear regression

2

3

4

5

6

7

8

Flooring number

Fig. 10.10: Average installation times in houses 1 and 3: (a) wall elements; (b) floor elements

time required for transportation of each element from the ground level to the storey on which it was to be installed increased with height above the ground. Potential explanations are that the time required was affected by: (i) adjustments made at the manufacturing plant or at the construction site that improved the efficiency with which elements could be fitted together onsite as the project progressed; and (ii) workers became increasingly familiar with the building system and, thereby, became more efficient in execution of their tasks as the project progressed. There was a likewise decrease in installation times for floor elements in building 3. However, for building 1 the linear regression analysis indicated a marginal increase in the time required for installation of floor elements as the working above-ground increased. The likely correct overall interpretation of what was observed is that experiential learning associated with manufacture of elements and site practices yields productivity gains, but when practices are stable, productivity decreases slightly with each change in the lifting distance from the ground to the work level. By implication, the project undoubtedly did not achieve optimal efficiency in terms of practicebased learning or storage and handling of elements on-site, but given the uniqueness of the project this is not surprising.

10.8 Additional comments The buildings in Limnologen, as the first modern eight-storey residential timber building in Scandinavia, represent new architectural and material use options. The new paradigm for timber construction represents results from the confluence of the transition of the Swedish code to a building performance basis, the government led Mer trä i byggandet initiative, and availability of CLT as a high-performance material for wall and floor elements. The integration of industrial capability with R&D activities by public and private sector institutions was also crucial for successful completion of the projects and creation of buildings that perform as intended. This can be a blueprint for creation of other timber building systems and technologies in Sweden and elsewhere.

Acknowledgements This chapter is based on information, diagrams, and photographs supplied by Professor Erik Serrano and Dr. Johan Vessby of Linnaeus University in Växjö, Sweden.

147 Chapter

11

Example Project 4: Björkbacken, a 10-storey hybrid building in Stockholm, Sweden

Summary: The building discussed herein is for residential occupancy by the elderly and has four reinforced concrete (RC) plinth superstructure storeys and additional six modular timber framed storeys above. Timber-framed modules are of a type that has been previously proven satisfactory for construction of an “all” timber six-storey superstructure. The primary reason for constructing lower storeys from RC was that such an approach was acceptable to building regulatory officials in Stockholm with respect to fire performance expectation of the complete building. Such a requirement reflected that no similar buildings had been previously constructed in Sweden. The other primary technical issues associated with design of the building were related to control of expected vibrations and static deflection responses of timber floors for serviceability. Force flows resulting from lateral external wind loads on upper storeys are channelled from the building facades to the timber floors (which act as diaphragms), then to timber shear walls, and then into the RC storeys, which are monolithic with a RC foundation. Timber modules are anchored to the storey below against uplift, overturning, and sliding due to wind by hold-down and shear connectors at their bases.

11.1 Background Until 1994, the maximum height of timber-framed buildings in Sweden was prescriptively limited to two superstructure storeys, because of concerns about the potential risk of fire spreading between buildings in urban areas. This mirrored restrictions that existed in many countries until quite recently (Chapter 3). Since 1994, the Swedish building code permits buildings of any height to be constructed provided that it can be demonstrated that the design meets minimum safety and serviceability performance requirements related to the intended type of building occupancy. In early 2011, Sweden adopted the Eurocode building codes with specific national requirements being contained in annexes to those documents. Design and construction of the 10-storey Björkbacken building discussed here was required to achieve a 90-minute fire resistance rating. Other general technical requirements were that the structural system (superstructure and foundation) be able to resist potential ultimate limiting states effects of the self-weight of the building and design loads associated with a residential building occupancy, wind and snow, and perform satisfactorily with respect to serviceability limiting states associated with the acoustical and vibration performance of the superstructure.

148

CHAPTER 11. BJÖRKBACKEN, A 10-STOREY HYBRID BUILDING IN STOCKHOLM, SWEDEN

11.2 Superstructure concept The featured building is located in Stockholm and has 10 superstructure storeys that have a total height of 30.9 m. The building consists of four plinth storeys constructed of RC and six storeys of timber construction on top ( Fig. 11.1). The four lowest storeys have cast-in-place walls and prefabricated hollow floor decks, with the walls being monolithic with a RC foundation. The top, elevated concrete floor (floor of the fifth storey) was precambered to counteract the deadweight of the superimposed timber construction. The six timber storeys were constructed mostly using prefabricated light-weight timber-framed modules (Fig. 11.2), that are interconnected using steel link elements mechanically fastened to interfacing modules/substructures. The upper storey floor plans replicate one another and consist of four residential apartments for elderly people that can be entered into or exited from fully enclosed elevator and staircase shafts that connect to the apartments by communal corridors, with the corridors and staircase functioning as fire-escape routes. The timber modules arrived at the construction site with windows, doors, services, and internal linings already installed and their interiors painted. The structural concept is that the building system minimizes concentrated force transfers as much as possible; meaning that relatively small

FA

FA

Fa

Fa

FA

FA

Fa FA

Pa

Fa FA

Fa

Fa

FA

FA

Fa

Fa

FA

FA

Fa

Fa

FA

FA

Fa

Fa

FA

Fa FA

Fig. 11.1 Building superstructure

Interface between RC and timber construction

Ground level

11.2 SUPERSTRUCTURE CONCEPT

149 structural timber wall framing elements sheathed with mechanically fastened wood-composite materials could be employed in the upper storeys. Prefabricated timber modules are, therefore, made of light-frame construction (also known as “2-by-4” construction in North America). The approximately rectangular shape and replicating nature of the floor plans of storeys containing apartments (17.9 m × 23.3 m outside dimensions) and stacking of modules facilitated even distribution of vertical and horizontal forces (Fig. 11.3). The fire design concept is that occupants could escape from any apartment either by the central staircase or via scaling ladders attached to fire engines; an approach acceptable to the local building regulatory officials. As shown in

Fig. 11.2 Prefabricated timber-framed modules 23318 4000

7559

6 2 0 9

0 0 0 2

6 2 0 9

3910

4110

7624

11759

0 6 0 3

0 6 0 3

6 2 0 7

0 6 0 3

4904

1 0 9 4

0 0 0 2

11559

0 6 0 3

7 2 9 7 1

4000

6 2 0 9

7424

6 2 0 7

7759

5104

0

7 2 9 7 1

0

11559

0 6 3 3

0 6 3 3

11759

0 6 3 3

1 0 9 4

6 6 7 4

6 6 7 4 6 9 4 6 0 0 0 4

0 0 0 4

2625 1 0 9 8

2000

1 0 9 8

9559

1 0 9 8

9759

2000

Fig. 11.3 Floor plan for upper storeys (black dots are locations of steel linking elements in the lateral load-resisting system)

150


Fig. 11.1, the site on which the building is located has a sloping terrain, such that one facade of the building has only eight above-ground storeys.

11.3 Fire compartmentalization Each apartment and the staircase were designed as separate fire compartments. This means that the separating walls and floors between apartments are designed for a fire resistance of 90 minutes. This is achieved in timber storeys by interfacing modules in a manner that creates double layer walls and floor–ceiling substructure interfaces. Building ventilation shafts are fitted with a fire-valve system that would automatically close the ducts in the event of a fire being detected by smoke sensors connected to that system. That would prevent fire spreads from one compartment to another. The risk of external fire spread via apartment balconies is mitigated by cladding the lower faces of balconies with non-flammable material. Except for the balconies, the building facades are faced with non-combustible material.

11.4 Vertical load resisting system The building was designed to resist the effects of a roof snow load of 3.7 kPa, and a reference floor live load of 3.7 kPa, which was adjusted to account for tributary areas of modules. Static deflection of timber floors was designed with intent that it would not exceed 1/300th of the span. In general, the sizing of the horizontal and vertical walls of light-frame elements to resist effects of vertical loads matched normal practices employed in the design of low-rise light-frame buildings. The floors and ceilings of the modules were attached to walls using coach-screw (also known as lag-screw) fasteners that embed in each wall stud. The design intent is that all vertical loads will flow through the outside walls of a module, and any walls dividing the modules were regarded as non-structural. Future remodelling of the apartments could, therefore, be done without further evaluation of the system capabilities, provided that the building occupancy class did not have to be altered to one having higher floor live loads. Notably, a module can have both external walls and internal separating walls around their perimeters. External walls are constructed from timber framing with cross-section dimensions of 45 × 220 mm 2. They are fully insulated and contain vapour barrier and two layers of sheathing on the inside. Walls that separate the apartments have timber framing members with 45 × 95 mm2 cross sections. In that case, the walls contain fire and sound insulation. The back side of the wall is left without sheathing to ensure good acoustic performance. Vibration damping material is inserted between top and bottom plates and stud ends to prevent flanking sound transmissions. Separation gaps that exist between adjacent modules to create double layer separating walls and floors and ceilings together with insulation material ensures good acoustical performance with respect to direct sound transmission. Floor joists spans do not exceed 4.0 m because module widths are limited to 4.15 m for transportation reasons. The joists are made from glued laminated timber (glulam) and have cross-section dimensions of 42 × 225 mm2 and are placed at a spacing of 600 mm, apart from bathroom floors where the spacing is 300 mm in order to control deflection. The most challenging aspect of designing the timber vertical load-resisting system was providing for design load effects to be spread evenly from supported elements to supporting elements,

11.6 CONSTRUCTION OF BUILDING

151

so that construction methods could be consistent throughout the superstructure. This was an issue, for example, where large openings were required in facades for windows. In such instances, the adopted solution was to frame the openings with glulam elements that were larger than other framing elements. Glulam elements were also incorporated inside exterior walls to support loads from the balconies. Roof snow load and self-weight is carried by timber trusses supported by the outer walls of the modules. Close attention was paid to avoid problems in situations where wall studs were loaded with horizontal framing members in compression, perpendicular to the grain. In some cases, use of double or triple wall studs was required to keep the design stress transfers at acceptable levels.

11.5 Lateral load resisting system In Stockholm, the maximum design wind load for this type of building is 0.79 kPa, and the city has very low risk of a seismic event. Therefore, the lateral load-resisting system only had to be designed to resist the effects of wind. The building superstructure has the maximum slenderness ratio (H/B = height divided by width) of 1.7 (= 30.9 m/17.9 m). In Sweden only superstructures with H/B greater than 5.0 are considered slender and required to be designed taking into account dynamic wind forces, which means that simple equivalent static force design methods were applicable. The RC walls in the lowest four storeys act as shear walls that stabilize that portion of the building against the effects of design wind loads. The light-frame perimeter walls of the timber modules were designed to act as shear walls of the upper six storeys. Where possible, vertical planes of RC walls are aligned with vertical planes of the timber shear walls, to avoid eccentricities in flows of wind-induced shear forces from one level to another, down through the lateral load-resisting system elements and eventually to the foundation and the ground. The maximum shear flow through the timber shear walls can be resisted by two layers of sheathing. Using a combination of one cement-bonded layer and one gypsum plasterboard layer was found to be sufficient for this, and also met the 90-minute fire resistance requirement for the building. Uplift, overturning, and shear flow force transfers are handled by steel plate link elements, as illustrated in Figs. 11.2 and 11.4, and placed at locations as shown in Fig. 11.3. Floor assemblies act as rigid horizontal diaphragms for the purposes of transferring wind forces from building facades to shear walls and maintain the building shape as it sways because of wind pressures on the facades. The connections between horizontal timber diaphragms and shear walls had to be designed to be sufficiently strong and stiff to transfer forces effectively but sufficiently flexible to avoid creating monolithic joints that would transmit sound and vibration directly or by flanking between storeys. The lowest timber modules are therefore connected to the RC storeys by steel angles, as shown in Fig. 11.3.

11.6 Construction of building The RC foundation and storeys were constructed by one contractor and the timber modules were installed by another, with each having suitable and necessary experience. The timber modules arrived at the construction site from the manufacturing plant (which was 900 km away) covered

152


2

100

0 8 2

A 120

A

Fig. 11.4 Example of hold-down devices (dimensions vary with load levels), units: mm by tarpaulins. As there was no storage space at the site, the modules were lifted from the delivery trucks directly into their final positions by a tower crane. This required that the modules be fabricated to correct design dimensions to avoid unnecessary site work. Dimensions were correct in all cases relative to the requirement that gaps left between adjacent modules measured between 20 and 25 mm (i.e. 5 mm tolerance per outside module dimension). Contacts between modules were initially through simple male–female joints, with steel lin k elements being added after final placement of particular modules (Figs. 11.2 to 11.4). Gaps between modules were made airtight and fireproof by filling them with insulation material. As timber storeys were added, the storeys themselves became the working platform for the next storey, with each level being made watertight as quickly as possible. This ensured that finishing work below the current working platform was done in dry conditions. Activities were planned such that a full storey was added in a single day. The first three of those storeys were assembled during Tuesday to Thursday of one week and the remaining three during the following week. Vapour barriers in exterior walls were made continuous on-site by securing overlapping projecting pieces of barrier material that had been preinstalled in the modules at the manufacturing plant. Building services (e.g. plumbing and electrical) were also preinstalled in the modules, and on-site internal work was restricted to installing various systems in service shafts and coupling those to modules and to external services. Fireproof sealants were used to block openings around pipework. The building is shown in nearly completed form in Fig. 11.5. As seen in that picture, the facades are clad in material that, except for exposed concrete in lowest storeys, gives no hint of the nature of the structural system beneath.

153


11.7 Additional comments The Björkbacken project demonstrates that it is possible to construct quite tall buildings in congested urban locations rapidly and efficiently, based on a combination of cast-in-place RC storeys below and modularized prefabricated timber storeys above. Although not discussed in detail, it should be mentioned that construction of this particular building was cost-competitive and provided a high quality living environment for the elderly. The low mass of the light-frame timber modules greatly reduced the structural demands on the RC substructure compared with other options. This approach can in general lead to substantial cost savings, especially in the case of foundations, and in some instances the incorporation of light-frame or other timber storeys will enable relatively tall buildings to be constructed on sites Fig. 11.5 Building close to completion where ground conditions would otherwise preclude similar height buildings from being constructed. This project is consistent with the philosophy embodied by this Structural Engineering Document (SED) that timber should be thought of by architects and engineers as a material that is suitable for use on its own or in combination with other materials.

Acknowledgements This chapter is based on information, diagrams, and photographs supplied by Professor Erik Serrano and Dr. Johan Vessby of Linnaeus University in Växjö, Sweden.

155 Chapter

12

Looking to the Future

Summary: It is impossible to be exact about what types of buildings will be constructed in the future, what types of superstructures they will have, or what construction materials will be used. However, it is possible to say with a high degree of certainty what will be technically possible in those respects. This chapter discusses likely achievable building heights for multi-storey superstructure systems of different types that employ primary elements made from timber and/ or timber-based composite materials. The LifeCycle Tower concept for constructing timber framed buildings with up to 20 superstructure storeys is mentioned as a metaphor for the types of systems that are already being envisioned in various parts of the world. Also discussed in this chapter is the nature of changes to structural and fire design codes that would facilitate engineering design of building superstructures containing timber elements.

12.1 Likely limits on heights of multi-storey superstructure systems The discussion in this chapter addresses what might become technically possible during the next few years based on the application of emerging design and construction know-how, new timber-based materials, and the continued development of Performance Objective-based design practices and codes. Maximum achievable heights of superstructures depend on a harmonious combination of mechanical capabilities of construction materials with architectural decisions. Consequently, there cannot be exact answers to questions concerning the inherent limits on numbers of storeys/heights of building superstructures separately from questions of architectural form. From an engineering perspective, the simplest building superstructures to design and construct are those having simple shapes and modest ratios of the height to footprint dimensions. Taken to the extreme, once voids are compacted, almost any material can support an infinite height of construction above it, if adequately constrained against lateral expansion and equipped with an appropriate foundation. Mountains attest to this, as do massive man-made masonry structures like pyramids and cathedrals. Unlike mountains and massive historical structures, modern buildings are typically expected to be highly voided internally, which places restrictions on harmonization of construction material and architectural choices.

156

CHAPTER 12. LOOKING TO THE FUTURE

For timber structural elements, as with concrete or masonry ones, the ultimate capacity attainable in compression depends on the degree of peripheral containment provided by other structural parts or internalized containment by embedded steel or other types of lateral reinforcing materials. In the case of material like glued laminated timber (glulam), measures as simple as inserting self-tapping reinforcing screws could increase axial compressive capacities of columns. Therefore, once the shape of any superstructure has been selected, the primary operable question in determining its maximum possible height is how porous it can be near the base (what the minimum void proportion can be) without disrupting the functionality of the building to which it belongs. This statement presupposes that designing portions of superstructures to resist combined tension and shear force flows is simpler than stabilizing them to resist combined compressive and shear force flows. Usually this is a quite reasonable presumption, and that is assumed in comments below.

12.1.1 Lightweight timber plate assemblies Previous chapters have not discussed timber lightweight plate assembly superstructure systems in any detail. Yet it can be supposed that they will be used for construction of multi-storey building with heights greater than what is traditional. This assertion reflects that such construction systems have dominated the residential sectors in North America, parts of Northern Europe, and elsewhere for centuries. In lightweight plate timber construction, wall and floor plate substructures act together as assemblies of folded and interlocking elements, with the plate elements consisting of structural woodbased composite sheathing panels that are stiffened by edge framing and intermediate framing components that are attached by mechanical fasteners (e.g. nails and screws) or glue. From a structural engineering perspective, the individual plate elements are rib-stiffened plates capable of efficiently resisting in-plane forces and out-of-plane bending forces. Many variations in the basic theme are possible with constructed systems mostly being compartmentalized assemblies that are similar in form to internally divided cardboard boxes. Structurally, the systems can be very strong and robust, but only as long as the substructures remain interconnected. Smith [137] addressed this issue of where the limits on the heights of timber lightweight plate superstructures lie from the structural engineering perspective. He examined two classes of construction methods suitable for multiple residential occupancies or mixed office–residential occupancies. The construction methods considered were the platform and balloon methods, as illustrated in Fig. 12.1. Within each method, possibilities considered are use of “standard” timber framing materials like sawn lumber and use of modern timber-based composite products like laminated veneer lumber (LVL) that have enhanced engineering properties. Also considered was how supplementing timber lightweight plate assemblies with “building cores” consisting of reinforced concrete (RC) shear walls, or other relatively massive systems capable of resisting lateral wind and seismic loadings on taller superstructures, would affect achievable superstructure heights. Table 12.1 summarizes those results, based on the relationship that each storey corresponds to a superstructure height increment of 3 m. The findings are that 10- to 20-storey structures are theoretically feasible, if best possible construction methods are employed, especially if there is a RC core(s) in the buildings to stiffen them laterally. In most countries, buildings with more than three above-ground storeys are required to have

157

12.1 LIKELY LIMITS ON HEIGHTS OF MULTI-STOREY SUPERSTRUCTURE SYSTEMS (a)

(b)

Continuous wall plate

Discontinuous wall plates

Fig. 12.1: Construction methods for lightweight plate assemblies: (a) platform construction; (b) balloon construction method (inserted photograph courtesy of Canadian Wood Council) System type ( Fig. 12.1)

Load Combinations: Dead load plus Occupancy + snow

Wind

Earthquake

Standard timber

6

6

2

Enhanced timber

10

10

5

Composite timber–RC core

10

10

10

Standard timber

12

6

2

Enhanced timber

20

10

5

Composite timber–RC core

20

20

20

Platform construction

Balloon construction

Table 12.1: Maximum possible number of storeys for lightweight plate assemblies [137] elevators, and most buildings with more than four storeys must have fire escapes and firefighting access staircases protected by non-combustible material. Therefore, all modern tall buildin gs must already have RC or other structurally substantial walls. Therefore, these could satisfy structural system requirements without the need for unusual measures. Platform construction has lesser height capabilities compared with balloon construction, because platform construction involves the transfer of compressive forces between storeys and from the lowest storey to the foundation by compression perpendicular to the weak horizontally oriented layers of timber (Fig. 12.1a). The balloon construction method eliminates the source of weakness inherent to platform construction by aligning vertical wall framing members such that force transfers are only parallel to their axes (i.e. parallel to grain). In principle, it is possible to reinforce weak zones in platform construction systems, but normally it is simpler to not select an approach that creates weak zones.

158


Despite what Table 12.1 suggests, there is much practical experience that indicates that the achievable number of superstructure storeys/heights using lightweight plate assemblies is constrained mostly by the ability to economically counteract problems associated with vibration and sound transmissions between building occupancy units (e.g. between residential apartments). Also, there are public and firefighter concerns about fire performance of lightweight plate assemblies. Consequently it is speculated that building superstructures constructed from timber lightweight plate assemblies will never be more than around 10 storeys (i.e. 30 m high).

12.1.2 Massive timber plate assemblies Massive timber-plate assemblies employing materials like cross-laminated timber (CLT), as discussed in Chapter 7, have strong potential for use as medium-rise and possibly taller superstructures. Figure 12.2 shows a low-rise CLT structure under construction. At the time this Structural Engineering Document (SED) was written, the tallest such building was a multipleoccupancy residential building in Melbourne, Australia, that has nine CLT storeys constructed over a RC plinth storey. Full-size shake-table tests were performed in Japan by Italian researchers on three- and seven-storey building superstructures constructed from CLT plates in the order of 100 to 200 mm thickness [15]. Neither of the test structures was damaged significantly by simulated maximum credible earthquake events. The smaller structure was destroyed by a controlled fire test after completion of structural tests. It can be seen from the Italian studies that, from both seismic and fire performance perspectives, honeycomb construction CLT buildings can behave excellently, if properly designed. The originally envisioned maximum heights for massive timber-plate assemblies for residential or office-type superstructures was in the order of 10 storeys (circa 30–35 m) with or without RC storeys below, with the limiting factors being economical countermeasures against inter-occupancy sound and vibration transmissions. Sound and vibration problems are always issues to which plate systems are prone to. Conceptually, heights greater than 35 m are feasible. Constructing superstructures that are completely assemblies of massive timber plates is technically feasible to heights of 50 m or greater, from a structural engineering perspective. At the time of writing this SED, detailed design solutions were (a)

(b)

Fig. 12.2: Massive CLT plate assembles: (a) CLT plates being hoisted for installation; (b) hotel extension under construction

12.1 LIKELY LIMITS ON HEIGHTS OF MULTI-STOREY SUPERSTRUCTURE SYSTEMS

159

being prepared for buildings with around 15 to 20 storeys built mainly from CLT, for diverse geographic locations (e.g. Milan in Italy and Vancouver in Canada). The ultimate limit on structural height will depend on the slenderness and regularity of shape of the superstructures. Most likely, the critical factors will be the ability to design connections capable of handling wind and seismic shear and overturning force flows between storeys and between the superstructure and foundation, when superstructures have regular geometries. In cases where the superstructure geometry is irregular, achievable heights are likely to be considerably less. This reflects the need to pay close attention to avoiding out-of-plane instability of perimeter walls and floor diaphragms, and controlling lateral sway and acceleration rates near the perimeter walls. Within the limits of the above comments and the discussion in Chapter 7, it is speculated to be unlikely that many massive timber-plate superstructures will ever exceed the originally envisaged height of 10 or perhaps 12 storeys (circa 35–40 m), but rare examples of buildings will probably reach heights of around 40–50 m. The suggested range of maximum heights does not include systems where massive timber plates are installed as floor slab and wall panels of buildings with frameworks of materials such as steel, RC, or glulam.

12.1.3 Heavyweight timber-framed assemblies As illustrated in Fig. 12.3, heavyweight frame assemblies employing timber as primary structural elements have been successfully constructed for many centuries, with the Sakyamuni Pagoda of the Fogong Temple being an iconic example (Fig. 12.3a). That pagoda is 67.31 m high, including an 11.46 m steeple. It was built in 1056 during the reign of Emperor Dao Zong Di of the Liao Dynasty and has five storeys, the top four of which have mezzanine floors, resulting in a total of nine floor levels. It is octagonal in plan and exceeds 35 m in width at the ground level. The superstructure gains it stiffness and stability from a combination of carpentry joints, geometry, and frictional constraint resulting from the enormity of its self-weight. The principles of construction are the same as those outlined in the still existent building code (the Yingzao Fashi) published by Li Jie in 1103. Reportedly, the pagoda has survived over 40 significant seismic events including a pair of magnitude 7 earthquakes and half a dozen 6–6.9 magnitude earthquakes, as measured on the Richter scale [7]. The pagoda’s primary defense mechanism against structural overload is frictional damping, which modern analysts would typically ignore. There exist many other examples, both historical and modern, of modern multi-storey superstructures that employ moment resisting timber frameworks as their primary structural system (Fig. 12.3). However, such buildings are not common because it is difficult and expensive to make moment transferring connections between framework members. Difficulties are associated with the tendency of framework members to split, unless they have complex carpentry junction arrangements, and/or members are reinforced at junctions, and/or use special connection hardware. Also, because timber is quite compliant and shrinks perpendicular to grain, even well-made carpentry joints do not tend to be reliable in creating rigid connections in frameworks. Therefore, it is most normal in multi-storey heavy timber framed superstructures to have separate gravity and lateral load-resisting systems in which framework connections approximate hinges or are simple bearing joints. As discussed in Chapter 5, lateral rigidity in complete systems is provided by cross-bracing, shear panels of various materials, or masonry or RC building cores. Floors and roofs act as horizontal diaphragms that prevent in-plan distortion of the frameworks.

160

CHAPTER 12. LOOKING TO THE FUTURE (a)

11.456

10.747

7.820

8.635

8.855

14.608

All measurements in meters 5.981 5.063 13.334 4.956 6.247

(b)

(c)

(d)

(e)

(f)

Fig. 12.3: Timber heavy-framed assemblies: (a) Sakyamuni Pagoda of Fogong Temple in Yingxian, China; (b) Venetian example of carpentry tying parts together under heavy gravity load; (c) Bolognese knee braces in a medieval porticoed residence; (d) Claremont Hotel under construction in 1906 (source: University of California Berkeley Library); (e) modern multi-storey car-park in Mura, Austria, with separate timber gravity and lateral load-resisting systems; (f) modern five-storey building in Baar, Switzerland, with separate gravity (timber frame with timber–concrete composite slabs) and lateral load-resisting systems (RC core)

12.1 LIKELY LIMITS ON HEIGHTS OF MULTI-STOREY SUPERSTRUCTURE SYSTEMS

161

Late 19th- and early 20th-century hotels in North America are believed to be the largest residential timber-frame buildings ever constructed, with the now demolished Yellowstone Canyon Hotel having been the largest [8]. The largest remaining structure is the Claremont Hotel in Oakland, California, which has 279 rooms and conference facilities that together occupy 10 floors (Fig. 12.3d ). The ridge of the upper roof is approximately 36 m above the foundation, and a central tower rises about 49 m above the foundation. Tall timber-frame commercial buildings were constructed in the USA in the 19th century, with many having six or seven timber storeys and the tallest having nine storeys. Such buildings were filled with vibrating heavy machinery. Reportedly, sprinklers for fire suppression were invented to protect them. Another interesting aspect of the history of tall commercial buildings in the USA is that it was common until about the mid-19th century for cast iron columns to be used. However, following some catastrophic brittle failures of such elements they were widely replaced by timber columns, which a halfcentury later were in turn widely supplanted by steel elements [138]. What can be said for certain is that heavy-frame, timber buildings perform well at heights of at least 50–60 m (i.e. in the range of 55 m which is the height of the Sakyamuni Pagoda minus its steeple). This is without requiring more than simple construction details. However, the theoretical maximum height using modern structural engineering approaches is much greater. For squat shaped buildings, the structural limit on height is in the order of 160 m, with the ultimate capacities of the columns in lowest storeys being the critical issue. This estimate is based on 1% of the lowest floor plan being occupied by columns. Slender shaped buildings of types more likely to be built will have maximum possible heights nearer to 80 m. This relates to inability to economically constrain magnitudes of translational and torsion movements caused by wind as the limiting factor. As discussed in Section 12.2, the LifeCycle Tower concept is proposed as a suitable way to economically construct superstructures to about 64 m (20 storeys). The suggestion that achieving heights of 160 m is technically possible without consideration of non-structural superstructure performance criteria may appear unrealistic. However, readers should bear in mind that on February 21, 1872, William Ferguson measured the stem of a fallen eucalyptus tree (Eucalyptus regnans) at Watts River, Victoria, Australia to be 133 m long. That dimension did not include the tree top that had broken off. At the break, the diameter of the stem was about 1 m and it was estimated that the tree had probably been at least 1 52 m tall. Structurally speaking, tree stems are not very efficient compared to what engineers can achieve using the same mass of material, because tree stems have architectures that support life functions of the trees, apart from achieving structural efficiency at various stages of tree growth. The 118 m Gliwice Radio Tower, shown in Fig. 1.2, demonstrates what humans have achieved even without timber construction capabilities as sophisticated as those existing now. Future timber heavyframe building superstructures will never achieve heights that take them into the skyscraper classification (> 100 m), because account must be taken of the relative proportion of the total supported weight that is non-structural components, and because humans have quite a low tolerance for building movements during events like wind storms.

12.1.4 Hybrid/composite assemblies There exist many options for using timber in combination with other materials to create composite multi-storey building superstructures. As already discussed in Section 1.2, historical experience proves that combining materials can result in exceptionally good service performance.

162


Roof

Secondary structural system: • Preventing disproportionate collapse • Fire protected

s t f a h s e c i v r e s g n i n i a t n o c e r o c e t e r c n o c C R

Thermal jacket Cladding

Subspace: • Single occupancy

Isolated timber compartments

• Isolated from neigbours • Internal isolations possible

• Modular prefab timber • Internally subdivided e r o c C R

Composite isolating layers for: • Fire separation • Vibration & sound separation • Structural damping

Superstructure to foundation isolation layer

Foundation

Fig. 12.4: Concept for high-performance multi-material tall buildings [20]

However, any combination of materials is effective only if it is linked to a thorough understanding and careful selection of the structural and other design concepts. Modern composite systems are designed with many performance attributes in mind. Combining materials can add damping to systems to improve their response during strong wind and seismic events, to provide toughening mechanisms against propagation of localized damage into system-level damage, to improve vibration serviceability, to minimize sound and vibration transmissions, to improve fire performance, or for other purposes. Smith and Frangi [20] suggest that the future of well-designed tall buildings containing timber substructures will be those that contain multifunctional interfaces made from multiple materials that are located between major substructures and between the superstructures and the foundations. Figure 12.4 illustrates the concept. Important to note is that when multiple functional objectives are involved, achieving high design efficiency does not necessarily mean that different materials will be combined to create structurally efficient monolithic constructions, because maximizing structural efficiency can impair attainment of other objectives. It should also be recognized that particular attributes and properties of combined materials have to be accommodated, or there will be problems of force-flow concentrations associated to differences in compliance and different sensitivity to temperature and moisture flows (e.g. associated with processes like curing, drying, and changes in outsideand inside-building environments). The term composite system/assembly refers to situations where two or more construction materials are designed to act as one unit. This can be synonymous with hybrid systems. From a purely structural perspective, it is known that using relatively low-mass timber components instead of some traditional heavy components within tall multi-storey framed

163

12.2 EXAMPLE OF PROPOSED SYSTEMS: LIFECYCLE TOWER CONCEPT Structural representation of 24 storey building

m 6 9 = m 4 × 4 2

Comparison of peak lateral seismic drifts

Roof 24 r 18 e b m u n y 12 e r o t S

SRC_centre (C) SRC_corner (B)

6

SXL_centre (C) SXL_corner (B)

Ground 0 0

50

100

150

200

250

300

Drift at top of storey (mm)

Comparison of modal periods 3.5 3

2 5 . 6 m

m 4 8. 3

) . c 2.5 e s ( 2 d o 1.5 i r e P 1

SRC SXL

0.5 0 1

5

9

13

17

21

Mode number

Fig. 12.5: Steel frame with XLAM massive timber or RC slabs [89]: (a) structural representation of 24-storey building; (b) comparison of peak lateral seismic drifts; (c) comparison of modal periods (SRC denotes steel framework with RC slabs, SXL denotes steel framework with CLT, XLAM, slabs; locations centre and corner are alternative plan positions on floor and roof slabs)

superstructures can yield significant gains. As discussed in Chapter 6 and illustrated in Fig. 12.5, using massive timber floor slabs instead of RC floor slabs in a 24-storey building markedly alters the dynamic response and reduces levels of lateral drift during strong earthquakes by about 40% [89]. Potentially, there is reason to believe that if properly designed and constructed, future building superstructures containing substantial amounts of timber could surpass the heights of current super tall structures by a considerable margin ( Fig. 12.6 ). This possibility reflects that the self-weight of the system would greatly reduce effects of gravity and wind-induced stresses in steel or possibly other materials in the primary framework, and help control lateral motions by passive methods.

12.2 Example of proposed systems: LifeCycle Tower concept The LifeCycle Tower concept is based on using glulam framed superstructure systems with minimum lateral bracing elements that work in conjunction with RC building cores to resist structural loads. The intent is to also accommodate various architectural configurations, keep construction details as simple as possible, and provide superstructure frameworks that are compatible with every available facade system ( Fig. 12.7 ). Structural systems have only three main components: the building core, the floor diaphragms, and the facade columns ( Fig. 12.8).

164


Is it really possible?

1000 m

800 m

) 9 0 0 2 (

All-timber buildings D A y r u t n e c

h t

1 1 y l r a E

D A y r u t n e c

h t

0 2 l a c i p y T

D A y r u t n e c

e e r t n o s u g r e F

h t

0 2 e t a L

m 8 1 8 – s b a l s C R h t i w e m a r f l e e t S

) ? 0 2 0 2 – n e h w ( s b a l s m a L X h t i w e m a r f l e e t S

600 m

400 m

200 m

0m

Fig. 12.6: Tallest structures of different types

Fig. 12.7: LifeCycle Tower modular systems at various stages of construction

The illustrated completed system has 20 above-ground storeys of a 24.8 by 38.3 m rectangular footprint and a total height of 76 m, including two below-ground plinth storeys. Design of the building core assumes it will be erected in two 32 m high above-ground construction stages, which facilitates maximization of lateral building stiffness and provides localized ductility, and

165

12.2 EXAMPLE OF PROPOSED SYSTEMS: LIFECYCLE TOWER CONCEPT

(a)

(b)

Fig. 12.8: Superstructure arrangement of LifeCycle Tower: (a) typical floor layout; (b) cross section parallel to minor plan axis

economic competitiveness with alternative high-rise building systems. Having only two horizontal interfaces (i.e. core to foundation and between construction stages) permits a high degree of prefabrication and shortens construction time. Although employing central building cores has well-defined structural benefits (Chapter 5), it can have disadvantages in terms of efficient use of floor area. A ribbed composite glulam rib and concrete slab diaphragm system was selected to transfer the very large forces from the facade columns to the building core, and secondarily for fire separation between floors. In addition to increased resistance to the spread of fire beneath it, the diaphragm arrangement provided space between the ribs to accommodate technical building services within the depth of the deck. Facade columns were designed to satisfy structural requirements associated with gravitational loads and to transfer facade forces directly to floor diaphragms. Notably, avoidance of application of compressive forces to horizontal layers in timber superstructures is critical, because materials like glulam are weak and compressible when stressed in the perpendicular to the grain direction. Therefore, the LifeCycle Tower facade column connections were designed to completely avoid such situations in the transfer of gravitational and other vertical forces between storeys and across the depths of the floor diaphragms. Attachment of the columns to the diaphragms was via single mortice carpentry joints that were not dependent on metal components for force transfer. The adopted connections produced close to pin-end conditions for the columns, thereby rendering their effective buckling lengths close to their physical lengths. During construction, several pairs of columns could be installed as a prefabricated facade unit, with secondary facade elements preinstalled, so that site work could progress rapidly compared with alternative high-rise construction systems. According to the fire protection approach the system had to be designed and constructed to meet regulations applicable to various building occupation categories (e.g. private residential, hotel

166


(a)

Elevator lobby

Offices Corridor

Stairwell

(b)

Primary escape distance 29m Secondary escape distance 29m

Primary escape distance 22.5 m Secondary escape distance 34 m

Fig. 12.9: Fire design of LifeCycle Tower: (a) fire zoning strategy (schematic), blue zone = fully fire protected building core, yellow zone = escape staircase; (b) escape route lengths within a storey (schematic)

and office) in various jurisdictions. Consequently, the building fire performance concept was designed to address the following considerations: •

Creating a simple and comprehensible layout of escape routes and provisions for firefighting access routes.

•

Using standardized prefabricated elements, so that design complexity will not lead to construction errors.

•

Provision of 90-minute fire resistance for exposed timber elements (e.g. glulam ribs in diaphragms and glulam facade columns).

•

Use of only mineral-based building materials to create junction interfaces between fire walls and ceilings (bottom surface of floor diaphragms) or tops of floors.

•

Ensuring that horizontally orientated building service conduits connect only to noncombustible building elements.

•

Encapsulation of building services within non-combustible materials.

•

Not creating enclosed spaces adjacent to glulam or other combustible materials that would facilitate undetected fire spread.

•

Not constructing facades in ways that facilitate fire growth between storeys (e.g. extension of concrete slabs in floor diaphragms right up to the facade).

•

Dividing floors into zones that permit escape of building occupants and safe access to each floor level for firefighters ( Fig. 12.9).

12.3 REFOCUSING DESIGN CODES

167

Notably, the choice of 90-minute fire resistance may not be consistent with requirement in various regulatory jurisdictions. However, satisfying other fire resistance requirements is not problematic. Taking contemporary Austrian building regulation requirements as an illustrative example, the maximum permitted escape route length within a storey is 40 m measured from any point in an occupied space to the stairwells of primary and second escape routes. Figure 12.9(b) shows application of this requirement to the LifeCycle Tower . As is clear, from this example, fire performance requirements may sometimes impose limitations on adoption of the discussed approach, but not necessarily on using timber to construct tall buildings in other ways. In this example, the building services include an integrated high-pressure water mist extinguishing system that self-activates to extinguish and suppress fire based on three principles of physics: (1) cooling, which is the most important property of water in firefighting; (2) inertisation (sudden local evaporation of the water mist inside the flames), which results in the water mist expanding to about 1700 times its original volume to displace oxygen inside the flames and suppression of fires; and (3) blocking/shielding of radiated heat, which because of the high number and density of extremely small mist droplets (micro-water droplets from 20 to 200 µm) absorbs heat very effectively. In addition to extinguishing the fire, the discharged water mist binds the fumes from fires, washes them out of the air, and carries them to the floor. Self-activating extinguishing systems are a primary tool for mitigating loss of life and property damage (Chapter 3). The LifeCycle Tower concept clearly demonstrates the feasibility from structural and fire engineering perspectives of constructing high-rise building superstructures that incorporate frameworks made from glulam members or other large dimension structural timber elements. As with any alternative system(s), the key to success is creation of a holistic overall concept and paying close attention to technical details during implementation of such a concept.

12.3 Refocusing design codes As the nature of structural systems evolves, so do structural design codes. This reflects many factors including alterations in capabilities of materials, engineering know-how, and occupier requirements for how building should perform in service. Nine centuries ago, when Li Jie published the Yingzao Fashi [7], structural use of timber was at the zenith of structural design capability. However, now the situation is somewhat different, and structural engineers use design documents that are predicated principally on the assumption that high-rise building superstructures will mostly, or exclusively, be constructed from non-timber materials. The following subsections briefly draw attention to some use of timber-oriented factors that contemporary design code committees may wish to consider. 12.3.1 General requirements

This SED is written from the viewpoint that there should be no special provisions in building codes related to the specification of design loadings or performance requirements/objectives particular to timber or other materials. Pursuant to this, general provisions in building and loading codes would include the following: • A definition of durability requirements for primary structural, other structural and nonstructural building elements. • Explicit requirements related to anticipatable sensitivities of structural elements to static and cyclic fatigue processes (which differs substantially between materials and applications).

168


• Replacement of remnants of prescriptive structural design criteria, like those related to rigidity of diaphragms and inter-storey drift, by precise performance objectives such as “diaphragms must tie substructures together so that they move in unison”, and “inter-storey drift and must not be of a level that damages structural elements or non-structural facade and fire containment elements”. • A requirement that material-specific design codes be fully compatible with assumptions embedded in the loading codes with respect to the basis on which characteristic design properties of elements and element “sizing rules” are derived. • Clarification of the acceptability/applicability of Capacity Based Design concepts in relation to and uses of materials in composite and hybrid construction. • Clarification of requirements when equivalent static loading methods are applied to seismic design of hybrid superstructure systems (e.g. applicable to podium construction; Section 2.6 ). The above-listed examples of reorientations of general aspects of structural design code requirements are likely to benefit engineers who design superstructures made of any material(s), and owners and occupiers of buildings who experience the results of design decisions. 12.3.2 Timber structural design

Concerning provisions of timber design codes, there are a number of ways in which they could be usefully reoriented to facilitate the reintegration of timber into the mainstream of structural design of relatively large and tall multi-storey building superstructures. The following are suggested as primary additions and clarifications to what currently exists: • Provision of alternative characteristic strength data that permit elements to be designed based on onset of an inelastic response (pseudo-yield). This would apply to elements within statically indeterminate arrangements which do not permit attainment of unconstrained element capacities. • Provision of guidance on determination of whether structural elements are expected to fail in a ductile or brittle manner, as a formal or informal aid to ensuring that design practices will limit potentially brittle releases of energy, if elements in load paths fail. As indicated in Section 2.2.2, the ability to ensure that connections are ductile fuse elements in load paths is particularly important. • Provision of guidance on the range of capacities that structural elements possess (e.g. 5 percentile and 95 percentile characteristic values), which is necessary information for the application of Capacity Based Design concepts. • Explicit statements of any expectations regarding utilization of proprietary structural elements in combination with commodity (non-proprietary) elements for which timber design codes already provide design guidance. Additionally, it is highly desirable that design guidelines/manuals give explicit information relevant to the calculation of deformations in structural systems and substructures, which is relevant to the estimation of internal force distribution and deflections. Especially, there is a need for guidance on the behaviour of connections that typically behave as semi-rigid spring elements. In the absence of other guidance, designers would be wise to satisfy themselves that they adopt a conservative interpretation of connection stiffness during estimation of internal force flows.

12.4 FINAL COMMENTS

169

12.3.3 Timber fire design

As discussed in Chapter 3, fire design may be the most challenging aspect of finding a total design solution for any building superstructure containing timber elements. Many building codes prescriptively specify maximum total floor areas associated with various expectations for the fire resistance ratings of building elements. Yet, it is often unclear how to rationally relate code requirements for performance of specific design solutions to “Root Performance Objectives”. As the example projects discussed in Chapters 8 and 9 illustrate, relatively large superstructure systems containing timber elements are currently required to be designed as Alternative Solutions. The basis of this is proof of their equivalent performance to Acceptable Solutions. Equivalency can be proven based on a mixture of R&D evidence, engineering calculations, and professional judgement, but it would be much more desirable that Alternative Solutions be directly referenced to Root Performance Objectives. The current practice amounts to product substitution rather than direct determination of “suitability for purpose”, which amounts to rule of thumb practices rather than fully fledged engineering design. Given the rapid development of fire engineering as a science in recent decades, it can be confidently anticipated that fire design of superstructures will soon be placed on a fully rational engineering basis, which will be of advantage, irrespective of the construction material(s) involved.

12.4 Final comments This SED was written at a time when available timber and timber-based construction materials, construction methods, design codes, available design tools, and training of structural engineers were all in a period of rapid transition. The influential fluxes are concomitant with transitioning from what became possible around 150 years ago (circa mid-19th century onwards) when steel products began replacing timber as a primary construction material for large and tall buildings, to what is becoming possible at the beginning of the 21st century. New possibilities are emerging, thanks largely to intensive R&D in various countries, in parallel with growing demands from societies transitioning from traditional building infrastructure requirements to buildings that better suit urbanized lifestyles. This document is, therefore, not anchored to specific building/loading code provisions or material specific design practices. Instead, the approach taken has been to anchor the discussion to concepts that reflect purely fundamental technical aspects of the structural and fire performance design of buildings likely to be constructed in the short-, medium- and longer-term. Uptake of structural ideas is never uniform. It will be up to nationally-, and sometimes internationally authorized (e.g. European Union) design code committees and regulatory authorities to endorse or reject ideas presented here and elsewhere. If concepts and suggestions within this SED are accepted that should be in the manner of regarding them as an integrated package, rather than as dissectible parts.

Acknowledgements The information and diagrams for the LifeCycle Tower concept was provided by Jakob Bonomo and Harald Professner of Cree GmbH, Bregenz, Austria.

171

References

Architecture hitecture. Yale [1] Fux, X., Guo, D., Liu, X., Pan, G., Qian, Y., S un, D. 2002. Chinese Arc University Press, New Haven, CT, USA. [2] Smith, I., Snow, M. 2008. Ti mber: an ancient construction material with a bright f uture. Forest Chron., 84(4): 504–510. [3] Hu, S. 1991. The earthquake resistant-resistant properties of Chinese traditional construction. Earthquake Spect., 7(3): 355–389. [4] European Committee for Standardisation (CEN). 2004. Eurocode 5—Design of Timber Structures, Part Part 1-1: General— General—Common Common Rules and Rules for Buildings. EN 1995-1-1:2004, CEN, Brussels, Belgium. [5] Yenne, B. 2003. Seaplanes and Flying Boats: A Timeless Collection from Aviation’s N ew York York,, NY N Y, USA. U SA. Golden Age. BCL Press, New [6] Rihll, T.E. 2007. The Syrakosia or Alexandris, Department of Classics and Ancient History. University of Swansea, UK, http://www.swan.ac.uk/grst/What’s%20what%20 Things /Syrakosia.htm. [7] Lam, F., He, M. 2008. Example of traditional tall timber buildings in China – The Yingxian Pagod. Struct. Eng. Int., 18(2): 126–129. [8] Langenbach, R. 2008. Building tall with timber: a paean to wood construction. Struct. Eng. Int., 18(2): 130–132. [9] Foliente, G.C. 2000. History of timber construction. In: Wood Structures; An East-West Forum on the Treatment, Conservation, and Repair of Cultural Heritage. Kelley, S.J., Loferski, J.R., Salenlikovich, A.J., Stern, E.G. (e ds.), ASTM STP 1351, American Society for Testing and Materials, We West Conshohocken, PA, USA, 3–22. [10] Canadian Standards Association (CSA). 2009. Engineering Design in Wood Wood . CSA Standard 086-09, CSA, Toronto, Toronto, Canada. [11] Smith, I., Landis, E., Gong, M. 200 3. Fracture and Fatigue in Wood Wood . John Wiley and Sons, Chichester, England. [12] Smith, I. 1999. Joints in timber structures: state-of-the-art knowledge in North America. COST C1 Int. Conf. Control Semi-rigid Behaviour Civil Eng. Struct. Connections . Liège, Belgium, September 17–19, 1998, Of fic fice Publications European Communities, Luxembourg, 255–264. [13] Foliente, G.C. 1998. Design of timber structures subjected to extreme loads. Prog. Struct. Eng. Mater. Mater., 1(3): 236–244.

172

REFERENCES REFERE NCES

[14] Langenbach, R. 2008. Resisting earth’s for ces: t ypologies of timber buildings in history. Struct. Eng. Int., 18(2): 137–140. [15] Ceccotti, A. 2008. New technologies for construction of medium-rise buildings in seismic regions: the XLAM case. Struct. Eng. Int., 18(2): 156–165. [16] Buchanan, A., Deam, B., Fragiacomo, M., Pampanin, S., Palermo, A. 2008. Multi-storey prestressed timber buildings in New Zealand. Struct. Eng. Int., 18(2): 166–173. [17] Institute for Research in Construction (IRC). 2005. National Building Code (updated 2010). IRC, National Research Council, Ottawa, Canada. [18] European Committee for Standardization (CEN). 2003. Eurocode 8—Design of Structures for Earthquake Resistance—P Resistance—Part art 1: General Rules, Seismic Actions and Rules for Buildings. EN 1998-1:2003, CEN, Brussels, Belgium. [19] Smith, I. 2006. Reaching for the Limits with Timber Construction. Reports, Vol. 92. IABSE Symposium on Responding to Tomorrow’s Challenges in Structural Engineering, September 13–15, 2006, Budapest, International Association for Bridge and Structural Engineering, Zurich, Switzerland. [20] Smith, I., Frangi, A. 2008. Overview of design issues for tall timber buildings. Struct. Eng. Int., 18(2): 141–147. [21] Knoll, F., Vogel, T. 2009. Design for Robustness, Structural Engineering Document 11. International Association for Bridge and Structural Engineering, Zurich, Switzerland. [22] Smith, I., I. , Snow, M., Asiz, A., A ., Va Vasic, S. 2007. Failure mechanisms in wood-based materials: discrete, continuum, an d h ybrid finite element representations. A review. Holzforschung, 61(4): 352–359. [23] Smith, J.W. 2006. Structural robustness analysis and the fast fracture analogy. Struct. Eng. Int., 16(2): 118–123. [24] Östman, B., Rydholm, D. 2002. National fire regulations in relation to the use of wood in European and some other countries, Swedish Institute for Wood Technology Technology Research, Trätek Rapport P0212044, Stockholm, Sweden. Research Notes 689 . Espoo, Finland. [25] Mikkola, E. 1990. Charring of Wood, VTT Research [26] König, J., Walleij, L. 2000. Timber Frame Assemblies Exposed to Standard and Parametric Fires, Part 2: A Design Model for Standard Fire Exposure . Report I 0001001, 0001001, Trätek – Swedish Institute for fo r Woo Wood T Teechnology Research , Stockholm, Sweden. [27] Young, S.A., Clancy, P. 2001. Structural modelling of light-timber framed walls in fire. Fire Safety J., 36(3): 241–268. [28] Frangi, A., Fontana, M. 2005. Fire performance based design of multi-storey timber buildings, Proceedings of the IABSE Symposium on Structures and Extreme Events , Lisbon, Portugal, September 14–17. [29] Fontana, M., Maag, T. T. 2004. Fire risk assessment based on Bayesian networks , Proceedings of the 5th International Conference on Performance Based Codes and Fire Safety Design Methods , L uxembourg, October 6–8, 2004, Society of Fire Protection Engineers, Bethesda, MD, USA. [30] Fontana, M., Favre, J.P., Fetz, C. 1999. A survey of 40,000 building fires in Switzerland. Fire Safety J., 32(3): 137–158. [31] Schweizerische Brandschutzvorschriften VKF, 2003. Vereinigung Kantonaler Feuerversicherungen VKF . Bern, Switzerland. [32] Buchanan, A.H. 2000. Fire performance of timber construction. Prog. Struct. Eng. Mater. , 2(3): 278–289.

173 [33] International Organization for Standardization (ISO). 1999. Fire-Resistance Tests— Elements of Building Construction—Part 1: General Requirements, ISO 834-1, ISO, Geneva, Switzerland. [34] American Society for Testing and Materials (ASTM). 1988. Standard Test Methods for Fire Tests of Building Construction and Materials, ASTM E199, ASTM, West Conshohocken, PA, USA. [35] European Committee for Standardization (CEN). 2002. Eurocode 1: Action on Structures Part 1-2: General Action—Actions on Structures Exposed to Fire. EN 1991-1-2:2002, CEN, Brussels, Belgium. [36] Hietaniemi, J. 2007. Probabilistic simulation of fire endurance of a wooden beam. Struct. Safety, 29: 322–336. [37] Buchanan, A.H. 2001. Structural Design for Fire Safety. John Wiley and Sons, New York, NY, USA. [38] Fontana, M. 2005. Fires in buildings, Proceedings of the IABSE Symposium on Structures and Extreme Events , Lisbon, Portugal, September 14–17. [39] Joyeux, D., Bonnot, S., Hietaniemi, J., Korhonen, T. 2004. Risk-based fire safety engineering approach applied to a public building, Proceedings of the 5th International Conference on Performance Based Codes and Fire Safety Design Methods , Luxembourg, October 6−8, 2004, Society of Fire Protection Engineers, Bethesda, MD, USA. [40] König J. 2005. Structural fire design according to Eurocode 5—design rules and their background. Fire Mater., 29:147–163. [41] European Committee for Standardization (CEN). 2004. Eurocode 5: Design of Timber Structures, Part 1-2: General–Structural fire design. EN 1995-1-2:2004, CEN, Brussels, Belgium. [42] Schaffer, E.L. 1967. Charring Rate of Selected Woods—Transverse to Grain, Forest Service Research Paper FPL 69 . Forest Products Laboratory, Madison, WI, USA. [43] Frangi, A., Fontana, M. 2003. Charring rates an d temperature profiles of wood sections. Fire Mater., 27: 91–102. [44] König, J., Walleij L. 1999. One-dimensional charring of timber exposed to standard and parametric fires in initially protected and non-protected fire situations, Report No. I 9908029, Trätek , Swedish Instit ute for Woo d Technology Resear ch , Stockholm, Sweden. [45] Frangi, A., Erchinger, C., Fontana, M. 2008. Charring model for timber frame floor assemblies with void cavities. Fire Safety J., 43: 551–564. [46] Gerhards, C.C. 1982. Effect of moisture content and temperature on the mechanical properties of wood: an analysis of immediate effects. Wood Fiber , 14(1): 4–36. [47] Källsner, B., König, J. 2000. Ther mal and mechanical properties of timber and some other materials used in light timber frame construction , Proceedings of CIB W18, Meeting thirty-three, Delft, Lehrstuhl für Ingenieurholzbau, University of Karlsruhe, Karlsruhe, Germany. [48] Frangi, A., Knobloch, M., Fontana, M. 2009. Fire design of timber slabs made of hollow core elements. Eng. Struct., 31: 150–157. [49] Erchinger, C., Frangi, A., Fontana M. 2010. Fire design of steel-to-timber dowelled connections. Eng. Struct., 32: 580–589. [50] Frangi, A., Knobloch, M., Fontana, M. 2010. Fire design of timber-concrete compo-site slabs with screwed connection. J. Struct. Eng., 136(2): 219–228.

174

REFERENCES

[51] Schmid, J., König, J., Köhler, J. 2010. Fire-exposed cross-laminated timber—modelling and tests, Proceedings von 11. World Conference on Timber Engineering , Riva del Garda, Italy, June 20–24. [52] British Standards In stitution (BSI). 1990. Structural use of timber - Section 4.2, Recommendations for calculating fire resistance of timber stud walls and joisted floor constructions, Standard S 5268-4:1990, BSI, London, UK. [53] Östman, B., König, J., Norén, J. 1994. Contribution to fire resistance of timber frame assemblies b y means of fire protective boards, Proceedings of the 3rd International Fire and Materials Conference, Washington, DC, USA. [54] König, J., Oksanen, T., Towler, K. 2000. A review of component additive methods used for the determination of fire resistance of separating light timber frame construction, Proceedings of CIB W18, Meeting thirty-three, Delft. Lehrstuhl für Ingenieurholzbau , University of Karlsruhe, Karlsruhe, Germany. [55] Schleifer, V. 2009. Zum Verhalten von raumabschliessenden mehrschichtigen Holzbauteilen im Brandfall, PhD Thesis - No. 18156 , Eidgenössische Technische Hochschule, Zürich, Switzerland. [56] Frangi, A., Schleifer, V., Fontana, M. 2010, Design model for the verification of the separating f unction of light timber frame assemblies. Eng. Struct., 32: 1184–1195. [57] European Committee for Standardization (CEN). 2003. Fire Classification of Construction Products and Building Elements—Part 2: Classification using Data from Fire Resistance Tests, Excluding Ventilation Services. EN 13501-2:2003, CEN, Brussels, Belgium. [58] Dock Tower. 2002. Urbane Vi sionen an der Swissbau 02, Architektur & Technik , B+L Verlags AG, Schlieren, Switzerland. [59] Frangi, A., Fontana, M. 2005. Fire performance of timber structures under natural fire conditions. Fire Safety Sci., 8: 279–290. [60] Frangi, A., Bochicchio, G., Ceccotti, A., Lauriola, M.P. 2008. Natural f ull-scale fire test on a 3 storey XLam timber building, 10th World Conference on Timber Engineering , Miyazaki, Japan, June 2–5. [61] Lennon, T., Bullock, M., Enjily, V. 2000. Medium rise timber frame 2000 stair fire test, BRE Report No 200-711, Building Research Establishment, Cardington, UK. [62] Hakkarainen, T. 2002. Post-flashover fires in light and heavy timber construction compartments. J. Fire Sci., 20: 133–175. [63] Standards A ustralia (SA). 2005. Specification for Preservative Treatment. Part 1: Sawn and Round Timber . Australian Standard 1604.1-2005, SA, Sydney, Australia. [64] International Organization for Standardization (ISO). 2008. General Principles on the Design of Structures for Durability. ISO Standard 13823:2008(E), ISO, Geneva, Switzerland. [65] Australian Building Codes Board (ABCB). 2002. Durability in Buildings. Guideline Document, ABCB, Canberra, Australia. [66] Bennett, A., Clark, S., Benge, C. 2001. Design for durability: A review of New Zealan d practice, Proceedings World Building Congress, Conseil International du Bâtiment , Rotterdam, The Netherlands (CD-Rom). [67] Foliente, G.C., Leicester, R.H., Wang, C-H., Ma ckenzie, C., Cole, I.S. 2002. D urability design for wood construction. Forest Prod. J. , 52(1): 10–19. [68] Leicester, R.H., Wang, C-H., Ng uyen, M.N., Foliente, G.C., Cole, I.S., MacKenzie, C. 2008. Engineered durability: completion of a 10-year project, Proceedings World Conference Timber Engineering, Paper No. 403 , Wood Utilization Research Center, Miyazaki, Japan (CD-Rom).

175 [69] Nguyen, M.N., Leicester, R.H., Wang, C-H. 2008. Embedded corrosion of fasteners in timber structures. Manual No. 6, Commonwealth Scientific and Industrial Research Organization - Sustainable Ecosystems, Highett, Australia. [70] Leicester, R.H., Ng uyen, M.N., Wang, C-H. 2008. Derivation of design equations for use with AS1720.1. Manual No. 2, Commonwealth Scientific and Industrial Research Organization - Sustainable Ecosystems, Highett, Australia. [71] MacKenzie, C.E., Wang, C-H., Lei cester, R.H., Foliente, G.C., Nguyen, M.N. 2012. Timber Service Life Design Guide. Forest and Wood Products Australia, Melbourne, Australia (ISBN: 978-1-920883-16-4). [72] Zabel, R.A., Morrell, J.J. 1992. Wood Microbiology; Decay and Its Prevention. Academic Press, New York, NY, USA. [73] MacKenzie, C.E., Wang, C-H., Lei cester, R.H., Foliente, G.C., Nguyen, M.N. 2007. Timber Service Life Design Guide. Research Report, Forest and Wood Products Australia, Melbourne, Australia. [74] Leicester, R.H., Wang, C-H., Cook son, L.J. 2008. A reliability model for termite attack on housing. Reliab. Eng. Syst. Safety, 93(3): 468–475. [75] Grace, J.K., Ab dallay, A., Farr, K.R. 1989. Eastern subterraneum termite (isoptera:rhinotermitidae) foraging territories and population growth in Toronto. Canad. Entomool., 121: 551–556. [76] Su, N-Y., S cheffrahn, R.H. 2000. Termites as pest of buildings. In: Abe, T., Bignell, D.E., Higashi, M. (eds). Termites: Evolution, Sociality, Symbioses, Ecology. Kluwer Academic Press, Dordrecht, The Netherlands, 437–453, Chapter 20. [77] Nguyen, M.N., Lei cester, R.H., Wang, C-H. 2008. Atmospheric corrosion of fasteners in timber structures. Manual No. 5, Commonwealth Scientific and Industrial Research Organization - Sustainable Ecosystems, Highett, Australia. [78] Nguyen, M.N., Leicester, R.H., Wang, C-H., Foliente, G.C. 2008. A Draft Proposal of AS1720.5 – Timber Service Life Design Code. Research Report, Commonwealth Scientific and Industrial Research Organization: Sustainable Ecosystems, Highett, Australia. [79] Kear, G., Wu, H-Z., Jones, M.S. 2007. Corrosion of ferrous- and zinc-based materials in CCA, ACQ and CuAz timber preservative aqueous solutions. Mater. Struct., 41(8): 1405–1417. [80] Leicester, R.H., Wang, C-H, Ng uyen, M.N, MacKenzie, C.E. 2009. Design of exposed timber structures. Aust. J. Struct. Eng., 9(3): 217–224. [81] Canada Housing and Mortgage Corporation (CMHC). 1999. Wood Frame Envelopes in the Coastal Climate of British Columbia. Best Practice Guide, CMHC, Ottawa, Canada (ISBN 0-660-17765-X). [82] Bassett, M.R., McNeil, S. 2005. The theory of ventilation drying applied to New Zealand cavity walls, Proceedings Conference of Institution of Refrigeration, Heating and Air Conditioning Engineers (IRHACE), IRHACE, Auckland, New Zealand (CD-Rom). [83] Burnett, E., Straube, J. 1995. Tests, Ventilation Drying and Pressure Distribution . Research Report, Building Engineering Group, Department Civil Engineering, University of Waterloo, Waterloo, Canada. [84] Cole, I.S., Chan, W.Y., Trini dad, G.S., Paterson, D.A. 2004. Holistic model for atmospheric corrosion: 4 – Geographic information system for predicting airborne salinity. Corros. Eng. Sci. Technol. , 39(1): 89–96. [85] Nguyen, M.N., Leicester, R.H., Wang, C-H., Foliente, G.C. 2013. Corrosion effects in the structural design of metal fasteners for timber construction, Struct. Inf. Eng., 9(3): 275–284.

176

REFERENCES

[86] Weckendorf, J., Smith, I. 2012. Multi-f unctional interface concept for high-rise hybrid building systems with structural timber, 12th World Conference on Timber Engineering. July 16–19, Auckland, New Zealand (CD-Rom). [87] Langenbach, R. 2010. Better than steel? The use of timber for large and tall buildings. In: Cruz, P.J.S. (e d.) Structures and Architecture, Proceedings of the First International Conference on Structures and Architecture , Guimaraes, Portugal, July 21-23, 2010). CRC Press, Leiden, The Netherlands, 103–104 (extended version on CD-Rom). [88] Macdonald, A.J. 2010. The changing relationship between architects and structural engineers. In: Cruz, P.J.S. (e d.) Structures and Architecture , Proceedings of the First International Conference on Structures and Architecture, Guimaraes, Portugal, July 21–23, 2010). CRC Press, Leiden, The Netherlands, 21 (extended version on CD-Rom). [89] Asiz, A., Smith, I. 2009. Demands placed on steel frameworks of tall buildings having reinforced concrete or massive wood horizontal slabs. Struct. Eng. Int., 19(4): 395–403. [90] Madsen, B. 1998. Reliable timber connections. Prog. Struct. Eng. Mater. , 1(3): 245–252. [91] Madsen, B. 2000. Behaviour of Timber Connections. Timber Engineering Ltd., North Vancouver, Canada. [92] Canadian Wood Council (CWC). 2010. Wood Design Manual. CWC, Ottawa, Canada. [93] Larsen, H.J., Enjily, V. 2009. Practical Design of Timber Structures to Eurocode 5 . Thomas Telford Ltd, London, UK. [94] Pampanin, S. 2005. Emerging solutions for high seismic performance of precastprestressed concrete buildings. J. Adv. Concrete Technol., 3(2): 202–222. [95] American Institute of Steel Construction (AISC). 2006. Steel Construction Manual, 13th ed. AISC, Chicago, IL, USA. [96] American Concrete Institute (ACI). 2008. Building Code Requirements for Structural Concrete and Commentary, Standard 318–08, ACI, Farmington Hills, MI, USA. [97] Wilson, E. 2004. Static and Dynamic Analysis of structures , 4th ed. Computer and Structures Inc., Berkeley, CA, USA. [98] Asiz, A., Smith, I. 2009. Structural connections for massive timber elements in hybrid structures. Proceedings of 33rd IABSE Symposium , Bangkok, Thailand. [99] International Code Council (ICC). 2012. International Building Code. ICC, Washington, DC, USA. [100] Lukaszewska, E., Johnsson, H., Fragiacomo, M. 2008. Performance of connections for prefabricated timber-concrete composite floors. Mater. Struct., 41: 1533–1550. [101] Newcombe, M.P., Carradine, D., Pampanin, S., Buchanan, A., van Beerschoten, W.A., Fragiacomo, A. 2009. In-plane experimental testing of timber-concrete composite floor diaphragms, Proceedings of 2009 New Zealand Soc. Earth. Eng., Paper No 19, p8. [102] American Society of Civil Engineers (ASCE). 2005. Minimum Design Loads for Buildings and Other Structures. Standard ASCE-7-05, ASCE, Reston, VA, USA. [103] Ju, S.H., Lin, M.C. 1999. Co mparison of building analysis assuming rigid or flexible floors. J. Struct. Eng., 125 (1): 25–31. [104] American National Standards Institute (ANSI). 2012. Standard for Performance-Rated Cross-Laminated Timber, Standard ANSI/APA PRG 320-2012, ANSI , New York, NY, USA. [105] European Committee for Standardization (CEN). 2012. Structural Timber—Strength Classes—Assignment of Visual Grades and Species. EN 1912:2012, CEN, Brussels, Belgium.

177 [106] European Committee for Standardization (CEN). 2002. Eurocode 1—Actions on Structures—Part 1-1: General Actions—Densities, Self-Weight, Imposed Loads for Buildings. EN 1991-1-1:2002, CEN, Brussels, Belgium. [107] European Committee for Standardization (CEN). 2003. Eurocode 1—Actions on Structures—Part 1-3: General Actions—Snow Loads. EN 1991-1-3:2003, CEN, Brussels, Belgium. [108] European Committee for Standardization (CEN). 2005. Eurocode 1—Actions on Structures—Part 1-4: General Actions—Wind Actions . EN 1991-1-4:2005, CEN, Brussels, Belgium. [109] Weckendorf, J., Smith, I. 2012. Dynamic characteristics of shallow floors with cross-laminated-timber spines, Proceedings of 12th World Conference on Timber Engineering, July 16–19, Auckland, New Zealand, University of Auckland (CD-Rom). [110] Uibel, T., Blaß, H.J. 2006. Load carrying capacity of joints with dowel type fasteners in solid wood panels, Paper CIB-w18 30-7-5, International Council for Research and Innovation in Building and Construction: Working Commission W18 - Timber Structures, University of Karlsruhe, Karlsruhe, Germany. [111] TRADA Technology. 2009. Cross-laminated timber (Eurocode 5) design guide for project feasibility. Timber Engineering Guidance Document GD 10, TRADA Te chnologies, High Wycombe, UK. [112] Gagnon, S., Pirvu, C. 2011. CLT Handbook, Special Publication SP-528E. FPInovations, Quebec City, Canada. [113] Pozza, L., Trutalli, D., Polastri, A., Ceccotti, A. 2013. Seismic design of CLT buildings: definition of the suitable q-factor by numerical and experimental procedures, Proceedings of the Second International Conference on Structures and Architecture, Guimaraes, Portugal, July 24–26. [114] Canadian Standard Association (CSA). 2001. Engineering Design in Wood . CSA Standard O86-01, CSA, Toronto, Cana da. [115] Canadian Standard Association (CSA). 2004. Design of Concrete Structures. CSA Standard A23.3-04, CSA, Toronto, Canada. [116] Canadian Standard Association (CSA). 2001. Limit States Design of Steel Structures. CSA Standard S16-01, CSA, Toronto, Canada. [117] Computers and Structures, Inc. (CSI). 1997. SAP2000 Integrated Structural Analysis and Design Software. CSI, Berkeley, CA, USA. [118] Computer and Structures, Inc. (CSI). 2004. ETABS V9.6.0 Structural Analysis Program. CSI, Berkeley, CA, USA. [119] American Forest & Paper Association (AF&PA). 2003. Calculating the Fire Resistance of Exposed Wood Members . Technical Report 10, AF&PA, Washington, DC, USA. [120] Musterbauordnung. 2002. MBO- Fassung November 2002, zuletzt geändert durch Beschluss der Bauministerkonferenz vom Oktober 2008, Bauministerkonferenz. [121] DIN EN 1 3501-2. 2003. Klassifizierung von Bauprodukten und Bauarten zuihrem Brand-verhalten – Teil 2: Klassifizierung mit den Ergebnissen ausden Feuerwiderstandsprüf ungen (mit Ausnahme von Produkten für Lüftungsanlagen), German E dition, December 2003, Beuth Verlag GmbH. [122] Fire safety concept 06G117-3. 2003. Dehne, Kruse Brandschutzingenieure. Gmbh & Co KG, December 2006.

178

REFERENCES

[123] Deutsches Institut für Normung e.V. (DIN). EN 1991-1-2/NA. 2010. Nationaler Anhang - National festgelegte Parameter - Eurocode 1: Einwirkungen auf Tragwerke 1-2: Allgemeine Einwirkungen – Brandeinwirkungen, German Edition, December 2010, Beuth Verlag GmbH. [124] Deutsches Institut für Normung e.V. (DIN). 10 52. 2004. Entwurf, Berechnung und Bemessung von Holzbauwerken – Allgemeine Bemessungsregeln und Bemessungsregeln für den Hochbau, August 2004, Beuth Verlag GmbH. [125] Deutsches Institut für Normung e.V. (DIN). 4102-2. 1977. Brandverhalten von Baustoffen und Bauteilen, Bauteile: Begriffe, Anforderungen und Prüfungen . Beuth Verlag GmbH. [126] European Committee for Standardization (CEN). 2012. Fire resistance tests - Part 1: General Requirements. EN 1363-1:2012, CEN, Brussels, Belgium. [127] Gesetz über die Feuerwehren im Land Berlin (Feuerwehrge setz – FwG), vom 23. September 2003, Senatsverwaltung für Inneres, Berlin. [128] Bard, D., Sjökvist, L-G. 2008. Sound transmission through a complete wood cross junction in a light-weight building. Proceedings of Internoise 2008, October 26–29, Shangaï, China. [129] Bolmsvik, Å. 2008. Evaluation of vibration distribution from a f ull scale measurement in an eight storey wooden house. Proceedings of International Conference on Noise and Vibration Engineering, September 15–17, Heverlee, Belgium, 2008, 593–607. [130] Gustavsson, L., Joelsson, A., Sathre, R. 2009. Life cycle primary energy use and carbon emission of an eight-storey wood-framed apartment building. Energy Build., 42(2): 230–242. [131] Klostermeier, C. 2008. Europameister - “Limnologen”: Achtgeschössige Holzhäuser in Väx jö/Schweden. Bauhandwerk , 7(8): 28–36. [132] Petersson, Å., Holterman, L. 2008. Vibrations in a 7-Storey Wood Building, Report TVSM-5157, Division of Structural Mechanics, Lund University, Lund, Sweden. [133] Rosenkilde, A., Jarnerö, K., A xelson, M. 2008. Flervåningshus med trästomme – Uppföljning av Kv Limnologen och Kv Rya, Rydebäck . Rapport 2008:18, SP Technical Research Institute of Sweden, Borås, Sweden. [134] Serrano, E. 2009. Documentation of the Limnologen project - Overview and summaries of sub projects: Results . Report 56/2009, School of Technology and Design, Väx jö University, Väx jö, Sweden. [135] Serrano E., Enquist B., Vessby J. 2010. Vertical relative displacements in a medium-rise CLT-building, Proceedings of 1st International Conference on Structures and Architecture. CRC Press, London, UK, 388–395. [136] Schauerte, T. 2010. Consumer perceptions on wooden multistory houses: Segmenting international markets. Proceedings of the International Convention of Society of Wood [137] [138]

Science and Technology and United Nations Economic Commission for Europe – Timber Committee, October 11–14, Geneva, Switzerland, paper MA-1, 11. Smith, I. 2006. Reaching for the limits with timber construction, Proceedings of IABSE Symposium Reaching to Tomorrow’s Challenges in Structural Engineering , Budapest, Hungary, September 13–15, 384–385. Swailes, T., Marsh, J. 1998. Structural Appraisal of Iron-Framed Textile Mills . Thomas Telford Ltd., London, UK.

Structural Engineering Documents Objective: To provide in-depth information to practicing stuctural engineers in reports of high scientific and technical standards on a wide range of structural engineering topics.

SED Editorial Board: J. Sobrino, Spain (Chair); H. Subbarao, India (Vice Chair); M. Bakhoum, Egypt; C. Bob, Romania; M. Braestrup, Denmark; M.G. Bruschi, USA; R. Geier, Austria; N.P. Hoej, Switzerland; S. Kite, Hong Kong; D. Laefer, Ireland; R. Mor, Israel; H.H. (Bert) Snijder, The Netherlands; R. von Woelfel, Germany.

Topics: The International Association for Bridge and Structural Engineering (IABSE) operates on a worldwide basis, with interests of all type of structures, in all materials. Its members represent structural engineers, employed in design, academe, construction, regulation and renewal. IABSE organises conferences and publishes the quarterly journal Structural Engineering International (SEI) , as well as reports and monographs, including the SED series, and presents annual awards for achievements in structural engineering. With a membership of some 4,000 individuals in more than 100 countries, IABSE is the international organisation for structural engineering.

Readership: Practicing structural engineers, teachers, researchers and students at a university level, as well as representatives of owners, operators and builders.

Publisher: The International Association for Bridge and Structural Engineering (IABSE) was founded as a non-profit scientific association in 1929. Today it has more than 3900 members in over 90 countries. IABSE’s mission is to promote the exchange of knowledge and to advance the practice of structural engineering worldwide. IABSE organizes conferences and publishes the quarterly journal Structural Engineering International, as well as conference reports and other monographs, including the SED series. IABSE also presents annual awards for achievements in structural engineering.

For further Information: IABSE c/o ETH Zürich CH-8093 Zürich, Switzerland Phone: Int. + 41-44-633 2647 Fax: Int. + 41-44-633 1241 E-mail: [email protected] Web: www.iabse.org

Use of Timber on Multi-Storey Building

Recommend Documents