\
~,
B!P3W SOO.Jod JO S3!JSD038
\ '--
/
.___~
\.....-
lnetltut fraricale-'Pll .. ole publlcatlone ''-
Thierry BOURBI~ lnstitut
Fra~
du P6trole
now with Sc:hlumbetger
'-'
Olivier COUSSY LabcJqtoire Central des Ponts et Chau..._
Bernard ZINSZNER lnstitut
~
du Nlrole
'--' "--'
'-./
''---''
'-~
"'---'
Foreword by
acoustics of porous media
Ama NUR · '-.~
Professor Director of the Sgnford Rock PllysQ Project Geophysics Depertment. Stanford Un~
Translated from the French by Nisaim MARSHALL
'-../
'~
1987
'--'
"
GULF PUBUSHING COMPANY •
BOOK DIVISION • HOUSTON, LONooN. PARIS, TOKYO
EDITIONS TECHNIP 27 RUE GINOUX 75737 PARIS CEDEX 15
'--'
technip
~
__J
/~
--
__.~~
;J-
.. ...-..,. ')
/ ,:-
.
....., '
~
-.,
.
•
~
'""
'" Translation of c Acoustique des milieux poreux • T. ~ 0. eot.y. B.,zirtii&IW'
. C ldkiona T~. P.W1111 .. ~
~
For (apra taka from· 1*\io- ~shed
~·8111hot·~:::.~ted intlefilurec;ap• . ~~ , list(title
~
and patitisher)c:ail be to.iod at the emt.of$be.book.
~
....__
.t"This Edition, 1987 Gulf Publishing Company Book Division Houston, Texas ~,
ISBN 0-87201-025-2 Library of Congress Catalog Cerd No. 88-82913 ---.,
e
1987 Editions Technip, Paris
All rights ..-ved. No Pllrt of this publication may be reproduced or transmitted in any form or by any means. electronic or mechanical, including photocopy, rac:ording. or any infOI'INtion storage and retrieval system. without the prior written permission of the publisher.
Printed in France by lmprimarie Nouvelle. 45800 Saiolt-Jean-de-Braye
"""""~
~-
"
~~
....--
·-·
· · .. fcreWord :_·~.=. -
•' •
'--' '-._.·
"---'
''-·
"-'
·-· '-----' .'-
'-'
'--"
'-
'-,..,...-
•r
!2'·
It has Ion& been recognized that~¥: wayc; ctifr~d~ristics as measured on the earth's surface can provide information not onl~--~bl>.!!UJae1l.ttitude and distribution of interfaces between rock types within the earth, but also about the mineralogy, as well as the state of the rocks present. In fact much of our knowledge about the internal constitution of the earth has been derived from seismic wave characteristics such as velocities and amplitudes. Althoup seismic methods, notably reflection methods in exploration geophysics have · been used most extensively, they were in the past applied mostly to delineate rock interfaces in the earth's shallow crust, to evaluate structures which might bear hydrocarbons. In contrast relatively little use has been made of seismic waves for the determination of the rock properties of direct interest to hydrocarbon recovery (e.g. porosity, permeability), or the direct detection of hydrocarbons. Even in acoustic logiing. only the estimation of porosity from velocities has been developed as a regular service. The estimations· of permeability or saturation are based on other, non seismic, methods. Because of the increasing value of oil, the growing complexity of recently discovered oil ftelds, and the growing realization that reservoirs and recovery are ·more heterogeneous than assumed. in the past, a major shift in the use of seismic methods has taken place during the past one or two decades. One of the central aspects of this shift involves the need to establish and understand the relation between the seismic properties of reservoir and reservoir related rocks, and their production properties (porosity, permeability) and state (mineralogy, saturation, pore pressure etc.). Some obvious applications are the evaluation of stratigraphic traps, fracture detection, and the spatial distribution of porosity and permeability. Seismic methods are almost never used in hydrocarbon recovery assessment, in spite of the growing need to better understand recovery. A major problem which has emerged in the area of reservoir evaluation and production is the realization of the complexity of most reservoirs, leading to great uncertainties in estimated total recovery, recovery rates, · and recovery method. Reservoir complexity is typically related to the signifiCant spatial heterogeneity in porosity, permeability. clay content, fracture density etc. The spatial variabilities cannot be inferred at any level of detail from well testing data, logs, or cores. They may only be obtained, hopefully, from remote geophysical measurement, especially seismic measurement. A direct consequence of the heterogeneous nature of reservoirs is the complexity of their recovery processes, ranging from problems like the migration of the gas cap in reservoirs with discontinuous shales, overpressure zones, and the tracking of steam or temperature in thermal recovery in reservoirs with large spatial variation of permeability.
_lj
VIII
FOREWORD
There is little doubt that seismic methods will play, in the future, a major role in helping to solve production and recovery problems. But we first need a better understanding of what it is that seismic waves can tell us about reservoir rocks, and how to extract the desired information. This book is an important step in this needed direction.
A. NUR
Professor Director of the Stanford Rock Physics Project Geophysics Department, Stanford University July 1986
.~,
-._,
~,
~.~
_, ~
,_ '--'
-~
First of all, we thank the lnstitut Franrais du Petrole ( IFP) for making: this book a possibility. Over and beyond material contribution from IFP, we benefned from the invaluable help of our colleague$ there. "--' "---"
We wish to thank in particular: P. RAsoLOFOSAON who provided us with friendly support and assistance in the writing of Chapter 6.
,j
'----''
':"' --'
''·--"
G. GRAu, M. LAvERGNE and P. TAIUFwho read the manuscript, which improved a great deal due to their constructive criticism. D. BELAUD, M.-T. BIEBER, M. GuEDJ, M. HuaTE, C. JACQUIN, M. MASSOI" and the IFPDocumentation and Publi<:ation Services who have contributed in their different areas of expertise. Editions Technip is responsible for the particularly careful presentation of the book. We must also not forget B. HALPHEN of the Laboratoire Centra/des Ponts et Chaussees who encouraged us, along with M. PANETand J.-P. PoiiUER, respectively of Simecso/ and the lnstitut de Physique du Globe de Pmis who kindly wrote the foreword to the French edition.
-~
D. GoLDBERG of Lamont Doherty Geological Observatory reread the English translation at great length, and his remarks enabled us to clarify many passages.
\,_/
\._;~·
~-
~
_.'
'-~
'--.--
A. Nua from Stanford University, to whom rock physics owes so much, kindly agreed to write the foreword for the English edition. We wish to thank him here most warmly.
I
~~
'-.-~
'-...-
contents -~
Foreword .•....•.•..•.••.....•.........•.•..•........................
'--.·
NOMENCLATURE ................................................. . GENERAL INTRODUCfiON . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . • . . . .
VII
5
Chapter l POROUS MEDIA '---"
lntrocluction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9
'---'
l.l. POI'Oiity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1.1. 111e . , • • , . , . or pon1111ty • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • •
1.1.1 Dellllldoa ole•••EUII_. 61ca•eued porGiides • • • • • • • • • • • • • • • • . • • • • • • • • 1.1.3. 8,edftc c-.: liiU ..!nlll ...... clay ponllity • . • • • • • • • • • • • • • • • • • • • • • • • • • •
1.2. 'l'1le pore space: aUcroseopk pometric analysis ....................... · 1.11. MediMI for 1i1 , .... die pore tpaee • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • '~
'---'
-~
1.11 Simple eXlmqlles or pore l!l•etrles .................................. . 1.111. lntcrpuular space in a packina of identical spheres •••••••••••••••• 1.111 Ideal VUIJY medium •••••••••••••••••.••..••••.•••••.••••.• 1.2.3. Act.al pore 1p11ca ••••••••••••••••••••••••••••••••••••••••••••••• 1.13.1. Choquette aDd Pray classifiCatiOn ••••••.•••.••••••••••.••••••.• 1.13.1. Application of mathematical morphology to the description of poi'OJIS media 1.13.3. SpecifiC case of crack and fracture pore spaces •...•.••.•..•..•....•
1.3. Tbe pore space: capiDary approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3.1. Capillary e~~ulllbria: ................ deflllitro. • . . . . . . • . . . . . . . . . . . . . . 1.3.1 Caplbry ~ c:.nes • • • • • • • • • • • • • • • • • • . • . • • • . • • • • • • • • • • • • • • . . 1.3.3. Afplic:lldouf C8flllbry ~ c:.nes to ,......_, . . • . . • . • . . • . . . • • • • . . . 1.3.4. ..... 4111billadoll at ... ~ Kale • • • • . • • • • • • • • • • • • • • • • . . • • • • • • . 1.3.4.1. Principles of .the visualization of fluids in capillary equilibria • . . • • • • . . . 1.3.4.2. Application to the aeometric description of pore networks • • . . • • • • • . .
. •
. • . .
I0
1o II II 12 12 13 13 15 15 15 18 21 22 22
24 26 29 29 30
···-'
CONTENTS
XII
Concept of relative permeability . . . . • . . . . . . . . • . . . . . . . . . . . . . . . . .
30 31 31 33 35 38 38
1.5. Problems of scale in porous media .................................. .
40
1.5.1 Defmition of minimiHII ~tion vol- •.•••.............•.....••..
40
1.5.2. MinimiHilltomogeaizadoa Yo111111e aad ph)'licalllellanor •••••.••.•••••••••••••
42
1.6. Example of a natural porous medium: Fontainebleau sandstone ......... .
43 43 43 43
1.4. Fluid flow in porous media 1.4.1. Cae of a single ftuid totaDy saturating a pore space: absolute or siagle-phase permeability 1.4.1.1. 1.4.1.2. 1.4.1.3.
Defmition, units and measurements ••••.••.••.•...........•..•. Characteristics of the pore space affecting permeability ...........••.. Porosity/permeability relationship in intergran.ular spares . . . . . . . . . . . . .
1.4.2. Multiphae ftows . . . . . . . . . . . . • . • • • • . • • • • . . . . . . . . . . . . . . . . . . . . . . . . . 1.4.2.1.
1.6.1. SoU4 skeleton ••.......•••......••.•.•••••...••...•••••••••.•••• 1.6.2. Pore space •......•...•.......•••.•.••.•..••....•...••..••...•. 1.6.2.1. 1.6.2.2. 1.6.2.3.
Geometric characteristics ...•....•...•....•............••... Porosity/permeability relationship . . • . . . . . . . . . . . . . . . . . . . • . . • • . . Total porosity trapped porosity relationship ..........•.......••..
•
~
44
47
Chapter 2
WAVE PROPAGATION IN SATURATED POROUS MEDIA Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
49
l.t. Review of elastodynamics . . . . . . . . . . . . . . . . . . • . . . . . . • • . . . . • . . . . . • . . . . 2.1.1. Straia teosor . . . . . . . . . . . • . . . . . . . . • • . . . . . . . . . . . • . . . . . . . . . . . . . . . . .
49 49
2.1.2. Stress teosor aad eqtdlibrl.m e.aaadolls . • . . . . • • . . . . • • . . . . . . . . • . . . . . . • • . .
52
u- elasddty .................. : . . . . . . . . . . . . . . . . . 2.1.4. u . . elasticity aac1 rock ..c~aaa~c:s. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
54 56 56
2.1.3. CollldtutiTe law of 2.1.4.1. 2.1.4.2.
linear elasticity . • • • . . . . . . . • . • . • . . . . • • . . • . . . . • . . . . • . . • . . . . Rock mechanics and effective moduli . . . . . . . . . . . . . . • • . . . . . . . . . . .
2.1.5. Wave propagation in an isotropic linear elasde mediiHII . . . . . . . . • . . • . . . . . . . . . . 2.1.5.1. 2.1.~.2.
Waves in 3D space . • • . • • . . . • . • • . . • . . . • . . . . . . . . . . . . . . . . . . . I D wave equation (elastic case) . . . . . . . . . . . • . . . . . . . . . . . . . . . . . . .
2.2. Wave propagation in saturated porous media: Biot's theory . . . . . . . . . . . . . 2.2.1. Assumptions . . . . . . . . . . . • . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.2. Equations of movement . . . . . • . . . . . . . . . . . . . . . . . . • . • . . . . . . . . . . . • . • . . • 2.2.2.1. 2.2.2.2. 2.2.2.3. 2.2.2.4. 2.2.2.5.
Strain potential and stresses . . . • . • • • . . • . . Gasmann's equation and Biot's theory • • . . . . Dissipation pseudo-potential . • . • . . . . • . . . . Kinetic energy . . • • • . . . • . . • . • • • • • • • • . . Equations of lliOvement . • • . • . . . • • . • • • • .
••. ••. ... •. . .•.
•. .. .. .. ..
. . . . .
.. .. .. .• •.
. •.•. . •. •. . •... ••••. ••. . .
. . . •. . . . . •. . . . . .•. ••••. . .. .•..
. . . . .
58 59 59 61
63 64 66 66 67 69
70 70
~
--------------------------·--
CON1'Pfi'S
XIII
QuaJitati,·e aspects of Biot's model. • • • • • • • • • • • • • • • • . . . • • • • • • • • • Experimental results • • . . • • . • • • • . . • • . • • . . . • • • . . • . . . • • . • . . . • .
72 73 74 81 84 85 85 88
2.3. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
94
llledium...........
95
l.%.3. Wa.-e pi'OIIIIIlldoe • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • l.%.3.1. :Z.Ul.
Existence of a slow P wave . • • • • • • • . • • • • • • • • • • • • • • . • • • • • • • • • •
.....................
Waw velocities and attenuations • • • • . • • • . • . . • • . . • • • • • • • • • • • • . •
l.l.4. Blot's......,. ... TftDIIIII'IIaw . . • • . • • • • • • . . . • • . • • • • • • • • • • . . • • • • • . • •
2.2.5.
~t.......
1.1.6. Expetl•e•hll wrif-cioe . . . . . . . . . . . • • . • • . . . . . • • . . • . . • . . . . . • • • . • • • . 1.1.6.1. 1.1.6.2.
Appenclix 2.1. Wave propagation ia a aon-ilotropie elasdc:
Olapter 3 WAVE PROPAGATION AND VIBRATION EFFECI'S
IN VISCOELASTIC MEDIA (IIIIWimeasioaal)
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
99
3.1. Delayed behavior of materials . . . . . . . . . . . . . . . . . . . • . . . . . . . . . . . . . . . . . . 3.1.1. Relaxlltloa tiiCI •.••••••••••••••••••• ·••.•••••••••••••..•••••.•.••
100 100 101
3.l. Unear Yilc:oelutic behavior ......•: . . . . . . . . . . . . . . . . . • • . . . . . • • . • . . . .
102
3.1.1. Creep t11C1 • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • •
·3.3. Dynamics of Ulli4imensional liaear Yitcoelastk media, fust coacepts of the quality factor Q . . . . . . . . . . . . . . . . . . . . . . . . . • . . . . . . . . . • • . . . • . . . . . . . . • 3.3.1. eo.,lex....... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.1. Han11011k pro~~~ea.
3.3.3.
. .... .... .. ..................... ....... .......
w~ propaaadoa . • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • •
3.3.3.1. 3.3.3.2.
Wave propagation and attenuation . ~.. . . • . . • . • • . . . . . . • • . . • . . . • Variation in time of a free wave packet: group velocity and phase velocity .
3.3.4. Qwality fader .........,. IIOtleM • • • • • • • • • • . • • • • • • • • • • • • • • • • • • • • • • • • • 3.3.5. EU!Dfle of ~or tltcoelllldc .......
to,....._.. ................
3.4. Important viseoelastk models .. .. . . . .. . . . • . .. . .. . . . . • . . . . . . . . . . . . . . . 3.4.1. Deolollcalaodels.. . . de(~ ... factor ......... :. . • . .. • • .. •
die.....,.
3.4.1. M«MMeel ..,_. from die fl8llty r.eter .......•... ; . . . . . • • . . . . • • • . . . • • • . 3.4.1.1. 3.4.1.2.
NCQ model (Nearly Ceustant Q) . . • . • • • • • • . • • . • • • • . . • • • • . • . • . CQ model (Constant Q) •••.•••.•••••••••••••• ·• • . • . • • • • • • • • •
104 104 lOS 107 107
108 112 114
ll7 117 123 123 124
CONTENTS
XIV
3.5. Vibrations in viscoelastic media . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . •. . • • • . • • • • •, • • • • . •
128 128 132 132 134 13 5
3.6. Conclusions concerning the quaUty factor and fmal remarks . . . . . . . . . . . . .
139
3.7. Conclusions..... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
141
Appendix 3.1.
142
3.5.1. TraveUiag waves 111111 ribrations . . . . . . . . . • . 3.5.2. Modal analysis • • . . . . . • . . . . . • . . . . . • • . 3.5.2.1. Nonnal modes . . . • . . . • . . • • . • . 3.5.1.1. Foroed vibrations and free vibrations 3.5.3. Defmitions of die quaUty factor 1l'5iag vibrations
•. . . . •. ••. .•••••• .••••••. •• •. . . . • . . . . . . . . . . •. •. ••••. ••
. .. . . ••. . . . . ••. •. ••. •••••. • . . .. •. . ••••••• . •. ••••••••••
The Kramers-Kronig relations.............................
Chapter 4
EXPERIMENTAL TECHNIQUFS FOR MEASURING VELOCmES AND ATTENUATIONS Introduction . . . • . . . • . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
145
4.1. Measurements using waTe propagation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
146 146 148 148 151
4.1.1. Diffac:tddes • . . . . • . . • • . • • . . . . • . • • . • • . . • . . 4.1.2. Measurement prindples 111111 experimental techniques . 4.1.2.1. Velocity measurements . . • . . . . • . . . . . . 4.1.1.1. Attenuation measurements . . . . . . . . . . .
. . . . . . . . . . ••••. . . . ••••.
. . . . . . . . . . . •. . . . . . . . . . . . . . . .. .. .. . . .•. . . . . . .. . •. . . . . ... . . .. •.. •. . •. . •
4.2. Measurements using vibrating systems (standing waves) . . . . . . . . . . . . . . . . .
Pendulums .•••.•....••• Resonant bar . . . . . . . . . . . . . . . . . . . . . . . • . . . . . . . . • . . . . . . . . . •
161 161 161 163 163 163
4.3. Methods using stress/strain ctmes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
170
4.4. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
171
4.2.1. Difl:.Culties • • • • • • . . . • • • • • • . • • • • . . • . • . . . • • . • . • . . . • • • . • • . • . . . • • • • 4.2.2. Geueral priadples • . • . • . . • . . . . • • . . . . • . . . . . . . . . . • . . . . • . • • • • • . • . • . • 4.1.3. Experimeatal setaps . . . . . . . . . . . . . . . . . . ~ . . . . . . . . . . . . . . . . • . . . • . . . . . 4.2.3.1.
4.2.3.2.
!....................... ; . . . . . . .
Oaapter 5
WAVE PROPAGATION IN POROUS MEDIA RFSULTS AND MECHANISMS Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
175
5.1. Results and mechanisms in the laboratory.............................
175 176 176
5.1.1. Velocities: results 111111 mecbanisms . . • • • • • • • . • . • • • • . • • . • • • . • • • . • • • • • • . . 5.1.1.1. Velocities and pressures . . . • . . . . . • • . • • . • . . . . . . . . . . • • • . • • • • . .
'-/
'XV
CONTI!N1S 5.1.1~
-~
5.1.1.3. 5.1.1.4. 5.1.1.5. 5.1.1.6. 5.1.2.
Velocities and aturations ..••••••.......•••.••.•.••• , ••••• , , Velocities and temperatures ••••••.••••••••••••••••••••••••••• Velocities aod l'roqueacy •••••••••••••••••••••••••••••••••••• Velocities and strains •••..•••.•......•....•••••••••••••••.. Summary •••••••••••••••••••• • • ••• • •• • • • • • • • • • • • • · • • • · ·
Ac.-...._: l't8lllts 5.1~1.
•..•••.••..•••••....•••••....••...•..•••....•
5.1.2.5.
Attenuations and pn:ssures .....•....•.........•...••.•...... Attenuations and saturations •..•...••....•...•.....•.•....... Attenuations and temperatures •....•..••...••••. , .•...•...... Attenuations and frequency •.•.••...•.•....•••••.••••••...••. Attenuations and strains ••••••••••••••••••••••••••••••••••••
5.1~
Summary •••••••••••••••••••• • •••• • •• • • • • • • • • • • • • • • • • • •
5.1.2.2. 5.1.2.3. 5.1~4.
5.1.3. A"rm .-.: ..._.. 5.1.3.1. 5.1.3.2. 5.1.3.3.
lnterJranular friction .••.••••••••••••••••••• , •••••••.•.•••. Attenuation mechanisms in "dry" and very sliJbtly saturated f9Cb ..... . Attenuation mechanisms in partiaUy or fully saturated rocks •••••..••••
5.2. -Results aad medaaaisms coacerning in litu measurements ...... _......... . 5.2.1. lldrodtlcdoa ••••••••••••.•••••••••.••••••.••..•••...•••••....•.
'----
•
~
•
5.2.2. Vllodtt. ••••••.. • •••. · • · • • • · · · • • · · · · · • • · • • • · · · • • • · · • • · · · · · · · · 5.2.2.1. In situ velocity measurements ••••.••••...•.••••••••••••••••.• 5.2.2.2. Velocities and porosity .•..••••........•....•.......•....... 5.2.1.3. Velocities and density .•......•............................. 5.2.2.4. Velocities and clay content ..•..........•....•............... 5.2.2.5• Velocities and'compection •..•.•••...•..•.....•....••.•....•• 5.2.2.6. V,./f/'11 and Poisson's ratio ....................•....•......... 5.2.17. Summary on in situ velocity measurements ••....•.........•.....• 5.2.3. AUraa .._ ••......•.•...•. · · · . • · · • · • • • • • • • • • • • • • · • · • • · • • .- • • • • 5.2.3.1. In situ atteouation tneaSUn:DlCilts • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 5.2.3.2. Results • • • • • • • • • • • • • • • • ' ~ • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 5~4.
Caacl d - • Ia dta
•r-r••• ................................ -.. .
187 193 198 198 202 202 204 204
207 211
215 215 217 218 218 220 229 229 230 230 230
233 233 237 239 239 240 240
242 242
~-
'-
Chapter 6
WAVF.S AND INTERFACES -----
Iatroclucdon ........................................................ . 6.1. Wave propqatlon in saturated porous media. Discontinuity effects ...... .
245
6.1.1• ........,.~ •••.•••..•.....•..•..•.........•....••••...•••
246
6.1~
'-
245
w- ~at die lilted- or two..,.....,._._.. 6.1~1~
Calc of normal incidence •••••••..•••••••••••••.••••••••••••
249 249
6.1~2.
Analysis of n:flection on the fn:e surface of a semi-infmite saturated porous medium •..•••••.••..•••••...•••.•...•••.•...••.....•..
258
......•......•..
.- .------'
CONTENTS
XVI
6.1.3. lllterfaee problems betweea ~..._.tell me41ia. Aptlladoa to aco8ltic logiq . . .
Interface waves • • . • • • • • • • • • • • • . • • • • • • • • • • • • • . • . • . . • . • • • . • Seismic source in the vicinity of a fluid/porous medium interface . . • • . . . • Conclusions . . . . . . . • . • • . • • . . . . . . . • • •. . • . . . . . . • • . . . . . . . . . .
263 263 269 281
6.l. Retledion and transmission in l'iscoelasticity . . . . . . . . . . . . . . . . . . . . . . . . . .
283
6.1.1. Wa.e equation in Yiscoelastic media • • • • • • • • • • • • • • • • • • • • • • • • • • • . . . • • • • •
283 287 287 288 288
6.1.3.1. 6.1.3.1. .6.1.3.3.
6.2.2.
Eaef1Y balaDce and quality factor 6.2.2.1. 6.2.2.1. 6.1.2.3.
....••.•.•...•...•...•••••........ ·. .
Energy balance • • . . . • • • • • • • • • • • . . . . • • • • . • • • • • • • • . . . • . • . • . Quality factor • . . • • • • • • • • • • • . • • . • • • • • • • • • • • • • • • . . . . . . • • . . Constant Q model in two dimenlions • • • • • • • • • • • • • • • • • • . . . • • • • . . Theory . • • • • • . • • • • • • • . • • • • . . • . • • • • . • • • . . • • • • • • . . . . . • . • • Interface effect of attenuation • • . . • • • • • • • • • • • • • • • • • • . • . . • • • • • .
290 290 29 5
6.1.<1. lllterfaee waves Ia 'rilcoelaldC Bllllla •••••••••••••.•••••••••..••..•.• ·• • .
297
6.3. General conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
298
6.2.3. Reftecdoeuad .........._ .. two ••uni- • . . . . . . . • • • . . . . • . . . . . . . . . . . 6.1.3.1. 6.2.3.1.
Chapter 7
SOME APPLICATIONS IN PETROLEUM GEOPHYSICS Introduction .........................•.......•........ ·. . .. . . . . . . . . . . . .
299
7.1. Low frequency seismie prospecting. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
302
7.1.1. ~ • . . . . . . . . . • . . . . . . . . • . . • . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
302
7.1.2. CoBYeetioaal •ilmic: ~ ••••••••..•.••.•... 7.1.1.1. Combined use of P and S waves . . . . . • • . . . . . 7.1.2.1. Signal analysis . . . . . . . • . • . • • . . . . . . . . . . . . 7.1.1.3. Three-component studies • • • • • • • . . • • . . • • . .
.•.• . . •. .... ••.•
302
7.1.3. a-,.o1r selsmics . . . . . . . . . . . . . . . . . . . . . . . • . . . . . . . . . . . . . . . . . . . . . . .
308 309
7.1.3.1. 7.1.3.2.
' . .. •. •.
••. ... ... ••.
. . . .
••. ... ... •..
•. .. .. ..
. . . .
.. .. .. •.
Variation in fluid phase or pore pressure . . . . . • • . . . . • • • . . . . . . . • . . Analysis of fractured zones • • • • • • . • . . . . . . . . . . . . . . . . . . . . . . . . . .
303 303 306
310
loggiD&.....................................
310
BmLIOG·RAPHY........................................ . . . . . . . . . . . .
315
AUTHOR INDEX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
325
SUBJECT INDEX. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
329
7.2. Full wavefonn acoustic
'"' ~------~---------------------------------------------------------------------------~----- - - - - - - - ·
',..,_/
nomenclature A subscript (letter or number) after a comma denotes a partial derivative with respect to the coordinate related to this subscript:
c2 uJC ...
iJu" ""·' ,., iJy
--~2
""·"- cy
au, 2 "'· ..
OXz ·•·
Time derivatives arc denoted by a dot (ftrst derivative) or two dots (second derivative): .
iJu iJt iJ2u
u==-
ii=-
iJtl
Vectors are represented by bold characters. · The notations div, and. curl and V 2 indicate the divergence, gradient, curl and Laplacian operators, namely in Cartesian coordinates: div •
"-
= iJcjiJC + iJcjl, + ocjl, iJx
oy
i}z
• +- (ac~~ ac~~ o+) ox • oy ' cz
an curl
+_(c+. _a+, cy
2
ac~~"
iJz ' oz
iJ2cll
_ac~~. o+, _i+,.) ex • OX
ol+
iy .
ale~~
v • = iJx2 + oy2 + iJzl '--
The real and imaginary parts of a complex quantity are indicated by: Real part =( Imapnary part ... (
)a or Re (
h or Im (
The system used is the Sl system. with a few exceptions (i.e. permeabilities).
The nomenclature below does not include the multiple constants used in the text. These are generally represented by the characters A, B, C, y etc.
A a ~
Co
attenuation vector. tortuosity (Biot theory) (see also t(c/>)). torsion constant. hydraulic dift'usivity.
c.
clay content.
c.(h)
Covariance function in the u direction for a length h.
c,ft,
elastic constant tensor.
i!
t
I '~
---"
2
NOMENCLATURE
c d, 0
o. D
e E F(Q>l
f
!c fo f(t) G
J ,.(x) " % %
kinetic energy. skin depth. dissipation potential. surface dissipation potential. thermal diffusivity. aspect ratio. Young's modulus. formation factor. frequency. critical Biot frequency. scattering central frequency. creep function. amplitude coefficient including geometric divergence effects. Bessel functions of the ftrst kind (n'h order). permeability (in mD). hydraulic permeability. permeability tensor.
relative permeability to water as a function of saturation. ,... ~-(S,.) relative permeability to non-wetting fluid as a function of saturation. K, interface hydraulic permeability. k wave number. wave vec~or. k k* complex wave number. K"'(S,.)
k* K
Ko
K, Kfl
K, l
L M D
p
PeG
P«c Pt
complex wave vector. bulk modulus. open (or dry) bulk modulus (Biot theory). closed (or saturated bulk modulus (Biot theory). fluid bulk modulus (Biot theory). skeleton bulk modulus (Biot theory). Lagrangian. length of sample concerned. elastic modulus (in general) or specific elastic modulus used in Biot's theory. normal vector. pressure. capillary pressure. access pressure. limit capillary pressure.
Pelf
Pc
p, p P,
p2 Q Q,o
Qs QE
Q.ll Qs, Qror Qbiph
Q.~
0 R; R,. Rh
R, IR r(t)
s SH
SV
s.. s.... s. ff T lr t, t u
u
-r. "r." ·f~
effective pressure. confming pressure. pore pressure. compressional wave. compressional wave of the f1rst kind. compressional wave of the second kind. · quality factor. P wave quality factor. S wave quality factor. extensional wave quality factor. Rayleigh wave quality factor. Stoneley wave quality factor. global quality factor. quality factor related to interface fluid flow. intrinsic quality factor (viscoelasticity). volumetric flow rate. radius of curvature (i = 1,2). mean radius of curvature. hydraulic radius. radius of gyration. reflection coefficient. relaxation function. surface area. shear wave polarized in the horizontal plane. shear wave polarized in the vertical plane. water saturation. irreducible water saturation. residual saturation. age of sediment. period. transmission coefficient. surface tension. tangent vector. displacement. mean displacement of continuous liquid phase. pore volume. total sample volume. solid volume.
-----------~--~~-~-~~-~-------------
----~-
--~···~
-
-------------------------------------------------------------------------------------~-
NOMENCLATl"RE
;!
f~
~
\1 V-..
V.,. V
v,. v,., J's VR Ys. VE
V,
v. V1
V.
w ~ !
z 2
r b(r)
b b
•
.
bii .II W
strain potential. maximum stored energy during a cycle. average stored energy during a cycle. velocity. compressional wave velocity or P wave velocity. velocity of compressional waves of the second kind (or slow waves). shear wave velocity or S wave velocity. Rayleigh wave velocity. Stoneley wave velocity. extensional wave velocity. group velocity. phase velocity. velocity in saturating fluid. velocity in rock matrix. fdtration velocity vector. depth. acoustic impedance. attenuation. gamma function. Dirac distribution. logarithmic decrement. logarithmic decrement at pulsation
£ t;j
, ~
(I
e I.
A Jl \'
.; .;,. p p,
PJ (I (1i_:
Cl,
r rtol
r cp cj! 4J
co•.
'I'
Kronecker delta. dissipated energy.
(:)
{}
3 strain. strain tensor. angle. viscosity. angle. volumetric strain. Lame's parameter. wavelength. shear modulus. Poisson's ratio. fluid content reduced damping at pulsation co., . density. matrix density. fluid density. stress. stress tensor. pure shear. rise time. tortuosity of current lines. specifiC function used in Biot's theory. phase. porosity. scalar potential. vector potential. angular frequency. integration volume.
---
-.---
____,
'~J
f
general Introduction
With a sharp blow ofhis hammer, he struck the smooth granite and almost immediately pressed his ear against the surface warmed by the sun, whose rays had shone on this very spot. He heard the sound diverge mtd scauer in far-flung reverberations mtd realized that tltere he would discover great things.
Boris Vian (Autumn in Peking)
The classic laws of elastodynamics and their extension to viscoelastic behavior are generally postulated for homogeneous and continuous materials. Their application to porous media requires a sweeping adaptation. Porous media are, by their very essence, composite and multiphase. Composite because the solid fraction - the skeleton -. is formed of grains whose chemical or crystalline features are often different and multiphase because this solid fractiort is always ;~ssociated with a gas or liquid phase that occupies the voids between the grains. This microscopic heterogeneity of the porous medium induces a complex macroscopic physical behavior sensitive to slight variations in fluid content or of the solid structure. The acoustics of porous media are intended to characterize their behavior by synthesizing between the.rigor of the laws of mechanics and the natural disorder of porous media. The task is difficult, but this difficulty is correlated to the importance of the applications to which even partial results can lead. In fact, acoustic analysis is one of the surest means available for the remote investigation of porous rocks and for the nondestructive testing of materials. The ideal fteld of application - at least for its economic benefits - is geophysics. bt the subdiscipline of acoustics, classic seismics. which predominates, was originally intended to provide a geometric image of the subsurface, by using the reflection or refraction of acoustic waves from discontinuities at the boundaries of geologic strata. This procedure was eminently successful. The prodigious groWth of electronics in the recording of signals and the computation facilities for their processing now serve to routinely obtain highquality geometric images of the subsurface. This result, which stems from the substantial improvement in the signal/noise ratio, is essentially achieved by the measurement of
6
GEl'ERAL
I~TRODL"CIIO"'
different transit times, interpreted in the light of elastodynamics. The acoustics of porous media as such is never an inherent part of this study. However, beyond the very success of these purely geometric methods, the need has gradually emerged to characterize the types of material the acoustic wave has traversed and even their fluid content. The lithologic and petrophysical approach to the subsurface by seismic methods, the prospector's old dream. is unfeasible today without a better knowledge of the acoustics of porous media. A second major field of applied geophysics is that oi acoustic recordings in oil wells, in other words seismic logs and well seismics. In this case. the need for familiarity with the laws governing the interaction of mechanical waves \\ith the porous medium is obvious. There is no better proof ofthis than the proliferation of experimental work on this subject. While petroleum geophysics, due to its economic importance, appears to monopolize the faeld of potential applications, one should not minimize the faeld of nondestructive testing of porous materials (concretes. vuggy plastics etc.). an area in which the applications are the most immediate, because measurements in conditions approaching laboratory conditions are naturally easier to perform. Owing to their composite and multiphase character, the acoustics of porous media constitutes the hub of several disciplines, and this is materially reflected by the three authors, whose specialties are physics, mechanics and geology. Chapter 1 is devoted to the description and visualization of the porous medium itself. One example of a natural porous medium, Fontainebleau sandstone, is discussed at the end of the Chapter.
The simplest multi phase case corresponds to the porous medium totally saturated with fluid. Chapter 2 provides a theoretical examination of wave propagation in this type of medium, differentiating between movements within the solid matrix and those of the saturating fluid. An experimental justifacation of this theory is given at the end of the Chapter. To distinguish between fluid and solid movements is to privilege global phenomena (at the scale of the sample) as opposed to local phenomena (at the scale of the pore and grain). One way to globally identify local phenomena consists of considering a homogeneous medium, equivalent to the porous medium observed in terms of mechanical behavior. From this viewpoint, Chapter 3 attempts a theoretical development of wave propagation in viscoelastic media, as this type of medium fairly closely simulates the behavior of porous media. The concept of quality factor is introduced with the use of the main viscoelastic models. Since velocities, attenuations and damping are introduced conceptually, it is necessary to defme them more specifacally. Chapter 4 examines a number of vital defmitions and their interrelationships. This Chapter describes the main experiments designed to measure velocities and attenuations in porous media. Chapter 5 presents the results given in the literature concerning various laboratory experiments on porous media. The dependence ofvelo-:ities and attenuations on different
GE~ERAL INTRODUCTIO~
7
physical parameters (pressur~ temperature, frequency) is examined in detail. together with their most plausible theoretical interpretation. A number of empirical formulas are given for fteld measurements, helping to identify the various types of formations traversed by the acoustic wave. Thus far, none of the foregoing Chapters has dealt with the prot-lem of interfaces defmed as the places of contrast of acoustic properties (elastic or anelastin Chapter 6 summarizes the theories of Chapters 2 and 3, and applies them to the problen of interfaces. The effect of permeability on volume and surface waves and the effect of attenuation contrast are examined in detail. And fmally, the fteld engineer's viewpoint is discussed in the last Chapter. A number of applications in petroleum geophysics are considered. Emphasis is placed on the difftculties currently encountered in fteld measurements and interpretation of the acoustic properties of porous media.
~
-----------------------------
---------
·--'--
1 porous media
INTRODUCTION Porous bodies are aggregates of solid elements (grains, matrix etc.) between which the voids form the pore space itself. These voids within the porous body give rise to the wide differences in physical behavior between dense solids (such as minerals) and porous substances, which are complicated assemblages in which the presence of a fluid, even in very small amounts, adds to the overall complexity. The ratio of void volume to total volume of the sample is called the porosity. This petrophysical value is usually easy to defme and to measure. However, the process become.s far more complicated when one attempts to make a geometric description of the pore space. While a few specific cases (such as the pore space existing in packed spheres of the.same diameter) lend themselves easily to quantitative description, most actual pore spaces are too complex to be dealt with in a strictly geometric manner, and only relative descriptions are feasible. One mainly tries to highlight how the medium investigated differs from the spaces most routinely observed. For instance, it is important to draw a clear distinction in the medium investigated between the pores themselves (namely the void volumes which store fluids and allow them to flow) and the grain boundaries or microcracks, which are mainly surfaces marking discontinuities between the ·solid elements, and which play a vital role in the mechanical properties. Faced with the relative failure of a strictly geometric description, progress is possible by using a physical phenomenon, the capillary equilibria in the pore spaces. The analysis of these equilibria helps to treat the pore space with a series of simplifications that set the stage for a more systematic quantification. Conversely, these capillary equilibria are, in themselves, an important factor in the study of porous bodies. In fact, if two or more fluids co-exist in a porous medium (as in petroleum reservoirs for water, oil and gas; and in the uppermost layers of the ground for water and air), these capillary effects become decisive for physical behavior. One of the major characteristics of most porous Ndies is to allow fluids to flow. We shall review the concepts of absolute permeability (porous media totally saturated by a single fluid) and relative permeability (porous media containing two or more immiscible fluids). We shall also show how pore spaces are liable to raise problems of scale. An absolute apparent scale of the porous medium does exist, that of the solid grain forming the
10
POROUS
~fEDL\
1
skeleton. For a better understanding, however. it is often necessary to replace this absolute scale by the concept of a scale relative to the physical processes examined. and this relative scale may be quite different from one process to another.
1.1
POROSITY
The measurement of porosity
1.1.1
By defmition, porosity is the ratio of the pore volume (1"-,.) to the total volume(1~) of the body considered. The solid volume (1 ~)is given by 'I~= 1; - 1';.. Hence it suffices to measure two of these three parameters to calculate porosity. The most common measuring methods are summarized in Table 1.1. Methods (ll and (4), (l) and (5), and (3) and (5) are generally used, and sometimes (l) and (7) or (3) and (7). TABLE
1.1
MAIN MErnOOS FOR MEASURING POROSin·
Measured volume
Total volume
Text reference
Measuring method
(I)
Buoyancy in mercury: Mercury, the non-wetting liquid, does not penetrate without pressure into commonly encountered porous media. This method provides aver~· accurate measurement of the total volume
(~)
Direct sample measurement of the d!(f~rent lengths: This method is only suitable for test specimens with very regular shapes
'Y,
Pore volume 'Y,
Solid \"olume
(3)
(4)
Compressibility of a perfect gas: A plot is made of the pressure vs. volume injected in a container that is ftrst empty and then contains the sample. The difference is used to calculate the volume of solid,-whose compressibility is ignored
(5)
Buoyancy in a wetting fluid totally sawrating the porous body: Solid volume is measured directly from the difference between the dry and immersed weight
(6)
Measurement of solid density: After fme grinding of the porous substance
(7)
Calculation of the solid density: By quantitative analysis of the constituent minerals
·r•
~·-·-
Setting of wetting fluid by total saturation under vacuum: Pore volume is obtained directly from the difference in dry and saturated weight
---
- -
I
ti
POROUS MEDIA
II
Monicard (1965) and Dullien (1979) furnish details about these measuring methods. In actual fact, these methods are not equivalent. While the determination of total volume does not raise a theoretical problem (provided that the size of the sample is very large in comparison with the pore size), the same cannot be said of the determination of the other two parameters which are closely linked. Techniques (3), (4) and (5) take account only of the pores connected to the exterior, while methods (6) partly and (7) entirely account for all the voids. -
1.1.2 Defmition of connected and disconnected porosities Depending on the method employed, one can therefore measure the connected porosity, namely the void volume connected to the exterior, or the total porosity, namely the void volume connected to the exterior or not. The difference between these two represents the disconnected porosity. Disconnected porosity is rarely found in most natural porous media. The clearest example of disconnected porosity is given by fluid inclusions in crystals. This disconnected ·porosity is often negligible in sedimentary rocks. In crystalline rocks with very low porosity, fluid inclusions may account for a large fraction oftotaJ porosity. In lavas with vuggy textures, the fraction of disconnected porosity is often smaller than normally believed. The vugs are interconnected by very fme channels or cracks, But, even in the case of pumice stone, a large part of the porosity is clearly connected with the exterior, but through channels that are so narrow that the air is trapped inside the vugs when the sample is immersed in water. This poorly connected porosity is called trapped porosity and will be examined in Section 1.3.2. Generalizations are more difficult in the case of artificial materials. Porous bodies produced by sintering (glass, alumina, steel etc.) sometimes exhibit a substantial fraction of disconnected porosity (up to.9% of the total volume in sintered nickel, for example). In conclusion, in many types of porous media, disconnected porosity is slight but cannot be ignored. Above all, one should avoid the confusion that may be engendered by more or less subjective terms such as effective porosity or dead:-end porosity.
1.1.3 Specif1c cases: incoherent media, clay porosity The measure of the foregoing quantities presume that, before any measurement, the connected porosity of the media examined has been emptied of the liquids it contained, and that this elimination (generally by heating) has beell carried out without disturbing the structure. This is often true of coherent media, but signifrcant exceptions exist. Incoherent media (soils, muds, recent sediments etc.) only acquire their structural properties by the presence of water. This means that (if the concept of porosity is fully retained) measurements and, above all techniques for the observation of the pore space will raise different,problems from those that we are developing here for coherent media. ·. If one simplifies to the extreme, clayey and shaly media constitute an intermediate case. They contain water in three main forms. These different types of water can be investigated
12
POROUS MEDIA
1
by thermogravimetric analysis, which consists in measuring the loss in weight of a sample by heating (on this subject, see Grim, 1953). These threeforms of water are the following: (I) Water located between the particles or clay clusters. This water occupies the pore space in the usual sense. It is eliminated by drying at low temperature (below 70"C). (21 Water located between crystalline folia (illite, montmorillonite) or in tubes existing within certain crystals (such as sepiolites). The expulsion of this water requires a defmite energy. and heating to temperatures approaching 1OO"C is necessary. Above all. this form of dehydration displays a degree of irreversibility depending on the type of mineral concerned. (3) Formation water of the crystals, which is only released by their destruction at temperatures above 200"C, and which therefore in no way corresponds to a notion of pore space. On the other hand, if one considers types (1) and (2), it is clear that the measurement of the porosity of a clayey medium may depend in some cases on the dehydration temperature. As a rule, clays are not dehydrated above 70"C. Heating above this temperature involves interfoliate water, at the risk of irreversibly altering the structure. This provides an example of the problems generally raised by the petrophysical analysis of clay bodies. The state and hence the properties of the sample may depend closely on the history of dehydration that it has undergone.
1.2 THE PORE SPACE: MICROSCOPIC GEOMETRIC ANALYSIS 1.2.1
Methods for visualizing the pore space
Setting aside quantitative macroscopic defmitions, if we wish to observe the pore space, we must have investigative facilities at the scale ofthe pore. For coherent media, it suffices to saturate the pore space totally with a liquid which can then be solidified. Use is often made of synthetic resins (including. epoxy) which ·are injected under vacuum and polymerize under the effect of time or heat. In this way, the pore space can be analyzed by the conventional methods of optical microscopy, with thin sections (Plate 1), polished sections etc. For observation by the scanning electron microscope (SEM), one can also fabricate moldings of the pore network (replicas) by destroying the mineral phase with hydrochloric and hydrofluoric acids, to leave only the resin (Plate 2). The replica technique offers the advantage over other methods of allowing observations at very high magnification. and offurnishing a virtually three-dimensional view."{he very wide depth offleld of the SEM serves to obtain stereographic photos [by tilting the preparation under the microscope between two photos (Plate 6)]. This makes it possible to measure the
...._____ __ _
--------~~~~~~~~~~~~--~~~--~~~~--
"
'--
l ~
POROUS MEDIA
13
thickness of very thin cracks, and even to consider the use of photogrammetric methods to observe pore geometries. These different techniques are discussed at length in Pittmann and Duschatko (1970), Caye et al. (1970) and Delfmer (1971).
·~,
~
1.2.2 Simple examples of pore geometries Observed at the microscopic scale, pore spaces display an extraordinary diversity (Plate 1), and in broaching the analysis of their geometry, it is wise to refer to simpler specific cases. We shall examine two types: (a) The intergranular space in a pack of spheres. (b) The ideal vuggy medium.
1.1.2.1 Intergranular space in a packing of identical spheres
·~
This pore geometry has been discussed in many works (by G. Cargill in 1984, for example), because it lends itself well to the preparation of experimental models and to computer simulations. This is also the theoretical case most closely approaching the pore spaces frequently encountered in natural media (sandstones) and artificial media (sintered materials). In fact, the dense random packing of isogranular spheres is very complex. It is therefore customary to describe only regular packings (which do not exist in reality for volumes larger than a few grains). Table 1.2 summarizes the main characteristics of the three packings most often described. · These regular packings offer us a preliminary view of the geometric complexity of pore spaces. For each regular packing, the diameter of two types of sphere can be calculated: (a) The largest sphere inscribable in the widest zones of the pore space (for a cubic packing, this sphere has~ radius of 0.732, where the unit of length is the grain radius). (b) The largest sphere passing through the narrowest pore cha~els (accesses), i.e. the largest sphere that can .. circulate" freely through the entire pore space concerned (for the cubic packing, its radius is 0.414). The radio of these two radii is 0.56 for cubic packing. In this case, it is evidently unrealistic to try to divide the pore space into two clearly distinct zones: the·pores (i.e. subspherical widenings corresponding to the largest inscribable sphere) and the · "accesses" to these pores, as is· often done on two-dimensional sections of the pore space. This ratio is about 0.3 for more compact packings. This shows that the distinction between •• accesses" and •• pores" is inore diffiCult to make than is often believed. Similar conclusions are drawn by considering the volumetric aspect of the problem. Table 1.2 shows that the largest inscribable sphere represents only a fraction ofless than 45% ofthe total pore space of the regular packings described. The notion of pores in the sense of
---- -
Compact hexagonal or tetrahedral
Simple hexagonal
,
J!
Q,
] ~
~. . ..,
I~ ~
,
~
Sohd phase crystal structure
I'~
I
I
I
Tetrahedral void
J-
..
'
nw . '
.
~
Simple rhombus void
@
Cubic void
.
.
{Wj
p ore structureHl
I
j§
about 36
25.9
39.6
47.6
p orost't y (%)
.
I
I I
z ,., ... , (' «
The unit or length is the radius or the sphere. (I) From GUILLOT (1982). (2) From CARGILL (1984).
Dense random packing or hard spheresf21
h
:=-
~ ~
~
~~
SintJ?Ie
. Packmg type
'.
t&
around 9 on average
I2
8
6
Numhcror con tac t points h per sp ere
+
I octahedral
2 tetrahedral
I ,
5
'k
at least f1ve main types Bernal's canonical holes
I
2 trigonal
. Votd type
TARLE 1.2 CIIARA
I
most rrcquenl radius 0.29
0.4 I4
0.225
0.528
0.732
Radius . o r maxtmum inscribable h 01 sp ere
Radius
I
tmj;.,
0.1 )~
0.414 and 0.155 curvilineartriangular pore access
pore access
~~
0.414 curvilinear-
I
I
I
r' .
ofmaximum sphere . passmg through narrowest pore channels
Fraction
27
45
43
.,
(%)
sph~re
con tamed . 111 t~em~xtmum mscnbable
ofpo~osily
1
15
'--
subspherical widenings that are relatively large in comparison with the .. accesses·· is only truly meaningful for the intergranular space if the porosity is signiftcantly reduced by the compaction of natural sediments or by the intense sintering of artiftcial materials. This points out the limits of a strictly geometric analysis. In practice, the intergranular medium tends to be considered mainly through a series of equations relating porosity to other petrophysical parameters. An example of this procedure is given for Fontainebleau sandstones (see Section 1.6).
~.
1.2.2.2 Ideal
\.._
'---
'-
POROUS MEDI:\
I__ (_ \.._ ~-
~
vug~·
medium
The ideal vuggy medium is defmed as a medium whose pore space is formed of spheres (a complement or inversion of the above intergranular space). This is often the ftrst geometry considered for porous media. although this case is quite rare in natural environments. This type of pore space nevertheless exists in vuggy lavas and oomoldic limestone, for example. It is more often found in vuggy artiftcial media (concrete, glass, plastic~ In addition to its great simplicity. it displays the feature of being totally devoid of inter-pore connections, so long as the spheres are not in mutual contact. Experiments and calculations related to percolation theory show that a minimum porosity of 30% (in the simple case of spheres of identical radius) is necessary for ·the porous phase to be continuous (Larson et al., 1980). This 30% threshold is very important when investigating the electric or hydraulic characteristics of this type of medium. For example, fluid circulation (permeability) remains zero (very low in natural cases) in media with porosities less than 20 to 30%. and then· rises sharply once the threshold is crossed.
"-..
~
....:..
.-' ; ·'
'"<>~ I :~
·;::
i~
i
I ·-
~:r
...,
'--
~~
1.2.3 Actual pore spaces We have shown that a presumably simple pore space, such as the intcrgranular space, may prove to be very complex. One can well imagine the problems raised by actual media (see Plates 1 and 2). In addition to this intrinsic complexity of commonly found pore spaces is the problem of the three-dimensionality of pore spaces as opposed to the twodimensionality of our means of observation and analysis. Even in the rare cases in which partial (caSting) or total (serial sectionst three-dimensional data are obtained, overall methods are lacking to describe and. even more important, to quantify the pore geometries. To attempt to resolve this major diftkulty in microscopic description, several methods have been developed and are derived from a wide variety of theories. We shall provide two contrasting examples. The ftrst consists in a classiftcation adapted to carbonate rocks, whose pore spaces are likely to display an extraordinary variety due to the chemical mobility of carbonates, among other factors. The second example corresponds to an approach based on the principles of mathematical morphology.
1.2.3.1
Choquette and Pray
classif~c:ation
;....
This method (partly illustrated by Plate l) is \\idely used by petroleum geologists. It essentially consists in defming the pore space as the complement of the solid skeleton
16
POROUS MEDIA
1
-~
4
-1-~~ ~
-~
~ 1
PLATE 1
VARIETY OF NATURAL PORE SPACES Examples of carbonate rocks Choquette and Pray classification
J1
Photographs of thin sections (30 J.lm thick rock sections) Illuminated by transmitted light
Before making the thin sections, the rocks were saturated with dyed resin. Hence the pore space appears in red in the photographs. Note that only macroporosity can be identified by this system of photography.
FABRIC-SELECilVE · POROSITY
Photo 1 - lntergranular porosity. Oolitic limestone. Photo 2 - Moldic porosity. The large pores are due to the complete dissolution of microfossils. Photo 3 - lntragranular porosity. Nummulite limestone.
NON F ABRIC-SELECilVE POROSITY
Photo 4 - Remnant of vuggy porosity in a totally dolomitized oolitic limestone. Photo 5 - Channel porosity. Microcracks widened by dissolution. Photo 6 - Porosity in stylolites. Stylolites are microcracks of a \·ery specif1c type, related to pressure/solution processes.
· . .J
WWL
9
I :Jl JI1J -
WWL
g
l 3H:JNV1d
~~~~-~--~-~---~~~~~~~~~~--~~~~~~-
•
• \.....
PLATE 2 '-'
PORE CASTS (Epoxy replicas) Scanning Electron Microscope photographs After the injection of epoxy resin. the entire mineral phase (solid) is destroyed by acids. Only the resin , representing the pore space, appears in the photographs .
...... ......
'-"
~,m
I
\.....
Ground Pyrex (Compare with photo 3 Plate 4)
Vuggy dolomite
....
]soo ,m
II ...._,
II I
.......
I I Nummulite limestone (Compare with photo 3 Plate I)
Altered feldspath
18
POROUS MEDIA
1
which is easier to describe, and whose shapes are familiar to sedimentologists. In comparison with this solid skeleton, the pore space is defmed as lying between the grains or the crystals (intergranular and intercrystalline). within the grains (intragranular), or formed by the total dissolution of a grain (moldic porosity). The description can be further refmed by using the dependence between the porosity and the initial rock structure at the time of its deposition. Hence two major types are distinguished: fabric selective porosity, directly dependent on the initial structure, and non-fabric selective porosjty, superimposed on this structure by fracturing (fracture porosity) or by weathering (channel porosity). Among other advantages, this type of classification offers the advantage of clarifying the development of porosity in relation to the history of the rock. But its main drawback is to minimize the quantification of the dimensions of the pore space. Thus the application of this classification to rock physics is often limited to implicit generalities such as: intergranular porosity is better connected than intragranular or moldic porosity. Yet classifiCations inspired by petrography offer the great value of highlighting the vital fact of the frequent diversity of pore geometries within the same medium. In a given medium. it is not rare to fand the pore space divided among two or more geometric types, with contrasting characteristics and hence physical ·behaviors. To. understand a petrophysical behavior, it is therefore usually necessary to overlook the total measured porosity, and to consider only the fraction of this porosity that plays a role in the process investigated.
1.2.3.2 Application of mathematical morphology to the descrlptioa of porous media Mathematical morphology is the mathematical and probabilistic analysis of shapes. To apply this theory to the description of pore spaces, use is made of digitized images processed by specialized computers (image or texture analyzer, Plate 3). We shall restrict ourselves to a very concise review of this technique developed from the theoretical work of Matheron (1967), and described exhaustively by Serra (1982). The shape of a porous medium observed over a very small area is merely the particular culmination of a more general process. One can therefore try to formulate this process by means of a probabalistic theory. Within a homogeneous medium. this process is always the same. Whatever the point selected as the origin of the observations, it is reproduced indefinitely, always equal to itself. It can be represented by a stationary variable. To describe this medium. it can therefore be analyzed by using reference fagures, i.e. points, segments, circles etc. by raising questions such as: do both ends of segment h or the entire circle R belong to the porous medium? The answer is expressed in terms of probability, and the assumption of invariability by translation enables us to apply the laws specific to stationary random variables. The covariance function is one example of this probabilistic approach: if A is the set of pores of a medium with porosity l/J (where AC, the complementary set of A, is the solid), the term covariance Cu(h) in direction u is the probability that both ends of a segment of given length h belong to the set of pores A when segment h is displaced parallel to a given
n
PLATE 3
lS
AUTOMATIC TEXTURE ANALYSIS
.y
·e
Example of hexagonal granulometry in the grain phase of a Fontainebleau sandstone
d ·e Video image
Memorized image
te
0 )[
;:
:-'
n
s, a
d
IS
A t elevision camera placed behind the microsope (thin sections, polished sections) or an epidiascope (photography) furnishes a video image which is exceptionally contrasted in the case presented . (epidiascope). A gray threshold is selected to separate the phases. An image formed of black or white points (around 400 x 400 points) is memorized. In the general case, this phase separation operation · (segmentation) is often difftcult, and requires a whole series of ftltration operations.
'o
Hexagonal granulometries
:s
;t
)f
,'Y IS
d
e. iP )l
IS
)f ...-'
I), )[ l.
•
-
Structuring element -
•
A series of additions/subtractions of translated images is used to determine the fraction of the phase analyzed (black phase: grain) which may contain a given structuring element. The ratio of the area of this fraction (white in the photograph) to the total area of the phase is measured .
20
l
POROUS MEDIA
direction oriented by the unit vector u. The assumption that the variables are stationary requires that this probability is exclusively a function of h: Cu(h)
= prob { M(x)eA
and
.\1(x +oleA}
(Ll)
where x is the position vector of point M. Hence Cu(O) = (jJ, and this independently of u, since, in this case, x and x + hu are identical, and the porosity (jJ is clearly the probability that x belongs to the set of pores. If h becomes very large. one can consider that x and y = x + hu are independent. Therefore, if: h-. oo and Vu C.(h) = prob (xeA et yeA)= prob (xeA) prob (yeA)= (jJ 2
(1.2)
2
The value of h for which C.(h) practically reaches its asymptote t/J is called the range in the u direction. Like the function c.(h~ this range may be independent of u, as in the case of statistically isotropic media This distance, which characterizes the zone this phenomenon influences, is very useful in understanding porous media Figure 1.1 gives the simplest example of the function c,;(h).lt often occurs that the shape of the curve is more complex, indicating the complex organization of the porous medium (the "nugget effect" indicates an abnormal concentration, for example).
~
c (hi
I l
h
Range
Fig. U
Covariance function.
Automatic texture analysis, which allows a rigorous and rapid quantification (automation) of the pore spaces, appears to be headed for substantial growth. However, it has to contend with difficulties at tw.o levels: (a) At the practical level, digitized images of the pore space must be obtained before any analysis. The principle is simple: a television camera is employed and operated behind a microscope. In practice, however, the acquisition of good quality images requires a lengthy adaptation on costly equipment (Delfmer, 1971). (b) At the theoretical level, it will be a major task to select, among the many quantitative parameters furnished by image analysis, those that are really useful for the petrophysical interpretation of pore spaces.
1
POROUS MEDIA
21
1.2.3.3 Specifac case of crack and fracture pore spaces ·,
The pore geometries described above display the common feature of being actual threedimensional spaces. This volume accounts for a significant fraction of the pore bodies considered. However, virtually two-dimensional pore geometries exist, such as planar surfaces of negligible thickness. These are closer to discontinuities in a solid phase than pores in the commonly accepted sense, and they include fractures, cleavage planes, and grain boundaries, which are generally grouped under the term of cracks. Thus defmed, these discontinuities can only be of very limited volumetric importance, and, if not, the medium would lose all coherence. It is estimated that 0.5% fracture porosity is a very high value. Yet their mechanical role is vitally important (see for example Chapter 5). The visualization of these cracks raises certain practical difficulties. Once observed, however, by treating them as sets of planes, they can be described more easily than pores. The low volumetric importance of these cracks often imposes more sophisticated display methods than for ordinary pore networks. Resin injection methods must account for the very low permeabilities often involved (low viscosity resins, high injection pressure). The pore cast technique is mainly used for actual observation (destruction of the mineral fraction by acids), with stereographic techniques. For observations at lower magnification. it is sometimes useful to employ ultraviolet fluorescence microscopy techniques, which help to detect very small amounts of resins containing fluorescent pigments. These crack type media a:Iso provide a popular fteld for sample preparation artifacts: crack propagation under the effect of temperature and pressure during drying, washing or saturation. But the most difficult point (for natural materials extracted from the subsurface in boreholes)is the effect on the microcracks of the sudden decompression experienced by the core when it is raised to the surface. It is often difftcult to differentiate between the microcracks really present in situ and those created or widened by decompression. Sometimes, a very meticulous SEM analysis of microcrack planes yields some indications: microcrystals, micro "pillars", etc., but this only occurs in rare examples. This alteration of deep borehole samples by decompression is of fundamental importance for the interpretation of petrophysical measurements (acoustics or permeability of compact rocks, for example), especially since this microcracking is not necessarily reversible under the effect of the confming pressure recreated in the laboratory. By contrast, this artificial microcracking may prove to be a valuable means of investigating the structure of materials (for example, detection of remanent stresses in rocks). Volumetrically "negligible", sensitive to artifacts, and closely dependent on the stress state of the sample, the crack and fracture pore spaee is often difficult to observe. Its vital influence on the mechanical properties of the porous mediumjustif1es the efforts required in this area.
22
POROUS MEDIA
1
1.3 THE PORE SPACE: CAPILLARY APPROACH
1
n I
1.3.1
s: sl
Capillary equilibria: general discussion and defmitions
g
When two or more immiscible fluids co-exist in a pore space, the interfaces between these fluids and the solid are the locus of capillary processes. Restricting ourselves to twophase mixtures, it may be observed that one of the two fluids spreads preferentially in contact with the solid (wetting fluid). By contrast, the other fluid tends to minimize its contact area with the solid (non-wetting fluid). In the case of water and air, for instance, the water is usually the wetting fluid. A surface tension t. exists at the interface between the two fluids, inducing a pressure difference that depends on the mean curvature of the interface. This capillary pressure Pea is governed by the Laplace equation: Pea =
r.(;l ± ;J .
(1.3)
where R 1 and R 2 are the main radii of curvature ofthe interface. The + sign corresponds to the general case in which the centers of curvature 0 1 and 0 2 are located on the same side of the interface. The - sign corresponds to the case in which the centers of curvature are lo~ated on both sides, as may be observed in convex/concave interfaces (Fig. 1.2). In p*ctice, one mainly speaks of the mean curvature 1 Rm such that: 2
1
1
=-+Rm Rt- R2
(1.4)
02
01
Fig. 1.2 Example of convex/concave interface.
The existence, between the fluids, of this pressure difference, which depends on the shape of the interfaces, implies that the equilibria existing within a pore space are not random. The analysis of these capillary equilibria is very useful from several standpoints.
·--· · - - · · - - - - - - - - ·
g a
1
(]
'V
r (
I
n
,_ r. s e
e
..
I
POROUS MEDIA
23
In the experimental procedures which attempt to reach stable and significant states. it is not possible to control fluid saturations completely. (The fluid saturation is the ratio of the volume of fluid f to the total pore volume.) For a given experimental procedure, in fact, saturation can only vary between defmite limits. The capillary pressure, and hence the shape of the fluid clusters, may also depend closely on the fluid saturation technique for a given saturation. This shows that these capillary equilibria are important factors in the general behavior of a sample, because they interact with the other physical processes affecting the combination of solid and immiscible fluids. Capillary equilibria also offer an excellent means to investigate the pore space itself. These capillary effects enable us to measure a number of characteristics of the pore space (porosimetry) or to transform the porous geometry allowing measurements on simpler volumes than on the initial space. We shall merely review a few definitions. The interested reader can fmd a detailed description of capillary effects in Morrow (1970) or Dullien (1979). Wettability. The concept of wettability has been d~fmed above empirically by the variable aft'mity of a fluid for the solid. It is customary to quantify this affmity by the value of the contact angle (} between the fluid interface and the solid (Fig. 1.3). This simplification is only valid for perfectly regular surfaces. In actual pore spaces, this concept of wettability may become extremely complex (refer to Morrow, 1975).
F~g.
1.3 Defmition of contact angle.
We shall restrict ourselves to the case of perfect wettability (0 = 0), i.e. where a fluid shows a dearly preponderant affmity for the solid. For commonly encountered clean porous bodies (rock, glass, etc.), this often occurs for a liquid/gas mixture, in which the gas is the non-wetting fluid. Drainage and imbibition. The term drainage is used when a non-wetting fluid, under the effect of a pressure which counterbalances the capillary forces, invades a porous medium, expelling the wetting fluid. If the saturation with wetting fluid increases (following a drop in pressure), the process is one of imbibition.
J
24
1
POROUS MEDIA
1.3.2 Capillary pressure curves The basic experiment for analyzing capillary equilibria in porous media consists in the determination of the relationship between saturation and ::apillary pressure for different fluid saturation methods. The most commonly employed method is that of capillary desorption (sometimes called the restored states method). It requires the use of a semipermeable membrane (such as a ceramic plate) whose pores are so fme that, once they are totally saturated with water, they oppose the penetration of air by capillary pressure (this pressure may exceed one megapascal). The sample, totally saturated with water, is placed in a pressure vessel, on the ceramic plate, whose lower face is at atmospheric pressure (Fig. 1.4). Since the ceramic is impermeable to the gas, a pressure p can be applied to the gas that is greater than that of the water in the sample, with the water remaining at atmospheric pressure.
Pr-.wizedgn
........... ........................... :. :. :. &a;np..·.:.:.: ........... ................................. ................ ................ . . ..........................
P
.. uw
_ . Water
Fig. 1.4 Schematic diagram of desorber.
The water expelled by the air (drainage) flows along the ceramic and the curve of capillary pressure (Pea) is plotted against water saturation (S..) (Fig. 1.5). The fmt drainage or initial drainage (from a water-saturated sample) only begins at a given capillary pressure (Pace) (threshold pressure). This is the pressure that must be reached for the air to be able to penetrate into the largest pores of the sample. By raising pressure Pea• the sample progressively empties, and drainage tends to stop before the wetting fluid (water) has been completely expelled. Although the pressure is raised, the water saturation no longer decreases. The wetting phase configuration is such that fluid movements therein become impossible. The water is found in the form of small clusters interconnected by water films that are so thin that their viscosity is very high. The sample is in a state of irreducible saturation with wetting fluid (S,.J
1
J
l
.~
he-
"
.rv :ti-
a
('!
p
s
~
.I e
! .!I
u
\,.,. ,, 1.!: \
l
If
j ;
;Q
\\'
Ifl.
of
e
,,,,,, i•
:n
lk if
25
I'Oit.OUS MEDIA
'\
il il
\~ \ \
I
\
,, \
~
PCOs8
Wetting Fluid: liquid
- - - Wetting
\
t
I _
~Juid: men:uty Y..,of
\
5
-pressure 0
20
40 60 Wetting fluid 11tur1tion
..,
100
Fig. 1.5 Capillary pressure curves.
If this experiment is continued by progressively releasing the capillary pressure, the water again penetrates into the sample (imbibition) but the process is not reversed. At a given capillary pressure, the water saturation is signifiamtly lower at imbibition than for initial drainage. Once zero capillary pressure is reached, the water saturation is lower than 100%, and, in some cases, this water saturation value is closer to that of the irreducible saturation than to 1OOo/o. Part of the air is trapped during the imbibition process. This air fraction is called the non-wetting fluid residual saturation. Its value, related to the total volume of the sample, is the trapped porosity. During drainage, when the non-wetting fluid invades the pore space, the volume it occupies obeys the Laplace equation. For a given capillary pressure, the mean curvature of the interface is f1xed. The increase in capillary pressure therefore ~ds to increasingly smaller radii of curvature. The capillary pressure curve during ·drainage hence corresponds to the measurement of the increase (in the pore space concerned) of a volume bounded by a surface with an increasingly smaller radius of curvature. This explains why drainage can continue until the complete invasion of the pore space by the nonwetting fluid (subject to the concept of irreducible saturation, as defmcd above, and corresponding to the break in hydraulic, and not geometric, continuity of a liquid wetting phase).
J
26
POROUS MEDIA
1
On the other hand, imbibition corresponds to a progressive increase in the mean radius of curvature of the wetting fluid 'non-wetting fluid interface. Marked discontinuities may appear, since fractions of the volume of non-wetting fluid may separate from the main mass. This can occur, for instance, if certain areas of the pore space are not of sufficient radius of curvature as would be required to allow the expulsion of the non-wetting phase located" upstream". These relative narrowings play the role of a capillary valve. When the capillary pressure drops to p1, function of the largest allowable mean radius or curvature at the narrowing, the non-wetting phase remains upstream, is disconnected. and forms a trapped cluster at pressure p1•
:c - .d
sar "h '11fi
1.3.3 Application of capillary pressure curves to porosimetry We have shown that, when water is drained by air, an irreducible saturation zone exists at high pressures, corresponding to the immobilization ofthe water phase. This zone does not exist if the wetting fluid is a rarefted gas(" vacuum") which disappears with the gradual penetration of the non-wetting fluid. This applies to the mercury (non-wetting fluid)/mercury vapor (wetting fluid) pair. If a vacuum is previously applied, one can cause drainage by the penetration of mercury into a sample. This is the principle of mercury porosimetry. Initially popularized by Purcell (1949), this method has formed the subject of an abundant literature. Wardlaw and Taylor (1976) can be consulted on this subject. To interpret mercury porosimetry, the porous medium is treated as a network of capillaries whose mean radius R is calculated by J urin's equation: Pe=
2t. cos () R
11.5)
In the case of mercury in contact with most commonly encountered minerals, in drainage, the wettability angle is close to 140". At a given pressure, the mercury invades the fraction of the pore space connected with the exterior by" accesses" with a radius larger or equal to that given by Jurin's equation. Hence porosimetry can be used to quantify the access radii, but not the pore radii. This restriction must always be kept in mind in the interpretation of porosimetric results. The distribution of access radii in common porous media is usually of the log normal type. The experimental results are therefore presented (Fig. 1.6) with pressures (and hence the radii) on the logarithmic scale. For the semi~quantitative description of pore spaces, the most useful representation is the access radius distribution curve (this is the derivative of the curve described above). This type of representation is often called the porosimetric spectrum (Fig. 1.6). Mercury porosimetry is an important tool for the quantiftcation of pore networks. However, excessive quantitative extrapolations should be avoided. In practice, the main use of porosimetry resides chiefly in its capacity to identify the existence, in many porous media, of several families of" pores" with highly contrasted access dimensions and hence physical properties.
J
ius
27
POROUS MEDIA
1
figure l. 7 shows the porosimetric spectra of various porous media. A single type of access radius can be seen to exist in certain cases (for example chalk). On the other hand, many other porous bodies display very clear bimodal spectra (bioclastic limestone, shaly sandstone). The existence of three or four types of pores can be observed in some cases. This shows why the interpretation of the physical properties of these media is heavily influenced by the consideration of this diversity of pore types.
13
!lJ:
enl !l~.
tb
ue
s ·....
,.'
0
100
10
n
0.1 Injection pressure (MPal
Le-. 'r (b)
e'
§
li
Ll
e
01
I..
.l:
;
Jo
7.5
75
750
7500
Accen radii (nm)
.
Fig. 1.6 Schematic porosimetric curve (a) and spectrum (b). '-
I
...I
Porosimetric radii (.um)
0.1
1
10
100
Rock type
q,
K
(%)
mO
White chalk
42
6
j
ra Micritic limestone
23
\u to
24
.• IC
Pelletal limestone
30
20
•. 3 Crinoidal limestone
24
250
- -1
m~
Oolitic limestone
33
700
N
~i~
Bioclastic . limestone
37
~~
2000
·e,
th Bioclastic limestone
44
10
500
tS
Oedolomitized limestone
33
._,
7000
dJ j( ~~
Slightly shaly sandstone
25
600
Kaolinitic sandstone
20
1700
Silt
30
n
d
0.1
10
v
100
c f .(
Fig. 1.7 Example of porosimetric spectra of sedimentary rocks.
..I
POROUS MEDIA
29
1.3.4 Fluid distribution at the microscopic scale We have examined capillary equilibria from the macroscopic standpoint by means of drainage and imbibition curves, but it is interesting to determine the geometric shapes of dusters of wetting and non-wetting fluids in various equilibrium conditions. This helps us to understand the physical behavior of these clusters. and also, conversely, to achieve a more accurate two-dimensional description of the complex geometries of actual porous media.
1.3.4.1
Principles of the visualization of fluids in capillary equiUbria (Plate 4)
Use is made of pairs of wetting/non-wetting fluids of which one of the phases can be solidified after having reached the desired capillary equilibrium. Two types of pairs are mainly used. Wood's metaljvacuuni (Swanson. 1979, Dullien and Dhawan, 1974, for example). Wood's metal is an alloy of tin and bismuth which melts at 700C and. above 1200C, displays capillary behavior closely approaching that of mercury. Experiments are then conducted with the molten metal that are identical to those of mercury porosimetry, and then, by sudden cooling, the capillary equilibrium thus obtained is frozen. Polished sections fabricated from the samples thus treated allow easy observation of the location of the Wood's metal, and hence the determination of the accessible zones of the pore space, ior a mean curvature that depends on the pressure applied. Polymerizable synthetic resins (Etienne and Le Fournier, 1967. Zinszner and Meynot, 1982). Before polymerization, synthetic resins are liquids that wet rocks perfectly (in comparison with air). Using the air/resin pair, this makes it possible to conduct the drainage and imbibition experiments described above, and, after having reached the desired state of equilibrium, to polymerize the resin and f1x this state that can then be analyzed on a thin section or a polished section. The wettabilities of the rock by water or resin in the presence of air can be considered as identical as a f1rst approximation. For drainage: a vacuum-injected sample of dyed resin undergoes centrifugation, draining the resin. The acceleration is selected so as to produce:
(a) either a zone in which a relationship exists between saturation of wetting fluid and drainage pressure, where the resin occupies the pore volumes whose access radius is smaller than a radius that depends on the imposed acceleration, (b) or a state of irreducible saturation, where the resin occupies the same sites as the water at the end of the drainage tests. For imbibition: capillary rise is carried out. The base of the sample is placed in the resin \vhich, by expelling part of the air, gradually iO\·ades the sample under the effect of capillary forces alone. At the f1rst order, the fmal result does not depend on the pair of fluids used (provided that the wettability contrast is good). Thus a capillary equilibrium is reached corresponding to the one existing at the end of total imbibition. An example of this process is thoroughly discussed by Pickell et a/. (1966) and Bousquie (1979).
M
30
POROUS MEDIA
1
In a more general manner, the air/resin pair can be used in many cases to simulate the experiments conducted with air/water and oil/water pairs, to observe and to quantify the geometries of the clusters of the different phases.
.. r
'fa1
.;OJ
1.3.4.2
Application to the geometric description of pore networks
Capillary equilibria depend primarily on the geometry ofthe pore spaces and not on the pair of fluids employed (provided the wettability contrast is clear). We have discussed this point for imbibition. It is also the case for drainage (apart from the irreducible saturation zones) if one uses the parameter Puft. cos 6. Based on a very complex pore geometry, it is then possible to subdivide the space into several fractions corresponding to the capillary phases. Our investigative resources (thin sections, polished sections) allow only a twodimensional observation, but the capillary effect, even seen in these conditions, integrates part of the data relative to the third dimension. Two types of preparation are normally employed:
he be
i I
,iz
1.·
l.•
flc ch
Drained samples (Plate 4). This helps to analyze the fraction of the pore space connected by a porosimetric access radi1ls that is larger than the mean radius of curvature corresponding to the. drainage pressure. Using a porosimetric curve for a sample, one can select the fraction of pore space related to the process analyzed, and then separate this phase by drainage to obtain images that are much more easily quantiftable (for example, by the image analyzer) than the total pore network.
i,n·
co
Totally imbibed samples (Plate 4). We have shown above that, after total imbibition, the non-wetting fluid occupies the zones of the pore space whose maximum mean radius of curvature was large in comparison with that ofthe "accesses". This reflects a defmition of the "pore" considered as a widening of the pore space. This notion of a pore, although very familiar, is actually very complex because it is three-dimensional (this point was raised in Section 1.2.1 for regular packings of spheres). Bubbles of residual non-wetting fluid show the complexity of the pore space. In pore size analyses based on twodimensional observations, it is often advisable to distinguish this fraction of the pore space.
wl
01
w
m
er e<
m
1.4 FLUID FLOW IN POROUS MEDIA
tr d
Fluid flow in porous media has formed the subject of intensive theoretical and experimental investigation, and Dullien (1979) can be referred to in this respect. We shall restrict ourselves here to reviewing the points that are useful for the rest of our study, and to provide measures for order of magnitude estimates of permeability. From the outset, it must be stated that, more than any other petrophysical property, permeability is sensitive to the scale factor. If one measures the flow rate of a fluid through
-1:
.r
i I
,I ----------
------------------------------
'-
POROUS MEDIA
l
31
a rock formation at meter or hectometer scale, one can realize how rare discontinuities (fault, fracture) can determine the behavior of the formation. These bulk permeabilities or in situ permeabilities are frequently virtually independent of the type of rock. The reservoir engineer or hydrologist calculates these transmissivities. If, however, one is interested in the petrophysical behavior of porous media, the natural permeability of these media must be considered. This is often called the matrix permeability, and is measured on centimetersized samples free of discontinuities. Our discussion is restricted to this scale.
t~
t
tJ- I.'S
"
1.4.1
t
trv
Case of a single fluid totally saturating a pore space : absolute or single-phase permeability
{I,_
t' .liv
1.4.1.1
Permeability was fiTst defmed by Darcy, in the speciflccaseofwater, by showing that the flow rate per unit area was proportional to the pressure drop in the porous body and to a characteristic parameter of the .porous body concerned called permeability. The introduction of the concept of viscosity helps to generalize this law for all fluids, considering only non-turbulent flow. Darcy's law in its simplest form is writtenct):
~eo 1~..
2
niS
....
Defmition, units and measurements
::-r
t
Q
=~SAp '1
h..
(1.6)
where Q =volumetric flow rate in a given direction through a slice of the medium of thickness. Aland areaS with a pressure difference L1p at the ends of the slice. " = permeability of the porous medium, '1 = viscosity of the fluid,
(
of
,,
Al
r
n~"
or its differential form:
'0.,.._
U= -
II
where
U is the flltration
K/'1 grad p
(1.7)
velocity of the fluid.
Permeability is treated as an area, and its unit in the International System is hence the m 2 • But this unit is little used today, because it is disproportionate to the values encountered in nature. The traditional unit of permeability is the darcy tO), which is equivalent to the square micron (l darcy is exactly equal to 0.986923 J.UD 2 ), but the millidarcy (mD) (;;;: 10- Is m 2 ) is frequently used, as it is better adapted to the order of magnitude of the permeabilities generally observed. In fluid mechanics experiments in porous media, the permeability is usually calculated directly from the measurement of the flow rate and the pressure drop. For a rapid
1(
dl
let(l) The expression of Darcy's law given above corresponds to the isotropic case (i.e. " independent of direction). In practice. however, it is important to note the strong dependence of permeability on direction for many pore spaces.
:v
~ll-
I
,j
PLATE 4 VISUALIZATION OF CAPILLARY PROPERTIES Photographs of thin sections of rocks selectively saturated \\ith dyed resins or Wood's metal by drainage and imbibition Photo 1
'j! '
Total imbibition. Vuggy dolomite (dolomitized oolitic limestone). Red resin. wetting fluid. Yellow resin, residual non-wetting fluid (trapped). The characteristic shape ("bubbles") of the clusters of residual non-wetting fluid can be observed.
Photo 2 - Drainage and imbibition. Crinoidal limestone. The sample, f1rst saturated with blue resin, was centrifuged to irreducible saturation, and then, after polymerization of the blue resin, impregnated with red resin. Blue resin represents irreducible wetting fluid. Red resin represents the fraction displaceable by imbibition of the non-wetting fluid accumulated during drainage. Yellow resin represents the residual fraction of non-wetting fluid. Note particularly a petrographic detail which plays an important role in the distribution of the displaceable and trapped phases of the non-wetting fluid. The fme layer of palisadic cement surrounding the crinoids has burst under the effect of compaction. and the fragments of cement part the intergranular spaces, favoring imbibition (red resin) without substantially altering the rock structure. The pore spaces devoid of these fragments usually correspond to trapped porositY (yellow resin). This compartmentation process, which reduces trapped porosity, is not an exceptional occurrence. Photo 3 - Drainage. Ground Pyrex. ~olten Wood's metal behaves like mercury (non-wetting fluid) and porosimetry experiments can be performed in which this metal is frozen by cooling. Black, Wood's metal (drained zones). Red resin, subsequently injected (undrained zones).
-,
I t
Ai
!
Photo 4 - Successil"e drainages. Crinoidal limestone. The rock. f1rst saturated with blue resin, was subjected to drainage by centrifuge, and theri, after polymerization; to' saturation with red resin, followed by centrifuge under low acceleration. It was then saturated with yellow resin. The color of the resin present in a pore depends on the access radius.
Blue
Red
Yellow
Acceu Radii (jlml -
The very small access radius zones (blue resin) correspond to microporosity in the crinoids (invisible in transmitted light) and also to intragranular macroporosity. A good intergranular porosity (red resin) is connected by flow channels smaller than 1 ~m. This is frequently encountered in well-cemented bioclastic grainstones.
'·
t
Afl .J
-
A
tnOJ ";l
L ~
m .. -.
'0
"lS
:11 J-"1cl - t 3H:JNV1d
l
POROUS MEDIA
33
·determination of the absolute permeability of a large number of samples, the variable-load air permeameter is often used. A column of water placed in negative pressure in a calibrated tube returns to equilibrium by aspirating the atmospheric air through the measurement sample (Fig. 1.8). The time taken by the column to fall between two predetermined marks is proportional to the permeability.
11
Sample
• I
Calibrated tube
......_. al
Constant level tank
Fig. 1.8 Schematic diagram of air permeameter.
1.4.1.2 Characteristics of the pore space affecting permeability The f1rst parameter with which permeability K can be correlated is obviously the porosity t/J. In fact, while examples of excellent t/J/K correlation exist (see Section 1.6.2.2), this relationship is more often than not ambiguous. Hence, for limestone rocks with 20% porosity, Figure 1.9 shows that the permeabilities commonly vary from 10- 3 to more than 1 D. Permeability is far more closely conditioned by the size of the flow channels inside a porous medium than by their relative abundance. For example, if one takes a very simplif1ed porous medium formed of straight capillaries of radius R and porosity t/J, Poiseuille's equation gives K = f/JR 2 /8. In this equation, the radius R appears as the square and therefore plays an important role.
e)
(2) Note that Poiseuille's equation relates the flow rate a capillary tube of radius R, this equation is written :
a to the pressure difference per unit length Jp/ .1/. For
a= 1tR~
s,
Isee, for example, Mandel, 1950).
Jp dl
34
1
POROUS MEDIA
5
tO
20 30 40
to
10"1
to·
to·
2
t0" 2
~ to·
3
to•3
e."'
~
. §
:c
"'
a.
10"
4
tO.
5
to•
6
10"
7
"'
~
tO," a
0.5
~
~
(.)
2
5
tO
20
30 40
Porosity (%I
Fig. 1.9 Schematization of porosity/permeability relationship, in rocks. We used data from Brace, 1984, for crystalline rocks, and from Scholle, 1977, for micritic limestones.
The problem of
-------------
I
35
POROUS MEDIA
porosity fraction corresponding to the largest type of access radius. The simplest means of identification (but experimentally the most costly) is mercury porosimetry. Figure 1. 7 gives examples of porosimetric spectra among which many are bimodal, with differences in radii often exceeding one order of magnitude. Obviously, only the large access radius family influences the permeability. In many cases where mercury porosimetry is unavailable, an attempt can be made to identify the different types of porosity by direct observation on a thin section or a pore network casting (Plate 5), followed by a quantitative estimate on a thin section. Once this analysis is completed, it is generally easy to explain the t/J-K relationship by relying on a pore geometry model such as the intergranular model, which is the most widely used.
1.4.1.3
Porosity/permeability relationship in intergranular spaces
The intergranular space displays a t/J-K relationship that is fairly well known experimentally, at least for grains of subspherical shape and constant grain size distribution (Jacquin, 1964, for example). We have shown that, for the permeability of a network of cylindrical capillaries, the combination of the Darcy and Poiseuille laws give the equation: R2 (1.8) K=-t/J 8 For a network of tortuous capillaries of any cross~section, an equation of the same type is found by introducing the hydraulic radius Rh, the tortuosity of the current lines -r:(t/J ), and a shape parameter A, that is relatively invariable:
e)
A
K
2
= r(t/J)
Rh
tP
(1.9)
.JS,
However, Rh is a function of t/J where Sis the specif1c area of the pore space. Since .jS is itself an inverse function of the grain diameter d, Carman and Kozeny established the following equation for the permeability of such a network (Carman, 1961): d2 K = Bt/J3 -r:(t/J)
(1.10)
where B is a constant for a given medium. Permeability is hence proportional to the square of the grain diameter. Figure 1.10 shows an experimental verification on intergranular spaces.
(3) The tonuosity of a capillary model quantifies the mean developed length (I~) of a current line joining the two ends of the model, in relation to the real length of the model (I..):
'~
rlt/>)=-~
1..
I
The exact expression of tonuosity varies with different authors.
·
36
POROt:S
1
~IEDIA
The parameter K/d 2 is a function of lj>", with n varying experimentally in accord with porosities from n;;?; 7 (l/> < 5~o) to 11 ~ 2 (l/> ~ 30%). In the Carman-Kozeny equation, if tortuosity is constant, the exponent n is 3. and this is observed in sintered glass and Fontainebleau sandstone for porosities ranging from 15 to 30%. Many natural intergranular media display porosities between 10 and 25%. Hence an exponent between 4 and 5 is often used for these media. Ghen the wide scatter observed in permeability measurements. this value of n corresponds to a" mean" between the low and high porosities. >
lx1o-•l
4000
400
40 Sintered giiS$ I
• 280 pm spheres o 50 pm sPheres 4
• 0
10
20
40
Porosity (%I
Fig. 1.10
c1>
vs ;
2
relationship in sintered glass (from Pellet, personal
correspondence l.
The high exponents observed for low porosities are related to a threshold effect. For the pores to be interconnected, a critical porosity must be reached (this point may be statistically modeled by the percolation theory). Experimentally, this critical threshold appears to lie at about 5% in clean sandstones and some sintered glasses. A certain terminological ambiguity of this threshold must be clarified. The threshold does not correspond to a transition from disconnected porosity to connected porosity (see Section 1.1.2) since, even at very low porosities. physical connections often persist between the pores. But these connections are so thin that they ~orne insignif1cant in relation to fluid circulation.
PLATE 5 EXAMPLES OF BIMODAL PORE NETWORKS (Epoxy pore casts)
f i
Scanning Electron Microscope photographs
I
...
10
TI
~-tm
1
Photo I - Oolitic limestone.
Detail of photo I.
This limestone consists of oolites formed themsel ves of very fm.: calcite cr;;stals, between which abundant microporosity exists. Intergranular macroporosity can be observed between the oolites. Fluid flow obviously takes place only in this fraction of the pore space.
+
20
l
~-tm
1 Photo 2 - Kaolinitic sandstone.
Detail of phoio 2.
Clusters of kaolinite crystals have developed in some of the intergranular macro pores of this sandstone, generating a specific porous medium .
38
1.4.2
1
POROUS MEDIA
Multiphase flows
The simultaneous flow of two or more immiscible fluids in a pore space is difficult to investigate experimentally, since it requires the measurement of flow rates arid pressures in all the phases. We have shown that multi-phase mixtures are subjected to capillary pressure effects. The superposition of this capillary effect on the dynamic pressure drop partly explains this experimental difTlculty. Hence, despite its great practical interest for the production of hydrocarbon f1elds and the abundance of theoretical and experimental work already devoted to it, this branch of fluid mechanics in porous media is still the focus of many discussions. We shall simply review the concept of relative permeability for two fluids, whose surface tension t, is high and wettability well contrasted.
1.4.2.1
Concept of relative permeability
If, in a porous medium, a combination of water (considered as the wetting fluid 1and a non-wetting fluid is caused to circulate, a steady state is obtained for each flow rate, characterized by the pressure drop gradients in each of the phases, the flow rate, and the fluid saturations. By analogy with Darcy's equation, it is possible, for these different states, to calculate the permeabilities relative to water Kw(S,..) and to non-wetting fluid K""'(S,..). Dividing these values by K, the single-phase permeability of the medium concerned. yields the relative permeabilities, for which one example is shown as a function of saturation in Fig. l.ll.
>
~
1 . B.
Non-wetti119 Fluid
0.6
-~
~
a:
0 0
SW;
sr
100
Water saturation (% l
Fig. 1.11
Relative permeability curves.
~
l
l
39
POROt:S MEDIA
Note ftrst that, as defmed above, the experiment can only be performed between two saturation states defmed in Section 1.3.2: irreducible water saturation s. . . and residual saturation of non-wetting fluid S,. When the sample is in a state of irreducible water saturation, the water is by defmition immobile and has zero relative permeability. The water is also limited to very thin ftlms on the walls or to clusters within menisci of very small radius of curvature. This does not hinder the flow of non-wetting fluid. whose relative permeability is close to l. In a state of residual saturation with non-wetting fluid, the situation is reversed, but not symmetrical: the non-wetting fluid is discontinuous and immobile and exhibits zero relative permeability. Moreover, it occupies the central parts of the pore network and considerably hampers the flow of water. The relative water permeability of a porous body in a state of residual saturation with non-wetting fluid is generally much lower than l (see Fig. l.ll). Between these two extreme cases, permeabilities depend on saturation and the fluid setting method. We have shown that, for the same saturation, the shapes ofthe phases may vary. For the relative permeability to non-wetting fluid, this is generally greater in drainage than in imbibition. In the latter case. in fact, the non-wetting fluid is partly discontinuous (clusters of trapped porosity). For relative permeability to water, which is usually very low, this difference is difficult to pinpoint. Table 1.3 gives a number of values of relative permeability to water and to oil (non. wetting) for natural media in a state of irreducible and residual saturation.
0
n
y p r -
ll
s
i)
TABLE
1.3
ExA~fPLES OF RELATIVE PER!IfEABILITIES
Type of porous medium
I ~:
1 f
. J l•
~
•
f
Clean sandstone ....... Coarse-grained shaly sandstone ............. Chalk. ................ Fine limestone ......... Oolitic limestone ....... Bioclastic limestone ....
Irreducible water saturation
Single-phase permeability (water) "(mD)
Sw) (%
Relative permeability to oil
ll
220
9
22 42 28 15 34
1700 2.4 25 80 1100
20 18 20 20 30
Porosity % (¢)
Oil residual saturation
s,
I Relative
(%1
permeability to water
1
58
0.04
1 0.8 0.7
40
I
32 20
0.3 0.15 0.03 0.18 0.1
1
33 60
Note that, in a state of irreducible saturation, the relative permeabilities to oil all approach 1. It is not rare to exceed 1 experimentally when the permeability to oil is slightly improved by the presence of a ftlm of water on the walls! On the other hand. in a state of residual oil saturation, the relative permeabilities to water are low and variable from one rock type to another. No clear relationship exists between single-phase permeability and relative permeability. For example, compare chalk to fme limestone. for which the relative permeability is much lower, despite a single-phase permeability that is ten times greater. These two types of permeability do not depend on the same geometric characteristics of the pore space.
40
POROUS MEDIA
l
1
From the practical standpoint, it is important to recall that values of relative permeability to water are generally low (even with very high water saturations). and must be taken into account. for instance, in the interpretation of the acoustic properties of twophase mixtures. More simply, from the experimental standpoint, these low relative permeabilities explain the slowness with which certain two-phase equilibria are established in "apparently" quite permeable media with respect to single-phase permeability.
1.5 PROBLEMS OF SCALE IN POROUS MEDIA The application of the macroscopic laws of mechanics to porous ltl'"..dia implies that these media are continuous, in other words that physical values can be defmed at each ·point; such as porosity. permeability, saturation etc., in the form of derivable functions of the point concerned. In fact, discontinuity turns out to be the basic characteristic of a porous medium, since, at the microscopic scale, a point lies either in the solid or in the pores.(limiting the matter to the porosity variable). The discontinuity problem is a classic one in physics. The originality ofthe porous medium resides in the fact that the dimensions o(the tlementary volumes, necessary for taking care of the discontinuity effects, may vary substantially for a given medium in accordance with the parameter analyzed.
'
i 1.5.1
Defmition of minimum homogenization volume
To defme this minimum homogenization volume, let us consider a two-dimensional example of the intergranular pore space (formed of grains of diameter d) (Fig. 1.12). If concentric circles with increasing areas are implanted at random, the change in porosity as a function of circle diameter can be measured. In the example given, the center is implanted in a pore, and the initial porosity is therefore 1. The mean value of the porosity of the medium is reached with circles of 2d to 3d in diameter. In this way, the minimum area required to homogenize the porosity variable can be calculated statistically. Section 1.2.3 defmed the range of the covariance function. This is the distance from which two points in the space concerned no longer display a statistical relationship between each other. This offers us a concept very close to that defmed above. The range is hence a means to quantify the size of the minimum homogenization volume of the porosity variable. The range is on the order of magnitude of one or two grain diameters for intergranular spaces with well-sorted grain size distribution. As soon as the pore structure becomes complex (as in certain limestones for example), the minimum volume to be considered may be very large in comparison with the unit grain size. If, presuming the example of the intergranular medium in Fig. 1.12, one analyzes the phase dispersed in the form of bubbles representing a low saturation, the same method of concentric circles leads to the observation that the minimum homogenization area (i.e. of the representativity of the phenomenon) has a diameter of 6 to 8 units. It is also observed
.M
l
1
e t
e
1
41
POROUS MEDIA 1
1-----C
t
' I
•
Cc 0111 ,.. :z:
(
i j
'
~
..:
Solid: black phaM 172"1 Pore: wllltelftd cr.,......tched ph-. 12ft) The cross-hatched ph- represents 10% of the white ph-. (satumionl
,
~ c
~-fd •en.,_
z~O
-!iii
I
,
80
~
60
·g
40
••
2::1
100
....,
l
~
\
/ II I
0..
20
16 •
t;
-
12
''---------
I
•
0
2
3
fi
-·
.i: ~
8
.t:; :I
4
u
0 4
5
6
7
8
.g
10
Diameter of meaturement circles
Fia- 1.12 Defmition of minimum homogenization volume on an example of porous media (top), The evolution of porosity and saturation as a function of the diameters of the measurement circles is given in the bottom ftgure.
J
I
that the drawing is not sutTtciently large to clearly defme this saturation. If we had a larger observation fteld to measure the covariance function, it would exhibit a clear nugget. effect (C(h) = 0 between O.Sd and 3d). Hence, for the same very simple medium, a sharp variation is observed in the size of the homogenization volume for the "porosity·· parameter and one of the "saturation" parameters. This may apply to all the remaining variables. As for permeability, quantification is far more ditTtcult because a two-dimensional simplification is unfeasible. However, it can be understood intuitively that the volume required to defme permeability is greater than that defming porosity in many cases. For the mechanical parameters
.. _,.
! l
f
(moduli), it is possible in principle to use a volume equivalent to that required for porosity, as long as the material is not cracked. But, if the medium displays cracks, it is the cracks that are essential for the "'echanical behavior and,. once again, the minimum homogenization volume may be significantly greater than that corresponding to porosity.
j
42
POROUS MEDIA
1.5.2
:-!
·
Minimum homogenization volume and physical behavior
Consequently, when a law of macroscopic behavior is defmed for a given medium, one must account for these different minimum homogenization volumes. Mechanical experiments are only .. macroscopically" meaningful ifthe sample is much larger than the minimum homogenization volume of the parameter analyzed. For cracked media, this volume may sometimes be considerably underestimated. In acoustic experiments, the wavelength must be compared with 1he minimum homogenization volume of the medium, or, more precisely, with the largest of the minimum volumes for the parameters involved (porosity, saturation, permeability): (a) If the wavelength is significantly greater than the diameter of this homogenization volume, the vibration behaves effectively as it would in a homogeneous ··macroscopic" medium. The vibration is not sensitive to the microscopic discontinuities of the porous medium.
1 (
'
f
(b) If the wavelength is of the order of magnitude of this homogeneity dimension, diffraction effects occur, radically altering the behavior. The main homogenization volume (which is no longer the right one) must be divided into smaller cells (which are themselves homogeneous), at the scale of the grain~ for example, and these cells must be used as the basis for analysis. (c) If the wavelength is much smaller, one is virtually faced with the initial problem: the porous medium no longer exists as such, and the analysis must be resumed at a much smaller scale. The medium to be analyzed in this case consists of the individualized grains and pores. This contrasting behavior of the relationship between the scale of the medium and the wavelength does not raise any basic difficulties as long as the homogenization volume does not vary during the experiment. However, this is not always the case, and a delicate ·problem may be faced when a parameter such as saturation is adjusted. The example in Fig. 1.12 shows that, in some geometric configurations, low saturations may give rise to homogenization volumes much larger than that of the pore space itself. In a desaturation . experiment, the homogenization volume may thus vary until the scope of the wave/medium relationship is changed, with the experiment passing progressively from ·'homogeneous medium" behavior to "diffracting medium" behavior.
-----------~---------------~
·i
l
1 ty,
k• 11'
:y.
I
POROUS MEDIA
43
1.6 EXAMPLE OF A NATURAL POROUS MEDIUM: FONTAINEBLEAU SANDSTONE We have shown that no global approach is available for the analysis and description of porous media, but rather a set of methods, whose common feature is undoubtedly to describe the particular features of the medium, analyzed in relation to simpler or betterknown typical examples. For practical applications, it is therefore useful to have such models. We shall now give a simplifted description of a typical example of a granular porous medium, Fontainebleau sandstone, which serves as a basis for many rock physics experiments.
-, ~
1e ill'-~ I~
i~ ·-~
1.6.1
h.__
Solid skeleton
1e '--
n"IS'-C"
\ __ _
I, l
[)'-
lt
sr \
e'--
t
I '
' f
j
•'
•
l·, ~
\_ '-·
~-
I
t
i
"
I.
I
'-
1'--
'''\..
1.6.2
Pore space (Plate 6)
Associated with this exceptional simplicity of the solid skeleton is a pore spaCe that is itself rather simple (apart from the problem of microcracks in certain samples, which is discussed in Chapter 5 Section 5.1.1.1), making Fontainebleau sandstone a good example of an intergranular pore space, especially since the cementation, which was highly variable, provides an uninterrupted range of porosities between 3 and 30%.
~ l
1.6.2.1
~r'
,l..__
Fontainebleau sandstone is essentially formed of quartz grains that have undergone long-term erosion and good grain size sorting before being deposited, during the Stampian, in dunes bordering the shore. The original deposit is formed of grains of rilonocrystalline sub-spherical quartz, around 250 ~in diameter. Following a geological evolution, still not fully understood, these sands underwent cementation (more or less intensive) by silica, which crystallized around the grains in the form of quartz in crystalline continuity with them. Fontainebleau sandstone thus displays exceptional chemical and crystallographic simplicity, and contain more than 99.8% quartz.
~- ' Y•
Geometric characteristics
Examples of hexaganal granulometry of the pore space are given in Fig. 1.13. This granulometry is obtained using a texture analyzer. The y axis gives the percentage of the pore space accessible to a hexagon of a given ''diameter". Although this provides a representation identical to a grain size distribution curve, it must be recalled, for practical interpretation, that this pore size distribution does not actually correspond to individualized objects. The porosimetric spectra are given in Fig. 1.14. Note the good correlation between porosity and modal access radius. Many measurements are available in the porosity range 8 to 22%, enabling us to observe a linear relationship of the type R,.. = aot/J, where R,. is the modal radius in Jim, tjJ the porosity in %, and a0 a proportionality factor approaching 1.
44
1
POROt:S MEDIA
j
100
\\,·..:\.
c
..... 0
lj
!8 0 c C.\!!
80
,..._~
\
60
O.!
s,! .. 0
,',··.... .•. ·.
~
c .. ~;e
~
-
~=
4.5%
,
'·.. '....:··.'...._:·~.:::::__ ---~ .:':":':-........·
20 0
0
50
I
--- ~= 9.5%
'
'·····.. '\.. ',···..::--.
40
~~
.ti
--- ~= 23% ··•··•· ~= 16.5%
.
200 100 150 "Radius" of reference hexagon (j.tm)
250
Fig. 1.13 Hexagonal (two-dimensional) granulometry in Fontainebleau sandstones for the porous phase.
; = 21%
\I
.2
~
! ..
~ = 14.5%
4
~
'
H ~
c
. -~
0"' ~
!
&.
•
0
1.
~
.., ~ ~ ·~
,. ·~
~
= 9.5%
I I
I
• = 5.2% I
.J
I 10 20
0.1
100
Pore access radius (j.tm)
Fig. 1.14 Porosimetric spectra of Fontainebleau sandstones.
1.6.2.2 Porosity/permeability relationship It is not surprising that this geometric regularity is associated with an exceptionally precise >-K relationship. This relationship was ftrst investigated by Jacquin, 1964, using the parameter K/d 2 , where dis the grain diameter. Figure 1.15 illustrates our results. This relationship corresponds to more than 400 samples taken from different outcrops of
----~-------------------------
....
~
l
~
.,.
•
PLATE 6 EPOXY PORE CASTS OF VARIOUS POROSITIES FONTAINEBLEAU SA~DSTONES Stereographic views of Scanning Electron Microscope photographs 500 11m
¢ = 28 %
r
¢ = 21 %
/
r'
·r
r r
r
r
r r r ('
,.. (
¢= 5 %
46
I
POROUS MEDIA
,I
n=8 :// / ),,.,n=3 1000
I
."1'
i !
'
; ...y··
s.§.
,r· 4! f:•
100
~
I
10
10"1
't
..f:•·
I- ·=···
$ '
t
1----;---~~-----.....-· 5 10 15 20 25 30 2 •
-"<
'1
1
Porosity(%)
Fig. 1.1! Porosity/permeability relationship in Fontainebleau sandstones.
/
15
~
.//
l: 10
l
~
u.
./··
5
~
0 0
5
./10
15
20
25
Tot1l porosity(%)
Fig. 1.16 Free porosity/total porosity relationship in Fontainebleau sandstones.
' '~ '---
'·
---~-------------~
l
-----
POROL"S MEDIA
i
t'
47
Fontainebleau sandstone. Slight variations in grain size distribution explain the splitting of the curve, which nevertheless offers a rare example of good correlation. Note the variation in exponent n of the equation" = f((jJ") discussed in Section 1.4.1. Between 8 and 25% porosity, the exponent is very close to 3, as in the Carman-Kozeny equation. This point should be observed in the light of the linearity of the 4J vs porosimetric access radius relationship, and of the equation K = a(j)R 2 for flow in straight capillaries. At lower porosities, n rises to 7 or 8 (for 4J < 5%1. This can be related to the problems of percolation threshold already discussed.
1.6.2.3 Total porosity/trapped porosity relationship
1 ~
l
r:·
Trapped porosity corresponds to the pore fraction of the space which preserves the nonwetting fluid at the end of total imbibition. We have shown that this trapped porosity corresponded to widenings of the pore space. By comparing the hexagonal granulometries (Fig. l.l3) with the porosimetric spectra (Fig. l.l4), one can observe that the pore phase granulometry decreases more slowly than the access radius with decreasing porosity. In this case, the relative importance of trapped porosity must rise. This can be observed in Fig. 1.16, and for samples with porosity lower than 6%, the porosity is entirely trapped (to within measurement uncertainty). A threshold effect is therefore observed. Threshold effects at low porosities are fairly common. For intergranular spaces (Fontainebleau sandstone), the thresholds appear to lie between 4 to 6% porosity, but for bimodal pore networks, they may occur at a much higher value.
--
---------------~-
2
'-
wave propagation in saturated porous media
~
~'
i
•
t,, t
.
"
. INTRODUCTION
~
;f ~
~
The application of the general results of continuum mechanics to heterogeneous materials implies that the wave phenomena are observed at the macroscopic scale. Hence the parameters introduced in the previous Chapter are presumed to be defmable on the elementary volume dx 1 dx 2 dx 3 employed in the mathematical description of this Chapter. In other words, porosity f/>, permeability "· etc., are continuous functions of point Mat the macroscopic geometric scale considered (i.e. f/>(M), K(M) ), while their local values depend on the microscopic, physical and geometric characteristics at point M. This Chapter is intended to broach the study of wave propagation using the tools of continuum mechanics. The first part reviews the results ofCiastodynamics Ul, which can be applied to porous media if the dissipative character of propagation is ignored. The second part shows how the two-phase character of saturated porous media introduces a dissipation, and how this factor can be taken into account. We shall then show the importance and the limits of the model introduced, and the need for further developments.
2.1 2.1.1
REVIEW OF ELASTODYNAMICS
Strain tensor
Subject to forces, solid bodies are deformed, and the distances between material points vary. Let us consider any point of a solid body, represented by its position vector x (components x 1 = x, x 2 = y, x 3 = z) which, after deformation, becomes the point
(l) It is well worth consulting the works of Landau and Lifchitz(l967), Ewing et al. (1957), Achenbach (1973), Germain (1973), and Mandel (1974).
50
i~
WAVE PROPAGATION
SATURATED POROUS
I
2
~EDIA
j
represented by vector x'. Displacement during transformation is therefore characterized by the vector (see Fig. 2.1): (2.1)
u = x'- x
-----
,,
...
.....
'
: ; ; -M /M' I /
''' . ,
'/" "
...
'
\
\ I
X'
';
I
I
'
I
I
'\
,/ .....
Fig. 2.1
-i
0
""-
{
1 d
'
I
I
2
___ , ,
I ,,
I
~-
I
E
A
•
I
.~
- VI
~
Deformation of a solid body.
li
g
t
The distance between two infm1tely close points before deformation was:
11
1
= (dxf + dx~ + dx~)2
d/
(2.2) ·~
After deformation it has become: dl' = (dx~2 which, according to (2.1), is written:
..... Jo
A
1
+ dx22 + dx32 )2
(2.3) 1
= [(dx 1 + du 1) 2 + (dx 2 + du 2 )~ + (dx 3 + du 3 ) 2)2
(2.4)
a
Substituting du 1 = u1.k dxk where the summation is considered to be on the repeated subscripts, as in the rest of this text, one then obtains: d/' 2 = d/ 2 + 2e1k dx 1 dxk (2.5)
v
dl'
where e1k is defmed by: 1
e;k
= :;- (u;.t + uk,i + u;.iul.k)
c p
(2.6)
tl
(e1k) is called the strain tensor. For small deformations, the only case considered here, the variation in distances between material points, and hence the variation in displacement, is small compared with the distance itself. In other words, the products of derivatives can be ignored in comparison with the derivatives themselves. Hence the linearized strain tensor is written: 1
eit =
2 (u 1,k + uul
tl
(2.7) £1
I
j
I
2
2
d
Let us now consider the variation in length in direction l. This consists in making in (2.5): d/' 2 = dx~2 , d/ 2 = dx~ et dxf = 0 if k -:I- I. This immediately gives :
I)
t
WAVE PROPAGATION IN SATURATED POROUS MEDIA
dx~
2
= dxi(l + 2£ 11 )
51
(2.8)
If b(x 1 ) is the difference in norms between dx 1 and dx~ divided by the norm of dx 1 , one obtains: dx~2
_,
1 ~ "'
l
(2.9)
Equations (2.8) and (2.9) then give: [I
;~
= dxi[l + b(x 1)JZ
+ c5(x 1)] 2 =
l
+ 2t 11
(2.10)
Assuming small deformations, b(x 1 ) is infmitely small and Eq. (2.10) leads to:
b(x 1) = e11
(2.11)
This shows that, assuming small deformations, the diagonal elements e;; are equal to the linear dilatation in the corresponding direction i. If we now examine the transformation of the scalar product ot two vectors dx and dy, which after deformation become dx' and dy', a similar argument as the one leading to (2.5) gives< 21 : dx'. dy'
= dx. dy + 2e;t dx; dyk
(2.12)
Assuming dx = dx 1 and dy = dx 2 , initially orthogonal, Eq. (2.12), added to the interpretation of e 11 and E: 2 , gives: (l - e11 )(l + ed cos (dxJ., dx2) = 2e 12 (2.13)
'.)
.-
Assuming small deformations: cos .)
(dx~. dx2) =cos
(i- ou) ~
012
(2.1~)
and (2.13) leads to :
:1
012 = 2e 12
(2.15)
Thus the non-diagonal elements characterize the change in angle between two basic vectors. As for any symmetrical tensor, the strain tensor has real eigenvectors. The directions corresponding to these eigenvectors are called the principal strain directions. These principal directions are orthogonal (property of eigenvectors) and remain so throughout the deformation, since, in the reference system built on these directions, the strain tensor is diagonal. In this reference system, the diagonal terms that we note e1 , e11 and e111 represent the linear dilatations in the principal directions, and are thus called the principal strains.
)
·)
e s
e r (2) Equation (2.3) shows that :;;.-= strain tensor ~ defmes a bilinear form such that: c:(dx. dy)
...,.j
= l/2(dx' . dy' - dx . d~ '·
~
2
·t
The elementary volume, built on the principal directions, is d"'- = dx 1 dx 11 dx 111 and is transformed into:
~;
52
WAVE PROPAGATION IN SATURATED POROUS MEDIA
d't"'
The volumetric strain
= (1 + e 1)(1 + e 11 )(1 + e 111 ) d't"
(.2.16)
e. to the nearest second order, is therefore: 8 = tr B = B1 + Bu + B111
12.17),
The trace of the strain tensor is a tensorial invariant. Owing to defmition (2.7), we therefore have:
e=
l
.;
(.2.18)
div u
Hence assuming small deformations, the volumetric strain corresponds to the displacement divergence.
i ·
2.1.2 Stress tensor and equilibrium equations If a body is deformed under the action of external forces, elementary forces, called stresses, are generated to oppose this deformation. More speciftcally, let us consider a point M in a deformed solid, and subdivide an elementary cube (Fig. 2.2). z
z ayz
axz
~I
I
dz
I
J-t::=::...;;.;.;._-x J
v
;
6
dx I
azx
v
v
Fig. 2.2 Stress tensor. Defmition of stresses applied to the faces of an elementary cube dx dy d:.
The contiguous parts of the body exert elementary surface forces on the faces of this cube. The j'h component of the force applied to the face whose normal is the ;•h direction is denoted u 1i. The series of (uii) constitutes a tensor called the stress tensor 131 : q
=
qJCJC
Uyx
Uxy
Uyy
Uz:x) Uz:y
Uxz:
Uyz
O'z:z
12.19)
(
(3) As defmed here, the tensor is the contravariant representative, whose components should be denoted u"" ... In Cartesian coordinates, however, the contravariant and covariant representatives of a tensor merge and we adopt the notation u,.,. or u".
'
j ---------------------------
2
53
WAVE PROPAGATION 1:-J SATURATED POROUS MEDIA
Due to the tensorial character ofthe stresses (Mandel, 1974), the i'b component F of the w tten:
force applied to a face whose normal is defmed by the unit vee F;
= u;ini
(2.20)
If the quantity F;n;. which corresponds to the projection of this force on the normal, is positive, this represents a tension, and. if it is negative, a compression. Projections in the { plane of the face (such as u;i for j ¥ i) are called shear stresses. Under the action of external forces, a stress f1eld develops within the solid. The f1eld must satisfy the local equilibrium. This represents the case in which volumetric forces are absent. Let us now write the equilibrium of an elementary cube (Fig. 2.3) subjected in the dynamic case to inertia forces - pu; dx 1 dx 2 dx 3 where pis the density. In the 1 direction this gives:
f t
dx 2 dx 3
uu -
•
t' I
(u 21
+ 0'21
+ u2 1.2
dx 3 dx 1 + u 31 dx 1 dx 2 - (0' 11 + 0' 11 , 1 dx 1 ) dx 2 dx 3 dx 2 ) dx 1 dx 3 - (0' 31 + u 31 , 3 dx 3 ) dx 2 dx 1 = pu; dx 1 dx 2 dx 3
(221)
Hence: O'u.1
+ a21.2 + u3,.3 =
(2.22)
pu;
with the corresponding equations for the remaining directions 141• 1122+ 1122,2 dx2 1121
~
x2
0 23 + 11 23,2
+ 0 21.2 dx2
dx2
11 12+ 11 12,1
dxt
~ uu+ott,1 dx1
dx2 11 13+ 0 13,3 dx3
dx1 0
22
L--------:-----x, )(3
Fig. 2.3 Equilibrium of an elementary cube (after Fung, Foundations of solid mechanics, p. 65. © 1965. Reprinted by permission of Prentice Hall, Inc., Englewood Cliffs, NJ).
(4) Note that, as it has been defmed. this stress tensor relates to the present (i.e. deformed) geometry. Strictly speaking, the equilibrium equations are all related to this geometry. However, assuming small deformations or, more precisely, small displacements, the initial and present geometries can be merged in writing this equilibrium.
I I
_..,1
l
r
~
t
54
WAVE PROPAGATION IN SATL"RATED POROUS MEDIA
2
The equilibrium of the moments shows that, in the usual case without volumetric distribution of moments, the stress tensor is also symmetrical:
= (jji
(fi)
(2.23)
.i i'
~" ~
!l
This symmetry allows us to write the equilibrium equations in the form:
= piii
uii.i
(2.24)
Since the stress tensor is symmetrical, a principal reference system can be deftned at each point in which the stress tensor is diagonal. The diagonal terms u 1, uii and um are called the principal stresses.
t•
2.1.3
Constitutive law of linear elasticity
C·
Equilibrium equations alone are inadequate to solve a given problem, because only three differential equations are available for nine unknowns (six stresses and three displacemc:nts). The constituent material is thus involved by its constitutive equation, which links stresses and strains. The simplest equation is the equation of linear elasticity: =
(fij
(2.25)
cijklekl
This equation is the simplest for the case of reversible behavior. In fact, it corresponds to a linear response around an equilibrium state. Owing to the symmetries of the strain and stress tensors, the tensor Ci111 necessarily satisftes: ciJ'"
= cjikl = cii'"
(2.26)
Furthermore, we obviously have: cijkl
= c"lii
(u;1e;1
= uk,ek1)
(2.27)
This reduces the number of components of the tensor of elasticity Ciikl to 21 in the general case. For the isotropic case, that is to say when all stress directions are equivalent, only two constants are required, and Hooke's law is obtained: uii = ),
tr
e bii
+ 2p. eii
(2.28)
where A. and p. are called Lame's coefftcients, J.l is the shear modulus, and the term Jii is the Kronecker's delta: Jij = 0 bij = 1
i=Fj i=j
(2.29)
The inverse Hooke's law is alternatively written: 1+ v e-.11 = -E-
v
11
.
E tr u tJ 11..
q .. - -
(2.30)
where E and v are Young's modulus and Poisson's ratio, respectively such that:
J
2
·i
I
55
WAVE PROPAGATION (]'; SATURATED POROUS MEDIA
£l·
E
). = (1 + v)(1-2v)'
J1
= 2(1 + v)
or
(2.31) v=
3). + 2p E = J1 ----:--). + J1
1.
2(). +pi"
The quantities introduced all have a simple physical meaning. In a simple compression (tension) experiment, where only one stress is non zero (for example u 11 = u, the other u;i = 0), it is easily concluded from (2.30) that:
•I..
u
'I
''I
e22 = e33 = -
= Ee 11 t;i
l'£ 11 ,
=0
(2.32)
ifi#:j
Hence E characterizes the strain in the direction of the applied stress, while v characterizes the relative extension in the orthogonal directions. This is represented in Fig. 2.4 for a unit cube.
:ll
'!
a 11 =a
,c l
1i I I !
alE=
En
, ....
,
,'
/ +t I
I
)-.
--,..,. ,
I
I
v , ,J
)
,.. ,
-- -- -
_____ ..J.. .... I
_,-''
e22.,-w 11
Fig. 1.4 Interpretation of Young's modulus and Poisson's ratio.
Let us now conduct a pure shear experiment, in other words an experiment where, for example, u 12 = u 21 = t, with all the other stresses being equal to zero. Equation (2.30) immediately gives: £21
=
£12
1+ v = ~:
t
= 2p'
other
~:iJ =
0
(2.33)
The corresponding deformation is an angular distorsion. It has been shown that the tra~e of the strain tensor represented the volumetric strain 8. Taking the trace of the two members of Eq. (2.281. we obtain:
2p) e = K8 31 tr u = ('I. + 3
.... 4
(2.34)
56
WAVE PROPAGATION IN SATURATED POROUS MEDIA
2
Equation (2.34) thus linearly relates the mean hydrostatic pressure< 5l (1/3 tr u) to the volumetric strain. K is called the bulk modulus. In conclusion to this very brief review, note that elastic behavior may be introduced by assuming the existence of a strain potential 'V(eiJ). For an infmitesi~al strain de;j• the deformation work is by defmition:
av
j
I
(2.35)
d'V = - deiJ OB;j
~
and the stresses are thus identifted with:
l
~~
av
(2.36)
0"··=081j
IJ
In the case of small deformations, this potential can be linearized around a reference state that is assumed to be free of prestresses by only using the quadratic terms, and assuming isotropy, the potential Eq. (2.28) is written: 2'V = (). + 2J.L)(tr e) 2 + 2J.L(tr e2 - (tr e) 2 ) (2.37) The potential is positive for any deformation, and thus imposes:
!
tl:t: ''
i 81 i
•• ~
2J.l A.+-~0 {
3
J.l~O
or
{
1 -l
2
-~ I
(2.38)
E~O
This was evident from the previous interpretation of the different coefftcients.
2.1.4 Linear elasticity and rock mechanics 2.1.4.1
J; iii
Jl I
''
!
Linear elasticity
The elastic constitutive law [Eq. (2.25)] is an equation between present-time values. In other words, the materials have no chronological memory. This is obviously due to the reversible, elastic behavior ofthe material. The question arises whether this reversible, and possibly isotropic, linear behavior applies to rocks. Figure 2.5 shows the main phases of the behavior of a rock during a simple compression test along axis ox 1 • An examination of Fig. 2.5 shows that characteristic stress values determine the behavior of a rock in a compression test. For stresses less than o1, thereis a crack closure phase. This phase is elastic (reversible) since no cracks are created. However. the stress/strain relationship is not linear (see Ftg. 2.5) at the macroscopic scale. In fact, the material tends to stiffen due to the closure of cn:.cks at the microscopic scale, and this closure is reversible. Yet the degree of crack closure depends on its orientation in relation to the applied stress direction. The deformation of rocks. whose complexity was discussed
, ,
(5) This denomination is due to the fact that (tr u)/3 = - p for a body subjected to hydrostatic pressure p.
I
.J
2
57
WAVE PROPAGATION IN SATURATED POROUS MEDIA
a,
a, 1 .~
Unstable crack growth
?
"I
fj
a2 =a3 :o
M a 1
Stable crack growth
CriCk initiltion
---------------------.q- ,, __ _ CriCkclosure
:f!
J!
.,
...
t:
;11
·I
I
; I !
!
t
,,
Str1in
Fig. l.5 Main phases in the behavior of a rock during a simple compression test (after Panet, 1976).
~
tl
e1
0
in the f1rst Chapter, has no a priori reason to be a linear function of stress in the crack closure phase. A linear elastic phase occurs for u 1 varying between o1 and uf, when the crack closure phase has been achieved. For stresses from o{ to u{-.,.a stable crack propagation phase occurs. This phase is not reversible since cracks are created, and the material retains the memory of this new cracking. These different phases are clearly exhibited in the curves showing the volumetric strain 8 = 8 1 + 2t3 or transverse dilatation 8 3 as a function of u 1 lreversible only between 0 and o{ and linear only between o1 and o{). Unstable crack propagation occurs above u{-, followed by fracture when u = uf'. The more homogeneous the material, the closer of is to uf'. By using Eqs. (2.30) to (2.~). the different curves help to determine Young's modulus E, the bulk modulus K, Poisson's ratio v, and the shear modulus p.. Some experimental values at atmospheric pressure taken from Angenheister (1982) are given in Table 2.1: TABLE
2.1
EXA~IPLES OF ELAmC MODUU FOR DIFFERENT ROCK lYPES
(from Angenheister, 1982)
Westerly granite .................................. Tholeitic basalt. .................................. Solenhofen limestone (dry) ......................... Ottawa sandstone (dry) ............................ -
.....
-- -
K (GPa)
(GPa)
Jl.
E (GPa)
v
19.8 59.6 53.2 0.52
18.8 31.9 25.4 0.54
43.0 81.2 65.8 1.20
0.14 0.27 0.29 0.11
58
WAVE PROPAGATION IN SATURATED POROUS MEDIA
2
The lithostatic pressure of in situ rocks generally lies. between 11~ and 11f. The slight variation in stress L111 due to the propagation of a wave is always such that 11 1 + L111 < 11f. In fact. Winkler (personal correspondence, 1979) showed that the strain amplitude L1e of conventional seismic sources (artificial or natural) was Jess than 10- 6 in the far field (i.e. approximately two wavelengths from the source). Given the stress/strain relationship, the corresponding variation L111 is thus less than 1 bar (0.1 MPa). Hence. by ignoring the effect of strain rate, it may be considered that elastic behavior can reasonably be adopted for the problem of wave propagation. This leaves the problem of the linearity or non-linearity of this behavior according to whether 11 1 , the experimental compression, is lower than or greater than cG· Considering the minimum value of L111, the approximation of the non-linear curve 11 1 (e 1 ) by its tangent to the ordinate 11 1 appears to be perfectly justified 16 l. The remaining problem is now that the different elastic moduli for a stress 11 1 < cG in the laboratory are not those in situ (lithostatic pressure, 11 1 > cG). This explains the need for pressurized tests in the laboratory. It must be stated that the experiments described above are only static experiments (zero frequency). The time dependence of the constitutive law for a given stress is a fundamental factor in the dynamic measurements discussed in this work. This dependence will be dealt with in subsequent Chapters.
2.1.4.2
Rock mechanics and effective moduli
In the foregoing Section, the rock was considered to be a homogeneous material with given elastic moduli (see for example Table 2.1). In fact, as we have pointed out on several occasions, a rock is an aggregate of solid elements between which the pores may or may not be saturated with liquid. Each one of the constituents (s.olid an·d liquid) of the rock is associated with different moduli. The shape of the pores is the most important property determining the bulk elastic properties of a rock. It is easy to see why a .. spherical'' pore will be more resistant to a uniaxial stress than a .. flat" pore or microcrack (see Chapter 5 for experimental examples). For a given mineralogical composition, the .. bulk" value of the elastic moduli and the effective value reflect the type and shape ofinclusions and pores in the sample analyzed. An abundant literature exists concerning the effect of inclusions on the elastic properties of solids. Eshelby (1957), Kuster and Toksaz (1974) dealt with the general problem of an ellipsoidal elastic inclusion. Walsh (1965 and 1969) and Wu (1966) tackled the problem of heterogeneities in the form of ellipsoidal cracks, and the results obtained offered a satisfactory way to reproduce the behavior of rocks under pressure (see Section 5.1.1.1). O'Connell and Budiansky (1974) and Budiansky and O'Connell (1976) resumed the same type of investigation, using a self-consistent method: the effective modulus is obtained by an implicit equation, with the pores already assumed to be included in an effective medium. Finally, Mavko and Nur (1978) discuss inclusions of a more random shape.
~
·;.I
'i I
-'!
·; i
tl
f:
IJ, > I ;J \
,,
_,- i\
~{
I
(6) Note that, in a homogeneous linear elastic medium (no defects and invariable elastic constants), the relative variation in wave velocity as a function of applied stress is totally negligible (the relative variation is comparable to the deformation created by the stress. Marigo, 1981). In a rock, however, the only signiftcant factor is the closure of the cracks.
.... ---------·
.
·------
2
i
''
f
W.... \"E PROPAGATION IN SATI:RATEO POROUS MEDIA
59
The theoretical developments associated with the determination of effective moduli aE use the same type of assumption, namely that each inclusion (gas or liquid) is a closed system. Hence the result is independent of the permeability of the medium. Biot's theory. which we shall discuss in Section 2.2, accounts for the permeability effects.. providing a so-called open system. Hence the theory of effective moduli is not a specifiC case ofBiot's theory. but a different way of dealing with theoretical problems. For a liquic inclusion (saturated rock) and a gaseous inclusion (dry rock), it is possible to determine values of effective bulk moduli, but these two moduli are in general not linked by a relationship such as Gassmann's equation (see Section 2.2.2.2). The theory of effective moduli must therefore be used with caution when dealing with any process in which permeability plays an important role.
.;, \';
i'
2.1.5 Wave propagation in an isotropic linear elastic medium 2.1.5.1
Wnes in a 3D space
By introducing (2.7) into the constitutive equation (2.28) and then the result into the equilibrium equations {2.24), the following equations of motion are obtained:
{). + 21-') grad div u -
11 curl curl u
== pi
(2.391
Let us now consider irrotational movements (curl u = 0) defmed by a potential iP such that: (2.4(11
u =grad 4> This equation introduced into (2.39) gives 171 : J724>
=
.
~~ c}S
(2.411
V,.=e: 2~-~y 1
(2.4:1
Equation (2.41) defmes waves propagating at the velocity V,.. These are called dilatational waves (or compressional waves) because they concern the propagation of volumetric strain. By applying the Laplacian operator J7 2 to Eq. (2.39) fc>r u satisfying (2.40), and using Eq. (2.18), we actually obtain: 2
V 8
= ~~ iJ
(2.431
These waves are also called P waves, where P corresponds to primary, because these aT! the fastest waves likely to propagate in an isotropic linear elastic medium. Let us now consider motion defmed by a vector potential , such that: u = curl ,
(7t ~ote that div grad~= 17 2 ~. hence div.
...4
=8 = r
2
~•
(2.~'
t
60
WAVE PROPAGATION IN SATURATED POROUS MEDIA
!
2
~
;·I
These movements correspond to motion without a volume change or equivolumetric motion, because: (2.45) 8 = div curl 'I' = 0
·h
J }
Introducing (2.44) into (2.39) gives: (2.46)
172'1' = __; "' Vs 1
(2.47)
Vs = (JJ./p)2
I
'
Equation (2.46) defmes waves propagating at the velocity V5 . These are called shear waves. They are also called S waves, where S stands for secondary, because they are slower than the P waves. Note that the ratio V~/V~ can be written as a function of Poisson's ratio v in the form: v~
'<
,2(1- v)
v~ =\.1 - 2v I
Now, studying the motion of material particles due to the propagation of the wave, by using the change in variables:
a
1
=t- V
(lx
+ my + nz) (2.48)
1 P= t + V (lx + my + nz) (1
with
+ m2 + n2
and noting that, for any function F(a,
= 1,
V
= Vp
or
Vs
I F ,x =-(Fp-F' v . ·"'
F = F,p +F.«
··-
P>:
we obtain that a solution of (2.41) and (2.46) of the type F(a,
etc.
(2.49)
P> will verify:
4F,,.1 =0
(2.50)
where F = 4' or '1'. When integrated, this equation implies that: F=
f(r-
lx + my + nz)
v
lx + my + nz) +g ( r+---===v--
(2.51)
For f1xed t, F is constant for the plane lx + my + nz = constant. Hence the waves defmed by (2.51) are plane waves whose wave fronts are planes lx +my + nz =constant. For F = 4', Eqs. (2.40) and (2.51) show that the particle motion takes place perpendicular to the wave fronts. The wave polarization is called longitudinal. For F = '1', Eqs. (2.44) and (2.51) show that the particle motion takes place in the planes constituting the wave fronts. This polarization is called transverse. These results, obtained for plane waves, are actually generally applicable to any wave fronts (see Appendix 2.1). Polarization is either normal to the wave front for P waves, or in
.J ~·~--
2
61
WAVE PROPAGATION IN SATURATED POROUS MEDIA
the tangent plane for S waves. This is why P waves and S waves are also called longitudinal and transverse waves respectively. At this point, we have assumed that the medium is isotropic (see Table 2.2 for relations between different elastic parameters). in other words one whose properties are identical in all directions. This is not generally valid as the properties of a rock are often anisotropic. The study of wave propagation in the most general linear elastic medium, thus satisfying (2.25) and not (2.28), calls for a development other than the one carried out in this Chapter. This development is given in Appendix 2.1.
'·
'~
t
2.1.5.2
1 D wave equation (elastic case)
Let us consider a rod element of constant cross-section, lying between sections I(x) and l"(x + dx) and bounded by the surface a..r, with the latter free of load (see Fig. 2.6).
I
l ;f ~
6I
z f
'-
i
~'
0
~
~ ::::1_
l:(udxl
~n +a,,..,
xx
Fig. 2.6
Bar element.
~
Let us now examine the propagation of disturbances corresponding to ftelds of uniaxial ~tresses along the ox axis and uniforin in cross-section. These ftelds are therefore defmed
by a single stress uxx. The dynamic equilibrium of the element gives rise to the equation of motion: ti)CJC,J<
(2.52)
= pii
where u is the displacement along ox. For a uniaxial stress fteld, it has been shown [see Eq. (2.32)] that: (2.53a)
tl"" = Etxx = Eu,x £ 71
= Su = -
V£""
Sij
= 0
if
i =F j
(2.53b)
Equation (2.53), introduced into (2.52), then gives: I
..
u,%% = v~ u
(2.54)
VE=l
(2.55)
where
' .....
.
•
F.
2)
3 ---·----
3R~ +I
9pV~R~
-
3K(I - 2v)
2
v = (Rf - 2) = (3R~ - 2) = 2(3R~ - 1) (Rf - I) (3R~ + I) (3R~ + I)
'
4 8 pV,.--V
( 2
3(1 - 2v)
..
-
Jl 1(1
2/1(1 + v)
(I t-v)(l-2v)
2(1 + v) 2v)
).
-
-
v
.
31 + 2J1 Jl~ K-A. 9K-3K- A. 9KJ1 3K + 11
E
3v
). I+ v
-
EJ1 3(3J1- £)
-
-
A.+ 2J1/3
K
TAIII.I'
2.2
2v ;>,.
-
p(V~- 2V~)
(I + v)(l - 2v)
3K ~--l+v F.v
I' I
-
E- 2J1 Jl-----3J1- E 3K -E 3K-9K-E
K- 2Jt/3
-
-
A.
.
K
).
See below •
-
-
-
-
3K-E 6K
Ef(2Jt)- l
2(3K 1- 11)
3K- A. 3K- 211
2).
-I'
v
,,'~.i.
--~··
------
(I + v)(l - 2v)
£(1- 1')
I- v 3K --l+v
2 - 2v 21· I' I
I
).---
4J1- E Jt·3J1- E 3K + E 3 K 9K-E
K + 4tt/3
3K-
A.+ 2J1
A. --2(1 + Jl)
pV~
-2 -{IV p
v
H: Young's modulus v : Poisson's ratio ). : Lame's constant V,. K 2 2 R• = -V., R2 = -V2' Rl p :density S (I S
K : hulk modulus JJ : shear modulus
(NuR, personal correspondence)
RELATIONSHIP BETWEEN ELASTIC CONSTANTS OF AN ISOTROPIC BODY
--
I
. -2 + 2v 1\
I - 21·
:!1• ---
21'
··,-..p~r;.,·.--.,.'
--------·--
2 + 21'
JK
).
3KE 9K -E
-
-
J(K- A.)/2
pV~ = J1
.:.r-~--,'!
----~-----~-
r ' i !:.
2
WAVE PROPAGATION IN SATURATED POROUS MEDIA
63
Equation (2.54) corresponds to longitudinal waves (without shear) with velocity 1'&, where E stands for extension. Note that, in comparison with an infinite three-dimensional medium, in which this type of wave propagates at velocity V,. [see Eq. (2.42)], we have:
Vi
. v; =
(l
+ v)(l
- .2v) 1- ,.
(2.56)
Note that this elementary theory ignores the inertial effects induced in the oy and oz directions by transverse deformations (2.53b). Hence the three-dimensional equilibrium equations (2.24) along these directions cannot be satisfted assuming a uniaxial stress fteld. In the case of sufftciently low frequencies (large wavelengths in comparison with the transverse dimensions), these inertia effects remain negligible, and Eq. (2.54) can be accepted as a fust approximation.
''
t 2.2 WAVE PROPAGATION IN SATURATED POROUS MEDIA: BlOT'S THEORY
i At this point we have only considered single-phase (solids) media. We shall now extend the study to fluid-saturated porous media. To do this, we shall present Biot's model and its experimental confumation by Plona. This will help us to identify an attenuation mechanism for compressional waves. In dealing with the problem of acoustic wa'l'e propagation in saturated porous media for dynamic analysis of the subsurface, two approaches are possible: • The ftrst approach draws on homogenization processes, which help to pass from microscopic laws to macroscopic laws. The term microscopic is used here to apply to laws governing mechanisms at the scale of the heterogeneity (of porosity in our case), whereas macroscopic laws refer to a scale related to the heterogeneous medium concerned, identifted as representative of the mechanisms investigated. So far as we are concerned here, this scale is in fact the measurement scale (see Section 1.5). In these preliminary approach methods, our discussion will be limited, and the reader can refer to the works of Suquet (1982) and Andrieux (1983) for a more extensive review. Briefly, we shall note that two homogenization methods essentially exist. One of them is based on an averaging process: a microscopic problem is fmt resolved at the level of an elementary cell containing an isolated heterogeneity (a fluid-ftlled channel in our case). From the solution to this elementary problem, we then infer the mean value on the cell of the quantity analyzed (stresses, strains, energies or relative flow velocity) as a function of the macroscopic value imposed at the cell boundary (strains, stresses or velocities). The actual heterogeneous medium is then replaced by a ftctitious homogeneous medium. The response of the latter to an imposed force is the mean value previously calculated. The function linking them depends spatially on the geometric and mechanical parameters of the heterogeneities existing in the actual medium. This method is quite effective for low and medium concentrations of heterogeneities, for which cell-to-cell interaction processes
''"'
l.
:'
..
64
WAVE PROPAGATION IN SATURATED POROUS MEDIA
2
can be ignored. This procedure was used in particular by Biot ( 1956 b) outside the framework of the theory that we shall present here, to characterize the flow of a fluid in a porous medium. The second homogenization method relies on the assumption of periodic repetition of the microscopic heterogeneous structure, imposing the periodicity of the solutions. By making this spatial period tend towards zero with respect to the macroscopic scale (small parameter asymptotic method, Bensoussan et al., 1978), the form of the macroscopic laws is obtained. This method otTers the advantage of mathematical rigor, its systematic aspect, and its absence of concentration limitations. It has been used successfully in recent years (Lhy and Sanchez-Palencia, 1977, Levy, 1979)for the problem of saturated porous media. In parti~ular, it h~l~d. to generalize ~ar~ts law in ~on-steady-state con~itions. The assumption of penodicity may appear hm1tmg, but th1s method seems to Withstand the comparison with experiments conducted in random media (see Suquet, 1982). The drawback is that it furnishes only the form of the macroscopic laws, unlike the method based on average procedure which can provide analytical estimates within its applicability limits for geometries with simple heterogeneities.
i fJ
• The second approach consists in deliberately ignoring the microscopic level and assuming that the concepts and principles of continuum mechanics (existence of potentials and stationary principles in particular) can be applied to measurable macroscopic values. This older approach is discussed by Biot (1956a, 1962) for problems concerning our subject. Strictly speaking, it is only justified a posteriori by the agreement of the results that it provides with those obtained by the above homogenization method. Despite its more heuristic appearance, we have decided to present the results concerning porous media through this macroscopic approach, which has more physically realistic assumptions. In the ftrst part, we shall discuss those assumptions on which it is based. In the second part, after defining the strain potential and stresses and the dissipation pseudo-potential and kinetic energy, we shall derive the equations of movement using Hamilton's principle. In a third part, we shall examine the propagation of waves, their velocities and attenuations. We will also discuss the existence of a second compressional, or slow, wave, in addition to the standard compressional and shear waves. We shall then see the extent to which the mechanisms identified play a significant role, and the experimental and theoretical developments that they suggest.
2.2.1
Assumptions
The f1rst assumption states that the wavelength is large in comparison with the dimensions of the macroscopic elementary voluflle. This assumption, which is normally always satisfied in geophysical applications, is required to make a description of the processes analyzed by the tools of continuum mechanics. Hence the wavelength is large in comparison with the dimension of the elementary channels where microscopic flow occurs. It can then be shown that the stress distribution in the fluid is nearly hydrostatic, although viscosity plays a major role in the flow (Mandel, 1950).
... -----------------------------------------------------------------------------------
rf
t Jf
2
WAVE PROPAGATION IN SATURATED POROUS MEDIA
65
The second assumption is that of small displacements both for the fluid and solid phases. This assumption is fully justified, as the strains in seismic studies (laboratory or field) are less than l0- 6 • Hence if u; is the ;th component of the mean macroscopic displacement, the components of the macroscopic strain tensor can be written to the nearest second order: ll;i =
,i
~t
1
2 (u;.i + uj,;)
(2.57)
The third assumption is that the liquid phase is continuous. Thus the matrix consists of the solid phase and disconnected pores. In the following discussion, the porosity under consideration is that of the channels in which the flow occurs. With respect to this porosity, which we assume to be isotropic and uniform, and which we denote cJ>, the medium is assumed to be fully saturated. Let U be the mean displacement of the continuous liquid phase contained in the macroscopic element. The elementary macroscopic flow rate dO through an area dS with a normal n and per unit time is given by: dO=
w
w. D dS
= c~>
(2.58)
where a dot denotes a derivative with respect to time. The vector wis the filtration velocity vector. Note that, for any macroscopic volume 0 with a boundary S, we have:
·.
I
w • n dS =
The local increase in fluid content
L
div w dO
~
is given by:
~
=- div w
(2.59)
(2.60)
The fourth assumption concerns the matrix which is assumed to be elastic and isotropic, with the understanding that the theory can be extended to the anisotropic elastic case. Hence all the mechanisms of viscous origin related to the matrix (such as those due to the presence of fluid in the disconnected pores) will not be dealt with. For the description of these mechanisms, the reader can refer to the works of Walsh (1969) and Budiansky and O'Connell (1976), or even the more formal work of Biot (1962). The anisotropic case was examined in a general theoretical manner by Biot and Willis (1957). The fmal assumption concerns the absence of any coupling and, in particular, the absence of thermomechanical coupling. Note however that this coupling has been discussed by Biot ( 1977).
......
66
2~2.2
2.2.2.1
WAVE PROPAGATION IN SATURATED POROUS MEDIA
2
Equations of movement Strain potential and stresses
Since the perturbation caused by wave propagation is a rapid phenomenon, the process is adiabatic and, from the general concepts of fluid mechanics, it is reasonable to presume (see, for example, Mandel, 1974) the existence of an internal volumetric potential V, such that its differential represents the deformation work in an infmitesimal macroscopic transformation. The previous assumptions give rise to the fact that the potential can only depend on the components ofthe strain tensor eii• and on the increase in fluid content e. These variables are normal variables, and for any infmitesimal transformation defmed by the increments deu, d~. we have:
av
d\1 = oeij deij
av
+a[ de
'i
(2.61)
where the summation, as below, applies to the repeated subscripts. The first term of the second member of (2.61) corresponds to the elementary deformation work at ftxed fluid content, and the second term to the work associated with the increase in fluid content in a transformation at ftxed macroscopic deformation. Hence it is reasonable to defme a macroscopic stress tensor a and a mean pressure in the fluid p by:
cv
Uij
= C~£ .. I)
(2.62)
cv P =
ce
·>
Assuming small disturbances, the expansion of V can be limited to the quadratic terms (see Section 2.1.3). The assumption of isotropy implies that this expansion only involves (for example see Germain, 1973) the flfSt two invariants of strain tensor 11 = tr e and 12 = 2(tr (8 2 ) - In, as well as the variation of fluid content One can therefore write:
e.
2V
= (A.I + 2p)lf + pi 2 -
2PMI 1
e+ M e
2
(2.63)
For P= ~ = 0, the single-phase case is obtained [see (2.37)]. The justiftcation of the form selected for the different coefficients will appear subsequently. The "positive defmite" character of the quadratic fomt associated with V, as a function of e;i and also implies (Biot, 1962):
e.
p>O
i. 1 -P 2 M+~p>O
M>O
(2.64)
From Eqs. (2.62), it can be inferred that: uii = ).1 tr
p = M (<) .• = { I)
---,------
01
8
Jii + 2p E:ii - PM ~bii
p tr 8 + e) if if
i"#j i=j
(2.65) ~\
--,--2
WAVE PROPAGAtiON IN SATURATED POROUS MEDIA
67
which can be alternatively placed in the form:
+ 2p. eii -
aii = i. 0 tr e ~ii
f3P~ii
1
(2.66)
e =-p+f3tre M
where i. 0 = i, 1 - /3 2 M. The interpretation of the different coefficients is thus straightforward. The coefficient Jl. is the classic shear modulus. As the fluid does not respond to the shear forces, it corresponds to a shear experiment for a system indifferently closed or open. The coefficient i.1 is the second Lame coefficient in the case of a closed system, in other words for constant water content [ { = 0 in (2.65)]. Hence it is linked to the elastic constants of the matrix and also to the Ouid compressibility. The coefftcient ),0 is its homologue for an open system [p = 0 in (2.66)]. Coefficient M is the pressure to be exerted on the fluid to increase the fluid content of a unit value at isovolumetric macroscopic strain [tr e = 0 in (2.65)]. Coefftcient f3 quantiftes the proportion due to the variation in fluid content in the apparent macroscopic volumetric strain tre for an open system [p = 0 in (2.66)]. It is therefore linked not only to porosity but also to the geometry of the channels where flow occurs. Biot and Willis (1957) have discussed the experimental determination of these coefficients.
2.2.2.2
Gassmann's equation and Biot's theory
Let us introduce the closed or saturated K 1 and open or dry K 0 bulk moduli as a function of the Lame parameters already defmed :
.
2
+ J Jl.
(2.67)
2 . K o=l.o+3JJ.
(2.68)
KI =
l.f
It is interesting to relate the four parameters {3, M, K 0 and K 1 , which are not independent (K 0 = K 1 - {3 2 M), and the bulk moduli of the fluid K 11 and the solid skeleton K •. Let us ftrst consider an experiment at zero pressure in the fluid, so that p = 0, and at imposed macroscopic hydrostatic pressure in the sample, so that - (tr a)/3 = u. From Eq. (2.66) and the previous considerations, it can be inferred that the actual strain is: tr e - {
= (1 -
f3) tr e =
-
q
(1 -
/3)Ko
(2.69)
In the fmal analysis, this strain is merely the strain of the solid and hence equal to
- afK •. This gives: K 0 = K.(1 -
/3)
(2.70)
Let us now consider a second experiment in which the mean macroscopic hydrostatic pressure - (tr a)/3 = p is also the pressure prevailing in the fluid. From Eqs. (2.66) and (2. 70), it can be inferred that: (2.71) - p = K, tr E
68
WAVE PROPAGATION IN SATURATED POROUS MEDIA
Recalling that
2
eis [Eq. (2.60)]: .; = -
cp div (U -- u) =
-
+ cp
tr
t:
(2.72}
and using (2.66), we obtain:
cp div U
..!!... = M
cp) tr t:
- ({J -
(2.73)
It is also known that, under the assumption of small displacements, the defmition of K 11 yields:
(2.74)
- p = K 11 div U
Replacing in Eq. (2. 73) div U and tr e by their values as a function of p [Eqs. (2. 74) and (2.71)], we fmd:
{3-cp
1
cp
-=--+M
· K,
(2.75)
K 1,
Equations (2.70), (2.75) and the equation: K 0 = K 1 - {3 2 M
(2.76)
give us K 1 as a function of K 0 , K 1 , and K,:
1
Kf
1]
1
1
cp [ K,- K 11 + K,- K 0
:0 [~. -:IJ + ~. [ ~. - ~]
(2.77)
This equation is known as Gassmann's equation ( 1951 ). Gassmann obtained it directly by considerations of elementary elasticity, without requiring the use of p and M. The reader can refer to White (1983 b). It should be emphasized. however, that it is implicitly understood that the porosity is uniform throughout the sample. Note that Eqs. (2.70), (2.75) and (2.77) are quite compatible with limit cases. For a solid medium corresponding to P = cp = 0, the expected values K 1 = K. = K 0 and M -+ oo are obtained. For a fluid medium corresponding toP= cp = 1, we obtain K 0 = 0 and KI=Kit=M.
We have stated that a fluid medium corresponds to P= 1 and K 1 = K 11 = M ( = ).1 because J.1. = 0). These equalities can help to understand the significance of the displacements U and u. In order for a medium to be a fluid. the macroscopic stresses given by (2.65) must be identified with - pbii• hence: ().1 - {JM)
tr t: =
M~({J-
II
(2.78)
Since this must be satisfied whatever the local increase in fluid content eand whatever the apparent macroscopic volumetric strain tr t: , the previously mentioned equalities are obtained. P = 1 and i.1 = M. The value of P = 1 introduced in (2.65), jointly with (2.58) and (2.601, where cp = 1 and tr e = div u, then leads to: - p = M di\' U
(2.79)
----------------------
1 2
WAVE PROPAGATION IN SATURATED POROt.:S MEDIA
69
which corresponds to the equation of perfect fluid behavior, for small displacements, where M is the fluid bulk modulus. This helps to understand why U, the average fluid displacement, is not identifted with the macroscopic displacement u, when the medium is totally fluid. In fact, ifU were equal to u, the rise in water content would be~== Oand(2.78) would only yield ).1 = (JM. Thus (J would not ncecessarily be equal to 1 and one could not obtain (2. 79). In fact. u is the macroscopic average on the reference geometric element considered, while U is an average of the displacement oi the fluid contained in this element and of the fluid which has left or has entered it 18 l. Simultaneously, in the general case, u is not the average displacement of the skeleton, but the average, on the reference geometric element, oi the displacement of the fluid part and of the skeleton part.
2.2.2.3
Dissipation pseudo-potential
Dissipations are assumed to result only from the relative fluid/matrix movement. In the neighborhood of equilibrium, it can be stated that the flow vector wand the associated force X are linked by a linear equation such as:
w1 =
:KiJXJ
(2.80)
Onsager's principle, based on microscopic considerations of symmetry, also stipulates that the tensor :K = (:Kii) is symmetrical. Based on (280), a dissipation pseudo-potential [)can be introduced (for example see Germain, 1973),which is a positive defmitequadratic form of representative matrix -!"- 1 , such that: 1r . • []) =- w:K- 1 w
2
Xi=
-
ao ow,
(2.81a) (2.81b)
The term pseudo-potential, due to Eq. (2.81b), recalls that the concept is only justified in the vicinity of thermodynamic equilibrium. In the isotropic case, the tensor :K - 1 is proportional to unity (-!" - 1 = :Kl) and the previous equations become : 1 .2 [) = 2:£ Wi 1 .
--w, X ,:K
(2.82a) (2.82b)
Darcy's classic law can be recognized in Eq. (2.82b) if :K is identified with the hydraulic permeability of the medium and force X with the opposite of the pressure gradient. As shown below, this identification is only possible in steady~state conditions.
(8) This difficulty stems from the fact that the macroscopic des."7iption is a priori Lagrangian, whereas that concerning the fluid is Eulerian. Nevertheless only small displacemnts are considered for the solid and fluid and the two descriptions are equivalent.
70
WAVE PROPAGATION IN SATURATED POROt:S MEDIA
2.2.2.4
2
Kinetic energy
Since the wavelength A is assumed to be large in comparison with the dimensions of the macroscopic elementary volume dQ (i.e. dQ - / 3 and l, .1 ~ 1), one can restrict the expansion of the volumetric density C of kinetic energy to the quadratic terms: 2C = p,ii;U; + 2p.,,..U;Ii·;
+ p,..li';li·;
{.2.83)
The absence of terms such as w;wi is due to the assumed isotropy. The specific case in which no overall movement occurs (i.e. w = 0) serves to identify p., with the average density in Eq. (2.83): p.,
= p = (1- c/J)p, +
{2.84)
where p. and p1 are the matrix and fluid densities respectively. The terms Puw and p,.. will be identif1ed subsequently.
2.2.2.5
Equations of movement
Let l be the volumetric Lagrangian density deftned by: 12.85)
D..=C-\1
Hamilton's principle (for example, see Achenbach, 1973) ~tipulates that, among all the possible paths between any two points in time t 1 and t 2 , the one that prevails will be the one that makes stationary the integral over time and space of the Lagrangian density and of the work ofthe dissipated forces, where the latter are derived from a (pseudo)-potential. Ifl depends on the parameters q;, i]; and qi,i• and the potential of dissipative forces UJ on i]; . . the variational calculus reflects this condition by the classic Euler equations:
a aL a aa.. cL aiD --+----+-=0 at aq_l
axj aqi,j
cq;
oiJ;
12.86)
where q1 = u1 or w; in our case. The application of (2.86), by using the expressions of ID (2.81a), C and V (2.83) and (2.63), leads to the equations of movement: uli.i =
- P.i
pii;
+ p.,..,w;
1 = P.,wul- + p,..w;- + .)('
(.2.87a) .
W;
(.2.87b)
These equations serve to identify the parameters Puw and Pw· If there is no average relative fluid movement with respect to the overall macroscopic movement (i.e. w = 0): the f1rst equation is reduced to the equation of movement in a continuous medium (see Section 2.1.2), while the second must be reduced to the equation of movement in a fluid. thus Puw = p1 . Now if the fluid is at rest (w = - )ii;-; il;
--
-
~-
- - -
-~~~~~~~~~~~~~~~~~-
12.88)
t '~
r I
2
WAVE PROPAGATION IN SATURATED I'()ROUS MEDIA
71
This shows that, if overall acceleration occurs. ~ force must be exerted on the fluid to prevent its average displacement. For its inertial rart, this coupling force is:
j
(p 1 -
p,..>)iif
(2.89)
To describe this coupling effect, similar to the qtass effect added in the analysis of the movement of an obstacle in a fluid, it is usual tO introduce a parameter a called the tortuosity parameter 191 such that: a
12.90)
Pw=;pP! I
One must necessarily have: a~
(2.91)
1 I
because the coupling force (1 - a)p1 ii1 must o~viously be opposed to the overall acceleration. This condition (2.91) ensures the n~n-negative character of the quadratic form associated by (2.83) with the kinetic ener*y density C. Like parameter {J, the tortuosity a is related not only to porosity but alsd to the geometry of the medium where the flow occurs. Hence the equation relating a to 4> ~s not biunivocal. Note, however. that a · must tend toward 1 as
sed: a
1[1 i:]
= l ;j; +
(2.92)
As already observed, the equation linking a w1th
F(c/J)c/J,
(2.93)
+ p 1 ii~ + p,.l\·i)
(2.94)
Equation (2.87b) is thus written: wi
=-
f(P.i
As previously noted, it is only in the case of st~ady flow ii = w = 0 that Eq. (~.94) is identif1ed with the classic Darcy's law. In the unstdady case, inertial effects are added and the comparison of (2.94) with (2.80) shows that the force associated with w1 is: Xi= - (P.I + P!iii 1+ p,.l\\)
(2.95)
(9) This tortuosity related to a dynamic process is differenl in principle from that defmed in Secuon 1.4.1, which is only related to the description of the porous mediuiil· In reality, these two quantities are of the same order of magnitude, because the same geometric characteristjc is involved.
72
WAVE PROPAGATION IN SATURATED POROUS MEDIA
2
Hence it should be noted that, even in a case of a zero pressure gradient, a fluid flow can be generated by inertial forces alone. It is customary to derive the classic Darcy's law (in steady-state conditions) from the Navier-Stokes microscopic equations for a fluid of viscosity '7, by making an analogy with Poiseuille's 'law (Mandel. 1950). This gives rise to the expression of the hydraulic permeability or mobility in the form: K
f='1
(2.96)
where "· the absolute permeability, depends only on the geometry of the porous medium. Note however that, as it appears here, Darcy's law only results from considerations of linearity in the vicinity of equilibrium. Hence, as already pointed out (Mat heron, 1967), the linear character of Darcy's law, obtained by the standard procedure mentioned above, actually only results from the linear character of the Navier-Stokes equations for flows at sufficiently low velocity. Only the form (2.96) of f involves the law of linear viscous behavior which these equations imply. To conclude, this law in the form of(2.94), as noted in the introduction, can be obtained by a rigorous homogenization method, once again based on the validity of the NavierStokes equations, but in this case in unsteady-state conditions. Moreo,·er, this method shows that for wavelengths that are large only in comparison with the dimensions of the elementary flow channels, and no longer in comparison with the dimensions of the macroscopic element, as required in the method developed here, the permeability is no longer absolute, but related to the frequency of the wave which generates the flow. Previously this result had been qualitatively found by Biot (1956 b).
2.2.3
Wave propagation
The introduction of the equations of stresses as a function of displacements u and U in Eq. (2.87) yields the equations of motion in the form:
+ 211) grad div u + Y grad div U - 11 curl curl u = P11 ii + P12 0 + b(u- iJ} Ygrad div u + R grad div U = p 12 ii + P:z 2 0- b(u- VJ (i.
(2.97a) (2.97b)
in which we noted :
i. R P12
+ Mcp(cp- 2P) = l 0 + M(P- c/J) 1 • Y= = Mcp 2 , p 11 = p + cpp1(a- 2)
=
i.1
= tPPJ(1- a),
P22
= acppf,
.\lcfJ(P- >)
(2.97c)
cp2
b
=f
Note that Eqs. (2.97) can be written for the limit cases of the perfect fluid and the solid. In fact, if the medium is a perfect fluid, the set of parameters to be considered is cp = P=a= 1, i.1 = M and 11 = b = 0 (%-+ + Xi, '1 = 01. Equation (2.97a) disappears, and Eq. (2.97b) gives the dynamic equation of perfect fluids under the assumption of small displacements: M grad div u = p 1 0 (2.98)
2
WAVE PROPAGATION IN SATURATED POROUS MEDIA
73
Note also that Eq. (2.97a) disappears naturally and not by assuming U = u (see Section 2.2.2.1 ). Now, if the medium is solid, letting P = (jJ = 0, a -. + oo, Eq. (2.97b) then gi\·es:
I
,o =- a
(2.99)
~I
while Eq. (2.97a) gives:
~I
Equation (2.100) is clearly the equation of the dynamics of elastic solids, since the second member is cancelled owing to (2.99). Note that this cancellation takes place independently of the manner in which tortuosity tends to infmity and permeability to 0, since the equations are not related to these asymptotic behaviors. Moreover, if decoupling occurs (see Section 2.2.4), P=
11
ij
(A. 1 + 2J.L)graddivu- J.LCurlcurlu- p,.ii
2.2.3.1
=a
1 ()
+ b(u- lJ)
12.100)
Existence of a slow P wave
Let us f1rst examine the case without dissipation (b = 0) and let us consider waves such . that:
(Uu)
= (grad 4>1) = ad cJ) grad 4> 2 gr
12.101)
These waves correspond to dilatational waves (P waves) that are irrotational (curl u =curl U = 0). By introducing (2.101) in (2.97) with b = 0, the following equation is obtained for the potential vector:
R172cJ) - MiD =
0
12.102)
where R and M are the rigidity and mass matrices respectively:
"""
"""
R = """
(A. + 2J.L Y
Y)• M = (Ptt P12) """ P12 P22
~
12.103)
As the matrix R - l M is positive defmite and symmetrical, it has two real positive eigenvalues which we denote V~, and V~ •. In the eigenvector reference system, Eq. 12.102) can then be written :
J72cJ)* -
_I 0) ( v~.
_1_
iD• = o
12.104)
o v2P2
where cJ)* is determined from cJ) by the change in reference system. We can identify here two equations of decoupled waves, corresponding to the two Pwave velocities Vr, and Vp 2 , and to the two characteristic movements defmed by the eigenvectors of R - t M. It can be shown (Biot, 1956a) that one of the characteristic movements corresponds to a movement in which overall and fluid displacements are in phase, and the second to a movement in which the displacements are out of phase. The
74
WAVE PROPAGATION IN SATURATED POROUS MEDIA
2
wave corresponding to the latter case is called the slow wave or the wave of the second kind. This terminology derives from the fact that the associated velocity Vp,, as shown below, is much lower than the velocity Vp, of the in·phase movement waves, called waves of the f1rst kind. These waves correspond to classic P-waves, with which they merge in the absence of fluid. Now let us consider the shear waves (S waves) or isovolumetric waves (div u = div U = 0) such that: u) =(curl 'I' 1 ) = curl 'I' (U curl '1' 2
(2.105)
Equation (2.105), introduced into (2.97), leads to the equations: 2
17 'I' 1
-
2
'I' = _ where velocity
Vs
is given by:
Vs
vz1 'I' 1 = s
0
(P12)'1'1
(2.106)
P22
=( ~ P12)i P11
(2.107)
P22
Since the fluid does not respond to shear forces, it only influences the shear wave through inertial effects. Its movement is evidently in phase with the overall movement [see (2.106) where P121P22 ~ 0]. We have only discussed the isotropic case here. For the non-isotropic case, the results naturally extend as follows: along the principal anisotropy directions if a compressional wave exists in the purely elastic case, two compressional waves, one slow and one · standard, must be considered within the framework ofBiot's theory. From the qualitative standpoint, the shear waves are altered in the same way as in the isotropic case, in accordance \\ith the processes that we shall now discuss.
2.2.3.2
Wave velocities and attenuations
Let us now consider the general case with dissipative effects (b =F 0). Introducing (2.101) into (2.97), we obtain: R 17 2 $ = A cj,
where
~
+ Mii>
(2.108)
is the damping matrix given by:
A=(_:-:)
(2.109)
Let us examine the case of plane harmonic P waves propagating in the x direction, such that: ~~>1 ~~>2
= 4> 10 exp [i(kx- wt)] = 4> 20 exp [i(kx- wt)]
(2.110)
1
1·
I I
T I
2
WAVE PROPAOATION.IN SATURATED POROUS MEDIA
75
where cf> 10 and c1>;z 0 are constants, and·w and k are the angular frequency and wave number respectively. The introduction ofthese equations into (2.108) gives a system of two equations for cf> 1 0 and cf> 20 • The Kramer's determinant of this homogeneous system must be zero in order for cf> 10 and cf> 20 to have non-zero values. This leads to an equation linking k with w. This equation allows two solutions k,., and k,. 2 , which correspond to the flfSt and second kind P waves previously discussed, and which are now complex owing to attenuation effects: k,.,(w) = Re k,.,(w)
+ i Im
k,.,(w)
(2.111)
Introduced into (2.110), these equations show that the imaginary part is the attenuation coefficient a.,., such that the corresponding wave amplitude is proportional to exp (- :x,.,x). while the real part is related to the velocity v,.,: (Re k,.,(w)) V,.,(w)
=w
(2.112)
The same procedure can be followed for the shear wave. Figures 2.7 to 2.12 show the curves relative to these calculations. The P and S wave velocities are normalized respectively by v,. and Vs whose expressions are: 1
V,.=C'I~2J.Ly Vs
=
(~t
(2.113) (2.114)
These expressions correspond to the velocities of the P and S waves if no relative movement occurs between the fluid and the overall movement (closed system). These are reference -velocities corresponding-to the limit velocities of the farst kind P wave and the S wave, wlie~ the frequency tends towards zero. Effectively, within this quasi-static limit, the effects of inertia contrast and dissipative effects become negligible. The frequencies are normalized by a characteristic frequency fc which depends on the ftltering medium :
(2.115)
The attenuation coeffacients are normalized for the compressional and shear waves respectively by :
21t/c
a,.=--
v,.
21t/c
(2.116)
as=--
Vs
The curves in Fags. 2.7 to 2.12 represent the relative variations of velocities and attenuations for the three wave types, with a constant ratio of fluid bulk modulus to solid bulk modulus (i.e. K 11 / K. = constant) equal to 0.06. This value corresponds to a watersaturated siliceous matrix (Wyllie et al., 1963). We varied the ratio ofthe open system bulk modulus to the solid bulk modulus (i.e. K 0 / K. = 0.3, 0.6 and 0.9). The broken, solid and dotted line curves correspond respecti\'ely to the porosities
76
1.0006
WAVE
PROPAG.-\T!O~
r-
Fig. 2.7 Velocity of P1 waves of the fvst kind vs. frequency for different values of K 0 K. (0.3, 0.6 and 0.9) and different porosities (10%. 20% and 30%).
Porosit~'ll.
.·
10 -20
1.()()()4
vi'
H- --I 30
2
I"!' SATL:RATED POROUS !\lEOlA
.6~.··
1
Vp
v 1.0002
v
1.0000 I
se; !1?.7:':':':'
0
.03
• .12
.09
.06
.15
f/fc
l PorosityT ~ % .0010 Li ......
I 10
......
j20
I
·~
- · - I 30
I
K
~- .9 K - .6
I
I
,
,,
5
.9
ap1 •p
I
I
,
•"
,-
a/
.0006
;', .3 , , 2~ .6/77
Fig. 2.8 Attenuation of P1 waves of the f1rst kind vs. frequency for different values of K 0 iK, (0.3, 0.6 and 0.9) and different porosities (10%, 20°1o and 30% ).
... ... ,_. ~'.9~······ -···
.0002
0
.03
.06
.09 t/fc
.12
.15
2
77
WAVE PROPAGA110N IN SATURATED POROUS MEDIA
.20 , . . . - - - - - - - - - - - - - - - - - - - - ,
Fig. 1.9 Velocity of P2 waves of the seeond kind cslow waves) vs. frequency for different values of K 0 ! K, (0.3, 0.6 and 0.9) and different porosities (10%, 20% and 30%).
~ I
-
.15
Vp2
.10
Vp
.05
.00~----~------L------~------~----~
.00
.03
.09
.06
.12
.15
f/fc
.6
.9
I
1 1
.6
.6
.9
1 1 1
.6
.3
I
I
.5
.4
ap2
ap
.3
.2
Fig.l.lO Attenuation of P2 waves of the second kind (slow waves) vs. frequency for different values of K 0 /K, (0.3, 0.6 and 0.9) and different porosities (10%, 20% and 30%).
.1
,
,.......... I -----
20 30
.0
.oo
.03
.06
.09 f/fc
.12
.15
78
I. p~~~s-i~~ .I
1.006 ~
Fig. 2.11 Velocity of shear S waves vs. frequency for different values of K 0 /K, (0.3. 0.6 and 0.9) and different porosities (10%, 20%, 30%).
9'>% 10
20 1.004
2
WAVE PROPAGATION IN SATURATED POROUS MEDIA
~------I
I
30
Vs V 5 1ol
Ko
Forany - Ks
1.002
.,. ,.,. .- "'
_,-'
_,
.,""
""
.--'
... -··
1.000 .00
.03
.06
.09
.15
.12
fife
.006 ~------------------------------------~ Porosity
I
¢%
.005
10 20 30
.004
erg
as
.003
.002
Fig. 2.12 Attenuation of shear S waves vs. frequency for different values of K 0 /K, (0.3, 0.6 and 0.9) and different porosities ( 10%, 20%, and 30%).
For any
_.... ____ .,..._,_-"":""..............,
.001
;'
....
··········· ·············
.000 .00
..
Ko Ks
.03
.06
.09 fife
.12
.15
T I 'Cl
·'
WAVE PROPAGATION IN SATuRATED POROUS MEDIA
79
remaining parameters of the theory are inferred from the above by the equations of Gassmann (2. 70), (2. 75) and (2. 77) for M, fJ and K 1 , and by Berryman's equation (2.92) for tortuosity.
~
().0\
For the three wave types, the velocity can be observed to rise with increasing frequency. This is explained by the fact that the inertial forces increase simultaneously. Indeed, these inertial forces being different for the fluid and the solid part, they generate a differential movement between the fluid and the fluid/solid combination due to permeability effects (Darcy's law). It implies that the overall movement defmed by u involves less fluid, whatever the type of wave considered. Hence as the mass involved in overall movement declines progressively with increasing frequency, the velocity increases with frequency. However, the differential movement thus facilitated by the increase in frequency causes increasing dissipation. This dissipation is proportional to the square of the angular frequency. For the three waves, this means increasing attenuation with frequency. Note also the more highly attenuated character of the slow wave in comparison with the other two (see scales), a characteristic that also limits the increase in velocity V,., corresponding to the high frequencies. This leads us to the following paradoxical situation: the more the phenomenon might tend to occur due to the velocity increasing as the frequency increases, the more this phenomenon would be attenuated.
l"t~
l~
15
2
.I
i
From the general standpoint, note also that permeability only affects the abscissa scale, normalized by frequency f. defmed by Eq. (2.115). More speciftcally, it can be shown numerically that, for commonly used fluids (water, glycerine, kerosene), it is the parameter ,, and hence the viscosity, that is mainly involved, independent of the density or bulk modulus K 11 . As the permeability~ = K/'1 approaches 0 (or towards infmity), the frequency f. tends inversely towards infmity (or towards 0). The rise in the curves with increasing frequency on an absolute scale is accordingly less (or more) pronounced. This agrees with the physicalevidence: the lower (of ~igher) the.~rmeability, the less (or more) are the differential movements (fluid,'matrix) privileged and the less (or more) Biofs effects are pronounced. Once again, however, a paradoxical situation results: the lower the viscosity 'I and hence the higher the permeability ~.the greater the attenuation. This is fairly easy to understand, because the lower the viscosity, the more the differential movement may be pronounced (i.e. less fluid is involved in the overall movement) and hence wis greater and the dissipation given by (2. 72a) (but tempered by the inverse of ~) increases. Actually, therefore, there is no para
80
2
2
WAVE PROPAGATION I]'; SATURATED POROCS MEDIA TABLE 2.3 SoME VALVES OF BlOTS CHARACTERISTIC FREQL"ESCY
~
s
Fontainebleau sandstonem .. Fontainebleau sandstone''' .. Tight sand m . .............. Cordova Cream limcstonel2l . Sintercd glass 131 ••••••••••••
~
Characteristic frequency Porosity IPermeabili t y {mD) (%)
Water
Normal oil
Heavy oil
80 MHz 30kHz 1 GHz 4.5 MHz 42kHz
800 to 4000 MHz 300 to 1500kHz 10 to 50 GHz 45 to 230 MHz 420 to 2100 kHz
8 to 400Hz 3 to 15 MHz 100 to 500 GHz 450 to 2 300 MHz 4.2 to 21 MHz
~
t
'I= I cP) 141 ('1= 10to50cP)(41 ('I= IOOto500cP)( 41
5 20 8 24.5 28.3
to-• 1000 2 10- 2 9 1000
(1) Bourbie and Zinszner (1985). (2) Carmichael (1982). (3) Plona and Johnson (1980). (4) Viscosity 'I is expressed in ccntipoiscs (1 cP = 1 mPa . s).
A glance at the situation depicted in the table tends to show that the very porous Fontainebleau sandstone and sintered glass display the same behavior concerning slow waves. In fact, the slow wave is also sensitive to the pore radius through the skin effect. This fact will be discussed in the experimental veriftcation in the next Section. The curves in Figs. 2. 7 to 2.12 also serve to analyze the influence of the ratio K 0 / K. and of porosity
-----
'
~ ! I .
'
2
WAVE PROPAGATION IN SATURATED POROUS MEDIA
81
a function of porosity. Note also that, as for the shear wave, the tortuosity effects decrease with porosity (less mass added). This merely reinforces the effects discussed above, albeit slightly (see shear wave). A number of considerations can already be drawn from the previous analysis to identify the existence of this slow wave experimentally. It is ftrst necessary for the system to be open and for the permeability to be sufftciently high (low viscosity) for average fluid movements relative to the matrix modeled by this theory to be possible, and for the slow wave to be able to propagate at sufftcient velocities. A critical description of an experiment in which this slow wave was identified (Piona, 1982) is given in Section 2.2.6. A fmal remark concerns behavior at low frequencies. In this range, the term Mi:b, proportional to jl, can be ignored in comparison with the term A$, which is proportional to f. Equation (2.108) is thus reduced to: V2 $ = R - 1 A$
where
R -
-1
~
Knowing that Cl» = (
(R + y
b
-
- (i.
+ 2p)R -
:J
(2.117)
-R- y ) i. + 2p + y
Y - Y- i. - 2p
(2.118)
and substracting the two equations for cP 1 and cP 2 obtained
from (2.117), one obtains : C 0 V 2 (cP 1
-
cPz) =
cPz
(2.119)
where we noted :
+ 2p)R - Y2 -JlM _ .v i. + 2!1 + 2 l' + R
_ (i. C o-
A.0 i. I
+ 2p + 2p
+ + 4f3p
_ .v K 0 4/3p -JlM__;;---:-':-:~ KI
(2.120)
Considering the equations relating pressure p, displacements u and U, and potentials cP 1 and cP 2 , Eq. (2.119) can also be written:
CoVzp = p
(2.121)
Thus at low frequency the slow wave corresponds to a diffusive type of propagation mode, governed by the scalar diffusion equation (2.121) for pressure with a hydraulic diffusivity coefficient CD. This remark is due to Chandler and Johnson (1981), who showed that the analysis of Rice and Clearly ( 1976) was also included in the theory of Biot. This is even more remarkable because Rice and Cleary introduced the fluid bulk modulus separately from the equation of state of the fluid, whereas Biot's theory treats the matrix . and fluid on the same level from the outset.
2.2.4 Biot's theory and Terzaghi's law For the case of total saturation. Terzaghi's law is frequently used in geotechnics. This law breaks down the macroscopic or total stresses qii into so-called effective stresses
at.
I
82
WAVE PROPAGATION IN SATURATED POROUS MEDIA
2
which are the stresses exerted on the solid skeleton, and into hydrostatic stresses - pi>ii• which are the stresses pre\"ailing in the fluid, so that: (2.122) (jjj = pi>jj
at -
Accordingly, one may well ask what are the connections between Biot's theory and Terzaghi's law. If an experiment is performed at zero pressure (p = 0), the total stress tensor is identif1ed with that of the effective stresses. Equation (2.66) then gives: p = 0
tr
at= 3K
0
tr
8
(2.123)
For low porosities (c/> < .20%), by an elementary theory of effective moduli, it is moreover easy to show that: (2.124) p=O tr = 3K,(1 -
at
where K, is the matrix bulk modulus. From (2.124) and (2.123): K 0 = K,(l- c/>)
(2.125)
It is now possible to use Gassmann's equation (2.77) and Eqs. (2.70) and (2.75) to obtain:
f3
= c/>
K1
= K,(1- c/>) + K 11 c/>
(2.126)
K,z
i\1 = c/>
Using (2.122) and (2.1261 and Biofs equation (2.65), the following general identity is obtained: (2.127a) tr afi = 3(1 -
(2.127b)
Thus Terzaghi's law is more restrictive than Biot's theory, since Eqs. (2.126) reveal the presumably additive character of the bulk moduli. In a closed system, the rheological model attached to the saturated porous medium in relation to compressibility effects consists of two springs in parallel, with. constants K,(1- , concerning the respective contributions of the skeleton and the fluid. In an open system, only the skeleton participates (spring with constant K,(l :.___ c/>) = K 0 ), while the share accounted for by the increase in fluid content in the apparent macroscopic deformation is cf> tr 8 ({3 = cf>). This assumption of additivity, which is implicit in Terzaghi's law, leads to a decoupling between the fluid behavior law (2.127b) and that of the skeleton (2.127a). To evaluate the effective stresses, it is therefore assumed that the medium consists exclusively of the solid skeleton, leading to the average pressure K. tr e. The pressure effect is subtracted and the result multiplied by the factor (1 - c/>) representing the effective volumetric proportion of solid skeleton (2.127a). The fluid behavior law (2.127) involves only the volumetric strain of the fluid, div U, which corresponds to the above decoupling. Note here that the strain tensor 8 concerns the average or macroscopic displacement on the solid/fluid combination (e is not related to the solid part alone). Moreo,·er, Biot's theory concerns only small disturbances, and in (2.127) 8 is assumed to be small, as well as U, which is a Lagrangian quantity.
_,
;f'
I
. ·t
2
WAVE PROPAGATION IN SATIJRATED POROUS MEDIA
The introduction of equations (2.126) in (2.97c) gives by (2.1041:
)2 1
v. = ( p,
).0
+ 2Jl
p.(l - t/>)
i;,~
(2.128)
1
Vp, =
"
83
(~;'Y
where the effects of inertial coupling due to tortuosity (a = 1) are ignored, together with any dissipation process. With these assumptions, it is clear that the wave of the ftrst kind involves only the matrix and occurs at zero pressure (open system). while the wave of the second kind concerns only the fluid (~i = 0). This is perfectly understandable, since Terzaghi's law, which is mainly reflected by the assumption fJ = 4> in relation to Biot's theory, assumes a priori that the action of the fluid can be considered as an external action exerted on the matrix or solid skeleton. The equation (2.122) inserted in (2.87a) for steady state conditions therefore leads to:
~at cxi
_ OX; op = o
(2.129)
and the action of the fluid on the solid skeleton is reflected by imaginary volumetric forces of intensity - grad p. It should also be noted that, with the implicit assumption ofTerzaghi's law (fJ = t/>), if the fluid bulk modulus is much lower than that of the solid skeleton (K 11 ~ K 0 ), and if tortuosity effects are taken into account and viscosity effects ignored, the wave velocities are given by: 1
i.0 + 2Jl
V. _ [
p.(l - t/>)
P, -
1
Vp
>
+ tf>p1 (1 -a
=[KJ,J ap
I
Vs =
[
p.(l- t/>)
Jl
+ t/>pf(l- a
·>J Jl
(2.130)
•)
The equations (2.130) can hence lead to an experimental estimation of tortuosity, if a sound choice is made of the saturating fluid (low compressibility and low viscosity). As we have shown, assuming small displacements both in the solid part and in the fluid part, Terzaghi's law applies within a more restricted set of assumptions than Biot's theory. As we know, however, the central value ofTerzaghi's law lies elsewhere and concerns flow problems. In fact, the decoupling assumption helps to analyze substantial fluid movements (flow in an earth dam, for example) independently of small strains of the solid skeleton. To do this, the pressure f1eld in the fluid is determined by Darcy's law (where gravity effects may be accounted for), and the law of conservation of mass (101 for the fluid which, in this type of calculation, is usually assumed to be incompressible (div U = O)uu. 110) Note that the laws of conservation of mass in Biot's theory are automatically satisfted to the nearest second order, since displacements are assumed to be small. I 11) In this case, U becomes a Eulerian quantity (fmite transformation), and (2.1~7b), which is only valid for small displacements, no longer holds and is replaced by Darcy·s law.
84
2
WA\"E PROPAGATION IN SATURATED POROUS MEDIA
By reinserting the pressure fteld thus determined in (2.129) (in which gravity forces are taken into account). we can obtain the equilibrium equations for the solid part alone, with considerable latitude in selecting the constitutive law. Hence, apart from the Eqs. (2.130), which are interesting for the determination of tortuosity, the respecti,·e values ofBiot's theory and Terzaghi's law are not concerned with the same ftelds of application, and these theories are complementary rather than competitive.
2.2.5
Geerstma-Smit equations
As we have shown. the correct solution for the ftrst kind wave velocity within the framework of Biot's theory is relatively easy to obtain, but not very practical for any actual application. Nevertheless, we have pointed out (Table 2.2) the high values of critical frequencies fc for commonly encountered rocks. This makes it possible to obtain an approximate solution for the velocities by developing the second order equations in f !fc (Geerstma and Smit, 1961):
vzp,
v!, + =
v•(!c) f
2
0-
(2.131)
v; + v~(~)
where 1-'0 and V.., are the zero frequency and infmite frequency velocities respectively within Biot's theor'y'"presented here (low frequency hypothesis). By implicitly using Gassmann's equation (2.77), these authors obtain the following expressions for V0 and Vx :
).!..
V0 = (
4 z Kf+-Jl 3 [see Eq. (2.113)] p
Voc=(p.(l-cf>)+~p,-ll-a- )
(2.132)
1
·
4
[ K, + J" +
(
P
K
)(
t/>-a -1 + 1 - Ks - I1 Pt K
)])y
_K(l_2cf>a-
0
' K ¢
(I - K:- ~) K, + K''
1
.
(2.133)
where K 1 is giYen by Gassmann's equation (2.77) and p = (1- cf>)p. +
p1 . It should be noted forthwith that, if tortuosity a is infmite, Eq.t2.133) is reduced to Eq. (2.1321. because there is no longer any relative movement of the fluid and solid (infmite
"""
----~--~-------
2 __
,
WAVE PROPAGATION IN SATURATED POROUS MEDIA
85
inertial coupling). and hence no permeability effect. Note also that Eq. (2.133) is reduced to Eqs. (2.130) using the assumptions:
i
'
Ko
1- K =
•
t/J,
K 11 ~ K 0
which have precisely been used to state (2.130).
I
lj 'f i
j
I -~
'"""
2.2.6 2.2.6.1
Experimental verifiCation Qualitative aspects of Biot's model
We have just described a method for modeling wave propagation in a saturated porous medium. The model presented reveals the existence of three propagation waves within such a medium, in other words two compressional waves and one shear wave. The particular feature and main interest of this model reside in the prediction of the second compressional wave, or slow wave. In fact, this slow wave does not exist in a classic solid and isotropic medium. Hence it is essential to prove the validity of Biot's theory to identify the existence of such a slow wave experimentally. We shall ftrst describe the qualitative procedure that enabled Plona (1980) to verify this theory experimentally, and then go on to the results obtained. A saturated porous medium is a medium formed by the interpenetration of two phases. One of them is the solid phase and constitutes the matrix of the material concerned, and the second is the liquid phase, constituting the saturating fluid. This interpenetration can occur in t'to different ways, shown schematically in Fig. 2.13 (Plona, 1982). In one of the two cases (Fig. 2.13a), a discontinuity exists in the liquid phase, consisting of a group of disconnected pores. with continuity of the solid phase. In the second case (Fig. 2.13b), both the liquid and solid phases are continuous. Two types of wave (a compressional wave and a shear wave) can propagate in an isotropic solid, while only a single type of wave, a compressional wa,·e, can propagate in a liquid. Since a saturated porous medium consists of two phases, one solid and the second liquid, it therefore appears logical to be able to observe three waves, namely two compressional waves and one shear wave. For this to be true, it is necessary for the liquid and solid phases to be continuous. In this way the waves can propagate within a given phase. This means that the porous medium must be of the second type described above (Fig. 2.13b). In fact, every natural porous medium possesses both types of porosity (disconnected and connected~ so that the liquid that participates in the motion of the slow wave is merely the fraction of liquid contained in the connected porosity. Finally, it is important to realize that the coarse image by which one considers that, among the two compressional waves, one moves within the liquid and the second in the solid, is false. In fact, the porous medium is a material constituted of a solid and a liquid phase coupled together. A more accurate image can be derived by representing the sample as a system of two springs whose eigen vibrations consist of one vibration in phase and one vibration out of phase. These two types of motion have been modeled and discussed previously in Section 2.2. This remark will enable us to understand more clearly the necessary conditions for the observation of a progressive slow wave. If the liquid is not
86
WAVE PROPAGATION IN SATURATED POROUS MEDIA
2 ~
t.
·~
Fig. 2.13 Schematic diagrams of porous media. a. The fluid phase (white) is discontinuous, the solid phase (cross-hatched) is continuous. b. The fluid and solid phases are continuous. (after Plona. © 1982 IEEE, New York, NY).
viscous, no viscous coupling force occurs at the liquid/solid interface, while, by contrast, if the liquid is very viscous, a substantial coupling exists, preventing differential liquid/solid movement. This clearly shows the importance of the viscosity of the interstitial fluid. Furthermore, as we well know, the intensity of the viscous coupling force depends on the incident wave frequency. At infmite frequency, a viscous fluid acts as if it had no viscosity, whereas at low frequency even low viscosity can give rise to substantial coupling. In fact, considering the intensity of this viscous force, it can be observed that it decreases rapidly "ith increasing distance from the liquid/solid interface (Fig. 2.14). Moreover. this effect can be characterized by a skin depth d3 • This skin depth:
{2i1 • .J-;;;;
d =
(2.134)
is proportional to the square root of the viscosity and inversely proportional to the square root of the frequency and of the fluid density. This clearly shows that the propagation of the slow wave becomes even more observable as the fluid viscosity decreases and the wave frequency increases. The density term in (2.134) is there to remind us of the inertial ·problems of setting any mass in motion. Hence it is very important for the observation of ·Biot mechanisms and for a given sample that this depth d. should be much smaller than the average acces radius of the pores.
"'
2
WAVE PROPAOA110N IN SATURATED POROUS MEDIA
!I
j:
'l 't.
FLUID-
,, I
SOLID-
t
f:
v
!l
0
Wall set in motion
y
f
t>o ,
~
v
Tr-ient motion of the fluid
Fig. 1.14 Fluid velocity proCde due to sudden movement of the walls of the capillary tube.
In sandstones, for example, the average access radius is relatively large, about l to 5 pm for permeabilities ranging from a few mD to 100 mD. This implies that, for an ultrasonic experiment, using water as the saturating fluid, the skin depth d. is about 0.5 to l pm. The ratio of skin depth to access radius is hence too high to be able to observe the slow wave directly, as the viscous effects at the fluid/matrix interface have prevailed over the possibilities of fluid/matrix motion out of phase. This means that the slow wave must be observed indirectly, for instance by analyzing the changes in signature undergone by the different recorded signals. In brief, we have shown that the following properties are required for the observation of a progressive slow wave(l2):
t, ,r
'libli.
th ...
itt;
(a) (b) (c) (d) (e)
l(
"-·.
eet
Continuity of liquid and solid phases, open system. High frequency content of the incident wave. Low saturating fluid viscosity (high hydraulic permeability). High saturating fluid density Oess important). High pore size and pore access radius, high absolute permeability.
For a clear observation of the two movements described above (in phase and out of phase), the difference in velocities between fluid and solid must be high enough to ensure signifacant separation between the two movements. Consequently, it is important to have a fluid that is much more compressible than the solid matrix.
:v
a{\,
l t
lV 6
tiar
l
87
~
~~~~
(12) The term progressive is used here in opposition to diffusive .
....
~~
88
2.2.6.2
WAVE PROPAGATION IN SATURATED POROUS MEDIA
2
Experimental results
*
Having made these preliminary remarks, Plona's experiment appears simple. In effect, to observe the slow wave, it is necessary to allow fluid and solid movements, and thus to transmit and record the waves far from the solid. This means avoiding bonding the transducers to the sample, which would prevent the recording of the slow wave. The experiment is conducted in a water-filled tank (Plona, 1980, 1982) (Fig. 2.15) and the angle of incidence of the signal emitted with respect to the porous medium may vary to observe
Shear wave
Compressional
~
waves in water A
i 0
8
I Receiver I
Fig. :us Diagram showing mode conversions and refractions at the different interfaces: the reflected waves are not indicated (after Plona and Johnson. ~ 1980 IEEE, New York, NY). the angular dependence of the process. The transducers employed have a central frequency of 2.25 MHz. The velocities of the different waves are measured with an accuracy of 3%. Figure 2.16 (Plona, 1982) shows the results obtained with a sample of sintered glass of porosity
(}~· =
.
SID
vw
-1 _ _
Vp,
(2.135)
~,
r ll
2
I
89
WAVE PROPAGATION IN SATURATED POROUS MEDIA
W-'er
Sample
0: 0°
1"'~1 m4t. A
C
D
E
F
I
11 1
A
c
A:OirctP 1 C, E. G: multiples
E
W-'tr
0: Oirct stow wave F: Convertld WIVe
G
Sampl•--+---,r0
(b)
e< ep· c
I
A
--- Slow mode __ Normet mode
1
....
..A o.A 50 mY
F
_;,.I
' ' I
B
A: Dirct P 1 wave 8: Oirct S wave 0: Oirct slow wave 5ps
0
Fig. 2.16 Signals recorded at different angles of incidence for a material made of sintered glass spheres. L 6 = 0". b. 0 < 6 < ~· (after Plona and Johnson. © 1980 IEEE, New York, NY).
·: ~
where V.., = sound velocity in water, Vp, = f1rst kind P1 wave velocity in the sample. Three arrivals are observed, whereas, in the case of a non-porous solid, only two would have been observed. Arrival A is the standard compressional wave, arrival B is the shear wave, and arrival D is the slow wave. Figure 2.17 (Plona, 1980) shows the variation in recordings as a function of angle of incidence. Fig. 2.17a and b resume Fig. 2.16. The recording in Fig. 2.17c was made with an angle of incidence() between the two critical angles related to the P1 and S waves, that is to say: where
O:•
o:· < () < (P,
(2.136)
. - 1•:.. _ 0: =SID Js
(2.137)
is given by (2.135) and:
where V5 is the shear wave velocity in the sample.
lij
90
2
WAVE PROPAGATIO!' IN SATURATED POROUS MEDIA
As observed, there is no longer a transmitted standard compressional wave. However, the S wave and the slow wave continue to be transmitted. If the angle of incidence(} is progressively increased, until it exceeds the critical value 0:. Fig. 2.17d is obtained, in which a single arrival is observed, namely the slow waYe. Finally, it has not been possible to cause the slow wave to disappear: no critical angle exists for the slow wave, since its velocity is lower than that of water [see for example the approximate formula (2.130)]. (a) I
: !
r
L.
,.;
~~
9 =0°
l
!
I
iI
I i I
F
G
I
A
A
ofo
C 0
B
E
I
0
mIMIll B
0
l
d)
9'
I
\ I
I
I
I
i 50mV\
L___,__
I
A IV l
.!_L.-
I
L
i
I
5 .•__
Fig. l.l7 Signals recorded after propagation in the water/porous solid/water system for different angles of incidence (sintered glass).
= 00. b. 0 < 8 < 1980).
a. 8
Of:' c.
~·
< 8<
Of
d. lfi: < 8 < 90" (after Plona,
Plona's results were obtained using various types of porous material. Figure 2.18 (Plona, 1982) offers a glance at the different materials used. The ftrst class of materials consists of different sintered materials I steel, titanium, Inconel). The second class includes ceramics produced by different manufacturers (Coors, 3M, Filtros). These ceramics are normally used as ftltration materials. The third class comprises materials manufactured by Plona in the Schlumberger laboratories at Ridgefteld. They are sintered glasses whose
·--------- -----
i1. l
2
'1:.'
4
Sintered glass beads # 1 beads # 2 beads # 3 3 \f-55 3 \f-40 Coors Porous steel Porous titanium Porous lnconel Filtros # 1
tP
r
v,.,
Ys
v,..
28.3 18.5 10.5 34.5 30.0 41.5 48.0 41.0 36.0 40.0
50 20 10
4.05 4.84 5.15 2.76 2.91 3.95 2.74 2.72 2.12 4.65
2.37 2.93 2.97 1.41 1.62 2.16 1.54 1.79 1.15 2.91
1.04 0.82 0.58 0.91 0.96 0.96 0.92 0.91 0.93 0.94
55 40
55 20 30 90 60
.
...,.
91
WAVE PROPAGATION IN SATURATED POROUS MEDIA
·-
=porosity, = average pore size (IUJI), I j. = fast compressional velocity (krn/s), Is =shear velocity (km/s). lj., =slow compressional velocity (km/s),
ll>
r
Fig. 2.18 List of the different porous materials analyzed (after Plona.© 1982 IEEE, New York, NY).
porosity can be varied. Figure 2.18 shows that all these porous materials display a behavior of the type described by Biot. The slow wave is always slower than the wave propagating in the saturating liquid, which is water in this case. The amplitude of the measured slow wave may be very high. Figure 2.19 provides an example of a seismogram recorded in a sintered steel with a grain size of20 ~tm. But~ we have seen, the amplitude also depends on the ratio of the wavelength ofthe signal transmitted to the pore size, or, in fact. for the materials analyzed, to the grain size. Figure 2.20 highlights this effect. To do this. the frequency - and hence the wavelength - of the transmitted signal is kept constant, while the pore size varies. In Figure 2.20a, the pore size is 15 ~tm. Considering th..: ~kin depth [see Eq. (2.134)], a value of about half a micron is found. already accounting for a significant fraction of the pore size. Thus an incipient viscous effect and a tortuosity effect are observable (see the beginning of Chapter 2: as
il'l
92
WAVE PROPAGATION IN SATURATED POROUS MEDIA
~lmtt!. 8
A
2
Fig. 1.19 Refracted arrivals at non-normal incidence for 20 J.LII1 sintered steel (after Plona and Johnson. © 1980 IEEE, 1\ew York, NY).
0
A:
Normal P1 wave
B:
S wave
0: Slowwave
n~.\tt= A
0
8
I'" I~ A
8
0
lcl
- r\\ !A. ~
50 mv
2 (L•
i
Fig. 2.20 Refracted arrivals at non-normal incidence for 3 \1: samples (after Plona and Johnson, 1980. © 1980 IEEE, 1\ew York, NY). . a. Pore size 15 j.lm, weak slow wave. b. Pore size 55 j.lffi, energetic slow wave. c. Pore size 175 j.lm, scattering: no slow wave.
pedormed with porosities greater than 10%, but a theoretical extension to the value 0 was obtained on the curve. Since the slow wave is due to the presence of a liquid inside a porous medium, it is perfectly logical for its velocity to tend towards 0 with decreasing porosity (a tends towards infmity as 4>· tends towards 0). The experiments described here concern artiftcial materials other than rocks. As we have alrea4y emphasized, owing to the pore size of a usual rock, it is very difficult, if not impossible, with usual fluids (such as water) to have a slow wave propagating within a porous medium, and the slow wave is generally attenuated immediately. Experiments are currently being conducted by Plona and Johnson and their colleagues to develop a technique to resolve this problem. The general principle consists in using liquid helium II, whose viscosity is zero, as the saturating fluid. The preliminary results obtained in these conditions have revealed the existence of a slow wave in a rock. The use of liquid helium as the saturating fluid is extremely interesting for two main reasons. Johnson (1980) showed that the slow wave mechanism was actually the
~'f:
2
~
WAVE PROPAGATION IN SATURATED POROUS MEDIA
93
generalization of the fourth acoustic wave in superfluid helium II. Moreover, with this material (i.e. negligible viscosity and fluid compressibility ~ matrix compressibility), Biot's expressions for the velocities are simplifted and are given by Eq. (2.130), assuming the applicability of Terzaghi's law (see Johnson and Plona, 1982). Hence it is possible to measure tortuosity a by using superfluid helium. for which the above equations are valid to the nearest 0.01% (Johnson et al., 1982). Johnson eta/. f 1982) also showed that these equations are valid to within 10% for water-saturated sintered glass. Incidentally, this parameter can be related to other types of measurements, such as the refractive index of the fourth wave of a superfluid He II (Johnson and Sen, 19811 or the electrical conductivity or the formation factor (Johnson et al.. 1982). 6
Fused glaa beads
s
~
::!!. ~
l!
;
2 200m\D
_,,1..---'"
•
•
•• • Slow wave
Ol,,~
0
10
20
30 Porosity (%I
40
50
Fig. 2.21 Measured velocities of·P, Sand slow waves vs. porosity for samples of sintered glass, The P ~ave and S wave velocities in solid glass are 5.69 and 3.46 km/s respectively (after Plona. © 1982 IEEE, New York, NY1.
In conclusion, Plona demonstrated the existence of a slow wave very close to the one predicted by Biot's theory. He showed that this wave could only exist as a propagation wa,·e if the following conditions were satisfted: (a) Continuity of the solid and liquid phases (i.e. possibility of differential fluid and liquid motion). (b) Sufftciently high incident wave frequency and sufl'lciently low fluid viscosity (i.e. weak wave attenuation mechanism). (c) Incident wavelength sufl'lciently large in comparison with pore size to avoid scattering, while the pore size must be adequate to avoid viscous effects at the wall (skin depth effect). (d) Very different fluid and solid bulk moduli in order to separate clearly the two compressional waves. Moreover, Plona and Johnson (Eq. (2.130)] revealed the possibility of employing approximate equations to determine the different velocities in the saturated porous medium. The equations make use of tortuosity a characterizing the geometry of a porous medium.
...
94
WAVE PROPAGATION IN SATURATED POROUS MEDIA
2
2.3 CONCLUSION We have considered a theory modeling wave propagation in saturated porous media. This theory, whose experimental conftrmation has been discussed above. identifted a number of mechanisms related to the presence of fluids and the permeability of the medium. It has been shown that these mechanisms occur to a signiftcant extent in the case of high mobilities and frequencies. The usual permeabilities and viscosities are generally inadequate to cause them to appear. The frequencies at which the mechanisms described are invoked tend to be too high (see Table 2.3) to apply the theory, since it is restricted to wavelengths that are large in comparison with the dimensions of the flow channels where ftltration occurs. For a wave propagating in an infmite two-phase medium (i.e. one which does not ··encounter" boundary conditions). wave attenuation should be looked for in the absolute local motion of the fluid, and no longer in the motion ofthe fluid with respect to the matrix, although certain experimental results (for example Morlier and Sarda, 1971) appear to demonstrate a relationship between permeability and attenuation. Note for the time being the theoretical work of Datta (1975), Mavko and Nur (1979) and Walsh (1976) who deal with these problems, which we shall discuss in detail in the subsequent Chapters. Howe,·er, it must be observed that, in these processes, the entire fluid part is involved, both the fluid found in the disconnected porosity and the fluid in the connected porosity. In addition, contrary to the framework ofBiot's theory, these mechanisms occur throughout the whole saturation range. One may well question the practical interest of Biot's theory for the study of wave propagation in an infmite medium. However, for the problems in which boundary conditions are imposed on the flows, pressures and displacements, the phenomena identifted by Biot's theory, which are mainly related to pressure gradients, must play a major role. These include the important problems of seismic reflection at interfaces, and the problems related to surface waves in sonic logging. Geerstma and Smit (1961) made a theoretical analysis of the problem ofreflection, and showed that a slow wave was always generated at the interface. This slow wave may account for a non-negligible portion of the total energy, although, due to its attenuation, it is not generally observable. Rosenbaum (1974) pointed out the importance of the role of permeability on the attenuation of Stoneley waves at the interface of a saturated porous medium and a fluid, in the plane case and in the case of guided waves in wells. These studies will be discussed in Chapter 6, which will demonstrate the necessity of a careful analysis of wave propagation for seismic applications. Biot's theory also justifies certain general considerations. The assumptions of this theory include the notion of scale related to wavelength and hence to frequency. The seismic profiles correspond to low frequencies ( ~ 50 Hz), while sonic logging corresponds to higher frequencies (~ 10 to 20kHz). In all cases, it is considered that these wa,·es propagate in homogeneous and generally isotropic and stratifted media. In nature, the media that the wave travels through are porous, inhomogeneous, and anisotropic to varying degrees. The use of exact laws of propagation should allow a better understanding of the complexity of these media, so that data can be available on their petrophysical properties and their fluid content. Strictly speaking, these exact laws do not exist. From a
r ;'l ~'
.j,:. '-
'
a
2
WAVE PROPAGATION IN SATURATED POROUS MEDIA
95
practical point of view, the laws used must always be chosen according to the specif1c problem investigated (boundary conditions, geometries) to the tool employed (frequencies, wavelengths, energy), and to the parameters to be quantif1ed (water content. permeability, porosity). ' One may be tempted to complicate a model such as Biot's model (for example, by introducing the viscoelasticity of the matrix) to clarify a number of measurements. In our opinion, this complication is extremely liable to be prolonged indefmitely, without any hope of success. In fact, the complexity of the porous medium (see Chapter 1) is such that it is totally unrealistic to try to construct a general model for porous media. On the other hand, it is necessary to account for in situ conditions and for the frequency range to be investigated to adapt the theory to the problem at hand. To each problem corresponds a different theory, which relies on a specific physical parameter to yield an observable effect. In the laboratory, it is essential to press forward with a qualitative phenomenological analysis of the problems (which is the preponderant effect in a given experiment?). an adjustment of simple models to the experiment, and a suff1cient number of measurements to acquire statistical knowledge that is virtually "rock type., by "rock type ... As we have already stressed on several occasions, for the time being, the complexity of the natural media under consideration dashes any hopes of a unique theory based on continuum mechanics. However, in the beginning of this Chapter, we pointed out that, for a given physical phenomenon, a porous medium could be replaced by an equivalent homogeneous medium with very specif1c properties. In other words, the porous medium is no longer · considered with its complexity, and only the result of its interaction with a given physical phenomenon is important. This sets the stage for a new trend in investigating porous media, which may fmally lead to a clear understanding of certain observed mechanisms. For problems of acoustic propagation, and for certain types of process, the rock can be replaced by a homogeneous and linear viscoelastic medium. The study of this type of medium will be discussed in the next Chapter.
· Appendix 2.1
WAVE PROPAGATION IN A NON-ISOTROPIC ELASTIC MEDIUM The introduction of(2.7) into the constitutive law (2.25) of a linear elastic medium, and the result of this introduction in the equilibrium equations (2.24) gives: cijkluj.kl
= pii;
(2A.ll
For isotropic media satisfying (2.28), Eq. (2A.l) is merely Eq. (2.39) set in another form. Let us now consider a wave front of unit normal vector n and normal velocity V at geometric point M belonging to this front. If f(xt, t) is the equation of the wave front, note that V is defmed by :
96
WAVE PROPAGATION IN SATURATED POROUS MEDIA
-j
V=
2
(2A.2)
1
[~ u:">2Tand thus represents the velocity of a geometric point moving on the normal n between time t, when it coincides with M, and a later time t + dt. Let us now consider acceleration waves, characterized by a jump A of the acceleration vector when crossing the geometric surface representing the wave front. We shall set: [ii~c]
= Ale
(2A.3)
where(·] represents the ..jump" operator and consists in determining the difference in the quantity· on either side of the wave front. If the second partial derivatives with respect to time satisfy (2A.3), we know that the second partial derivatives with respect to a space variable and to another space variable or to time must satisfy the equations of kinematic compatibility stated by Hadamard (1949). These equations are: n"n'
[ U·~cl]=-A·. 2 1
}.
V
•
.
U·~c= }.
nt
--A. v 1
(2A.4)
If the second derivatives of a quantity satisfy an equation, their discontinuities or jumps must satisfy the same equation when crossing the wave front. By applying the jump operator [·] to Eq. (2A.l ), and taking account of (2A.3) and (2A.4), we obtain: AljAi-pV 2 A;=0
(2A.5)
Alj = C;"i'n"n'
(2A:6)
where (Alj) is defmed by :
This represents the acoustic tensor relative to direction n considered. This acoustic tensor is symmetrical and the associated quadratic form is positive defmite. Consequently, the endomorphism associated with the acoustic tensor has three mutually orthogonal eigenvectors associated with the positive eigenvalues A1 = p Vj (J = 1, 2, 3) satisfying: det (Alj - A 1 <5;) = 0
(2A.7)
Hence, at any point of a wave front, three mutually perpendicular polarization directions exist, and each is associated with a velocity V,. These directions display a complex dependence on the direction n concerned, through the associated acoustic tensor defmed by (2A.6). In the isotropic case, Eq. (2.28) leads to:
i.<5;i<5"' + J1(<5;"<5i, + <5u<5j,)
(2A.8)
The introduction of (2A.8) into (2A.6) and (2A.5) yields: (A. + J1){D • A)n + (/1 - p V 2)A = 0
(2A.9)
cii"'
=
This equation is satisf1ed either ifn and A are colinear, or if their respective coefficients are zero. The former case corresponds to longitudinal waves. In fact, by setting A = An, which corresponds to a longitudinal polarization, we have: P
v; = i. + 211
(2A.l O)
..,....-
----------
2
------------·--·----
WAVE PROPAGATION IN SATURATED POROUS MEDIA
.
97
The second case corresponds to transverse waves. The transverse polarization n . A = 0 thus leads to : pV~ = Jl
(2A.ll)
The results found for harmonic plane waves, concerning polarizations in an isotropic elastic medium (see Section 2.1.5), are hence valid for any shape of wave front. In the most general anisotropic case, in which the tensor Ciitl displays no particular symmetry apart from the natural symmetries (2.26) and (2.27), the characteristic polarization directions are normally random and depend continuously on the vector n. It is only if the material exhibits certain additional particular symmetries that certain polarization directions are preferential. Let us thus consider a case of standard anisotropy which is the orthotropy'of revolution (or transverse anisotropy). Let us assume for the purpose that the orthotropy axis is parallel to the physical direction OxJ. This means that all the directions are equivalent in plane (Ox 1 , Ox 2). In this case, it can be shown that, ofthe 21 independent components for tensor cij/d in the general case (see Section 2.1.3), only five independent components remain, which are C 1111 = C 2222 , CJJJJ C 1122 , C 22 JJ = CuJJ• C 1313 = C 2J 2J· The remaining components that are not inferred from natural symmetries (2.26) and (2.27) are zero, except C 1212 given by: 1 (2A.l2) Cu12 = 2 (Cuu- Cuu) Considering (2A.6) in the case of the orthotropy of revolution, it is easy to see that two types oflongitudinal wave and two types of transverse wave can propagate. To begin with, when n = nJ is oriented along the orthotropic axis OxJ, a longitudinal wave (A= LlnJ) with velocity .jC 3333 /p and two transverse waves (A. nJ = 0) with velocity jC1313 /p = JC 2J 23 /p are seen. Subsequently, if n belongs to the plane normal to the ortbotropy axis(n. nJ = O),alongitudinalwave(A == Lln)ofvelocityjC 1111 /p == JC 2222 /pandtwo transverse waves, one polarized along nJ(A =An) with velocity jC 1313 /p = jC2323 /p identical to the f1rst type of transverse wave, and the second polarized in the plane normal to the orthotropy axis (A= Lin x n3 ) with velocity 12 ufp. Finally, the four independent velocities are:
JC
• P waves:
Vr==/¥
n= n3 A= Lin
I
n . n3 = 0
Vp=/¥=FF
(2A.l3)
• S waves:
A. n =0
n = n3 n . n3 = 0, n . n3
= 0.
rc;;;;
A • n = 0 } Vs == [C;;;; == A=LinJ ~p ~p
A = Lin
X
OJ
J's = ~
T '
I
wave propagation and vibration effects in viscoelastic med1a (unidimensional)
3
INTRODUCTION We have shown in the previous Chapter that the average motion of the fluid with respect to an elastic matrix led to invoking a dissipative mechanism and hence to wave attenuation. However, we showed that, apart from the case in which boundary conditions are involved, this mechanism could be considered as negligible in most cases. The cause of dissipation is to be found in the absolute local motion of the particles, both solid and liquid. This Chapter is devoted to the mechanical modeling of this phenomenon by the introduction of viscoelastic models. It is not intended to provide orders of magnitude of the parameters that we shall introduce (which can be found in Chapter 5), nor to discuss the relevance of these models concerning the behavior of rocks at the passage of a wave. We again consider the medium to be homogeneous. In other words, the models developed take account, from a macroscopic standpoint (that of the measurement), of the average dissipation on a representative volume of fluid and solid, without drawing a distinction between the share accounted for by the fluid and that relating to the solid. The fme analysis (microscopic) will be dealt with from a qualitative standpoint in Chapter 5. The motion considered here is hence the average macroscopic displacement u of the fluid/matrix combination introduced in the previous Chapter. In Chapter 2, for the matrix itself, we considered a rigorously elastic constitutive equation; in other words the material has no strain memory. This is the simplest case. Real materials, and rocks in particular, macroscopically display irreversibilities of behavior: they dissipate energy when subjected to deformation. This dissipation, apart from the Biot type of mechanism (Chapter 2), stems from many sources. The most probable mechanisms involved for rocks are discussed in Chapter 5. They include capillary forces, thermal effects, intergranular friction, and local fluid movements. These microscopic, irreversible effects are therefore numerous and complex,
100
WAVE PROPAGATION IN VISCOELASTIC MEDIA
3
and any macroscopic model (in the sense intended here) that would attempt to describe them fully, even if available, would be of extreme complexity and of very limited scope. Consequently, we shall only consider the simplest models here, namely linear viscoelastic models. In fact, as we shall show in Chapter 5, these models are well suited for the description of a broad class of dissipative processes, resulting from rapid, smallamplitude variations in strain due to waves that propagate in rocks. These models require the knowledge not only of the present stress and strain values, but also of past values, and are therefore called memory models. To investigate and quantify these effects in actual materials, it is essential to understand what is generally called their delayed behavior. In this way, the time effects, particularly those associated with strain and stress velocities, are introduced quite naturally. The following study is deliberately restricted to uniaxial or unidimensional behavior of the materials that we will examine. The reader interested in an exposition of the general case can refer to several works (Fliigge, 1975, Christensen, 1982, Salen.;on, 1983).
3.1
DELAYED BEHAVIOR OF MATERIALS
The experimental study of the behavior of materials as a function of time can be undertaken by two main types of test: creep tests and relaxation tests .
.
3.1.1
Creep tests
In this type of test, the sample, previously at rest, is subjected to a constant load a 0 from time t 0 • An analysis is made of the variation with time of the strain e(t), or its equivalent, the creep function f(t) defmed by: e(t) = f(t, t 0 ; a 0 ) ao
(3.1)
In the following analysis, we shall consider only materials whose properties do not vary with time (non-aging materials). For the wave propagation experiments dealt with here, this assumption can be accepted without restriction in view of the measurement time scale (i.e. wave passage time). Under this assumption, the function f depends only on the interval t - t 0 and (3.1) becomes: e(t)
-
ao
= f(t - t 0 ;
a0)
(3.2)
Figure 3.1 shows a standard creep experiment. The loading time sequence consists in the application of a given stress a 0 which is kept constant for the time interval (t0 , t 1 ) during which no additional stress is applied. The strain response is shown schematically in Fig. 3.lb. At time t 0 , an instantaneous strain e0 is usually observed such that:
T
I I
3
WAVE PROPAGATIOl' IN VISCOELASTIC MEDIA
ao
101
(3.3)
eo=Mo
where M 0 is the instantaneous elastic modulus (Young's, shear, bulk, depending on the experiment). After t 0 , the strain increases with imposed load, giving rise to creep. This is the phase that determines the function f defmed by (3.1). At time t 1 , when the stress is e
0
Creep
p -----1, =-~~---~-~ -----_1:::
ao ~---
--
0
to
0
t 1
to
tl
(a) Stress
(b) Strain
Fig. 3.1
Standard creep test.
removed, an instantaneous unloading occurs such that, if e(tl) is the strain just before unloading. and e(t just after, we have:
n
Lie
= e(tn -
e(tl)
=-
e0
(3.4)
Equation (3.4) is only true because the material is assumed to be non-aging (if not, Lie :;C - e0 ). After time ti, the strain continues to decrease, leading to recovery. This recovery may be partial, in which case the strain tends to residual non-zero strain e~ (see for example Maxwell's model, Fig. 3.10a), or may be total, in which case e00 = 0 (see for example Voigt's model, Fig. 3.10b).
3.1.2 Relaxation tests In this second type of test, the experiment consists in imposing a constant strain e0 on a sample initially at rest, from time t 0 • The stress a(t), required to produce this strain, is then analyzed as a function of time. This is equivalent to analyzing the so-called relaxation function r(t) which, assuming the materials are non-aging, is defmed by: a(t)
-
Bo
= r(t -
t 0 ; e0 )
(3.5)
Figure 3.2 shows a standard relaxation test. To produce a strain e0 at time t 0 instantaneously, a load a0 must be applied instantaneously. After t 0 , to maintain the same strain e0 , it can be seen that the required stress decreases, because stress relaxation occurs. If zero strain is wanted after time t 1 , it is
102
WAVE PROPAGATION IN \-ISCOELASTIC MEDIA
Relaxation
E
~
oo Eo
3
1--0
0 to
(tl -,
rOto--~
a.,
tl
1----
O(ttl
(al Strain
(bl Stress
Fig. 3.2 Standard relaxation test.
necessary to apply an instantaneous jump Au in the stress to be imposed, which, for a nonaging material, is .Ju = - u 0 . Subsequently, the stress to be imposed becomes increasingly low, and tends to a non-zero stress value (partial stress cancellation) or tend to zero (total stress cancellation).
3.2
LINEAR VISCOELASTIC BEHAVIOR
Equations (3.1) and (3.5) only assume the non-aging character of the viscoelastic constitutive law. In addition, linearity consists in assuming that the creep function f and relaxation function rare independent of u 0 and e0 , namely:
= u 0 f(t- t 0 ) u(t) = e0 r(t - t 0 )
e(t)
(3.6a) (3.6b)
Equations (3.6a) and (3.6b) help to obtain the law oflinear viscoelastic behavior for any loading history. In fact, linearity implies Boltzmann's superposition "principle" which states that the effects can be added. This gives: e(t) =
'f 'f
du
f(t - r) -d dr r
- :o
u(t) =
-x:
r(t- rl
de
d
r
dr
+ ~ f(t- r;)(Au);
(3.7a)
1
+~
r(t- rJ(Ae);
(3.7b)
I
Equations (3.7a) and (3.7b) state that the history of stress u (or strain e) can be considered as the superposition ofinfmitesimal steps du(r)[or de(r)] and fmite steps (Au); [or (Ae)J at time r;. The term step means that the quantity involved is imposed and then
~ 3
WAVE PROPAGATIOS IN VISCOELASTIC MEDIA
103
kept constant from time t considered. One can then formally write the Eqs. (3.7) more briefly in the form : e(t) =
'I 'I
da f(t - r) -d dr r
-:r:
a(t) =
r(t - r)
-::c
de
d
r
dr
(3.8a)
(3.8b)
where the integrals and derivatives are implicitly understood in the sense of distributionsu 1• This convention for derivatives and integrals is always presumed in the following discussion. Moreover, in accordance with geophysics convention, we shall use Eq. (3.8b) and not Eq. (3.8a), although these equations are obviously equivalent. Causality implies that stress a(t) will not be influenced by the future of the strain (i.e. by e(r) for r > t). Equation (3.8b) can then be rewritten in the form : a(t) =
I
_,
d r(t - r) de dr
-x
(3.9)
't
where it is presumed that, because of causality: r(T) = 0
for r < 0
(3.10)
Deducing (3.9) from (3.7b) may appear somewhat artificial. However, the introduction of the formalism of distribution and the constraint of causality is extremely useful in dealing with discontinuities and for using the Fourier transform instead of the Laplace transform in applications to dynamics, as we shall show subsequently. Integrating by parts (in the sense of distributions, both on rand e) and assuming that e(r) = 0 fou < - t 0 [i.e. e(- oo) = 0), (3.9) and (3.10) yield the equation: +oo
u(t)
I
= _x
m(t - r)e(T) dT
(3.11)
where dr dT
m=-
(3.12)
Equation (3.11) shows that a is written in the form of a convolution product (again in the sense of distributions) that we shall denote more briefly:
a= m•e
(1) A discontinuity .1/ at timer; for the function f corresponds to a Dirac mass Jfo(rdf/dr, if the latter is considered as a distribution.
(3.13)
tJ for the derivative
104
3
WAVE PROPAGATION IN VISCOELASTIC MEDIA
3.3 DYNAMICS OF UNIDIMENSIONAL LINEAR VISCOELASTIC MEDIA, FIRST CONCEPTS OF THE QUALITY FACTOR Q 3.3.1
Complex modulus
Let us consider the Fourier transform F(w) of a function f(t) defmed by 121 : F(ro) =
f
Conversely: f(t) =
+:c
21 1t
(3.14a)
f(t) exp (- iwt) dt
_ 00
f
+X>
(3.l4b)
F(w) exp (iwt) dro
-:10
The use of the Fourier transform in viscoelasticity problems is quite convenient because we know that the Fourier transform of a convolution product is the product of the Fourier transforms. Hence, by applying the Fourier transform to Eq. (3.13), we obtain: (3.15)
I(w) = M(w)E(w)
where M(w), the Fourier transform of m(t), is called the complex modulus. M(w) can be separated into its real and imaginary parts: M(ro) = MR(w)
+ iM1(w)
(3.16)
For a linear elastic medium. Eq. (3.13) is nothing other than Hooke's law, which is written: (3.17) u= M0 e This shows that a linear elastic medium satisfies : m(t)
= M 0 t5(t), MR = M 0 , M, = 0
(3.18)
where t5(t) is the Dirac function. Equations (3.16) and (3.18) suggest that the imaginary part M 1(ro) characterizes the dissipation of viscoelastic models, while the real part MR(w) is associated with the instantaneous response (at angular frequency w). The following Chapters therefore develop the study of this complex modulus. Before dealing with the unidimensional problem, it should be noted that the three-dimensional (isotropic) generalization of (3.9) is: +::c [ d de; .. ] (3.19) uii(t) = _"' J.(t- t) dr (tr e) t5ij + 2f.,l(t - t) d;J dt
will
I
giving rise to the defmition of complex Lame coefficients. (2) A capital letter denotes the Fourier transform ofthe function considered: F for f, M form, :E for u, E for e etc. This transform is understood in the sense of distributions. Hence the Fourier transform of a harmonic distribution: exp (iw 0 t) is 2m5(w- w 0 1(see Eq. (3.14b)]. However, this transform does not exist in the sense of ordinary functions.
r I I
3
105
WAVE PROPAGATION IN VISCOELASTIC MEDIA
3.3.2 Harmonic problems Let us assume that a strain e(t) is imposed on a sample such that( 31 : e(t) =em exp (iw 0 t)
(3.20a)
Inserting (3.20a) in (3.13) and taking the Fourier transform of the equation obtained yields: I"(w)
= emM(wo)27tb(w- Wo)
(3.20b)
where the Fourier transform of exp (iw0 t) is 2nb(w- w0 )(see footnote in Section 3.3.1). Equation (3.20b) can alternatively be written in the form: I"(w)
= 2ne'" IMI exp [iq>(w0 )]b(w- w0 )
(3.21)
where the phase (/) and the modulus IMI satisfy: .\f.(wo) tan q>(w0 ) = \1 ( ); MR = IMI cos •
q>;
. M 1 = IMI sm
q>
(3.22)
R Wo
To return to the time domain, it is necessary to take the inverse Fourier transform of Eq. (3.20b). The real stress a(t) is the real part of the expression obtained. The inverse transform (3.14b) then gives rise to:
.
(3.23a)
a(t) = Re [e'"M(w 0 ) exp (iw0 r)] 1
a(t)
= Re { e'"[Mi(w0 ) + Mf(w0 )]2 exp 1
a(t) = e..,[Mi.(w 0 )
[i(w 0 t
+ Mf(w0 )]2 cos (w0 t + q>)
+ q>)]}
(3.23b) (3.23c)
In the harmonic problem. stress and strain are hence out of phase by a quantity q> directly related to the viscosity of the medium. This viscosity therefore causes a lag between response and excitation in the steady-state problem defmed by (3.20a). As the angular frequency w 0 tends towards 0, the influence of viscosity becomes lesser and the static limit(41 is obtained by (3.22) and (3.23):
a= e..,MR(O)
(3.24)
Equation (3.24) should be compared with (3.17). Hence it is natural to associate. the viscoelastic behavior examined with the elastic behavior defmed by modulus MR(O) = M 0 , called the relaxed modulus. However, it is very important to stress that this
(3) The complex notation is e'ident: the physical quantity is the real pan of the complex quantity e(t) = £, cos w 0 t. This will always be understood in the following discussion, if not mentioned. (41 Note here that, since the inenia terms are ignored in principle (sample of negligible mass). the term static means that the time is suiTiciently long (pulsation w;;; 1/r very small) for relaxation at imposed strain e., to be complete, and MR(O) is merely the relaxed modulus (Section 3.4.1).
106
WAVE PROPAGATION IN VISCOELASTIC MEDIA
3
in no way prejudges the elastic energy (reversible) that can be stored by the viscoelastic material analyzed for a motion defmed by (3.20a) and (3.23 ). Thus it is not because: a ( t,
= 2mr roo ;
n
1, ... )
= 0, e(t,)
= e.,.MR(w 0 )
(3.25)
=o
that it can be stated that the elastic energy at times t .. is equal to e!MR/2. The evaluation of this energy can only be undertaken with additional data about the '1 iscoelastic behavior analyzed. This could, for example, be the case of the rheological models that we shall examine in Section 3.4.1. Let us now consider the energy dissipated over a period. This energy is written: LIW
=
!: ae dt
f
(3.26)
0
since the elastic energy is reversible over a period T = 2n 'w 0 . Considering that the quantities concerned in (3.26) are real, one obtains from (3.20a) and (3.23): Ll W = ne!M 1(w 0 )
(3.27)
The energy dissipated per period is hence directly proportional to the imaginary part of the complex modulus. Note that defmition (3.26) shows that Ll W can be measured by the area of the closed curve described in the plane [e(t), a(t)] during a period. Eliminating the time between (3.20a) and (3.23), it can be shown that this curve is an ellipse (Fig. 3.3). In the plane related to dimensionless coordinates (e(t)/e ... , a(t)/em IMI ), the main axes of the ellipse are plotted by the major and minor bisectors. The ratio of the half-axes is 1/tan 2 ({J/2, the major axis being plotted by the major bisector and equivalent to 2 cos 2 qJ/2. CJ (t)
Fig. 3.3 Stress-strain cycle in the plane (e(t)/em. u(t)/tmiMI).
3
WAVE PROPAGATION IN VISCOELAmC MEDIA
In the elastic case, the ellipse is reduced to the major axis because, in this case, lf' IMI = MR. The intersection of the ellipse with the major axis is given by:
e = ± e,.. cos lf'/2,
107
= 0 and
u = ± IMI e,. cos lf'/2
The fact that A W depends on the frequency is characteristic of viscous behavior, because this implies that the power dissipated depends on the strain velocities which are proportional to w. For instance, in the case of dry friction, which depends only on the limiting stress, this dependence disappears. In conciusion, we must stress that, for harmonic problems, the expressions of phase shift qJ (3.22) and of dissipated power (3.27) depend only on the data of the complex modulus.
3.3.3 Wave propagation 3.3.3.1
Wave propagation and attenuation
In dynamics, the unidimensional equilibrium equation is written (Section 2.1.5.2): (3.28)
u.x = pii
The introduction of the behavior Eq. (3.13) into (3.28) gives the equation of unidimensional waves in linear viscoelasticity by replacing e by oujox. This gives: 13.29)
m•u.xx = pii
Let us now consider a displacement of the form : u(x, t) = u0 exp [i(w 0 r- k*x)]
(3.30a)
In this expression k* is a priori a complex quantity and Eq. (3.30a) can also be written: u(x, t)
= u0 exp [- :x(w0 )x] exp
[i(w 0 t - kx)]
(3.30b)
with k* = k- icx, where k and ex are real quantities. A solution such as (3.30b) thus corresponds to a traveling wave with attenuation :~(w0) at angular frequency w 0 • We fmally obtain the equation: k 0 V0
= k(w 0 ) V(w 0 ) =
w0
13.31)
where k and V are the wave number and phase velocity respectively at angular frequency Wo.
The introduction of Eq. (3.30b) into Eq. (3.29) and the derivation of the Fourier transform of the result obtained gives: - (MR
+ iM 1)(k- icx)2 + pw'f, = 0
13.32)
By separating Eq. (3.32) into the real and imaginary parts, the following two equalities are obtained: k
= Wo
/P(IMI + MR) 21MI2
(3.33a)
108
3
WAVE PROPAGATION IN VISCOELASTIC MEDIA
oc
p(IMI- M.J
= Wo
(3.33b)
21MI 2
1
where IMI =(Mi. + Mf}2. Assuming weak viscoelastic effects, i.e. M 1 ~ MR, Eqs. (3.31) and (3.33) lead us to: V(w 0 ) =
JMR~wo)
2oc
(3.33c)
M1
(3.33d)
-;:= MR
Thus the wave's phase velocity at angular frequency w 0 is given (to the nearest second order) by the same equation as in elasticity, if care has been taken to replace M 0 [refer to (3.17)] by MR(w 0 ). Moreover, the ratio M.JMR is again observed to appear in the expression of attenuation deriving from the dissipati,·e character of viscoelasticity.
3.3.3.2
Variation in time of a free wave packet: group velocity and phase velocity
Let us begin by analyzing in the elastic case the qualitative behavior of u(x, t) for a very simple specific case: u(x, t) corresponds to the superposition of three waves of amplitudes Ak Ak u(k 0 ), lf2u(k 0 ), 1 '2u(k 0 ), and of respective wave numbers k 0 , k 0 - 2 and k 0 + 2
(angular frequencies. w u(x, t)
= u(k0 )
{
0,
w0
-
Aw T
and
Wo
+
Aw) : T
exp [- i(k 0 x- w 0 t)]
+~exp[ -{(k
0 -
~k)x-(w0 -A;)t]]
+~ exp [- {(k, + ~nx
= u(k 0 ) [ 1 +cos ( x Ak 2
Aw)] exp [- .
- tT
1(k 0 x- w 0 t)]
-('"•
+ ~;)r]J} (3.34)
Let us firSt consider the initial timet = 0. In this case, the modulus lu(x, 0)1 is maximum at x = 0 because, for this value, the three waves are in phase and interfere constructively (Fig. 3.4). As we deviate from the value x = 0, the waves mutually go out of phase, causing a modulation oflu(x, 0)1 which decreases the amplitude away from x = 0. The interference becomes completely destructive when the phase difference between exp (ik 0 x) and exp [i(k 0 Ak/2)x] is equal to ± nand lu(x, 0)1 = 0 at x = ± Ax/2 where Ax is given by:
+
AxAk
= 4n
(3.35)
This equation indicates that the width Ax of the function lu(x, 0)1 (distance between two successive zeros) increases with decreasing Ak. In a more general manner, for a signal
T
3
109
WAVE PROPAGATION IN VISCOELASTIC MEDIA
I !
I
II
ko+~ 2 ko
~
ko-T
llx
-T
Fig. 3.4 Example of the composition of three waves. The result is given on the lower curve [see Eq. (3.34)] (after Cohen-Tannoudji et al., Mecanique quantique.© 1977, Hermann, Paris).
whose spectrum is different from 0 over a band Ak or .1w, this means that the narrower the band, the more spread out the wave "packet". Let us now consider a later time. Equation (3.34) shows that the maximum of lu(x, t)l, which was found at x = 0 at time t = 0, is now at the point :
=
XM
Aw
(3.36)
Ak t
and not at point x = (w 0 fk 0 )t = V0 t. The physical origin of this result appears in Fig. 3.5. Part (a) of this ftgure represents the position at time t = 0 of three adjacent maxima (1), (2)
Ak ko+T ko Ak ko--· 2
1,,) 1(1) 1(1)
1(2) 1(2) 1(2)
1(1)
1(3)
1(1)
1(3)
1(1)
1(3)
1(2) 1(2) 1(2)
1(3) 1(3) 1(3)
X
(a)
X
0
0
t
(b)
xM(O)
t
xJt(t)
Fig. 3..5 Constructive interference effects: group velocity concept (after Cohen-Tannoudji et al., Mecanique quantique. © 1977, Hermann. Paris).
IIO
WAVE PROPAGATION IN VISCOELASTIC MEDIA
3
and (3) for each of the real parts ofthe three waves. The maxima identified by subscript (2) coincide at x = 0, and constructively interfere and therefore correspond to the peak ju(x, 0)!. In the case of Fig. 3.5, where the velocity increases with k (the most frequent case), the maxima (3) of each wave merge. After a certain time interval, the situation shown schematically in Fig. 3.5b is thus obtained: the maxima with subscript (3) coincide and give the position of the maximum xM(t) oflu(x, t)l. One can therefore clearly see in Fig. 3.5 that xM(t) is not equal to V0(t) but is given by Eq. (3.36). In the more general case in which the signal corresponds to a Fourier spectrum of amplitude U (k), varying slowly with k and being non-zero over a narrow band centered at k 0 • the center of the corresponding wave packet can be obtained in a similar way by the socalled stationary phase method. Without going into mathematical details, which are unnecessary here, the displacement in the time domain is actually given by the transform [Eq. (3.14b)]: u(x, t) =
;1t J_+oooo ["{k) exp [i[kx -
J
w(k)t] dk
(3.37)
Since U(k) is assumed to vary slightly. the most significant contribution to the signal corresponds to constructive interferences in time t at position x. or, for a properly stationary phase:
c
(3.38)
ck [kx - w(k)t]iko = 0
Hence for: XM
= V,(ko)t;
V,(ko)
= dw dk
I
(3.39)
ko
where V, is the group velocity of the wave packet centered at k 0 and corresponds to the velocity at the peak value. One may well question the relevance ofthe foregoing remarks for the viscoelastic state. The quantity U(k) is transformed into U(k) exp [- cx(k)x]. To apply the method of the stationary phase, it is necessary for this quantity, U(k) exp [- cx(k)x]. to vary slightly in the frequency band considered. This condition is satisfied if the value of cx~k) is small, an assumption that has been experimentally justified for most rocks (see Section 5.1.2). In this case, the foregoing argument leading to 13.39) is valid. If not, no conclusions can be drawn without additional data. Note however that it is the interference effects that led to the concept of group velocity. The group velocity V,(k 0 ) is different from the phase velocity V0 when the phase velocity is frequency-dependent. The medium is then stated to be dispersive. In elastic media, the interferences leading to a dispersive character are of geometric origin and are generally produced by multiple reflection effects (wave guides). Hence, in an infmite homogeneous elastic medium, w 0 is always equal to k 0 V0 , the group velocity coincides with the phase velocity, and the frequency content of a given signal does not change. For an infmite homogeneous viscoelastic medium, this does not apply, because of dispersion due to intrinsic dissipation. The group velocity does not coincide with the phase velocity, and,
3
WAVE PROPAGATION IN VISCOELASTIC MEDIA
111
when a wave is no longer monochromatic, but centered around an angular frequency w 0 , the signal peak propagates at the group velocity, hence:
V, = dw dk
I=
V(ko)
lo
+ ko dV dk
I
(3.40a)
ko
or
_!_ __ 1 __ ~dvl
V,-
V(w 0 1
V 2 (w 0 ) dw
(3.40b) wo
As a rule. in an elastic medium in which dispersion is due to interferences of geometric origin (wave guides, Rayleigh waves), the group velocity decreases with frequency (its minimum is called the Airy phase velocity). When the group velocity decreases with frequency. dispersion is said to be normal. If not, it is said to be inverse. In a viscoelastic medium in which dispersion is of intrinsic origin, the phase velocity usually increases with frequency. and dispersion is therefore inverse( 51 • Note however that the concept of group velocity, obtained by the stationary phase method, imposes a minimum propagation distance for the wave to be grouped in packets. Before this minimum distance is obtained, this concept is irrelevant. In a viscoelastic medium, owing to the preferential attenuation of the high frequencies, for relatively high attenuations this grouping in packets may not necessarily occur, and the notion of group velocity no longer makes sense. Hence the velocity given by the arrival time does not correspond to an easily identified velocity. This problem will be examined in the next Chapter. The foregoing defmition of group velocity is based on a kinematic reasoning. For a wave propagating in a homogeneous linear elastic medium, it can be shown, by applying the Hamilton principle used in the previous Chapter (see for example Eringen and Suhubi. 1975, Achenbach, 1973), that the group velocity is also the energy propagation velocity. i.e.: V
= (P)
9
(IHI)
=
>.
(P (V+C)
(3.41)
where P is the power per unit area, and IHI the total energy density (sum of the kinetic energy C and elastic energy V), while the notation ( ·) indicates that the average is taken over a period
T((·)
=
~
tr·
dt). In the unidimensional case, (3.41) would be written:
( uu '> vg = (lf2pu2)
~,(1 '2u~)
(3.421
On the other hand, since the stationarity principle of the Lagrangian C - \ no longer applies to viscoelastic media due to dissipation, the energy propagation velocity of a narrow-band centered signal is neither the group velocity nor the phase velocity. In the two- or three-dimensional case, for a pure monochromatic wave (i.e. no interference effects) that is not homogeneous (planes of equal phase not parallel to planes of equal amplitude. see Chapter 6). the energy propagation velocity once again is neither the group
(5) An example is the case of Biot's media of the previous Chapter.
112
WAVE PROPAGATIO!'< IN VISCOEL...STJ(' MEDIA
3
velocity [i.e. defmed by (3.40)] nor the phase velocity. Only in infmite homogeneous linear viscoelastic media do homogeneous monochromatic waves (as always satisfied in unidimensional conditions) have an energy propagation ,·elocity equal to the phase velocity (Borcherdt, 1973). From the experimental standpoint, with respect to rocks in an infmite medium where no interference effects occur (as dispersion of geometric origin is obYiously possible in viscoelastic media), one must consider the relative importance of the concept of group velocity. Since attenuations are often such that the parameter "1. is small (low or medium attenuations), dispersion of intrinsic origin is also low, and the group and phase velocities may be merged as a ftrst approximation. Moreover, ifthe ray path for a given experiment is short, signiftcant dispersion, in effect, has no time to occur. Using Eq. (3.30b), the attenuation of a monochromatic wave can be obtained by: 1 dA d oc = - - - = - - - (In A) 16 ' A dxM dx.\1
(3.43a)
or a:= _ _ 1 _ I n A(x.\1,) x.\1 2
-
xM,
for x.\1, < x.v,
A(x.\1,)
(3.43b)
where A is the wave amplitude, and x.v• xM, and x.v, different observation positions. For a wave packet, the general concept of attenuation cannot be extracted from the dispersion effects discussed above. For moderate attenuations. however, Eqs. (3.43a) and (3.43b) can be appplied, taking the signal peak value equal to A. The attenuation determined can then be considered as the attenuation at the central frequency of the signal.
3.3.4 Quality factor, prelill)inary notions The importance of the breakdown of the complex modulus into real and imaginary parts has already been discussed. It is now possible to defme a quality factor Q which quantitatively characterizes the dissipation of the medium by: Q(CJ) = MR(w) M,(CJ)
(3.44)
Note that this defmition involves only the complex modulus. It also applies to any medium modeled by linear viscoelasticity, although it does not draw on any particular viscoelastic model like those which we shall examine in the next Section (Section 3.4). The quality factor is dimensionless. Since the imaginary part of the modulus is zero for a nondissipative elastic medium, the factor Q is infmite. By contrast. a zero quality factor implies an infmitely attenuating medium, without any transmission quality.
(6) The unit of attenuation is the neper per unit length. :x can also be expressed in dB unit lengtl:. and one can write: %dB ur:it len,C:
= 8.686
2:~p
unit 1entth
~
I
3
WAVE PROPAGATION II' VISCOELASTIC MEDIA
II3
Note that deftnition (3.44) is intrinsic and makes no reference to experiments which lead to its measurement, experiments which will be examined subsequently. For the most general linear viscoelastic medium, the developments in the foregoing sections enable Q to be related to other apparent parameters. The factor Q is fust related to the phase shift cp(w) between stress and strain measured on a sample subjected to a harmonic excitation [see Eq. (3.221]: Q(w) = tan q>(w) l Q(w) = cp(w)
Q~ l
(3.45a) (3.45b)
By (3.33a) and (3.33b), it is also shown that attenuation at angular frequency w 0 is related to Q by:
~-1 a=l;,J2 Jl+ 1 ~
Wo
(3.46at
Q2
or alternatively
Q = w0
[t _cx ~~] 2
2aV0
w0
(3.46bl
For Q ~ 1, these equations are written: Q ~ Wo = rcfo 2cxV0 aV0
(3.46cl
B = 2nC""'"
(3.471
Let us consider the value: LIC""'"
where C....," and LIC""'" are respectively the maximum kinetic energy and the decrease in this energy over a wavelength A. C sing (3.30b), it can easily be shown that: B = 2r.[l - exp (- 2aA)] - l
(3.48al
B = 2rc{ 1- exp (- 4n(ji+Q2- Q)]} -1
(3.48bl
which can also be written :
and for small cx(Q
~
1), we therefore have: B~Q
(3.491
114
3.3.5
3
WAVE PROPAGATION IN VISCOELASTIC MEDIA
Example of application of viscoelastic modeling to porous media
Porous media and rocks will be examined in the light of the viscoelastic models and discussed in the following Chapters. However, we should like to present here an example in which the macroscopic origin of viscoelastic modeling can be explained microscopically. To do this, let us consider a medium in which the porosity consists of spherical cavities of radius r, isolated from each other (Fig. 3.6). DiffriCted wave
z
,"::/ --
""'
I
··o, ,_,, 1
...
' _,
_..... I
.,- .....
.....
I
- ..... 0\I
... I
I ' .....
-' lo' I
'
'
I
I
..... I
,-
l(
-,/
'o, '.... _........
(p) . . -.. , I,0 __ \
p
Source
...
2r
I I
0\I
' _....
I
....
""'
Receiver
/
Fig. 3.6 Schematic porous medium consisting of spherical cavities in a homogeneous matrix.
Let us now assume that an incident P wave (see Chapter 2) propagates in this medium. The P and S wave velocities of the medium corresponding to the supposedly elastic matrix· are denoted Vr and J--S. The incident wave is expressed by: u
= u 0 exp
[i(kx - rot)]
(3.50)
where kVp = w. The incident power per unit area is written: P0
= Uxxu
(3.51)
Stress u xx is determined from (3.50) by: q xx =
(A
+ 2J-t)U,x
(3.52)
where A. and J-1 are the Lame constants (see Chapter 2). The average incident power over a period, defmed by: 2"
POv =
w
27t
Jol"' P
0
dt
(3.53)
can then be written from (3.50), (3.51) and (3.52):
POv =
-
~ (A. + 2J-t)wku~
(3.54)
3
liS
WAVE PROPAGATION IN VISCOELASTIC MEDIA
When the incident wave meets a spherical cavity, an elementary scattering process occurs. The scattered wave is the superposition of two waves, an S wave and a P wave. This scattered wave radiates energy in all directions, and this energy is subtracted from the incident wave energy. The latter is accordingly attenuated. To quantify the elementary scattered power, it is customary to introduce the notion of scattering cross-section SE which is the ratio of the average power scattered over a period to the average incident power per unit area. Hence it has the dimensions of an area. Many authors have investigated scattering effects by inclusions. Ying and Truell ( 1956) in particular determined the expression of the scattering cross-section for a spherical cavity. If the incident wavelength is large in comparison with the cavity radius r, they found: SE = gk•r6 m (3.55) If the spherical cavities are sufficiently distant from each other, i.e. if the number of spherical cavities n per unit volume is small (in practice, if the porosity is lower than 20%), it can be considered that no interactions occur between the scattered waves (no interference, and no multiscattering effects). The scattering cross-sections can then be added (see for example Waterman and Truell, 1961). This additivity of scattering crosssections allows the approximation to the f1rst order of multiple scattering to be made (see for example Ishimaru, 1978). In this approximation. it is considered that the direct wave is no longer the incident wave defmed by (3.50), but a wave attenuated by the elementary scatterers on the path already traveled. More specifically, consider a direct wave of average power P"v per unit area. In the volume dx d}' dz, the number of spherical cavities is ii dx dy dz. By defmition of the scattering cross-sections and from the principle of their additivity, the average power lost - dP"r by the direct wave on a path dx satisfies: - dP"v dy dz = iiSEP"c dx dy dz (3.56) from which, by integration: P"v
= POv exp (- 2cxx),
u
= u0 exp (- cxx) exp [i(kx -
rot)]
(3.57)
where attenuation oc is: 1 -s oc=2nE
(3.58)
Equation (3.33d) and the definition of the quality factor Q (3.44) then give rise to:
Q-t = iiSE k
(3.59)
(7) The precise value of the coeiT!cient g is:
41t
g=9
~
[ 3+ 40
(V.)s
2+32 Vs
L
nfVrYl
2
J
-2(vJ +3(vJ +16C,J 3 Vr
2
2 Vr
9
Vr
4]
If the wavelength is no longer large compared with the cavity radius, the expression of SE differs from (355t and the frequency-dependence to the fourth power disappears.
116
WAVE PROPAGATION IN VISCOELASTIC MEDIA
3
Hence, when the elementary microscopic process corresponds to scattering effects, Eq. (3.57) shows that a viscoelastic effect occurs macroscopically. However, the limits of the analysis should be noted forthwith, apart from the limitation on porosity mentioned above. Equation (3.55) corresponds to the so-called Rayleigh approximation or long-wave approximation. It assumes that the wavelength A, defmed by: 2n Vp A=-=-
f
k
(3.60)
is large compared with the inclusion radius, or more precisely:
kr < 0.1
(3.61)
An example of a radius of 1 J1. and a P wave velocity of 5000 m/s lead to a frequency limit fc of about 4 MHz, a limit that is generally an order of magnitude or more above applications of classic seismics (fteld or laboratory). Moreover, ncan be approximated by the expression :
cP
n = r3
(3.62)
where ¢ is the porosity. This consists in isolating the cavities in the tangent spheres of radius R = r/¢ 1 ' 3 • Combining (3.55) and (3.59) this gives:
Q- 1
=
c/Jgk3r3
(3.63)
Equation (3.63) shows that Q- 1 is proportional to the cube of the frequency, and also to the cube of the cavity radius. Hence, at the seismic to ultrasonic frequencies, scattering processes are negligible. They only become important at the very high frequencies. Emphasis must be placed on the fact that it would be misleading (or useless) to attempt to reverse the law of viscoelastic behavior defmed by (3.63) in the time domain. Apart from the high frequency limit fc of the analysis, other dissipative mechanisms with far more importaJ.l,t effects occur in the low frequency ranges (i.e. f ~ fc), where scattering processes are negligible. Thus, if the cavities are ftlled with a fluid, it is the viscosity of the fluid that plays the most important role in dissipation. Walsh (1969) pointed out that a fluid inclusion in a homogeneous matrix satisfactorily obeys the standard model developed in Section 3.4.1. The central angular frequency c.o, [see (3.76)] is then equal to ep.f'f, where e is the aspect ratio< 81 ofthe fluid-ftlled cavity, '7 the viscosity of the fluid, and J1. the shear modulus of the matrix. Other authors investigated the problem of wave propagation for a concentration of cavities ftlled with viscous fluids (in particular Datta, 1975). While these investigations are in themselves interesting in attempting to relate macroscopic viscous models to microscopic effects whose physical origin is clearly understood, they are nevertheless limited, because the experimental conftrmations are not conclusive. This is due to the fact that many other dissipation processes occur. They are analyzed in Chapter 5.
(8) The aspect ratio is the ratio of the two extreme dimensions of the inclusion: for a sphere this is l, for an ellipsoid with a major axis a and minor axis c. this is cfa and e ~ l.
.,-
:I
i
3
WAVE PROPAGATION IN VISCOELASTlC MEDIA
117
3.4 IMPORTANT VISCOELASTIC MODELS 3.4.1
Rheological models, new defmitions of the quality factor
Rheological models are often used for theoretical support in describing uniaxial behavior which displays dissipative effects of viscous origin. We shall show their general interest later. These rheological (viscous) models consist of networks of two elementary models, the spring and the dash-pot shown in Fig. 3.7. 11
E
~
IINIIIIN
•
(a)
a=Ee
(b)
Fig. 3.7 Elementary linear viscoelastic models. a. Spring b. Dash-pot.
If the force acting on the element is denoted by a and t denotes its extension, the equations of beha\ior are:
• For the spring: a= Et
(3.64)
,t
(3.65)
• For the dash-pot: t1
=
In its simplest form, this equation represents a linear and viscous behavior which linearly relates the force (or stress) t1 exerted on the element to the extension rate (or strain rate) e to which the element is subjected. We shall now examine the case of a commonly used model, the so-called u standard" or Zener model, shown in Fig. 3.8.1f t 0 and t 1 are the extensions (strains) of the two springs, using Eqs. (3.64) and (3.65) one obtains:
e = t 0 + e1 a= Eoto a= Ettt +'let
(3.66)
Eo
11
Fig. 3.8 Standard or Zener model.
118
WAVE PROPAGATION IN VISCOELASTIC MEDIA
3
where£ is the total strain of the model. It is then easy to deduce from (3.66) the equation:
a+ r'X)a
= E 0 r 00
e+ E
00
(3.67)
£
where we have set: 1: cc
=
I
'I
1
1
(3.68)
-=-++ E 1 ·Eoo E0 E 1
E0
The parameter r is called the characteristic relaxation time as it is representative of an experiment in which the strain is imposed [ d~:fdt = 0 in (3.67)] and in which the relaxation of the stress (to be imposed) to maintain this strain is observed (see Section 3.1.2). The parameter Eoo is the delayed elastic modulus; after an infmite time interval, it linearly relates the stress and strain (a= t = 0). On the other hand, £ 0 is the instantaneous modulus. It can then be shown that the solution of (3.67) can be written: X>
I
a(t) =
0£ r(t - r) -d dr
t
-oo
(3.69)
1:
where the expression of r(t) is given by: r(t)
= E + (E0
r(t)
=0
00
E aol exp ( - r
-
~)
t
~0
(3.70)
t < 0
The function r(t) is hence the relaxation function of the standard model, which had been defmed in general by Eqs. (3.5) and (3.6b). Proceeding as in Section 3.2, one can determine from (3.69): a(t)
I
+oo
= _
with
m(t - r)~:(r) dr
(3.71)
00
. dr m=dt
(3.72)
(3.70) and (3.72) can then be used to obtain the complex modulus relative to the Zener model: M(w)
= M0
1+r 00 r 0 w 2 •
+
t
2 00
W
2
•
+ zM 0 w
r 0 -r, • · 1 + ! 200 W
(3.73)
where we have set:
Eo _ !!_ M 0 = E-r_
to = r :c E
-
E1'
(3.74)
00
The complex modulus .\f(w) tends towards M 0 = £ 00 as w tends towards 0. M 0 is then called the relaxed modulus. Motion has become infmitely slow, and the dash-pot does not play a role (t = 0); the Zener model (Fig. 3.8) behaves like two springs in series E 0 and E 1 , or a single spring with a constant E'XJ defmed by (3.68). If the angular frequency tends towards infmity, the complex modulus tends towards M oc = E0 • As motion becomes very fast (t = oo ), the dash-pot blocks the movement ofthe springE 1 and only spring E0 reacts.
tr d
~~
3
119
WAVE PROPAGATIOS IN VISCOELASTIC MEDIA
The quality factor is determined from the defmition (3.44) and Eq. (3.73) by:
Q=
1 + t:/0 t 0 ci 0(t 0 -
t
(3.75)
,,J
The maximum 1/Q representing maximum dissipativity is obtained for the central angular frequency rom: w
m
0.76)
= .ytot:r: ~
and is therefore 1/Q(ro,.,)=
to- tao ~
2-y t 0 t
(3.77)
00
Figure 3.9 gives the variations of Q and MR(ro) as a function of angular frequency ro.
.------..----r-----.----.----....-----.
1.1
0.16
,.0_,._
2-;r;;:;:..
~MRP(w)
1.0
0.12
0.9
0.08
a-1
0.04
0.8 1
""'m ==
-;:r;;;: I
0.7
-=r=
1
-3
-2
I -1
I I
0
I
7't:
I
2
3
0
Angular frequency (log IClllt)
Fig. 3.9 Three-parameter solid (Standard) model.
The Zener model represents the network of three elements. Two other simpler models are defmed by the limit cases. One of them is the Kelvin-Voigt model, obtained for E0 ~ oo. This is the prototype of solid rheological models, because the delayed elastic modulus Eoo is non-zero. The second is the Maxwell model, obtained for £ 1 ~ 0. This is the prototype of fluid models. because the delayed elastic modulus tends towards 0; in other words the delayed strain is infmite and the material "flows" indefmitely (see Fig. 3.10).
i~
120
3
WAVE PROPAGATION IN VISCOELASTIC MEDIA
~~ ;I I
~
'I
Eo
1-f
If-IN
•
a
I lal
Time
.E~
'I
::!:J---
I!
en
a
El
~=I I
I Time
Fig. 3.10
lbl
Maxwell (a) and Kelvin-Voigt (b) models and their creep functions.
The Maxwell and Kelvin-Voigt models are very important because it can be shown (Mandel, 1966) that any linear viscoelastic solid can be represented by a series network of Kelvin-Voigt models and a spring (constant M 00 ), representing the instantaneous elasticity (Fig. 3.11 ), or in an equivalent manner by a parallel network of Maxwell models and a spring (constant M 0 ) representing the delayed elasticity. If the representation by Kelvin-Voigt models is adopted, in a similar manner to (3.66), this gives the equations: II
e = e0
u
=M
00
e0
+
L
i= 1
(3.78a)
e1
= E 1e1 + rue1
(3.78b)
For elements in series, the inverse of the resulting complex modulus is the sum of the inverses of the complex moduli of the elements. For the Kelvin-Voigt model, (3.78a) leads to: MkJJ) = E1 + i'71w
(3.79)
Hence the complex modulus satisftes: 1
1
1
= - + : LM-(w) M(w) M"" 1
(3.80)
1
For harmonic problems, it has been shown (Section 3.3.2) that, for a harmonic strain e = e,. cos wt, the stress is written :
u=
Bm
lMl cos (wt + qJ)
(3.811
I
I
ir 3
121
WAVE PROPAGATION IN VISCOELASTIC MEDIA
llj
llj-1
111
-1
M_
E1
eo
----+ Ej-1
a
Ei
Mj
e1
Ej
ei-1
Fig. 3.11 Representation of a linear viscoelastic solid using Kelvin-Voigt models.
The introduction of (3.81) in (3.78b) gives: IMI
+ cp- cp1)
e1 =e.., IMJI cos (c.ot IMI
e0 = e.., -
Moo
(3.82)
+ cp)
cos (c.ot
where we have set:
+ iM1;
M = IMI exp (icp) = MR
(3.83)
M 1 = IM11 exp (icp1)
The elastic energy at time t is written:
_1 1 \1-2 M:ceo
2 + "1 4- 2 E1e1
(3.84)
J
Hence from (3.82): 2
_ 1 2 2 [cos (c.ot + cp) \1 - 2 e.., IMI Mao
£
"
1 + 7' IMl
2
cos (c.ot
+ cp -
J
(3.85)
cp1)
From (3.80) and (3.83) it can be inferred that:
1
. .
IMI (cos cp -Ism cp) =
l
"
£
."
'IJc.o
1 MCC) + 7'1Ml-' 7'1Ml
(3.86)
Equations (3. 79), (3.85) and (3.86) then lead successively to:
\1
=~
2
2 [IMI 4 e... M QO
2
"
+ L...j
IMI £ 1 cos 2(c.ot + cp) IMI 2 IMJ·1 2 + M QO 2
IMI~ £ cos 2(c.ot + ~ IM 2 1
1
~ e2[cos 2c.ot Re ( Mao+ Ei MJ V =- e!Mo + 4 • M
2
1
4
1 e!MR \1 = 4
M
2
[(M2 M2) Moo + MJ £1
J
cp1)
(3.87a)
(M2 + EJ M2)] Mf
)
· 20Jt 1m M
-SID
+ 41 e! Re
+ cp -
exp (2ic.ot)
•
J
(3.87b) (3.87c)
122
3
WAVE PROPAGATION IN VISCOELASTIC MEDIA
From (3.80) we obtain: 1 eM
1 oMj
- =· IM~-ow M2 cw J
(3.88)
J
Hence with (3.79): oM OW
M2
~! 2
w - =I - 2 j Mj
i'1jW
=Ij -Mj2 (Mj- Ej)
(3.891
Equations (3.89) and (3.80) then lead to: 2 M2 M2 M2 E-M oM 1 M=-+ I-=-+ I-2-+wM oo i Mi Moo ow 1 Mi
(3.90)
Finally this equation introduced into (3.87c) gives the following expression for elastic energy:
W=
~ e!MR + ~ e! Re [ (M- w ~~) exp (2iwt)J
{3.91)
This expression had already been obtained by Bland (1960) and by O'Connell and Budiansky (1978), but in a less direct manner. Equation (3.91) gives the maximum elastic energy W""'" in the form :
wmax = 41 e! {
M R+
[r\ M R- w eM CWR)
oM ) -p:} J 2
2
+ ( M, -
w aw'
(3.92)
Thus, as stressed in Section 3.3.2, the maximum elastic energy, contrary to a classic error which consists in treating it as 1/2e!MR, depends in the general case on the real and imaginary parts of the modulus and their derivatives. Based on (3.91), however, the average elastic energy Wau is written very simply in the form: Wa., =
f T 1
2Jt
T=c;-
\1 dt =
1
4 e!MR
(3.93)
0
This is valid in the general case of viscoelastic models assuming a discrete breakdown as in Fig. 3.11. Since the energy dissipated per cycle is (see Section 3.3.2): JW
= ne!M1
(3.94)
Eqs. (3.93) and (3.94) immediately lead to a new defmition of Q:
Q = 4nWa., AW
(3.95)
In the general case of a rheological model, Q can therefore only be defmed from the average elastic energy and not from the maximum elastic energy. Nevertheless, this is possible for low and medium attenuations (Q > 10). In fact, the quantities MR and M 1, the real and imaginary parts of the complex modulus, are related, through their very defmition, by the Kramers-Kronig integrals (Nowick and Berry, 1972). These relationships are linear, and it can be shown that, if M 1 = 0, then oMRfow = 0 (see Appendix 3.1). This result can be understood by noting that the existence of an attenuation
wr-
•1
il
3
WAVE PROPAGATION IN VISCOELASTIC MEDIA
123
in a material (M1 :f:- 0) imposes the existence of a velocity dispersion [V = V(w) or what is equivalent MR = M~,(w)]. The functions M,(w) and c.\I.Rfow therefore display equivalent asymptotic behavior when attenuation becomes low: Q > lO ou M,
~
MR
<=>
oMR) ( M,. ow ~ MR
(3.96)
Accordingly we have:
Q > lO
1
2
\/,.." ~ 2 e,.MR
(3.97)
which is valid for only low dissipations, contrary to (3.95). Hence in the case of low attenuations, an approximate expression of the quality factor is:
2n\/,.." Q 1 Q ~ LtW ; ~
(3.98)
However, the demonstrations that led to (3.95) and (3.98) concern viscoelastic materials for which the representation by Kelvin-Voigt models corresponds to the discrete breakdown of Fig. 3.11. The term discrete does not imply that the elements are fmite in number, but that their characteristics E1 and 'IJ are isolated values on the positive real axis. The creep spectrum is said to be discrete and the material is said to have a short memory, because only the present stress and strain time derivatives are involved in the constitutive equation [see for example Eq. (3.67)]. If this is not the case, the representation by rheological modeling loses its value, since in particular, Eqs. (3.95) and (3.98) can no longer be demonstrated (although often accepted) because the discrete summation (3.84) of the energies by means of hidden parameters such as the strains e0 and e1 is no longer possible. The material is said to have a long memory, because all the past stress and strain values are involved in the constitutive equation which is available only in the form of a convolution product. To overcome this problem and to account for experimental observations that will be discussed, we can therefore, as in the following Section, defme the model directly using the qu.ality factor, rather than infer it a priori from a rheological model. The value of these models (for Constant or Nearly Gonstant Q) is obvious at the operational level, but it must be kept in mind that. in this new defmition, we lose the advantage of an explanation of apparent macroscopic behavior by hidden parameters, an explanation that leads to the development of rheological models.
3.4.2 Models defmed from the quality factor 3.4.2.1 NCQ model (Nearly Constant Q) Certain experimental results (for example, Murphy, 1982, Spencer, 1981) have shown that the quality factor remains constant over wide frequency ranges. The idea of Liu eta/., (1976) was hence to construct a mathematical model for which Q would be nearly constant, by direct superposition of Zener models, all having the same relaxed modulus M 0 but with different central angular frequencies [see (3.76)]. This yields the frequency dependence for 1/Q and velocity as shown in Fig. 3.12.
..·~I 124
3
WAVE PROPAGATION IN VISCOELASTIC MEDIA
~I
I
10' 8
r
0.01
1/0
10 4
10'4
0.005
0 4.8
v (km/sl
4.7
Phase velocity
4.6
4.5 10' 4
10'.
1
10 4 Frequency (Hz)
Fig. 3.12 Typical dispersion relations for the inverse of the quality factor and velocity for the NCQ model (after Liu et al., 1976).
The same result is obtained if the Zener model superposition is no longer discrete but continuous (Liu et al., 1976). The NCQ modc;l's dispersion equation is then: V(wl) ~ 1 + _1_ In col V(co2) 1tQ w2
(3.99)
Other authors have reached this conclusion mathematically by assuming Q is independent offrequency over a wide band (Azimi et al., 1968, Strick, 1970). Lomnitz ( 1957) also arrived at the same result by using an experimental creep law of the form: 0 t
\
where q, d and M 0 are experimental constants. 3.4.2.2
CQ model (Constant Q)
The naturally following mathematical step consists of building a model with Q strictly independent of w. This model was derived by Kjartansson (1979). The practical advantage of this CQ model in comparison with the NCQ models is clear. For the NCQ models, the dispersion relations and the frequency dependence of the quality factor are merely approximate relations, valid only for quality factors over 30. Moreover, the NCQ models all imply the introduction of a parameter related in some way or other to the frequency band over which Q is constant. The width and cutoff frequency of this band appear to be perfectly arbitrary, and the physical implications of this cutoff frequency vary according to the models (Lomnitz, 1957, Futterman, 1962, Strick, 1967, Liu et al., 1976). The Constant Q model is very simple mathematically, and is completely specif1ed by two parameters,
.;;r3
125
WAVE PROPAGATION IN VISCOELASTIC MEDIA
namely the wave phase velocity at a reference frequency and the value of Q. The creep function used by Kjartansson ( 1979) has the following form :
0 f(t) = CQ
{
t~O
1 M_rtl
t
-1-
?v\ Co)
27
(3.101)
t~O
where r is the classic gamma function (see, for example, Abramovitz and Stegun, 1972) and M 0 is the modulus of the complex modulus at the reference angular frequency w 0 = 2n/t0 • This creep function has already been analyzed by Bland ( 1960). It implies that the complex modulus is: 2
iw ) M(w) = M 0 ( Wo
2 Y=
w M 0 I Wo
1 Y exp
[i1t}' sgn (w)]
9 1 )
(3.102)
The quality factor Q is then given by: 1 - = tan (1tf')
(3.103)
Q
Figure 3.13 shows the behavior ofthe creep function vs. time for the Constant Q model. It is interesting to note the absence of instantaneous elastic strain for this model, unlike the previous Lomnitz model (1957). Let us examine the form and properties of the impulse response of this model. To analyze the impulse response amounts to examining how a Dirac delta function c5(t) is propagated: (3.104)
u = u0 c5(t)
This pulse is emitted at a reference abscissa x 0 in a medium satisfying the wave equation (3.29). Since the plane wave solution is exp (-ax) exp [iw(r- xfV)], the Fourier transform of the impulse response (after a travel distance x) is: H(w) = u0 exp (- tXX) exp (- iw
~)
(3.105)
For the Constant Q model, it is shown using (3.33), (3.102) and (3.103) that: ex = tan
~ sgn (w) ; ' ~ = Vo (~oy Vo =
I: r
(3.106)
....!...:.---'--
cos
1t}'
2
hence the function H(w).
(9) The sgn (·) function is the function which yields the sign of the quantity inside the parentheses.
126
3
WAVE PROPAGATION IN VISCOELASTIC MEDIA
0= 10
0=1 1.2 3
-!
-8 1.1 -~
-~ ~ 2
c.
..
i
1
c
c
-~
"£
u;
1
II)
0
.9 .8
0
2
6
4
8
0
10 Time
2
1.002
"B
1.01
1.001
~
'a.
i
10
8
0= 1000
0=100
c ·;;
6
Time
1.02
i
4
c.
i
1
1
c
!
~
.999 .998
.98 0
2
6
4
10
8
0
2
4
6
10
8
Time
Time
Fig. 3.13 Creep function oftheConstant Q model (after Kjartansson. © 1979 AGU).
The impulse response h(t) is then the inverse Fourier transform of H(w). This response is shown in Fig. 3.14for a unit pulse(u 0 = !)emitted at x 0 t 0 = QV0 as a function of time tft 0 [see (3.101)]. An important property of the impulse response is i
(X )t"=""Y
1 TocLfoc-oc -
A
V0
(3.107)
where T is the wave travel time for distance x, A the signal width, and A its maximum amplitude. This relation can be written: T A= b(Q) Q
(3.108)
where the function b(Q) depends on Q. It can be shown (K.jartansson, 1979) that b(Q) is virtually constant for Q > 20, which allows the development of a measuring technique for Q called the measurement of the signal width or measurement of rise time (see Chapter 4). To compare the Constant Q model with the NCQ models, the creep function (3.101) can be written in the form :
{~) = Mor(! + ly) exp [ 2y InC:)J
~-----
-------~~--~--
(3.109)
~
3
127
WAVE PROPAGATION IN VISCOELASTIC MEDIA
0.6r-----;:----------, 0=1
i' "
0.4
II~
0.2
0
5
o'
I
10 Time
5
~
5
..i
0.4
i
!c
0.2 01....---'-...L.--===~---L..J
105
100
______ ,
10
,,
15
20
0.6r--------------,
0=100
95
1:
Time
0.6 r - - - - - - - - - - - - ,
I
Oc 10
i !G 0.2~'
0.41
i
0.2~ 0·
110 Time
0= 1000
~
995
II::=-±-
1000
1005
II
1 010 Time
Fig. 3.14 Impulse response for the Constant Q model (after Kjartansson. (C) 1979 AGU).
Assuming that expansion:
y
is small (i.e. Q > 10), /(t) can be approximated by its ftrst order CQ
~~) ~ Mor(~ + 2y) [ 1 + 2y ln r:]
(3.110a)
1(1 + 2y) = 1 - 2yf3 to the fust order
(3.110b)
Moreover: where
p is the Euler constant: f3 ;;;;: 0.57721 (Abramovitz and Stegun. 1972)
(3.110c)
~~) ~ ~0 [1 + 2{1n (:J + P]] = ~ 0 [1 + 2y In rrJ
(3.111)
and hence
In the Lomnitz equation (3.100), the quantity d. inverse of a time, is very large in comparison with the time resolution of the experiment. Therefore 1 + td ~ td, which when introduced into (3.1001, implies that: 1 /(t) ~ -M (1 NCQ
0
which is clearly of the same form as /(t). CQ
+ q In td)
(3.112)
128
3
WAVE PROPAGATION IN VISCOELASTIC MEDIA
The same thing may be observed by starting with the dispersion equation (3.106):
v=
Vo
I:J' ~ vo(l ;J +yIn
(3.113)
in which Eq. (3.99) can also be recognized. The Constant Q model is hence the same as the NCQ model for quality factors greater than 10. This is not surprising. The Constant Q model is a limit of the NCQ model constructed from an inftnite set of Zener models. The Constant Q model offers the advantage of being easy to use in various situations, due to the frequency independence of Q. Physically, however, its validity is not generally greater than that of any other models.
3.5 VIBRATIONS IN VISCOELASTIC MEDIA 3.5.1
Traveling waves and vibrations
Let us consider a cylindrical rod of an attenuating material, of length l, initially at rest and ftee at both ends. For a short instant, let us now impose a force at the end x = 0. for example by an impact, and receive the signal at the other end x = l. The recorded signal is shown in Fig. 3.15. A series of pulses at intervals Tis fust observed. These different arrivals repeat the shape of the ftrst pulse, but their amplitude decreases and their duration lengthens with increasing distance from the original instant. A damped sinusoid of pseudo-period T gradually appears. This signifies the transition from unsteady state conditions to damped vibrations or pseudo-steady state conditions. The equations of the problem are the unidimensional wave equation (Section 2.1.5.2), for an elastic medium, which is recalled here :
u.xx
1 ..
(3.114)
=Vi u
where VE is the extension wave velocity and the boundary conditions: u.x(O, t) = f(t); f(t) = 0 u.x(l, t) = 0
t<0
(3.115)
where f(t) is the force exerted at extremity x = 0, normalized by the product ES (E Young's modulus, S rod cross-section). Since the derivative u,x also satisftes the wave equation, the analysis will be based on this quantity. By making a change in variables: X
Zt = t -
VE
X
z2
= t + VE
(3.116)
~ 3
129
WAVE PROPAGATION IN VISCOELASTIC MEDIA
0.4ms
.........
I I
I I
I
I
I
I
T
I
I : • T .:. T --+--'--~
1 ms
........... Fig. 3.15 Experimental signal recorded at x = I. The source is a short pulse at the origin. The sample is a Plexiglas rod. The top recording is a magniftcation of the frrst milliseconds of the bottom one.
the· wave equation then becomes: (3.117)
u,x%1%2
This equation implies that the general form of the solution of the wave equation is:
u,.,(x, t)
= t/l(t-
;J
+ q>(t +
;J
(3.118)
The function t/1 corresponds to a wave propagating in the direction of increasing x, while the function q> corresponds to a wave propagating in the direction of decreasing x. These functions thus correspond to the successive reflections occurring at both ends of the bar (at x = 0 for t/1, and x = I for q>). More precisely, it can be shown (Courant and Hilbert, 1962, for example) that the solution to (3.114) and (3.115) is: 00
X ) u,.,(x,t)=f ( t - -
vE
+L
•• 1
[
-f
(
2nl) ( v£
2nl)] v£
X - - - +f t -X- - t+
v£
vE
(3.119)
The f1rst term represents the propagation of the signal imposed at x = 0, while the second and t~rd terms correspond to the n•h reflections at the ends x = I and x = 0. The beginning of the signal in Fig. 3.15 thus corresponds to the successive reflections of the original pulse. In relation to (3.119), however, established for an elastic case, the dissipative processes deform the signal in two ways: on the one hand, energy is lost, and the peak
130
3
WAVE PROPAGATION IN VISCOELASTIC MEDIA
amplitude decreases, and on the other, the signal spreads due to the dispersive character of the waves (i.e. the velocity depends on the frequency). This spreading corresponds to the one observed for a Dirac delta function and a CQ model in paragraph 3.4.2.2. The question arises how to interpret the later appearance of the damped sinusoid. Letus particularize the function f by a brief excitation modeled by a Dirac 15(t). Equation (3.119) then gives:
( VEX) + L
u.x(x,t)=b t - -
[
X -2n/) -b ( t + - +b ( t -X- -2n!'J] (3.120)
VE
n;!
VE
VE
VE
Taking the Fourier transform [Eq. (3.14)] of this equation, we obtain: GO
U,x(x, k*) = exp (- ik*x)
+
L
(exp [- ik*(x + 2nl)]- exp [ik*(x- 2nl)])
(3.121)
n;l
where k*, the wave number, is given by: k* VE = w. The equation can be expressed alternatively in the form : :X:
U,..,(x, k*) = 2i exp (- ik*l) sin k*(l - x)
L
exp (- 2ik*lp)
(3.122)
p;O
which, taking the sum of the geometric progression, leads to: U ' x(x, k*) =
sin k*(l- x) . , ., [1 - exp (- 2ik*lp)], p Stn .
-+
+ oc
The integration of this equation gives: u (x, k*) = ucolt + uincolt
(3.123)
(3.124)
where ucolt =cos k*(l- x)
k* sin k*l • lncolt _ _
U
-
exp (- 2ik*lp) cos k*(l- x) for P -+ k* sin k*l
+ 00
(3.125)
The term ucoll is the coherent part of the signal representing the contribution of constructive interferences of the reflections to the general signal. The incoherent part uincolt, however, has no limit. Its phase is random, except precisely at frequencies such that: k* _ mn I
(3.126)
m-~
At these frequencies, ucolt tends towards infmity. This fmally gives the schematic spectrum in Fig. 3.16, whose irregularity results from the contribution of uincolt which varies at each reflection. · Equation (3.123) corresponds to the non-dissipative elastic case. If one considers a viscoelastic attenuating medium, the wave number becomes complex (Section 3.3.3.1): k*
= k- ia
a> 0
(3.127)
r
3
WAVE PROPAGATION IN VISCOELASTie MEDIA
131
1-
-!
.~
..
Q.
E
~ §
~ i5
),
.J."-
w,
____, ·'-"'3
"'2
· Angular frequency
Fig. 3.16 Schematic diagram of the spectrum in the elastic case for any point x.
When introduced in (3.123), this expression shows that the incoherent part now tends towards 0, and we obtain:
u = ucoh = cos k(l -
x)
k sin kl
(3.128)
Hence, in the viscoelastic case, U has a limit whatever the frequency, with local maxima at frequencies close to:
w,
m1t
k, = VE(w,) = -,-
m = 1, 2...
(3.129)
For sufficiently low attenuations, the velocity Vt:(w) can be considered as independent of the frequency, and the Eqs. (3.129) give the local maxima to the nearest second order. After a sufficient number of reflections, the signal is mainly composed of damped sinusoids of angular frequencies w,.. ·As a rule, attenuation increases with frequency, and the contributions of the sinusoids to displacement disappear sooner for larger values of m. The results are shown schematically by the spectra in Fig. 3.17. Note that the period T in Fig. 3.15 is simply 21t/w 1 = 21/Vt:, a time which clearly corresponds to a round-trip of the transient wave, and explains the spacing Tofthe pulses of the ftrst part of the signal. One can thus qualitatively explain the experimental result in Fig. 3.15: the original signal propagates, is reflected successively at the ends, and, due to dissipative effects, decreases in amplitude. Only the frequencies corresponding to W 111 contribute significantly to the signal, because they correspond to constructive (in phase) interferences of the different reflections, and thus give a signal with a sufftciently high amplitude to be able to propagate without being attenuated too rapidly. To directly attempt to derive the pseudosteady state asymptotic solution requires making a modal analysis ofthe system, in other words ignoring the transient conditions and developing the solution on the normal modes
132
WAVE PROPAGATION IN VISCOELASTIC MEDIA
3
defmed by the characteristic angular frequencies w,. This modal analysis will be examined in the next Section.
~
a =a ~ E ~
§
"§. i5
"'t
"',
"'2
Angular frequency
Fig. 3.17 Schematic diagram of successive spectra (p .1') in the viscoelastic case.
3.5.2 Modal analysis 3.5.2.1
Normal modes
Let us consider an elastic rod and determine the standing waves of angular frequency w which may be established therein. The corresponding displacement can be taken in the form: u(x, t) = Re [u(x) exp (iwt)] (3.130) The introduction of (3.130) in (3.114) gives:
u,xx
+ k2 u =
0
k =
(I)
VE
(3.131)
The solution is straightforward: u(x) = A sin kx
+ B cos
kx
To obtain the asymptotic solution for a bar free at both ends, it is necessary to satisfy the boundary conditions: or
u,..,(x = 0 or x = l, t) = 0
(3.132)
A=O, sinkl=O
(3.133)
leading to the characteristic angular frequencies (or eigenvalues) and to the normal modes: . mn mn mnx k, = - -, w, = - - VE, u,(x) =cos - -, m = 1, 2... (3.134) 1 1 1
r
3
WAVE PROPAGATION IN VISCOELASTIC MEDIA
133
l
I
which are precisely the angular frequencies ro, discussed in the previous Section which correspond to constructive interferences. The term .. normal modes" arises from the fact that these modes are mutually orthogonal, in the sense of the integral scalar product: (2/l) J:u,u" dx = {
~
m :F n m=n
(3.135)
This orthogonality means that the inertial forces in modem, proportional to w!u,, do not contribute to the displacements u" of mode n :F m. Let us now consider a viscoelastic bar subjected to linear external excitation p(x, r). The equation of motion is (Section 3.3.3.1):
a2 u
m(r) * ax2 (x, r) = pii -
s1 p(x, t)
(3.136)
where Sis the cross-section of the bar. By applying the Fourier transform we obtain: d2U
M(ro) dx 2 (x, ro)
pw 2 U-
=-
1
S P(x, ro)
(3.137)
Let us now consider the associated elastic bar defmed by the modulus M(w) = M(O) = M 0 1
and the velocity VE = (.\10 / p)2. Let us break down the solution of the viscoelastic problem to the normal modes of this associated elastic bar : U=
L a,(w)u,(x)
(3.138)
"' where u..,(x) is defmed by (3.134). We multiply (3.138) by u" and integrate. The orthogonality of the normal modes (3.135) gives the following as the expression for the generalized coordinate a,.(ro):
ff
U(x, w)u, dx
a.,.(ro) =
(3.139)
u! dx
By similarly multiplying (3.137) by u., and integrating, we obtain: M(wl
2
f
2 ddxU 2 u., dx = - pw
f
1 Uu.., d~ - S
f
Pu., dx
(3.140a)
Integrating by parts and taking account of boundary conditions (3.132), we obtain: M(w)
f
2
u dx U ddx;'
+ pw2
f
1 Uu., dx = - S
f
Pu., dx
(3.140b)
The normal modes satisfy: d2u ~-"' dx 2
2 + w., v2
E
u,. = 0
(3.141)
134
3
W.WE PROPAGATION IN VISCOELASTIC MEDIA
Equations (3.138) and (3.139) introduced into (3.140) thus give: 1
a,.(w)
f
P(x, w)u,.(x) dx
(3.142)
= R,. [M(w)/M0 ]w!- w 2
where R,. = pS
f u! dx
(3.143)
For the extension modes examined here, R,. = p Sl/2. However, we shall leave R,. aside, because the previous developments remain valid for other types of vibration (bending, torsion, etc.) which do not necessarily satisfy R,. = constant. These other types of vibration will be analyzed experimentally in the next Chapter.
3.5.2.2
Forced vibrations and free vibrations
Let us ftrst consider a harmonic excitation of angular frequency ro 0 exerted at point x 0 P(x, w)
= 27tP0 b(x- x 0 )b(w- w0 )
:
(3.144)
By (3.138) and (3.142), the response is: r· _ '- -
~
L..
.. = 1
27tP0
u,.(x 0 )u,.(x)
2
2
~( _ ) u W Wo
2
R,.w, M(ro)/M 0 -ro fw,.
(3.145)
By introducing M(w) = .\fR(w) + i.\11(w) into (3.145) and inverting the expression, we obtain the time domain expression. where we have formally replaced w 0 by w: U (X,
t)
=
~ 1
.,-;;: 1
u,.(x 0 )u,.(x) cos (wt - ({),.)
P0
-2
R,.w,. [(MR/M0
-
w 2 /w!) 2
1
(3.146)
+ (MJIM0 ) 2]2
where the phase difference ({),. is given by : MJIM 0 tan({),.= MR/Mo- w 2 /w!
(3.147)
Let us now consider free vibrations, obtained by an impact. The excitation, assumed to be exerted at x = 0, is thus:
= P0 b(x)b(t) P(x, w) = P0 b(x)
p(x, t)
(3.148)
Equation (3.148) introduced into (3.142) with (3.138) leads to:
U=
L _!l_
u,.(O)u,.(x) .
,. R,.w! M(ro)/M 0
-
w 2 fw!
(3.149)
Without any additional assumption about M(w), it is impossible to proceed further. Let us therefore consider the case in which damping is slight. In this case, in the vicinity of an angular frequency w,., one can write: (f) ::: (1)'" '
M(w)
~
MR(ro,.)[1
+ 2ie,.J
(3.150)
r I
3
I~
WAVE PROPAGATION
135
VISCOELASTIC MEDIA
where ~ ... is the reduced damping defmed by:
2e .. ~ .\.f,(w.,.)
(3.151)
.\.f~~.(w.,.)
•
Due to the slight damping, the maxima of(3.149) are clearly separated, and it is possible to invert (3.149) in the time domain:
u(x, r) =
I
m:l
RPo
2 mWm
u..(O)u.,.(x) exp (-
e. . w.,.t) sin co,..t
(3.152)
This equation corresponds to the qualitative result of the experiment in Section 3.5.1.
3.5.3 Defmitions of the quality factor using vibrations When dealing with vibration methods, the quality factor can be defmed using Eqs. (3.146) and (3.152). Let us fust consider free vibrations in the case ofs1ight damping (Q ~ 1). By filtering, each term of the series in Eq. (3.146) can be isolated, and each of term then corresponds to a damped sinusoid (Fig. 3.18). The logarithmic decrement b., is then defmed as the logarithm of the ratio of two successive maxima, or as the logarithm of the ratio of two displacements, for a time interval equal to a pseudo-period: f> ... =In
u(t) u(t
)
+ 2nfw...
(3.153)
by (3.151) and (3.152): 1 Q(w...) ~ 2e ...
·1 •
-!
£
l
Tm=211/W
m
.-1
1[
= b,.
Q~1
u ltol•- ~mwm(t-tol •
Llto+T I /
~mf
Fig. 3.18 _ Damped sinusoid corresponding to the mth characteristic frequency [Eq. (3.152)].
(3.154)
136
WAVE PROPAGATION IN VISCOELASTIC MEDIA
3
For slight damping, the quality factor is thus related simply to reduced damping and to the logarithmic decrement. The quality factor is obtained in the vicinity of the characteristic· angular frequency Wm· In the case of the impact experiment previously described, the first characteristic frequency is the ohly one to make a significant contribution [factor 1/w! in (3.152)]. Hence only Q(wd can be determined. To obtain the quality factor at higher frequencies by the logarithmic decrement, it is preferable fJCSt to subject the bar to forced excitations near the desired angular frequency wm, and then interrupt the excitation. The free vibrations then occur at the desired angular frequency given the adequate initial conditions. Let us now consider the case offorced vibrations. Each term of the series (3.146) can be placed in the form: Po
u..,(x, t) = - R 2 cos (wt - rpm)A..,u,.(x)u,.(x 0 )
,.w,.
(3.155)
where A.., is the amplitude term: 1
A,.=
1
[(Ma/M0
-
(3.156)
w 2 /w!} 2 + (M.fM 0 ) 2 ]2
Assuming slight damping (Q ~ 1), we know from the Kramers-Kronig relations that Ma ~ M 0 (see Section 3.4.1 and Appendix 3.1) and the amplitude term is reduced to: 1
A,.~
1
[(1 -
(3.157)
w2 /w;.) 2 + (MJ!Ma) 2]2
Assuming slight damping (Q ~ 1), the contributions corresponding to each characteristic angular frequency are clearly separated, which means that if the exciting angular frequency is close to an angular frequency w..,, only the term A,. has a significant contribution in the series (3.146). Figure 3.19 thus shows the influence of modes 2 to 5 on the response with excitation close to the flfst characteristic frequency in the case of a slightly attenuating material Q =50 and strongly attenuating material Q = 3. For the case of Q = 50, it is apparent that the higher harmonics can be totally ignored, but this becomes less valid for higher frequency or lower Q (Fig. 3.20). Assuming Q ~ 1, one can then proceed with a classic measurement of the quality factor, called the frequency sweep. Let us determine two angular frequencies w 1 and wu (see Fig. 3.21) such that:
A!(Wr) A!(Wu) = w.., w,. =
2
A (1)
2
(3.158)
We can then show that:
Q~
w,. wu - w 1
= w.., = _1_
A..,w
2e,
Q~ 1
(3.159)
where A,w = wu- w 1 quantifies the spectrum width and rises with increased damping.
----- ----------------------------
T
3
137
WAVE PROPAG.... TION II' VISCOELASTIC MEDIA
I
I
E
1
0.4
~
0.2
ol
I 1000
. "'
:r
1500
.
=:e'
I
1
2000
J
2500 Frequency (Hz)
Fig. 3.19 Contribution of harmonics 2 to 5 to the shape ofthc f1rst harmonic. The model used is the Constant Q model. The modulus M 0 is 5 GPa, the reference angular frequency is 1000 s- 1 • The density of the material is 2.5 gjcm 3 , and the sample length is ..W em.
0
F---;----=w-· I 3000
'
1 3500
'
J 4000
r
:
1
~
F.-quency (Hzl
Fig. 3.20 Contribution of harmonics 1, 3, 4 and 5 to the shape of the second harmonic. The parameters of the model are the same as in Fig. 3.19.
The measurement of the phase difference between excitation and displacement offers another means to determine ru1 and wu. In fact, for MR;;;; M 0 , the expression of tan cp.,. (3.147) immediately shows that: tan cp.,(w 11 ) = -tan cp,..(ru1) = 1
(3.160)
The excitation and the response show a phase difference of ± :: 14. In the case of rheological models. we have shown in Section 3.4.1 how the factor Q could be defmed from the energy dissipated over a cycle and the average or maximum elastic energy involved during a cycle. These defmitions apply to an experiment in which the excitation is the stress and the strain is the response (i.e. no inertia terms, sample dimensions negligible). One may well ask whether this applies to a rod in which the dimensions are no longer negligible.
I I
l
t
138
3
WAVE PROPAGATION IN VISCOELASTIC MEDIA A({l)
1
a= 011 _n
AM(O}
1
VV2AM(0}
w
n, Fig. 3.21
n=-
"'m
nu
Principle of the measurement of Q by frequency sweep.
To answer this question we consider the assumption by which damping is sufficiently slight for the response to be given by (3.155), when excitation occurs near the angular frequency w,. Owing to the conservation of energy, the dissipated energy A W is the energy furnished to the system during a cycle, because the elastic energy (reversible) is totally restored at the end of the cycle. The dissipated energy, for the case of excitation giYen by (3.144), can be written:
-; f 2"
AW =
du
(3.161)
P0 dt"' (x 0 , t) dt
0
The introduction of Eq. (3.155) for u,(x0 , t) into the foregoing equation gives: Ll W
=
1tP~
R
u!(x0 ) sin cp,.
2
,.w,. [(MR/M0 -
1
(3.162)
ru 2 /ru!) 2 + (M1/M 0 ) 2 ]2
For rheological models, the average elastic energy V,., in the case of slight damping, for which MR ~ M 0 , can be written from (3.93) in the form:
V,.,=4M 1 0
f'
s!dx
(3.163)
0
where s, is the maximum deformation at point x, namely: s,. = max ou,(x, t)
•
t
OX
Introducing the expression of s,., taken from (3.155), in (3.163), and substituting the expression ofsin cp, deduced from (3.141), we obtain, after a few calculations: 47tV,..,
Q ::::-LIW
Q~
1
(3.164)
-
~------
T
3
--
------~-
WAVE PROPAGATION IN VISCOELASTIC MEDIA
139
I
I I
I
This shows that the local expression (3.95) can be generally applied to the resonant bar with the difference that Eq. (3.164) for the resonant bar is valid only for dampings such that Q ~ 1, whereas (3.95) was valid assuming rheological models alone. Simultaneously, it can be shown that: 27tV""'.x
Q ~ AW
Q~ 1
(3.165)
where \1 ""'x is the maximum energy.
3.6 CONCLUSIONS CONCERNING THE QUALITY FACTOR AND FINAL REMARKS Table 3.1 shows all the ways to deftne or approximate the expression of the quality factor Q, whose reference defmition is M .. 1M 1, and notes which experiments underlie these defmitions. Each of the experimental techniques (resonance or propagation mechanisms using standing or traveling waves) allows the direct or indirect measurement of a speciftc quality factor Q. In fact, every value of the quality factor Q = (M .. I M 1) is associated with the type of excitation analyzed, and hence with the elastic modulus concerned. A bar in extension allows the measurement of QE = E~./E., where E =E.. + iE1 is the complex Young's modulus. A measurement performed on shear waves allows the measurement of Q5 = JLRIJL~o where JL = Jl1. + ip1 is the complex shear modulus. We can also defme Q.c = Ka/ K 1, where K = K .. + iK1 is the complex bulk modulus even if a direct experimental measurement seems ditTtcult to realize. Winkler and Nur (1979) showed that, for slight attenuations (i.e. by ignoring the products ofthe imaginary parts of the elastic moduli) and by using the correspondence principle (Fung, 1965) (i.e. by replacing the relations between real terms of Table 2.2 by the same relations but in complex terms), it is possible to obtain simple relations between Qp, Q5, QE and Q.c: (1 - v)(l - 2v) 1 + v 2v(2 - v) Qp =~Qs 3 1 - 2v 2(v + 1) -=--+---:-
QE
QK
1+ v
3(1- v) ~= Qp
Qs
(3.166)
2(1 - 2v) Qs
The Poisson's ratio vis obtained from the velocities by the equation: v=
v~ - 2 v~ 2 2 = 2(Vp- V 5 )
Vi -
2 v~ 2 2Vs
(3.167)
TABLE SUMMARY OF QUALITY FA<-TOR
Parameters concerned
Ct,
w,
v
tiC.,.,., C...,... A, et
Propagation
Propagation Harmonic excitation on sample
Vav• tiW Harmonic excitation on resonant bar
\!max• tiW
Harmonic excitation on sample
DEFINITIONS
Viscoelastic models
Sections
All
3.3.4
Q ~ 2etV
All
3.3.4
Q ~ 2nC..,.. LICmax
All
3.3.4
Q ~ 4nVa• LIW
Rheological
AnyQ
Q~l
I Q=tan cp
Q~-
Q =-~(I- C(2~) 2etV w2
Possible experiments Harmonic excitation on sample
cp
3.1 (Q = M,jM1)
c_..
--=(1-exp [-2etA])-
Lie_..
Q
4nV•• =---
tiW
I
cp
(I)
1
3.4.1
3.5.3
2nV max -- tiW
Q~-
3.4.1 Rheological
Harmonic excitation on resonant bar Ll.,w
15.,
'"'
Harmonic excitation on resonant bar Free vibrations on resonant bar Free or harmonic vibrations on resonant bar
-----------~- ---~~---·
35.3 Q~ w,.,
Ll.,w 1t
Q~-
15.,
l
Q~-
2~ ..
All
3.5.3
All
3.5.3
All
3.5.3
3
WAVE PROPAGATION IN VISCOELASTIC MEDIA
141
where the different velocities are related by 2
V,. = V2
_
v:(4 Vi 3Vi-
Vi)
Vi
V~(3V~- 4V~) v~-
E-
Vi
(3.168)
2 2 4 2 VK=V,.-3Vs
The ratio Q,./Qs can also be written as a function of (V,./V5 ) 2 :
Q,.= Q5
"'' K
4
I+ Jill
v~
-2
Vs
(3.169)
Finally, it can be shown that one of the following relations always occurs:
QJt > Q,. > QE > Qs Q" < Q,. < QE < Q5 Q"
for high V,./Vs ratios (e.g. in dry or totally saturated rocks), for low V,./Vs ratios (e.g. for partially saturated rocks),
= Q,. = QE = Qs.
3. 7 CONCLUSIONS In this Chapter we have considered the problem of wave propagation in viscoelastic media mainly from the phenomenological standpoint. More speciftcally, experiments (creep, relaxation) ftrst led us to the laws of linear viscoelastic behavior defmed by the convolution product (3.9). In some cases, it was possible to obtain a representation of these laws by the rheological models developed in Section 3.4.1. In most cases, however, the creep and relaxation functions are only implicitly known through the quality factor Q, and only over limited frequency bands. Nevertheless, the physical origins of the validity of these different models are not generally established in an indisputable manner. Note that the use of a linear \iscoelastic model enabled us to defme the quality factor Q in the form MR/M 1, and to relate the different ways to defme various quality factors which, a priori cannot be deduCed from each other by the equations in Table 3.1. Hence, for the linear viscoelastic model. the quality factor concept is quite independent of the type of experiment concerned, and is an intrinsic parameter of the medium. For linear non-viscoelastic models, the defmitions of Table 3.1 normally allow a characterization of various damping effects rather than one damping effect. Thus Biot's theory clearly leads to the notion of attenuation ex in propagation, and hence to a quality factor (c.o/2cxV), and yet this quality factor is not equivalent to the one that could be determined by an experiment with a resonant bar (White, 1983). From the practical standpoint, the "quality factor" parameters obtained by experiment can subsequently be compared with each other in the light of the assumption of linear
142
3
WAVE PROP:\GATION IN VISCOELASTIC MEDIA
viscoelastic behavior. It is clear that this is merely a model that may not necessarily represent reality. Laboratory experiments nevertheless show that, as a ftrst approximation, this model is satisfactory and can serve to analyze field problems, for example. The many causes of dissipation, and their relative preponderance according to the frequency range analyzed, normally preclude any attempts to make a too sophisticated defmition of a complex modulus. and hence a factor Q (and even more the creep and relaxation functions), by refming the models developed in this Chapter. The procedure required can be identif1ed clearly. It proceeds from a dual standpoint. experimental and theoretical. From the experimental standpoint, it consists of analyzing the variations of the quality factor for each type of rock, which appears to be the most interesting quantity to measure, apart from velocities, as a function ofvarious parameters (frequency, porosity, water content, etc.), for which care has been taken, as much as possible, to isolate these effects in an ad hoc experiment. From the theoretical standpoint, it consists of a careful study to understand the major physical effects responsible for the observed attenuation at the frequency studied. This makes it possible to model the processes, or rather represent them by viscoelastic models such as those developed in this Chapter. These models thus appear more as working tools rather than as an end in themselves. Adapted to the rocks analyzed, they can therefore provide guidance for a mathematical modeling of the processes involved, not only at the level of the laboratory sample, but at the level of f1eld seismics. We shall discuss these points in their various aspects in the subsequent Chapters.
Appendix 3.1
THE KRAMERS-KRONIG RELATIONS It has been shown in Section 3.2 that: u(t)
f
+JC
= _cc
m(t - r)e(r) dr
(3A.l)
where m is the derivative in the sense of distributions of the relaxation function r: dr m = dt
with r(r) = 0, r < 0, r(O+) = r0
(3A.2)
Since the complex modulus is the Fourier transform of min the sense of distributions: +or:: dr M(w) = MR(w) + i.\11(w) = d exp (- iwt) dt (3A.3)
f
-0::
t
the expressions (3A.2) give: MR(w) = r 0
M1(w)
=-
+[
f -
+:x: Xl
o::dr -d cos wt dt 0::
d
t
..!... sin CJt dt dt
(3A.4)
--3
WAVE PROP:\GATION IN VISCOELASTIC MEDIA
143
since the discontinuity of the relaxation function rat t = 0 is a Dirac delta function r 0 c5(t) for its derivative, considered in the sense of distributions. In (3A.4), unlike (3A.3), we deal with integrals of ordinary functions. Similarly, one can introduce (see Section 3.3.2) a function such that : ~>+ao
e(t) =
I
df
j(t - t)O'(t) dr
.. -:o
By introducing the Fourier transform J(w) = la(w) Eqs. (3A.4), we obtain: Ja(w) == io
f f
+
+ao
+ iJ1(w)
simultaneously in the
df
-d cos wt dt t
-ao
+:o
J 1(w) = -
(3A.5)
j=dt
(3A.6)
df
-d sin wt dt t
-X)
Note that Eqs. (3A.l) and (3A.5) require that: J(w)M(w) = 1 Hence in particular : J(oc).\f(oc) = J 00 M 00 =jor0 = 1 J(O).\f(O) = 1 0 Mo =j 00 r 00 = 1
,_
(3A.7)
In fact, we know (see, for example, Sections 3.2 and 3.4.1) that the instantaneous or unrelaxed modulus r 0 = r(t -+ o+) and the delayed or relaxed modulus r x = r(t -+ oo) correspond respectively to the moduli Mao (inf10ite angular frequency) and M 0 (zero angular frequency) in the Fourier domain. By eliminating the relaxation and creep functions between (3A.4) and (3A.6), we obtain (see, for example, Nowick and Berry, 1972) the so-called Kramers-Kronig relations: 1
2w M aW ( ) = Mo+-
2 M 1(w) =n:
la(w)
J 1(w)
i
0
n:
ao
1t
00
o
M 1(«) d« -----
«
(Ma(«) - Mo)
f
1 +2 . = v.YJao 1t 0
'w = :_
i
i"' (
00
J•(«)
-
Mao
d«
«2 - ( l )
« d«
1 ) la(«)--
0
(l)
2 OC
«
wl -
.-
-
(l)
(3A.9)
- 2d« --
W
(3A.8)
-«
•
where the integrals are taken as principal values. Thus with Eq. (3A.8), one can show that, if M 1 = 0, then Ma(w) = M 0 is a constant.
---
·~
.....1 1.
<
~
•
<':1
;;,1~
1.
.").-,~f~H·
~-..,.·
~,J
.,;,
4 experimental techniques for measuring velocties and attenuations '-
INTRODUCTION
, __ "--•
.....
·
'-
The prnioul Chapter attempted to show that the concepts of velocity and attenuation were complex, even at the lewl of their dcf'mitiou. We have stressod that eac;h of these definitions stemmed from a different experimental technique, and that. mathematically, the deftnitions could be coqsidered to be equivalent for low or medium attenuations. This statement is true from the mathematical standpoint: in other words, for each experiment, each measuriag metftOCI· was idealized formodeliil;.ln fact. the many interference effects, or, more simply. experimental dift'tculties encountered, mean that the measured quantity is not exactly the one anticipated or that it contains a substantial error. For instance, the l'DC81UleiDCIIIt of aueauatioa :M quilc d6cult. ovea ita the laboratory. It is nec:enary to extract tho.._ we WMl ftoiD . a·Witolt.,..:of illlelfenuas-~ ~ multiple nleclioM. ........ etc.)wlaoletl'ccts MOIRORI or less foraeeable and imply the aeed lor ~ This llmlals tlae impoetaace. of a souad knowledp of expcriJnoaaalteeluaiquaforiiWUUriaavelocityandaueauaboaforaaoocJ undentancliDJ oldie reliability- ~ oftbe .... aad to he able to GOmpue various sots of uta found in tlac literature. Three maia caM&oria o(.meuuremata can. be dist~: ·. (a) Measurements us~ travolioa waVes. (b) Measurements usin& vibratiaa systems. (c) Measumnents usia& stress/strain curves which are distinguished from the second catqory, althoup it uses an cquivttent ~~tation system. For each~ we shall examine tbt·eom:ctions to be applied to the raw data;
~
~
146
4
TECHNIQUES FOR MEASURING VELOCITIES AND ATTENUATIOJI;S
4.1 MEASUREMENTS 'USING WAVE PROPAGATION 4.1.1 · Difficulties Measuring methods using travdiag. waves are especialLy interesting because, by defmition, the mechanisms involved are propagation processes.~imilar to those of seismic exploration. Naturally, for laboratory measurements, the frequency range is totally different (around I MHz) as compared with the frequencies used in the fteld (50 Hz in seismics and 10kHz in logging) (see Ftg. 4.1). The wavelengths are hence quite different, and the mechanisms responsible for the deformation of ultrasonic and seismic signals are not necessarily the same. Furthermore, the velocity dispersion processes associated with the presence of attenuations become significant. Exploration geophysicists are thoroughly familiar with the problems encoun~.in tryW, to ~late a seismic section to a sonic log (Goetz er al., 1979). The extrapolation of the results obtained in the laboratory to the fteld is therefore a difficult problem. e.mq.,•• I
10'1
•
100 SEISMIC
~it- 4.l
lchoiiiiUI.... ·, .....,..
EJqltoq1ioft
I 10
,'o; ACOuSTIC
. ·;I
tal ..
I
1o-
l.illiloreiDry
I 101
I
~,
10"
ULTRASOtiiC
.FUQUCDCf.~ of ~.~ur-.cn&L,
~
· NeverthelesstJaemetllodsthat\Vemddcscrib'eareWOtrllall)ta,..blc·tof'llllet__.. aswellaslaboratorylifnals.lnthe'fieli.·a.--Gfeft'ectstlfstwebotltdle.,.....tioft and shape of tksipal itself. ·For scillllic prGpaptkm, tile naaitt di«tcuuciee . from the ignorance of the distance traveled by dlle wave. 11lil waw.ce.uinlfdtlft·not follow the straipt ray path generally 8JIUM, 'but rather a eutVeCl path~ to''thc prOifeSSive variation in the acoustic pMpertiesoftlte formationsaw:ouatered. Moreover, even if straight ray paths are assumed, the depth of the different interfaeet is unknoWb. At best, if the multiple refte«iOM have been elinlllrated cort11atly. one may determine the travel time to the different interfaces. The int~~l vel~y ~t~~n two i~terfaces is then determined by the maximum likelihood enctgy metftod. Fin-~:Y, iq ~g,J seismics, the existence of a poorly-known weatb~ :z:one (WZ), w~icb, .is iDJwnioaen~us and . attenuating, further adds to the djfftCul~ in ~Cnninins the vel()Qty. In brief, it is assumed that we can distinguish variatiOnS of 1'Yo to 2o/o on seismic velocities, but that the absolute value of these velocities .is only known widaill 100.4. ·As lot acoustic loging, the formations encountered by the propaptins wave are better known, but mechanisms such as wave scattering at the fractures may give rise to errors in the velocity (e.s. cycle skipping). In the case of very high attenuations, the signal received may be very weak and the velocity measurement very disturbed. We shall be discussing a laboratory example below.
.ra
...
•
"'·
j
'
----T -..__..-
·-/
•'~'
~·
I - }t"·I I
I -
~
;
4
l'l!dfNtQul!s POl( ~·~AMD1A'I"n!NUATroNs
147
We have jUst descrit'Jed some- of· the dilfle'bltiet eac6Untered in the fteld in the · · measurement of velocity; Hence itisdear that the~t ofattftuationwiHbe ~ motedift1cvltlto adlieft.lnfact, tile attenuation ofbtlerest to us is the tnrrlPISic attenuation
of the medium, namely that relafe(Ho·flle interadion between the wave and the porous medium and its saturating flui~ u opposed to the extrinsic attenuation. which depends on the geometry of the beds and on the source (scattering. internal multiples. geometric divergence, etc.). In the fteld, in fact,•the simple propagation of the wave in the rock formation is modified by,pttering. and by internal multiples that constitute energy losses unrelated to the lois dUe 'to the iM1uticity c:Jf tfN,media tl"''fetsecl. It is~ difficult to correct these drects (Schoenberger and Levin. 1974~ ~is why reliable attenuation measurements in the freld are taken either in well-known aDdllomoseneohs regions, using well-to-weU propap~a. for cupaple. or by .~JeCOI'diaas obtained ia.t.he wella and by ~rial siinaJa that have. travclo4 . . . .._t.patbs. A second major source of dilf~eulty is tbe ~of~ ItisacocraUY NIUDlCd that the initial wave.ftoJ&til~whereK.U.:r.it~~tnJir~ Tbismattcris welllaaown to sipal pr~l .ptb~ w~ to ~ tlaeir nKX)rdiags for seomctric diver~, apply a U.. A»m!dioo·io r' (~here /l. f!l Oaad ~the t ~n«Pectcd for a spboric:al wave. The ~- tbc qJIJDCrical value of the factor , remains CJDpirical..AAotllef ~of problems io ~a. .ualio~ais4be biah·noise level in.
di-..or
scisaait~ TM~of-.iad~~oot~lyrclatcdtotbe desired~ For~plc.iathe~,..U~ica,~wavcs(pstudo-RayJeiab or
Stoneley) could constitute .. noise" if one were interested in the refracted S arrival. This rapid description of the processes involved in the deformation of the signal points out the extreme difr)tUky obtainina meaain,.ul resUlts in the raeld.
or
·--
·,_
With rupect to laboratoryJDeUurements, some of the problems mentioned above c:an be eliminated. To the ..-lium analyzed is well(!) known, the interfaces deftned, and the ray paths relat~ely clc8r, pftll'llly yieldina accurate Velocity measurements ("itbin J'!. ~ Yet a 4iftlculty arilea ila ,..._.,accurately the dme or the.Jirst arrival in an atteJtuatina ~ Weshall.,....i~~~~t ttu.l.~ in detaiHt~ the next Sec:tion. As for attenuatioa meuumpenta, multiple reflections-.cl iatcrference arrivals can be eliminated 1 MHz (for ccntimctric · (e.a. by usia& samples of *'*tuate size). For~ saqaples), however, the lfODletric ~vcrpnce effect. ~Y ~important, and, as a rule, operations are conducted at discanoes &om the ICtUfCC JUch that the emitted wave is vi11UAUy a plaac wave. the raaJe Qt. this wOrk\fta iet~l de.pends on the site ·of the transducer, tho waveleaath emitted, and the distance between the emission and obsavatioapeints. In this zone, sinoetlle wave front is virtuaUy planar it is assumed that the amplitude decay of the displaeement due to the anclasticity of the medium is exponential.
beam "'tb,
leas....._
'-._.../
If the wavelenath becomes comparable to the transducer diameter, diffraction may occur. The correction required can nonnaUy only be made empirically (Trudl et al., 1969). Another problem, typical of laboratory propaption measurements, concerns coupling problems betweeaube·ttaasctucer AJMlthe sample(ltt6 a the reprodUcibility oflossts due to'coupliaJ). Mete ....-. tbe frelllim 1b8UW tolwd'empirically (Truell n al., 1969). when -~ tllvdw.l ift ·ultrasollic measurements, the Finally, O'WiOJ· to samPles must haw :pr«!11efy parallt1 Skies. This l'loblem ft5 ilr¥estiaatcd experimentally by Truell atld Oates (1963), wllo' shM.red th-.'\he ~of imperfect paraiWism
th..,
,_
~
•
148
TECH:"lQl:ES FOR ME!,SUIUN(i
4
VELOClT~ AND-~TIENUATIOl"S
depended on the inverse of the quali.Ut ~r ofthe ~rial used. and on the frequency employed. A lars,c; part of~ diff¥:ul~ listed abQvct can be .solved experimentally. Ne,·ertheless, attenuation measuremc:a.ts by wav• pr9DQ&tiOO remain a ~ult matter, albeit easy to design and implemeat. The reliability of!~ methods and the acx:uracy of the measurements of the quality factor are at bc;lt .10%.
4.1.2 1\-leasuremeat prbaciples _. 4.1.2.1
~xpe~
tecltaiques
,Ai
Velocity m_easantaeats
As shown in the prtvious Chapter, ~ion 3,.3.3.2, tfwrfne8SUI'enlent of velocity in an attenuating medium by means of tile Y ·a X/t (where j and· r are the travel distance and propagation tinie respectively) jives usi hmriOgelleous qdantity for velocity, but one that has no particular phyliead'signifrcance for t~signal analyt.ed. Contrary to the measurement of the distance x, Wbidl is nQnllally acetrrate anct easy to make, the measurement of time r is ditrteult becau_, a l*tje number'offac:tors add to the inaccuracy, such as length. shape and repett'tion rate of the etecttk:at pulse, 1he break of the emitted wave (directly related to the rise time 11etifted in Fig. 4: 12), the cbarteteristics of the · rertiver system. and \-he·rransducet aditl'fttstenill'l ~ A nUmber oft~ problems can be etiminated by using a reference *ftlple witll welMmoWil acoustic pioperties.
eetuatidtl
,•
.1~
~.·
-1
t"·'
l
t
t 0 .........
. ."'
.
,. '" .
?'lf'
-1
T-. ..).
0
0.5· -!
1
.......,.._.(MHZ)
Fig. 4.2 Signal in time (left) and in the frequency domain (right) used for the model. ., i~. i
J-
;
'
Nevertheless. the p:oblem.. ~~ ·with wa~ riso .time JDoasurcmenu r"mains uQSQlved. Let us consider two ~pies.
~
;;,--
'-'
"-----' "-../
..
·~Mix.M_.,ItiYI~IitlMfli'WlMSAf.f*"'*tllliS
Note also that, for the Constant
'-
Q 1\,lQde~. d1e phase velocity is pven ))y:
v-
'-
Vo(;;y
l
l
i' =- tan- 1
'----
'-'--
._.,-
'"-../
'-
,_
-
~
"--'"
'-
(4.2)
-
Q
7t
'-
(4.1)
with
'-~
'-
149
The greatest di«crence in phase velocity for the two samples is therefore obtained at the maximum sipikaftt bquency. Frequencies lbove t»0 /2n propapte faster in tile more attenuatina . mediUm (Q • 10) than in the less «tenuatina medium (Q • 100). The ()ppOJite occurs rot frequencies below m0 /2x. The lipal, which bu traveled throUJh dMUDOCiiUDl orquality factor Q • 10 hence anivesbefore the ODe that bu traveled tbrouah the medium or quality factor Q ... 100. and will be wider. This is lhowu in Fig. 4.3. To~~~~UU~Wdae~illpropapU. tiiDcbotwecn tbc twosipall.and hence the dilenmce in propaptioo velocity, it is necessary to set a level bdCMV which it is impossible to distinpish the signal from the noise. H this level is set at 1°/o, the relative differeDce ill velocitic$ is 4.2%. whereas if the level is set at 4.%. tbis difference is .-educed to 1.7%. l'bis~ is obsetft'Me on actual slPIJs,lll
amvaiS
(FiJ. 4.4).
or
.
.
even
.
.
·~--~~--~----~~----~----~_.----~~
Ju
·~
~
oI
I
0
1.1 h2>r
I
'-
''-
-G.I
~'-
'~
-1.' 0
'
I
I
5
I
I
•g
1
"'
I
,
tO .
•
..
I
I
I
I
I
15
I
I
I
20 TiiM (Ia)
'-
life. 43
Sipali recGrdedafter p~ itta,COMtat Q thateial. their maximal amplitude being nonnalu.d ten.· ·
'---
L
j
' 150
TiCMNIQUES FOR MtiASURJNG Va,OCITIES AND .4 TIENUATIONS
4
!
l
0.5
--
----
Initial signll
"
0•100
(l_·~"-
~
~,
'
0
0.6
0
Ffc. itA
F~CMtfal
Spectrum of sipals recorded ·after propaptioll in a constallt Q
material.
Hence ia the~ of Q == 100. ~Jlc farst ~v~ cop-esJX>nds to an averaae frequency ro 1 higher than the a~ frequency Wa l,)f tJie second aiTival in the case of Q == 10 : The contri.,utipn of the frequ.cncy co, ha$ ~smaller than the,averase nqtse. This makes it very difticult to C~~e velodty ~sur~entS _obtained-for a given s&inple, whose
"
quality factor has varied (for example, by changins the saturatiot\). For average attenuations (Q ~ 10 to 20~ an .. error" of the foregoing type of about l% can be t!xpected due to the technique employed. Nevertheless, the determination of propaption time remains the. CISSential problem. Figure 4.5 shows an exnmee~ple. TJte actual signal emitted has propagated through
II
" o.s 0
., - -----------
I I I
.....
-O.S ..
-1
s
~
~
~
a
I•
•I
51'S
~
•
_,
~
G
~
Time ~Psi
Actual 'lilaal tta.t lw ;W.voled throUJh 40 DlJP of FontaineiNeau sandstone with a measured Q ;; 2.
Fil- 4.5
"
.-,
"
'_,
,--
~--~--~------------------------------------------------------------------------------------------------------_J
'-
..
'~ "--'
~~~...-.·-~~~
1$1
a sample of Fontainebleau sandstone with a quality factor Q ~ 2. The uncertainty on the starting location of the ftrst arrival is about 1 p.s. while tbe travel time is aboqt 22 ps, makin& a relative error of So/o. . Hcacc it can be considered t"-j&Jae ~aventiolaal ~~ bcrc to measure velocities yields velocity values with an uncertainty of about l to 2% according to the attenUation·~ the material in~tipted: : · Yet 1llCUIJ'femcftts ttrat are~ more pllysic:ally·nteanla&fUI (phase \'elocity) and more accUrate can be perfOrmed by develepins a -spedftcsy*'in \'dt)'cities (see, for example,
for
Trueftdal.,I969);Wewillbritflyd~prepolidbyRoaezaadBader(t984)
i
for the measutement of phase telodtieS in l . .(sie'flla. '4.6). A moaochromatic wave is emitted in two different liquids, a non-attenuatiq reference liquid wit1r lmoWII wlocity and the test sample. As a ftrst approximation, the velocity cliftierencc A Y is related to the difference in propaption time At. This difference in propaption time is meesurcd by a calibrated Pbuc analyzcrusins the tM- ....... Scattering and tempeqlWR6etJ ~rtm.lly aqlipblc sinee-.... ~.-..- llellfc>r the sipal piopaptins in the refcftace liquid and ~n the_tcst ,..pic. Fiaally, this technique requires the emission of- lipals having prcc:Uc frequCnc:y Control to eliminate attenuation etrects. The velocities are measured with accuracy of about l.S crn/s.
cere..........
-~
an
~-
Acot.dc
'-
lftllliftlltlniiA
..-.
ChlnMIA
'-'
,._
)
cl'nlctor
........
'--
, .. ,,
'-._/
"---
'-
'-
4.l.a.2 AWrlfir.,... ...- •' ••••• Two main techniques are used to analyze wave propaption in an attenuatina medium: pu)IMcho methods and tra~ssion m~tlloda. n.Jmere~ principles are the same, but tbat application reveals •differences in det,Ut . , -•
hJu-«luJ metlunJs In pulse-echo methods, the sipal emittec;l by a''pie:~XN~Iectrie-ayltal bonded to the 1 sample underaocs muffipte refleCtiotls at. t~ ~ ~·plane the attenuation is determined from the ~ 1Un&ifkude o( two · sutcessive multiple retlec:tions : .. . L
ittt41'face.
'-
•.-.l..tn(~) .· 2L r
'-'
• \_
~~
;,
'
wa•
(4.3)
__________
' lS2
4
TECKNfQUESFOR twt!AtiURDrGV!Lotmi!S' AN'D·ATI'ENUATIOMS
where L = sample length, A 1 = spectral amplitude of a multiple reflection, A 2 = spectral amplitUde of the next trmltiple refleCtiOn.
· ·,. ' ,·
Note that, as we showed in Section 3.3.3.2, thisequan~~ valid only for tl:te QaSC of an atteo.uation «that isJtQ~ t091alp"'and ~a m~odlr.o~tiC siB~.~ PlCthod 8$SWDCS negligible losses. at the, d~ intc!f{ac;cs. e~i~y the. tr~/bOod/sample interfac:es. ~ver, iUS4UI.QOS~ ~t l<.ls~ a conditio~t r_.y rcalizcd for e~riJneats without any ~~ ~ FF1o 4:7 S:ehematically s~ws the experimcatal setup. ~· -,
•
SLm:::I=~~I
.>•
f. l
-~ .'
~
l J
·'·
'
Fla. 4.7 Diagram of a pulse-echo ex~ with q'*tz tr~uc:cr. ' .:~.- :\ "~ '.'c:. :::-.. :.. . / ' -·\,, ..,->~:.- ; ·-. . ~- " This method raises an McJi!fOpa( ~- .,t ia......... id ri.ake Q1casurcments on hiply attenuating samples, be<:ause all tile energy, •.•bsc)rbed in t"' paths inside the ~pi~, 1'tle solution is to reduce the loo.llb of the sample. It is also . . . .., ~e lure
·1i. .
'tted .. by theFinally transducer. propaga .. tioo dit~'nam . ely thatthickness thew.aveol the th' sample. one ID\IIIt ·e ·~that:tbe rcftec:ti.S on the edges of the sample res.W..,~.&MCMCU.Vff\;'1 it · . ·· ,·:·" ftoalbe.toua:e arri"-,arv.* direct signal, 01' have negligible energy'm Com it. Tliis problem is in fact fairly difticult to solve, and involves the theory of bounaecl ~- The interested reader can refer _to HostCD and ~mP,fJ (l~f4~ . ; , •. ~v. ,.d . • . . ;r ,1.. , . W10kler and Plona (1982) used a Jllodirtcation of'dlis technique; Tbaii' ·~pd consisted in comparing the reflections obtained at the two interfaces formed by a fnt bUffer and the top of the sample, and a second buffer and the bottom ofthe sample (Fig. 4.8~ The overall system could then be placed in a hip-pressure ce1f. for dffs setup, attenuation is Jiwn by.: enu.··
I
I
.
.
I.
. F ·t·t~.Yt J4(f») ' 2' . oc = 2L ln IR 12 (w)l :A'(ro) tt - R, 1{~)lt ·
(4.4)
where L • == Jength of ~e,~ .. , ... A(w).. A'(co) oo;,~~t~ of tpe tw~ ~1ons tQ be a>m~d. ~1 2 (w) = retlcct4oa ~t 1'1 ~rftce 1/2. R23 (w) == reflection codttcicnt at interface 2/3.
It should be noted in Eq. (4.4) tha~ tbC ~tion coefficients depend on freq~y. This is due to the anelastic character of' tlie materials concerned (see Section 6.2). Various --..
-,
•
'\
.......
¥[. :
.:
ns
~1~·· '9L61 "sRWpwdwd .S) ~ ~ liDW ~,...'*!P 3!-11~ (:) pue q ._6.t "I!.:J) as1l:) J(W UJ •laAta»J pn JOU!1D511~ ,qtuilii~ . ....., lUO»JP.P OIAl lOJ J)Op.JOOQJ SfHI!I oql lupedQIO:) ~rp UOf,lRQll1t ·amttOQl. Ol a{qiJIOd l! Ia~ INJ. .lpOJCl
af.J1f , , 1IUOJ a1lp ~~r. .... ~~ ;.trr~J ut.~*-"~~·~··,.,., IMU~lct~t! ..-ftlt~Rt~WtJPftMJ~'IJCM1·11f·~
lft,Pl'«) pn ·~ • • ~~saoo~~ ui1lllb fPfi ~N 'M6t 'D!WfU?ft) poJ 1'3!JPU!fA3 8 .8f· ~ OJ BJ~ ~.a. p»JtDUal DA81A IK(l pua •. _ . " ' Q! 'lw.A ~ llf.dar~ ~ 'M·t "1!.:1 uJ ~q!lftlla p:llappao3 :Kl 08:) atefuan :M{l JO sap!5 ~ql UIOlJ SUO!P'IJN pUlr ·~ QJ U1lql lsaf IJ' l(d1llal 01(1)0·~ Mil 86•t "~!.:1 UJ ·(n.\p:».t pn JOUJUII'Wl) UO!'~ Ja)t\p5U11Jl pue
Pm"'.
sr.-...
ms:.rcfiiawforttlg.~~·-·~---A·*••*-*.tlJO-'tw• aqi·ao ,.nq·~ub,sdlaoii.Mt1«'.fq ~~IJ·,..._,,.....WM·M.f·~ fJUO*·~ (Mip·,q.,,...,., ~),.,.., ...,.,....,.l 'q .
.·
•
d
·uOflvnuaue ~ :M{l JO tuauwnsam JO .(:)e.Jn:X)y ·~a~dum__.,..., .........._,UWJ•~'fqfnod aa suo~~ ·saovpaJUJ JU~:,)JP.patp Jlt su~oo lu!Jdno:) puw iUJl:,)ll1tOS ft1pi\I~OO~ lsnm suo~
lOJ o/e01 tnoqe pue %1 Jnoqe · SJ D!JJ."''(aA
e
~)l:lel'
--+k-...---tt I I
I
~.1--~.....-
I I
I
: l ..._ ,
-.-.~-~~~~~~~ .... AIICidJ--ov-·
~~·~----~====r
TECH~"IQt;'t!S l'<»t 'MEASURTNG VELOCITtES :"ND .4 TT£NUATIONS
154
4
Tarif, 1986). In fact. for all transmission methods, one of the most reliable techniques for obtaining the atten~tion from the recorded signals consists in comparing the spectral ~lUnplitudes at dift'~ren.t frequencies. Because attenuation implie5 a preferential loss of the h(gft" f~quencies. a change in the total spectrum will therefo~ occur. The spectral amplitude of a wave can be written (Ward and Toksoi, 19711:
(4.5)
.4(/. Xo) Ill= Gx0 A,(/) exp [- 2(/lxo]
.,
where G
= coefficient including effects of ~ometri(: divergence,
x0
= distance traveled,
transmission and reflection, rx(f)
A,(f)
= attenuation coeft'tciettt. = receiYer response.
The ratio of spec:;tral amptit~
In It is also
A2
(f
'x2
) = rx(f)[x 2
x 1]
-
G1
(4.6)
+ ln -G ;
,.
tnowa tbac.,at the intorclel' in a:-ol: .
1tf' rx(f) = Q(f) V
-...
(4.7)
of
For a given source, the width of tM! 'frequency speCtn0n the transmitted signal is rela~ivelytitnited~ so1b·t~ft em lSt ~!Mto be iftd:epend'eM el'fteqBeftCY. WC shall pro\ide an expet'bnetrtal jdifie:atift· Ofttfisiamtmpt'ioe itt Chaplet S; ·'Bquation (4.6) is theref'ote' writtea·:' · ', "· ' ·· ·
.... -+•U x,> • (x A2(f, :cJ), ,Qf ;1C ·
.,,
The term
In
GV~
is
1 .... :
~~v + 118 ~
.
G:
-'
(4.8)
in~ ~ f~· for a s-plierical ~- ActuaRy,
experitnentat.UU~ts co~ tilt validity of'this bjpottiesis m'liiaJtl111tteftuating samples. The quality faCtor Q'~is therefon~dbtaiMd~· me8suriDg tht Slope of the curve.:
.
. g(ft-~ tf;(/, ~!)
·:r.
. ,·
'
.4:2(/....~:2)
The same method. stiahtly modified, was applied successfully by Tok50z et al. (1979), (Fig. 4.10). They comp8ted the spectra obtained for a reference sample and the studied sample, both samples being of the same length and geometry. These spectral comparison ~niqueJ are routinel~· employed.in, the laboratory and in the fteld. They require a signal as uncontaminated as possible by otkr arrivals. Here also, if the studied material is highly attenuating, the signal recorded after propagation widens, the frequency content decreases, and the attenuation measurement becomes less accurate. In Fig. 4.11, the signal width is 3 J.lS for a Q of 200 and 9 J.lS for a Q of 3.5 after propagation through a Co~t QIJ}e
.,
/''\
..-..,
---.
:;;,----~ '-.../,
~'t'dlt-ill.iSiJ1IIMI~~--
4
! c
''-
I
~
l
~
.I i
'-·
.5
\
I
I
\
-
' .... ..... __
i
I
'-
)
'---'
.,__ ~
lJ
1St'
s-w-
1.0t II
" ::®
!
o.n: v\
.5
c
I I 'I
"--
I I
I I
~
ot
~
0
I I
...
s-w-
I
1.0
0.5
1.5
,_.Mf!CY CMHzl
Fr~CMHtl
'---
'-·-
1.0
Fit. •Ut Example of attenuation measurcmenta by spectral ratio in Navajo sandstone (after Tok.Oz ettd., 1979).
"---"----' '--
'--
-
'-'-~
•___, ''--'
betomes annoyina. For sliJhdy attcnuatina materials, the spectral ratio method is inaccurate, since the llope of the rclfC$Sion JUte if _toot~. the main diftic:ulties or these measurements lD the laboratory aacltb the faelchrites &OIIil eouplina problems. In the field. another difficulty may arise· ftOaa CIODtamiaatioD by iaterferina tipals in borehola:,H...-ver,the major ~ty,~_ .,_$PDCIJalJ&tio tectmique..arisea with pometMdiYfflllalll .-,~--J\on-planar wave fronts. Corrections arc therefore Dledecl for ablolutc 0 measun:meots. These corrections are lbeoretically siJDple, bat involw: tedious c:aJculationa. Papadakis (1976) computed these types of correction for bomopoeousliquid media and his mults do not seem to tit the data for solid materials (Tarif. 1986). Hence. the abtolute accurac:y of Q measurements Jiven in Table 4.1 must'- iaterpreted cautiously, especiaUy for hip Q values. It is also important~ mention d)e tilJC • -~Uf: bued on an empirical equation of Gladwin and Stacey (19741: ·
o.e.or
pr..._ ..........................
'-
'--
- f -
'--• ·~
where Q can vary in· SJ*le but
tnne
to +
r:
c {r l
d~Jl41K . . . . on
• wave rl5c '(5ee FiJ. 4.12), • travel time, t 0 , C • two constafttl de.,..._t ea tbo ...._
t
T '---
'--
--.· '--
ck -
--
frequency and
.(4.9)
156
TECN~E,S Fell MEASUIUNG
4
VELOCITIES AND ATTENUAl;IONS
0•2IDO
•--~ fi i · I I I\ I
I
<
I
-~.5b
_,I
...
0
I 5
I
I 10
I
'
I
I
w
I
I 15
I
I
I
I 20
I
Tlmo Clal
O•U
~
I Or----__} -OJ ·.·
.:.u
I
I
0
I
·~i'
!,,1
e',~i:.,"f'•·•·c~-
l
I"
•-'-'
Tlm!l""'
:.
·~. · • ·· in of dift'emlt attenuation and' for. tlle>illiM'flli&ilt siltill ~ ·Q IIIIOdllt;
.fla. .f.tl Synthetic Jipak •rec&ded afWi
m ·a · teria · ls.
"" ··?
l'uu4.t 0RDEil OF MAG~1Tt:'DE OF ACCURACY OS THE MEASUllEMENT OF Q BY SPECTilAL RATIOS
-
....
.fQ.'Q (%) Remark
s
3
Q
>
}()00.0
SOOAI
Signal prac- Weak sipal tically unobservable because hillhl - y attenuated
so 10 to 20%
Preferential
operf.tint range
100
;;; SO%
1
>SO%
The slope of the spectral ratio on 'flllicb Q dcpl'llds il Vf6Y low, and itsI determination is : inaccurate
I
very inaccurate ~
.--.
"" '"
'-------,---------------------------------~-·-~--~---
.,
~
"---~
;4
:II!CDIN,...,,_..
.....,'""!!I~llttp_.~~~
IS?
-'-----'
"--'
I Tm. ~
Fie- 4.11 Possible ddtnition of the rise time of-a signal. '---"
·~-
Kjartansson (1979), showed that a similar theoretical equation could be derived for the Constant (2 model (i.e: a quality 'factor independont off~) and for a Dirac pulse, namely: T' Q =C(4.10) T
'---....-"---
where T' = pseudo-period of the wave, C ;:; constant (for a Jivcn source). As a aeneral nlle for BllJ, source, Eq. (4.9) remains empirical. In general, the timeintegrated form of the cquatieo is used : . T ,T ""' To + C Q (4.11)
·---'----'
-----
Blair and Spathis (1982). showed thaf...guantity To is a flUlCtiOft of the source, but cannot be relate4 simply to its rise time,t nil: ~o-constant C is a sliptly variable function of Q (f~ Q > 20) and alsO depends ·on,_. source (Blair and Spatbis, 1982). Stewart, 1984, theoretically calculated Eq. (4.11) for a llarae number of artificial sources (window, Dirac, and intermediate sources). and showed that. the constant C could vary from 2 for a Heaviside funcmon to 0.~ for a Dirac delta rui.ction. Hence, assuming a relationship ofthc.type~.11) between rise time and quality factor, an attenuation measurement c:an be obtained from the rise time, if the values of the constantl --r 0 and Care lalowllbya-tlpetimatoaa rd'erelltl sample. Uafortunatcly, the relationship betWeen- me time and- quality -Mfer it· not as simple as Eq. (4.11) implies. Figures 4.13 and 4.14 show a synthetic example (Tarif and Bourbie, 1986). The relationship between r and Q is clearly a one-to-one correspondc~ and approximately ·ill die slope C aM ill the ordinate intetolpl (t 0 ) with linear, but a chaaae dccrcasit'lg Q : T Q < IS t ~ o.G7l + 8.295 Q
oecun
Q > ~s r
"-..-
=o.tS6 + oJs7 QT
158
4
TECHNIQL"ES FOR ME.,Sl"Rf:SG VELOCITIES' AND -' lTENL" ATIONS
f =i 0.5 .1
j
l
~
c)l
I
'rd
0
I
l
5
1
----.,
10
F~IMHzl
TimtU&sl
Fi&- .C.U A theomical reference .,aiJOill used -in FiJ. 4.14. The central frequency is 1.35 MHz. Time domain signal (left I. Frequency domain (right).
2
""'
11.8 ~
1J1.2
T
= 0.114 +0.311 T
0
a:
,
••
T
, ,.-o~+o:~1e 0
"'
.. 0
:r
3 " · ,.,.., tlmlfQ .,., ' '
~-
;s
8
Fit- 41.C Rise time t~t as a f..... of T/IJ. The diffenma rise times computed using the Constant Q~ aD4 the signal of fig. 4.13.
~
In practice-; this makes the applieatioa of th4: method di~uk. It becomes necessary to know the entire cur1·e t = t(Q) to be really able to obtain an absolute value oti Q in all cases. Naturally it is always possible to compare the quality factor Q for two measurements : the greater the rise time, the greater the attenuation. Tarifand Bourbie(l986) point out a possible solution to this problem. They show that the curvet= t(Q) can be simulated point by point using a Constant Q model. and that ~
"'
~~-------------------------------------------
., -"' '
-:;rr--
_D ci_\-..~......:v-....,
...___~
I '--../
'--../
'-.....-
'--.__/
,_/ ''-
'-'
'~
,__ ~'
4
TECH~JQ~FOR ~~~l.ko V£i&!tiES ~ti ATTENtJAilONs c:=:::;;
159
this simulati9n is el~y comparable to ':the true curve, the essential point bcins tha't the initial souni in the simUlation m~t bC as close as possible to the soul'
to
or
desree
_/
--
1.8 1.4
'-.._/
~~
1
11.2
.! a:
1
'--'
.8 ~
.8
'-
...
2
0
'-
'-
""'
..
5
8
·,
~
Comparison between spectral ratio and ri$e ,tbk methods to c;alc:ulate (2. Measurements of Q by the spcc:tra) ratio technique are taken from Bout. . and Zinszner, 198~. ,
'~
3
Fie- 415
'~
'-
1
Trwet*-/Q . .
'---'
I
It is oblervable that both ;mtt~ are more or less .valent. tn fact, the abscissa (value of Q) of the pob•tJ in 4.t'$1S o~taineC;l ~y lilc;aiuriDJ tbupec:tral ratio, whereas the ordinate (t) is aleasured on the reCorded sipaJ. A one-t~ ~pondcnce is obtained between t aDd T /Q which can be approximated by the straiaht line:
r...
t-
0.451
T
+ 0.208 Q
""-
Note that for high values of Q (Q > SO) both methods are equally inaccurate. This is because, at a Jive:n sampliaa rate, the rise time is too coanely sampled to measure slight attenuation. Similarly, the slope of the spectral ratio is very low, and it is therefore diflicult to make an accurate measurement, as we have already noted. For values of Q < SO. the accuraey of both measurements appears to be more or less identical. However for large attenuations, the rise time method, which utilizes only the ftrst '-'
'-' ~'
'--'
'---
,_
160
4
TECHNIQL'ES FOil MEASURING VE!pCITIES AND ATTENUATIONS
quarter period, is much less subject to the pro~ of jnterfering signals that may influence the spectral ratio method.. I~ must be. kept iq ..Und that a quarter period may not be representative of, the frequency content of si,aluUs several periods in leilgth. Hence, the rise time analysis before and after propaption ~)' not accurately measure the signal as a whole. One physically unrealistic example is :sh'en in Fia. 4.16, in which the rise time decreases after propaption throuah an ~tte~uating m~um. As a rule; however, the situation is much more favorable. Let us recall that, as for the spectral ratio technique, aeometric diffraction (irregular wave fronts ~ue tp the fQ1ite size of transducers and samples) remains a concemfor rise time measurements (Tarit 1986). Finally, in the fi~ld, it appears that the application of the rise tune ptcthOd. yields better results than spectral ratios (Arditty et al., 198.2). ·
---,
------
'
I
·~
,, .cJH l
.----.
~
j·:·
A 1\ vo
I
>:,
"
~
~
''-',-.i_{+...
,,,.
,_
"
Fia- (.16 Schematic example of a sipa1 before and after propagation in an
at-.atiaa
"
.
~
. ---......_,
-._
·~
·""'\
'
-...;p-
..
-~~
'J
'-.._/
4
ft!alf·--~~
l61
4.2 MEASUilEMENTS VSJNG;\1BitATING SYSI'BMS. (STANDING'W·AVES) '-
'-"
1'hese meth()ds have- OR1y becD used ia the laboratory.
4.2.1 ·~
·~
-··
'--'
'
:
Diffacaldes
As for propaption methods, we shaH fint dalya the dift'tculties of application of these measurements from the aeneral standpoint. The main handicap of these methods results from the fact that, by their very clefmition. they involve standing waves. whereas the only waves anaJtaed in the ftetd ..-........ •aws. It may prove dift'teuk te apply the results obtained for standing waves to traveling waves. espcfcially for attenuation measurements. Nevmhllea,tlte J*Yious.a.pter lhowedaat it •asf!OIIible torelaw two. t)!pCS of meatOfelllent eM a if &be attenua&ioll was aot too great. In practice, many ..iftt~M'fem~Ge" ll&tltl illaybee.,citidby ,_...... •vet. For-example. in a poNMtatwated • ..._....._. Wbta1i1ataayaiw rile tor..-a,.W~DCDtsofthe pen~ fluid ilt reletion to dae skeletal lluae or liot typo. aovemeats (W!Ute, 1983) w:~ influence the intrinsiC' at11eoalioa.••••Htt nt·(IIIM Section-4.2.3.2). However, it is customary to operate at frequencia of about 1 to 10 kHz, which happen to be near the frequency of acoustic louinJ. It is .V.~ to operate at much lower frequencies near SO liz which will be of interest in . . . . stta4ies.
t._
bid•••ioul• ...,_,__.rod
·~
·--· "----'
'--
"-·~'-....-
.i"'
4.2.2,. ~ ....~•.. Two methods of measurement are discuAed, one \ISing forced vibl'atiou and the other Ulia1 free vibrations. For free oscillationl, the sample il vibrated at one of its NSOnant frequencies for a short period of time. ~s the ~t of the decay in displacement amplitHe for,~succeuin )I!ICUdo-perioda, as ~ showed in Chapter 3 (Fig. 3.18), aives the value of attenuation u a function of the logarithmic decrement or the quality futor Q: ,}c .. ,
1C .
A2
'IC
·~:·In-•;;
",
where A 1 and A 2 are the signal amplitudes for two consecutive oscillations. Assuming that the displacement amplitude decay isexp (-Ill), which is reasonable for attenuations that are not larp (see Section 3.$.3), one call also write:
Q-::,·: -: '-....-
...__,.
~
'-
(4.12)
J
(4.13)
162
T~tQUES
4
FOR MEASURING YELOCm£5 AHD ATTE!!Ilh\TIONS
As we have stressed above, the use of free oscillations to measure.attenuation is only feasible if the sample. Is oxcdled at o~ of its~. freq~ In general, if the excitation frequency is dift'erecat from. ~. . ., :the ~ance frequencies, the signal amplitude is too small to obtain a reliable measurement. Vibration at the resonant frequency allows for a measurement of velocity [see Eq. (4.14)]. The second experimental technique involves fomcd O$CilatiQOs. TIM tat sample is subjected to a continuous sinusoidal force of predetermined frequency. The sample is in the form of a rod with a diameter much smaller (if possible) than its length. The frequencies at which the bar resonates depends on the elastic properties of the test material and on the length L of the rod. The relationship between these parameters is pven by the equation: '
,;.,
'·'
2L/
Y=Af-....;.;...... . n
.-
(
-
(4.14)
. (.
-~)
- IP\
ror exteasionat modes v. - v. - ..;; · and me.r ~ _v- ~ -, 'VI'f •
-
In this equation,
•
,·
J.,.
'
•
•
I
<
f is the ~-- f'reqacnc.y, A is die -viwatioft, waveliRt'h. E the
Y~smod'ulus,ptheshearlbOd..,,.aa,lhttbe~Theterm~·exteaeional
mode" means that the rod is subjected to a longitudinal extension/compression force. The term .,.shear JBOde" mea~~~ that the rocl·it subjec&ed te a ·shear Ioree. For the ftfturahnode-(the 1'0d issetJ.jeded m •lMDdi.q.foroe). the ~-vo no lORBer equidilcMt.l'hc-tnPII8t. . :wlbcity is tllM of..lbe extouional w~ aJ14l,.&he frequeneyfvelecity relationship is .Wen t., SdniiDer « al. (1~.7~: : .
. ,. , ·_; -~<-. -~ Y£
27t1Jf
:;;. ,'·, -{J =-. -"fPFr
1111 .......;2'•
J;,
'<,
~-
::.
-·
(4.15)
where
R, =-= radius of gyration,
f
= resonance frequency,
.
m - a constant depending on the order of the rcsonabc::C ~and, r•ttt~e.:rn:e boundaty condiUt>Qs. C
,. ..,.,.
To obtain the attenuation value with forced oscillations, whatever the mode anafyzed, the frequency spectrum is scanned in or4tr to d~be the resonance peak entirely (see Fig. 3.21), and the value of Q is thea given by one of the equations of Section 3.5, for example; . (t)
Q=~
(4.16)
where co is the resonance angular frequency and Aw the width of the resonance peak of the displacement at 1/.j2 times the maximum height of this peak.
..--..
..._.-....._
/
....
'
~~·T .. '------' "-...--"
~frbjl~~,--~~
lli3
:
4.2.3 Experimeatal setups The experimental setups for harmonic: cxqtatbl of a rock sample can be divided into two main categories. pendulums and resonant krs.
'"-../
'-
4.2.3.1
Pendulums
The farst category has been used for many years tor rocks (e.g. Pesclnick and Outerbridge. 1961 ). The experiment consists or subjleting a vertically suspended rock bar to a harmonic excitation, with a hip ..,... ofillatia mass, possibly attached to the lower end to increase the resonance period of the system (the sample{mertia-mass combination) (tee FiJ. 4.17). ~To avoid plal'inB tile 8l1llple uacler tellsion, some experimenters balance the sample(mcrtia-mass ~bination by a co1111ler'WeiJht acting through a pulley by \\'bich alterations of the vibration of the system are avoided. Two vibration modes are possible, the torsional mode and tbe bcndina mode. In the torsional mode, tbe fundamental resonance frequency oftbe ovenllsystcm is in~ly proportional to the square root of the system's moment of inertia:
1-ft "-../
"----'
where I is tbe moment of polar inertia, and~ is a torsioa constant. The moment of inertia and torsion constant are related to tbe cbaracteristicdimcuioas oftbe sample (length and diameter). The veloclty is measured by means of the fi&i.4itY modulus which, for a cylindrical rod clamped atone end and free at the otbcr one ~b Land diameter d, is:
'~
3.
'-
The
r
Sl21L 2
,~
"-'
(4.17)
p=
'~
.
(4.18)
measured~· v··= .jiiiP is the phase ~It the ...... frequency f.
A$ we previously pointed out, the resoftiMI'. .. . , cap be deaeucd bJ incccuina - inertia of the system. This system penDfts'it'Nit• ....... at ~ low frequencies
( < 10- 2 Hz), and is rlbt normally used for frequllltildl. . lhaft tO Hz. Theexprasion
for the resonancefrequencyoftbebendinl . . . . . . . . . . . . .~~oftbe
"'--'
system is aaueb more corpplicatcd and theafeN will 1lOt t.e tiveft. The mtensted reader can refer to the work'ofSehreibcr et al. (1973~ In cOildusion, the use of pendulums is relatively cas)', and ODe oftJae oftlJ dift'aculties
consists ofminimilina all tile encrJY losses other tbaa._.OCI.WIIiiaam the rock. It is also possible to use small salftples and to make meuurements at ~ low frequencies. However, the current trend leans more to the use of the resonant bar technique, which we shall examine in detail below.
'---' '-._,
,
__ -...'---'
'-~
"--"
4.2.3.2
Resonant bar
As we have just emphasized, thiS it aarretldy tbi Molt' widely used experimental setup today. The experiment consists in subjectiq a rock bar,ltept horizontal either by supports or by suspensions (Fig. 4.18), to a harmonic excitation.
164
TECH:-.
fO~. r,I~Sl:RING
4
VELOCITlES AND.-\TT£Nl:A TIOSS
'
I
" R«kiii!IPfe
L•gemoment of inertia
,
•. ?J
Simplified
L
dilaru1
;
(by
sz:ti pc~n
tTtm$111ission and 4ZUIU,UJI4""by
La..nns from
Sf:lsmit
waves: r(lt/imiM,
WJUtc. C 1965.McGraw HilL SY).
----..
0 I I
----..
f .
~/~
~
I I I
fi
1
2~. 4, •• 7
•
R. :•Rockllllllllt A W1 W2 : We9!tund Inertia erm c ::Air- coli '· J :: l'ivoc tnd IIIPIIbire
ROCI
~-c.n. ...............
. SWe~encll*l.....-.
S
M
Soriile clamp
G
.(b)
: Light source
:!Minor
:Scale
(c)
b. and c. Actual diagram (after Peselnick and Outerbridge.
Fia- 4.17
© 1961 AGU).
T~on ~ndulwns.
-~
' '-------,-----------------------------------------·--~------
...
---
•(.' ®
·.· .·
.'
.
.··
.............. ·piii.(IJu..........., It! JMIW -.met UOPwtPP n R OSIIIIW:) ~uq ~dW!S. 'sUO!...WUlJ IAOf.puw 1110!11lf1PIO INJ JO ;no ttp 10.:1 i6J't '1!.:1 all) IUOpWffFISO poo.IOJ 'PfA\ J0 (~"J ...S ~· J Act J*!W!I I! fwul!s a~pu -n) SUO!PJIIPSO aq 11)J4l ~ lflt{f110 -u -,a o-.ocJc1o a.p Ol ~Uoq )(1!1' p:llS lfiWS V .
'*" aq
*'· _,
AI~
5! UQPWl~Xa 3p't~ 'q?Ol J0.:1 "UOJtlnl-.lflltls.(JM qp lfW!JaPRU
,JO sprpw JOJ ~'lSD ~(U.Q AIJBJaUaJ ~· SlaOJn,ps ~ JAOf Na~&~Jnq ~ 3fl8l'OJJ»P JO (S~6J 'lUVH,O ~~Ul '.\quqn0J. ~~l.a!'i .(q palp~ aq 8'1:) IUOfl11.J'l!A ~'1.1
Q w:;;..===~-;~~-=----_-7,1 Ii
~
1~
---
166
TECHNIQUES FOR MEAS\HUNG. VEL<)CffiES AND ATTESUATIONS
4
As we have shown, the fundamental resonance frequency is given by:·
v
J=2L
(4.19)
where V = wave phase \'elocity.
L
= bar length.
.---.,
For a given sample, it is possible to alter the resonance frequen.:y by changing the length ofthe bar. As we have s,bown. three~ ~n be excited: the lolljitudinal mode (the rod is subjected to a t.ensioa}c:olnprcssion ~~.~ torsio~;tatmoc,te ud the bendina mode. The loagitudinaland headina modes cmable the tnC!MUrement of Y ouaa's modultll E thrOUJh velocity"£=
fp.an,dthe~rrespoudin,a'l\la"lityfactor.~(seeSection 3.6~ The torsion
mode enables the measuremat of the shear modulus (p) lhroup velocity Vs, and quality factor Qs. The values v,. and Q,. are calculated from the measured vatue1 using the equations in SectiQIJ 3.6. A. aumbor 9( "'""'ectiOJIJ must be applied to· the rcsoaaace frequency values to obtain absolute volecity ~t$. By usia& the uniclimcasional wave equation, we have ignored the inertial effects in transverse directions, and we must correct for lateral extension. Strictly speaking, it is necessary to analyze the bar in three dimensions. These calculations show th•t a ratio of wavelength to bar diameter determines the mapitude of the eo~TCCtion to be made to a.e·frequency and modulus values. Equations and charts oftbese corrections can be foun4 Uitpinner and Tefft ( 1961) and in Schreiber er ol. ( 1973). A .. slenderness" ratio of 10 (length diameter) issuflkient for these corrections to be negUgible for harmonicS or a -..atively to~.- order. The samples can also be jacketed for tests under ,pressuwe aMt "M a $"'en saturatiod, a jacketing which requires further e«>.~ti~~ (Ga~r: ~! oi••J 964 and Winkkr. 1979). In the laboratory, it is difl'acult to analyze samples ~ tbiJJJ nt .lo~ and consequently to 10 down to frequencies lower than SOO Hz. Trttmq~JJ97'1) l'CCOIIUnended the use of the addition of end masses &o efl'cdively lower tbe frequency by chaftliQg the momeet of iaedia and only slightly alter the fiaidity of the S)'Jtem. This apin requires specihc corrcc:tioas (Birch and Bancroft, 1938). To study variation :::=nifal PfOperties as a function ofexcitation frequency, it is necessary to analya b . . of a hilh order. However, these harmonics interfere with each other and tbe inftuence o( one harinonic on its neighbors can be observed to increase as its order rises and (2 decreascis. He~. the resonance peak, which was previously symmetrical, becomes· asymnietrlcal (FiJ, 4.2C)t. It may be observed that the effect is m~able (as)'Dlltl~ri4:&1 peak) iftbc quality factor is less than or equal to 10. Other corrections must be m•de when working with rock samples To begin with, differential movements may exist between the;fluid and solid in a saturated sample (see Biot's theory, Chapter 2). Ia a resonan,t bar, f&teral exteasion processes are capable of generating these movements (W..ite, lft3). f.ig.are 4.21 illustrates a test to identify this phenomenon. The attenuation and veloctlyirl a homogeneous bar of Vosges sandstone of porosity 22% and permeability close to 100 mD were measured accurately as a function of saturation (solid curve). This bar was then cut lonjatUdinally into four plates. These plates were then bonded together and a very thin aluminum foil inserted between them to
'
'
~
.-~
~.
_,
.---.,
'"'' ·~
OOl ,
•
...
~
•
..... 01.
. .. t"L
••
·-t-..............,...
I'L
................ 00&
•
01
01.
"'
I
I
•
l I
01
OS 0 OOL OOO &
•
OSL ~
JO
-.n
,.,.l}lallatll:) lOJ~ ~qlftu tttt .toJ ·.-.r OOUtiUOIU , . ac~etts _... 'lt4 lZMt ..,.....,~
~lllml JMIOI*
c
P:~::r:~:..,.....r~....,..~~...,....,.;.::;:.-.,.
L91
:·
168
TECHNIQUES FOR MEA.WJJNO Vf&LOCITlaAND ATTENUATIONS
4
guarantee total impermeability from one plate to another. The attenuation and velocity were then measured on this foliated bat (dotted curve). The lensth and diameter of both bars (homogeneous and foliated) were the same. It may he observed that the attenuation at very high saturations(> 85%) has~ by 35°/o, whereas for saturations less than 80%, the decrease is only 17%. Certain problems were added to the measurement, such as rigidiftcation (and hence an increase in velocity) of the bar due to~the introduction of the . aluminum foil Nevertheless, this result appears to ascribe some importance to the Biot process in this type of experiment, with high water saturation (i.e. high relative permeability). One must also note that a rock is not a truly homopneous ()bject, and the velocity and attenuation measurCd using the ~~91= peak may not necesSarily be indicative of the average velocity and attenuation Of tllC'sample. Figure 4.22 shows the manrement ·taba 011 • · •sample oonsistiq of a piece of aluminum and a piece of plexipas oftbetame''fenath bonded tOfetber. The velocity and attenuation measured on the ftrst harmonic are very close to the velocity and attenuation of plexiglas alone. It is therefore essential to have samplef·as homogeneous as possible, a criterion of homogeneity being tlae regularity of the frequency distance bet~een resonance peaks. (
I
3045
I 10.5 ~
R
12040
;I
I Jl o~.L ..., 3GIIIt
I I ~ I!
c.·:.
1020
- .. ·':A
, _ _ (Hal
,.., ,..,
181100
Fig. 4.ll Experimental recording of resonance peaks vs. frequency for a twoelement bar (aluminumfplexipas).
Another problem related to measurements on rock samples isthe·non-linearity of the Qstrain relationship. If ~ deformation caused by excitation is excessive, the Q-strain relationship becomes non-liflear and ~g: iatriuically different from fteld studies far from the source (see next_ Chapter)..Fisure 4.~3 sh()WS one example of the change in a resonance peak at high amplitude of deformation. 'Qais ch8Jlle shows the dependence of velocity and quality factor on t)ac."tr,ain &JQPlitude (Fig. 4.24). The resonance bar method, as well as the pendulum method (see paragr. 4.2.3.1~ is reliable and matively easy to it'rlJ'le.-t. Aceuraoy oliM meaatare111ientd ~ iS between 5 and 10% for Q leSs tllan 1~. P'orlmver •uenuations(fl > tOO;forft.amPle), the accuracy declines. In the extreme case, for metal~ (or Which Q is nearly infinite ( > 10,000), the
......,
.........
........ lOU 1! pmpita 'Jwq laftOIOJ ~l pn '=trn<*U • ~ Ol ~!S$0dd. Apl\lJ!A~i! l!'((' > ()) ......WIH• .tfltl!q 1! 1lltlp1ll MpJ! •uA~ •suo!ltm~l· n II""' n la!l~ so:~·ot > JOj 'At {lftOqV pta ot < ~O.J tnoq•
D .;.,·o
Z'
: lUOJP3D 1!~ S!'P UJ top!OOJM ~~'(ON& ~(I'A A1pop\ MpJO hwlft038 ~tt.L
-.sGf tm)!lftd1 llqlO
·
. ptJW UO!}V!PWJ JO S~ ~ Ol Ullm)WA »pUll ~npuoo ~
ptnoqs lDW!JadX:J
•(6L61
'nwu..A\ nt.JW) apt\)!fdunt U!'J.QI ... IJO!I.IOI U! UO!fii'Gallll . . .
t
I
{ZHt kluenbai:J
~
..----.,
~
i.
'
'""" '"""'-
-'
--.
I
I
..
l
'"""
"
,..-.., -~'""'
170
TECHNIQUES
FO~
MEASURING VELOCITIES AND ATTENUATIONS
4
""'
4.3 MlTHODS USING STRESS/STRAIN CURVES These methods ~e qtJite silnilar to lhe foregoing. since the types of excitation are the same. As we have already stated above(see Fig. 3.3), the ener&flosS per oscillation cycle can be measured directly from the hysteresis of the stress/strain carves for different loading and unloading cycles far from the reSonance of the system (Gordon and Davis, 1968. McKavanagh and Stacey, 1974). The U$C ofthis method by these authors has been limited to torsion experiments and uniaxial stresses. The major diffu:ulty is that the adaptation of this method to small strains requires very high measu~t aa:Uracy. To obtain a reliable numerical value, it is necessary to achieve excollent repeatability of the excitation stress function. The value of the quality factor Q is then pven by the equation: 4nV.,•.
Q == L1W
"' -, ~.
~-
(4.20)
where V.., is the average energy stored during a loading cycle and L1 W is the energy dissipated per loading cycle. Spem:er ( 1981) (Fig. 4.25) directly measured ·the phase difference between stress and strain at even higher frequencies and lower strains~ For the phase difference measurement, the quality"fact()r is given by: I Q•tanqJ
"' ',
"' .......,
(4.21)
The velocity is obtained by measurilll the elastif; modulus relating stress and strain. The ratio between stress all4l stnllti lives tlie uttplitude of Young's modulus (dilatation waves)
I
from which the velocity is determined (Spencet, 1981):
I
V=!J
with
lui = lEI tel
(4.22)
"'
i
I ,, 1,
In this equation, p is the tkosity, u the st~ and e the strain. This velocity Y is, slisll_.Y lower than the phase velocity:
f
21EI 2
I
""'
i
V.== ,
~[lEI+
£a]
[see Eqs. (3.31) and (3.33a)]
(4.23)
Here also, an accurate representation of the sinusoidal excitation funetion is aecessary. These rnethods are relatiVely diff'JC'illt to implement, ·given the accuracy required for the small strains involved (s <' to-•). lltetrequency range for seismic applicatiOn spans from a few Ht to a few dozen ·Hz(and eVeri a few hundred Hz for the Spencer apparatus). The accuracy of measurements reported by Spencer is le5s than s•;o for values of Q'of about 20. Accuracies of 1 to 2% can be expected for velocity measurements though error increases in the case of rigid materials. In this type of experiment, it is essential to operate far from the resonance frequency.
I
'-
_!
' ----..
~,,
-, """""''"' --.,
'---'-,_~
4
TECHNIQUES FOR MEASURING VELOCI'f1t!S ANn ATtENUATIONS
171
"-
.,
SECTION THROUGH THE LABORATORY APPARATUS A.ftoclc$1111111e B.,__......._,
-,_~
C. m.lace:ment ltlnlduclr
D. Elaltromtldl...-. Jbaker E. Shaker uble F. Tllble extender
G. Tlble -.-nsfon H. S1llel billet . I. 8aetam _ . .
J.
T.. tndplate .
K. Cleinping ri. . L. Filllurefortlla~
'-~
........_..tubbar,...
M. Flelcllle
'-
~....,_,
\..../
F ... 4.15 Diaaram Of a pendulum t"or measumnents Of phase differ· ence between strefi .aed strain (ifter Spencer. C 1981 AGU).
'-..._/
'~
-~
~
'---·
4.4·' CONCLUSIONS
~
'-.../
~
"--' ~
'--
"~
____.
The many ttclmiques for measuring attenuatioa are dift'acult to ~pltmcnt, both in the fteld and jn tile laboratory, and the accuracy or.-.ul11ment is atncrallY not very high. Moreover, Che freqt;ICilcy range uied in these methods varies, as well as the type of wave involved (ltalldina or travelins waves). Velocities, however, can be measured quite accurately (a f%1 Neyerthelcss,.tlteJypc of V41locity IMUure4 (phase velocity. group velocity) -.aries from one technique to another, and the analysis of velocity dispersion is only feasible for atandina-wave experiments in which the ·phase velocity is measured directly. For propaption experimejts, any accurate mcasvmnent of velocity require$ its own experimentat setlp, since systelnS for the sim\lltaneous measurement of velocity and attenuation arc not suffiCiently acanate for both parameters. The. main results are. summariied in Table 4.2. ·
"-·
"-
'-· ~
...._)
r TABLE
4.2
MeTHODS FoR MP.ASURtNu veuJCtTII'.S AND ATTI!NUATtoNs. SuM~ARY
Field of application.
Type or method
Method
r.eid
Laboratory
Standinj wavel ·. Vibration ..
Ease or implementution
Pendulum
LF, a rew Hz
No
•esonant bar
a''tHz • 10kHz ..
No
DkHz to S MHz
No('l)
~::.
..
Puhc...M:1.0
Fairly easy
thue.wcity
Fairly easy
Pbue~y
Easy
Oill'erent from follow- Rcnection amplitude ratio or spectral ratio ing velocities • phase
\
~~
...
...,,<
... ~
-,,,
Tfll vCiina wavea
&rOUP .•.,.,
Trasll!li111illut:
S)(l kHz to
5 MHz SeismM:at
well-to- Very easy well In wells tt· a"h kHz
"
'··
..
tT-£
.. dift'elilnoe
Di~
cycle
A rew Hz to a few No dozen Hz
l"luhc: ·
) ' ) )
)
)
m•umncnt Up to 400Hz ofphuc
)
---
)
OlciJI•ionl I • forCed •. •.. • free
"'
prttpqa·
Dilerent &em follow- Spectral ratio. Rise ingvcl~
)
No
'
..
time
.phue
.poup:' >"-.J
·.:~11oft
-Stl'fl!l/ftrain
Oscillations • foR:ed • free
lion
....
"Pr~tien
Measuring tcchftiquc
Measured velocity
propaga·
'
Dift'R:Uk
- phase velocity
Di5sipated 'energy
Difftcult
- phase velocity
Phase difference t1 -I:
)'
i J
)
(
(
(
(
I
(
(
(
(
(
---------------- -----
{
(
(
(
(
(continued)
Reliability
Stras/strain Plwe difl'erence
Suwr.tARY
Reliability and acx:uracy oaQ
and accuraq
Method
oilY
Corrections
Remarks
Yes 0.5% ifQ > 20 I %ifQ<20
Yes
Yes 0.5% ifQ> 20 l%ifQ<20 1 ifQ< 3
Yes Yes but tabulated ... 5%ifQ < IOOanciQ > 3
PuJie.eeho
2 .. 3% depeadiq 08 Q. ReliabiB if Q aot too low. ltelatiw mees•rementJ more HOCUratc
Coupliaa problema''' imply uareliatlility. Accuracy > 10 %. Relati¥e ntea!Mtremenl5 more ~KX-'Ufllte
Many and empirical = difFICUlt
With buffers, possibility of study under pressure and better reliability
Traft~~millsion
I to 2 ~. depeftdina on Q. ltdiable if Q > 10
Unreliable 111 > 1o-1•. Accuracy depends on Q
Many and empirical = dill'lcull
In the f~eld, ri5C time appc:11rs to he better
,,_,.cycle
YCII lto 2
s to 10 •;.
YCII
Sliafll
YCII
Sliaht if accurate
Standing wave~~ Vibration lt.esollant bar
Travelifta waves Propaption
(((\
~
Menioos FOtt MEASURING VELOCmES AND ATTl!NUATJONs.
Pendulum
((,
--------
TABLE 4.2
Type of method
(
(
~.
Direct meuurement Yes lto2% ofplwe
II) The aa:ui'IICy values
Yes but tabulated
Torsion. bendina Low frequency measure-
- S •;. iCQ < lOOancl Q > 3
ments
5%forQ~20
ai\lefl for YtltltlltJflio: prOfl"plion MCthoola refer mtJf'C to
¥Blues
if
Torsioa, bending Low frequency measurements
system
very Fur from resona~nce Torsion and uniaxial stras
system
very Far from resonance Extensional s
accurate
ul' rcprodlleibility, sitK:c it ill very dill'oc:ult lu ublain 1he nh•nlull: value nl Q.
~
~
/i,}.:.;.!>"':!.:f. .. y:f~i' {!~I· ~:·!_J·J:jr'[_"f~"'
'---"
5
'-
'--
''---'
WcNe
propagation in porous media
resu~s
and mechanisms
''---
'
'---'
~J·
'-
" ·~
INTROOOCI'ION We have 10 far ill&roduced a number oftlleorctieai(Cbapters 2 ud 3) and experimental (Cbapter 4) toelseaabliaa usto raodelaad •raNK dal lllllduaaioal propcrtios of porous media. We also dlowed iD Chapler 1 that to .,..U of~ JMdia in the absolute IOMC wu relatively ........... A timple conaept·sucrh•-tltat of porosity must be split up ialo . a large number of categories. In this Chapter, we have decided to begin by .,..,..... a mambcr of result$ of acoustic measurements in the laboratory by hiahliahtina the influence of a number of important parameters. Usina this qualitative knowledp, we shall then try to understand the mechanisms responsible for the process~~~.-.-.. deal with 'tile inverse problem, that is to say in situ measurements for which the unknowns are the properties of the propaptiaa media. We shall show how certain lderatuq aauks cu be used and provide some of the most widely employed CRJpirk:al ~~ns. '
j"
"
•
'-
'-'--.-'
'-
"-
----,___. ·___. -"~
'-,...·
'--
5.J RESULts ANn 1\li'.CIU.NISMS . IN THE LABORATORY The earliest syltematic meuurements or variatioM In the acotlltic properties of rocks appeaftld in die 1950(twos. At the·time, muy f1JIIrl1llis,illduditla Wyllie, GfiJory and
Gantner, inYeltiptedthe variations in the wlocitiet
or...... waves in porous media as
a funcrioll of such pltl'lltftetert a porosity, ·aturat.ion aild prasuro. 'f'hey also cleak with the problems raised by the measurement of intrinsic attenuation. At the same time, Nafe and Drake hepn anatp.ina ~tics. in 09IU RQOI' lledimcnts. with a partic;ular focus on the preblem of colllf)IClion. This type or ......... ~ down ia dae 19(i()s. but similar ia~iJations were revived in the early t91Gs·by_ re~ inchtdina Domenico and Nur, whose studies of variatiOns of velOCity as i fUnction of prdSure, saturation and
176
RESULTS AND MECHAI'ISMS
s
interstitial fluid viscosity are well-known. At the present time. many areas draw on the knowledge of velocities and, though more rarely, of attenuations. These include many subjects in civil engineering and geophysics. Concerns and motivations are different in each of these areas: they involve the analysis of thermal mechanisms for geothermal energy, high pressure, high temperature and time measurements for nuclear energy, zero pressure measurements in soil mechanics, and quality control of concrete for construction. Yet the primary area for research on velocities and attenuations remains petroleum geophysics. The considerable growth and the pioneering role played by the petroleum laboratories arc nat rally grounded on the fundamental importance of seistnic reflection surveys for petroleu exploration, and on acoustic logging in wells. This also explains s, and especially sandstones (whose importance as reservoirs is well why sedimentary r known), offered the ost popular area for experimental research. In the following pages, we shall try to avoid estricting ourselves to experimental curves available for sandstones. pointed out that few investigations dealing with carbonate rocks, However, it must clays and shales exis in the literature. One must realize that the results presented concern clearly defmed samp es, and that these~rnalts.can only be extrapolated to other samples with all due precaut ons. We shall fust examine the results obtained by velocity investigations, which will enable us to highlight a number of trellds. We shall thea demonstrate the difftculty of deflfting a siftlle "WJocitY'•parameter •oharactcri:re the bdha,ior of a giWil sample. We shall then go on to ~ts of afteiR'Ja!Wn. The interpretation of theBe resUlts by means of physical -meciNMtsms will follOw. FiftaDy ,..., ,shall take the .tewpoint el the investigator and develop some empirical laws for velocities and attenuatioDI and give their limitations, for various. in~. ·
'""'
-~.
'""'
•.v«
~
S.l.t 5.1.1.1
Velocities_. Jlllllt es •
Rocks in the earth's crust are srib,;ect't6 high stresses. To understand the processes observed, it is therefore essential to understand the role of pressure on acoustic parameters. To begin with, we shall examine the effects of different types of pressure on velocity, namely confming pressure Pc (pressure to which the sample is subjected), pore pressure Pp (pressure of tJte. flp1.·4. i.,nsia.F the por~~ and the differ~ntial or ••effective pressure" Pefl ( = p • ._ pP)fH. We Wilt~ ~bow the e1fes'Ct of a uniaxial stress.
a. Effective pressure Figures 5.1 and 5.2 aivo a ~"r of c~amplcs of V.Jriati
.
--..
...__
-.
but this increase-dcpends.svb!;~ly ~n·~ type of rock co~oed. PractialUy no (1) The eft'e<:tive Strc:IS is normaRy ~ U the stress to which the $01id skeleton is SUbjected, i.e. the difl'eftllee betWeen the lithostatiC ·streu-...ct tk l)'drostetic prtSsure. This is derivWII from nrza,m.._ law diKUSSed in Scetion 2.2.4. At • rule, tlais ~vc ~trcacaa be trtatca u tile differealial pressure, Wbic:h is~ diffefCJl<:e between confmins pressure. and the. ~re press!'re (see ~ ur and B~ :rice. 1971).
~---
·-"
""
......
- T
I
\.....'-'
...............
-
7
1
I
5
8
v,
-;
l4
::!
js
~
l:
8
Vs
2
Vs
3
0
100 2110 300 EffectiM ,....,. CMhl
408
2a.--....a......---'--......1'--...J 0
(a) 7
Webnlck 'dolomite
i::! '-c~
'-
fs ... >
'--'
4
108 200 300 Efhctivt ..,_.. CMhl 1iI'Oytrlrutt .
v, 8
Vp
J.
--
~ 3
.,..
400
(I)
I
"
...
I
~
I>
I -----'-·
S.t.
4
Sat. llld dry
vs ~,
100 200 EffMtlve ,.._.. fMPal 5
\_
300 (a)
I
!3
Sat. lftd dry
l&
Vp
•
~
'~
......_,,_..,.. .,
I
4
I>
.
.,, .,....:..,__ .:.,
3"-----.,a.----'----..J
2
'4
s.t.ll'id dry
3
0 '-
0
10 20 30 Effectlwe.,....,. CMPtl
'
• 40
<')
vs ___.__
2 0
•
I
100
200
Effective,_,,. {Mhl Fit.~~ J~UCDQ; of poreaape o• ~14ft«tivc pr•ure ~hips (after Nur and Murphy, edited by lrutin and Hsieh, 1981 and Nur, personal correspondence). a. Microcracked rocks (ultrasonic measurements). Ia. Massillon sandstone (resonant bar), Solcnhofen- limestone (ultrasonic measurements). '-
-.___
4 Pierre shtle
Sit.
3
i
~'...
~
.~
!
........ "'
.,.,.
..
",..P Dry
_......-o--.......... - .. _-o Dry
2
~s.t.
Vs
110
100
150
&ftec:tlve pressunt (Mh)
Ch..k 3.5
sw
•=306%
~~~~··-==;-1~
::::;
-
i
10
seo
3 }-,
,"1,<
fu >
$w
s
0
2
10 . 40 20
80
~to
100_ _ _..___ _ _ 1.5 .__ 0
10
~..-
_ _ _;r..;,;.;_ _ _L---.....J
20 30 Effective ,..._,. IMflll
40
Fia. 5.2 Velocity/etTec:tive pressure relationships for Pierre shale (after Tosaya. 1982) and chalk (after Greaory. 1976} (ultrasonic measureinents).
"
i".'.
,...~u~,·
~:.f.i
h'L~.
·~~-{11,
J;·;~ ;<:t..,n;,
h~: :Y~:r~en~
J
._.~
'----'
~
t
·:·
'-
~.
'-
179
increase occurs in Solenhofcn limestone, whereas cbc P wave vcloacy rises from 4 to 6 km/s for dry Westerly granite. 1'he iftereue in I' waw wlocily vs. e«ectiYe 9fC8IUie is much smaller for a saturated sample than tot a dry ....,te.. and tbis applies to all the samples in Fia. 5.1. On the other hand, the shear waw veloCity is vinually unaffected by the presence of water in the medium. Also observable is the existence of a maximum pressure above which the Velocity reriWns constant Howewr in Piem shale (FiJ. 5.2), no plateau can be obSetwd at maximum pressure applied: The samples shown in Fij. 5.1 are 6lirfy weU4cnowtt porous media (see Chapter 1). Granites, dolomite and Bedford limestone are well known for their essentially microctack porosity. Solenhofen limestone and MusiiJon S8ftdttone, however; exhibit a pore apace consistina of pores with aspect ra1irGI dOle to t··.mc~~ we call ~·spherical" porosity, aad cracts are uncommon. The samples in Fit. 5.2 display c:oniJ)Iel pen apaoea, eitber because of the presence of clays or due to the pbysicc:M:hemical sensitivity of the c:alcite making up the chalk. Til~ Fig. S.l shows-that the effect ofeonftaiq ~is direCtly related to the number of cracks (two-dimensional pores whid'r tend to desteuily) existinJ in the sample. Denis et al. (1979) measured the ultrasonic velodties of dry samples at atmospheric pressure and succeeded in calculatiag a fraeturina index (eontinuity illdex). Fiptt 5.3 illustrates the
'-
'-
~A'MI)~
!
'--~
'--
''-,_...-
0.&
'-'
o.•
~
~
"---
~ Yo
~
0.1
~·
0.2 '--'
'--'
.
0.1
1
2
S
•
1
I
Crack~ tX tO~ol · • "--"
F'IJ. 5.3
Influence of microcrack porosity on tbe velocity/effective pressure relationship (compressional wave) (ultrasoaic: measurcmettts) Iafter Nur and Murphy, edited by Brulin and Hsieh, 1981)..·
--'----' ''-.--
Vo = velocity at atmospberic: pressure, V1 == velocity at 1 OPa,
&V- V1
-
V0 •
0, Oak. aJl ~tone; FO,, duqitt; Jl~.·. 9!~.ap; TO, Tro~· sranitc; WD, Webatuclt. dolomtte; WG, Welterly JfU'ife: "'-'• "Stone Jranitt: CO, Casc:o Jl.
,_,....-..../
gnmfte.
'-.-
'-··~ '----'
J
·
180
~LTS.
AND MECHANISMS
5
extent. of fracture porosity by the inc~·ia velocity ,with ~~·The .ordinate in the diagram represents the normalizl0d4i&rcnce bctw,eeothe compressional velocity without pnssure and that uader a prcuurc of 1 .PPa for dry samples: the ,greater the number of crackS; dto more the velocity. v.aric$ with pressure. As already noted, the P~rrc ~pies used in FiJ. 5.l display a different velocity function dllan Fis. 5.1. The~ RlateJu as a fuiKltion ofp~ure, which represents the closure of its .. last" crack for the ~pic ;~nc::emod. is not oluervcd for this rock at 120 MPa. In fact, Jones and Waoa (l981) .obMfvcd this phe1lome.non continuing up to 0.4 GPL It is possible. &Nt tbc .~ in velocity as a function of pressure shows. in addition to thccontinuousdosurcofthc ~a.:ks,.an aJipment of the day crystals in the miDimwa shear streaJtb plueflosaya, 1982)-;aa.Jk. tlae$CCOnd examplein Fia. 5.2. is very c:omplicated ~ the stnacture 8f ~ material ~ ac:cordiq to the pressure applied (creep). We shall .net 10 .into detail for Qjthcr Pierre shale or chalk. assuming that the variation in velocity versus pressure are only important if the strw:ture of the material iavestiptedis aot GMqcct itq:venibly: by the ~L For clays as well as c:balk, tbore is goect reason. to s~ :At~ in the CUfVe of velocity vs. cft"~ive
-.le.
'
furtlaer
pr.essurc. .
We have mown that the increase in velocity with prasure resultS from tbc closure of the cracks, and that this closure is reflected by a greater rigidity of the material under pressure (i.e. an increase in the correspondins elastic modulus~ In fact, it must be remembered that any velocity V can be expressed in the form:
v=
Jo/;
(5.1) ~.
where M is the elastic modulus, aad p the density, and that, consequently, at constant p, an increase in elastic modulus implies in velocity. The behavior of cracks and pores under confmina pressure was modeled by Walsh (1965, 1969) and Wu (1966) for a pore or crack included in a matrix. The equations ~ by K, the bulk modulus. and p., the shear modulus of the rock for the dry sample are- the follo\\illl:
arise
(t + !.)e , _!_ (1 + !.) ,
_!_ == _1 K K1
A
.!_
B
JJ
=
~
(S.2) (5.3)
e
JJJ
where K 1 and p 1 are tbe solid moduli, e the aspect ratio oftbe pore or crack (e = 1 for a sphere, e ~ 1 for a crack), tj) the porosity, and A and B are constants depending on the characteristics of the medium an4 close to 1. For the saturated sample: 1 1 . (5.4) K ~ K; (l + A'tj))
.!_ JJ
~ _!_ P1
(1 + !.),e B'
(5.5)
where .A.' llDd I!" arc consta~ts deP.eqd•rw on K 1 and Jlt ($D4 ft,U.id bulk .modulus for A') and close to l.lt can be seen that, if e = ·t, the eff'ect of the pore on the moduli is negliJible
-.,
'-'-
-
r ·f. ; ~
' !
-- .. ,. 'lti!Siiii$ ~ M£Ct4ANiiMS '
5
f8l
to-
2, the effect on for low porosities (about the order ofmasnitude of • ). However, i(e == the bulk modulus of the dry sample isconsiderable. On the other hand, it is nepgible for a saturater.i sample. For a livcn sample, it i$ necessa11 to integrate the effects of type (5.2) to (S.S) for all the pores and cracks. The more pressure is applied. the more cracks are closed and the less the moduli are altered. EquatiOns (5.:zrn,~s.S,.als9 bclp to understand the di8'ercn
"'
o•ine
0.35.,_,.
;.·.
0
.... lllldllone
A Dry DW....,pp•O • W..,pc•toMPa
20
...,
to'
- ..
10
£ffwtlw ......... ......, '
lnftuenc:e of saturatina fluid_.~ pressure on Poilsn'• nttio of Berea sandstone Iafter Nur and Murphy, edited by Brulin and Hsicll, 1981).
Ftc- 5.4
... hN,....,. . . etlll/itlilll~ The resulti hi Fijs. S:,, S.~ .~ 5. 7 ~w clearly that the important parameter controlling velocity is the. .ivc pressure. tletail, it is observed that, at a pven effective pressure, the higher the pore pressure (i.e. the higher the confaaing pressure), the higher the
In
4.0
4:0
MassiHon undstone
'·'~· ••
~ 3» ~
t
Vp
>
.1::
8
' ....
•
Pc•80MI'I· _,
~· Vs
1~.:,:...__ _
1.5
0
·a
:
• -
•
..
13.0 f
-
~ 2.5
2.oL,.
...
•
v,
u
2.5
"s
2.0
1.5 '--_ _......_ _ __...._
20
40 60 lttectM.,.... ....,
80
20
0
_,.._~-------''-'
..........
40
~
60
60
5.0
3.8
4.5
i
v,
3.4
• .,p...,
4.0
~
13.2
St ,._, SlndttoM
•
Ys
13.0
13.5
•
>
>
2.8
3.0
Vs
2.5
0
2.8
.J.l
L 20
40
60
..,~o
80
20
0
Effective ..,._. MPa
40
60
110
Effective .,.._. MPa
FJa. 5.5 y.locity/efl'ective pressure relaUQRSh,ips for cases of zero pore pressure (p., ... ()) ana '(l()Ditant confaning pressure (p, = 80 MPa) (ultrasonic measurements) (after Jonea; 1983).
,;,
...... ....._ 4
~Pett=Pc
13.5 ~
PeffilinMJita Peff=21 MPe
J
Peft•14r.'IPI . Peff=7 MPa
-
3 Peff=O
.
. 0
20
40
80
Confining pressure (MPal
~;
fitJ 5.6 "••prn•ionat ,..w velocities in
aetea ..-ollC VJ. ~fipjq~te and eltective p-cssure (ultr&sonic measurements) (after Wjftie er lll., 19$8). ·
,_.,-"
r
•
J 4
3.8 . - - - - - - - - - - - - - - ,
...
-a
13.4
,, .. ,
3.5
t
Pp""41 MP•
.-,=o
~
i
m·
--·-~
v,
Drv
""_,.,. t
I.
I
--4
~
1.1L
Pp = 0
p,•1MP•
"
100
• ,,•41 ...
lu
O.IL-.._ _,_____...__ _..__
Pp =41 MP•
• ,,-,1 w.
-a Ju fJA,
"p,=41MP•
50
e
1:. ,.
2.5
12
32
..
, , .. 1 . ..
__..~
21
150
Cclnflnlnt,...,,. ......""
50 7t •• IH.ctlw.....-MP8
l'frl. 5.7 lducace of port .,-JPlHmd ceafiiUoa pressure OJI ~ io Piette .... (ultrascmic inebu~ (aflfl' Tosaya, 198~. ". _ ,·
'~
velocii1(FJ8. 5.S). Y« the d~betweeo Ute vdocity~ ~ut"ewrves for the cases of zero pore pnsaure aDd constant confanina pn:aa.e is very sliaht, and in many cases lies within thcacwracy ol the ox~t. Fi~Urcs $.6_11l4 S.7 show the increase in velocity II a fuilction of codaina 'fhiSUN for dilrereat..- presiUl'CS. The general conclusions are the _ ..... ofelfective pressure. It is also ~ that, at a pven con(apina ~ure. the vclec:it)' is.~ for flow pore pressure ia cc:qpariiOft witla a IU~ .,.,. ~ · ~-~ tbe increase in velOcity wida dKtive prcuure. " "" From a 4I'JIIIitaeiw tiewpeint, ads JK.a.llble dlat t~~e~cliPt iDctcale in velocity correspondiaJ to hiP pore pRSSures with a pven effecti\ie .....re can Mtelated to the correspondiftl increase in confming pressure, which slipdy increases the riJidity of the matrix and hence the velocity of the medium.
......_,.....,_,_.fllaalion 1'bi:
c.
~
. , _ , - . . . . . . . .,_..~
The uriatioasofvelocity as a fuactic.aof"Miexial'ltn:ls_..doscly 4opeadeQt OD two parameten:tlle rau.G'thisltals tofhefaih,....,...ofdle...aplcaadthedirec:tioaof application ol&he atfeU in relldiDD.*the-lWOPI. . . . cliaetiaa 1oftbe waws analpled.ln
fact,dependialoo th&mapliUickof*cllHII~.._ theesistiqcr~ks aredosod (low stress) or new cracks arc created (high stress).
184
5
RESUI..'R A!'fD M6CHANISMS 0
VN 1~
------------------
'
Fig. SJia 11\ftuonce of a uniuialatress on P and S velocities ill a sample of Westerly granite (after Lockner~ al. C 1977 AGU). Left: schematic view Gl' ~xperimJntal setup. Right: normalized velocities as a function of failure strength.
a!WII
i
uf 4.4
25 20 ,,
·i
4.2
tO
-
~
4.8
' sv 3:1F-._
p
SH
3.t
\
30
:::4.015 ~.8 L
o <;.
3.1
-~
L-.a..
o•
..
:::· &
,...•. .,., ;·····-·...................,............... j_
••
~·d
~~ 2.7~:: :?;-;;tee;:
-,,c
·td '.to!
••
to•
F.. UJ .. V~I~ty ~&Jopyill4~ oll.•~Jranitesampk bl a uniaxial sttesr. n.e stm~tsapPHed mtt.ra~·D • ft'. The SH wave ts polarized
perpendicular to the stress direction for any 8. The SY •'&ve is polarilled in . tile pine coaeainiaa•tbe:*- direetion<(aftcr Nur an4 Murphy, edited by 8Nlia aDd IUiola. 1"1).
At low stresses, the cracks perpendicular to the stress are closed preferentially, inducing a clear anisotropy in the sample. In Fig. 5.8a, at stresses lower than half of the failure strength of the Westerly granite investigated, the Wteeity inereases (closure of the crack-s). It then drops sharply at high stressel. . . .mt ef craeks). The velocities are lllClUUrCid in a direction perpendicular to the strea·la As. Ub, byeon.tnst, the wlocities are measured in all the direeticms afid.at IDlJICh ..._. lttelacS (pndudias any aution of cradcs). The anisotropy induced by the uniaxiAl sUms· is dearly shown by the diffe~WJCC in velocity measurements in each direction.
----Stereoscopic setup PAV, qJ= 6%
sw
100%
sw
100%
() 1 Atm.Press : 5 MPa
v
(m/s}
1Cl0qo
5400
I 5435 20
32 500/).m
z 220,
qJ_= 3.8%
(5"
1
Atm.Press. 5 MPa I
(m~sl 4358
1001o
172
14800 84 I
I
Fig. 5.9 SEM photographs of two pore casts of low-porosity Fontainebleau sandstones obtained by the technique described in Chapter I (after Bourbie and Zinszner, 19851.
186
. t-
~
s
RESULTS AND MECHANISMS
..V~ONfflii-Dry
GN£1SS-0ry
..,. ...............
-~· .\:~_ _
~6
f)t 14
(J
'
·~~-·
o i~ ~·tl Unudalt
~
.L•'AtiliaU.
Oil ~iatioft; l L........ '
,
;;'a ,
'
1
,
1
,
tl· · ,
....,..... ~~" .40 .: . •
'
'
,
'.:,'.~~:....:w .,,
1
' .,
Cal
...........
1110
~
er-------~------~----~~
.-,·.~~
Sw•100% 3 Vp4SO
.-v,4fSo
5
~
• _v33
l::':!. ~
i>
v22 = Vn v21
4
L:.'-: •
2
• v,3 =V31
't
2 0I
I
"~
..
'
)!rf.J .. J'il ·.. lioJ
,
....
Aaother~~-.pp&~~~ un~~~tC,.,of~~pvcn in S.9. N-~--
DI•n twoF~iaiD-~~able
po~ a·~·&fti:Jilatk:al chediiltry (99.9%q~;w.r.. ......... by F"
ultrasonic methods at a central frequency of 500 kHz with and without uniaxial stress of SO MPa. The velocities and atteauations behaved differently under the effect of the uniaxial stress, and the velocities ·are quite different for the two samples. Figure 5.9 allows a qualitative explanation. The pore casts (see Chapter l) are also quite different. In addition to a "three-dimensional" porosity, one of the two samples (Z220) shows abundant grain boundaries that act like cracks. At zero stress, these cracks make
~
r'
,,~-
'- f·~·t·
·'.
" "--
atsOl'tt'AMY' r.M.if..\t...MiP'
5
18?
the skeleton compressible, hence the low \·elocity (and high attenuation). The application ·
"-.;!!"
of a uniaxial stnsspartly dotes the cracks (those perpendicular to the stress direction) and the velocity riles substntially. These induced anisotropies arc naturally quite di«erent froiJl the intrinsic velocity anisotropies frequently observed ill rocks. three examples of which are given in Fig. 5.1 0.
'-' '-
5.1.1.2 Velocities aad saturatiells
"--
Natural porous media are always saturated with fluids and the influence of these fluids on acoustic propertia is essential. We will prescnt farst the effect of water/air saturation, and then ao on to the other saturating fluids, cmphasizittg the role of the viscosity of these fluids.
'-
'-
"-'- '-·
~
"--· "-·
I
s.t.ndi011 As already shown (fi&- S, 1 for ~x~ple). velogty measurements arc quite different for ••dry"< 21 or 100% lablrated samples. Fipre 5.11 gives aa eumple of velocity measurements venus pressure for dry, partially. and fully saturated Massillon sandstone. For P waves at partial saturation, the velocity is lower than the .. dry" velocity, which is lower than the velocity at l
L,
5
'-
.
Masillon undrtone
•
'-.._-
4
Dry .
4
...
Fullyut.
Dry
;:!a
''--'
I I I I I
' -~;··n' '._/
''-
"--
0 0
10
20
30
Efflctlve pressure IMPel
•1
l'fl. Ul Veloeity/W*f .......UOD relatioubip ia M...won u.adrltonc 'VI. coalaainc,...... (~ ~·~\at et Ill.. 1910). waves uad. f9f sp~ wr~ t~ onq.eflc:Qt d~ t~ in.trod~o~ of water jnto the dry sample is a density inc,rease lq~DJ··to a~~ in velocity Eq. (5.1)]. By eontrast, when COIIJI),tele sat\aratio,a is ~ me '~. become more' difftcult to compreis,
t'he
t•
(2) Ccmlictirallie Ulbltuib· ~aallle lCim ..dry .. Pfe\-ails amoq the cfilerent tUdlots. 'For sOme. tlae t.n. ... , .ellfllidl·to.....,yeliMditMMIIJ(I.e. ,;m tbe,.....4f......_ifa dae rock.lldlol1lld ia _,...
ao••••bcric
we,..un
..,._IICIC:!ItfCiiq oo~. For~ "Po~ ~,UIIdv VICIIUIL As llaow further on in this Section, the water adsorbed plays a fundamental rote on the velocities (and attenuations). This ambipity or te!'JftinolOI)' may explain certain apparent c:ontradictions in the variOus experimental results.
._.,
-~,
•188
5
RESULTS Al'll) MECHANISMS
4
,.......-...........
1-
~
·-'"'
l
11.5 -1311Hz
~
~
8 3 ~
40
20
0
......,_._ ··:Ia
eo
.............. .,, "''
2~--~~--~----
100
80
.......,.....__ .-:a.
2 ~ ~.,
1.8 '111.8
l~
VE 15911-1187 Hzl
~o-o--o--o-o •
11
tto ......1
1.4
-~
Ju ::>
v5 1315 . eu Hd
•
w-..aw.tionl"l
~
v 5 13115-385 Hzl
~
~
4
Ve 1571 ·847Hz)
L...
•
------
0.8
....~
to
...
~·~
z
40
~eo
ro
80
~
~
',
W.... ldlrlltion I,. I
till .
Fla. !.ll Velocity/water saturatiob relatic.-hip fot different materials (resonant bar). Curve (IJ) after Bourbie and Zinszner, 1985. Curves (a) and (e) after Murphy, 1982.
increasing the velecity..F:or S waves, it ·may be ummed that the liquid bas M effect on velocities and that the eft'eet observed is exd~rvely a density effect; For granite, the experiments ofNur and ~urphy (FiJ~ 5.12) show that the "dry" velocity is lower than the velocity at partial saturatfop, which iS lo\11er than 't~ velocity 100"/o saturation. In this case, the porosity is too low for the density effect t
at
"' ~
'-'~
,-
~ -~
'-- !
s
'119
·~
Nr 111UmiGn (S) 100
50
0
Pc!l35MPa
'-'--
'-'-....·
'-
I·
'---'
10.
I
·'---' '-'-·~
'---'
0
'-
'-'
.............. "', !10
100
··~" behavior of l'lltlocitY .a a Juacqon of wa~r · saturation for c:ontOiidated lediments 8ncl a-~ ~.of 35 MPa · (alter Greaory. 1976). .
Ffa. U3
'-' '--·
''-...-
With rcsptet to puita. ~ty JDa'lllll with ••utur.U.n for satuutions above 20o/.. For all dterects..._., tho.,.,.,wa~;~ it virtQily~t of water satutatioa tlftor the &au. •tNds 2%. ~_Nutilio,n ~aadAoao, ~ ia very porous. a sliabt demucl in velocitil,a iJ observed d¥C to, &be densilJ effect~ JSq. (5.1)]. Ia dba,orous saacilloae.lor r~a& bar~ Umilarbebavioris observed to daat clel4.:ribe4 ·by .Gfesory (1976).1or ~-waves (Fi&- S.l3) :even if the loqitudinal v~ia Fil- S.l2 are•tonsional veloeiQaaad aot P wave vcloQtic$(~ QMptcr4). ltilntativelycuyto,apltliatbtbelaavlorof~ia . . . .iUoa-.saadstoneforwater saturations .t 2 to 100~•. From the qualitative standpoint, in fact, the clastic modulus is iadependent of saturatioD for saturation froua 2 to 95%, 'ililereas the density (p = (1 - fl)p. + t;p1 S. p. is the dtrtsity of the solid, p1 the density of the liquid, S.., water saturation)ibcreases with saturation, leadin.s to a slipt decrease in velocity. At total ' ~aturatioa. dae water ~ tends to . . _ the material (for lonlitudinal wavet).aad thewtocity·risa.:For Sftftlnoluidc«ectooeun, asw U.C altudystated.
«"
......
,
(3) 1k dality of aid• aelfiafHe ill .oaapariloll.with dlole of the hid and solid. '-
~
190
5
&ESVLTS AND MECHANISMS
Gassmann's equation [see Section 2.2.2.2, Eq. (2.77)] serves to quantify this type of process. In the case of total saturation, this equation can obviously be adapted by replacing the fluid bulk modulus by: 1 1 l /(=(1-S,..)K+S ... K (5.6) /It
/1
H
/H) and Silanoi ions
0
..
,
flz
where K 11 , and K 111 are the bulk moduli of the two fluids, and S,.. is the saturation with fluid 2. This presumes that one of the two fluids (fluid 1) is totally included in the second (fluid 2). The use of Eqs. (2.77) and (5.6) jives a behavior similaT to that observed for Massillon sandstone at water saturations of 2 to ·100%. In fact, these equations constitute the zerofrequency approximation of the eqvationsdeveloped by Biot (sec Chapter 2, Eq. (2132)]. For measurements taken at ultrasonic frequencies (sec Fig. 5.13 for example), the application of Eq. (2.132) does not ,yield a very good result (sec Fig. 5.14a~ In Fig. 5.14b, Domenico (1976), uses the Geerstma-Smit .. high frequency" equation [see Section 2.2.5, Eq. (2.133)] in which be varies the tortuosity parameter from 1 to inftnity. The result is not better than itt the case of Fig. 5.14a. Various reasons can be postulated for the disagreement. To begin with, the equation used does not account for permeability effects, and, at a jiven tonuosity, the permeability effect is negli@ible only if f/1. is greater than 5 (Gecrstma, 1961). Possibly even more important is the fact that this equation presumes a por~ity uniformly distributed throughout the sample, and that any gas is totally included in the fluid. In fact, partial saturation of the sample is not uniform (see for example Bourbie and Zinszner,1984). Gas may~ occur in a partly continuous form, in which case the sample is truly three-phase and requires a more sophisticated Biot theory (Brutsaert, 1964). Even so, the non-uniformity makes analysis difftcult. The Geerstma-SIIlit equations are interestJna to describe overilt1 behavior, but are defmitely not intended for systematic use., · · Note that neither Gassmann's nor the Geerstma-Smit equations help to explain the behavior of velocities at low saturation and at frequencies in the neishborhood of 1 kHz (Fig. 5.12). To explain tbe~ble ftrilltion in velOCity at very lew sataration, Oark er at. (1980)invokedehydrati0ft ndstifreftins Of day miaerals that are in~tact with the grains. This explanation is ~. tieeause a velocity variation df'!M same type was observed 1ft porous gl•ss (\tycOt) and in sand (Hardin et al., 1963). materials from which clay mitlenils are afMieiftt. Two mechanisms may explain the behavior of velocities at very low saturation. To 1»e8in with, the surface ora quartz snriD possesses a suffteient negative ~to ioni!e the tltinftlm ofwater covering ittParts, 1984). HelrlCe it appears that the pore surface of all sands and sandstones is hydroxylated. A doUble, charged layer is fOrmed betWeen the wateniht Ole silicate SUffiCe (Fig. 5'. ts and Murphy, 1982b) with. molecules of water
'
6 'si/
'si/
-~
~
.The .
hydropn bonds between the surflce b)!droxyls ...amoleeulcs of water are broken which leads toaclissipatioD.·ofentrl)'(teeSeeticm S.U.lb; Tittl8anll«dll.,19B0,5peaccr.1981), and the free surface energy, namely the work required to produce an increase in free surface area, is decreased by the breakaae of these bonds. The. adsorbed water also exerts a signiftcant effect in reducina this free energy(Murphy, 1982 b). The twofold deereate in free
' ~
' ~
~fi'J.jfj; ~~
5
t9t'
2
ju
~j
'-· s*.. ~-
·.tij
••1.1 , ....
1-
2
I
.~
,,_:,,.
8 ~
I
t
0
20
~;
. ..
JO
100
I
.....
I
I
'
......,...Vp I
4.
0
-.~'
fu~
I
'
',·'
f Coltlpulllll
:
-
.,_"'*
i
i
v,
.I
'
I
waw UQ1r11!on I"I
''
I
181
I
,, '---
-·-
Fie- 5.14 Com.Puison of velocities · - ur...... calatl&ted· Ia wandlt~ ··~ computatioa . . . a........ ..,..., [100 Eq. (2.132)). J.iaht, computation usiils Riot's theory [see Eq. (2.133)]. In this latter case a denotes the tortuosity (after Domenico, 1976). .
~
10
so
·- t
•• s ·- tt'
10
·--~
eo w.... _.....,",
JO
.,
too
tbl .• ~_;
or the material and hence a 16.-er wlocit;'. A 1eeOftd catqOry of fofcft 1My explaitl tOck beltavior at low satutation; aamely ca'pilatyforees. lk-rele ofcapi~Wyfot'Ch itmati'Vely well blown for uDCODSOlidated tedimentl. A schematic aad qualitative examplcjs .provided by a pile of dry sand to which wawis ~ually ..... Tt.-elttllllplc is only a qualitatiw4escription and cetramly cannot be usld as a proof, sitiCC muay parameters are not taken into enerJY(adsor~ 'iltld elc!Ctrochemieal) implies •'JI'e*fcomptiance
'-
'-
.,__._.,.
,,,.
.......
.,._........
PiJ. 5.15 ~~orWUitadtorbedonsramsotiilica(lfter Murphy, 1982b).
·~·
192
R~UL~
5
ANQ MECQANISMS
consideration, such as the adhesive forces between the grains. Initially, the sand is dry and the solid skeleton non-rigid: velocity is therefore.low. As soon as a small amount of water is added, capillary forces act between the grain., the solid skeleton exhibits greater rigidity. By adding water further, these forces disappear and the apparent rigidity (and hence velocity) decreases and then becomes virtually ~t until high saturation. At high saturation, the incompressibility of the flui4 itself plays a role and the velocity again rises. The qualitative variation in velocity in theund pile is hence dosely comparable to the variation observed in laboratory experimonts on rock samples (see for example Fig. 5.12) except for "infmitesimal" saturationsf'very dry.. to .. dry"). In lalldstones, a capillary mechanism of the type d~~ abowe is therefore plausible, as the capillary preesure at low saturation guarantees better cohesion of the material. FiaaUy, it should be noted that ttae ~ ~-~ PJ:OPQIICCI (cbemiRI boPds and capillary forces) are both related to the speCifiC surface atea of the porous medium, and we shaD show that this remark raises a; problem for attenualiott at low saturation. Ia conclusion, for dry rocks (saturation less than 2%); substantial variatiort in velocity is observed, which can be explained by a decrease in the sui:f'a6e energy, resulting from the combination of a variation in capillary pressure and an adsorption mechanism. In all cases, we have seen that the·~ ofvelocitl on saturation,-is v~y Jliabt, both for P waves and for S waves. aa4b ·water saturatiOIIS bonreea >10 ed 90%. Couoq._.tly, velocities are not aood indicators of the quantity ef ... pniiMlDt in .... pons.
aoo
tree
b.
Ty~ ofm~~rati•g fluid
The type of saturating fluid influences the acoustic wave velOcity. if only because of the compressibility of the fluid. An example of this variation in velocity is given in tippee. tbcftuidv~y.(andhcnce Fig. 5.16. ~urthermorc,u is*-t.IJ ~-... its temperature) intl'4Cqa:s tbl: v~ cyeo _,o thafl !be ~ type of aaturatina fluid. This is wby ~~Y}tcuapcf.'ture relat~ are diseuiiCMl in the aext Sef;tion.
in.,.
i
........ ... .--,..
·---
•
......
~._~--._~~~~~--._--~-----.
.,
·.
if~
~................
~
3
-·
~
I
i: 0
,, .. 3 ...
Pp•3MI'•
Yp
o
aoc
0
~.12PC
• tSOOC 0
10
20
,_,-3 ...
'1, '·
30
40
50 0
2S0e
0 -~
~ '~ o tSOOC
10
20
250C 1250C
-, ISOOC 30
40
50 0
10
•
30
40
50
DiffemltW ,...,. IMh)
Fia- 5.16 lnfluenee of temperature on P wave velocitJ -in Venezuelan sands Jor dim:rc~saturJlliD& flWds (ultruonK: ~relDClltJ) (~ Tosayt. et Ill., 1985).
~
;:--
---s
':.·
"
JIIUIIfl AN»~
193
'-
5.1.1.3 Veloddes and te.,enawes The burial of rocks in sedimentary basins implies that tl\eY are not only subject to pressure e&cts, but also to temperature effccts;' Tile 'averap temperature Jl'lldient in the crust is around 1-c per lO m of ICdiments. Weshall tirst discuss the ~effect in the broad sense, and then go on to the proble!Jas associated with the chanp in viscosity, and then ftnaUy deal with problems of phase chanp under the efrect of temperature. '----"
'-"--·._..-
a.
Tempa'tlt,.e
Velocity varies slightly with temperature. For example, the velocity in distilled water increases by 1•1o as the temperature rises from 10 to IOOOC (Kaye and Laby, 1973) and the velocity in quartz decreues by 0.4% as the temperatures varies by IOO"C (Carmicbacl, Vol. Ill, 1984). This shows that temperature exerts a very slight intrinsic eft'ect on velocity. In fact, a number ofextrinsic factors cause the velocity to vary with temperature, and these indude di&renees in compressibility between the clitferen.t matrix dameatl aad banc:e di«erences in pressure, chanps in the viscotity of tlie'tauttatHra fluid, and phase cbangcs.
.......
'---'
~'~
........
I
1... Ju~ 1-
•.o ... '--
'~
'-
'--
0
f1l. 5.17.
'-
'-'-../
'---"
_......
?iF
..........
_.
. . ___.:
,::po-_..,. ' ~p '• ~-~-!_ Wet
1
1Clo
-
--- '
••
-
I
•
c:.tlnlnt......,. CW.)
.
'
!
I
400
1100
Compressio. Col • aaJ ~tics vs. temperature in Water.ly puitc.
Compariloa hetwccD dry aDd satwated cues for different confaaiaa pnuurea (ultatoDic _,utemeota) (after Nur, .1980).
For lowporosityroeks such as pauite and pbbro (FiS. 5;17) the decrease in velocity with temperature is nearly alwaytieu than 5% for a tcmpanture inaeasc of about lOO'C (Carmicblcl. Vol. II, 1912). F . , . 5.18 and S.l9 o&r a nuaber of•...,... of velocity variations with temperatUI'f'for Jaip porOiity sandstones. It may be observed that, for dry and water-saturated ..-tones, vciOClity ~with risina temperature. The~ in velocity observed in Fig. S.19d 'at partial~,:._. from aa irrewnible structund chanFin the r9'k, ~ aJ¥lwn VfltY clearly by the hytt~ f.ucpt in this specific the deeroale ia 'ftlleaity is .-er•y fairly slipt. For a tempcratu(e rise of lOO"C, the velocity varies by a maxiuium ota fC'fl perwnt anoon as the confuainapn~~SUre is~ . obecnred in sandltoncs ._, ~ be explained by the dilremn Velocity thermal expansion ~ts betwcea the~ COilstQ,nts of the solid, wbich include clays and .,artz. ·
case.
vanalioris
~-~
3CJCIIDc
0
• 4000C
3.5
_Fit- 5.18 P and S wave ·velocities vs.
Vp
temperature in brine-saturated Boise sandstone (u_ltrasonic measurements) (after TiMUi, 1917). .. . .
J...- cr--o_o_O.....C.....a....,.
....... ~-•.s•a'
13.2& ... ~
I! .2
il
.........~~
r
t
·c
;;:
s.
'---...o.
13
•
~1.75-
,. •
I
Vs
(.)
•
I
0
&
eo
40
20
US·
10 100 l20 140-
Fia. 5.19 Variations in velocities in Berea sandstone vs. temperature. a. Dry Berea sandatone (ultrasonic measurements) (alter Mobarek, 1971). b. c. Fully saturated Berea sandstone (resonant bar) (.rter Jones, 1983). d. Partially water saturated Berea sandstone (resonant bar) (after Jones, 1983).
Temper.cu,. C0c1
.
""'-;,:.
........._
4
)1,. 3.8
i
~
3.6
J3.4 I I
w..rlllVt'atMI
~
.
1.80
··J.,.
~
•
'
it . ·'
'ii ... J
p,=4.5MPa
...............
~'c-10Mh
.:l
0
3.2
20
3
-_..
I
I
4&
(al
~~
rflna.
..... •
10
~
1.2
>
r-..
•
0
220C 0.95 cp
•
&aC'C
jl.~.I:C'
e . IMat 0.34cp 6
_Pp•O.SM,_
5
lei
120
I I
1.8
1.4
100
2.2t'
:I
J
2.0
Ju J
10
T....,.cur.t'cl . Pp"'Pc •O.S·MPa
...... ..,...
20
eo
40
(b)
A.
Ve
10
itooc· o.aq, 15
Confining prwan (M.-.1
20
20 (dl
40
60
10
100
T.....,...,. lOCI
140
~.
"-'~
'---' '---" ~
'-'-../
~
·~
'--
"·~ ·~
---
J
(~'
5
RI!SUL1i
AMI)
"-··-
""'-
~-••'··.
195
ViSCfl6ity ofs,..-, /11M lnvestiptions on velocity variations as a function of saturating fluid viscosity are conducted by varying the fluid viscosity by heatina the sample. Velocity variations in solids as a function of temperature~ ftellilible, and. as the variation in velocity of the fluid as a function oftemperatu,re is small in comparison with that orthe viscosity, raising the temperature is equivalent to observiq velocity variations with viscosity. For glycerol, for example, from - 77 to + 100"C, the viscosity decreases by ten orders of magnitude, while its compressibility only varies by a factor of 3 (Nur and Simmons. 1969~ Figures 5.20 and 5.21 provide a number ofexamples. In Barre granite or very low porosity Bedford limestone, and in the sandstone samples, a rise in velocity is observed with increasins vi$cosity for saturatins fluids sudl as pycerol and oil. In the case of water (Fig. 5.21) the velocity is inclepenc:lcet of temperature to within measurement accuracy, and the viscosity of water is virtually invariable in the interval concerned (in comparison with that of oil). The biJher the saturation of the medium with viscous liquid, the Jllore important the vitcolity Clffoct (Fip. 5.20 and 5.21), 1be slisht variation of voloc.ity in the fluid, as opposed to the high variation of velocity of saturated rock with tanperature (20 to 60%) shows that the etrect observed is clearly that of viscosity. From a tbeoretical staadpoiat, Walsh (1969) de-veloped£qs. (5.4) and (5.5), assDmins that the inclusion .CODSickred was 'riiCDIII. To .do this, he replaced the ricliaity ot the fluid phue by korf,_ wJlere «»is the anplar frequency and 'I the viscosity. He fouad tbat the c«ectivo shear mocJulus I' is liven by:
b.
1 + (CO'f/1Jlo) 2 p • l'o l'oll't + ((J)'f/tl'o)'-
·~
(5.1)
'-../
·~·
'~
"-"
while the bulk modulus K varies sliptly.ln this equation Jlo is the relaxed shear modulus (at c» :1:: 0) equal to the solid modutus, 1'1 the relaxed modulus oftbe composite, and e the aspect ratio of the iaclusion. This shows that an inclusion of viscous ftuid in this model behaves approximately like a standard (or bner) model (tee Ctlaptcl' 3). Qualitatively, increasins the temperature is equivaltnl to loweriq the fluid viscosity, and thus. reduces the rigidity of the material and its velocity. · VISCosity etrects are still noduli)' W1dentood, and recent investiptions. especially those of Nur et al. (1984) may laelp te clarify thee results. Hence, we will leave tbe subject for future discussion. c.
~
'-...-
'--
~~
'-·~·
"---'
'-" '---"
r~MM,cJi.,ge
The ftnal effect on the viscosity variation fl a phue chanae, either by vaporization or by solidiftcation. Timur{1968) observed an increue in velocity in water-saturated rock when the water fr'*(tee Fia- .5.22). Spetzler; and Andenon(l.)eblierved a wide variation in P and S wave velocities when partial meltins bepn. Experiments due to DeVilbiss (1980) and Tosaya et al. (1985) on different types of rock show a signifant increase in velocity at the water-steam transition staae (Fia. 5.22). _ · This effeet lw beea oblervtd to be stroapr in lou porous and less permeable materials (e.g. Jranite), than ill porous·end permeable materials (e.J. -sandstone). It is possible that the specifac behavior ~compact materials result from a non-uniform distribution of the of the pru. cha.nac.ln low ~rmeability media and for relatively pore pressure at the •
Bedford •irnest-
1.0
t.O
Vp
~
o.• -
t
0.7
J!.u
-· J
0.8.
Vs
c
0.8
I
' j
I
I
0
-2
2
4
•
6
Log v~ lpoiMI
(a)
10
2.5 ~---------------.., 1Qil" GivaN~
~0
I
I
G.7
I
P-6
- - - IO"Giyowol
-··- o.v-...
2.25
i
•
I
v,
>
1.5
-··--........... :: -·. a -~
-20
0
20
4
•
II
,._,.,_. Varia~ in velocity vs. ,;scosity in JIJarol amples. L Normalized. P addS wave vel.x:ities in 1ledfenl liDI!IstOtle (ultrasonic :nasun:-
••rated
··-
~~Nur,l~.
Ia. Nonuatized shear modulus in Barre J,ranite and Bedford limestone (ultra-
some measufements) (after Nur. 1980}. c. S wave ·velddty in Boile saadstoae for
1.25
(e)
2
Lot~--·
f')
,,
-80 -60
0
-2
-4
2 1.75
~
~
80
80
100
various glycerol saturations
TemperatureiOcl
~~esoaant
'"-
bar} (after Nur et al., 1984). ~
.
1<.-Riw .... 3.6
' V..,..,....oil ....
p.•JOW..
'·•10.W. u ----.
-
l
3.2
1• 3
•
II
2.8
100Sell
1
.....
f J ~J
-
•.
•
-.....:: •
...101* lr 50% oi
10
1 1CIO flO ~... (fig
-bl'
................. ell 2.4
. . ·-- ,..,.. ..
2 0
50
100
150
T......-tocl
-
2
f.5
.o
lOG
hence
Fig. 5.21 Influence of tempei-ature. (and of saturating ftuid viscosity) M P wave velocity in water- and oil-saturated sandstone (ultrasonic measurements) (after Tosaya et al., 198S).
~
-
,_
,_.-
r _y
y I
y y
'
I
r :
I
.
• 4.1
,''.:/?U.·~
I
t_P
'I
........,
i
~
!'4.8
l
6Temp./
t
.. 4.4
4.2
1
ICI
WATIR
4 ~
34
20 tl
11
I
II
0.
'
~- -1-12-18-20-24-.21
....... t"Q
,.,
1.1
WATER
SftAM
i !u 1
............... '0......, ......
I
&SIIftt ............
. ,M
0.7
Jl'
~u
0
0.2
0.4
...
..... ..,..,.. ,
0.1
0.8
1
1.2
1.4
1.8
.. lbl
f1a. 5.22 ltdhaencc ofpbase'ehaqe ohaturatinl fluid on wloclty (ultritonic measurements).
a. P wave velocity in Berea sandstone (ultrasonic measurements) (after '----' '--->
"---
Timur, 1968).
Ia. P and S wave velocities in StPeter and Berea sandstones and Westerly aranite (ultrasonic measurements) (after DeVilbiss, 1980).
198
RESULTS AND
5
MECHANIS~S
'
high-frequency waves, the assumption that the pore pressure is not uniform is reasonable. Finally, the compressional vei9Cities display virtually no variation at the time of the water/steam transition in samples which are only partially water-saturated (Tosaya eta/., 1985).
5.1.1.4 Velocities ud frequency The foregoing Chapters have shdWn that, in attenuating media, velocity is slightly dependent-on frequency: the media are said to be dispersive. We have compared a large number or results, without going into details about the frequency used. The frequencies used for the measurements do vary widely, and, for laboratory experiments, range approximately from 1 kHz to l MHz.ln Fig. 5.23, we show both ultrasonic and resonant bar measurements in similar samples. Ia this ftgure, it is not possible to compare directly the two compressional velocities v,. . d VE. These; velocities correspond to different boundary conditions but, knowina the Poisson's ratio, are related by Eq. (3.168). For an average Poisson's ratio (0.2-0.3), J'£ is smaller than v,. of 5 to 15%. in low to moderately attenyatin& samples (Sierra White aranitt an4 Fort \ltticMt ~), variations of l 0 to 20% betwcon ultrasonic and resonant bar measurements are oblerved. These variations can be explained partly using a Constan&Q model (see Fig. 5.24 for granite) and partly because ultrasonic: measurements have been made under a small uniaxial stress. For the highly attenuating Massillon sandstone, the difference between ultrasonic and resonant bar measurements is important and much bigger than the one we would get· from a Constant Q model (see Fig. 5.24)even if 10 to 20%4flhis difference is taken into account by the effect of u•xial stress. A Zener model as discussed previously may reconcile ultrasonic and resonaJl! bar ~Cifs. This reconciliation problem is an experimental example of bow diffiCUlt it is to measure a reliable velocity dispersion on a . wide frequency ranae. ... Finally, if a high frequency is used, effects~& to scatterin& are observed (Fig. 5.25, Winkler, 1983). Velocity then decreases with rising frequeac)C. and this is an extrinsic: effect. The higher frequencies .~ p~ and llence travel a longer path. The velocity calculated on the basis of a dircw;t p,tdljs thus artiftcially decreased. By contrast, the lowest frequencies are only slightly scattered, if at all, and their apparent velocity is the true one. In conclusion, velocities vary rather slightly with frequency ifattenuation is not high. We shall show, on the other hand, that variations of attenuation as a function or frequency may be significantly greater.
I I
-
I
I I
Ii
5.1.1.5 Velocities and strains The resulting strain from· a passillg \\'ave f. dependent on ttae amplitude of the wave. Excessive strain may involve sevetalmechanisms liable to alter the elastic properties of the materials analyzed. Figure S.26 shows the effeCt of,Jttain ampti4Kie on velocity. When this amplitude exceeds a valueofapprQximatelylO"'•, tb,e-velocitydecreases, but only slightly. Moreover, dry material is obsen'~ to display a higher threshold than partially saturated material, and the dependence ofvelocity on deformation above the threshold is areater in partially saturated materials. Pressure also has a similar effect. The higher the pressure, the higher the ,threshold. and the lesser this strain amplitude dependence. Mavko (1979)
-,: •._.'
•{ ,(,.,'
~"..Jt~'~~-~ ~
~.~,.dii / ~.(-f
'"' ··:r
~:.·J:J'>
~·);
.1li
,j:
1:·iUJ) \~!Ji:~~·t.:.-hth:-~:Cfi ~-·lt"i(:.·%S'tr:--,r ')1.-i''-'; ·.1
.d
~-
"(Z861 '.(qdlnl'( ~l,P) ~um l111JUl!S lOJ (lqi!J111) (ZH~ J11q lU11U~l pu11 .(q palft~ ~!=KJpA JO sUOI!JlldUIO::J £rs"lt.!l
(U~(111) spoql~ 3!QOS8ll(n (~)
CIOl
08
UO!lt..,... *'M 01 Ot>
.,
t'
0&
I
.......
~
t1U UO!Itlllltl JePM
• •
0 l:
(ZHll 00Z'") SJ\
l"l:
(ZH ttl 1· LLOll SJ\
~
Ot>
£
<
<
c) CZH toiOl: • DLU 31 J\
r 00&
1i'
l
•
I ~I UO!ltJftltl *'M
•
.,
...
tatiiiA tat14MIWIIS,
r
5 ........ "t'IM -.s
•
001
•
........- ...... _._...
-•
IZHliOil"'l SJ\
.
(ZH £"·Uti 311\
-.-(~)
-
l
.
(ZH!il£·11£)$1\
001
.I
1i' 1"£
•
oao-•
o ooo•-(ZHII OK "'I clJ\
O.,c& "dllieJ.
~"'-·
uopi.IIIPI ......
01
001
•
,_ 0.1'-...J.
""-·
IIIIIIIIPWI UIIIUn- WOlf ,
--------------~------~.
tzH1I o!n ..., s"
.. -·
o.u·._l. ~,
........ UO!Ut\UO:I
J
200
1.6
2.9
l~·~
"i j
0
.
2.&
~
0
1.5
SlemWhite.,..,..
~•
Ultr~ ~
27
li !u
J
5
RESL'LTS AND MECtv.NISMS
i
::!!u
UID"IIJRic
f1.2 l j'·' ~-= I
~-.red
pr.tic:bd
~
.....,..
6-~-,. ~o-
2.4 100
eo
eo
1.4
~0
... 0.9
0
btt
/;'-
1.0
20
40
~ ;/C,.,.t
0
10
.... ..,.tlon "')
20
I
'
. . . .8Niitone
30 40 Timefhounl
50
eo
70
--, Fig. 5.14 Influence of frequency on velocities as a function of saturation. Baaed on the low-frequency velocity ....ta (resonant ber), the ultrasonic veiQc:iaies are computed with a Constant Q DiCrdel Note that Sierra White granite appears to display Constant Q behavior contrary to Berea sandstone. In Berea sandstone, water ~turation is not computed, and only drying time is given (after De Vtlbiss. 1910). · ~.
_.......
4 . . . .11011....._
---Dry<,
..,.cMPal 3.1
~
........_........
20
"W'
............., ........
-........
f3.8
20 .............
::!!
10-:::.
13A 5
u·
....._,
__..
..............
' ......=::::.........
-..
I
.....
2.5~
I
..... 10
3 0
500
1000
1tao
2000
21100
Fr.qu~~~Cy CkHzl
Fit- 5.15
lntl~
of frequency on oomprcuional velocity in dry and
sabN'ated Massillon sandstone (after Winkler. C 1983 AGU).
l !
iI i
I I
~
.--.... ~
~
I •(1861 "P.SH pu1t U!JIU& -'q ~u.J» '.(qcilnV( puv Jnl'( JatP!)~,i'Ba.p p.,_ UO(I!ftWW "41
_,1JPU111
. . . . . . 1Pl'1 spwno~lm'BW 6L6l ~ ., 111 l~I~A\ nl.J'B) ff!:top.qno!tUOlD pD ·900 . •nt . . ·.()~UO~Utf~ ~ ~1t.f . • . ·' : • ' JO '
'* .
...
·-
1
;
...............
·-
~.
(
I 0
I
I
-so
'~ftc
i ·: .
'
• '"'
'~~ (e}
.•
....... ·1:
l j
~--------------------------~--~st
L-Ol '
~;
.. OLtrl:
~
ON&
OILZ '• < ''
"'--..vC---.---.-oa----~9~----~ 0
-
c
•
§
- '""
' .WIIfiiN
I
Oltrt"!
ooa
r---~~----r----------------+~
'-·
OLK
202
llESULTS AND MECHANISMS
5
showed theoretically that the 8{ain-to-sfain friction proc;ess was a threshold mechanism involving a non-linear relationship between velocity and straia nu- will be discussed further in Section 5.1.2.5. To conclude, Winkler's calculations (personal communication) showed that the strain amplitude of standard' seismic sources was less .than to-• in far f~eld for any source employed (artiftcial or natur~and hence for the range of uses in geophysics, velocity can be considered to be independent of strain.
II
~
I I I
5.1.1.6 Slmunary We have shown the following properties concernin& variations in velocity as a function of physical paramet~rs: • Velocity increases with coafiDia& ·1'ft'SSU£e aftd differential pressure. The pressure parameter is the ditTerential or efl'ective pressure, which is equal to the difference between the conftning pi'CS$ure and the pore pl§SIIre. Grain contacts and cracks play an essential role in this increase; hence, for crack-free rocks, velocity is virtually independent of pressure. • Velocity depends sliahtly on water saturation for two-phase water/air mixtures and water saturations that are neither too IQW ( < 10%)nor too·biab (> 90%). For dry and 100% saturated samples, velocity increases signiftcantly. • The viscosity of the saturatina fhttd is a parameter that strongly influences velocity. Velocity increases with viscosity. • The role of temperature is slight. An increase of lOO"C in temperature causes a decrease in velocity of only a few per cent. • Phase changes of the fluid in a porous medium cause the velocity to vary substantially. • Velocity depends on frequency ils.JD~liis meditiln; and thus increases slightly with it. • For the low strain usplitudes used in seismic experiment$, velocity is independent of strain.
5.1.2 Attenuations: results We have deliberately decided to present the resultS of attenuation measurements here without any explanation oftbemecbaains involved, other than descriptive comments. In fact, if the behavior observed for velocities could be explained, either using models or qualitatively, it is because the velocities have been,studied and measured for a very long time. By contrast, attenuations are more ditTtcult to measure. They are far more variable and the few mechanisms involved ar~ mostly slibject to consid~ble debate. We shall therefore present the te5ults first, and then~ in Sectipn $.1.3, )Ve 'ShaH try to understand what causes tbis attenuation by interp.-etation _of the data;
I
I
l f
I I I
i'
I j I
I
I I I I
~
'-../
~r---------------------------~
.'----'
Gqnfte
.,;_ ·pv
t'~·
1- ·':
-~r---------------------------~ i~"'t2
..............
'-· 20
~
Q
10
40
.30 \.....-.
20
.., .
Confitllne .....,., CMPel .
:fr·;; f •
w~------~------_.------~--------J 0 tiG wo ttiG 200 Conflllllle ,._. .......
fbi
~
•• ~1
Colorldo oil ......
-~
0
1... :' !::::-
15
~~
tO
I-~
I
I " J..
5 0~~----~------~------~~-----J 50 .. 110 aoo Confinilll~twa)
lei
f'il- 5.f7 Influence of confmins pr~ure ~n quality factor. \ .. Pbue dift'erei!OI ..........~ qat, ,..am~ (after .GOrcloo aa4 .Davis.
C 1968 AGU). .. Berea sandstone and c. Colorado' shale (ultrasoaic ..-uremcnts). A clistinetioll is tnade .lle:NIIiin wa• plltdel alld ~ to- tile baldiJt8 (der .Jota&oa adlolcako e,lM ACM.ij.
---..s
lldfonll.__
80 110
a,.. '---
!10
tOO
200
Olfferenliat .,._.IMI'al.
FiJ. 5.28
Influence of pressure on quality factor for Bedford limestone (ultrasonic measurements). Note the sipiftcant hysteresis, probably due to the creation of microcracks (after Johnston and Tolts<>z. C 1980 AGU).
-~
204
5
RESULTS AND MECHANISMS
5.1.2.1
Attenuations and
pre~&Ures
As in the case of velocities. the essential parameter controlling attenuation is the eft'ective pressure. A decrease in attenuation is normally observed with increasing pressure for P and S waves (Fig. 5.27). One exception is that of the Colorado shales, in which attenuation is virtually independent of pressure. Considerable anisotropy is nevertheless observed. Also, as for velocity, the be!havior of attenuation as a function of pressure depends greatly on the presence of microcracks. Figure 5.28 iUustrates this problem. In this experiment, the Bedford limestone sample undergoes deterioration of its pore network by the creation of new microcracks. Tile hysteresis curve shows that attenuation is greater in the presence of microcracks, and when they are able to open. This very important role played by microcracks is demonstrated in Fig. 5.29, in which the two Fontainebleau 150 120
~ p
~ PAV= 6%
5001&HI
3001-
~
00'1 =0
90
1p ·1110
60
30
l
..
~··
..
.,~1'1-=l 40
60
80
100
W1ter satur1tion (%)
Fit- 5.29
p
120
.
,~.
~=3.8%
cr 1= 0 cr = 5 MPI 1
0
o -o-, -io
.,.J .....~.. -.....
60
0
20
0...~ ~.4>
240
ecr1 = 5MPa
z 220
X
Qi
500kHz
o L--'--~--'--~ 20
40
60
80
100
W1ter satul'ltlon (%I
Influence or microcticts on attenuation in Fontainebleau
sandstone (same samples as m: Fit- S.9Hafter Bourbie ~nd Ziftii'Zilef, 198S).
sandstones of Fi&- 5~.are aulyzocl feN- atteauaU. hy tlw ultrasoaic •thod. As for the velocities, it can be observfd that the lllDChtollewitlaout pain contalctsil unaffected by the uniaxial stress, and that its attenuation is very slight By contrast the sandstone with grain contacts and very high attenuation is very sensitive to the uniaxial stress.
I'
I I
5.1.2.2 Attenuations aad saturaftoat a.
Wflter/.U sat,.atiOII
Figure 5.30 gives two examples of variations in attenuation as a function of saturation. The attenuation of P and S waves is slight fordryrdeka:~efootnote( 2 1 in Section 5.1.1.2a), but high for partially saturated rocks. Attenuation in sat~ted rocks is intermediate, and the quality factor Q5 always appear to be lower dlan Q,.-' 100% saturation. Figure 5.31 shows various types ofvariations in attcnuatil'taas a function of saturation. These measurements were taken around 1 kHz by the resonant bar technique. For granite, it may be observed that S wave attenuation increases steadily with water saturation.
.;,;
\
..
H
··~,;';~;! ,;:;'.;.:;.~h.'j .·~
I
I
\
'--'' '--;,·~tr>'
j..J': ~
HIJ<.ifi'\';: 9"i;·.~< ~ ..r.~,;;e~?'
·-· u..Hion~
MassiUoa..-...(~ bar).~ W'aatJer U!Cl Mur.
L
eo
C
40
1000 Op
'-'
1979 AGU).
30
• 0
tO
20 EHwtM...-.CMh)
lff20
30
!' OL-------~------.L------~~----~ 0
w
20
30
EffwttM.,.-CMh) . ..,
..
70r-------~--------------------~
eo ,..,., sandnDM
50
Dry Q
25
•
76
40
tOO
Confilliftl , _ . . (MPal
30
.. Navajo sandstone (ultrasoll\£
Nllilliolllldnone W..UIIntld
measurements) (after Johastoti ·
aad TobOz.
C 1980 AGU).
so
100 150 Dltt.rantial , _ . .....
·--"------"
Fie- 5.30 pressure.
InRuenc:e of saturation on attenuation as a function of effective
20
206
5
RESULTS AND MECHANISMS
~ r-----------------------------~ • - 18.1"
Fontainebleau sandstone (after JouriMe and Ziasmer. 1985).
L
320
tt:'-
240
180
r
(1.5 ·3kHz)
~1•
~·-·
_•..._ •....__.••./ •J FontliMbleeu~
80
0.
f I
I
I
.,
«»
0
I
• I
t,
eo
e
eo
:
J
100
watar uturation C1' I
100 Sierra Wl'ltte granite
.,
._ Sierra White granite (after Murphy, 1982).
60'
1¥ 40
20
0 0
20
40
80
80
100
·.Watar......,..~)
10
r---------------------~----------, Vyoor porous gl•
e. Vycor porous glass (after Winkler and Nur, 1982).
lfle 0 0
Fit- 5.31 Influence measurements.
of
saturation
20
on
40 60 Water taturation I" I
attenuation.
Resonant
80
100
bar
-------,--------------------------------------------------------------------
----
'-·
5
:!OT
RI!SUL'TS AND M!CHAMISMS
.........._
'-.
20
o....... ....._, 0 c:-.lno ..,._:
10
~
15
I
~ 10 0
I
·-----., 2 3 .--.... volatl..
4
,,.,.. 5.32 loRuence of slight traces of liquid on 1.5
w.w --...c,.ttto..., •........, Cal
b.
attenuation (re10out ball.
a. (after Clark et al. C 1980 AGU). k_ (after Tittmann et al., published by AGU, 19~).
J'ny low stltlll'lllitM: "tlry" rMics (.,.,. ,.,.._ory collllltioiU) Attenuation is very ~itive to.tbe.presence of~aces ofliquh~ ia the porous medium.
Figure 5.32 provides an iHustration ·ofthis. AttenUation rises with the addition of liquid. and increase$ accordina to the type 0( sample. Fiaure 5.32b shows t~ this increase depellds on the types of fluid present in the porous medium, Fluids consisting of polar moleeules inc:rease attenuation more substantially than other fluids.
c.
s.tw.n. tee~
Several techniques are available to obtain a Jiven saturation. For a water lair mixture, for t~xample, the saturated sample can be dried, eeetrifuaed, or depi'CSiurized to introduce the liquid. All these techniques yield a different distribution of the two ftuids in the porous medium (sec Fig. S.33bJ. Attenuation measui'CIDWltS depend heavily on the saturation tecbniques (see FiJ. 5.lla), Ud the attenuation peat shifts or disappears dependin1on the saturatien technique employed~ · 5.1.2.3 Atteauadoas aad temperatures •
a. ,,...,_.e Fipres 5.34 and· S.35 show a decreaic..il} attCJWation with risinl temperature. This decrease is much pnter than that of velt'Jtify,. sift<:e attenuatioo varies by a fKtor of S between 20 aftd 1OOOC (Fig. 5.36}. Tbe eti'Kt Of temperature appears to be less significant at . partial saturations. b.
J'isctnity
Chanaes in viscosity of the saturatina ftuid are pnerally obtained by chanses in temperature. Fiaures 5.37 to 5.39 summarize the results obtained. althoush the dependence of attenuation on viscosity may be hiJhly complicated in detail (Fig. S.~9a);
L
50
100
ll; IliA)
200
50
• 112 250 ~ : Ill
300
., 10
;..,.,
~t··~ #;t . .-...;
.~
~
0
s
..
JiG;
I
J
157
'
10
•
·~'
•
..
~
·. ••
"
.......
.
.. . .• •.•. .
4 p
3
2
1
mm
50
100
150
:
tm
CENTRIFUGE
•
•
(b)
(a)
·.·•......,,,•
'
..
·~
"
• • •. '• A " •••
'\
•·.
__
~
•• ~ti.J., )-· •: •
.
10 58
117
o Ill
J•
liO
.
--
. . . . .#)
..
•'It.
•
. .1
-a,
·-·"' ... ,.
-~:i.•:t
•
~~··':_.,,
·: f
• •
·~-:-·,~..:
• "'
•;·.,;. ;.,-7~·"'
•
.
'#
, ••
::·.... ... ,...•.•...:.,.... ,. . .. .-· ~ : ., ....... ..!' ........,. ••• ~ ., ' ... ,~~ ...~..... ....,. '... .,r. ..,.,. .,·':··..... .. :··· :.--::
~·-
~!'.:.:··· -~
,~
-,- •.•.-.-,,.1·•
DEPRESSURIZATION
5w"'7~
-~
..... ''7... •.., •••. .'..........·'·..... f!-. ,---::;: ..... ::,. ...... ..... .. ' . . ~..,, .'·.., ..: ',' ... ............,. . .
-~ '1-
--;. p · .
I 100
100
200
2110
300
I ~150 :ws
~·. t • ,• •\ if• ~· •A .~
;----
.•~ • •
... .•
90
I
0
GENTRfl=UGE
.........atloft '"'
itt
~---~-~<::"'"
DEP.RESSURI%ATION
Sw=SR
...• ., . •:.:
100
~
200
2liO
300
Zl
..:
lt. Redrawn SEM photographs. Fluid distributions in the pore space are displayed by epoxy injection. Air bubbles are shown in black.
· a. Ultrasonic measurements.
F'IJ. 5.33 Inftuence or saturation technique on atte~uationjsaturation relationship (after Bourbie · and Zinszner, 1984).
v=:
90
CENTRIFUGE
............... "',
70
DRYING
10
~
ll!
J>\i J...
Water-ion"'l
70
DEPRESSURIZATION
I!!
90
100
'\.....~
~
_,
'~ '-
40
50
P11 •0.SMPa
·i 40
Pc"'10MPII
"
•
•
~30
Pp•4.5 . . .
G.IScp 0.52cp 0.34cp O.»cp·
~30
·~.
~•
20
... "-·
10
•
fl·
tO
_,. I *IV
20
I
I
•
Etteetiw ..,_,.tMN '-
f1a. 5.34 Influence
t-
•
I
40
'
,
.,
10
r........,,.eoc,
Pc = Pp•O.S MPa ''-
40
6 E e S
.0
lf!l30
~ Ttmf*ature /
E t
s 'r..,..,._'
20
'-
10
10
100
120
t40
.t......... tocl
'-
FiR- 5.36 lnftuem:e of temperature cydcl on attenuation (nsonant partially saturatlkf 8era samlstcme (after Jones. 1983).
AI
...... ,,.,
confming pressures, p, • 4.5 MPa and p~ - 10 MPa (alter Jones, 1983).
50
.,
I
100
lnftucnc:e of temperature on attenuation (resouant bar) in a water-saturated Berea sandstone (constant pore and
'---
40
I
Fia- 5.35.
of temperature on attenuation (raonant bar) in a water-saturated Berea sandstone. Pore pressure is O.S MPa (after Jones, 1983).
20
I
_,.in
a
I
120
~.
--..
11
0.5
I~ o -4
-2
4
2
0
6
8
s
l
-----
'I
!-~
3
j
p
I 10
• ...;4
2
0
-2
Viscosity
Vilc:osity (Poise ·log scale)
4
8
6
TO
!Poise- Log 1C81el
Fig. 5.37 Influence ofsaturating Ruid visclosity on attenuation (Barre granite), The attenuations have been normalized to the atteauadon of the dry sample (ultrasonic measurements) (after Nur and Simmons, 1969).
$
'
Fi&- 5.38 Influence
of saturating fluid viscosity on attenuation. Bedford limestone (ultrasonic measurements) (after Nur. C 1971 AGU).
100
--.
'
1000
-a 10
-4
-2
•
0 4 2 8 Viscosity !Poise • IOf*:elel
10.
140 ~==30%
120
'
2-4kHz Glycwot 11turation 80%
.-100
ipso Fig. 5.39 Influence of saturating fluid viscosity (temperature) on attenuation in a partially glycerol-saturated Boise sandstone (JCIOnant bar) (after Nur et al., 1984).
'In 40
.,
20
. o' , , -80
,
, -20
1
,
,
20
,
60
,
, 100
Temperlture (°CI
.,
"
'
~-----,--------------------------------~--------------------------~----------·--------
~T I mut~-~,,~~
~
I
J2
I
p
P
-"-v..ltlon
1-
11I' z
2tl
V..-.laoilund
0.11-
1.
1
.............,......
4
~I
...._s
j
13
1.
l
!1 J
Pc•10MPe Temp.110·c
2
..r I . . • ........ ..........
J
ASelnt ......
lrine awreM
I
llu
0.4
J,fl I
I
I 0
0.2
0.4
0.8
0.1
1
1.2
1.4
1.8
Pore..-..IMI'al
ol I' 0
1•1
, __
I
I
I
1
2
3
4
Pen,_. IMPel
5
8
lbl
Fit- !AO lnOuepce of phase ~hange of a saturating fhrid (waterfsteam) (ultrasonic measurements). L SandstOPCS and granite (after De Vilbiss. 1980). ._ Venezuela 011 ~d (after Tosoya et til., l98S).
'--
GO
attaauation
The existellCe of one or more attenuation peaks appears to bedeilonstrated clearly by the results given. These peaks are intrinsic to the liquid itself. c.
Pluu~ clttllw~
As already discussed. temperature can cause phase changes within the porous medium. As in the case of velocities, changes in attenuation as a function. of phase ~F of the
saturating ftuici are sigaif~nL Fipre SAO ofers two examples. A wicle. variation in attenuation is observed at tbe vapor/liquid transition. These wide variations are only observed on samples that are totally saturated with liquid. For partially water-saturated samples, these \-ariations are not observed (Tosaya et al., 1985).
5.1.1.4 AtteautltldM _. ........,. We have shown that velocities depended sliptly on frequency over a limited frequency range. On the other hand, attenuation may vary considerably (see Fip. 5.41 to 5.45). Measurements taken by Gordon and Davis (1968) reveal a slight variation of attenuation in granites and quartzites at frequencies ranging from 0.6 to 50 mHz. Quality factor peaks
J
212
5
ll£St;lTS Al\10 MECIJAI'ISMS
•
Gr.,ite +·GIY<*ine
~
30>oG•.,;N 6
Quartzite
~20
~a
0
•
10
0
~
0
e-
..,.
QD
tb
COil
,_
10
0.1
100
Frequency (mHzl
Fia. 5.41 Frequency depeJKtence of attenuation. Granite and quartzite (after Gordon and Davis.
I
9 Dry a.r.. sandstone
[
8~
7t-
E
.. -- 4~~1 s
50
I
Wet Berea sandstone
~
""' -
6
30
o-: ~ 3J
5
tp· '
:
;J"O%
4
-o., s
~ 2ti'
E 20
3
~·-
lPc' · =0..5.MPa · aS Pc., 10 MPa
2 ..,. •• E S
I 0 . 0
oE I
I
. . I
2
I
I
4
I
I
6
I
I
s
I
10
=t • '
5
0
Pp tip ..• ·a5
J .·
•
tip ··o
.
Gh
0
Frequency (kHzl
Fi&~
1tc • 10 Wa
5.41 Frequency dependence of a~n. sandstone (after Winkler and Nur, 1982).
2
4
8
8
Frequency (kHz)
OfJ; and saturated Berea
"
r ~
125
·.;.
100
.oe
t..
"-') )\
Mlllltlon •ldltone
• E
• s
75
I
1¥ 50
'---
':t:~
25
1--:: 0 O.Q2
1
0.1 0.2
2
20 ~~---'---'---'-=,__ 0
10
f'*IUifiCY {kHz) L
11
tp
Massillon sandstone with different laturatioas.
I
-
-- -·
12
8 4
,.20 r[-~w--.;_100-,.----_,.;...,J b. Porous glass (Vycor) with different saturations.
11
/-...., s
12
//~ F /• E
8
FiJ. 5.43 Frequency dependence of attenuation (after Murphy, 1982).
'-
-~
4
0
I
t
3
~·
Vyoor,poraul .... I
f
5
7
I
'
fr~Q~MMY
I
I
I
11
13
15
CkHzl
'--
50 WinptiiWldiUIM lletlltiUd
40
•-7kHz • -200kHz
10
~20
'--·
10 0 0
10
20 30 40 50 . 5tlllc:IM ...........
eo
10
Y .. 5..u Frequency dependence of attenuation as a function of e8'ectivc pressure in a saturated Wingate sandstone (after Tittlnann et al. C 1981 AGU).
...
214
RESULTS AND
"l
---
70
5
MEC'HA~"ISMS
I •
~
I
50
,.,.,.,.//\! .--~: '""m~• ...•..• ...... ..• ... .0.... T::-· l ......
~]
_~
n
3D
I
20
-
~·
• •, - . .
I
I
I
eo
40
10
I
•
10
100
20
eo
40
w- uturMien I")
10
100
w..., _.,., 1%1
F'JI. 5.45 Frequency dependence of attenuation as a function of saturation. Massillon sandstone (after Murphy, 1982).
125 MaliliO'I undstone
1()0
---Drv _:
75
~ 50
/
.,
..,"
25
,., ,/""
""
0
0
0.5
1 1.5 F,..-y lltHI)
2
2.5
Fit- 5.46 Frequency depeD4caae ofatteauation. Scattering effect, Massillon, sandstone (after Winkler. C 1983 AGU). ----,
r".
....______
-,-
----~-----~~~~-~-~--~~~~-~---~~-~~~~~-----~~---
r '--'-
---
~
~
"----
1 $
'I )1
~ fl I I
....:::::::.
215
arc ob&crvcd for pal1ially and fully aturatcd ~ at t'Nqucndcs in tbe
'---
·~
...a.u AMD MICIIANIDIS
5
neilkborboocl of S to 10 kHz (appiOXialately the frcqucacies employed ill acoustic Vyoer porous Jlass also cliapUn--beha¥icrr ·timiJat to s*dstones, twnely atteauation is inllepeftdent of freqUIIM)' for the a., sample. and biPIY dependent on · frequency at around 10 kHz with increasing sample saturation. Figure 5.45 shows that the position of the attenuation peak as a function of saturation varies with frequency. Figure 5.44 shows that the difference between _ attenuations at two given frequencies louin~).
decreases with rising effective press~ At sufficiently high frequencies (in comparison with grain sizeL scattering effects are observed. as in the case of velocities (Jee Fla. 5.46), and scatterint induces an increase in attenuation with rising frequency at frequendes ......- daan 1 MHz. The behavior observed corresponds fairly closely to a scattering-type mechanism (~ion 3.3.3.5, Sayers, 1981, Winkler, 1983). This mechanism OCICUrs only if the transmitted wavelenith is comparable to the arain siD (Devaney et al.. 1982). Tbc _<:entral frequency of the scattering proc:ess is given by: -L
~
-~
fD
3(M)! P.
=R
1
(5.8)
where
R • p-ain radius, P. = density of the material, M = elastic modulus of the wave, and, in the case of sqdstoneS. the fttquencyf.o is around 3 MHz. 5.1~5
AttelmatioM _. ltnias
As we ba"' seen, ~ocities arc independent o(sttam for the strain range observed iR ppbysics ( < 10- 6 ). Attenuation bebaves silaUrly (Fip. 5.47 to 5.49): "---'---
(a) For deformations lower than a threshold '• of about 10- 6 • Q does not vary with strain. (b) The threshold e. increucs with risiaa prcssu~e and decreases with inacuina water saturation. (c) For e peater than e., attenuation decreases with risiag pressure and increucs with incrcasinJ water saturation. The dependence of attenuation on strain resultia,a from adtreshold.e«eot, namely gftinto-&rain friction (Mavko, 1979), appears to be Jlelligible at seismic strains.
'---
Variations in attenuation as a function of the ,various above can be llimmarized as f'oii
'--
'-._.-'
.'--
Physical paraaeters considered
• AtteauatiOa decreases db risina efteetive pressure, sinc:e the effects of pore pressure and conf'taina pressure are not inetependen( The p n ·oontacts and cracks play an important role, and their presence causes su\Jitu~ attenuation and considerable dependence of attenuation on pressure.
~
100
10
't::=:: :~·: ·. =:. a-~
~ ~::: 1
:
-~
:
:
:
~~
1
f .
FiJ. 5.47 lnflueru:e of strain amptitude on attenuation of different rocks and crystals (lonpmdinal ..CUation at 90 kHz) (lfter Oordoa and Davis. C 1961 AGU).
.
' ~\
"
j
0.1 ~
~
Quartz single crySUf
0.01
to-&
to-7
to-"
,~
...............·,.-s
10-8
..,-3
.----._,
10 • - llndstone
Pc•1MPa
8
~ Fig. 5.48 Influence of straia am• plitude on attenuation. Berea sandstone under different confining and pore pressures (resonant bar) (after Winkler et al. Cl 1979 Macmillan Journals Ltd~
8
0
-o
..... 0.
_ _ _ _ _...,;,.,...,..-:.------• Pc • 5 MPa
4
...,. Strain amp! itl,lde
50
t e M...mon Sw • II S- 70Hz 6 M-illon Sw = 0 S - 1 kHz
40
~o
OttfWII
'~w
•
o" lltolt tem
1~30
Fie. 5.49 Influence of strain am20 10
r.
-·
o~
~~ i
•
~ <,.-oo"''
__________._________~._--------L---------~ -7
-8
·-s
plitude on attenuation for different saturatiQlls (t~nion pendulwn~ Massillon sandstone (after Nur and Murphy, edited by J,ru.lixl and Hsieb. 1981) and Ottawa sand (alter Stott, 1979).
__
~,
-4
Strain .mplltude (tOil tettel ~
~
~
"'
~"-'
rj I:;
5
"
• - - dqlalds coasidorably
I - '
J
'~
·~
'--
...,.,.,~ND'~Nars
217
_,oil
or of P*r liquid· ia 1tie pomus medium atteouatioa Spifaatly, and the atteauetioa of *Y rooks is low. The quality factor Or tcema atwa,s than Qs at l 00% saturation. Attenaatioa also depends on the distribution of air bubbles (or of aonwettins Ruici~ within the porous medium. • Rising temperature detrcascs attenuation signiftcantly. • Attenuation is hilblY sensitive· 'to the viSCQSity of the saturatins fluid. One or more attenuation peaks are observed. • Phase cbanps of the fluid cause substantial cbanses in attenuation. • The quality factor depends stronJly on frequency at frequencies dole to tO kHz, and at· f~ ptator tbu l MHz. 1'hc i,avene of the ~uality factor 4isplaJS a peak arowad lO kHz. • Ia lliaic aptdmeuts, the quality factor is illdependeat of strain amplitude. • Figure S.SO SUlbnlari7.a some l:.l these resUlts for a ••tjpical* sandstone, Massillon
.-ter
sandstone.
-
f
'---'
'--
'--
..... 5;11 . . . .tic reprt~eotation or aaturatiob (after Murp~ty, 1982).
actcmuation vs. frequency and · ·
5.1.3 A.__:_.lllisms We aucaeclod in aeasuria& the attenuation)IIII'WIDOtet' eheracterizinathe loss of energy a ,.._ wave duriJtl ita. propqlltioa du'oulla a &iven materiaL We thus obsened that this pa,....r, jult like velocity, varied: cxmsiderably and depoDded strODIIy Oft lecll ~ tetaplntUe ·udnother physical dfccts. It remains to undentaJHlho9i thiS loss ofeaeraya:cars. How does tMinteraction between the wave and underpM by
·~
218
5
RISVLTS AND MECHANISMS
the porous medium dilsipate enersy tlarough a· Joule dfect? What is alae souroe of the intrinsicaUenuatioa?Which mechaliismsarc N8pOnSiblc under the coaditions that are of interest to teismic exploration? UDJite velocity, attenuation il still.poorty uaderstood. and many mochanisms that we dcscftbc in attempting 10 answer diose qwatioos are merely hypotheses. This. area ofrodc physics is: jn a staa of constant cbansc. We will ftrst examine the role ofintergranularf'rietion. which was tbefocusofimportant work in the 1970s (Walsh, 19(4 Jobnston et al., 1979). We lhall then exlnsidcr ~everal linear mechanisms, distinguishing those that apply to .. dry" rocks (in laboratory conditions) and those applicable to partially or fUlly saturated rocks.
5.1.3.1 lnterpuutar f~ Recent ·experimental measurements (Wiftkler et Jil., 1979, Wialcler and Nur, 1982, Stewart et al., 1983, Murphy, 1982 b, Stoll, 1979) were made at strains ranghlJfrom to-• to to-•, and revealed the iadoponcloace of velomy aiJd ~tioa .............. from strain amplitude, if the ta~r is less than 10- 6 (fjas 5.26 agd 5.47 to 5.49). Moreover, we have mentioned that Mavko (1979) shows that intergranular friction is a .. non-linear mechanism with a threshold effect. This mechanism is triggered only if the deformation amplitude is higher than a given value. Finally, the Winkler and Nur calculations (1982) also served to show that the strain amplitude in seismic experiments was always less than 10- 6 • Hence all these .facts suggest that intergranular friction is not an important phenomenon in the attenuation of seismic waves. Yet Figs. 5.26and 5.47 to 5.49 bave-lbotbown.thejmportancc=ofthe presence of water, even in very small-proportions, in altering both the threshol$1 and the amplitude of variation in mechanical properties as a fUnction of strain. this implies a sort of lubrication achieved by the ftrst layers of water 'between the dift'ercnt ·arams.
5.1.3.2 Att----'••••mws
f ~dry" ... very d&fatly tU!h ..,. rocks mq..-fN
The term ••dry" rock li..a..mat dndear. The fOGies not totally free of moisture. In fact, they ate dry~ under laboratory ~ns (atmospheric pressure and a given humidity) and not rMJP~en f~~~ oven. The degree of water saturation of a ••dry" rock hence UCS:~efil-"and a maximum of 0.5%. Typical experimental results obtained on a Massilli!llandstone with 0 to 10% saturation are given in Fig. 5.51. The very low saturation valUQS are o~~ by ~lo.Wing equilibrium to ~ur between the sample and an atmosphere with a given relative humiditf.. ~ mc:Uurements of velocity (Fig. 5.12) and attenuation were taken by Murphy (1982) using a resonant bar. The measurement frequencies are the resonance frequencies of the bar and are therefore nV/2L(n- orderofresonance,L- bar length, V""' volqe,it,at~Jiequcncy);Tbe variations in measurement frequency shown in Fig. 5.51 (namely 599/997 Hz and 385/653 Hz) result from the wide variation in wlocity from 0 to 1 "• sablratiott (see Fig. S.12). By exlntrast, the resonanc:e,fnlqueQcy for water satimttio111 of I to 10'/o is virtually constant; For saturations of 1.5 to 10%; attenuation is constant, whereas for saturations less thaa 1.5% the variation in attenua:Pon is sipificant. Attenuation rises from 60to 70% for saturation varyinl'from6%to sligbtly-leDthaa 1%. A peat appears at
~"
/"
'-..---'
'---'
~I
-...
5
~
2-19
lOr----------------------------------,
''-'
...
-......--.;.--
10
-,-; ~
''-
50
'-'
E (ti88 • 917 Hz)
lf940 3D
t
'-
20
'toL---_.~--~~--~~--~~--~
0
2
4
•
8
,10
.... llllntioft ~) '-
n.. 5.51
Extensional aad sbeat attetauatioos (1000/Q) iD Musillon sandstont vs. saturation (0 to tG-!.)(after Murphy, 1982). '
'-'
-\_./
1%. 1bis peat pollibly results from the ,,...._..,. depeDdeace of 1000/Q. ·If ibe were .taken at coastaat lro.qucac:,, tllc _pe¥ would probably be absent. However, tbia explaaation requires ~ ~af~PUtioa. We shall now cumioc a munber of pouible mcd1anisms offering aa qualitative explaDatioD of attaluatiop behavior at low ~n. ~
.
DUJoeiJtitM ill parlt Mason n crt (1910) SlltJest that~ oftbe titieatc surface are set in motion by tbeacoastitwaw. This is a rather debatable~ sbicetbe leftldl ofdislocatiobs in quartz and the forc::es rcsistina the dislocatiou involved are such that tbe corresponding resoaancc &equencics are mudl too high to be excited. L
'--
'-
....._,
'---
'-' '-
'--
c,.,._ ,_.
lt. lluMIIP sf lu shown above for velockjes, the porous surface all suds and sandstones is bydroxylated. Tittmann et al. (1980)and SpeDCOf{l981) augcstcd that part of the energy is diuipated by the bnakagc of hydrolia ~.-.eca tbe . awfagc'h)'droxyls and the water moleculc:s, thus subJtantially doc:rouiag the fRJC surface encrJY. This attenuation model is directly depeadent on the specific surface aQa of the porous medium: the greater the specific area, the Val~ of,apeciftc area and attenuation in Vycor (Fig. S.43) and in Massillon sandstone are as~ follows:
or
Ji*•* . ......,
"-----'
~ate&
Attenuation
(tr/Jf
"--~
'"-----' ·~
'-../
~
Vyoor ............................ . Mauilloa~ ........-....... .
a :ZOO •
10
1000/Q ~
2
;; 10 to 25
1,. 220
5
RI!SULD ANI) MECHANISMS
~.
Their incompatibility is easily ob$ervable assuming an attcnuatio~tmecbanism exclusively at the surface. Hence, to explain Che results iJl Fig. 5.51 exclusively by means of the breakage of hydrogen bonds is inad~uate. We shall discuss in the next Section one possibility of explaining the incompatibility observed between speciftc surface area and attenuation.
C11piOtuy forc~s We have shown for velocities that in the case of low saturations capillary forces may play an important role. For attenuation at very low water saturations, part of the energy dissipated may also result from capillary mechanisms. The mec:banism involved is again associated with the specifK: lPlfaa: area of the porous medium, and the Vycor/sandstone problem (see previous paragr.) again appears to arise. In fact, the v•lue of attenuation due to capillary forces depends considerably on the microstructue of the porous medium (surface roughness, aspect ratio and size of capillary tubes) and the specif1c area per se plays a minor role. . Attenuation is hence the combination ofmicrocapillary hysteresis (viscous dissipation on rough surfaces combined with the breakaac of ch~ bonds) and of mechanisms linked to the motion of Ouids (-' saturations pater than 1 to 2%).
c.
~.
--"
~.
~
5.1.3.3 . Atteaaatlea ~~ ._ partially 01' f.,Uy .._.... roc:b Figure 5.52 shows the variation in attenuatiOn at water saturations ranainJ from 10 to 100%, and concerning the same MassiHoti sail'istone'as in F~ S.St.ne variation -in extensional measurement frequency results here from a sharp rise in velodty(see Fig. 5.12) at saturations approaching 100%. The lfteasurement f'requeftcy is hence mually Constant for all saturations in shear mode, and for saturations ranging from 10 to 95% in extensional mode. Fipre 5.52 re~ dlat at~. is coD$idcrabJy affectod bJ water saturation. ExtcDiional atteauation ·readies a . . . .\Uil around ~% Aturation, wbile abear
~
-----..
70
Miitalllon sandstone
•·2a
80
~.
60
!.!2• "
30 20 10 20
40
60
80
100
Wdlttltiii'Mion~l
Fla. 5.Sl Exten•ional and sbtar attenuations (1000/Q) in Maalilloo sandstone va. saturation (10 to 100%) (after Murphy,1982).
"
., "
....,
5
'-../
----,:1
~
'--
_,
:t
ld!IVlti'
"* tid!a~A. . .
121
attenuation shows a maximum at 100% saturatiqn. Moreover, extensional attenuation
(Qi 1 ) is always greater than shear attenuation (Qi 1 ) except at very JUab saturation ( > 97% ~ Also observable (Fia. s;43) is a wide variation of Qi 1 and Qi 1 as a function of frequency, with a peak at around 5. kHz. Finally, as we have shown (Fig. 5.4S), the attenuation peak as a function of saturation varies in position and amplitude with frequency. These properties (peat at 5 kHz, satur~on, depebdence) arc approximately independent of the type of rock analyied, or rather, the dcpendetl4:C on the type of rock is very slight in compari~n with cQnaes in porosity and permeability (see Table 5.1). We shall review tbct.111eehanisms that arc m01fcommonly used today to explain these experimental L
res~tlts.
DUI«tditHH ;, ,....,, As discussed above in Section 5.l.3.2a,: tbis mechanism appears to be inadequate.
i!t
b. llrHk"6e of elwlltkld 6ollb . At high saturations, this medllnism, rolateclto the; spccif'tc surface area of the porous
medium, is undoubtedly nealiJiblc. Moreover, tlle frequency ~ce of attenuation at around 1 kHz appean to be cHfBcUtt to ezplaip by an ioctelse in the number of broken chemic::al bonds, as these bonds raoJtate only at Vt:r'J biab {Nquenc:y. The attenuation peak at hiJh saturation also CiUillOt be ftplaioed ·by this mechanism. The brcaka1e of chemical bonds is therefore unacceptable as an attenuation mechanism at high saturation.
·--
c. Cqilltlry forces . At low saturations, the iPfluaiC:e 'O(t;:apillary:forccs was due to a sort of microcapillary
hysteresis. At high saturatio• tlia mechanism is heace <*taialy less active, and cannot· possibly explain 'the attenuation pt8k observed (see FiB- S.S2). '
d.
Tlln11t0-rela..,
An acoustic wave propaptins in a pven me4ium acts by ~poling a sudden change in stress on the medium coDCel'Did.. Bccause it is a rapid pt~ wave propaption may be considered a process that is controlled macr~y by t~ adiabatic properties of the material. Microsc:opically~the ~ is veryheteroaaneeu.. siDce each grain and each pore with its Ouid respollds adiabatically in accordanco with its own thermomechanical , properties. The rapid strain variations due. to tbc wave front impose temperature variations throup thermomecbanical coupliq. These temperature variations are heteroaeneous due to the heteroaeneity of the microscopic constituants of the porous medium. Macroscopically, tcnaperatW'! equilibrium' is obtained through thermal conduction. A tllmnal relaxaticlll.dtus takes'))lace an4 iavolves a phase shift between macroscopic stress and . strain. &foreovcr, if two phales of the same component (waterjsteam)~prelent,masatiaasks(vapoj!izatio8/conclensation)takeplacefromone phase to the other to auaraaU~Q tlermodyuamic equilibrium. This mechanism was analyzed theoNtjcally by kjart.,.,.... (1919 b); He prtdkted a dcpendctl4:C of attenuation on saturation and on freQuency. The cenn.t i.laxatfon frequeacy is pven by:
''~
'--"'
hl
h-~
'~-
..__._
....
~~
fg
TAIIU:
5.1
FRF.QIJF.N('Y Df.I'P.NUf.N('E 01' ATI'IINIJATION
Per meP()tosit' ' ability (%}
Type
of sample
(md)
Maximum attenuation
....
lnvcstljatcd
rrequency
freq~ncy
"
Bere8 saadsto.ie .......
20
617
2kHz
Berea ....dstone .· .. ' ...
20
tooo
· 2 k,Jfz
Navajo sandstone .....
ll
;;l:O.S kHz
·u
;;. O.$.kHz '().017 kHz ;;.0.5 kHz
Expc;ri.-entat
conditions
'"
0.4 kHz-3 kHz
4>.4
Reference
kHz~3.5-:.tHz
Jones and N ur, Warer · sat IH'ated (1983) and Pc aad p, up to 20 MPa . Jones and Nur, As &oo-ve· (1983)
Nav¥» sandstone . ·• ..• Spor~ liniestone .•..• OWa *!a aruite .....
14
t
3 Hz-500 Jk
.. -. 3Hz-~ HZ 3 Hz-'OOHJ 3Hz• Hz
...
23
737
4to.kUz
JO Hz-tOk~
Perous Blass. (Vycor) ...
28
0.01
7109kHz
1 kHz-f2 'kHz
Barca sandstone .:·......
20
500
3 to 6kHz
1kHz-8kHz
Ma.'lllilkm
~~andatonc
From JOnes (1983}.
Spencer, (1981) Saturated (various Ouidsl Spencer (1981) Low saturation Spencer (1981) Water .satura&ecl Spencer (1981) W•r saturated Murphy (1982) Vau:illble · satUration Murphy (1982) Variabtc saturation \Vater saturated Winklcr~d Nur (1 2) (10 MPa}
!i
I
-
1111
• ) )
)
)
)
~r .._..
-;,
~i
·:, $
~
~
~·
h - pore half-width, D • &bermal dift'usivity of the composite material.
In the case of Massillon sandstone, fr is approximately 10kHz. which is close to the resonance frequency observed (Fig. 5.43). Figure 5.53 Jives the frequency dependence of attenuation obtained by thermo-elastic effect ill relation to that of the standard model previously analyzed It may be observed that the thermo-elastic peak is much broader than the peak of the standard model. This iS normal. because the thermal relaxation process is a diffusive one. The width of the thermal relaxation peak is incompatible with the narrowness of the peaks observed on Massillon sandstone (Fig. 5.43). Figure 5.54 shows tl\e variation of attcauation witb •saturation. The ~um is obtained around 99% saturation. which docs not correspond to tbc experimental results. Figure 5.55 shows that attenuation increases with rising temperature. However some recent data reported by Jones (1983) show the opposite behavior. Finally, the thermal relaxation process fails to explain the caergy ~ obJet •ed in shear. Nevertheless, the foregoing comparisons between theory and experiment must be adjusted appropriately. In fact, the theory is developed for traveling waves, and the experiments employed standing waves (resonant bar method). In the present state of our knowledge. it is diffiCUlt to 8SICSS the scope of this difference. Yet it appears that thermoelastic processes are negligible, at least at temperatures below lOO"C. CluurKe ;, chellfictll ~llilibriltm Acoustic plane waves propagating in a liquid generate local variations in temperature and pressure in the range of 2 • l 0- 3oK and 3 ·• l 0 3 Pa ccspectively (ltalsiag, 1975). Any chemical system in equilibrium subject to sucb· a dist\U'bance will fnld its equilibrium changed in tbe form of a modification of onc.or more of its thermodynatftic constants. This modiftcation is possible by the passaae Qf individual dlllllleculcs from~ state to another, aad hence a dissipation of ~IY· Thi~ dissipation clqJends on the frequency of the disturbance transmitted. At high freq~ the chaJliiS are too rapid for the system to dissipate energy, and at low frequencies the reaction has the time to adapt so as to be nearly reversible. Thus an attenuation peak will be· observed for an intermediate frequency. The type of chemical reaction propolcd by Schmidt et al. (1986) is the proton exchange reaction between free water and bound water, and whose resonance frequency is about a few kHz. It is worthwhile noting that Schmidt also succeeded in modeling the ditsipation of .tectromapotic waves due to the e~mical reactions. He showed that the fnqueaty atteuation peak occurred at the same frequeacy as for the acoustic waves if changes caused by conductivity are ipored. Fipre · 5.S6 shows that similar behavior was observed for clcctromagaetic and aoouatic waves. This model raises two problems: it is a r~ of the speciflc surface area (see foreaoing Sections) and it fails to explain variatioas as a t\mction of saturation. Nevenbeless, it helps to explain the behavior of granite-lit-e rocks as a function of saturation (Fig. S.31b). In this case, attenuation increascs.with risin1 water saturation. Qualitatively, the porous medium of a aranitic matcritl coasists essentially of microcracks. The additipn of water to the porous medium, as satwatioa iacreases, gradually saturates a growing numhcr of crac:ks. The attenuation, which results from proton exchanges between free water and boWld water, can only increase as the number of saturated cracks increases, in other words with an increase in
e.
"--'
~
'-
'--
'..._..
2l3
where
~·~ '--
amfl.n iA~-~~
5
I'' o.sr---~r--------~------~.--------r--------,----,
Fig. 5.53 Comparison of theoretical attenuations of standard model aDd thermoelastic dil'uaive model (after Kjartans-
aA
son. 1979b).
01
:=:
.d'::
, ---
,
~---
1
,
71== :=;=--
L
'
J't ~:
' t
Ftwquiftey Cwrl
!fJ! Fit 5.54 Tlleoretical thermo-elastic ..uenuation vs. saturation and for dilrerent cfepdls (alter KlartaassoD, tmb~ .
.
....
8.1
~ flwlian ... "''
Fie- .5.58
Theotetical
t~
attenuation vs. temperature in three watersaturated These three samples have aa identical matrix and dil'enmt porosities. The dry velocity is the same (1 kmfs) and ttac satu-
samples.
lp
rated velocities are 2, 0
200
300
3 and 4 km/s (after Kjartaasson, 1979 b).
Tempemure C"CI
~
------~~~--~~-~-~~~~---~---------------'
;;:-
~
"----'
5
,_,..,.._~
l2S
''---"
•r--r--~-r--~-r--,-~--~~~-
o
..,..tGHI
• tlllli'Os , .. - 2.1 kHz
'-'
50 0
----
0
40 "----
I·
"--'
'--
'--"
20 '-
'-
10 '-'
.,
0..___,__....___._ _,__.__....._...._....,.,___.___, 0
-~
'-
'--
'---
,____.'---" ·~
~
100
Comparison of attenuations of e~omaptic and a<:oustic waves vs. saturation. Massillon sandstoae (after Sclmtidt et til., 19&6).
saturatioa.lil tile case fl FOillaiadlleau saaclstooe(fi&. 5.3la) we fmct one example of the same ptOCIIII. 'l1!le additionofthltftnUte tOOAI ofwaterftlls the ~fain contacts of the pore networt.leadilla to a wry apid rile iD attftuatkm. In an interpretative sense, Fis. S.3lb for aranite is analoaous to the very low saturation behavior of Fontainebleau sandstoJle. At very low saturations, as we have shown. other attenuation mechanisms are also invoked (hydroxyl bonds. capillary forces~
Ft.UI/IOU IMrtilll c~ This mechanism, associated with Biot's theory (Chapter 21. accounts for relative movements of the ftuid and solid. It is a bulk mechanism •bic:h, u we haw streaed, requires a hiahly permeable sample. This theory predicts _a substantial frequcacy depende~ of attenuation in the ranse 10 to 100kHz for materials such as Massillon sandstone. At the seismic fiequenclcs and at common permeabilities, the losses predicted by this ~octet are insisftiftc:ant. Moreover, MOchizuki (1982), recently showed that this mechanism implied attenuation values that were too low in comparison with the fortaoina data. The equations of Biot ( 1956) and Mochizuki (1982) show tbat the c:titical frequeaey ia proportioaa1 to tbe fluid ~· aacl that attenuation clecreaset with tisina viscosity. Thil result ia in eontradictic,.urith,tbc experimental results of Jones and Nur (19U) for oumple (FiJ. 5.57). t
"-
....,_.....•
40
Fla- 5.56
\,__..,-"
"----
20
226
5
RESULTS AND MECHANISMS
ll
(GPIII
~--·
615
40
g.
I
~30r 20 10
0.2
Pc= 10MPa pp=4.5 MPa
·' 0.4 0.6
"
1
2
fNqiMncy X
Dyn~~nic
vilclllty (kHr • d')
Fia- 551 Attenuation of S waves arid dyl\lmic modulut as a function of the product of frequency by dynamic viscosity (resonant bar) (after Jones and Nur. C 1983 AGU).
~
~
'
Finally, this model predicts a dependence of the c:cntral frequency on the inverse of permeability, a dependenc:c that has never been actually observed (see Table 5.1).
~-
re#utltiolt of tM "*''ltiiW p.i4 If the fluid phase is sufl-.ciently viscous. viscous stresses opposed to the motion are added to pressure. A .unplc cakulatioq (Nur, 1971, \,\'tWa. 1N9} -~ lhows that the riequcaCies at a~ oporatioaal arc \"et)' biJb for fluids such as water, and for liaht oils. In fact, ~- ~trallrequency of the mechanism is g.
1
JrUCOIII
! ,
which
Jiven by:
eva
_mecba.., --ep
f/
Q)e- -
(5.1())
where
e - aspect ratio of the pores, p .... shcaE modulus of the matrix. '1 - viscosity of the intentitial fluid. -
For water, cue is about 109 to 1012 Hz in satura1ed -sandstones. However., for rocks in the state of partial melt, this mechanism may~ important at the seismic frequencies.
._ Rep.djlu This is the fluid flux between the peaks aad trdups ofa strain wave, or the flux from a high stress •• regioll" to a low s&ntss ••repon ". 'Fhi$mochanism concems a group or several pores. It depends on the permeability of the rock and the wave amplitude. The transmitted
',
~
----
-
=T ""'~.....
'
..........
-....----
5
227
wave period lllust be or the l8llle order ofmapaudc utho PRBSurcrcluation time for the mechanism to be siplifacant. MaximutD aatenuation is obtaiaed when the period alld relaxation time are identical. or for a frequency given by:
y2
I= 4n2Co
(5.11)
where Y is the seismic velocity and C 0 the hydraulic dift'usivity. We have for the hydraulic diffusivity«41 : CD= KKt,
"
(5.12)
where " is the pcrmeab,ility~ ~-the porQSity, K 11..the bulk moclulus of the fluid and 'I its viscosity. -' For the usual values o( the fGftgoia)a parameters: ._, : ;
'
I a too ~Hz '--
This frequCncy is hiper ~l'l t~'one fouo4 in,·~ abQve Jne4$Urements. Furthermore, it depends linearly on the inverse permeability, Wbidl is not experimentally observed (see Table 5.1). Fipally~ the depeadelieeof'frequcncton\tiscosityiS theinverseotth&t observed by Jones and Nur (1983){Yaa. SS1). Hence~-~ does not appear appropriate.
or
'-·
i. "Stpiln" or "MJf'b/1 "jlllw " This is allOCher rnedla1dsi'D of~ relauticm, but tile exciting mechanism is external, the ''squirt"or ''sqvilh"ofthelui4iatllcpotbutcavity. The .. squirt low .. or the "squish flow" ._...... . . . mvolw:~lieleal1'110tioa d the -fluid -ia the porou cavity with a hiP Reynolds - • · Tbis mecllbi• it i.Utialiad bJ tbe ,....,.. of the wave. In fact, the ~of the . . . • tate capila., tubea , . _ .... a1ocat pore pre~~ure which iltl the fluid in -...loa __... lllle ,.,._, 1",_ • tllitoMical ~ it 'has boell deAtonlttated that Ibis t)tpe of metlllanilm'tOUld by a standard Yitcoelastic model (a.,ter 3) (MaW. aad Mur, -1979; Pa1altr ldld Tra.totia, 1981, and O'ConneU aad BlldiMiky, 19'71). The reaent.aperimeatal data efJoaes and Nur(l983)(Fig. 5.57)or tlloseofSponcler(t98l)a'PPe&Uooodtma;belraWor'oftbeZenertype.ln this model, the liquid low can Oftly Ufke place iftbc ~is tower tba1l a limit frequellC)' (Murphy, l982b):
be..,......
'---
Ji .!
f&,
b"-;;
~5.13)
where b • dime~'ofthe fluid dr()P, K [I - bulk modtd"s or the ftuid,
p1
-
ftQid density.
(4) c, coeft'tc:ient is actually the one cWmed by Eq. (2.120) if TerzaJhi's hypothesis is made [M ""' K fl{•• Eq. (2.126)] and 10 - i.1 •
'-
228
s
USULTS AND M£CHANIINS
This limit frequency is about l GHz ill sandstollCI and sands. The central relaxation frequency is given by (Palmer aad Trawma. 1981): s /,= K e (5.14)
.-,
where
e - aspect ratio, K. =bulk modulus of the skeleton, 'I = fluid viscosity.
For high porosity sandstone, this frequency is about 1 to 10kHz, and the attenuation amplitude is approximately the S&lDe as that in Massillon sandstone at 500 Hz. This mechanism thus qualitatively explains tlie frequency dependehce obsei-ved in rock samples. We shall now examine the clefendence of attenual;ion on saturation. At total saturation, in compression, the intentitial liq~id offers resistance due to its low compressibility, and low pressure gradients are aenerated inside the porous medium. The resulting attenuation is therefore sli&ht By contrast, in s~. the pr~ure gradients generated are ~ter and the R$ult~ llttelluation is hiaJlef. Henee the result that Qi 1 is greater than Qi 1 at total saturatio~ (Fi,P. 5.31 aild ·s.52). , At partial saturation, the problem is totaUy dift'ereilt. The wat«jr,as mixture in the pores is highly compressible, and the extensional attenuation (Qi 1 ) dominates (Q£ 1 > Qi 1 ). Also the existence of an attenuation peak as a function of saturation is observed. This peak occurs in the Massillon sandstone investigated by Murphy,1982, at;a-water satwation of about 80o/e.lt can al.sobo._.Wi&. S.
......
of'_.._
Hence the ··squirt flow" process appears to partially and qualitatively explain the attenuation behavior observed. However, no theory i~ ye~ ,.dequate to provide a more quantitative 'iew of the process. It should therefore come as no surjrlse that, given the complexity of the porous medium, no single global or local model based on arains and spherical cavities can be adapted in detail to the fme description of mechanical behavior. The question remains whether it is necessary to complicate the models used ad infinitum.
j ~
--..
~~T \_/
·s
ialttif.AWJ~
229
5..1 RESULTS AND'MlCHANisMs CONCERNiNG IN SITU MEASUREMENTS 5.2.1
~
Intrecluction
The measurement results that we presented in s.ction S.l Wet"Cobtaincd by &aboratory experiments. The value of this type of measurement is obvious both for an understanding of the mechanisms oblerved and for the applicatioll of the results in the fteld. Nevertheless.
"---'
it is essential to be able to measure velOc:ities and atteauations in situ for many pnctical reasons. For example, the reconsttuction Of tile subsuiface acometry requires the blowlecfae of velocities in situ. We areiliterested here in the r¢1ationship between acoustic measuremelltl ad petrophJiical dwactotistic:t Of in Jitu materials. In fact, considering the rite in oil prices UftOaJ other factors, the need 'bas arisen in petroleum popb)'lic:s to determine not only the 'pometry of the beds, l>ut also their lithology, fluid content, porosity and permeability. Sonic weD louin& ori&fnally intended as a .. simple" conftrmation of depth for seismic sections, bas been 0.0 to approach those problems due to its finer depth resolution. It is clear that the in situ mciasurement of acoustic properties is an inverse problem widt ftiiiPICl to laboratory JRea~UN~DCDts. In the laboratory~ the petrophysical proportiel(porosity aadpenatabitity.ofthesampleare weD known and She meuuremeat is·intonded to oblerw the wriatioa in acoustic: propertie$ as a fuaction of physical parameters (twessule. temperatare). Ia the fteld, by contrast, the· s;hysical conditions (pressure, temperature) are relatively well known, and the· geopbytic:ilt is illterestcd in determining the petrophysical cbaraeteristics of the formations from the
·.._
'-
'-./.
acoustic properties measured.
The comparison ~ field meuuremera.tl aad laboratory measurements is not an easy one. It is often tacad witll_problelas such ..- dia~ induced by the temperature and stress reoond.itioniq of t.bc samplei 1a fac:l. AOt uusual to ftnd that laboratory measurements display· a lt)'ltereais iadic:itiaa -~·. inevcmbiiity of a structural c:banp durin& a measurement cycle. Tbia mca01 tlaat tlae results obtained cannot be representative of the behavlot of rocb buried \lDdel' kBOmcten of iediments for millions
'~.
"..i•
of years.
~
'-
~
ddcb:••••
.._. ~
i
i
'-
But the most important poiD& is &bat the frequencies employed in the laboratory are often quite different from 'dlele employed in situ, namely about SO Hz for seismic proapectina and 10 tlb for Well -lolling. The ebaraetoristie waveleftgth in seismic prospecting is thus quite dilrcrbt from the ~ waYCJenath in the laboratory, aDd, since the resolution of the measurement is proportional to this wavc:!ength, in situ meuuremeats tlwt itftelrate fonutioa that are tarely wUform. Fmally, the c:onstderlable frequeftcy ·depeDCieMe of the pt«111ill (espeCially attenuation, aee for example Fit- 5.4Sr~tt ~ direCt uttapdtalten of labOratory ·rei\llts to the fteld; Howewr, laboratory ~ ~- the existence of Jimit vahtea for velockiel ad a~ uader.-(Ap. S.l 'ad J.27foreumple). These limit values are usually ealklid ·tetmibal wloch:les . - tltmtdal atte~ntatkms (Wyllie tt al., 1951) and occur at efrectiw pNIIUI'II of abOUt ·lOO'M~ ~n1 to sediment thickaeDes in kilometen. ·
I
j -·
~-___,
230
5
RESULTS AND MECHANiiMS
In practice, and particularly for velocities, users have developed a number of empirical laws, enabling them to deal with thcr inverse prot)~ they face. We shall examine the results obtained for velocities, amt theft 'for attenuations, and draw conclusions on the usefulness of these empirical taws. -~'
5.2.2 Velocities ::.,;..-...,4
5.2.2.1
Ia situ velodty measunlllelltl 0
""·
In seismic prospecting vdocities are determined by intervals, usina the reflections obtained on a single mirror point. The relative error in the determination ·or these velocities is about 1 to 2%, whe~ absolute accuracy is about 10%.1'hese measurements ~ taken with P and· S waves . ~ to dle t}rpe of source employed. The reader
--.._
wisbina to familiarize ~If with _seismic re~on measurina techniques can refer for example to the work of CQrdier (1~83). In wdlloain& the determination of velocity is standard, as the distance traveled is ac:cura~ly known. The absolute accuracy is very hiah,
about 1%.
-----~
5.2.2.2 Veloclties aad porosity measure~Detlts
of Wyllie et GL. (1956,. .1958, 1962) reveal a simple relationship between velocity and porosity (Fip. 5.58, 5.59 and 5.60) for saturated samples under suft'ldent stress (terminal velocity). aacl of similar mincraiOJical oamposition, naJDCly sandstones and ICdimentary rocks in this case. Tile equation for P waves is called the Wyllie equation:
The
l • 1-t; -==-+-v v, v,.
-,~
(S.lS)
where 4> is the porosity, ~·the \'etocity ofJhc qturated rock, ~i the velocity in the fluid, and V,. the velocity in the rock matrix. This meaDS ~bat, if the type of rock, saturating ftuid and velocity are kno1m, the porosity can be catculated. The matrix velocities are given below for three major families of 'fOCks (from Schtmilberaer Co., 1971).
v. (mfs)
Sandstones .......... ; •...•.•... Limestone& ................... .
Dolomites ................ , · · · ·
S488to S950 6!tOO to 1000 7000.to 1925
This extremc.l.y ~pie equat.ioD noaed1eless requires many precautions for its use. The vcloQties are :quito different at shallow cicptbdrom those aivcn by Eq. (5.15). Fiaurc 5.59 shows that, at biah.poroliUu. thee~~ poiJUadeYiatefrom theaverase.curve (it ia not possible to apply a suftlcieot stress to~ uoconsolidatod.samplc,s, and the concept of terminal velocity boc:omes mea~). Fisuro 5.61 dearly shows that, at atmospheric pressure, the velocities of Fontaiacbleau ~roaes differ substan&ially from the average velocity (up to 40%). The apP.ication of a uaiuial stress of S MPa reduces tllis deviation from the average velocity to a maximum of 15 to 20%.
i'
I I',
I
I
I
~,
~
~
'-'-
I
'i.··''"'· t,: ·~ (
Velocity Cllmltl
3.5
3
4
4.6
'~<
:·~
Ytllollty lkmltl
5
35 110
•• •
30
80
25
~
_60 ~
20
~so
~
·g :
15
40
............
10
e Tripolhl
5
·W-......S 90
.,
Trwel
tm. Clts/ftl
0
120
110
100
eo
70
QI
5()
I
I
I
I
I
110
140
120
100
10
F~&-
,
1.1'
30
25
!:zo ~
J ·~
11 10 6
0 ''
f
1).2
I
II
o.a
Poillon's ratio
·~.
FJa. 5.60
lS" I 10 40
5.59 Compressional velocity/~rosity relatioasbip for siliceous rocb UDder . uniaxial 1tre1s (ultrasonic measurements) (after WyUie et al, 1958).
'
35
I
,.,
Trft time Cltslftl
Fig. SSI Compressional velocity/porosity relationship. Comparison between laboratory measurements and Wyllie's equatioll (ultrasonic measurements) (after Gre,ory.
1977).
I
200
,
110
I
160
I
I
140 , .
J . . ' .. •
t
'
I
..
'
I 40
.
Travel tin. Cltslftl
Experimental ret.tionship. Dear waft veloc:ity and Poisson's ratio as a function of porosity (ultrasonic measurements) (after Wyllie et Ill., 1962).
232
5
RESULTS AND !otECHANISMS
•• ••
5.5
5
•• 1·· • •
?:
j
.: .....
•
Sw=100%
4.5
~
• •
I
4
500kHZ
I
•••• •
3.5 3
•
•
s
0
20 •.:·
15
10
• 30
25
PorCIIity "')
,,. .:1
..
!5
8?: ...5 l
•
...t
b
... • I
•
4 3.5
•
Sw•100%
.... ••
500kHz <..',
• •
3
u 0
5
15 20 Porosity I'K. I
10
25
30
Fit- 5.61 Co~~ velocity. porosity relatioubip in Fontainebleau Sandstone (ulpsonic Dle&surementa) (after Bourbie and Zinszner, 1983). L Atmospheric ple8SIItC. ft. S MPa uaiuia1 sttess.
The only intrinsic velocity being teqnina} velocity, it is the only 'one that should be introduced into an equation such as that of Wyllie. Thu&~ndercompacted materials fail to satisfy Wyllie's equation. Moreover, it has been pointed Qut that, at very high and very low porosities, Eq. (5.15) is 'inadequate. Jd,Oy authors have proposed a modified Wyllie equation, such as Nafe and Drake-(1963) arid more recently Raymer et a/.(1980). The latter propose the following equatipns :._, · Consolidated rocks:
.<35%
V =(I -
t/>) 2 V,+ tj>V1
:
(5.16)
Unconsolidated ocean floor sediments:
t/> > 45%
1
t/>
l-t/>
~::~·*""'"?VI" pv• p1 v 1 p.Vr
(5.17)
.,
>"
~
-....-
1 I
·s
~tt~tliD~~
133
The comparison with Wyllie's equation is given in Fig. 5.62. Figure 5.63 gives a porosity /velocity Idationship for uncoasolidat61 ocean floor scdimcats, and Fig. 5.64 shows a correlation with experimental data and average regression lines for sandstones, limestones and dolomites. To a certain degree, it is therefore possible to relate velocities and porosity in a biunivocal manner. The extrapolation of porosity from velocity will have to be performed with precaution, as Eqs. (5.1 5), (5.16) and (5.17) are empirical experimental relationships and not physical laws.
S.l.l.3 Veloddes aad dentlty The densities of the materials encountered in ~ry basins vary relatively sliptly in comparison with velocities. Nate and Drah(t963)nd Gardner er al.(t974) proposed a relationship between the termiDat P wave velocity for saturated samples and the density. Figures 5.65 and 5.66 show this experimental relationship. The curve ~ be approximated by the follo\\ina equa~on: p- 0.31 V0· 25
(5.18)
where p ==density in aJcm 3 , V = velocity in m/s. Moreover, we know that the density of a given totally saturated material is related to the densities of the matrix p., of the saturatiq Ruid ·p1 , and the porosity t/J:
p .. p,(l -
~
+ tflp,
(5.19)
Equation (5.18) .:an therefore be transformed by means of £4. (5.19) into a Wyllie type of equation: (5.20)
V- V(•• p., Pt)
Wyltie'sequatiea c:u be-expralltdiChcmatically in thefortll V = V(t/J, V,, V1 ) where V, and Y1 ate tbeiftatrisud ftuictWiotities respectively. Hence Eq. (5.18) involves only the densities and not the elastic moduH, unlike Wyllic"s equation. It is therefore a less accurate equation than Wyllie's, and its application is mainly limited to the determination of an averaae density from the measurement of a terminal velocity.
5.1.2.4 Velodties ud day eonteat Itcccnt measuten~oents(De Martini et al., 1976, Tosaya and Nur, 1982. and Kowallis et al., 1983) show that -at.ionsbips exjst for water-saturated samples between compfessional or sbcar.velociticS,and p<»rQSity or clay content. The experimental results arc given iB Fip. 5.67 an4 ~.68. 11ae relationshipi between Vs, porosity • and volumetric clay content (c.) are(To.-Ya and Nur, 1982):
v,.,
V,.{km/a)• - 2Aca - 8.()(p J's(km/s) ~,..::.. ~~Jc.- 6.3q, for a differential pressure of 40 MPa.
j
+ 5.8 + 3.7
(5.21) (5.22)
234
R!:SULTS AND
5
M~CHANISMS
100'
80
~60
·i2
j_40
20
O''"
I
80
60
.., I
I
I
120
140
,., I
I
I
180
200
T.....t time (l.cs/ft)
Fig. 5.62
Modifacations ofWyiJie•s equation proposed by Raymer et al., 1980.
Velocity (kmlsl
100
•.•
•.•
·-
\
Fft. $.0
Experimental comparison of travel
time .-d porosity in ocean floor sediments
\
(after Raymer et al., 1980).
\
\
4!
~
Rt- 1M
I
Porosity/travel time Nlationship
Cor clil"oreat typeJ oC rock (after Raymer et
C\f» oo/
al.. 1980).
c:4,•
Vlloclty lkmls) ·
% /O oolacD
7.5 8.5
•
8
," I
3.5
(
3
2.5
,,
... -----. --
4o
e~
i3G ~
40~----~----._----~----~ 180 110 200
210
Travel time (la/ft)
4.5
so~~~~~T-~---,-----r------~----~
0/0 0
' o
u s
J
l~
~
... ,
.,..
1
~ le1Urftld
. , ...
cv,, • 100 mlsl
10 0 40
50
80
70
80 90 100 Travel tm. fj.q(ft)
110
120
130
140
....,7
--- ,, --·
-"~- ~-
f1a. 5.65
~ Compreuional velocity/ daiJity rdatioaabip iD various sedimcau (after Na(c and Drake.
8
I
C 1963 Wiley ~Dterscieacle).
' I
5
i:!!. -~
J
4
I 3
2
"'~----
I
1~--------------_.--------------~ 1 2 3
I
llulkchNity 7
• I
~
14
f.
.... '
3
2.5
2
..... " ' Compreuional veiotJM.y/ density reJatioDibip for clift'ilpat types rock (after Gardner et Ill;
1.5
or
1974).
1.1
2
u
2.4
2.1
lulk IMnlltv (log ....l
'-._..
2.8
3
6r-------------------~
3.5
Vs
Vp
P-eff • 40 MPI
6.5
Pett=40MPa
s.-too"
s.-=100"
i
i
.>/.
i
~
4.5
"jj ~ 2 3.5
31
0
I
Jh.
20
40
~
'
80
1.5 1
·'
•
0
80
20
40
"\
60
CIIY content ('\ vol.)
Cliy content '" vol.l
(a)
(b)-
i
'
80
Fia. 5.67 Velocity clay content relationship as a funetion cl porosity for saturated samples. L Compressional velocity (after Tosaya and Nur. C 1982 AGU). b. Shear velocity (after Tosaya., 1982).
5
I
v,. Pc•tOMh
0" 4 .............
5
.f
I
I
3
2
10
11
'20 POI'OIIty
Fft. 5.68
"'J
25
30
Compressional velocity 'clay content relationship as a function of porosityfor dry samples (after Kowallis et al. C 1983 AGU).
'-'-
5
USULTS
AND~·
237
These eqations allow a better estanation of porosities if the velocities aaa tbe ~verage day C01ltent • kaowa. They olwioully depend on the differeatial. preuure applied. Finally, h1ce the fonlaoiDi eqpations, ·. tltey empirically enable the eVIIIaation of one parameter if the mnaininJ two are bown.
5.2.2.5 Velocities .... CIIIDpAction The burial of rocks in sedimeritary basins, for example, generates compaction processes due to the lithostatic pressure. Compaction is the~ in porosity due to the effect of overburden pressure. Different geological bodies obey compaction laws that are more or less woO kaown but aenerallY distinct. The knOwledge of these compaction laws and the meuuremeat of velocity variaboaa with depth can thus help to provide an idea of the geological characteristics of a layer or a JI'OUp of layers investigated. Faust (1951) developed an empirical law from some 500 velocity surveys (Fig. 5.69). He showed that, in shale and sand sections, the relationship between compressional velocity and depth was: 1
V,.==B(~n
(5.23)
where ~ • depth in meters, B = constant equal to 46.6 for Faust's samples, v,. - compressional velocity in m/s, ~ == age of the sediment in years. In carbonate sections, a general equation of type I5.23) is difftcQit to determine. This is because tu role of compaclioa is-leas uniform titan in shale and sahd sections. J•owsky (1970) nevertheless proposed a qepession of the'ame type (Fia- S.70). or sands, Gardner et al. (1974) and Domenico (1977) proposed the ,-elocity/deptll relationships shoym in Fis- S.7L.:Siilhl~amobservable in comparison with faust"s rclatioe that result from specific pressure aacl coatpacd6n elwacteristics.ltiliatefestinj to note that a power
f.
I
'--
I
dependance on depth. ~.was also found by G~ (1,51) for the P wave propagation velocity in a helagonal packing of spheres (COrdiet, 1983). The usc of Faust's relation or an intCJf&ted law: s t==B'~
where tit the vertical propaption time, offers a auess at the type of lediments analyzed. If the law applies, it is pneratly a normally compacted shale and sand section. If not, it may consist of carbonates or evaporites. or even of hi&hlY tectonized rocks, or series that have undergoae erosion after b\lrial. In this latter case velocities are higher than the ones given by Faust's relation. If velocities are lower, the formation may consist ofundercompacted shales or very biJb porosity series. Finally, as we pointed out for~ previous relations deanna with in situ velocities this is only an empirical law to qproximaiHhe depee of compaction of deep sediments. Several variations of Faust's relation exist, and these are neither more or less accurate, but merely indicative of slightly differetrt sediments.
5
Fig. 5.69 Compressional velocity/depth relationship for different ~ series {after Faust, 1951).
-;;4
l
~
f
l !
3
D .luralic·Tri.aic
.
oea.-us
A~OE-
•
2
"-••~
0'-'lln
0
o Tertlety
......._,
·3
2
4
Qlpth ClPnt
Fig. 5.70 Compressional velocity/depth relationship for differeat .lithoJoaics (after . Jankowsky, 1970).
0.5
.,,
Shaly 1u...-,..._;on liM
e ~
1u 2
...('
...-.onn.. ,
•
''
2.5 '---'---''----'----'U...--.....;L..-_ _ _ _.....J 800 500 400 300 200 100 TliiMl tllne .,.,..,
e ~
J ._.,,... lrlne-...ct
Fig. 5.71 Compressional velocity/depth relationship for different types of sand (after Domenico, 1977).
2
• 2
2.5
3
, _ OM!ocity (kllllsl
• 3.5
'-"~
·~
s 5.2.1.6 ,.
~
I
dSULW'Atlt
~--'
239
V,.f Vs and Poisson's ratio
The ratio ( v,. /V5 ) of lonsitudinal velOcities ( V,.) to transverse velocities ( V1 ) corresponds one-to-one with Poisson's ratio v: Jj.""
1$
J2(l -
Y)
(5.24a)
1 - 2v
or
,. = O.S(V,./Vsf -
l
(S.24b)
(V,./I's)l.- 1
The c:alculation of Poi110n's ratio._ requiftlll, tbe limultaacous measurement of the velocities 1',. and J's. The compilatioa made by LawrtM{pcnoul c:ommuaic:ation) helps to show that the different types of roCk display. rather different Poisson's ratios (see Fig. 5.72). ·-'-
·~
............
~
.we•• •
Dry••~_. . . . . . . . . . .
'-
. 0
0.1
l'-
•
0.2
•
.~~
~.3
-
•
OA
o.a
......... ,.tlo
J11e. 5.71
Averqe Poiuon's ratios for di«erent lithoJoaiea (after Lavcqne,
penonal correaponcleDee~
._/
'-
Hip values of Poisson's ratio ( > 0.35) (or high ratios Y,/l's) correspond to UllCOIUOlidatcd rocks. Compact rocks dispJe;y P~'arati9!1~ 02 aaci0.3S.Yt'bile ps sands have a very low Poisson's ratio, about O.l.ln water-saturated sands, by contrast, Poislea's natio ia about 0.4. A awnbcr of average values of P aod S wave velocities are pvea in Table 5.2. 1be for:qoiaa c:orrelalions bctweea ~ type of rock,. saturation and Poissen's ratio (Fig. 5.72) are merely indicative values .Uowiaa the ppbysicist to focus on the polosical cbarac:tcristics of.tbc rOICks inVC!It.i8ted.
'-
5.U7 Sa-mary • ia :sial Wllodty ......,...... ''-
r=
3.E
.__/
.__/
All the empirical equations that we have pointed out show that, if v,. is known, or preferably v,. and J's, a pas can be made u to the Bthol()ay arid the porosity (but virtually nothins about the saturation) of the material stUdied. This is not illtended to determine
precise values but to set limits between which dW: true nsponse. should lie. It is also important to point out that, strictly spcakins, tbe foreaoina equations are strictly applicable only to the samples investigated. since variations exist from one type of sample to another.
240
s
ResuLTS AND MECHANISMS TABLE
5.2
AVERAGE PROPAGATION VELOCITIES OF p AND AND AVEitAOI! ROCK DENSmi!S
Type of formation Scree, vegetal soil ............ Dry sands .................. Wet sands .................. Saturated shales and clays ..... Marls ..•......... , ......... Saturated shale and sand sections Porous and saturated sandstones Litnestones .•..........•..... Chalk ...................... Salt ........................ Anhydrite ..............•.... Dolomite ................... Granite ..................... Basalt ..... , ...............•. Gneiss ...................... Coal ....................... Water .... :: .....•.......... Ice ......................... Oil ............. ·.............
S
WAVES
Pwave velocity (m/s)
S wave velocity (m/s)
Density (g/cm 3 )
300-700 400-1200 1500-2000 11 ()()..2500
100-300
1.7-2.4 1.5-1.7 1.9-2.1 2.0-2.4 21-2.6 2.1-2.4 2.1-2.4 2.4-27 1.8-2.3 2.1-2.3 29-3 2.5-2.9 2.5-2.7 2.7-3.1 2.5-2.7 1.3-1.8 1 0.9 0.6-0.9
~~
J
100-SOO 400-600 200-800 750-1500
500-150 800-1800 2080-3500 3500-6000 2000-3300 2300-2600 1100-1300 4500-5500 2500-3100 4000-SSOO 2200-3100 3500-6500 1900-3600 4!00-6000 .2500-3300 28410-3400 ~ 4400-5:200 21c».3200 2200-2700 1000-1400 1450-1500 3400-3800 17004900 1200-125<> ...
'·
Density of constituent crvstal (g/cml).
2.65 quartz 2.65 quartz
2.65 271 2.71 21
-
quartz calcite calcite halite
··~
-
(Ca, Mg)C0 3 1.8-2.99
'
------
After Laverpe, personal correspondence.
·~
!.2.3 Attenuations ~.1.3.1'
Ia situ .......................
Measuring in situ attenuation is a delicate problem, and fteld measuretneftts found in· the literature are oftetl debatable. More6ver, the attenuation measured by various authors is generally the total attenliatiotl, in other words the sum of the intrinsic attenuation directly related to the porous medium, and the extrinsic attenuation resulting from the geometry of the subsurface and of the source, scattering, etc. The most reliable field measurements are taken either by means of weiJ.loging tools such as the EVA (SI tool at frequencies around 10 kHz, or by means of VSP (vertical seismic profiles) at the seismic frequencies. Table S.3 Jives some ...rences aQd results of field measurements. We shall _ discuss these techniques ip the tuwCbaptcr. .In this Section, we. have decided to present so~e results and ho~s conq:J-nina t~ use of ahen1,1ation m~urements.
~'
'~
(S) The EVA tool (Elf Aquitaine trade-mark) has S transmitters and 12 receivers.
\
j
r
~
I
(((
(
..._
(((((
(
(
(
,'.),'.
~~~--
...~
--
V.U.IIF.S Of
Type of rock
Location
Limon (Colonulo)
Depth 1m)
Measurement frequency (Hz)
Q,. apparent
50-450
32
SO-«ll
2 181
Sandy day Clay/saM
()..3 3-30 30-150
IS0-300
50-400
Clay/saad ·•
117D-1770
Saadlaad . . . .
Same .,_ _,. saMr
1'710-2070
" 125
So.at.....·T8aJ
Sandbaab.Jftll tiltyahale
Southeast Texat
Mostly . . Sand (23%} aad clay Sand (2C)-4l afld
2010-2850 9$0-1560 JSf0-1800 t8Ci).2100
Oult Coast (30 km south or
.. Ofl'shore lovisiana. (~ .
.
/1':
Southeast Teaas Hcauf~t
Loam/saitdfcl&y
Sands and..._
Houstottt.
--
clal
l.ime!etotte an4 clta Sand(4S%, qct day
Sand(~· aifd'~)'
Sea (('anada)
otfshore
SiliccoU. chaJit
6GO-IS60
From Carmichael (1984) and Goldberg fl98S).
50-400 50-400
"125
"us " 80 " 80 c;80 c;80
15 136 67 > 273 28 52 >'273 ., 30· 41 > 273
>1020
._ RO 15-40 4()..70
!i4CJ- 1193 945-1311
12S 425
278-442
5000-1 5,000
442-5H2
on avcraJlC 5000-1 5,000 2H
1590-1755 ~1320
Baltimore
Silieco111 Chalk with porcellanitejoina.·
~
--
'_:
___
TAILI! 5.3 Q MF.A.~IRF.Il IN SF.Il1MI1NT!I
0-225
Pierre sllale
-
28 55
References
corrected
McDonald (1958)
..
Tudos (1969)
et
and
Reid "5 ·,.:
·( ,,;
:;·
Haup(I9Bt) ;:
46
> 273 34 94 43 67
Oanlcy and Kanasewich (19110)
68
on average
at.
'
67 > 273 31 •·109 > 273 37
Goldberg (1985)
( l, ~
242
5
RESULTS A"N"D MECHANISMS
5.2.3.2
Results
The pessimism expressed above must be moderated by the results obtained in the laboratory. In fact, attenuation is a difftcult parameter to measure. but provides considerable infonnation concerning the medium investigated. especially with respect to. saturation. Figure S.73 shows that knowledge of attcmuations and velocities provides a better defmition of saturation than that allowed by knowledge of velocities alone.
--.---4 ~
,r
0 0
Vp
0
I
I 0 01>0 oO
I
° ~
0
0
r...itlon Andst-
•Drv
6 . .,
2 1-
6
• P-'illly a1Ur~ Fully a1Ureted Sierr• White gr.nite
, t::
•
0 Plrt;.Hy a1Ur~
o
•Drv
I
2.0
I
1.5
-
• Drv
l I
I
I
I
::j
1.5
2
2.5
3
~to
0,
0~
0.0
-
• Pattilllly . . .reted 0 fully~1MI
Sirrll White gr•nite
6• '(6 I
1.5
~
•Orv 0 f'ltrt;.Hy •1Ur111MI
-
.... 0 0
0
I
9_
2
0
I
2.5
Vp/Vs
Vp/Vs
C•l
(bl
3
"""te
FiJ. 5.73 Massillonsaa4ltoac and Sierra graaite. The arrow indicates increasing effective pteflures (resonant bar) (after Winkler, 19791. L Relationship V, as a function of V,.fV5 • . b. Relationship Q; 1/Qi 1 as a function of V,./V,.
Although. oa the averap.~tenua~ rises with porosity, Figures S.i4 aad S.1S show that attenuation/porosity. relationships· appear to be less simple than velocity/porosity relationships. Hence, with. the attenuat;iort parameter,lhe possibility exists of obtaining comple11lentar)' data ab"Gut the porous medium. Nur n Gl. (1980) pointed out that S wave attenuation and permeal>ility couldtJe correlated; Wesball discuss a possible explanation of this in the next Chapter. . . .
S.2.4 Conclusions on in situ measurements Field measurements lead us to derive a number of empirical laws for seismic velocities. These empirical laws enable the geoP-hysiciSt to pinpoint the petrophysical properties of tbe materials analyzed (porosity, debsity, -~Y content, compaction). It has not been possible to establish empirical laws fot attenuations, because in situ measurements are still too few in number and inaccurate. Ultimately, it is likely that reliable in situ measurements of attenuation will be obtained. These measurements will then allow an evaluation of other petrophysical properties, such as permeability and fluid content.
~-
.4
,__,~
5
' j
-
243
USULTS AND MECHANISWS
1000r---------------------------------
o,
·~
I
0
L-1.0
tocl
I
\00
t
.
' ...
....
lgneoulll'ld rocks
*Me_...e.lc
101-
t
o
..
Ll"*'-
....._
....... _ol 10
1
0,1
Porolity
FiJ. 5.74
(~I
Attenuation vs. porosity in a number of rocks (after Johnston et al.,
1979~
'-~
~
500 L 400
1-
1~:t 100 L
•
Op (500 kHzl
•
a1 =0 sw = 100%
•
• • •• •• • •• • •
.,..
10
.
•• •
'
15
•
Op (&00 kHzl
200
= 5MPa Sw= 100~
a,
•
!..!!!9 150 Op
•
260
,.
!10
•'
0
20
25
30
0
...
5
•• 10
Porality(~)
•
•'••• •
• ,• •
100
•
• 15
20
25
PorOiity (~1.
J'ia. 5.75 Compressional wave attenuation/porosity relationship in saturated Fontainebleau sandstone (after BourbiC and Zinsmer, 198S).
...
30
•
v~r ~ ~
. i
i
'f.
:':l
',,;,
6
.. ;
~~
i waves and intertaces
'--.../
A
INTRODUCI'ION
'-.-/
'-.."-_
"'---'
·-·
_/
"---'
In Chapters 2 and 3, we have studied wa¥e propaption in i'ffinite poroelastic and viscoelastic media. In the subsurface. acoustic waves are reflected and transmitted at interfaces which are characterized either by contrasts in elastic properties (acoustic impedance contrasts) or in anelastit properties(attenuation or permeability contrasts). To interpret field data, it is esscali8l .to UIMlentaaclllow theloioterfaces modify the reflection and transmission of waves. Moteover the existence of interfaces allows the aencration of interface waves (Raylciab or Stoneley wa•) w~b may also be used for the detcrmiDation oflitllolopcal -~ petrophysical propcriies of the subsurfaee. In this Chapter concemia& iataface p:~~ ~.,_..,we will study firSt •UI'ated porous media, and secondly viscoelastic media. Finally, we will show that the two models are complementary, and must be used jointly mcertain cases to correctly interpret fteld
....
-..__/ ~
"-../
6.1 WAVE PROPAGATION
IN SATIJRAT£1> POROUS MEDIA DISCONTINUITY- EPFBCTS
---~
'--' ....,
____
_/
~
we--... in the«JJldusionaaelaapter 2 th&&.c.t'~ porous mediamodelod by Biot's theory, permeability plays a role essentially in the presence of discontilluities. More precisely, the discontinuities, by the bo\lndary conditions that they imply, induce pRSSUre aractients which modify the ftow. Heace they jive rise to attenuation. The poasible existence of a second kind P wave also implies sipifac:ant chances in correspondina clutod)'UIIIie ~. . In the flnt Sec;&ion of dais~. wethaU ~the boandary con4itioDs tMt must be satisfied at the interface of two saturated porous media. The IOCXIIld Seotion will deal with wave reftections at this interface. The fmal Section will show what happens to the
~,
246
WAVES AND INTERFACES
c~todynamic
6
interface waves. We will also examine a typical problem in borehole
seiamics in which a source is immersed in a fluid and acts near a permeable interface.
6.1.1
Boundary conditions
Let us consider an interface S separating two saturated porous media denoted by superscripts (1) and (2) respectively. The unit normal attached to this interface and oriented from medium 1 towards medium 2 is denoted n, with components "r The direct normal vector ton is denoted t with components t1 (see Fig. 6.1 ). The notations are those of Chapter 2: displacements and macroscopic stresses ui'1 and ai7, pressures p
-,
lSI "
M.dlum1
•n
Medlum2
Fil- 6-t Notations at a Jiven mterrace. At the interface, in additiOn to the fteld equations valid in each medium and developed in Chapter 2, ne1v coaditib..S 'ftltlit 'be satisfied. The first deals with the continuity of macroseopic displacements. 11Us is tile tmemadc cbndffion:
'' 1411 .,., ui 21
(6.1) '
The second condition is the CODICI'Vation of fluid mass and hence the continuity of' the flow through the interface. This amounts to writins the continuity of the filtration velocity component in the direction n, which we shall denote w,.:
w-
11-
.:.ell,. = W!lln "'I I I I
"
(6.2)
The fmal conditions to be expressed are the so-called natural conditions, derived from Hamilton's principle and ~with the~Mcroetopk:ltresseso17 and pressures P''' at the interface. These conditions were obtaieed by Deietiowia aad Sk.alak (1963). We shall adopt a presentation here ,tbae is close te their oWn, but in agreement with that of Chapter 2. Just as in Section 2.2.2.3, where we defaned a volumetric pseudo-potential ofdissipation, we shall also define here a peeudo-pomuial of dissipation per unit a~ea o. relati¥e to the interface, defined by :
D,=-1-~
2rc, "
(6.3)
where "• has the dimension of a hydraulic permeability per unit length. The parameter "• characterizes the permeability of the interface,, and hence the interCOftftCCtion of the two porous media (iee Fig. 6.2).
J
'""
'--'
-
r l
i
.,.,..~ 't111'1t1VMB
8
2i47
Fipre6.2acorrespoadstoacueinwbidlaUdacbannelaaamllllicate(thl'illterfaceis said to be opcB) aDd "• is theNfore illfmite (K, • co, ao dissipation). Fipre 6.2c n:p~aents the cue of aeoaled iMerfacoaad K, 0, 10•tbat, u shown lator on, O(no filtration). Figure 6~ (0 < K, < co) com.ponds to an iatermecliate cue. Tbaa ill a borehole. the parameter K, can model the effect of a ''mudcake";-orof a tlooded zoae. if the latter is sufkitntly thin to be treated u a resion without thickness.
w.-
=
Open lnwflce
hrtillly open ..........
................ Medium 1
lllldium2 --._,
Cal
rzz::J sOlid
(bl ph-.
(cJ
c:JUquldpt._
Fi&. 6.l Simp~ cJil&rap ohn intorfacc botwee&. two porous media (after Deresiewic!:i and'Stllal; 1963). ·· ' · ,. ' ' , The variational formulation of Chapter 2 can then be resumed, addint the surface potential ·t() the velumettio 4illip.don poteDdaa dclft1led by &q. (4S.3~ · Hamilton's principle al'wAy& leads to-thnalile lieN eqMdont reiMed to tk medium concerned [i.e. uu- t111 Mld p • ~. w,.·"f'', ., - flf, I • l·or 2· m (2.87}]. FurthermoN. under conditio•(6.1)and·(6.l). dletoll...._ift1e8ral iiiMMIId IIIICUGelled:
'-
J.
'-
d.t
d.t - 0
(6.4)
~vc*the v~tiou W..-4Jw•. ~f;lwe.vanatioAaare~' ud the
portioG of dac, 4atc.rf8cc S is IUtJittarJ, iq. (6.4) daw:_. ...,....t41J ~ the nullity ol the two intcaraJa. Thus from tbe lint oae.: ~Jln1 •
o1J•n1
(6.S)
Condition (6.5) represents the 'continuity of the macroacopic stress vector at the interface crossing. This is the equivalent of:
~~-~ on a surface where the surface force ~ is. imposed. The W11 a - K,(pCll- p(ll)
-~
~~ ~
condition gives: (6.7)
Condition(6.7), in which iJp - ,C2l - p(l) represents the difference in pressure prevailing on either side of the interfac:c (or of a mudcake), is nothiq other than Darcy's law governing the fluid flow across the interface. It may therefore be noted that, for
'-;'
.
2118
6
WAVES AND INTERFACES
(open interfKC~·the fmitcness ef:tbe flow gives rise to a zero pressure difference (~p- pl 21 - rP 1 - 0). For "•- 0 (sealed interface). no flow is obtained (w. - 0). It is dear that "• depends notoaly on the eeotaet geometry between lbe two media; but must also depend oa the frequency (iloanbaum. 1974).. if the latter cornsponds to a wavdeagththat is not large in comparison with the dimeosioDJ of the elementary volume, but only in comparison with those of the. channels in wbicb the Oow takes place. In this case. permeability is no longer absolute but relative to the frequency, on which it depends (BioL 1956 b). The experimental determination of "•• for application to .actual cases (mudcalce for example), is nonetheless a delicate matter. Hence aU the theoretical investigations consider oaly the limiting cases "• "" 0 and "• = (Geerstma and Smit, 1961. Rosenbaum. 1974, Feq aiad Johnsoo, 1983-). . Equations (6.1), (6.2), (6•.S)and(6.7) are knee the so-called nat1,U'al boundary conditions that must be satislied at .tbo interface ·or two saturated porous media. It must be emphasized that, contrary to the inconect preaontation adopted by Deresiewicz and Slcalalc (1963). for a Hamiltonian fQnnuJation of the problem, o~ Cannot, independent of this formulation, introduc:x: the constitliti\'ic equation (6.7) govetning tbe ftow at the interface. In fact, this equation is a consequence of Hamilton's principle. To apply the principle, it is necessary to account for all dissipations, adding the potential of dissipation per unit area (6.3) to the volumetric poten&ials of dissipation. Giveri their clear physical signiftcance, one could obviously hav~ introduced (6..5) and postulated (6.7). However, a pan from the fact that this presentation woUld not bcin~t with the Hamiltonian presentation of the theory developed in Chapter 2, the procedure would also have obscured the so-called natural character of (6.5) and (6. 7), by conferring on them a more ·heuristic value. Two specifiC eases "iH be~ below. F~ eoaaiGor the case or a {fee surface. Equations (6.1) and (6.2) then . . . . . .r. Equatioa(66) iueplaced .by Eq. (4.6). where we ICIC j, == Ia t6.:7), P' 21 • 0 and bene~ p .. p'- 11 • 0, ~\IICtolal interc()niiiOetion exists (open interface, '"• •· cc ). Ia the cue :Of a. frce. •udae.. ;mC i*lCiitio• arc. tilus:
" • - JO
o.
tln=O Free surface { . 11 1 p==O
(6.8)
The second cue tonsidered ii that Of the ·reftection. at·tlle iBtetfaee of a pOrous 'Bledium (medium l)ilndofaftuidmedium{tliedhlm 2).1nthisc:aseEq. (6.1)cfisappean.Moreo~. since the fluid corresponds to unit porosity, the ftltration vector seen from thetluid side is reduced to \\1 21 = Ul 21 - fl1 1 ' where iq 2J is the average fluid velocity. Furthermore, the stresses in the fluid are reduced to a hydrostatic pressure, whereas (6.7) remains unchanged. Finally we obtain: ·
w = W. 11 n- = (UP1-
l
II
Interface Porous mediUm l~fluid2
l
I
o1] 1n1n1 = o1]"n1t1 == 0
w. =
l
p<21
......
il.ll)n. I I
(6.9)
- "•(p(l) - pill)
.
' ., ........
'=-'-'r ·~·
,
tl49
WA¥81 A.-'IMIIli'MIIS
...
~·
·~
'--·~
j
'---'
6
~ -~,
''--'
•v--"
·-.'---'
'---'
6.1.2 Wave reflectloas at the interface of two saturated porous ...U. In this Section it is assumed that the problems dealt with are two-dimensional. the interface is planar, the waves_. p~anecu • harmonic wa• aad that the incident wave travels from medium 1 to medium 2. Ia dMsic elutodyumic:s. an iDc:ideat P or Sl' wave reflected anbe interface of two clastic tolids paerally lives rise to four typea of wa'Ve, two transmitted and two nlected. P aDdS I' (for example. Ewiot a al., 1951). Tbc SH wa~ is of less interest because of the at.nce of convenioa. and will not be conaidered further. In poroelutic:ity, tbe famewcn of the study, the possibility exists of pneratina six types of wave, tbla. traaaitted aad three r•c:ctl
6.1.11 Case of .,..lladdeaee
"---'
Let us farst introduce the ratio Z of acoustic impedances correspondinl to the incident
~
wave: "--'
z _!.e.!!!
.__.
In (6.10). Vis the wave velocity for a closed meclium (i.e. no relative motion between the ftuid and the overall movement). With the notations from Chapter 2, it may be recalled
(6.10)
(pY).
-
~=
C''; l#J)' l
Incid.ont P1 wave: V • V,.., ~'--"
l
Incident $ wave:
V•
Ys • (~Y
(6.11)
~
(1) It is always possible to approaima&e ,vave frpD~ ~ ~y low curvature loc:aUy by tlMir taaaent planes. For any wave front. a couJ)Iitla OCICUfs ~'thf lriSftt pometry and the iaterface, and it is Uceual')' to employ 6Der theories such u til<* oh&)'S (Cerveny ~r Iii., 1917).
---~
'---'
·-·
...
250
6
WAVES AND INTERFACES
i
Let us also introduce a reference angular frequency (Chapter 2t defmed by:
al
.
41
= 2nfc =-;;;:
(6.12)
PpJf
where~
is the hydraulic p'Cnneability.lt ma;v be recalled (see Chapter 2) that we is usually very lar,e, thus allowing expensiODS with respect to •tw<. One can now determine the rdlection C()Cft'tcients R and traasmission coefficients T. These coefficients are dehned as the ratio of the amplitUde of the displacement corresponding to the wave a1Udyzed (reflected or transmitted) to that of the displacement correspondiq to the incident wave. Note that, coatnry to the dastodynamic; case, these coefficiems are complex because of aUIID\Iation processes dar to the two-phue character of the materials involved. tbis at:tataation ·leadiBa to differeaces of phase. The computations corresponding to the different typc5 of reflection were conducted by Geerstma and Smit (1961) and Deresiewicz and Rice ( 1964). We shall only comment on their results here. L
CIIU of- ltu:i*W S
""'
WCN
For an S wave at normal incidence, only the S waves are evidently transmitted and reflected. lbe coefftcients are then given by:
zz (. ~w ~w )z 2[ z2. .( w·. w )2 ·T··• -:--1 + Z I +.4(1+ zl' 7t al,_ Q7 R-
1- z [ T+Z 1 + (1 + Zf
y1
-
y2
exp (i tan
where }' = Pt
1
.-..,
8,.)
(6.13)
,.
2
.
1
.
for media l and 2
p
9.1 d
1
1
exp (i tan"" 8y)]
i2
'
_ ]
:.Zzi (¥• ;; -"yz ~)
(6.14)
(J))
(J) OT = -z- ( ¥2-1'1 --;-·
1+
z
~
(1)1
These equations correspond to an expansion to the nearest second order in wfw< [i.e. up to (w/wc) 2 ]. Note, to begin with, that, if w - 0, the classic reflection (1 - Z)/(1 + Z) and transmission 2/(1 + Z) cocfticimts are obtained: Moreover. they depend on w2 , thus differing from the same problem ·in viscoelasticity (see Section 6.2.3.2). The flow conditions at the interface, characterized by the parameter Ks. have no effect on reflection and transmission of the S waves. This is nonaal. ~¥cause shear waves induce no pressure and hence no flow at the interlace. The influence of Biot's mechanisms is only felt by the dissipation induced by the competitive effects of permeability and inertial forces, thus the only forces responsible for flow in the media. but not across the interface. Thus transmission is complete (T = 1) in the ftrst ord~r term of w d if Z = 1. In the second order term of wfw<, the order in which the two-phase character prevails for the case of S
~
""' .-..,,
--.., .-..,
·--.., ~
""
-.
~-,
'- ,i "---"
:..
2Sl
ft"'WI•·AD~
'
waves, it is also necessary that the quantity ,Pp1 /pof = :Kp}lp be continuous [substitution of(6.12) and(6.14) into (6.13)]. In fact, this quantity quantifies the foregoing competitive effects. These d'ects ·must .then act similarly in both media in order for the phase dift'crcncc between Ruid motion and macroscopic motion to be the same on either side of the interface, and for transmission to be complete.
b. CIIIC of a iltciMIII P1 w••c The tr&Mmisaion and rctlcction codlieieata corrcspOildina to aa W:idont P1 wavoand a transmitted or rcftected ~wave arc denoted yw and ~. Tbc sublaiptj is 1 fot waves of the r.m kind and 2 for dlc wave of the seconcl kind or slow wave. For rcftected and transmitted P w~ves, we !)btain : " .. . . .. ., . . 1
R'u •
.
fUl -
1-Z 1
+ z (1 + 60 ) exp (i tan - 1 80 )
,m 1- z) exp [- i tan-• (1~ - z8 )] +2Z•·(~1-8 1 0
0
(6.1S)
.
where (6.16) with
'[(1 - pm)rrryJ fJ l
. . m== fJM ·g= .1.1 + 21',
.
(6.17)
for media l and 2 and where fJ and M arc the coefftcients of 1Siot"$ theory (see Chapter 2. Section 2.2.2.1 ).:;· , ~• For reflected ud tralll-.itte&t 1\ waYCS: ·
.· R'1• • 91(1-ft ;D,)exp{i ~- 1 ·(1-lB:a- ~s>] (6.18)
TC 21 •
-
/
~ f 1(1 -92 + 8,-IJ•J eltp [t taft- (82 ·+ 84 )] 1
"'•
where
81 =
Jj(t- ~ Z
m1
m1
-
m2
80
Bz == (~ :-. 1 . ~~·2 ) 1 - z2 Bo 1 + Z rc.p 1 V,, m1 - m2 ~· . .
~ ~dico/aft -
.1 fla ~-:flit .. ·----.,zZ (lz oft
8
-
,~
.
(6.19)
2S2
6
WAVES AND INTERFACES
with
4-, t/131fPt p(p,fp- m)
for media. 1 and 2
(6.20}
In (6.20). q is a shape factor that depends on the geometry ottbe pore network. It from a more elaborate theory than the one developed in Chapter 2 (Biot, 1956 average value is Among other factors, it is related to tortuosity" by ah, acoefticient eharacceriziog thecapitlary CI'OSS•aectioa (Ia ~ 8 fot' ein:uJar JeCbOIII, h for infmite penny-shaped cracks).
r-
JS.
The foregoing equations correspOnd to an expansion ·with respect to (J)/(.1)~. expressions thus show that the influence of the two-pbUe dtaracter of the 1
concerned is felt to the first order in (Q)/0>1i for tb.e ftlflection of a P1 wave, as opposed to the case of the reflection of an SV wave, for which the influence is only felt at tbe frrst order in (.1)/0Je. This inftuence is mainly exerted through the contrast, of tbe quantity m defmed by (6.17). Ifthis quantity]J continuous (i.e. m1 - ,.2 ), Eq. (6.16) gives 80 = 0 and Eqs. (6.17) then show that the reflection coeft"tcicnts AC 1 ' and transmission coefticients TC1 1 are reduced to the classic cocfticicnts of elastodynamic:s (1 - Z)/(1 + Z) and 2/(1 + Z). This is understandable considerina tJle ~QCC Let us recall the defmition of the macroscopic stress (see Chapter 2):
or ....
t~,1 =
If we set
t~.,
A.1 tr e ~iJ + 2p eu- PMe~,1
(6.21)
= 0 in (6.21), oy being the direction ~ to the interface, we obtain:
tr ~ ==
mt
(6.22)
siru:c tr • is,. in this case. equal to ~· The quantity 1/m thus quantifiCS the share accounted for by the inaeue io water content in the apparent volumetric strain tr .... zerO 4W!JQ~ . . . .M:opie--' stuss (u., == 0). Ifm 1 = m2 , this share is the same on both sides of the interface. relative to the respective voiUJJleitric ttrain: Thus, for m1 • m2 , dae traJ1Jinissioa and Rlloction of the P1 wave of the first kind can take place with the same elastodynamic coeft"'cients, because of the absence of au atUitioltalsta;cu at tbc iater(ace with respect to the elastodynamic case. Simultaneously, note that in the absence ofan impedance contrast (i.e. Z = 1). the contrast of the quantity m leads to the generation of a reflected P1 wave dependent on .;;;;!Oi. Mechanisms of this type were identified in the analysis of the effect of attenuation in viscoelastic media (Bourbie, 1982, Bourbie ami Nar, 1984). This will be discussed in the second part of this Chapter.
e
We have just shown that the contrast Am= m 1 - m2 uerted a considerable influence on the reflection and tran~on coeftic:icntuelative to the P1 wave corresponding to the classic compressional wave in elastodynamics. Thus total transmission occurs for .1m = 0. However, even when m1 = m2 • slow waves er MCODd kind P2 waves are always generated. This is understandable because it is tho ~tual movement of the interface (and not the additional macroscopic stress, which is 'zero for m1 == m2 ) which generates slow waves in return. For m 1 = m2 , this effect nevertheless depends on w/O>c to the first order (factor 8182) whereas it depends on (w/w1 112 in the general
case. .
~-r;_ '---' '
~".;.
'-' ~
'
253'
WAVII . . DSI~·
It shoulcl ftnaUy be noted that the permeability .of tbe interface .rc, bu ao eft'ect (to the tirst onlcr in CIJ/of} ncept t1uuaJb the factor f 1 82 [aee (6.18)]. aad hence only for slow waYeL This is UDCientaa4able becaute it is always· the mac:t'OICOI'ic moW~DCtDt of tbe interface that can generate a fluid flow MI'OII:dlc iDtelface ill the ·ctirectioa Gppt)Sitc to the propaption direction of the iacident waws. Equations (<6.12) and (6.1S) to (6.19} pvc Dllults correspoadiag to a number of limit . cases. Let us therel'orc auume thatoeeofdle two media is impermeable. Tbis correspoads to Ulumina for this medium aa iBflnite ~ anautar frequency [.1'" -= 0 in (6.12)]. Now let u assume that the iDc:ideat waw canes from tlae permeable medium (i.e. crfz ..,. + GO). This does not aeacnto U}' aipifiamt~YC ChaJlF for tbe problem of the iacideat S wave.(aft- w.~ .... +eo (6..UO•. Fer th8 problem of an im:idcnt P waw from. a pcrmcabk naedium,. _Ecp. (6.1S) ta {6.1sa) show that, to tbe aearat second order in ...;;;;r;;i:
a
H«ll==l-Z
1 + z•
y
p; R«ll _ ~
; ..,...
(0 ( .d ...-..·up l+Z of
Jt)2
(6.23)
Is is clear that no P2 wa:ve is ~ in the impermeable medium (T' 21 - 0). The signiftcant cft'ect is that the two-pbue daancter is only exerted on the reflected P2 wave to the second order in and no lonsef to the f1rst order, as in the case of two permeable media. This is understandal\le in the tllht oltbe forqoina explanations, since, with respect to the elastic case, a macroscopic:.,..._. efthe iatetflleecuealy have an influence on the P2 wave aencrated i11 tile ~ '1iledium. The lesser sipifteance of the result is shown by a transfer to • ~ order -"'· ~· Now, if the wave CO~DCJ from the impermeable medium(~ • Of, oft .... -+ GO), the efl'ec:t is obviouSly reversed (R( 21 - 0 and
...;;;;r;;i
ym ~ on (J)/iff). Another interestinlspeciftecase is tbe one in which one of the two mec:f'1a is a'ftuid.ln the limitina cue ol a. fluid, the parameters introctuced into Biot'a theory (Chapter 2) are
·~
reduced to:
-·
ll ==0
no c:ouplina force. unit tortuosity unit porolity infuute permeability, no dissipation no shear
;.1 == M
fl-1
a •l
~ ... +ex> ..,-
These conditions have an evident physical sipifteance which was discussed in Chapter 2. In ~. an infinite hydraulic permeability correspoo4s to an inftnite absolute permeability rc (see Chapter 2. :II" • "'"· where If is the viscosity). The boundary conditions (6.9) correspond to a perfect OWd, IUdl that the viscosity is zero. ·We observe a certain paradox. becaue the porous medium is saturated by the same fluid, aad its viscosityistbesoaroeofnon-nealiliblediaipativee«ects(flnitepermeabitityoftbeporous medium). But in fact, this paradqx -~ - ""rWtly exist. The diSsipative elfects are actually due to the fad that fluid. Dow occu'n within a complicated pore network which
'-
,_
(6.~4)
•
'~ 254.
6
WAV£5 AND INTERFACES
~·
slows down its proBR*ion by giriag rise to shear stresses of viscous orilin at the liquid solid interi'ace. In tbecue of a luid medium. this interface does not exist, and the assumptiOA of a perfect luid can be ablined. because disaipatiOA within the Ruid is nqligible far from the walls lor I~ viscoJity Ruids. Substituting (6.24) into (6.1 5) to (6.19) to determine the case of the Ruid/porous medium interface is in fact irrelevant, becaual, accordins to (6.24), w" -+ 0 and the limited developments in w/aJ~ for tludluicl pbue are ao loftFr valid. The entire problem lllUst therefore be reconsidered, butitasedoaboulldary eonditions(6.9), to develop a function of wlaf whtn of is the characteristic aagutar frequency of the other pcnneabte medium. Geerstma and Smit (19tH) pafonaed the Corresponding computation. The qualitative results agree with the tJC110l'81· reaul&s, .ia other words the pneration of reflected and transmitted P1 waves with a firit order eotrective term in j(i{ai, the generation of a slow P2 wave in the permeable medium with a ftrst order coefftcient in j(i{ai.
Approximllte 1fU18IIitlllks oftlll ,_,.,.,._ We have used the experimental data of Wyllie et al. (1962), concerning three types of sandstone with widely differing pcnneabilities:
c.
porpsity t; = 29.7% ·permeability K = 1900 ml>
Teapot sandstone Berea
~ndstorie
porosity
f
=·
pel'JDel.~ty
i 9o/o K
= 200 mD
FoxAills saa4t~ ..·' porolity • ... 1.4% ~y K- 32.5 mD The saturating fluid is water. The~-- data a~ stun~ in the following table: wa~ <~turadaa ~) . . VISCOSity ...................•...•. ·...................... . Density ........................ ' ....................... . Sound velocity .....................•.....................
1 cP to* kl/arl lSOO m/s
~
GraiDS (silica) Density ...............................•.•.•••..•••....... Bulk modulus ............•.......•... ~ .........•......... Matrix Berea sandstone P wave velocity S wave velocity T~pot sandstone P wave velocity s wave velocit:r F oxhills sandstone P wave: velocity
2650 kg/m 3 3.19 x 1010 kg/m/s 2
................................ · · .. · · · · ....................................... .
3670 mjs 2170 mis
.. ' ....... ; ........ • .... · ... · · · · · · · · · · · ·
3048 m/s 186S m/s
. . • . . . • . . . • • . • . • • . . . • • . . . . ••.•••.. • .. · · ·
44SOm/s 2S1S m/s
...........,...•••....• ' ................. .
S wave velocity •.•••.. ~ ••••..•••••.•, ••..•••.••••.••••.•
Tortuosity is assumed .to be constant and ~ual to a == 3. . . ' '~
...
:;;r·~-
"---' '-....-
·~~~yft)'~OW..
~
255
We have considered two types of ititerfage~ (a) The interface between two porous media (two of the above tbree Ulldstones). (b) The interface between water and ono of the above .lbree sandstoDCI. ·
In the ftrst case. we consider an incident P wave (Fie. 6.3) and an incidentS wave (Fig. 6.4) coming from one of the two porous media. P1l
lA(1)
Par-medium1
tRC2l
JOl
'--·
yC11
~
rnecllurft 2
6r--------------------------, '-
4
1-<2)1
3
-'5
Prop.t•f2
I-J2
~-
,, 0
--~--'!.~!~_:..
···········-·········"" ..................................... -~
rl' . . . .
•••···•••••••
.
0
·~
F......,....CIIHzt
28
30
ofthe~lilileclioecoetftcleM
R'-dtbe 1\ wave ....._.toModulwa the elutk rel'ectioa ...-..(ot tbt P, wawt • a fuDctieD (cue an wave). Tbe curves identilted as
... U ''--
\0.
..,_lilftdatJFOldtills Uflllltt.
or frequency or incident P1 Me follows: porous medium (I)/porous II'IO!Itium (., The wave approaches from medium (1). Note that. for the reftec:tion and traDsmission <:Oeffic:ients 11< 11, ym and 2l, the quantities A.. A. ., , and are lower
T<
R
I
IT(2)1 T_.
than 10%. The symbol el stands for elastic: and meus that the reflection or tnnvri,;Oil coeffiCieats are~ at-. ia.-.betwelm two dutic media or that daae ooctT&<;icnts are QQIDPIUid.at ZeFO frequency.
'-
ln-tbe IICODCI caK, we consider an iacililent I' wave comiat from the Suid medium (water) (Fig. 6.S). To identify the differences betweea poro-clulielllobavior aa4 purely elastic behavior in terms of wave roflectioa and tru•i!lsioa. Fip. 6.3_to 6.S show. as a functioe, ofliequency (up to ~ut 30 kHz. i.e. frcqueQel used in aco~ wdl-loqiq expcrhDcnts): ' (a) The modulus of the mative .deviations of poro-'elastic Rfltction R", and
~
256
6
WAVE$ 4ND lii.IERFACES
transmissiQn T 1il coeft'tcients ofthe P1 wave(i = l)and the Swave(i = S)in relation to the elastic reflection (R.,) and transmission (T.,) coefticients (i.e. at zero frequency hence zero Biot effect) (Fags. 6.4 abd 6.S). (b) The modulus of the reflection R', and transmission "fC3tcoell'lcionts of the P2 wave related to the elastic reflection and transmission coafticients respectively of the P1 wave (Figs. 6.3 and 6.5).
s RfSI ...._medium1 ...._medium2
4r---------------------------~~ jR(S) _
RJ
3
laJ "2 ·0· 1
0
Fie· 6.4
~
- .1 'f ...... ----~~fioldlit ..," ............ :r.;-;r.~-
• 30
20 10 fN~a~lncvlkHz)
Modulus of the relative deYlation
IRfSlR~ l
$...,_
A.,
or the poro-elastic
reflection coeffacient in relation to the elastic reftection c:oe8'1Cient as a fuJM:tionoffreq~ofaa..._a
Tbe~ideQcif'IOd:u
follows: porous aaediual (1)/porout ....._ (2J.l"he waw apprcwfaea (rom Jlllldium (1). Note tJiat: jl"'t - T.dfiT..! < 1% We shall emphasize the following points: • Effects on P1 and S wawi are anater inldection thaD in transmiaion. In the case of an incident S wave (Fig. 6.4), the n6lctiot) deviations, although sli!)lt ( < SOlo). are always greater than the transmission deviations ( ronounted for the ·s wa"" than for the P1 wave. Thus the relath·e deviatien of the poro-elastlc' teftection coelftcients in relation to the elastic coeffiCients. although slight in the cased an incident S.wave (Fia- 6.4) ( ~ 4o/o at 30 kHz),
~.
~
•/ :uu ' •• );•' i•
?::,·:_~.if
J'J ,
...,....,........,..
yl21
·~----------------~~--~~~ jattl_ ...~"20
~
"-·''"
_____ ... _... .....1......_~=----.,__.. ..... ...,,............... :;..,...-"................-..
..
-·--
0
---:········~-·--···············-
10
20
•
Fnquency lkHit
(a)
•r---------------------------~ 10
trt''- yJ,. jTJ
\
Pro.,.'t '" 5 -~
-~"::-·--·~~
___ :. .. ,
,. , _ .... - - ---.:.,,,~-~-~............... ·····-···-····-·············--··········-··· 10
<•)
10 f~·Hzt .•
•
40
.--~-------------------,
!;.
~"(e)
,..,,•12 ~-~~-...... -------... ----~..-n. ........... --·-····~··-·--······-······-··················
to
io Frequency lkHzl
F1a-
6.! Modulus of the ftllative deviatiens of porCHiaStic mlection coeft'tcient IR'1l - R.d/IR..I (4) . . and transmiuion coeft'tcient OT111 - T.,) /IT.d) {It) or the P~ wave in relation to the elutic reftection and transmission coeft'tcieniS ~ively, and modulus or the por04laatic transmission coelftcient Tl 21 (e) or the p 2 wave related to the elastic transmisaion. ~ . . . fuactioA of rrequeacy (cale or an incident p 1 wave).
"--'
•
258
WAVES AND JSTERFACES
6
is greater than the same deviation in the case of an incident P1 wave (Fig. 6.3) ( ~ 1%) which is not shown. • As predicted. the effects are much greater if one of the two media is fluid (see scales in Fig. 6.5) than if both media are porous (see scale in Figs. 6.3 and 6.5). • The reflection Rf2 , and transmission T'·n coeff'tcients of the P2 wave are not at all negligible (Fig. 6.5c and also Fia- 6.3) compared with. the reflection and transmission coefficients of the P1 wave. Thus, in thecase of Teapot sandstone (fig. 6.5c), the ratio of the transmission coefficient T 121 of the P2 wave and the elastic transmission coefficient ofthe P1 wave reaches 30% at 30kHz in very permeable Teapot sandstone (K = 1900 mD). • Tho deviations are proportional to the square-Qf the frequency (Fig. 6.4) in the case of an incident S wave, aad. to the square root of the frequency (Figs. 6.3 and 6.5) in the case of an incident P wave. Moreover, the elfocts are stron,er with more permeable media. • It must be emphasized that dle orders of magnitude given above for the transmission and reflection coefficients concern an open interface. The effects are slight for a sealed interface (one order higher in w/af).
In the light of these results, it appears that if both media are porous and permeable, or if one is fluid, the P 2 wave reflection and ta,lnsmission coefficients are not at all negligible. The generation of reflected or transmitted P 2 waves participates in the total energy balance and contributes in a non-neJligible manner t~ alter the P 1 wave reflection coefficients. However, owing to the high attenuation of P 2 waVes in comparisoo with P 1 waves (see Chapter 2), direct observation oftb~former appears to be impossible. Hence the non-negligible signature of the P2 waves must be found on other types of wave: In conclusion, it should be noted that the main influence is exerted on reftection and not on transmission. and that it must themore appear .Jti the recordings. Th.e two media considered must nevertheless both be permeable and the interface open for a P2 wave to be generated with a noticeable effect:
6.1.2.2 Aaalysis of reflecdoa CNI die free surface of ll seml-inf•nite saturated porous medium We have selected this specific case to IUgblight another important property of reflection
processes in saturated porous media, namely the existence of inhomogeneous waves. The computations raise no theoretical difficulties, and it suffices to satisfy the boundary conditions (6.8) by superJ,esing tbe incident wave and the reflected waves necessary to carry them out. However, the computations remain cumbersome and we shall only stress the qualitative character of the results obtained by Deresiewicz and Rice (1962). A wave is defmed by its potential II such that :
II
= 110 exp i(wt -
k* • r)
(6.25)
where k* is the wave vector and r the position vector of the point considered. Due to attcnuatiQn JllCCbanisms,,k• is a ~plex v~r: k* • k - iA
(6.26)
• • 110 up (-A. r) up [i(an- k. rt]
(6.27)
which gives for (6.25):
-,
~
~-
-T
I
·~
I
6
2S9
~AV!SANDI~
The norms of vector A and k arc written:
I
tAl== A
(6.28)
fkl- k
where A, inverse of a Jenath. is an attenuation, whereas k is the wave number such that: kV = ro (6.29) where V is the wave velocity. A wave is said to be inhomogeneous if the \"ectors k and A arc not colincar. In other words, the planes of equal amplitude (A • r = constant) arc not parallel to the planes of equal phase (k • r - constant). These inhomogeneous waves arc encountered in another coatext '**ly comparable to the one examined here, that of the reflection at the interface of two viiCOelastic media (Bourbie, 1982, BourbiC and GonzalezSerrano, 1983). This problem wiU be dealt with thoroughly in the next part of this Chapter, which can be referred to for a more detailed analysis of the inhomopncous character of the waves. The most important qualitative results for the problem examined here are the following. In the case of an incident P1 or SY wave, the charactcristM:s efrcftected waves of the same type as the incident P1 and SY wave dift'er relativfiy IJiPtly from the same waves rcftected in elastodynamics. They are homoaencous, sliahtlY dispersive and dissipative, and this dissipation depends on w/of to the lint order. By contrast, the reflected P2 wave is highly 1
dispersive and depends on (ro/of}¥io the fu'st order. ltsattenlMtion is proportional to rofw• l .
.
in a dircctioR panllel to tbe free surface, and to (w/of)-l' in the normal direction. Deresiewicz aftd 6.'e (lf62) auaamcaD)' analyzed the expcrimtwltal da1a ofFatt (1959) correspondina to a kerosene-saturated ~."tortuosity was assumed to be 1.01. We Jive here the results~ to an iDddent 1'1 wave 4:w wbidl.thc Biot effects arc particularly important, especially at low anatcs. of illCidcace (Glose to DOtlD&l incidence). This is due to the polarization of the wave involvina fluid phase movements pcrpendic~ to the free surf.,.,. For an incident P1 wave, the variations of phase velocity normalized by velocity Vr as a function offrcqucncy are in the neiabborhood of 10-" and 10- 2 respectively for reflected P1 and SV waves. For the retlected P2 wave, these.variations arc about to-• (see Fia. 6.6). Apin note the sliaht difereDces between the dispersive eharactcristic:s of the inhomoaeaeous waves (see F"tJ. 6.6, solid curves) and the dispersi:re characteristics of the homogentous 'WtWeS (dotted curves in the same ftgarc), analyzed in Chapter 2, and propaptina in the same porous medium (kerosene-saturated Sandstone), albeit infmite, namely in the absence or diJcontinuities. The attenuations arc much hi8her in the direction perpendicular to the free surface thart in the parallel direction. This was forelceablc, sinc:e. durina these .reflections, the Biot effects occur, especially for flows normal to the discontinuities. It seems clear that, during reflection, the free surface condition will considerably favor the differential movement of the liquid phase in relation to the solid phase, essentially in a normal direction to thC surface. The difl'erential m~ in tbilntiNc:tionwill theft be the main souroc of energy dissipation. For the rcftectcd P1 wave, attenuation characterbed by the ratio A. k is too low to be signiftcant, this ratio being about to-•. For the reflected SVwave, this ratio is about 10- 2•
'---
•• 1.()()04
.3 Yp __:1..2
v,
-.::..1 Yp UXI02
lnhofncla-neaus-
Vp
'
0
1.0000
I
s«:
I
t
0
r
3
Vt/fc
3
2
vm;
2
0
I
I
.I
:::.. Yp .49
.48
0
1
vm;
2
3
· Fia- 6.6 P . ve1ocily of il'l~ r~ P1 wave (Y,..~ P2
wa~ (V,.2 ) and S V wave (J'.)(solid lines) compared with the phase veloCities of the saae types of. wava(P1 , P2 ancl·SY)ill the eae olullo........- IDII)de.of propaption .ia aa iaMR. I&'IJ8dium (dotted ·~ 'l'boee ~ are
~ .~· tJac vcloatf of '• ~vca (DO dissipatiOil). ~.abtcissa is the square root ol the bquenc:y.~ to the~ hquency fc (cue of aft incident 1'1 wate). 111e iatilo VJ.!Yr is ta'bll eqlilld to 0.483 (noadiMipati-ve c:ue) (ilftei: DeaMwicz aad .Rice, 1~ .
~ :t \ 2L
l 0
0
I ""'
r,-o~ 1
I
=
;
•..~.
t__.. ...
~
2
~
Fit- 6.7 Normal, ceaaponeot of thc atten..UOn Vfllilt"' for the ..aow wave normalized to the norm of the propagation vector as a func:tion of ,1flfc (incidence P1. wave) (after ~ and Rice. 1962).
.----..,
~
-~...___
-
~-·
6
261
WAVES AND IHTUPACES
30
r I -
r3 p1 p2
I
I
0
30
10
r,
sv
10
·~--------------------------~ flo.U
20
\fiif:.,: ·4t
r2
'
·.
' .. ' . .
-
Vffc· .1
~or~c,or <•
thAI Pa wave and{3 of
SV wave
vs. anateof~ { 1 ohbe P1
1Rve 1962).
Deresiewicz and Rice, .
ViTfc. 0
8
30
10
to
r,
The intcrestiq qualitative. aspec:t is that, at the
.alnc frequenq, it rises with ana1c of
iDc:icJeDce (defmed in relation to - normal to cfie free surface). This results from the peiMilatiea of perticle 1DotioD ·J*pCDdicular -~ tbc propaption direction. For the rcflccted P2 wave, the normal component of attenuation is very hip (see Fia. 6.7) b1U is pradically independent of ansJ.e of incidence.·
or
or
Fipn 6.8 shows the variations anales reflection { 2 and { 3 of reflected P2 and SV waves as a function of anpc of iaddence C1 of the P1 wave. It is clear that the angle of reflcction ef the P1 wave is tbe .... as tUt of the iac;idcBt waw. Moreover, DQte that the variation of ar&ale of nflec&ioa ', .. a ru.:tien of antic of incidence differs from the clanical SaeiJ.Descartes law of er.t~: This eift'ect 1Mcomcs more pronounced with increasing deviation from· aiOimaJ incidence. The anJies of reflection { 2 and increuc with the ratio . the ratio with which the effects due to the two-phase character of the studied- media increase. This can be explained by the dispersion of the reflected waves which, conversely, impoee an increase in the anpe of reftection as a function of frequency at faxed analc of incidence, to satisfy the Snell-Descartes laws.
'1
.;c;r;;,
c,
1.0
-l
I
182 0 (I)
I
IR1>/ 0.2
...
/
.........
"'
L
I
I
e,
I 180
178 90
f1
30
0
80
30
0
0 90
80
r1 0.06
0.04
IR2'!
r
I
fife• 1.11
.~
0.021
.........
I
.:t------90
A .1
of
. a:t<:: 30
0
., 80
r,
o.e
0.4
~
m,:m
~
'
-
··~-_::I o____ ~
---
30
0 .·!·
"'··
80
90
r1
'""'"~·.-.
180
jacslj
113
171
l---.
tl
I
Z1.a-=-
..
30
80-
I
I I
'. I
I 90
r1 30
80
r1
Fla. 6.9 Moduli fRt"' and phde 11of reflec:tiot1 ~ts (in potential) of P1 (i • 1), P1 .(i• 2) and SV (i • l ar S) waves u a function of anJle of incidence ' ' (~ 1'1 wave). ~~Yel ~e in4exod by the nonnaliad frequency f [. (after Deresiewicz. qd Rice, 1962).
.,
-----
"---'~
r I
-~f '
I
I
'WA\U'Aktr~Ad!s
6
Figure 6.9 shows the
,-ariation of moduH and phases of refteC:tion coetftcients RU1 U= l(P1). 2(P1), 3(SY)) vs. the anJle of incidence { 1 • Here the cOefticicnt All' is ctef'tned from tbe potentials related to the wave studied by:
T
-. ....
-I
-,
I
::r ~
..-
'-
...,-
_1
:1
.J
263
j
= l, 2, 3
Ru1 =
i; = IR
(6.30)
1)
where ()0 is the incident wave amplitude. Note that the reflection coefficients corresponding to the P1 and SV waves differ slightly from those calculated in elastodynamics (i.e. w/ol = 0 in the curves). The only signiftcant difference is a wider phase d~rence whidl is independent of the angle of incidence. as compared with the elastic cast for the SV wave. This increase (in absolute value) results essentially from dissipative effects which slow -down particle motion. Moreover, as above, for a ftxed angle of incidence { 1 , the more the conditions for twophase effects are favored (rising w/Cif), the larger amplitude P2 wave is generated at the free surface, which was foreseeable. In fact, coetTICient IJIC 2 ~ increases with w/ol' for a ftxed angle of incidence and by contrast, the reflection coetTtcicnt IJIC 1 ~ corresponding to the P1 wave decreases to satisfy the enersy balance. The SV wave reflection coeft"tcient IA«511 displays slightly more complex l;)ehavio~, which we can S11DUD&rize as follows: At high angles of inc:idence (about > 650), and due to its polarization, the SV wave sets partides in motion in the solid phase. esscntialy in a direction perpendicular to the free surface, while the liquid phase rerbains practiallty at rest, since it does not transmit shear forces. This has the eft"cct of generating difrerent'-1 movements between the solid phase and the fluid, which induce two-phase effects, fuoring the generation of P2 waves and consequently the attenuation of SV waves. As ptedicted, the effect is intensified with rising frequency. At angles of incidence close to no~l, the mechanism described above is no longer effective for the s1· wave, but for the I\ wave. To summarize, an incident P (or SY) wave at the-interface between two porous media gives rise to three types of reflected wave (P1 , ]\,and SY) and to three types of transmi_tted wave (P1 , P2 and SY). 1bt wavtlitt the iatcrfacle: ..........., ildaolnOI'ftCO\JS (planes of constant phase diltmct from planes iJr CODitUt'amJiitude) except c:asc of the refJccted wave, which is of the same type as the incident wave. Moreover, the Biot eft'ccts on reflections and transtnissions are more pronounced at aqles of incidence approaching normal, and hence ~ no tcmlt; be ipored. -
in-.
6.1.3 laterfaee probltllll ltetwee11 porous ua.atetl metlia Applicadoa to aeOUitic ......... 6.1.3.1
Jaterface ·wues
•• ll•ylft6h ··~~·
In classic elastodyrWnics at the f~ surface of a _._infmite .,lid, a so-callocl surface wave can propagate, namely the Rayleijh wave (see for example Lord Rayleigh, 1885 or Viktorov, 1967). This is a non-dispersiye inllemopneous wave whose amplitude decays exponentially with depth. Particle motion O<:f\-1" flong two transverse components with a phase difference or 7r./2 and contained in the Ulittal plane (plane deftned by the wave
'--
264
6
W.\VU.AND JNTERFACES
vector and the nonn,alto the$~). The ~~t~tmnity ofthe polari,zation vcctor.describcs an ellipse in the direct.ioa retfoar• near the~ aad PfO&rade in 4cpth. The horizontal component is cancelled at depth Q.2 118 , where Aa is the associated wavelcagth (sec Fig. 6.10). MJJJJJJJJJJ JJJJJJJJJJ
JJ)JJW)j)))))))
~
JC
Prclpegition
~~
nnn JJJj ns;nn;j;;ilnn ;;;;; ;;;;;; Jh;sJ; --... x
l
ftrOIMIJition
'
z
AW2
lbl
'
f'll, 6;.. DistriHtioo.otveklfity fleW (a) ud 4i~ <')of particles ol Ill isotropic scmHafmi!e medium at the passage of a llayleiah WJVe (after A\lld, 1973).
.
..
We shall now examine what happens to this Raylciab wave if the propaptioo medium in Fig. 6.10, for the same geometry, is a saturated porous medium described by Biot's theory. Let us consider potentials of the form:
== t;1 ~ (- r1z) exp [i(kJx- a>t)] j == 1, 2, 3 (6.31) Each potential defmes a P1 U == 1), P2 U a; 2)aad S{i • 3)waveasindic:atedin Chapter 2. ~J
To identify a surface wave of the Rayleigh type consists of determining whether the boundary conditions (6.8) can be satisfted by a superposition of waves dtfmed by potentials of type (6.31). If this is possible, the real part Re(kl) and the imaginary part Im (kl) in (6.31) respectively defme a Rayleigh wave phase velocity V8 and a corresponding attenuation ~. in the ox directioa ~ (()
~e (kl) ""' v.(w) 1m 1kl) == «a(a>)
(6.32)
''
WAWiAiM~Aa!s
6
285
The general eafculatiolll oorrespondins ~ dais procedure were performed by Deresiewicz(1962). Based on these c:alculatiolll, oneoan clemoBstrate that an expansion in w/of yields the equations which must be satisfied by Va and «.a in the form: 1
2
1
1
../i ,.2q, v. (•- v!)1(w)1 _4 (•-~)2( v~)l- (2 _ v~) 1 _ Y,. g V,. V of Ys . V 5
5
(
6.33>
vl t - 2 V2) - .fi.«a {(3 vivi V2vi- 2(V2vi.,. . 1)} ==
_(w)
l
2
of
~'-<
'----'
"---
''-.--"
'---'
·~·
~ ,.~1-(t- v~)·(•- ~)(2- ~) v,. Va
g
V5
Y,.
V,.
V5
(6.34)
Rayleigh's classic equatioll(wbere ro/af • O)can be RJCOpized in the second member of Eq. (6.33). The ftrst member, where m and (J ~ de(ined in (6.17), corresponds to the twophase effect. This ftrst member depends on theJrequcncy. and heDce, contrary to classic elastodynamics, Rayleigh waves in saturated porous media are dispersive. Figure 6.11 shows the Rayleigh velocity as a function of frequency, and also attenuation, or, more precisely, 1000/Qa where Qa is the quality factor derived from (6.34). The ratio of the bulk modulus K 11 of the fluid (water) to the bulk modulus K. of the solid constituting . the matrix (silica) is kept constant and equal to O.OS9(- K1 ,JKJ:. This value is the same as for a water-saturated Berea saadstoac(Wyllie «&.1962). Asia Claapter 2, the ratio of the bulk modulus of the open system to the bulk modulus of the solid has also been varied (i.e. K 0 /K.- 0.2, 0.4 and 0.6). The broken line, solid line aad dotted line curves correspond respectively to the pGft)lilies t; ... 30, 20 and. 10%. The other parameters involved are derived from Gassmann's equation •d Berrynia11's equation for tortuosity [sec Eq. (2.92) in Chapter 2]. It may be observed ia Fig. 6.11a that the velocity fint decreases with rising frequency, and that, at a pvea frequency, it increases witb the ratio K 0 /K, . .This is easy to understand: liace the surface is free~ the pressure p ia ..., fluid is :lAII'O on this surface and a depth gradient op/o: is created, c:ausina differential m0¥011leftt between .the ftuid and solid. The movement correspondina te the·~ wave iS obtaiaed by the superposition of three potetltials(6.31),eacbc.:oi'reSpondina to a P1 waw,an Swave and a slow P2 wave. As sboWD ia Chapter 2, however, the P1 and S waves have veloc:itiel which rise very slowly with frequency, whereas tbo P2 wave has a vetodty wbicla inCreases rapidly (from zero at :lAII'O frequency). Hence as the frequency rises, the contribution of the P2 wave dominates. Since the velocity of this wave is much lower than Va(O) (the velocity of Rayleigh waves at zero frequeacy), the velec:ity oldie Jlay,lei&h WM1J4ooroa.scs-' the frequency rises. This initial dec:nue is mdeady leis pronounced if K.ofK.. approacba 1, because this causes a less pronounced difFerential movement of tho ·ftukl. At ·hilber frequencies, the Rayleigh, wave velocity increues as a function of frequency. This result agrees with the ftndinp of Dercsicwicz, 1962). This is explained by the fact that, at the higher frequencies, the P2 wavei are too attenuated to be sipifJCant. In this case, the increasing effects of the P1 and S wave velocities prevail. The curves corresponding to 1000/Q11 (Fig. 6.1lb) vary in the oppotite way to the curw:s correspondia; to the velocities (i.e. inereasing at low frequencies, decleasina at hilb &equencies for frequencies 110t plotted in Fig. 6.11 but studied by Deresiewi<2, 1962). This is ~rtnal. becaUte the more significant the P2 wave amplitude the jreater its attenuati~ and "'" wrsa.
266
6
WAVES AND INTERFACES
,.,......, 1.012
.. _,' .
~,
'"''
/
10
,.,.....
20 30
f..·~
"'"" .......·:;/1 ... ··
/
- - ...... -:...: ... ··:·-:;;:..·,""
VR
Vo
7
.4
~
1.000
I
········
.2 .. ""
,t;"
-- , ...
.1198
.4
.2
......... ....... ' _,______ . --..... -··· ..~.2
~
Cal
~--···-~ '
~
.oa
0
.08-
.•
.12
'Ks
l
.15
fife
I i
20
~ity
...... 15
-
""'' JO
~.
~---~~-L . . ~...
!...9!!!
/
OR
/
10
,./ -~ ,
(
.2/ ., ,.,
, ..· "' .. ··
'
.... ······
.6
(
········.4 ·······:·····
( I
5 (b)
0 .0
.03
.08 fife
••
.12"
l
.15
F'ia· 6.11 Phase velocity 'fa) aad ittvreild ctuality factor<') of the Rayleigh
wave. Tbe phase Velocity is~ to the phase velocity in the absence of dissipation (i.e. at zero n.~). the frequency is. normalized to· the
characteristic frequency fc. The curves arc drawnfor varyina ratio Ko!K•.
b. Stouky
WIIHS
In classic elastodynamM:s, at the iate~ of two. elastic half-spaces, a aeneralizcd Rayleigh wave can propaaate, wh~ amplitude c,iecreases ~xpooentially with distance from the interface. This wave. identified by Stooeley. does not-always exist. Its existence was investigated by several authors, and particularly by Cagniard (1962). The reader can ~
'""'
'-'-·
'-
t/Jfhl$ ANfi~Aeil
6
261
senerat
'-
'-
refer to MikloWitz (1978) for a diiCUuion. For the ease ofa solid/fluid interface, the Stoneley wa~ always eXists. . Now let us consider the case in'wbich one of the two mtdia is a perfect fluid and the second a saturated porous medium. To identify the existence of an interf8ice wave of the Stoneley type consists in determining whether the bOUndary conditions (6.9) can be satisfled by a superposition ()f waves defined by potentiak similar to (6.31) [thn~e potentials (P1 , P2 and S waves) for the porous medium, and one potential (P wave) for the fluid medium]. "fhesecalculations raise no difl'reulties but are tedious, and the reader can refer to Rasolofosaon and Coussy (1985). Figure 6.12 shows the variation of velocity V5, of the Stoneley waves and the ratio 1000/Q51 where Qs, is the ctuality factor, as a function of frequency for different values of
K 0 /K•.
1.CIOit'' . YSt
~ Open
1.ooi0
Vo
.til (a) 0
.01
AMI
.12
.()8
fife :q~:,
'-
.a
_....,_ ::ru~l .
'!Sa.
Ost
. .3•KofK.
---. . --------L-?tttz=--=--····--··· .....
_.... -~-
..
..
.oe
fife
Fie- 6.ll Phase
.
a;;-:;.t J (It)
.12
velocity (a) and invene,cpility factor (lt).of the. StoadcY .· wave. The phase velodt¥ is •onu.lizecl tp the pbase velocity in the a~ of ~ip&tio!l.
'
268
6
WAVES AN~ INTERFACES
The PQrosity is assumed here to.~ 19o/.. and tortuosity is 3 (see Berryman's formula. Eq. (2.92) in Chapter 2). The ratio of the bulk modulus K I«. of t)lc Ouid (water) to the bulk modulus K. of the solid constituting the matrix (silic.~t.) ~ kept constant and equal to K 11 /K, = 0.059. The solid curves comspolld to ~e case of tbc open Ouid/solid interface (i.e. JC1 = ao) and the brokeo.liQCS to the case of the scaled or the closed interface (i.e. "• - 0). Like tbc R,aylciah wave in the case of porous media. the Stoneley wave is now dispersive. For a better unders&andjng of the curves obtained. Fig. 6.13 shows schematically the geometric representation of the increase in the ratio K 0 / K, at constant porosity. Poroua medium
fluid
Poruulllllllium
Fluid
Fluid-=· Fluid__.
Fluid
Fluid
KJKs "\.1
lowKJK,
Fi&. 6.13 Schematic diagram of a Ouiclfconstant porosity porous medium interface. Note the influence of K 0 /K. Oft 40 pore sh'e". ·
-,,
To presume increastDJ K 0 K •• at constant t/J and hence at co~tant saturated connected pore volume, corresponds in r.ct to tic ~x.i:stence of illCldsinglyJme pore channels (see Chapter 2). Let us ftrst consider the case efthc scaled-mterface. f"br this sealed interface, no fluid exchange occurs between the porOU& medium and the fluid medium. Hence there is no contribution from the P2 wave in the porous medium, because of the absence of any possibility of movement out of phase between the overall system and the fluid part. Accordingly, the fwo-phase effect will only be exerted througb 1'1 and S waves. Since the velocity of these waves increases with frequency (see Chapter 2). the velocity of the Stoneley waves will'also increase as a functioll of ftequency. These dift'erential inertial effects, increasing with frequenq, will incrnse IB~dilference between the motion of the fluid and the overall system. This differential movement is respollsible for attenuation phenomena and therefore implies that l088LQ61 increases as a fUnction of frequency. At a given frequency, these effects Will obvioutly be less pronounced for a rising ratio K 0 /K•• since, in this case, dift'erential movement will occut tt.ith lesser facility. Let us now consider the case of the open interface. tn this case, fluid exchange occurs between the porous medium and the fluid medium. In the porous medium, this fluid exchange allows for a contribution from tile P2 wave, for which the Ouid movement and the overall movement are out of phase. Thus the case of the open interface for the Stoneley wave leads to the same phenomena as for the Rayleigh wave, which explains the variation in velocities: an initial decrease, followed by an increase as a function of frequency. Signif1caot difFerences rtevertheless exist. In fact, the pressure gradient prevailina in the ftuicJ at the interfa~ is l(,)wer for smaller ratios K0 /K •. For a suft'1ciently SQ1all KoiK•• the channels arel10t so. thin (riC- 6.13a) and
•
..-....,
'
---~ '---' "-./
'-
''-' '----'
'-"-./
'-
'---
"-
"-----~
'----
-~
"---'
'----
WA'WIAHB~
'
:!69
the reduction in cross-section throush which the fiUidftiJws toWards the porous medium is leis drastic. Tbe- pteiiUI'e 8f'8dield is tbenlf'ore llal'fOWW, Iivia~ rile to a sanaDer Ouid cxchaftae between the two mediL The contribution of the-~\ wave, as well as the dilsipation d• to Row across the iaterface, is ......... ICCOfCiiatly. 1'llis aplaiDt why, in contrast to Rayleip waves, the velocity eurveaconapondinsto K 0 /K, • 0.3 lie above those corrapondins to K 0 K, - 0.6 and 0.9, and why, at low frequencies. attenuation (1000/QSr) is less for K0 /K, 0.3 than for hisher Ke/K, ratios. At hip fn:queocies, however, with increasins K 0 / K,, attenutiOn decNases because the attenuation due to P1 and S waves dominates in this cue. The velocity curve correspondins to K 0 / K, • 0.91ies above that correspondin& to Ko/K, • 0.6. because the channels become exceaively thin, and the pressure sradient prevailins at the interface is inadequate to auarantee sipiftcant flow as K 0 /K, approaches l.
=
c. Slulutulrl As a general conclusion concernina interface waves, we can state that the two-phase cbaracter of saturated porous aaoclia ~Mba 1M RaJleilh and Stoneley waves dispersi\'e. However, the effects can be iporecl fer the ltayleiala wve, both for Yeloeities and for attenuation (see scales in Fia. 6.11). The same applies to the velocities for Stoneley waves. On the other band, the attenuation ofStonelcy waves due to the tv.·o-pbase character is no longer neJlilible in the case of an open interrace. At low frequencies, in fact, an open interface leads to Ouid excbanacs IJetween the media, and hence a sipiftcant contribution of the Pz wave, correspondint to out-of-phase fluid/matrix movements. Dissipation and attenuation then reach a maximulllat a liven -frequellcy for intermediate values of K 0 / K, (K 0 /K, • 0.6). allowing large pressure ~tadiea:ts anCf SGbstantial fluid exchanges. Inertia effects become domiaant at IBP fN\quencies, and ateteetl on P, and S waves. 'fbeseefl'ects are also aecentuated by the open dtaracter of the. ietotface. For applk:ations to wen loainr. it is iftttrestial-to aulya the claaqe cauled by the two-phase character of the medium in comparison with the elastodynamic framework, wbea a seismic source is plaeed mtt.e·Yicitdty of a'I\IM/penlable rneclit.tm interface. We shall examine this problem in the next Section. ' -
6.1.11 .............. wJcilllty ., 11 fl:dtlfOIMI - · - .Wterface We have just examined the aeneraJ properties of sutface waves (ltayJeiab and Stoneley) propaptina at the pfaae surface or a semi-inf'mite saturated porous medium, and at tbC plue interface betweea a fluid and a saturated porous medium. These waves were a~ iadependently, ill other Words 1rithout poltUiatiDa any hypothesis on the source generatina them. We shaD now examine problems correspoJl'dinS to sources immetsed in a Ouid, near a petmeable interface. We shaD examine the thanacs aenerated by the permeable porous media in comparison with the classic elastic media. We shall see that the two-phase eft'ects alter the waves propaptinJ at the iatedace. irrespective of the cylindrical or plane geometry. The study wiD be carried out with the aim of fmdint a suitable method for determinint the permeability of porous media, tbroup its sipature in the ,enerated waves (i.e. modification of the recorded signal). We shaD f1nt describe an analytical method of resolution of the problem, and then analyze the ~results of sewral numerical simulations.
270
'
WAVI;SANP lNTEI\FACES
a. Alflllytical resolutitM qJ tlw ,...,_,
~
The problem is defmed by the pollletry of the propqating media and by the source characteristics. The geometries aaalyr.ed corrapond. for example. to the follo•ing problems: a plaae interfacx betwcea a 8uid and a saturated porous medium t Fig. 6.14._ a layer of fluid of constant thickness lyina between .two identical saturated porous media (Fig. 6.1 5), or a circular borehole filled with Ouid in a surrounding formation consisting of a saturated porous medium (Fig. 6.16). The sources are defmed by their aeometrical characteristics (point source. line source or cylindrical source) and the time sipal omitted (Dirac delta function, Ricker wavelet). All these problems can be dealt with in the same way. We propose below an analytical resolution algorithm. ~
Choice of a system of coordiDteL For example, cylindrical coordinates in Figs. 6.14 and 6.16 and rectangular (2 D) coordinates in Fig. 6.15 for the case of a lint source.
Defulitioll ef potelltial fuc:tioM. Tlaese potential functions serve to completely describe the vibration state in each of the propapting media: (a) In the fluid, assumed to be perfect and incompressible, and where the source is immersed two scalar potentials must be defined, one corresponding to the solution obtained without any soura; and the secoi)d coticsponding to the superposition of a source. (b) In the porous GJ.edWm. it iua~~ry to dell~ three potentials, two scalar ( 41 1 and 4Jz) and one vectorial 'I' ~5ec Ch~ 2~ Due to the spccif1c symmetries. the potential 'I' actually has OAly one aon-zero eoptpOnent, the orthoradial component in the case of cylindrical,aymmctry (FiJ. 6.1~ aad the normal component to the plane of the figures in the bidimcJWonal case (Fip. 6.14 and 6.15). ~doaofdlecrlltidoala...;....,._...,_• • . _
and the porous medium (Section 6.1.1 ),
"
The complete resolution of the problem hence amounts to the determination of all these potentialfuactions.lt is very coammietlttouse mecbods based on Fourier-type transforms (Bracewell,1978, for example), in die tilDe and spac:edomainin the two-dimensional case, or of the Hankel type (Oitkinc and Proudnikov, 1978. and Bracewell. 1978, for example) in the space dom~n in thC three-c:timeJlSional case \\
,...
..,t -
v12
~
'il't
u d). 4S(r) • == fi(t,.,(z iJ 11
1
where /(t) VI
= source fu~tion. = wave velocity in t~ fluid,
nr
for i
= 1,2
(6.35)
~"
-~
~
__
,
"\
z
-. P!nt:l
Wft!I*JI-.IOof
IJII/J/Il/1111//J!IIJII//f«U
..... --1
------
:
s
z
P!ftl:l
p
z 9
272
IS(z- d)
WAVES AND lNTERFA(;ES
6
;_r> =Dirac disrbution product and, ' {1 = Kronecke s delta 0
6 11
ifi=l if i :#= 1.
Equation (6.35) ex that the potential fl 1 destftbes the vibratory state in the infmite fluid in the presen of a so...rce (second member of (6.35) non-zero) and the potential fl 2 corresponds t the ··reflected" waves (second member= 0). After the time Fourier tra sform (t -+ w) and then the zeroth order Hankel transform (r -+ k) of Eq. (6.35), ~ utions of this equation are eafily put into the form: •
k 1
fl2
where
h} = k2
-
'
) _ F(w) exp (- h1 1z- dl) hI
z, w - 2w2
k, z, w) = fl2 0 (k, w) exp (- h1z)
(6.36)
1· ··
w2 V2 with Re(h1 > 0,
~notes
fl1(k, z, w) the ouble transform (Fourier and Hankel) of the potential z, t), 11>20 (/c, w) is an arbitrary runction ot k and w which are the conjugate variables of r and t in the two foregoing transformations.
fl1(r,
The necessary uniform behavior oflthe poklnlitl1s at infmil)' (radiation condition) justifies the condition Re(h1 ) > 0 and the rejection of the solution fl 2 in exp ( + h1 :). The complete resolution of the problem amounts to the determination of the unknown function fl 20 (k, w) which allows the determination of potential 11>2 related to the fluid, and the determination of potentials •., •~ aad 'I' related to the porous medium. These unknown functions are computed t.y bbfe.Vansforming the equations and determining the boundary COnctitfou to be;atisfied at the m.terfaces. These conditions are expressed in the following matrix fow: [M](X) '"" [}f)
(6.37)
_where • [X] is the unknown column matri'lt' whose elements are unknown transformed functions [for example fl 20 (k, w)] to be dttermined. • [Y] is a column matrix of the same dimension as X, whose elements depend exclusively on the source spectrum (freq\acncy and wave number). • [M] is a square matrix depending exclusively on the propagating media (geometrical and mechanical characteristics) and totaUy independent of the characteristics of the source.
"
In the absence of a source, the matrix [Y:J is zero and syatem (6.37) is then reduced to: [M][X) = [0] (6.38) where [0] is tbe column matrix composed of demnts that are all zero.
"
I
i
_j
...._,
"
, - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - ----------·- ---
T .._,''---'
,_ :~
'---'
m
ftJmfftiij~'
'
In thisQSC, to deviate from the trivial sOlution {X] - O,correspoildiDJ to all potentials equal to 0. or to a si*e at rat, it is llecellary tocancdthecletermiuilt ofsystem (6.31)or: dct [AI) - 0 (6.39) Equation (6.39) actually constitutes the velocity dispersion relation ohhe system, which
serves to dctcnnine the characteristic propaption modes of w,vcs corresponding to the pomctry analyzed. Equations of this type were discussed in Section 6.1.3.1, enabling us to
~
describe the dispersion and attenuation functions of the Rayleigh and Stoneley waves. These characteristic modes are obviously independent of the source. In the presence of sources, the sotutions to system (6.35) are given by: '---'
X
det (WO] , - dct [M]
(6.40)
where
is the transform of the fla unknown potential. • [~ is a square matrix obtained ltrnplacina the ,.. column by the column vector (Y].
• X1
'----
Since the unknown transformed functions are all perfectly determined by Eq. (6.40), the vibration of the space is determiDed by takina tbeir inverse transforms. Thus, for example, ia the cue C'll F'll- 6.14, tbe.,..... dillribUtioll iac6eeuid is determiDed .. fDUows: By clebition: Pr(r, z, t) • - Pr[11 (r; z, r) + 4S~(r, z,' t)] '
'-
(6.-41)
where f'tckll Mel the fluid dcDiitJ (AcbaaiNichi lJU, for-ple). HCJl4:0,., Fouriw and zerotb o.- Hankel uaasrona cl (6M). we .obtaia!. ,
pf{lc, Z, fD) • p1aPf.t(lc, t-, Gf) + .2(~ %, m)J
\.._-
(6.42)
where •• and • 2 were previously detcrmiDed by Bq. (6.40). The pressure distribution in the ftuid is then obtained by inverse trauaform of Eq. (6.42), ~: ,
~
, A,r, z, t) - ;.
'-
f J
_+..,.., _+..,.., p/co•r•. (i, ~~ tiJ) ....2(t, z, co)]Jo(kr)t dk dco (6.43)
.. .....,.. of,.,.,...., ..........
'---'
The simatations that we pnwat were performed to.y lleeeabaum (1974). based on the expcrinaeatal data of Wyllie n Gl. ( 1962) (aee Section 6.1.2.lc~ Only two spedficconditions
at the iDtelface were ualyakt: (a) Free interface (b) S.W. irlterfaGe
·-..
t~'-
·
'
"• • oo.
"• -
o.
To aDalyze the eft'ect ofpermeabDityJllore pndlery'ia each simulation, the results for each type of sandstone were rcplacecl by die ·rMitl obbtibed for a fictitious equivalent material Qf the same dlarKt...... but widllowcr pmacabilitY.:
,_
(a) S mD instead of 1900 m'D for Tupot salldltone. (b) s·mn instead of 200 mD foi'lerea taridstofte. · (c) 0.1 mD instead· of 3l.S, 10 or' I iaD fOr Poxhilts Andstone.
~'---
_... '~
'
274
WAV~
AND
INT~RfACES
6
The source employoci is a pressure so~ who~ tfequency spectrum is such that its phase is zero and its moclulqs.c:oostaat bet~ frequencies / 11 and .f•• and zero.on either side of the frequencies f 1 ( < f 11) and f 1 ( > f.). Between f 1 and fs and between / 2 and / 8 the spectrum amplitude rolloff is a cosine taper (see Fig. 6.17).
I
l
'a
.
,,_..,.. 'H
'2
Fig. 6.17 Frequency spectrum of source function.
Two JeOIDetries wcte iiM:dipted by ROICilbaum: plaae (fia. 6.14) and cylindrical geometry (Fig. 6.16). However, it should be recalled that the Biot's effects occur essentially at discontinuities which, by the bo.undar)' conditions that they imply. induce pressure gradients which alter the flows. This shows that, on one hand. the geometry of propagating media can only play a secondary 1'0~ ·and why, on the other, pcrmeabi1ity and ftow conditions at the interface have the ~same .qualitatift iRflueac:e for both aeometries. We shall therefore mainly discuS$ the l'esults concernins a plane interface, but compare them to the ones obtained for cylindrical syinmetry to emphasize the foregoing remarks.
Different wa•e type& Several type$ of waves boundary:
are
senerated by . . immened sourc:e ncar a fluid-solid ~
• Body waves: (a) A conical P wave rcfr~ P oa .the interf:ac:e (dcoot~ P below) constitutina the f~rst arrival. (b) A conical P wave refracted Son the interface (denoted S) constitutina the second arrival: this wave only exists if the S wave velocity in the solid is greater than the P wave velocity in the fluid. (c) A direct P wave in the fluid (denoted D), which does not .. meet- the interface: this wave is often muted by tbc. StQilcley wave whose velocity is (see Section 6,1.3.1) close to the wave velocity in~ fluid. • In the case of the cylindrical ~ry (F~g. 6.16) and of the equivalent plane geometry (Fig. 6.15), ''multi-reflected n piQed. wa,ves which display an i.Qfinity of modes, all very dispersive and attenuated, e~ in elastodJ1W11ia (Paillet and White, 1982): the zeroth mode of these guided waW~. slightly dispersive, is also called the Stoneloy mode
~
'
"~T
~ Ji
'
- ~I
'-;
--'-" '--
'"-
,, 1
----· '-'-.,..-
'-.,..-
~
-·-
1:~
·- ·;4-
WAVES AND~
6
21S
because, in elastodynamks, the velocity correspoading to the high frequency linut JJ tbe velocity of the Stoaeley wave (the Wllvelength beiDa very small. the. cwvature Gf· the interface doll not have any eft'ect). Its low frequency limit is the familiar tube wave in borehole seismology. • In the case of a plane interface. a Stoneley wave (denoted St) previously described in Section 6.1.3.1, which is non-dispersive and non-attenuated in elastodynamics. In the simulations and actual recordings (acoustic logs) these multi-refiected modes (except the 2CI'oth mode) are very often masked by other arrivals due to their small amplitude. Hence we shall only analyze the conical P wave, the conical S wave and the Stoneley wave, whose amplitudes are significant.
Results of 101M . . . . . . . . Different theoretical sipals are shown in Fip. 6.18 and 6.19 to simulate signals recorded in Berea sandstone. As a ..-. the anivals correspond to rather low amplitudes for the P wave, moderate for the S wave, and very energetic for the Stoneley wave. Variations of the amplitude ratio obser'Wid fortb.e thtee types ofnvc(P. Sand Stoneley) in the case of a Jiven porous material, and of those observed in the case ofthe equivalent material having 11M sune cba.racteristics but very low permeability, are shown in Figs. 6.20 and 6.21. The sipal is recorded in lbctfluid., the receiver beiDt at 3m from the source and at the same depth. Infl~nce
of the source frequency spectrum.
~
For Teapot and Berea sandstones with high and medium permeability respectively, high frequency and low frequeacy sources w~ tested. The phenomena are more pronounced (see Fig. 6.20) at the hiper frequencies, which is expected because high frequencies favor two--phase eft'cCts.
....... '-"" '-..-~
-
lrt}lvence of interfae. flow cmuliJioM. (open or 8flled interface). The effect of an opea or sealedioteffacc is prKtlCitly neali&ible Oft the P wno and the S wave (sec F11- 6.20~ However it iuadical on tbe stoneley wave (compare Figs. 6.19a and
,__,_
6.20 to Fia. 6.19c:~
·-.__.
Infl~nce
'--'
'--
of permetJbility.
Rosenbaum's results are summarized in Table 6.1, which gives the calculated attenuations per unit lonath obMrved for each type of wave. .....__._
The effect on the P wave is plllactioally nepaibJe. Therefore for example. in a very permeable sandstoae (J'apot) atltlnlation is lea than 1 dB/m. The effect on the J wave is sipificant and relativoly independent of interface Oow conditions. This is a particularly intmlltina result for the inverse problem of measuring the mobility of the ftuid directly from fteld data. The effect of the Stolldey wave is considerable. bul de~nds heavily on the interface ftow conditions. For the hiJh permeability Teapot sandstone, open interface attenuation is 20 dB/m and sealed interface attenuation is 1 dB/m. Note that &a results pwa fol: low~ Foxbilli sandstone (JCC Table 6.1) are only refOI'ellCIO iafoanadoa, aad aN aot .at all OCJIDIW~ to the other raults (Berea and Teapot saDdttones). 'I'M fnlquency specm.. of the source is much richer in high
,___ '--~
"--' "-~
~
--~
"---'
--'-.,..-
"--'
41
,
•I
I
,
O+St
•I
I
I
D+St
I s.lecl ....._
I~
..t J •I 1110
IC
J
:=_
~i: .I
I
J~
·~
0
-
...
1100
"*limO
- .
10110
-
...
1000
100
Ti!lllelft ...........
1 ~
l~ • ~ j~ ~
1100
~0
~
l,
Fi&- 6.18 Tbcoretical sci.smopuDI for a plue ~ted~ between a saturated
~~ere& sandstone aad water. The tranamitter and receiver are near the interface. and the transmitter/receiver distance is 3 m. The source frequency spectrum is aiven in fia. 6.17 (/1 • 0, f•- 10kHz, 1... lO kHz, / 2
J
.J.
• 30kHz) (after Roaeabum, 1914).
J
J,_
,
•& •••
I s
,
O+lt
·-
.....,_
s-led
-.
'
I tr-
I
&
---~AA.._____L,
.----AAA-
~,,.---~-----, ~v~
-40«11
- -_, .. -,__ ....... r
I
I
1
r. + -- - -I
I
····mO
I
-li
='l
11::-
~i
--t: -1
•
!
I
J_,
....
1-:J
J 1400
Theoretical seismOaraJDs for a plane interfaee between a saturated Berea lalldltone an4 water. The U'aMiDithr aacl reoeiwr aN aear the interface, and the traasrDitter/reatiYW distance-is 3 IQ. The IOUia (~ spectrum is pven in Fis. 6.17 (/1 • 15 kHz, f. "" 25 kHz, In = 35 kHz, / 2 ... 45 kHz) (after Rosenbaum, 1974).
Ji1c. 6.19
~
~:
·=t·~ limO
I
~
l·~
• •
~---.
TliMinmer-lds
~
......, ....,._
d
c.
'
l
.I
.t-··UIIfll---
-=. 2001110
...("""
O+lh
I
1100
_f
-l
~ --.
........,
-----.. '\
~~
i
(
(
(
(
(
('
( (\
(
(
(
(
\
(
(
0
(
.~
II
"'r (
(
._.....
----------------· BEREA SANDSTONl (IC • 20D mOl
,......,_... 1.0
II t·' I!,=,
r
:.
":·
.01 ~
.01
11
-
__ ,
1.0
.........,. 75
dlst-c-t
Openlnt.t-
.___,
0.5
' : :::·· li r
·.
~s
1.0
~~ ,.Q1
~
1.0
lu I
._..........,
.
......
Open ......
TEAI'OT SANDSTONE I " : I 900 mOl
31110
._
I!,=
J
.1
J
.1
I .. f1
I
St
. ) t.o
.01
,01 ______ L 11
31110
...............
dlltMiatlcml
.5 .1
cllltllllltcml
•
.01
i
r ~~ 1.0
li
r-
-P.St.S
r .01
=-
Il ·' ,.~ J
75
300
Sou___._
~-
,
.01
.01
~:
.1
J•
.01
!-
1.0
1.0
j
.5
i-:r l
. 75
..
.......... .......... 300
1 I! a!
"!:::
.1
___ ,
1.0
I
.5
I1
.1
's
j.OI
.01
I
-e
,
1
I
z
.Ot ~-"-'-'-"""--75 300 Source-r-w
\
,011
1
1
0
0
75 31110 Soun:e-realwr
~-1
Fig. 6.10 P ~tnd S rerracted wave and Stonelcy wave (St) amplitude variation!! Rll a (unction or transmitter/i'eceiver distance. These amplitudes are normalized to the amplitudes computed for an equivalent low permeability (S mD) medium. Case of plane geometry in Berea and Teapot sandttones (after l.oseabaum, 1974~ L
,L.-.
,.
._lcml
II
(
( (\ ..( ....,..,..... ____ (
I
•
',. ~
FOXHILLS SANDSTONE
St
1.0 ,.... 0
E
"' ~ H I(
1·5
t::t::!:::
-
•
1.0 ~
~:
1
l .5 rJ
) I
.1
20
40
eo
"'
::!!
"@0
...
-:z
_j_
.1
2040
so
~Neeiverd..._
>
"'!
St
l
I z
i
Open int..tKe
SNied inwt.ce
10
~
eo
~""
Soun:e-,_w., disu.ce (em}
(em}
0 __:~
..,
1.0 ....
I
.I
I
i
..
2 I(
• .5
.
...
~:::::!
:· St
1.0
-!t"'
I~~ ~: I:"
Q
"'~ ~ ;;l l,j
St
)
J
..
1 I
I
.1
•j!:
.
r-
1
Ill
'
.1 20
Carnl
-=-~ . ~
.<·· :>-
2040eo80 ~iverd..._
_J "' Ill
40
~~
10 80
S o u - - - d - . - (em}
-.0
-~ z
..'_...l
<
i ~
I
I
i
u
I(
1.0
L"
• • •
I
r P,St,S
I
•
1.0
.I 1-
• .5 ~
.1
I
J
!
~~
:';
I
I
20
40
Sou~vrr
I
I
eo80
d.._ (CI'II}
•
~
I ::::::
-Ill
P,S St
£
-·"' ··.~ s "'
~z
~ ~< ;;l
.1
I
20 40
_j_
eo
80
Sour~~lcml
P ~ Stefrac:ted "!ave and Stoneley wave (St) amplitude variations . as a function or transautter/receiver distance. These amplitudes are normaliad to the amplitudes computed for an equivaleat low permeability (0.1 mD) medium. Case o!plaacpometry iD Foxbilksandatoae.. The.sourcc frequency spectrum is given in Fig. 6.17 (/1 =60kHz, / 2 = 160kHz) (after Rosenbaum. 1974).
z
-E<
F~~o 6.21
-------r----------------------------------------------------------------~
' ..
~
(
~
(
(
(
I
(
,..
(
(
(
(
(
'
(
(
\
(
(
(
(
(
(
(
(
(\
TAKLF. ATTENUATIONS OF INTEIFACI!
P,
6.1
SAND STONI!LF.V WAVF.S C"ALC"ULATF.D I'OR THF. C"ASF. 01' PLANF. (if.OMF.lRY
I'Oit VAIUOUS TYPF.'I Ot' SANI>!."TONF. WITH lllt'I'F.RF.Nl" pt;Utf.I\IIIUTIF.S
.
Type
orroct
Teapot •ddstone
1900
BerCallliMktone
200
frequency
spectrum
(mD)
Attenuation
Cetttt-al
F~equency
Permeability
ofiOUrc:e
~
IS_, 45kHz
1.2
Pwave
Swave
Stoneley
Open
0.4
3.5
20.s·
Sealed
8.9
4.0
0.9
Open
8.9
4.1
11.6
Sealed
0.9
3.5
0.9
Open
ts
5.9
17.1
Sealed
1.S
6.2
~o
Open
~o
3.2
10
Sealed
~e
3.2
~o
Open
~o
~o
J.5
Scaled
~o
~o
~tO
frequcney
IS to 45kHz
0.2
·32.5
(dB/In)
Interface
3
60to 1601Hz ,.
Poxhilts. samistcine
'"7.
3
60 to 160kHz
3
60 '*' 160
tO
I
T,uu 6.2 CoNI'A&IION Nl
Sowce
spectrum
BI!TWII!N A'l'fafUATtoNS CALCULATED IN I'LANI! AND CVLJNOiliCAL OBOMITitY
Bal,4 SANlJSTONI! 011 ~
Pwave I Swave 0.9
0.9
Stonefey
3.5
Source spectrum
11.6
4.1
IS to 45 tHz Sealed
20(). mD AJIID POaOSITY • •
Attctauatlon ~B/mJ
Interface Open
I( •
I
19 %
AND
P0a //f.
• 0.2 Attenuation (dB/m)
Interrace P wave
Swave
Stoneley
I
4
3.4
0.7
2.7
()pea IS to 45kHz
0.9
Sealed
-~---~
Plane geometry
Borehole (24 em diameter)
(FIJ. 6.14)
(fiJ. 6.!6)
0.6 ....
-------
I
,
'---~-~
----
(
280
6
WAVES AND INTERFACES
freqacacies (central spectrum frequency/cha!1lcteristic frequency of Biot's theory ;; 3), since it is also necessary to use very bish frequencies to observe sipif1cant effects at low permeabilities (see Table 6.1 and Fia. 6.21). Furthermore. for comparison with the attenuations previously observed for plane geometry, we have summarized the same results for the attenuations computed in cylindrical geometry (see Table 6.2) in Berea sandstone. As we previously pointed out, the effect of the source spectrum, permeability, and intctrface conditions of the different waves is similar. However, the attenuation of the Stoneley mode is smaller for the cylindrical case than the attenuation of the Stoneley wave for the planar case (see Table 6.2). The interpretation of these observations is given in Fig. 6.22. In the case of the refracted P wave at the interface the movement of the ftuidis more or less in phase with the overall movement (tee Chapter 2). This means that the differential movement between the two phases (the source of attenuation in Biot's theory) is relatively small Moreover, particle motion, due to its polarization, iS parallel to die iatert'ace, so that the interface conditions are unin:lportant because the filtration velocity vector is also parallel to the interface.
p._.,. li'ropegetlon
fluid
..
s-w-
-~-
fllulcl
Solid
Solid
,.. ........llltllliGft
...... lloc'.......
........... .................... ......... Dllfhllllllt.......... to .
................ _........., ......,..._ .......... •..
.. .................. lkrt
.-,
StoneltyPrQPIIIIdan
Proplgltlon
..................... •
·r.
'
.~
~~~- ~ r o
Dlplh
"''
\
.....,
tafluence or relative movement or fluid phase in relation to solid phue aad interface conditions on refracted I', S and Stoneley waves.
Fta. 6.21
Phyiical iaterpretation.
In the case of a refracted S wave at the interface, since the liquid phase does not respond to shear forces, no pressure aradient occurs and hence there is no fluid exchanse across the interface. Thus the open or sealed cbal1lcter ofthe interface has a nealigible effect on the S wave. However, since the interface is set in motion, flow occurs in the porous medium by inertial coupling (Darcy's law and inertiaUerms), and not by the pressure gradient etTect, which is zero. Thus differential movement occurs between the fluid and solid skeleton of the permeable porous medium affectina the S wave attenuation. Finally, permeability and interface conditions exert a considerable influence on Stoneley waves, as discussed above (Section 6.1.3.1). However, the effects in the
'--, ~
~
"'
.._.-
• 6
·~
. . . .ANI)ftim!lt.ti'.«:!!S
neighborhood of the interfa<:e arc incrcasinpy pronounced for the Stoncley.wave and at. higher tiequency (lower penetration depth). since Stoae1cy, particle movement is elliptically polarized and hence ditfcrential movement ocx:urs between the fluid and the porous medium. I tifluence of distance
·~
_/
'~
'--'
'-
·-..__._ '-'
,_ '.._
of transmitters and receil'ef"S from the interface.
All the simulations described above apply to transmitters and receivers very close to the interfaces. In Fig. 6.23c. the sourc:c is lcK:atoclat 10 em from the interface. and the porous medium is a highly permeable Teapot sandstone. The effect ofdittance between the sourc:c and interface is clearly undetectable for the P and S waves(c:o.aaparefigs. 6.23a and 6.23b with Fig. 6.23c). wlaercaa the same distance (10 em) implies that the Stonelcy wave amplitude is attenuated by about 30 dB. Stoneley attenuation is approximately the same as that due to the difference between a sealed interface and an open interface (;;; 20 dB, Table 6.1). This is expected, because Stoncley amplitude dccreucs exponentially from the interface. whereas P and S amplitudes decrease as the invcnc of the distance from the interface.
6.1.3.3
.....
211
Coadullloas
To conclude tbitaalylil, aucl from a practical studpoint; if research is guided in the direction of invcrtiilg acouatic lhcasurem•u for mobilitJ ofdadluid, independcndy of the interface Dow conditions, it is of primiry importance to Ute bish frequency sources (that is to say frequency dose to the chaaeteristic frequency). However at these high frequencies, the Stoneley wave is strongly attenuated and hence diffiCUlt to observe. In practical applicaqons, it will be necessary to compromise, in terms of tiequency, if the Stoneley wave is desired. AI it has been thown, the Stoneley wave is most influenced by permeability, in the case of an open inttrface. It is important to note that the fmt artivals (P wavea) are •• inlensitive" to permeability for any interface conditions. F"mally ·4- S wave is especially interesting because its attenuation is sensitiw to the :variation iD pcrmcability and i~t of interface flow conditions whicb arw,. . .t for in tiWIIliU1JI'elllCRt. . The transmitters and rec::civers must be placed relatively'*- to the interfaces for these c6c:ta to be observed. It is nevertheless difttcult to evalute permeabilitics leu tbaQ to .aD'(see Fig. 6.2le and Fia. 6.2U) cspccially if the interface is -.Jed. Laboratory experimcnu on borehele models of non-porous epoxy and sand (Schoenbera et al., 1911) or porous concrete (Cben..l9U) materials have revealed wide gaps betwoea the RtuiU olailild by ~ aDd tlaose predicted by theory. The possibility of sample hctcrepoeity has been suaested as an explanation. For in situ acoustic loging experiments, the problem of the inversion of acoustic measurements for permeability still remaiaa :poorly .....~ For the. aiAae bling. it .appears that, in many ca1e1, ue is oftina maa-of a ftUJDiaer of e&Dpirica1 folmu&as, dtlvoid of any solid physical foundations, which work only in rare favorable eonlipratiOM. However, the importance of this inverse problem for hydrocarbon recovery and reservoir evolution justiftcs a major effort in this area to understand the physical phenomena and develop an experimental proaram. Some references on field applications of permeability effects on interface waves are given in Chapter 7, Section 7.2.
. •(tL6t 'Umwefua80lf £ 5! ~S!P »A!**J'»U!UJIUQ ~)--- pu1J NOl. . . . ~~ puaua U;»MlaQ acnJj»J9} aQid 11 10j ~~oaq.J. .£r9 'lt.!i
---owtps
Jltp}{W
c-. tlllt.L
i" i"' 'i'
I
I
0
l
01
woo,....-.
-.-ut flll.UI04
'---"'ln---...;..-~o
,__.....
·.~
t
lS>Q
I
t
s
d
01
....._,_,.,
•.-.u!"'lctl
• t
t
..,.,..UI pejft$
q
lStQ
IIWIOIIIOS
01
'
s
d
01 -...ul"'lOl
.........
-p-.-,.
'-
0
-.-ut
liMO
•
·~
t
lStO
9
t s
01 (JH'Iot=HJ:~,.IJ) ZH'I 9t>-O allll\4adl801f10$
t d
S3.JV.:fll3l.NI QNV S3AV.'A
QIIIOQ8L•,.
1' L'&t• • , 3NO~NVS .LOdV:i.l.
-
~
.,
·*AVIS AN&~IWAd!s
283
6.2 REFLECI10N AND TRANSMISSION IN VISCOELASTICITY
·~
We have dilcussod the intemum in Biot type media ia tbe tint pan of this Chapt«. In this second part. we shaD examine the chanJCS made by the introdactioa of iat~ in a viscoelastic medium. We shall develop the arguments in a thne-climeasioul space, and then. for quality-factor-related problems, we shall limit ourselves to the two-dimensional case. We will then calculate reflection and transmission coeft'Jcients and examine in detail the ditterences with respect to,the elastic case. Tbe problem ~ ielerface waves will be discussed very rapidly.
-~
6.2.1
Wave equation in viscoelastic media
We showed in Chapter 3 that the unidimensional c:dl!lstitutiveequation between stress a and strain e of viscoelastic materials is written in one dimension :
a=m•e
(6.44)
The extension to three dimensions yields:
a.1(t) = C1.;t1(t) • £111 (t)
(6.4S)
In this equation, Einstein's aotation is implied,. with IUIIIIDadon OD the repeated subscripts. In the isotropic case, Eq. (6.45) is reduced to: ·~
qlj(t) =
·-
'•J( K(t)- 32Jl(C)) ·~~(t) + 2p(t)Hu(t)
wbcrc 61} - K.roooc*cr 4clta;
·{ 8,1 -o ~I}- 1 ~
(6.46)
iflt'J if i • j
K(t)- bulk modulus, p(t) = shear modulus. The equilibrium equations are:
a,1J"" pu,.
(6.47)
where u1 is the f'A component of the displacement Equation (6.47) caa be ICWI'ittlln by iDtroducina the constitutive Eq. (6.46) in Eq. {6.47):
•
pii1 - 61{ K(t)- ~t)] u•..t&' + p{l) •ui.JI + p(t) • "JJJ "--·
(6.48)
284
6
WAVES AND .U'ITEJt.f'A.CES
In this equation, displacement u is a function of the space [r = (x1)] and time (t) · variables. Takina tbe Fourier~ 1Jf (6.4&), we obtain: - pw 2 U
= ( K(co) +It(~))
an•
8
+ p(co)
V2 U
(6.49)
where K(co) and p(~are tlte FoariettnnaformsofK(t)andp(t)R"JSpectively, and 6 is the ~olumetric 1train - div U. U - U(r, 0)) is the Fourier transform of •(t, t). Let us fmd U in the form: U = arad 4> + curl .P (6.50) with the condition ctiv 'I' • a By combining Eq. (6.50) with (6.49) we obtain:
172 4>
V2 ,
..
+ kJ 2 4> = 0 + kJ 2 '1' = 0
(6.51a) (6.5lb)
with
14
pw2
2
K(co)
•
(6.52)
+ 4/3Jl(co)
kJ2 = pcol p(co}
(6.53)
or
1
1
kJ == ~~ V,.(QJ) • (K(co) + 4/3p(co))'l • ().(co) + 2p(co))2 Y,.(co}
.
p
co kf • Ys(a>)" V,(m)
.
=
(
Jl~)
r !
P
(6.54)
(6.55)
The quantities K(co), p(co), V,.{co), V,(co), kJ and kJ are complex quantities. The chosen branch for the square roots in Eqs. (6.54) and (6.55) correspond topositive real parts. This choice will be understood in the ~g discussion. The general plane wave solution in the time domain of Eqs. (6.51} is:
4> == •o exp [i(cot - k• • r)]
(6.56)
In this equation, k• is a complex vector which we can separate into its rCa1 and imqinary parts: k• = k- iA
(6.57)
where k is the propagation vector and A is the attenuation vector. The solution (6.56) to Eq. (6.51) is then written:
4> = 4>0 exp (- A • r) exp [i(cot - k • r)]
(6.58)
As a rule, the vectors k and A are not panllel, and in this case the wave is inhomogeneous. If the angle between k and A is zero, the wave is homogeneous.
'
-,
-;;or ,_.~
,.,
-~
-6 '"-'
· . . . . . . MIDINUIIPAC!ES
PhyiM:ally, the wave ampliludc must no& Daile in dac propaption diRiction.:implying that the anale 7 between dae wctoa k aacl A must satisfy:
-~
·~
O~y<
2•
1be existence of a non-zero anale y expNISCS that the planes with constant phase and planes of constant amplitude are not parallcl. Fiaure 6.24 shows schematically the behavior of a homogeneous wave (y - 0) at the interface between two attenuating liquid&. 1be transmitted wave is automatically inhomogeneous.
·~
·~
---
...
~
.... 6.JA ......... ud .IIIW"watt fll &.-pilule medium (Schw•tjc a..,..~
'-'
WMe Uta vilc:ociMtic
'-
,_
Equations (6.Sl), (6.56) and (6.S7) live us: '-...-'
k• • k• - .., ..... fAI 1 -- '211AIIkl COI1
(6.S9a)
~ /X»1 k•. k• • - • iiiif [Ma - iMJ M IMI
(6.S9b)
and '-
\-
where M • M(QJ) is the ooaplex modulus of the wave, namely:
""'-
'-
4 M(QJ) • K(QJ) + j p(w) • l((l)) + 2~t(w) for a P wave, and
'-' ~~
\_
·~
M(fO) • for an S wave.
~t((l))
286
6
WAYI!i!h\ND JNTERFACES
In two dimensions (x,
z coordinates) Eq. (6.58) is written:
tfJ = tfJ0 exp [- (Axx
+A:z)] exp [i(rot- (k"x + k=z)]]
(6.60)
and Eq. (6.59) can be expressed in the form: 2
(kx - iAs-)2
+ (k: - iAJ2 =
p:;
(6.61)
or
k!2 where V is the complex velocity
+ k: 2 = pro
'
2 _ w2
(6.62)
M- y2
~med
-,,
by:
V=l=;_ Equation (6.59) enables us to calculate
lkl IAI =
= { ~ [ Re kt {
lkl and IAI as a function of kt- and 1
+ (lYJ 2
2
~ [ - Re kt
+ (I::J~J
2
2
+ (
2 2 kt )
r
~.
y:
1
1
YJ
(6.63)
y 1
(6.64)
The phase velocity of the plane waves considered is equal to:
v
=
' (0
k
(6.65)
lkl 2
By means or Eqa. (6.63) and t6-64), it c:aa be 'Sbelwn that this volocity is· higher for a homogeneous wave than for an inhomogeneous wa'Ve (Borcherdt, 1973). Finally, it is interesting to note that, in the case or homogeneous waves, the foregoing equations are simplifaed. If 6 is the angle propagation measured from the vertical (Fig. 6.25) we obtain:
k! =
~- iA,. •
(Jkl- iiAI) sia 8 == ro
j{;
sin 6
(6.66)
II
X
~r"'k z
---...
A';
~
II '
-k -,
z
Fig. 6.25 Attenuation and propagation vectors for homogeneous Oeft) and inhomogeneous (right) waves.
-----.. -~,
"'
.,........
WAV!f'MftntftdFAees
6
28?
The aDJic tJ is given, whether· the wave is homogeneous or iohOJno~ by:·
""
(6.67)
lkl
sin 6 •
which is written, in the case of the homogeneous wave:
. 8 = Vi + Yf -k"
v..
SID
(6.68)
(I)
where Va and V. are respectively the real and imaginary parts of the complex velocity V = foJP and equal to: (6.69)
v. 6.1.2
-('.vt;M·t
(6.70)
Energy balance and quaHty factor
6.1.1.1
Energy balance
We have shown that the displacement uWi8flcd Bq. (6.48), or that its Fourier transform was the solution to Eq. (6.49~ for a plane wave, it is always possible to write Eq. (6.49) formally as a function of u (see for example Bourbie and Gonzalez-Semno, 1983):
pi= (K(w)
+ p~td)] arM 8 + p(w)V2u
(6.71)
with 8- div u.
__
The actual displacement vector is the real part Ua of the displacement new equation verified by Ua: ·
11.
By deriving Eq.
(6.71), we obtain a
,
pl. - ( Ka +
~) .... 8~ + l'tV2. . +
![
(K, +
~) .... Ba + Jlt vzu.]
(6.72)
The .sublcripts R and I indicate the real and imaginary parts respectively of the quantities considered. · The~ ~ua~ is olQined \lY detenniaina the scalar product of Eq. (6. 72) by lia (Linds&y, ~~~After tranSrormatioq (llorctierdt, 1973), we obtain:
Pa) 8a2 + 21lallaa.)llaJ•J 1 ] ot0 [p2 ~2 + 21 ( Ka +. T
1 [(~
+(I)
I
+
Jl•) 8a. l + llilla,..JUaJ . ] 3
. [(Ka+3Pa).Baila+ •1 ( Ka+3Pa) 8atia · )
-dtv
+ p i [ ,..(U. • div 11a +
!
Pt(litt • div
(6.73)
lia)]
288
6
WAVES AND INTEitFA(;ES
The intcaration of (6.73) on. a vol&UU. D ofsurfacc SJi~ us the equation:
:r L
E dO+
fn 0 dQ = -
II•.
dS
(6.74)
where E == sum of kinetic and potential .energy densities, 0 = dissipated energy per unit volume, I =energy flux per unit time.
6.2.2.2
Quality factor
The knowlcdae of these quantities now enables us to calculate the quality factor defmed in a manner compatible with one of the 1D defmitions. We have decided to defme the inverse of the quality factor in the same way as Borcherdt (1977), namely:
Q_1
= _1 loss of eneraY denlity per forced oscillation cycle2x
peak energy density stored during a cycle
( ) 6 75 ·
Using the results of Borcherdt (1973), the values of the different quality factors are obtained: (a) For P and SV waves:
Q_ 1
M1 + ll•Mf/IMI 1 tan1 y - Ma + ~taMl/21Ml1 aaa1 y _
'·
(6.76)
where for P waves:
4 M == M(
(b) For SH waves, Q- is given by: 1
1
Q _1 _
lla[1
6.2.2.3
+ Jll./114 1 tan1 YaP 1 + (1 + Ill 11Jll 1 tan 1 YsHJiJ
2p.[t
(6.77)
Coostaat Q model in two dimeosioas
We showed in Chapter 3 that KJartansspn (1979) bad developed a riaor~ thcotetical model for which the one-dimenSional quality factor was ripously ilickpcrldent of frequency. The quality factor considered by Kjartansson was the one defined by:
Q""l == : :
(6.78)
Equations (6. 76) and (6. 77) are the same as Eq. (6.78) in the case of homoaencous waves (y == 0). For inhomogeneous waves, these equations are different, as the angle y between the planes of constant phase and oonstant amplitude atfects the value of the quality factor.
'--
(
\
(
(
/
(
\
I
(
(
(
(
(
(
(
(
(
I
(
\
(
i \
(
(
(
(
(
(
(
(
'
(
{
(
(
(
(
(
( I
~~
CottsTANT
TAIU! 6.3 Q llroDBU IN ONI! AND TWO DIMI!NSIONS
Constant Q mqdelln two dimensions
SH waves
p(al} = p(CIJo)
QiJ-
SV waves
,.(01} _
(w)z'
p(m) •
0).
tan.,
OiJ = 2 tan JCI
p(wo)(':.Y'
+ lin2 Kl tan 2 .,,.lr I + (l + sift 2 . , tan2 "'1111,. ,t (l
p(Ofo<:r'
lw)"' p(Q)) - p(We)(C»t
tan--
Qi.,' • tan .p l + 1/2 smi •P tan2 "/sv
Qi.J-
1 + ltin*"' tan
(t)
M{GJ) •
M(oto{:)~
2
"/sv
.... lictuid medium
Q;' •lan trar
P waves
JI(Q))- Jl(t»o<:r·
(2)
M(w) • M(roo)
Q;·---
(iw)a. Wo
p(w) • p(wo)
and
1 + l'(a~o)fM("'o)
,-a I'
Al(w)
(4) Trivial
CllliC
= M(•o)(;iw)l
a;'- tan •«
M• .. o Ma
2
•)2'
~· It{«») • p(~Gi;
J
Wo
2 •
sm2 •ar tan2 y,.
"' -tan" 1 + p(aJo)/2M(Q)o) sift (3)
(fw)
Q;'-o
XIX
tan 2 "/r
, tan u • 2 tan
•P
290
W.-\ VES A:>;D
I~"TERF ACES
6
This is hence no longer an intrinsic property of the material, but a combined property of the medium and the type of wave analyzed. It can be shown (Bourbie. 1982) that, in the general case. a rigorous Constant Q model does not exist in two dimensions for an inhomogeneous P wave. In very specific cases, where the material analyzed reacts in a non-independent manner to longitudinal and shear waves, a Constant Q model exists. The results are summarized in Table 6.3. Nevertheless, it is important to realize that the variation. as a function of angles i'r· i'sv. y58 , of quality factors Q-p1 , Q.S/ and Q5J. defmed by Eqs. (6.76) and (6.77) is slight. Figures 6.26, 6.27 and 6.28 provide an example of these variations for a particularly unfavorable case in which attenuations are extremely high. It may be observed that Q -;.1 , Qiv1 and Qi,} are independent of the angles }' if these anaies are smaller than 70 to 75 degrees. The hilh values of y, as we will shoW in the next Section. are only obtained in the case of an incident homogeneous wave for angles close to the critical angle. Hence we can consider that the quality factor is virtual~· independent of the inhomoaeneity angle y, which amounts to stating that the quality factor can be assumed to be equal to the defmition (6.78) even for the two-dimensional problem. The Constant Q model of Kjartansson can be considered as a two-dimensional Constant Q model.
6.2.3 ReOections aad transmisllons in two dimensions 6.2.3.1
Theory
The details of certain calculations, particularly for the case of homogeneous waves, can be found in Cooper (1967), Lockett (1962), Borcherdt (1977) and Bourbie (1982). Let us
consider an interface between two attenuating media The variables associated with the medium in· which the incident.wave propagates will be denoted with a subscript lor 2 according to whether an incident of reftected ~·ariable is involved. The '-ariables associated with the medium in which the waves are transmitted will be denoted by a prime('). The computations will be carried out in potentials, since their use simpliftes the wave equation. Finally, time dependence will be understood and equal to exp (iaJt). The incident wave in the most general case is the sum of a P wave and an SV wave. If x is the horizontal (i.e. along the interface) coordinate and z the vertical coordinate, the system of equations to be resolved is the foJiowing :
• = A 1 exp (- i(kx + dz)] + A 2 exp (- i(kx- dzl]
(6.79)
+ B 2 exp (- i(kx- hz)]
(6.80)
'I'- B 1 exp (- i(kx
+
hz)]
., ==A" exp [- i(k'x
+ d'z)]
(6.81)
'I''= B' exp [- i(k'."C
+ h'z)]
(6.82)
R.e td) ;il: 0
(6.83a)
k2
IV•ll
+ d 2 = -.,_-_ I.+ 2p
k2
+ h2 =
pwl
-
J.l
Re (h) ~ 0
~,
(6.83b)
,,
'--
'-· -·
', __
I
.30
.-.·;
I
,.
~
·-~'j:;r
.· '-:'t·:,-.
Vp=2
hp = 0) =5
v$ = 1.2
0s lls =ot""s
:.._~~"· ~.~
f'il. 6.26 Inverse of quality factor 1JQ1 vs. angle between attenuation vector and propqation vector (after Bourbie, 1982).
•18 1
Osv
.t21::----------.. . --.06
~~_.--~--~~--~--._~~-L~
0
w
~
~
~
•
~
~
•
~
Propegation . •tt8nUitlon ..... (dig.)
~r---------------------------------~ Vp =: Op hp "'01 = 5 v 5 = ;,2
.24
.18
cl; .12
FJe. 63!1 . lnvene ot qalifJ facft,t
1/Qsr va. aqle.,...... attimua- ; o
tion vector and propagation vector (after BourbiC. 1982).
..oe OI 0
I
w
I
~
I
~
Propeption.
I
~
I
I
I
~
~
~
,....rian...........
I
~
I
~
30r---------------------------------~ v5 •1,2 Ostts .,., ·•
.24t .18[
..;.. .12
..t
'
"'""'
.oe o~--~~--_.--~--~~~~--~--~ 0 w ~ ~ ~ ~ • ~ ~ • Propegation • Anenuetlan engle Idee-I
Fi&- 6.l8 Inverse of quality factor l/Q111 vs. anp between attenuation vector and propaption vector (after ~urbie. 1982).
292
WAVES_ANP ,
y + d'l ...
2
•.P_(I)_ • ,
k'l
6
I~RFACES
+ h'l = p w
Re (d') ;;11: 0
(6.83c)
2
Re (h') ;,:: 0
p'
(6.83d)
In these equations, the quantities k. d, h, k', d' arul h' and the amplitudes are complex quantities, and are related to the propagation k and attenuation A vectors. Displacements and stresses are given as a function of potentials by: (1) Displacements:
a.
o'P
(6.84a)
u% -ax - - iJz -
u
:r
a• o'P
==-+az ox
(6.84b)
(2) Stresses :
... a.. (1 u
. a2• A.
_
02• 2
oz
+ 2P ax cz
a2•
0 2 .,
a2 ")
iJxl + (l + 2")
a2"
(6.85a)
= p ( 2 ~~ az - az'-' + ax 2
(6.85b) -~,
The reflection and transmission coefftc:ients are obtained by appling the boundary conditions at the interface. Three alternatives are avMlable. namely a liquid/solid interface, a solid/liquid interface, and a solid/solid interface. We shalupidly examine the results of the computations in the three cases.
a.
LitiiiiiiNIItl -~
Since the incident wave propagates in a liquid medium, we have p Continuity of displacements and stresses implies: U:r
= 0 and B 1 = B 2 = 0.
•a(
= a;x a.,.- u;z
(6.86)
0 We obtain k • k' and:
A2 p'd[(h'a- k2 ) 2 + 4k 2 d'h']- pd'(h' + P)2 Al == R == p' d[(h~:t - k2)2 + 4k2tl' It'] + ptl'{lt'-» + k2)2 2
A'
2pd(h'• - k') + 4k2 d'h'] + pd'(h':
+ P)2
<6·88)
4p4tl'k(#l'a + k2 ) = Ts = p'd[(h',t- k')2 + 4Jc2d'h'] + pd'(h''
+ k2)2
(6.89)
As • T r -.p'd[(h' B Al
(6.87)
1
-
k2)2
·
I
·w.wM-MellftiiU'AaS
6
293
b. Soll4//i411Ul hltajla Sinc:c the transmitted wave propaptcs in a liquid. we have p' • 0, B' continuity equations arc written:
= 0.
The
"• =,; flu= t1.,.
=
0
(6.90)
a;.
Here again, this implies that k = k'. Two alternatives arc available for the incident wave, namely an incident P wave or an incident S wave: (1) Incident P wave,.. Bt Az
= Rrr =
= 0: p' d(hz
+ k2)2 + 41c;2dd'hp- pd'(hz- kz)z h~
(6.91)
Bz _ 4kptl4'(h 2 - k2) At • R,.s = p'i.(Jil + ~ti)f'+ 4k2d4'hp + pd'(hi _ k 2 )1
(6.92)
At
A' A1
.t
Jn.'i .. • 'h'i .
H
4 •
2pd(h4
111.
•
•u•'i
1
k")
-
= Trr = p' d(hi + ti)z + 4kiUi;p + pd'(hl _
k'2)z
(6.93)
(2) Incident S wave • At • 0:
!!! s.
R,.
_,k:(U'hp :- pd'(h2- A:lt- p'd(hl + kl)l _,
JIL4
•
..1.1 .
s. A'
Bt
'-
------~-
• ••• . ... 4 ••,.
- 41ul'hp(h 2
Az '~
..,. 4
R,- p' d(h2
-
k2 )
+ P)l + ,4'(hi - k2)2 + 4i(!M' lap _ 4/ulllp(h 2
(6.94)
(6.95)
+ k2 )
= Tsr- p'd(h2 + fc'1)2 .;·pd'(lt2- k2)1 + 4k2dd'hp
(6.96)
e. So#U/HIU later/"" The continuity equations are writteft:
u., = ~~~ "• = u; a•.,= a'" ••• - a'., Here aaain we find the c:ondition k = k'.
~ ..
__
(6.97)
294
'
WAVES AND INTERFACES
In the case of an incident P wave (B 1 = 0), we deftne the quaat.itics: A2 Rp p = A,
B2
Rrs=-
A,
(6.98)
A'
T,,=A, B'
Trs=-
A,
solutions of the linear system: dfftp = 11
(6.99)
where
k
h k.
-d d=
- k -d' - lk4' p'
-2k4p ~ (h2 - k2 )/l ( (h2 _ P)p 2 - 2khp - p'(h'
~
~,=
(
R,.,.) Rrs
,g I'Jfl
. T, T rs
In the case of an incidat S
wave (A~ •
=
-
k2 )
J
h' - k (h' 2 - k)p' - 2kh'p'
( - dk J.·.. -
- 2kdp - p(h2 - k 2 )
0). we dCODc the quantities:
A2 Rsr = -
B,
Ba
Ass='B,
(6.100)
A'
Ts,.=Bt
.
B'
Tss=-
B,
solutions of the linear system : dfft5
= f!J
(6.101)
where
Jls =
RsrJ (Tss Rss Tsr
{!} =
- hk J (p(h2k2)
- 2kla,4
~
.,__.--
----
'1
,,
:.t:·;
WAVasANDI!\iEaf:ACES ·-· .}
29S
It is interestin& to note that, in aU cases, the condition k == k' must be satisfied. This implies a generalized Snell-Descartes law. In fact, for a single type of wave (P or S), we have: -~
ka = ka => lk1 1sin 81 = lk~l sin 82 k1
= kl
=> IAtl sin (8, - ;·,)
-~-
= lk'l sin 8'
(6.102)
= IA2I sin (lJ2- 'Yl) = IA'I sin (8' -y')
(6.103)
From which. by means of Eqs. (6.63) and 16.64) we obtain:
82
•d,
(6.1048)
l'l
= i"t
(6.104b)
An inhomogeneous P (or S) wave is therefore reflected as a P (or S) wave with the same inhomogeneity angle and with an angle of reflection equal to the angle of incidence. The values of 8' and i are more complicated, but can be iaferred similarly from the foregoing equations. Figures 6.29 and 6.30 show two examples of transmission angles (8', i) in viscoelastic media, for a solid/liquid interface and a homogeneous iQcident wave. The angle 8' for the attenuatilla case is very close to the angle 8' for the el&Jtic cue except near the critical angle of incidence. which in efcct does not exist for the attenuatina case. The angle "'( increases with angle of incidence up to the .. critical" angle for which y' is approximately
90". -~-
'v
'-
The foregoing calculations show that the different reflectioa and transmission coeftcients are complex numbers irrespective of the angle of inc:ideooc. In Fig. 6.31 we have compared the retlection coeft'ac:icnt for a non-attenuating liquid/solid interface with the one obtained for the S&me interface, but with an attenuating solid with Constant Q behavior. The incident wave is assumed to be homogeneous, and the reflection coefac:icnt for the intcrt'ue betwceaatteDU&tial~ is cak:ulatld for a~ aqular frequency (co0 - 1 Hz) in the Constut Q . . - . . VutuaUy-u c:Uenmae is obeenedbctween the two reficction coelftcients for angles less than the critic:al angle, thoup a larger difference occurs at anaJes greater than the critical angle. The interested reader can fmd other examples in Bourbii and Gonzalez-Serrano 11'983) for solid/solid interfaces. -
6.2.3.2 laterfaee effect of attenuatioa As we have shown, attenuation modiftes the modulus and phase of tho. reflection and transmission coefticients. BourbiC (1986) showed tbat. for any linear vitcoelastk model. the reflection coeftcient at normal incidence for a plane wave could be expressed in the ftrst order (i.e. for moderate attenuations) in the form: R
~ Pt V,o- P2 v20 +..!.. (-•-- _1_) In t~l + ~ L! Pt V,o
+ P2~'2o
2lt Qot
Qo2
CUo
4 ~t
- _1_, Qo2
(6105) ·
The-subscript 0 indicates-that the velocities .v, and atteauatioDI (ljQ,) arc taken at a · . reference anplar fr~ w0 • This gtfteral formula clearly explains the effect observed in Fig. 6.31 for a Constant Q lllOdel, ia other words the modikation of tbe modulus and pbue. However; it also shows
110
""
Vp •1.2 V'p•4 v·s=1
80 .., 70
l -eo fso 5
:[ I
~leo t SOL J I
,..,..,.
):
.I
Cp•10 O'p=50 o·5 =20
I
I
'
,_
40
30,
,__
v,= 1.2 V'p=4 V's = 1
I
'
....
o
ro
20
80
so
lnclc*>ce .ngte
Idea-l
30
40
ro
80
o
110
ro
20
30
40
80
ro
eo
80
110
lncidel)ce ..... 14111-1
f1l. 6.29
Transmission angle vs. angle of incidcacc for a mud/solid interface (after Bourbie, 1982). Left: elastic case. Right: viscoelastic case.
r
110
lt ..~ 70.
I ~~
Vp•1,2
0,.•10
V'p=4
O'p=80
V's•1
O'sa20
I
I
.._
' "
J
J .ro
ro
so • ...................
20
30
40
•
110
f1a; 63t· Aap~ attenUation tector anlpropapdoinec:tor vs.eJie · of iJieidence lor • mud/Solid iatert'ace (after 8oudK6. 1982); 0
I
-
•
i ...
.a
L~
i
I
. 1
I
'5
!
j
0-135
J 01
o
I
I
I
I
I
to
20
30
40
80
C)<-f
80
lnclc*>ce engle ldeg.)
70
I
I
80
110
-1101 o
I
10
!
20 .
I
.1
·I
I
30 40 10 eo lftcldenoe .,.ie ldll-)
-r---r=--' 10
80
'
110
Fit- 6.31
Modulus and phase of reOection coeffiCient vs. aqle of ioeideD:cc for a liqWd/IOiid interface. The solid curve is for the viscoelastic case and the dotted curve for the elastic one. The velocities are v,, ... 1.5 km/s, v,. = 2.5 km/s and Vs. = 1.2 km/s. The attenuations for the viscoelastic case are liven by Q,, = oo, Q,. - 14 and O.ra • 10 (after Bourbie, 1982).
~
-----------------------------------------~---~
~
F j:
6
wAVIS'·.\111)~·
297
that the clastic effects (a
Flaure 6.32 summariles tbe results .obtainecf m aniplit\tde. Ro li
'
I
I
No .........
'UnaoiiiOiidltld ....
lnterl- efftc:t
I
I
A-.~··&~= a:::::a--
I
elatic
pnltM I I"' Ti~ ... uncls
1
reflection _.,.,._ O.ot
ol 0
~;-........ ·~
'-lr!Ft 111Gb ...a.....JL....-....__.__
,
~
40
•
,.
10»10400
Q
,.. U2
AOU).
Jntcnaee .
effect .•
or a,tteauatiotl (._ lloutW· and Nur. C '
·"'
I
6.1.4 ltlterlaa .wA'fes ill
r
I I
.o..-• '-"
·
.· J:
y
r
1984
Left: sbftP66ed·dlalts li'htJ the . . . . . . . . . . ....UCucl vilcoelutic releetion ~.. (%). . . Ript: some "cbatacteriltic" VU. olila .,..._..._
I
·
i)
~
..._
s.-.
We havt lhowa in Chapktt 3 aa4 in 6.2. J &bat die CQQ&Doas of elasticity are easy to write mathematically when tbl .....aptioa of olaaticity is replaced by linear vitoooluticit)'.lndeed, tbe co,........_. bo,applied to replace the real clastic moduli by complex fr~-4ep • deat oaas ia aU e&utodyaami~ equations. The modiftcations of the equations for the propaption pf Rayleigh and Stoneley waves, assumina viscoelasticity, are then easy to derive. No other fundamental phenomena other than propaption of these waves in viscoelastic media need to be taken into account This was not the case in poroclastic media where fluid flow at the interface has to be taken into ac:a>unt. More theoretical details on in~ waws in vilcodastic: media can be found in Borcherdt (1971, 1973).
,..,..._,uld
. ~--.
298
WAVES AND INTERFACES
6
6.3 GENERAL.CONCLUSION Two types of constitutive laws for rocks have. been cJ,iscuss,ed in this boQk. First we have considered rocks as a combination of a solid skeleton and pore fluids. In this case, displacements both in the. fluid and in the solW skeleton were considered (see Chapter 2 and Chapter 6. Section 6.1 ). But the extreme complexity of a porous medium makes this approach too simplistic in most cases. The macro~c behavior of. rocks could not be obtained from microscopic laws. Attemativ~ly, we can consider an equivalent homogeneous material which WCJuld react to acoustic waws like the rock itseU: Linear viscoelasticity is an appealing model since it ~nts the e«ect of fluid inlide the rock and it is relevant to the small strain amplitudes encountered in seismic exploration (see Chapter 3 and Section 6.2). For wave propagation in~~- and for the Creque~~ ranae relevant to f~eld measurements, macroscopic fluid motion with respect to the matrix has a negligible Cft'ect (see Chapter 2). Rock behavior can be modeled effectively by viscoelasticity. The anelastic part is then quantified by an intrinsic quality factor Q.,e· In the presence of interfaces such as a borehole wall, macroscopic fluid flow is no longer negligible. At the mud/borehole interface, the free surface conditions ~e important, and fluid flow effects modify the wave propagation. Viscoelastic phenomena also add to macroscopic fluid flow effects. Therefore, to interpret acoustic data in a borehole, it is necessary to separate the two ~ts. To the r1rst order, they can be considered independent:
Qf..' .• flM:, (2) + Q;. t
~.
~\
(6.106)
This formula is simila,r:to the one obtained by Goldberg et al. (1984) on simulated data for P wave refraction. In their example, the total apparent attenuation is approximately the sum of a borehole iluid. at.tc;nua~p ao4 &Q .attenuation ,ip the viscoelastic medium. This observation implies that, even t()f p waves, there is Sonic tluid interface effect. The detenaination of~ aUQW4 a meaaure .of intrinsic ~~ty and mud cake effects. The determination of Q;;, 1 , which is an intrinsic; parameter,· aliQws a measure of porosity, porous structure and saturatillg fhdda·of the rock. If both terms can be obtained independently from the data. then several properties of the rock can be understood. .It is, then, essential for the interp~ of~ acoustic·loss to model the porous media as a viscoelastic Biot medium. Unpublished experimental results (Rasolofosaon, personal correspondence, 1985) support dw1.hypothesis (6.1 06) and show that Q;. 1 sheuld not be neglected in the interpretadoft of the data. In situ, it may be diffiCult to oBtain ill4ependefttmeuom of Q; 1 and Q;;;,. The double effects of attenuation may then become an *tade practical ,apptieationa. :,.
m
~,
Q.,,.
(2) is delmed from attenuatiotus in Yi~asticity. This def'mition is·soalewba1 abusive: this quality factor cannot be related to the other deftnitions, as was possible in viscoelasticity (see Chapter 3}.
.......
'-~
''----' "----'
7
some applications in petrdeum geophysics
INTRODUCI'ION (
In the previous Chapters, we lhoYt'!DCl the extreme compleuty of the relationship
I
between acoustic waves and porous media. Both velocities and attenuations vary substantially with porosity, ftuid contont, temperature, and pressm, and it is not always possible to der1ne the boundaries between the inftuence of the different parameters. Hence
I
.._
!I
~
' 'I I
'
~
~
"-...--
.,__ I
I
~
I
II
I ~ : I
.._
..._
'-~
it is easy to realize the dift'JCUity of the inverse problem, which consists in characterizing the porous mcdiulll through the chanps occurring in che w.ave traveling through it. Yet this very diffiCUlty is rdated to the amount of information that we can anticipate from solving this problem. The followilla table (fable 7.1) summarizes ~ salient features of the ~elations developed ia tht ~ Chapters between ~· tbnlc major families of parameters, namely ex.terDal CODditions (fresat~e, tempctatuu); the characteristics of the porous medium, and its acoustic ~ (veloi:iity, atteaUation). Of these three poups of parameters, oaly two are iadepeadent In fact. pressu~ and temperature conditions inftuence the~ oftbe porounl'liediUID without truly altering the mechanical propertie!lol'the eoDIIi-.tive ~ Tlaedrec:t of pressure and temperature can therefore be replaced by an eft'ect on the fluid daatacteristics or O$ the pore structure. Faced with the abundanc:e of ~elationships which to discourage any detailed analysis, it is necesaary toproOeed by successive eliminations of variables. Depending on the major application imolved, this elimination assumes various forms. The Simplest case is that of JDCaSurenaeats aimed at the nondestructive testiag of materials. In this case, the external eonditions can be accurately controlled. Knowing the ftllid content, one can then focus on one cbaractcriiUc of the porous medium. For those analyses, the measurement of acoustic properties is rdatively a«uraae and easy, provided the sample aaalyzcd is in direct coatact with the measuring iastruments (as opposed to remote measurement in f~eld . studies). Oae can ~ deaip a laqe number of experimeats to explore a specif1c relationship between particular parameters, keeping the other parameters constant. Nondestructive testing of materials has been developed, mainly based on velocity measurements. The additional quantitative consideration of the attenuation factor can
soem
~
·--
-~--
.
.....___._:.___
)
)
)
,,
•••
-
-
·-
~-
7.1
-~~
--'--~~
~lftellum
Characteristics or
~
••
..
•••
•••
-
-
•
•(?)
•••
,,
Attenuation
•• ••
••
Velocity
averap, ••• strong.
-
------.-.. -------
neaHJiblfl. • sliJht, "
~
Gat/liquid aaturation. Geometric distribution or phases ....
01P (plwc diaiaM ......
Type and physical state of
fluid content
Peftneab_iiity . ~ ............
Porosity •.... • •.•.........
Teatw!e: pn contacts, c:racb ...•.....•.........
-
Comptessional
-
-
• -
•
••
Velocity
Body waves
)
••
•••
•
*(?)
••
-
Attenuation
Shear
Acoustic properties
8tJMMAlY Of' NAIN POROUS MEDIUM/ACOUSTIC PROPERTY INTERACTIONS
TYpe oholid pbua •....•..
- - --------.,,---,-·
IDI1uedce: ("l) indeterminate, -
'
..
-
-
--
-
-
-
Temperature
•••
-
- -
prcaare
Effective
External conditions
~
TUU!
?
••
••
?
-
-
Surface waves
.,
'---'
10MB~ . . . . . . . . .
-
MIIIIJIICII
''----"
only help to improve future testins. We give below two examples of possible applications for petroleum-~ &abo..,...:
• We showed (in Chapter 5) that microcracb. under low effective stress, substantialy
affect the quality factor Q of a porous medium. Thus attonuation measurements on core samples from a well can serve to determine the orientation of the micrcxneb" anisotropy, and consequently furnish data on the state offractur'iq of a reservoir or a description of in
--~
situ streslel. • We also pointed out how the velocity and espec:ially the-attenuation of ultrasonic waves could vary as a function of ps-saturation and the distribution of this ps for a given saturation. This. ofrers a much needeed analytical method to study twopbue flows iD porous media. ·
or
aeometric
In this Chapter, we have decided to focus our presentation on applications iD petroleum aoopbysica. Firat .- all, the taraota ill ........... IIOPhYtics as aftoa far from the ll.lC8IUiiJII aad rer:ordia& IJiteiQ. It il .~ - M Y for &be acoustic waves tr&Dimitted lO adeq-.Jy .,...U.te tM IU~ 'l1lil ,...uatioa depth illimite4 by the IDCfiJ lOIIIIIUitained by tho wa• as ia tile~ (eaer&J lolaes which we have called the iDtriasic ud utri1tir .........uo-). For deep penetration. attauatioa lll\llt be smallaad the IOUNC ~tow. Oadle Giber baed, for louin& IDMIUia~•• which thc ••riaatlltcm. tile aco.UC frequcaciet ar:e hiaJaer. Tile follewial table(Tai*1.l) liltl._i_..,.of,.......ofthc wavclcaltha used in · the dift'ereot IOiamic tedmiques.
it,....,....,. taqet.,.._ •.,_ •• .. T.uu 7.2
5uMMAilY OF
.........,_
:~~e..
-~
.I ''----"
'j
lI I
,.,.., ..., ....
DIFFiaENr. ~ ~·,,··t.IIIO IN ioaAcrtcAL AI'I'UCA~ ·r. '
•x JOl
•
•
14~\~';
10'
ID
, .... ·~IMIIII:.
,,
~· .....
..
........... plelpictlii ,,
.....
•• ..~a
.,.
w
; . .....-c ..... .
4
xto·•
.. ,...... ....... 106
t.ow••....Y ...... i,jla,_·
Two major aroups emcrp, separated by neatly two orders of mapitude in frequency:
(1) Low-frequency seitmie prospectia& (arouad 50 Hz) il intended to detect iDterestina aeoloaie&l horizons under sedi~ ........ea ap to ...,.a •tilol8etcrs. This technique, whose main application ia the aeometric description of the subsurface, is ~to . . for tMMIOIIIiC ............ oftbiaW. widl respect to
..a,-,
the~~
(l) Hi.....,..1HIIlCY-.. proepe;ti•a (a w Hz)u..., fOf .....ICIDIIlta ill well$, .-...a at die puUcular WI ef ........., MCl it very siiDilar in tcduaiquc to. eleelria8lloaiDJ. 1111oWIII-. . . .tiwlJ . .y to . . and aM resolatiGJl is much bctecr tbata low-freftun.cy Pl'GIIMIDtiu. However the rook vohlmes·teated are lialited arouacl tbc wellL
''---'
302
SOME AM.ICAT1i0NS'IN PETROLEUM GeoPHYSICS
7.1 7.1.1
7
LOW FREQUENCY SEISMIC PROSPECfiNG
Geaeral
To analyze the changes undergone by an acoustic signal traveling through a bed. it is necessary to be able to distin,&uish between the reflectiom at the top and bott~ of the bed. In other words layers ofminimum thickness clo~Jo the wavelength (- 20-30 m) must be distinguishable (Widess, 1973). Therefore limits are imposed for interpretation of largescale exploration seismic prospecting (conventional seismic prospecling> in which the large wavelengths employed are not sensitive to the reservoir beds most frequently encountered. Hence any increase in the maximum UDble frequency in the recording enhances the knowtedse of the subsurfaa. nus is adlN\Ikt in high-resolution seismic prospecting. in which frequeli<:ies in the 1'11DJC of lOOHz'llfe 'still ob8ervftle after propaption in the subsUrface. It is ob\lious that tile illvestiptiOn depth Of hi@h-reselation seismic prospectint is less than that of~· seismic prospecting, but hip resolution is essential f~r an accurate iderttikatien fJf the JitholoJy: of a thin bed or· of a sha11ow reservoir. The fleqUeacy eentelltd' tlt6 teeerclochignat Clln also be mcr.sect tJy limiting the distance traveled in the subsurface. By recording in a borehole, vertical•a«d multiple offset seismic proftles·permit shOJ!ter'trav.d·patblfo neorders inside the borehole. Hence the technique of reservoir seismics emerged, very dose to. conventional seismic prospecting in its methods of measurement and analysis, and whose initial purpose was the calibration of seismic profiles. Its bet~er def1niti~ makes it easier to acquire high ·· resolution data about the' acoustic properties orreservoirs. These problems of ticaleean be partly limited in certain cases by using the-knowledJC of the interfaces obtained from Chapter 6. Whatever its frequency, a wave i$'icll:cted ®m discontinuities of elastic or anelastic properties, for instance when travetm, from a gassaturated bed. to a liquid-~~ bed: ,_.. rMion can be detect~ even ·it' the waveleaath bed thickness ratio is less than favorable (Widell.. 1973). This variation in the reflection, ·as a function of vti'hltions in • or~ .for exalftJ'Ie,"elln provide valuable data about the interface. w~, abaU provide an ~xample for bright spot characterization in· the followifti·Sectioa: ·
7~1.2
The··expNSiiott c.»tmtntioaatMilmic prG~p~Cting applies to tdeetion seilmics used· in petroleum exploration. As a f1rst approximation, the soarcaand receivers an located on the ground ,or at sea leveL Itt ~ seismic piOipeeting, ()Wing to the tarp number of shots and geopbOMS,· several amvals ant ftJCOI'ded for tbc sanle ·rellectioa point correspondillg to daerent transMitterlftleliver distallces (ol&ets). The Ule of. the first arrival times helps to 4otermine the 'Velocity aad depth of each bed (see, for example, Waters, 1978 or Grau, 1985). As a rule, seismic prof1les are obtained with compressional
"' ~
' 1
L_~~------------~------------------~--~~----
'-
s6Mi 'AHIU&1isM$<'tif'~ !~<;
7
·-
wave sources. However, a number of sources exist (Marthorw, S wave Vibrators) that enable the recordiq ·of shear wave seismic sections.
7.1.2.1
'-
303''
Ca. .laed .aeof P .... SWIMS
The combiacd usc of P and S waves allows the measurement of the P and S wave velQCitics for a given ~ and hcn4::e the detcnnination of the Poisson's ratio. As we showed previously (Section S.2.2.6), this ratio provides information about the type of formation traversed by the wave and helps to limit the problem. Nevertheless, to obtain a more thoroush knowledge of the litholoQ, it is necessary ,to employ more elaborate techniques, which no longer Use only the information given by propaptioa time.
7.1.2.2 Sipal .....ysis
·-
''-
The mea~Un~~Deat of.Mwh velecitiea ia reOectioa aacl refraction prospectiaJ requires oaly tJae kaowJedJe of ·the arrival tiJDel ol tbe ~ lipb. Aa we have shown (Chapcer 4), a sipal wbicb ptopaptes in an .uenuatins 'JJOilOUI medium undcriJOCS. cl1aaaa in spectral COBteaC, 10 that the fGna of tbis dwale ia tbe lipatare of the io.-actioa between the wave- and a poroua medium ol given :properticl.(porosity, permeability, saturation). In face, a wave ~ ia the subsutfloo'uadclrsocs other typos of tantformatioa, cveo. ill an elutic 111111-iilfmite lllldiUJL . For a wave witlaanoa-plaMwavetn.t(aplawi&W or c:yliadriallf'otaa•ple}, the wave amplitude decreases with propagation distallCC. The total cnerJY, proportional to the square of the displa<:emcnt amplitude at the souro:, is distributed on the wave front whose area increases during propagation, and hence the displacetWDt ampitu4,e decreases. . Thus, for a spherical wave front (!of i~~ an explosive soun;e), the. total area of the wave froat is 4d'1r1 at a given time t, for a ~ of the medium V, ~ beli<:e the wave amplitude per uait area is proportlonal to 1/t and thus ~ :Withdme.'t:tiis etrect is known as ~di~.ltsipUWlytkenall t~frequeQCies. tt~wa~ no loa,Fr
propaptes in a sinsle. ~•.b\Jt iit·.~·.series fl·a~· btcl$, tll,e.~ _transmitted undergoes other changes.~ CbanFI ~teto traasmission and1'eftection i.Mchanisms at the different interfaces. Many iavcstiptioas on fteld seismocrams have revealed that these processes affect not only the amplitude, but also the frequcaey. c:oatent of dte tr&QS!Dit&cd •ve (O'~y a.od AMtc)', 1971, -Sdloqaberpr and ~ .~974). Other more com~ pr~sucb as ac::atteriq. . . . dcct the s.,ctral abape or,~ sipal received. · Two types of attenuatioa caJ1 be disti~ The tint, which we sbaU ~ intrinsic atteauatipn, has 10 far bcp.our maiawbjfa diacuMion.lt clw~ the aaelasticity of the material the wave' has thro\lp._ t1lc second. v.:ltidlwe Shall cafl extrinsic attenuation, isclw'ac~ ofthe-pomctr)' of'~ su~ue and Qfthe soUQ:C.lt has two etrects oa the seismic sipal; an apparatt ndueliOn in amplitude (the missinctner&Y is not lost but has beea delayed ~ale to.. tbe . . _ , peth.s traveled by the wave). and a frequency fllterin& dl'ec:t for mul&iple rc~Ject.ioas.. siuai~N. to .~ dJect of iatrinaic a~ten~ti9n (Schoenberpr aad LeviQ, 1974. M.odol et Gl..lt$2).1~ fact, it is therefore vc[y dift'tcult to separate intrinsic: and extrinlic aUCB~~ b ~pte Spe0ccr et al, 1?82).
aonc
or
<-
-·-
(tl ·~
b&istered lfllde..mark of lrrstitur F~ts tlu Mrilte.
304
SO,_E
APPUCATIO~
IN
PET&O~UM
Gl!OJ'HYSICS
7
a. Seismk #rfllicr.,¥ Seismic stratigraphy uses the fact that a Jivca ~ ~Cquence (altematina shale/sand, delta deposits, seabed slumps) corresponds to a given sipature on the seismic signal, making it possible to obtain qualitative li&hol~·te. 1lae IUin iatfation is to use the extrinsic attenuation of the signal. In fact, a cyclic ~ namely one consisting of a regular success..ion of beds (fo.r instance sha.le and sand~'ons),. possesses a clearly defined spectral signature. The low frequencies, those whose w gths are tona in comparison with ~ thicknesses, travel as if they were passing thr up one homoaeneous medium, whereas the hip frequencies undergo multiple ret1 •ons within the sequence. The resulting frequency content is characteristic of the so· and the sequence analyzed (see for example Morlet et al., 1982). The morphological analysis of seismic sections is iCOJlductcd by means of various techniques of amplitude restitution and signal processjng, such as the aJlalytical signal (Bracewdl; 1971~ Various examples are available in (Tanaer « Ill., 1979, Sheriff and Tanaer, 1979, Lacaze « Ill., :1978, et til., 1979). Note that this tecbuiquc is oaly applicable to clearly defined p · teq\IOftCIS, aDd that it requires priorknowJodpeftbepolouiDtMrepoaiiWCICip · siaceiO\'eralsequenoeseaahave the same sipature. Where it does apply, however, tbia ~ terVCS to determine lhe contours of1be nsenroic and of deposita. Finally. altiMJUih it mainly involves a sublluface geometric effect (and hence extriasic·to the porouuaodiuat), die euminetion not ODly of arrival ~ "- alsolipal shafes is UICiful for the remote lithOlotic determination of the subsurface.
*litemule aadiuey
b. .AIUI/y$1$ Dj /iriglt 6]1DfS In a seismic section, the amplit\ICles ot Cfle retlectioas observed are prally fairly constant, with m~or co~oiu for elects of ae'ometric diveraence. In certain ~ · however, bilher ~ncrgy· refteetioD$ are observed in the form·oflocal amplitude anomalies (see Fi&- 7.,1). 'ThClle ~ c8Ued ~ .,o~. ~ JeOPlayslcists have observed for a lona time-t~~· bN,bt spoti ocCasiOaally identified the pretOilOC or ps at the interrace concerned. ·· · · VelocityCODCI .... The investigations of Domenico (l~4) showocl that tbe refteetiVity of a Shale/Oil sand interface was quid: different from that of a shale/gas' sand interface. The seismic velocities vary when a very small ;pnount of gas (S% or even less) is introduced into the system (see Fig. 7.2~ The reflection amplitude rises liipiftcaatly'hithk·cae. Tbis amplitude is not a linear functioll of the amount of gas preiCilt in the reservoir. Unt'Ortunatcdy, no biunivocal relationship·exisu between bright spots and the preseilcC ota ps reservoir. Some wells drilled on bright spots proved to be dry. The local amplitude auoJnaly results in this cae either from taterat facies variations, or from constructive (or destructive) interference between near tefld:tors. Hence it is essential' to be able to distinpnh betwee~~ bright spots with gas and without gas. To do this. all the traCes·~ for tbe same reflec:tion point (corresponding to different transmitterfftlCeiver distances allCl heD<:e to dilferent angles of incidence) must be used. Recent investiptiOils (Ostrander, 1982 and Backus, 1982) show the importance of a fmc analysis of amplitude variations with angle of incidence. Ostrander (1982), for example, pointed out that the gas sands have low Poisson's ratios
"
'
7
SO~!E
APPLICATIONS IN PETROLEUM GEOPHYSICS
X
...
l
j 1 I,
1.0
e s .I I,
"' "' E
i=
"
•.
s s
e e
e
.r e
Fig. 7.1
Example of bright spot (after Kjartansson, 1979 b).
y
3r------------------------,
;,
3048m
s a
e
~ :s e a tl 's .e ;e
:s It
>f N
IS
2.5
e: ?:
-g
.,..../1
--------ro;;:-./ ., =..------ --
--
.....
1828m..,}
1828m
2
---- --
-,;
>
~"'
-------610m
1.5
"-610m ~~---+----4---~~--~--~ 0 20 40 60 80 100
Water saturation (%)
Fig. 7.2 Longitudinal wave velocity vs. water saturation for gas (solid curves) and oil sands !dotted curves) at depths of 600, 1800 and 3000 m (after Domenico, 1 9 -~).
305
306
SOME APPLICATIO!'IIS IN PETROLEUM GEOPHYSJCS
7
(see Fig. 5. 72 in Chapter 5) (Poisson's ratio;: 0.1) with respect to shale and sand section (Poisson's ratio;: 0.3 to 0.4) and causes the P wave reflection amplitude to rise with increuina anp of ;ncideDcc. He also provides an example. In the case of a lateral facies variation. no uaiform ¥aria1ioB ill ttfllcUoa amplitude witb an,te of incidence should be
observecl. since tbe Poiuon~s raa of tbe different;becb are similar. Another approach. lot ~tin,a,.s eft'ectt in briabt Spots.~., comparing the P and S wave recor(Jinp correspondiafto the saJDe profde. Since S waves ate less sensitive to the presence ofps tho Pwaves,a,.. bript spot visible on the:Pwaveteetion will not appear on that of the S wave. Naturan,Y. inthe·c:ase of a lithologic variation, the bright spot will exist on both P aactS wave -'iOns (Ensley, 1985).
-'
A_._ oa•1rr11r
The invcsti,ations so far d~ wore only concerned'. with the analysis of the properties of seismic veloc:ilies.Yet wbow (see SectioQ $.1.1.2) that the P and S wave v~are relam-elyml...._..ltr.&bt~ ofpsforuturations~from 5 to 95%, and thereforeCODSlitute. . . .v.IJ.fODrsaturat~ i.,._(Ors. By~ acoustic attenuations are biJhf1 --~ to ~ differen~ saturations (see Sectioa 5.1.2.2), and attenuation readies a peak for a ps ·sa.-atkm from 20 to ~4. Bourbie aod Nur ( 1984); showed that ~ waves exertea:cwo au.uarion e1rects. •·t:ft'ect on propaption and an e«ect at tbc iaterface (Section 6.1.'.2).: The p~ of an attenuation contrast at an interface ~ the roftection....... which woukl haft been aecordcd in the absence of attenuation coatrast 'l"hh ..pitude of dris elfec:t, depends 011 the absolute value of tbe .. elastic.. Rflection~ ~~ atteDuation).(scc Fig.. 6.9).1'bit attenuation contrast is only ob~rvable on the amplitude at angles of incidence smaller than 20 to 30". Moreover. attenuation corttrasts act on the phase of the reflected signal, altering its behavior as a function of anate of incidence. In the purely crlastic case, a reflection changes polarity abruptly after ~ing through zero, whereas, in the attenuating case, this change in polarity occurs as a cqnge of phase (Bourbie, 1982). Figure 7.3 gives an example of this behavior. A detailed analysis· of ttte behavior of relections as a function of angle of incidence is also necessary. tot in situ atten~ation contrasts, the measurement of the absolute value of intrinsic attemultion is no longer relevent since the relative variation of attenuation at the interfac:e suftidently describes the effect. This can then be implemented easily in synthetic seismoarams (see for example Jones, 1983).
Conclusion on bright spots For the analysis of bright-spats by velocity and attenuation measurements. we have shown the need to study the variations of reflections as a (pnction of offset. To do this. it is essential to have a high sipal/noise ratio and to work on unprocessed data. Few analyses ofthis type have yet been conducted to determine whether it is truly (statistically I possible to differentiate between bript spots_with and without gas. Nevertheless, the few results obtained so far are promising.
.,,
'
_, ~.
I
l
~,
--.,
7.1.2.3 Three-component sa.dle$ Conventional seismic prospecting generally records a single component of particle displacement at each rec:eiver. Recent investigations (i.e. Prunier, ·1981) have demonstrated the value for onshore seismic prospecting of a simultaneous recording of the three
'f
'1
~
j
J
' "' ~
SC>~i ·j.,.:t:it.i~'N$' 1w ;PEfi&l~M ~Jdft.vlfkS ·
1
301
Offset
• I ; II 00
•
s I
-~~
..
I
'
I
I •
_l_
f
.....
..
~
··mfi ..~;
1-r• ~
I
:
~~~·:i: (
-.:;! . .
I
·r •
.• ~-H-tt+f--t-t-lt-++t+t-t+4H::Tt-
L~~· . .,. . :!"t- ~ . : : ,(~:~:n ~ :-n~:r ~ ~
•
.. ~
. -~{{·"r ..,.
w~a~-·to·
·-~
~M
-: .•
.. • ···u
~' t{• ... _{
2
~
\{ . ~"SS • .
-~~;_~ -!
..
•
"'
.•
,..•.
~~-t
••
.. .
Ill
, ·' ... t-1 ,
to
• •
t
.
,
•
~
J
~
..-:,..~:
~~.c."·
~> :. ... .
~
,.
..
;~··:.;,a-.
·J
~
~·
·;
... .
..._.. ,~ r~.;
t.{!t:l ,. .... ..
~~- •.
. ~..
.:~.,
~.: ....'.~'-~
..
t-~
::..
~·~Uh.•
~:
.;
..
.
:
,:.:. ...-., ;,:::
": ,!"~~~..
a
: ' .
I
~R
:
&
~
~0
}
•
.~. ..-:.s:: .
"
'!II
•. "' ~
'
.,.
)
": <::"
"j .
;.,.
...
u. •
t
" U_
• ·'---·
.;
!) -~~
..;
..... '7.3 Example of common mid--point gather. The arrows indicate the arrival for which a ~bangc in polarity is observed with offset (in Bourbie, 1982, after Kjartansson, 1979·bt. ·
~~
compo~nts .·
of the velocity vector. However the application of three-component reoordiQp to two- and three-dimensional Seismic prospecting requires an understanding of the propaaation m~hamsms iq the weattlefed zone (WZ). This undercompacted and under~nsolidated zoiie displays hip attenuations(~ ~ 10 to 20) and any interpretation of three-component particle l'ilodon must takt thc$e attenuations and th~r anisotropy into account. Recent examples (Dubesset nd tliet, 1984) reveal differential behavior
.'--
----A
308
SOME APPLICATIOSS II' PETROLEUM GEOPHYSICS
7
between the different components of velocity, which is not yet fully understood but appears to result from differential attenuation. Therefore the future of experimental threecomponent analyses should provide a better idea of the anisotropy of the propagating media.
7.1.3 Rese"olr seismies
* and hence higher This expression is applied to a hisher frequency('> 100-.lSO Hz) resolution kind of conventioaal seismic pros~ J»caetration is shallow, and the dcimition ia reservoirs. application of this method mainly adlio\ra bed« The frequency iDcrcase c:an be obtained by wttJcaJ SeisDaic profiliaa (VSP). For these profiles, the SOUI'Ce remains located ()a tfte .JI'Ound, but the receiver is pia~ in a borehole right below the source. This ~- ope travel path throop the hi,lhly attenuating weathered zone (WZ). However, Iince tbese .travel paths ate subvertical, tbe zone analyzed is limited to the formation doSe te .~ well. An ofsct VSP tcdmicp.re was developed to overcome this drawback. in Which the s~nirte is no ~aer near tbe borehole but offset by up to a few hundred meters~ In this. c;asc. the zone analyzed iJ cxtend.fd to formations that are further from the borehole (see Fit. 7.4~. ·'
-...uc
Fig. 7.4
Diagram of an offset VSP.
In VSPs. velocities and attenuations •re easier to measure than in standard seismic prospectina. Measurements can ~ QlJde iteratively relativ~ to. depth locations. Since the subsurface JCOII\etry is well defined, it is possible to determine the extrinsic •ttenuation from total attenuation. This m~es it much easier to obtain. infqrmation about the formations and their fluid content from signal analyiis. ' .
~
..
~-----------------------~----------------------------------------------------------------------------~------_J--------~ ~
'--'
., 7.1.3.1
309
Variation in fluid phase or pore
Dllllttre
a. Ste11111 jrtHII tUt«:titHH TltemeasURme~~ts presented in Chapter S showed that the liquid/ps phuecbanp was marked by a sipiticant chansc in velocities anchttteauations. Fiprc 7.5 siva an example ofa tiekl recording obtained by McEvilly er .,.(1978~ Thueianicdata COilCCrninga single event were obtained at diferent stations. The station located immediately nat to the caldera (station 3) shows a 10\\• ratio of P to S wave amplitudes. This is similar to the experimental results by De Vilbiss (1980) and c:orresponds to the minimum of the water/steam transition (sec Fig. 5.40). This is not an example in a reservoir, because there the procc11 oec:urs at a much laqer ICile, but it IUJIIIls UICful. results that ·can be aaticipated for steam front mOnitoriq.
~-..._/
''-·
~
·~
...___
,.... 7~ Seismic RCOfdiq (Mexico) abo.· • . _ oC a celdera. Tbe aour<:c is at a diatucc_ .of about ~SO~. ~ numbon rise with increasina distance from the caldera (after MoEvillY, et ill., 1978).
'"---'
·~
·''-
~-----
Nur0912)SUUOSfOdtileucofaetwocbofroc:eivendurinasteamOoodinaoranyothcr enhaJK.Od racovcry operatioa. inorclcr to4Cler~Pinc ~ ud attenuations within the reservoir. The monitorin& of ,-ariations in velocities· and attenuations would otter a knowlcdp of the gas phase distribution within the raervoir. A procedure of this type, adapted to an undc!JrOUnd gas storasc problem, is discussed by Blondin and Mari (1984).
1». A.IHtonncl reunoir JWe.,., We have shown that velocities and attenuations depend on the effective pressure, namely the dift'erence · bet\¥een CODAllins ~ (litbostatk) an.4 J10re pressure {hydrostatic). In the cMe of Jrilh-pressUte raervoin, the bi8h pore pressUre couatcrbalaaces the confiDills prts~t~te yie14iaa atmMIUIIIow I' and wave velocities and abnormally high attenuations. The determination of the acoustic parameters or their relative variations at the resen·oir jaterf~. &:Ould thus help to identify. hisb·pressure
s
·-·~
zones.
310
SOME APPLICATIONS IN PETROLEUM GEOPHYSICS
7 .1.3.2
7
Analysis of fractured soaes
a. Nlllurtll fr«turi~~g We discussed in Sections S.l.l.l and S.l.2.1 the int~nce offrac:turina on velocities and attenuations. In a saturated rock. fracturina reduees the velocities and increases the attenuations. This is one reason why fractured zones are acoustically distinc:t from the surrounding zones. Other processes are also involved, such as wave scatterina by the fractured zone, if the transmitted wavelength is about the same as that of tbe fractured unit block size. Well·to-well seismic prospecting is the ideal tool for analyzing these shallow fractured zones. Fractured zoaes arc more eeoustically vitiblc with lower effective pr~un:. in other words at a given depth with increasing pore pressure. This may also cxplaia die hiah amplitude reflections sometimes observed in deep seismic prospecting in crustal tectonic zones. If the permeability of the rock surrounding a fractured zone is sutftciently low, the fractures may preserve overpressures over long time IC81es (see for example tbc Wind River Range overthrust (nearly 30 km_deep), Jones, 1983). b.
-----.
''----
MollitM1116 ofltytlralk jiwt:tlrilg
The technique of hydraulic fracturina to improve hydrocarbon recovery in a given reservoir involves injecting a pressurizeG Ouid into a well to fracture the reservoir rock. These pressurj2ed fractures exert a 'tiplacaat effect on the transmitted waves. and also generate reflected waves due to the contrasts in acoustic properties. ReOections are generally greater with ~ waves ~ with compressional waves across a hydraulic fracture, since the aheat moduli aa. ......Y tow; The ideal seismic cocbaique for .....,... ~ hydra~ fr~urina process is well-towell seismic prospectiaa (see Glf~,1914 In ibiS ·metJaod. acoustic waves are transmitted and ft'Jeeived iR two ..ens.. a:h huladrad meten •part. This elimiaates all problems associated with the weathertd'zone, so that various horizons can be observed with relative a uracy. The maximum frequencies received are on the order of a few hundred Hz fo well-t&-well distances 'of 200 to 800 m. By means of wlocity and ·attenuation me urements, it is p0$si~ to delimit the horiZons and to determine the length and strike of the fractures induced during injection. Unfortunately, the well-to-well technique has so far been used very little, due partly to the possible damage caused in the wells by seismi shots. Recently, due t& efrset VSP dewlopments, this technique is reappearing in hysies and sbould. prove its drectivcncss in the years to come.
7.2 I FULL WAVEFORM ACOUSTIC LOGGING ~~
seismic prospectins in boreholes is a broad term for sonic logging with analyses. The $ODic sonde, up to 15 m long (like the EVA tll tool) has ers and receiveTL B9th tr~milllion and reception take place in the
---..,
(2, EVA, Evaluatlon or Velocities and Attettuations, Elf-Aifllltlllne trade-mark, has 5 traftllllitten and 12 recetvers. -r
'
' ~
'
--.--,
-::'-
1
311'
SDM1 'AiftlcATIONiffM ftfteta!IM CIIIIMV..S
borehole at frcquenacs of 10 to 20kHz. After propagation through the formation and mud, the full waveform is recorded. Two types of waves can be distinguished. namely the two waves refracted along the borehole wall (P and S wave). and the surface or guided waves due to the presence offluid and to the cylindrical geometry (see Fig. 7.6~ In perfectly elastic media, P and S waves are non-dispersive, whereas guided pseudo-Rayleigh and Stoneley waves are dispersive. An example of a recording obtained with the EVA tool is shown in Fig. 7.7. which reveals t~ coinplexity of a recording of this type. In some cases. neither a refractefl~ ~vc aor a_ Stoncley wave is obeened. while. in other zones. a refracted wa...a ~ ... aoStoaeley'WPe; Muy theOretical studies have been conducted to illaodet ........ ~ ~don iD .~ media (see for example Rosenbaum. 1974 or . . . . 6.·ll. ~fl• dloiU&:~ for Qan1ple Cheng and Tok.OZ, 1981 or~,a~.;l-~ ..... ~also ~used to test the validity of the . . . (!"i~*d:E~ ltJ3.adChen. 1984).
s
·I
.
..
,
.
;i' 1
jr,·~ ~·.,·~ .,;r1' ~~·
:·.. .:-. i
li' . '
I
~
,..., .......
ii
'.~ ~
.
'"" :.....1:-....,...... .....
·~
.,· . . . . . . . . . . . . . . . . . . . . . .
n.T.f
~.F
Wa•~>~~ an -~c tool~ the EV~ type. .l
~~ the ~boratory
Experimental .-esults ud in situ show that it is Q/wi/lys possible to obtain the compressional velocities v,. and $bear velocities Vs of the formation, by using either a standard source (as in the EVA tool) or by using a dipole or quadrupole source such as in the Exxon tool (Winbow and Rice. 1984). For the time being, measurements of attenuation Q,. and Qs are not routine, although interestins results have been obtained in some speciftc cases (Goldberc et al., 1984b, Huang and Hunter, 1984, Mathieu and .
312
7
SOME APPLICATIONS 11' PETROLEC\1 GEOPHYSICS
..
0
~ r;::
"§
~
·:; <::-
-.::: :::-, t.
"'
~
::: 0
~
8 ~
c:
l
.2 00
.
~
..,"'
.c
c.
0
c
6
.~
~
< > LLl
0 0
0..
E Oil
>< LLl
~
r-
r-.:
~
~
8 :!
~
!:!
0
N
•
. (S / W) aW!.l
..
5!
~
<:-
7
SOME APPLICATIONS IN P£TaOLIUM OEOfHYSICS
313
Tok50z, 1984). These authors clearly point out the relationship existing between the amplitude ratio of P and S arrivals (Goldbcq et al., 1984) and the Stoneley wave amplitude (Huang and Hunter 1984, Mathieu and ToksOz, 1984). and the fracturinc and hence the local fracture permeability. These results should not surprise us. In faa, they agree completely with the conclusions of Chapter 6, showing the importance of attenuations (and hence amplitudes) of refracted S and Stoneley waves for the measurement of a permeability that may not be due to fracture permeability only but to matrix permeability as well.
The relatively short wavelencths (a feW dozen em). the knowledce of the P and S wave velocities, and hence of the Poisson's ratio, and a certain approach to permeability, make these techniques the most promising of all those that have yet examined. It is in this type of approach that a sound knowledac of acoustic problems in porous media is essential to allow maximum use of recordinp of the type shown in Fig. 7.7.
we
<~
'~
bibliography REFERE~CE
BOOKSUI
Ac:hmbach. J. D .• WD~ propogtllioll in «Mtic solids, Notth Holland. Amsterdam, 1973. Blud, D. IL, 1M tlwory of lbwr riscotltutlt:lty, Perpmon Pless, New York, 1960. Ccrveay, V., Molodtov, I. A., PSCDCilt 1., Ray IMthod ill uismology, Univ. Karlova. Prape, Czechoslovakia, 1917. Dullieo, F. A. L., Pof"fnu nwd/11: jiJiid trfiMIIOI't tllld por~ stf11Ctlll'e, AcademK: Press. New York, 1979. FunJ, Y. C., FOIIIfdDtiOIU of solitJ ~McluMics, PmltKe.HaU. New York, 196S. Miltlowitz. J., 1M tlwory of ~lastlc WfWU lllld K"tlrr p;Ms. Nonh Holland, Amsterdam. 1978. Nowick, A. s.. Berry. 8. S.. An tltlstic r~hvcDtion in crysuii/IM solids. AcademK: Press, New York, 1972. Schreiber. E., Anderson, 0. L.. Sop, N., Eltutic constants llltd tlwir IMliSIN'.,..,.,, McGraw Hill, New York,
1973. /nJil6~ QIIQ/ysis lltfllm~~themtltiCtll m~~rplrology, AcadanK: Press. New York. 1982. TnaeD, IL. Eltlaum, C~ Chick, B. B., llltrwMMic,.dwJd,r in MIJ sltll~ plrpics. AAdemic: "-.New Yorlt, 1969. Vilttorov. I. A., Rll.t·/rigll tiiUi 1.-rb -..ts, Plenum Preis, Nev.· York, 1967. Waters. K. H., Reflection uismology: a too/for energy resource exploration. Whitey lntC!fiC. Pub.• New York, 1978. White, J. E., Ulftlngrotllfll6(111111/: lfillictuion of uismic K"D\YS. Elsevier, Amsterdam, 1983.
Serra, J.,
REFERENCES Abramovitz, M., S~.l. A., HtiiWiboolc ofmt~tlwmtllicfllfi/llctiotts, US Govemmeat Printin1 OfCK:c, 1972. Achenbach. J. D., lfa~ [lf'OINI6tllloll ill «•tic IOiids. Norda Hollaad. Amst4ltda1ll, 1973. Alldrieux. S., "Un JDOCiilc de matiriau ~: applicatioa aux rocha et aux bCioal", Thais, ENPC, Paris, 1983. Anpheister, G., PhysicQ/ propertin ofroc/u, subvolume b. Sprinaer-Verlaa, Berlin, 1912. An:hie, G. G., .. Elecuic:alllllislivi&)' uanaid iDeo~ analysis iDtapmation", TrlfiiS AIME. I"'· S4-63, 1942. Alditty, P. C., AJwns. G., Staron, Pia., "State oftbe art,.Qil EVA data pnx:essiq: an improwrment iD subsurf.ce imaamg•·. presented at the 52nd AlmwJ/Int. SEG Jleetillg. Dallas, 1982.
~~-
Auld. B. A., ACOIIStic /abllffll-NS in solids, VoL-~. Wile)"lnlcl¥., New York. 1973. Azimi, S. :\., Kalinin. A. V., Kalinin, V. V., Pivonrov, B. L .. ··Impulse and transient characteristics of media with linear an4 quadratic absorptioa laws". Physic. Solid. L;uth (Eiljlish Ed.). 88-93, 196&. Bldtus, M. M., ''The reflection seismoJram in a solid layered earth", PaperS 16.S, presented at SEG -~ring, Dallas. 1982. Bec:quey, ~-. LaverJDe, M .• Willm, C., "Acoustic impedance lop computed from seismic: traces··. G«Jpptys.,
'~
316
BIBLIOGRAPHY
Bensoussan. A .• Lions. J. l., Papanicolaou, S .. As.mlptotic Amsterdam. 1978.
ana~rsis
for periodic structures, Nonh Holland.
Berryman. J. G .. "GoDfumation of Biot's theory", Appl. Phys. Lett .• 37. 382-384, 1980. Biot. ~- A.N "The theo~· of propagation of elasric waves in a fluid-saturated solid, I lower frequency range ... J. Acoust. Soc . ..fm., 28. 168·178. 1956a. Biot. ~- A .. "Theory of propagation of elastic waves in a fluid-saturated porous solid. II higher f~equency range". J. Acoust. Soc. Am.. lB. 179-191. 1956b. Biot. ~-A .. "~fechanics of deformation and acoustic propagation in porous media ... J. Appl. Ph.•·s. 23. 14811498. 1962. Biot, ~- A .... VariationallaJ!'angian-thermodynamics of nonisothermal ftnite strain mechanics of por"us solids and thermonuclear diffusion", Int. J. Solids Structures. 13. 579-597. 1977. Biot, M.A.. Willis, D. 0 .. '.'ThcclasticcodTJC:imtsofthcory of consolidation", J. Appl. Mech., 5~1. 1957. Bin:h. F., Bancroft, D .. -EJasti&:ity and iat~ fJictjon iaa loq column of...Ute" • .81111. S.ilm. Soc . .-em., 38. 243-254. 1938. Blair, D. R .. Spathis, A.P .... Attenuation of explosion generated pulse in rock masses". J. Geoph. Res.. l7, 38853892. 1982. _Bland. D. R., The theor.t· of linear viscoeltJsticit)·. Pergamon Press, N.Y., 1960.
-~
Blondin L.. Mari. J. L. "DeiAICtion of pz bubble boundary mo,•ement", Abstract, 46th EAEG .\feeling. London. 1984. · · Borcherdt, R. D .. "lnhol"IIOFftesous body and surface plane \\'liVes in a pneralized viscoelastic half-space». PhD. Thesis. Ulliv. Cal. Berkeley, Calif., 1911. Borcherdt. R. D .... Rayleip-type surface wave on a linear viscoelastic half space", J ...fcoust. Soc. Am.. 53. 1651-1653, 1973. Borcherdt. R. D .... £~· and plane waves inliMar "'iscoclastic media" J. Geoph. Res.. 71, 2<442-2453. 1973. Borcherdt. R. D .... Reflection and refraction of type II kaves in elastic aad anelastic medium ... Buli. seism. Soc. Am.. 67, 43-67, 1977. Bourbie, T .• "The effects of attenuation on reflections". PhD. Thesis. Stanford Univ ...Calif.• USA. 1982. Bourbie, T.. .. Reftection coefftcient at normal indicence between tWo YitcOelastic: media". submitted to Geophysical Prospecting. 1986. Bourbie, T .. Gonzales-Serrano, A .• "Synthetic seismoJ!'3MS in attenuating media", Geophys.• .a. 1575-1587. 1983. Bourbie, T .. Nur. A., "Effects ofattenualioaoa reo.aiolls: eaperimental test",J. Geoph. Res.,l9. 6197-6202. 1984. Bourbie, T., Zinszner, B .• "Saturation methods and attenuation '·ersus saturation relationship in Fontainebleau Slllldstone", 54th SEG Me,ing, Atlanta, 1984. Bourbie, T .• Zinsmer, B.• "Hydraulic and acoustic: propenies as a function of porosity in Fontainebleau sandstone", J. Geoplt. Rn.. 91, 11524-1'1532, 1985. Bousq~. P .• ''Texture et porosite de roches catc:aifts''~ Thesis. Mines c1e Paris, 1979. Brace. W. F., "Permeabj]ity of cryst,alline rocks: New in situ measurements", J. Geophys. Res., 19, 43 27-4330. 1984. . Bracewell, R. N .• Tlte Fmirier transform altd irs tJPPikations. 2nd Edition, McGraw MiD. New York. 1978. Brutsaert, W., "The propaptioh of elastic: wave5 in ililCQnsolidated unsaturated aranular m«tiums •·. J. Geoph. · Res .. 69, 243-25-. 1964.
'
Budiansky, B., O'Corinell, R. J., .. Elastic moduti of dry and saturated cracked solids", brt. J. Solids Structures. ll, 81-97, 1976.
------..._,
Cagniard, L.. Reflection and refractitm of progressive seismic K'Mes. McGraw MiR. NC"· Yorl, 196:. Cargill. G. S., Radial distribution functions and microg~ometry of dmse rattdom paclcings of liord sp/r£res, in: Physics and Chemistry ofporous media. D. L. Johnson and P. N. Sen editors, American Institute ofPhysics, New York. 1984. Carman, P. C., L'koulement des ga: a travers les milieux poreu:c, Bibliotheque des Sciences et Techniques nucleaires, Presses Universitaires de France, Paris. 1961. Carmichael. R. S .• Handbook of physical properties of rocks. Vol. II. CRC Press, Florida. USA, 1951.
~.
lt~M
317
C8rmichael, R. S.• Hant6ook ofpltllktll f1'0II'W'#n ofr«b, Vol. Ul, CRC Ptas. florida, USA. t984. Olye, R., Pietrot. R .. Pre'fOIUtu, J. M.,llqot, J.P., MMHaodesde~des milieux poreuxen 'uede leur analyse quantitative au mic:rosc:ope", /Jull. Soc. Fr. Crista/log.• 93, S?r-311, 1910. Cenefty, V., Molockov. f. A .• Psencik, I., .. RayftlCthod in ldiAIIGioiY..,IJmy, X.Wwi. Prquc. Cadtoskwakill.
19n. . Chandler, N., Johnson. D. L., ~Theeqllivalenceof quasi-static flooN in ftaid-uturated porous media and Biot's slow wave in the limit of zero frequeucy", J. Appl.. Pltys.. 51. 3391-3395, 1981. Chang. S. K .. E\·emart, A. H., .. A ttudyofiOftic loaint in cued botebole",J. P~t. Tech .• 1745-1750. Sept. 1983. Chca, S. T., "The full acoustic wave~ ip a laboratory model of a borehole", Guphys.• 41, IS 12-1 S20. 1982. Chen, S. T., "Experimental studiesofon-lineshearwave loain1". PtlfiBS 21.4.54th SEG. M«tillg, At!anta,
l91J4.
a-,. C. H., Toks6z. M. N., .. Eiuticwavepto.,plioaiaaftuid-Mid~IDd l)'lltlleticacoustic lop",
------'
Geopltys.• 415, t042·10S3, 191U. Chcag. C. H., Tokliz. M. N., Willis, M. E., "Dctcrmiaation of in $itu attenuation from full wavefonn aooustic: lop", J. Geopll. ks.. rl, 5477-5484, 1982. Choquette, P. W., Pray, L. C., ''Oeotosic llOIIlendanire and classifiCation or porosity in sedimentary c::arboaa11S", A.APG Bull.; 54, l07-250, 1918. Christensen, R. M., TlrHry of riscodtulkity: ..t ifttrodtlctloti, Ac8demic Press. 2nd Edition, New York. 1982.
T.
Clark, V. A., Tittmann, B. R., Speacer, W., "Effect of volatiles on attenuation ({l) and velocity in sedimentary rocks". J. Guph. ks.,15, 'Sl90-S198, 1980. CoheD-Tannoudji, C .. Diu, 8., Lalise, F., Allctllllqw Henrwm. Paris, t9n. Cooper,M.F.Jt., "Retlectioallldtrl8llllitlion-of'obliqueplanewavesataplalleillterfacebetween~ media", J. A.COIISt. Soc. A.m., 41, 1064-1069, 1967. · Cordier, J. P., us rittsns elf sismlfW Hjfe:dMt, Ed. Lavoitier, Paris, 1983. Courant, R., Hilben. D., M~thods ofmat/wmo!ical physics, lntencience Pub., New Yorit, 1962. Dattll, S. K., -viscoelaitic: propenies or • . . . . ~te4,by viseoQs thaicHilled cavities'\ IL,-.-\M, Symposium 011 meclumics of•iscomutlc Me4itJ lllfll bodies•. Sprinpr-Verlag. 19'75. ' Delfmer, P., "ttude morpboloaiquc des milieui pomix et au&OIIII.dsation de. mesures en plaqua mU:,-u", Thais, Nanq I, 1971. Denis. A., Panet. M .• Tourenq, C., "ldcntifatkm des roches par l'indice de continuite~·, Cslrgris lift. Mic. du Rlx:lta, Montmax. 95-98, 1979. De Manilii, D. C., Beard, D. C., DaDbuia,J. S., Robinson. J.H.• ;,Variationsoraalmicvelocities in sand.-t<>QCS and limestones v.ith lithoJoay and pore ftuid at simulated in situ tonditions ", Proc. EG PC Explorati011 Seminar, Nov. 15-17, 1976. ' 1: ·• · Deraicwicz, H., "The effect of boundaries on wave propaption in a liquid ftlled porous solid, IV. Surface waves in a half space ... /Jull. Seism. Soc. Alft., 12, 627-638; t962. DeMiewicz, H.,ltiee. J. T., "Thee«ectoflleuadarieioe._..propaption ina liquid filled porous-solid. III. Reflec:tion of plane waves at a free plane boundary (pneral case)", Bull. Seism. Soc. Am•• 51, 505-625. 1962.
,..,,;qw,
~"
Dereliewicz, H., Skalak, R.; "''n uniqueaets in dynamic ~lastk:ity ", Bull. Srism. Sbc.
Am.• 54, 783--783, 1963. Deresiewicz, H., Jliee. J. T., •111e eft'«t ~boulidarics Oil wave propaption hi a liquid ftlted JlOrous-solid, V, Transmission across a plane interface", /Jull. Seism. Soc. Am., 54, 409-416, 1964. Devaraey, A. J., Levine, H., Plona, T. J., "Attenuation due to scattering ultnuonic compressional waves in lfiDular media", ill: Elastic Wa•e ScGtterillg tllld Pr0JH18atioh, (\'. K. Varadan and V. V. Varadan, edsl. Ann Arbar Sci., A8ll Arbor, 131·141,.1982. DeVilbiss, J. W., "Wavt dispmiOil and ablorption iii panially satutated rocks", PhD. Thesis, Stanford thliwnity, c.iifomiL USil, 19110. . Ditkine, V., Proudnikov, A., Transfomtt~tiolu ifttlgraks et calcul operatiohnel, Mir Publishing. Mosco•·. 1978. Dolneaico, s. N., "EffeCt ofwater•turation on seitmlc: ~ty or sand reservoin encased in shale", GHpltys.. 39, 759-769. 1974. ' ·· · Domenico, S. N., "Effect ofbriae1u mixture on velocity in a UftCOMOiidated sand reservoir", Geoph_n .. 41, 882-894, 1976.
'-._...
318
BIBUOG~PHY
Domenico. S. N .. ··Elastic: propeniesofuncoasolidated poroliJII!Ind mcrvoirs".Geop#l}·s..4l, 1339-1368, 1977. Dubesset, M .• Oict, C.. "Three CIOIIIpQIICilt m:ordi~Jev. interest for land seismiuoun:c: study", Paper praented at 54th SEG meeting. Atlanta, 1984. , Dullien, F. A. L .• Porous nwdia: fluid~~ lllfli.PfJT~ sllliCUift'. Academic: Press. New .Yort. 1979. Dullicn. F. A. L.. Dhawan. G. K .. "Characterization of pore structure by a combination of quantitative photomicrography and mercury porosimeJry", J(JWNJI of Coi/4Nd and intetf«e science. 47, 337-349. 1974. Ensley. R. A .... E\'aluation of direct hydrotarbon indicators through comparison of compressional and shear wave seismic data: a case study oftheMymam Gas Field. Alberta". CWophys., 58, 37-48. 1985. Eringen. A. C.. Suhubi. S .. Elasrodynamics, Academic Press. New York, 1975. Esbelby, J.D .. ''The determination of the etasricfleld of'an elfipsoidalinclosion and related problems", Proc. Roy. Soc. London, Ser. A .. 241, 376-396, l9S7. Etienne, J ~ Le Fournier•.. Application des resioes synthetiques colorees pour !'etude des proprietes de reservoir des roches c a l - minces", Jlrr. lldtiltd.li'rtlttc. tlu PilrtM; U.. 595-629. 1967. Ewing, W. M .. Jardetzlc.y. W. S., Press, F., Elastic M"a•-es in layered medio. McGraw Hill, New York, l9S7. Fatt, I., "The Blot-Willis elutic coeffiCients for sandstbne", J. Appl. 26, 296-297. 1959.
a
Mech:.
Faust, L. Y .. "Seisn,Uc velocity as a function of~~ aeoloP: tiaJ¥:", ~phys .• 16. 192-206. 1951. Feng, D. L .. Johnson. D. L., "High frequency aco~tic propemes of a tluid/porous solid interface. I. New surface mode. II. The 20. rellection Green's function",/. A~ust. Soc. Am., 74,906-924, 1983. Fliigge, W., Viscoelasticity, 2nd edition, Sprinaer-Verlag, New York. 1975. Fung, Y. C .• Foimdations of solid meclumics, Prentice-HaD, New York. 1965.
---,------,
Futterman. W. L.. "Dispersive body w.aves''. J. Geoph. f"s •• 67. ~279-,mll, 1962. I~ Vertictll Hislnk profiling. S«iety of Exp/MIIio#l .Geophy6id6ts,.~ Publication No. 12, Tulsa, 1974.
Gal'Peritt, E.
Ganley. D. C .. Kanasewich. E. R .. "Measurement of abs9rption and ~n from check-shot surveys", J. Geoph;•s. Res .. IS. 5219-5226, 1980. Gardner. G. H. F .. Wyllie. M. R. J.. Oro~ D. M., "Effects of pressure and fluid saturation on the attenuation of elastic wav~ in sands", J. ·P~t. ·rt;lr;, 189-198, 1~. Gardner, G. H. F., Gardner, L. W., Gregory. A. R., "Forma~on nlocity and density. Thediapostic basis for stratigraphic traps'', Geophys., ~. 770-781); 19'14. · 1
Gassmann. F., .. Uber die elastizitit poroser medie~ ", VierteljtJhrssclrrift der Naturforschenden CWsellsclroft in Zurich. 96, 1-23, 1951. ·· Geerstma, J., "Velocity-log interpretation: the effect of rock bulk collij)I'CIISibility", SPE Jour .• 1. 235-248, 1961. •. I • • • . ;
Geerstma, J., Smit, D. C., "Some aspects of elastic wave propagation in fluid-saturated poroUs solids", Geophys., 26, 169-181, 1961. Germain, P., M~canique des milieux cantinus, MIUIIOD, .,_ris, 1973. Gladwin, M. T., Sta.cey. F. D.," Aaclastic 4cF'adation of acoustic pulses iJuQI::k ", Phys. E4rth Pltaet.lnwr., 8, 332-336, 1974.
.
Goetz. J. F •• Dupa. L. L., Bowler, J .. "A» investiption into di~ between sonic log aod ~mic shot velocities", Schlwnb. APEA. 1979.
.
Goldberg, D., "Sonic attcDuation in coasolidated and ~solidatcd sediments", PhD. Thesis. Columbia Univ., USA, 1983. .
Kan. T. K.. Castagna. J.P•• ".1\Ucnuatipn measurements from sonic loa waveforms", 2.5th SPWLA Logg. S,,·mp .. llllle 1984. . . . .. Goldberg, D. S .• Gant. W. T., Siegf1ed, R. W., Castagna, J.P.. "Proc:elsing and interpretation of sonic log waveforms: a case study". Pa~r BHG 2.4.. 54th S£G ~tiJ18, Atlanta. Dec. 1984b. Gordon. R. B., Davis. L. A .... Velocity and attenuation of seismic: waves in imperfectly elastic ruck". J. Geoph. ks., 73, 3917-3935, 1968. Grau, G., "Appon des techniques silmiques a l.'ctu® cJes gisement$ d'hydrocatbures ", Ji- Coli. Europkn sw /'amelioration de Ia recuperation du phrole, Rome, April 1985. Gregory, A. R., "Fluid saturation effects on dynamic elastic: propenies oftedimcntary rocks", Geophy.s •• 41, 895-921. 1976. Goldberg, D. S.,
~
\
~
--~
'-
''--
~
'-
ll9
Gre,ory. A. R., •• Aspects of roclt physics from laboratory aad lof data that are imPQrtant to seism~ mterpaetadoe". ~islrrk mvn~F AAI'G . _ _ , 116, lS-46, 1977. Grim, R. E., Cllly miMra/DfJ·. McGraw Hill, New York. 1953. Ouillot. D., "PropriWs physiques et texturales d'un modele artifiCiel de matmau aponJiite iateraranulaire ", Thesis. Mines de Paris, 1982. Hadamard, J ., ~ :Jflr Ia propt~glllion dn Olflks ~~ ks eq.ations dttl'lr.•'drodJ-i4w. Cbelsea Pub. Co.• New York. 1949. Hardin. B. 0., Asce, A.M .• Richart, F. E., A~ee. F., "Elastic wave velocities in panularsoils". J. Soil Mech. FotWL Di•.• " · 33-65, 1963. Haup. P. S., ''Measurements of attenuation from vertical seismic proftles ". GHphys., 46. I S48-15S8. 1981. Hosten, B., Delchamps. M~ •· Etude de Ia transmission ultrasonore en faisceau bomC d'une interface plane a l'aidedu spect~eanplaired'ondesplaaa",Prlbl. till lAb. Mec. l'ltp.. VA 861. CNRS, Univ. Bordeaux I. 1984. H-.. C. F., H1111ter, J. A., "The ltllleowave llletbod of....... ill titu rock fracbn permeability in Ruidftlled boreholes", G~xpltlrtltion, D, 24S-2S9, 1984. lshialaru, A•• w~ ~tion . - t~t»~t~ 111 rlltfdont lftltdM. Academic: Press. New York. 1978. Jacquin, Ch., "Conelation entre Ia permeabilite et lescarac:teristiques ,eometriques du . . de Fontainebleau", h•.IIUt. Frt~~tt. tiM Pllrt*, XIX. 921-9!7, ""'· · . . Jankowsty, W., "EmpiricalinvatiptioaofSDIIWiilc:torulfectinlelulic:wavevelocitie&incatbonateroc:lts", Geoplt. Proq•• II, 103-118, 1970. Johnson. D. L., "Equivalence between fourth sound in liquid Hcl..t lowtemperatareaadthe liot slow wave in porous media", Appl. Pltys. Lett., T/,1065-1067, 1980. Jolmsoa, D. L., Sea, P. N., "Multiplca:atteriqofac:oUJlicwa!llll withapplK:ationohbe index of refraction of fourth souad", l'l!ys., kr. B .• 24, 2416-2496, 1981. · · . Johnson, D. L., Plona, T. J., "Ac:ollllic: slow waves and the eonsolidaaoo trausition ", J. ACOtUt. Soc. Am.. '71, SS6-S6S, 1982. Johnson, D. L., Plona, T. J., Scala, C., Pasierb, F., Kojima, H., "Tortuosity and acoustic slow waves", Phys. Rev. Lett.• ~. 1840-1844, 1982. Johnston, D. H., Toksoz, M. N., Timur, A., "Attenuation of seismic waves in dry and saturated rocks, 11. Mechanisms", GHphys.• 44, 691·712, 1979. joJmston, D. H., Tob6z, M. N., .. Ultrasonic P and S waw attenuation in dry and saturated rocks under pressure", J. (kop/1. hs.,l5, 925-936, 1980. Jones, L. E. A., Wang, H. F., .. Uitruonicvelocitieslnc:retac:eoussllalesfromthe Williston basin'', Guphys.• 46, 288-297, 1981. Joaa, T.,Nw,A., "Velocityaad ......tioainuadst9Mat.._.tlllllperattai'CUndpressures",GHift./la. Lett.• 10, 140-143, 1983. Jones. T., "Wave ~ill porous rocks ad modclllor CQilallh'Uetare", Pial). Thesis, Staafont University, California, 1983. Kaye, G. W. C., Laby, T. H., Tlllblu ofphysiclll fJIJd cltemklll CtiiUitMrts. LoDJ1D811. New York, 1973. lC,iartaDssoa, E .... ~. C?-waw propaptioa and ~ ... J. Geoph. kn.. 14. 4737-4748, 1979. K,iaJUnuoD. E., "Attcauation of .a.ic: waves in rocks" Pbl). n.is. Stanford Univ., California, 1979b. KowaiJis, B. J., Jones, L. E. A., Wang, H. F., "Velocity-porosity clay coatent .,.aernatics of poorly~ated eandltoua:•, prcscm.l to AGll lfWIUW. San F~ Dec. 19&3. Kuster, G. T., ToksOz, N. M., "Velocity and attenuation or~ waves in two-phase media, I. Theoretical formulations", (kophys., 39, fi07-6111, 1974. · Lacaze, J., Louis, P., Rqnaud. R., "Deltaic facies rec:opdion on seismic cfa,la ", Paper presented at 48.th SEG Meetlirg. 1978. Landau, L., Likhitz. E., MicQIIique, Mir Publishina. M~w. 1967. Larson,R.G., Scriven. L. E., Davis, H. T., ••Pen:o..tionllwieryel&wophueftowinporousmedia", CltmUc.l Enfhtemllg Scietlce.l6. 57-73. 1980. Levy, Th., "Propaption of waves in a fluid-~aturated poroUs elastic:: solid", Int. J. £n«. Sci.• 17, IOOS-1014, 1979. Levy, Th., Sul;:kz-Palencia, E., "Equations and in~ ~lions .for acoustic phenomena in porous media", J. of Math~matical Analysis IJIId A/lfllicalions. ,1, ··13-834, 1977. ·
c:onsolida-'
'· ~
.'----' ~
--../ ~
--../
~~
~·
·J --~/
'_/
'._./
_.J
'
J ..__)
.__) _)
.J
----·->
----,
320
BIBJ.IOGRAPHY
Lindsay, R. B.• Mechanical radiatior.. McGraw Hill, New York. 1960. Liu. H. P .. Anderson. D. L.. Kanamori. H .• ''Velocity dispenion due to uelulicity: implications for seismology and mantle composition··. Geophys. J. Roy. Astron. Soc•._tf'l, 41-58, 1976. Lockett, F. J., "The reflexion and refraction of waves at an interface betwcetl viscoelastic media", J. Jlech.
Ph:rs. Solitb. 11, S3-64. 1962. Lockner, D. A., Walsh. J. B., Byerlee.J. D., ··Cbanaes in seismic velocity andattenuationdurinJdeformationof granite .., J. Geop. ks .. 8%, S374-5rs. 1977. Lomnitz, C .. "'Linear dissipation in solids··. J. Appl. Pllys .. 28, 201-205, 19S7. ~lcDonald. F.J., Angona. F. A .. ~fiJ:;. R. L.. Sengbush. R.I., Van Nostrand, R.G., White,J. E .• ''Attenuation of shear and comp~onal wa,·es in Pierre shale", Geophys.• 13, 421_.39, 1958. McE,illy, T. V., Mahood, G. A.• Majer, E. L.• Schechter, B. S.• TnaeaclcU. A H., "ScismoloJical Geological field study of the Sierra la Primavera ptbermal system", Uni'r. of Clllifomill, Berkeley, 1978.
McKavuqla. B., SUay, F. D., .. M..,...aicill ..llystelais in roc:b at low strain amplitudes and lllilmic frequencies ... Plrys. &rth., Plonet Int .. 8, 246-250, 1974. McSkimin, H. J., "Propqation ofloqitudinal waves and shear waYCS in cyliaclrM:al rods at llilh frequencies". J. AcOIISt. Soc. Am.,ll, 484-494, 19S6. Madel, J., Essai sur Ill mecanique dts pseudosolidu, ~des Poets at ChaUIIiel, 1950. Mandel, J,, C011rs de mkanique des miJitna CMtiluls, Gaathier-Villlm, Paris. 1966.
-~
~"'
Mandel, J., Introduction ti Ia mecan!que des miliewc continus, AcadCmie Polonaise des Sciences, serie B de M«anique Appliquee, &titioas Sc:ientiftquet de Polope; 1974. Mariao. J. J., "Microflssuration du betou et propaption des ondes", Thesis, Univ. Paris VI, Paris. 1981. MUOft. W. P:. Beshers, D. N., Kuo. J. T.,"Iotemalirk:tionittWestedyaranite: relati011todislocationtheory", J. Appl. Plrys., 41, S206-S209, 1970. Matheron\ G., tt~nts J1(JUr wre rhJorle des milieux poteux. Massou, Pam, ·1967. Mathieu, f., ToksOz. M. N., "Application of full wavefoan acoustic login& data to the .:stimation of racrvoir permeability", Paper BHG 1.5. 54th SEG Meeting. Atlanta, Dec. 1984. Ma\'ko. G. M., "Frictional attenuation: an inherent amplitude dependence, J. Geop/1. ks .• 84, 4769-4776, 1979. Mavko, G. M., Nur, A., "The effect ofnonelliptical cracks on theCOIJI.PRISibiJity ofrocks",J. Geop/1.• Ru.,IJ, 4459-4468: 1978~ Mavko, M., Nur, A .• "Wave atteuuatioD in partially saturated roc:b", Geopltys.. 44, 161-178, 1979.
q.
Miklowitz, J., The theory of elastic .. aves and M-ave pUles, North Holland, Amsterdam, 1978. Mobarek, S. A. M., ..The eff'ect of temperat\lre on wave wlocities in poreus rocks", Master's Thesis, Univ. of California, Berkeley, USA. 1971. Mocbizuki, S., "Atteauation ill penial1y saturated rocks", J. Geop/1. Ru.. 81, BSM-8604, 1982. Monicard R., Caracttristiques des roches riserroirs. Anolyse des carottes, Editions Tecbnip, Paris, 196S. 1 Morlet. J., Arens, G., Fourpau, .E., G~. D., "Wave propaption and samplioc theory,l: compleuignal and scatterin& in multilayered media", Gllop/lp., tf'l, 203-221, 1982. Morlier, P.• Sarda, J. P., "Attenuation des ondes elastiques dans les roches poreuses lliturees ", hr. Int. Fran~. "'Petr«., 26. 731·756. 1971. Morrow, N. R., "Physics and thermodynamics of capillarity.., Inti. and &.g. Cltemistry, 8, 32-S6, 19'70. Morrow, N. R., "The effect of surface roughness on contact 8ftlle with special references to petroleum recove.ry", J011r. of Canadian Petrol. Tech .• Oct.-Dec. 1975. Murphy, W. F. "Effectsofpartia! water saturation on attenuation in sandstones",}. AcOIISt. Soc. Am., 71, 14S8-l468, 1982.
m.
~
--
----,--
Murphy, W. F. Ill, "Effects of microstructure and pore Ouids Oft the acouStic properties of aranular sedimentary materials", PhD. Thesis. Staaford Univ., USA, 1982b. · · Nafe, H. E., Drake, C. L.. Physical J"'Operties of marine sediments, in:
1963.
11ie Sea, Vol. 3, 794-SIS,Iaterscieoce,
i
No\\ick, A. S., Berry, B.S., Anelastic· relaxation in crystalline solids, Academic Press, New York, 1972. NUT. A., "Viscous phases in rocks lllld the_ low velocity zone", J. Geoph. Ru., 76, 1270-1277, 1971.
.
~
......_, ·""\
<.___,..
~
Nur, A., "Seismic vdoc:ities ill low por8litJ NCb'', ilt:
321
s..·..-...um •d· .,..t/tqiMJU f/ftdktiolt, In:
Hommap au Prof. J. Coulomb, Ed. CNU. Ptril, 1 • Nur, A., "Seismic imatillliaeahllaced oil ftiOIMI)'", SH;I)t)£ M•tllw, Tulsa, hperllo.10610. ApdJ 1982. Nur, A., Byerl.ee, J. N., ··All exact effective stratlawforelasticdetOrmadoa ofrocb willa ftuicl .., J. GHpll. bs., 16, 6414-6411, 1971. • Nur, A., Murphy, W., "Wa,·e velocities and atteauationinporousmedia with fluids", Proc.ofl#lh Int. Conf. on synars, S~ )I 1-32:7, 1911. · Contlnlutr lftOtll& of Nur, A., Simmons, G .• "The effect ofviscolity oh Ouid phase on velocity in low porosity rocb", Earth a1111 PloMt Sc. Lnt.• 1, 99-108, 1969. . . Nur, A., Tosaya, C., Vo-Thanh, D., "Seismic DlODitorln& of thermal enhanced oil recovery processes'\ Paper .RS6, 54th SEG M«tin6. Atlanta, 1984. Nur, A., Walls, J. D., Winkler, K.., De V'tlbils, J., .. E«ecu of fluid saturation on waws in porous rock and relalioDt to liydraulic permeability", SPE }tltlnlltl• .tSD-4!8. 1910i. Nur, A., W'mkler, K .... Thenrlulfficltiea.*laict. . in rock&" ( .U.I), ~. 45, S91·S9~ 1980. O'Connell, lt. J., Budiansk.y, B.. "Seismic wlecitica ill~- sa&!Dted crackocholida", J. (hop/1. /l.u., 1t, S412-S426, 1974. .
dist:r'"
ill..-........._
O'Connell, It J., Budiansk.y, 8., ''Vill:od··~~~~ IOiicls",J. ~.IW., 76, 2022-2034, 1977. . . . . . .. . ·. .
O'Connell. ll. J., Budiansky, B., "Measures ot~ in ~illedia", O.Oplry. Ra. Lftt., 5, s-8, 1971.
J
O'Doherty, R.. F., Anstey, N. A., "Rettections on amp&~". GetJplt. P.rosp•• 19, 430-4S8, 1971. O'Hara,$, G., "l.nu-ce of~~ ud ~JhiiciOP tJao frequmc:y-dqptDIIIeat attcuuation of elastic waves 'n Berea sandstone", Php/l:ill rnfew ..l, 32, 413-488, 198!. · Ostrander, J., '6Piane wa-Ye rerlectiOD coelfldenta ror'ps sands at non normal anates ortncidence'\ , . , No. SJ6.4 p~ted at 52nd SEG Meetilfl, DaJiu, 1982. Paillet, F. L., White, J. E., ··Acoustic model of propaption in tile borehole and their relationship to rock properties",~~-· 47, 1215-1228, 1982. ' . Palmer, I. D., Traviolia, M. L., "AtteauatiOD by ~quirt flow in under-saturated ps sancla ", Geophys., 45, 17801192, 1981. ! . Pallet, M., et IIJI., La mkaniqw des rochu qplilplh IIIIX 011vr1Jfft de Gblie Civil, ENPC, Paris, 1976.
w.
Papadakis, E.i P., "Ultrasonic difl'rac:tion frOm apert~ willa applications to pulse measurement& and crystal physics", ill: Physical ..lCOtUtic, XI, Ed. Muon Tlwnton, Ac.-Preu, Chap. 3, lSl-211, 1976. Parts, G. A., "Surface and intcdacial free . . . . . of' ........ J. G.o,/t .. ICes., 8, 3997-4008, 1984•.
liilalc
·~
''-./'
f....,ill,..
.PadDick. L., ,Outcrkidp. yt. F.. "Ill . ...a lUI4 sbcu . . .u.o(Soleabofcn JiiDestoae over a frequmcy ranp of 10 cyclci ... ....t",/. ~,.,... "- 581-SII,.t96t.
WJ.J .. ~ B. f .• ~ W.J.. "~ofair-~udoil-aircapillarypreuurcdatain the study of pore struc:ture and ftnid cfistrmutioJa , $1'. JOfiT1IIll, $5-61, 1966. . · ·
etutic
~
h.
Picltett, G., "Equations for computinJ coMtants flquraJ ,.cl19rsioaal RIOUilt &cquciacies of vibration of prisms and cvlindcrs", Am. Soc. Ten. A#(IIB. Proc., 45. 846-860, 1945. Pjumann, E., Dusclultko:R.. "Usc ofpQrc casts S'!M io pore pometry", Sell. Petrology. «l.
115!-1151, 1910.
study
ad
Jm'u.
. ' . .. ' Plona, T., "()bscrvation of a JeCOncl bulk~-~ an a poMUS medium at ultrasoaic frequencies",
..lppl. Phys. Lett., ._ 259-26t, t980. · Plona, T. J., "Acoustics of Ruid-saturated pi)I'Obt inedia.... Vlfraonlci-Sprrtp. IEEE, 2, 1044-1048; 1912. Ptona, T. J ., Johnsoil, D. L., .. Eipeiimelital Mudy' Of the two bulle COibprcSSional modes in water-satunted porous ltnlc:tures .., Vltrt1Sottk2 S1f'IP. 16EE, ...,!12. ..-,_ · Pnalicr, A., "RecMrcbe en tnlitaacat -.....•r1adidtdea_..SlNIIIDll troiseodlpolaDtcs, Application sur un exemple reel", Thesis, Univ. de Purcell, "Capill8ry ,_..,.... 'l1lllir ••• IMI'CUIY ....the caii:Watioa of penlleltbility thercf'nrm", J. of Petroleum Tedurology, 39-47, 1949. Qnimby,S.L.,"OatJaeexperimeat.tdetenlliDatiollefdll:viNosityofvibratinatolidi".P•,,.••.• zs.ssa..s73. 192S.
SUa.,.., ••· t
-~·
....
322
IIIIUOORAPHY
Rasolofosaon.. P.. Coussy. 0., "Piopaption des ondes ~urs dans les milieux poreux satures - Effets d'intenace. ~ panie··. Rttr. lnst. Franr. du Ntr*, 40, 78S-t02. l98S. Rassing. J .• "l:ltraSonic maxation specbomefry'' ;,: Clwrniad tlllll biologietll IUJI«U II/ rBaxlllion Sf'«trotnetr_t·. Ed. by Wyn:Jons. Reidal. nouan.l !~7S. Lord Rayleigh. ··On waves propapted along the plane surface of an elastic solid''. Prot:. Loru~Df~. Matlt. Soc .. 17. 4-11. 1SS5. Ra~mer. L. L.. Hunt. E. R .. Gardner. J. S., "An improved sonic! transit time-to-porositytransform",SPWLA. :Jst .4nn. L.ogg. Symp .. July 1980. Rice. J. R .. Cleary. M. P .. ··Some basic stress diffusion solutions for fluid-saturated elastic porous media with compressible constituents··. Re'·· Geoplr. and S~ce Phys.• 14. 227-:41, 1976. Rogez. D .• Bader. M .. "Tltrasonic ,·elocity dispersion in liquids betwea .3.3 and 330 MHz llling a high ~lution phase measurement techPique", J. AcOJMI. Soc. Am•• 16. }67-172. 1984. Rosenbaum. J. H .• "Synthetic mic:rosei~: loUidl ia.pon*S foRDaticla", Geo,ltys•• 14-32, 1974. Sa~. J.. YUcoilturtctte. PNaa de t•ac. NatioUIIt dll Poatut Cbauslies. Paris, 1983. Sayers. C. M .• "lltrasoninelocity dispersion in porous materials". J. Phys. D. : Appl. Phys.• 14, 413-420, 1981. Schlum~ Co.• Log intn'pretatitllt ~. Ed. by !dtiU11lb!!tJer Limited 1977. Schmidt. E. J.• \'o-Tbanb. D .• Nur, A.. Auld, .B., "AA:oustic and electromapetic energy absorption in clay bearing rocks at low fluid saturation ... to be published', (kop/iys.. 1986.
~·~
».
Schoenberg. M.. Manetta. T .. Aron, J., Porter, R., "Spa<:e-t.ime .lependance of acoustic waves in a borehole ... J ..·koust. Soc. Am., 70. 1496-1$07, 1981. · Schoenber,er. M .• Le'in. F. K., "Apparent attenuation due. to intrabed multiples", Geophys.• 39,278-291, 1974. . . .· . Schone. P. A.. ·• Chalk diagenesis &Qd' itS rdaliori tO 'pettOiell!Ji exploration". _.APG. S, 982-1009, 1977. ~hreiber, E~. Anderson. 0. L., Sop, N., Eitutic ~tants and their tneiUIITemetrt, Mc(Jraw Hill, New York, 1973. . . . . . . . . . .
..
Serra. J.• /mage analysis and mathematiCDI !1f0rphology. ~ fress, New York. 1982. Sheriff. R. E .. Taner, M. T .. ··Seismic attrlbute me;.$~reshelp defnacreservoifs". Oi/andGtuJourrull. 112-186. 1979. . Spencer. J. W .. "Stress relaxation at low frequencies i~ Ouid saturated rocks: attenuation and modulus dispersion". J. Geoph. Res .. 1303-l812. 1981. Spencer. T ..-\ .. Sonnad. J. R., Butler, T. M., "Seismic Q. Stratiy.lphy or dissipation", Gef!phys., 47, 16-24, > • .• •·• . • . . ' 1982. . ' ..
~tzler._.H.. :'M~· D. L.. "nC elf~ o( ~ture ~partial mellinJ Oil velocity and at$eouation in a · sunple bm&r) system . J. Geoph. J&s .. 73, 6051~. tMII. Spitmtt. s.. Teft, W. E., .. MethdclfOr~lllfidlmi<:lfn:sorwtc:drequeoc:ieallblltorcalc:11latingelastic moduli from these frequencies", AM.. Soc. Tm. Mam. Proe.• 61, l:!!l•tm. 1961. ·
rise
s«.. "·
Stewart. tt. G .• ··Q uct the uct a~ ptil~e", aeoPb,, J. JtD.r. Astr. 193-805, 1984. Stewan. R. R., TolcsOz, M. N., Timur, A., "Strain dependent attenuation: observations and a proposed medumism... J. Geoph. Rn.. •. S46-SS4, 1913. · StoU. R. D .• "Experimeptal s t . of~ttenuation in sediment$", J. Acoust. Soc. Am.• 66, 1152-1160, 1979. Strick. E.. "Tiie determination of' Q, viseosity arid creep ·cun·es from waw propaption measurements", Geoph. J. Roy. Astr. Soc., 13, 197-218! 1967. . · Strick. E.• " A predicted pedestal effect for pulse propagation in eonstant Q IOlids ". Geoplv•s.• 35, 387-403, 1970. Suquet. P.•. "Piasticite ct homopgeisation". ~ U~versi¢ P. et M. Curie. Paris 1982. Swanson, B. F .... Visualizing pores 8JlCi nonwettina phase .in por:ous rock", J. Pet. Tech.. Jan, 10-18,. 1979. Tauer. M. T .. Koehler. F .. Sheriff, R. E., "Complex trac1ranalysis". GftJPhJ'I.• 44, 1041-1063, 1979. Tarif, P., Bourbii. T., •• Experu.atal ~ botwten apeetral ratio and rise time teclmifrues for the measurement of attenuation··. to be published, G• . Prosp., 1986. Tarif, P.. -M~de I'attenuation par.prop~~ptioa d'ultruoa: application aI'etude de l'aailotropie ",Thesis, Unhrersite Paris \11, 1986. Tsmur. A.• "Velocity ofcompreuional waves ill pomusmedia at permalrosttempcraturee", Geop6ys .• 33, S84595. 1968.
ran or
,
..
dYnamic
~~
~.
~~~~~-------------------------~--~-------~---
.:
''-
'---'
f
I
:i
''l'".t.J·
:i·\
·t ":-
J.di-.i~
3~3
BIBLIOGRAPHY
Timur.A.. MTcmpaaturcdcpcndcnccof~onalandslaqrwaveveloc:itielinfocb",~ys.• 4l.9S09S6, 1977. Tittmann. B. R .. "Internal friction measurements and their implications in seismic Q-struc:turc models of the crust''. AGU ~oplt.t·s. MOllO,., Ja, The earth's crust, 197-213. 1977. Tittmann. B. R .• Oark, V. A., Richardson. J. M .• speitcer. T. \\' .. "Possible mechanism for seismic attenuation in rocks containing a small amount of ,·olatiles". J. Geoplr. Rts.. IS. 5199-5208, 1980. Tittmann. B. R .• Noadler, H .• aark, V. .-\ .. Ahlberg. L.A.. Spmcer, T. W .. "Frequencydependenccofseismic attenuation in saturated rocks", Geoph. Res. Lm .. 8, 36-38. 1981. Tok50z. M. N .. Johnston. D. H., Timur. A .. "Attenuation of seismic waves in dry and saturated rocks: I. Laboratory measurements", Geophys.. 44, 681-690. 1979. Tosaya. C., "Acoustical propenies of clay-bearing rocks". PhD. Thesis. Stanford Univ., Calif., USA. 1982. Tosaya.C.. Nur. A., ••EffectsofdiaJtneSisandclaysoncompressional velocities in rocks", Geoph. hs. Lett•• 9, 1982. Tosaya. C., Nur•.A .• Vo-Thanh, D .• Da Prat. G .. "Laboratory seismic methods for remote monitorins of thermal EOR". -~ Cnll. Europ. ·"" l'am;/fnrarinn de Ia rkrtf'<'rarinn tht phrnle. Rome. Arril 19RS. Truell. R.. Oates. W .... Effect of lack of parallelism in sample fa.:es on measurement of ultrasonic attenuation .., J. ACOIISI. Soc. •.fm., 35, 1382·1385, 1963. Truett. R.. Elbaum. C., Chick, B. B., Ultrasonic merlrods in solid sttJie physics. Academic Press. New York, 1969. Tullos, F. N., Reid. A. C., "Seismic attenuation of Gulf coast sediments", ~ophys., 34, 516-S28, 1969. Viktorov. I. A.. Rayleigh tmd IAmb "·a1·es, Plenum Press, New York. 1967. Walsh, J. B~ "The effect of cracks on the compressibility of rocks". J. ~oph. Res.. 10, 381-385, 196S. Walsh, J. B., "Seismic wave attenuation in rock due to friction", J. Geoplr. Res.• 71. 2591-2599, 1966. Walsh, J. B., "New analysis of attenuation of panially melted rock", J. ~- Res., 14, 4333-4337, 1969. Ward. R. W., t ok50z. M. N., "Causes of rqional variation of magnitude". Bull. Seism. Soc. Am., 61, 649-670, 1971. Wardlaw. N.C.. Taylor, R. P., "Mercury capillary pressure curves and the interpretation of pore structure", Bull. Ctm. Petrol. Geol., 14. 225-262. 1976. Waterman, P. C.. Truell, R., "Multiple scattering of waves". J. of Math. Physics. 1, 4, 512-538, 1961. Waters, K. H., hjlection seismology: a toolfor energy resource t.Tploration. Wiley lntersciencc Pub., New York, 1978. White, J. E., &ismic wans: rtJdiotion. trtDUmission and atlenUIJiion, McGraw Hill. New York, 1965. White, J. E., "Q of fluid saturated rods", PtJper S 4UJ6. AGll Meeting. May 1983. White, J. E., c:rulergrotUUI SOillld: wlictJiion of seismic t\'Ql'es. Elsevier. Amsterdam, 1983 b. Widen, M. B.. "How thin is a thin bed?", Geophys .. 31, 1176-1180, 1973. Winbow, G. A .. Rice, J. A., "Theoretical performance ofmultipole sonic loainl tools", PtJper BHG 2.7, 54th SEG Jleetittg, Atlanta. Dec:. 1984. Wmlder, K., "The effects of pore fluids aacl frictional slidina on seismic attenuation", PhD. Thesis, Stanford University, Calif., USA, 1979. Wmlder, K., "Frequency dependent ultrasonic properties of high porosity sandstone ... J. ~ph. Res.• a, 94939499, 1983. Willkler, K., Nur, A., "Pore fluids and seismic attenuation in rocks", Geoph. Res. Lett .• 6, 1-4, 1979. Willkler. K., Nur, A., "Seismic attenuation: effects of pore fluids and frictional slidina", Geophys.• 41, l-IS, 1982. Wia1tler. K., Nur, A., Gladwin, M., "Friction and seismic attenuation in rocks", .\'ature,l77. 528-531, 1979. Winkler, K., Plona. T., "Technique for measuring ultrasonic velocity and attenuation spectra in rocks under pressure", J. Geoph. hs .• 17, 10776-10780, 198:!. Wu, T. T., "The effect of inclusion shape on the elastic moduli of a two phase material", Intern. J. Solids Structures, l, 1-10, 1966. Wyllie, M. R. J .• Gregory, A. R., Gardner, L. W ., .. Elastic wa\·e velocities in heterogeneous and porous media", Geoplrys .• l, 41-70. 1956. Wyllie, M. R. J.. Gregory, A. R., Gardner, G. H. F .... An experimental investiption of factors affectins elastk wave velocities in porous media", Geophys.. 13. -'59-493, 1958.
s-s.
·~
..___
'·~·
---~
•
324
BIBLIOGRAPHY
Wyllie, M. R. J., Gardner. G. H. F., Gregory, A. R., "Studies of elastic wave attenuation in porous media··, ~•• 17, ~S89, 1962, lllrd d&t:ussi011: G#opltya., 21, 1074, 1963. Yin& C. F., Truell, R., "Scattering of a plane longitudinal wave in an isotropically elastic solid .., J. of Appl. Pltp. 7:1, 1086-1097, 1956. Zinsmer, B., Meynot, Ch .• ··Visualisation des proprietes capillaires
~-----
•
~
~
"---"
author index '-..__
"'--
'-
Brutsaert, w.. 190 Budiansky, B., 58, 65, 122 Butler, T. M., 304 Bycrlee, J. D., 176, 184
Abramovitz, M., 125, 127 Achenbach, J. D., 49, 70, Ill. 273 Ahlberg, L. A., 213 ADdenon, D. L., 123, 124, 195 Andenoa, 0. L., 162, 163, 166 Andrieux, 63 Aqenheistcr, G., 57 Anaona, F. A., 241 Anstey, N. A., 303 Archie, G. G., 71 Arditty, P. C., 160 Arens, G., 160, 304 Aron, J., 281 Asce, A.M., 190 Asce, F., 190 Auld, B. A., 223, 225, 264 Azillli, A., 124
s..
"'--·
.._
.
Capiarcl, L., 266 Caqil1, G. S., 13, 16 Camwm, P. C., 35 Carmichael, R. S., 80, 193, 241 Cutqna, J. P., 298, 311, 313 Caye, ll., 13 Cerveny, v .• 249 Claandler, N., 81 Chang, s. IC.., 311 Chell, S. T., 281, 311 Chalg, C. H., 311 Cbkk, B. B., 147, 151 Qoquette, P. W., 14, 18 Ouistensen. R. M., 100 Cart. V. A., 190, 207, 213, 21P Cleary, M. P., 81 aiel, C., 307 Coben-Tannoudji, C., 109 Cooper, M. F. Jr., 290 Cordier, J.P., 230, 237 CouraDt, ll., 129 Coussy, 0., 266
s.
Backus, M. M., 304 ·~
Bader, M., 151
BaDcrof't, D., 166. "'--
'---'
~
'---' '-
~
'-.-c
Beard, D. C., 233 Becquey, M., 304 Belllouuan, A., 64 Berry, B. S., 122, 143 Berryman, J. G., 71 Babers, D. N., 219 Biot, M. A., 64, 65, 67, 72. 7J, 22S, lin:ll, F., 166 Blair, D. ll., 157 Blan4, D. R., 122, 125 .Bioadin, L., 309
Borcherdt, R. D., 112, 286. 287, 288, 290. 291 Bourbie, T., 80, 157, 158, ISS, Ill, 190, 204, 206, 208, 232, 243, 252, 259. 287.290, 291, 295, 296, 297' 306, 307 Bousquie, P., 29 Bowler, J .• 146 Brace, W. F., 34 Bracewell, R.N., 270, 304
--
248. 252 Danbura, J. s.. 233 Da Prat, G., 192, 196, 197. 207, 211 Datta, s. IC.., 94 Davis, H. T., 17 Davis, L.A., 170, 203, 2ll. 212, 216 DelfiDCI', P., 13, 20 De Martini, D. C., 233 Denis, A., 179 Demiewic:z. H., 246, 247. 248, 250, 258,260.261, 262,264 Desclwnps, M., 152
'._/
_/
·._/
~
326
AUTHOR INDEX
Devaney, A. J .. 215
Grau, G., 302
DeVilbiss, J. W., 187, 195, 197, 200, 211, 242, 309 Dhawan, G. K., 29 Ditkine, V., 270 Diu. B., 109
Gregory, A. R., 75,177,181,189,130,231.133,235. 237, 254, 265. 273 Grittl, R. E., 12 Guillot, D., 14
Domenico. S. N., 190. 191, 237, 238, 304, 305 Drake, C. L.. :!32, 233. 235 Droschak, D. M .• 166 Dubesset, M .• 307 Dullien, F. A. L., II. 23, 29, 30 Dupa, L. L., 146 Duschatko, R., 13
Elbaum, C., 147. 151 Ensley, R. A., 306 Eringen A. C .. Ill Eshelby, J. D .. 58 Etienne, J ., 29 Everhart, A. H .. 3lt Ewing, W. M., 49, 249
Fatt, 1., 259 Faust, L. Y., 237. 238 Feng, D. L., 248 Flugge, W., 100 Fourgeau, E., 304
Fung, Y. C., 53
Hadamard, J., 96 Hardin, B. 0., 190 Hauge, P. S., 241
Hickman, W. B.• 31 Hilbert, D., 129 Hosten, B., 152 Huang, C. F., 311, 313 Hunt, E. R., 232, 234 Hunter, J. A., 311, 313
~~
~
~~
lshimaru, A., liS
Jacquin, Ch., 35, 44 Jankowsky, W., 237, 238 Jardetzky, W. S., 49, 249 Johnson D. L., 71, 80, 81, 88, 89. 91, 92, 93, 248 Johnston, D. H., 154, ISS, 203, 205, 218, 243 Jones, L. E. A., 180, 233, 236
Jones, T.,l82, 186,194,209,222,223,226,227,306,
~
310
Futterman, W. L., 124
Gai'Perin, E. 1.. 310 Ganley, D. C., 241 Gant, W. T., 311 Gardner, G. H. F., 75, 166, 182,230, 231,233,235, 237, 254, 265, 273 Gardne~J. 8.,232,234 Gardner, L. W .. 230, 231, 233, 237 Gassmann, F., 68, 237 Geerstma, J., 84. 94, 190, 248,250, 2S4 Germain, P .. 49. 66, 69 Giard, D., 304 Gladwin, M. T .. ISS, 201, 216, 218 Goetz, J. F., 146
Kalinin, A. V., 124 KaliniD, V. V., 124
Kan, T. K., 298, 313 Kanamori, H., 123, 124 Kanasewich, E. R., 241 Kaye, G. W. C., 193 Kjartansson, E., 125, 126, 127, 157, 221, 224, 218, 305,307 Koehlet, F., 304 Kojima, H., 93 Kowallil, B.J., 233, 2.36 Kuo, J. T., 219 Kuster, G. T., 58
Goldbug. D., 241, 298, 311, 313
Laby, T. H., 193
Gonzalez-Serrano, A., 259, 287, 29S Gordon, R. B.• 170, 203, 211, 212, 216
Lalc)e, F., 109
/-
..-......
~-----
Lacue, J., 304 ---.._
-----~
_...,
A~'iWolix
Nowick. A. S., 122. 143 Nur, A .. 62, 94. 139. 176, 177. 179, 181, 184, 187, 188,192,193.195,196,197.201,205,206.210, 211,212,216.218,222,223, 226,227,233, 236. 242, 252. 297, 306. 309
Landau. L., 49 Lanon. R. G., IS Lawrpc, M., 239, 240, 304 Le Fournier, J., 29 Levin. F. K .• 147, 303 Levine. H .• 215 Levy. Th., 64 Lifehitz, E., 49 Lindsay, R. B., 287 Lions. J. L., 64 Liu, H. P., 123, 124 Lockett, F. J., 290 Lockner, D. A., ISS 124 Lomnitz, Louis. P., 304
m.
Oates. w .. 147 O'Connell, R. J .. 58. 6S, 122 O'Doherty, R. F., 303 O'Hara, S. G., 165 Ostrander, W. J., 304 Outerbridge, W. F., 163, 164
c ..
Paillet. F. L., 274
Mahood .• G. A., 309 Majer. E. L., 309 Mandel, J., 33, 49, 53, 64. 66, 72, 120 Mari. J. L., 309 Marigo, J. J., 58 Marzetta, T., 281 Mason, W. P., 219 Matheron, G., 18, 72 Mathieu, F., 3tl, 313 Mavko, G. M., 58, 94, 198; liS, 2'18, '~
Me Skimin, H. J., 153 Meynot, a., 29 Miklowitz, J., 266 Mills. R. L., 241 Mobarek, S. A. M., 194 Mochizuki, s., 225 Molotkov, I. A., 249 Monic:ant, R., t I Mor1et, J., 304 Mortier, P., 94
,_
~~'
"---
'~
22'7
Me Donald, F. J., 241 Me Evilly, T. V., 309 Me Kavanash, B., 170
'~
l
Morrow, N. IL, 23 Murphy, W. F. III, 123, 177, 179, 181. 184, 187, 188, 190,191,199,201,206,213,214.216,218,220, 222,227,229
Nafe. H. E., 2~2, 233, l!S Noadler, H., 213
'327
Palmer. I. D., 227, 229 Panet, M., 57, 177 Papadakis, E. P., 153. ISS Papanicolaou, S., 64 Parks, G. A., 192 Puierb, F., 93 ~ L., 163, 164 Pickell, J. J., 29 Pickett. G., 162 Pierrot. R., 13 Pittmann, E., 13 Pivovarov, B. L., 124 Plona, T., 80, 81, 85. 86, 88, 89. 98. 91, 92, 93, 152, 153, 213 Porter, R., 281 Pray, L. C., 14, 18 Preis, F., 49, 249 Prevosteau, J. M., 13 Proudnikov, A., 270 Prunier, A., 306 l'leacik. I., 249 Purtell, w. R., 26
Quimby, s. L., 165 Rqot, J. P., 13 RuolofolllOD, P., 266, 298 Raising, J., 223 Lord Raylcip, 263 bymer, L. L., 232. 234 Reid, A. C., 241
-~j
328
Al:THOR INDEX
Reynaud, R .• 304 Rice, J. A., 311 Rice, J. R., 81
Rice, J. T., 2SO, 258, 2~. 26l, 262 Richardson, J. M., 207, 219 Richart, F. E., 190 Robinson, J. H., 233 Rogez, D., 151 Rosenbaum, J. H., 94,248,273.274,276. 277.278. 282.311 Salen~n. J., 100 Sanchez-Palencia, E., 64 Sarda, J. P., 94 Sayers, C. M., 215 Scala, c., 93 Schechter, B.S., 309 Scblumberger, Co., 230 Schmidt, E. J., 223, 225 Schoenberg, M., 281 Schoenberger, M., 147, 303 Schone, P. A., 34.
Schreiber, E., 162, 163, 166 Scriven, L. E., 17 Sen, P. N., 71, 93
Sensbusb. R. 1., 241 Serra, J., 18 Sherift, R. E.. 304 SieJ(Jed, R. W., 311 Simmons, G., 195, 210 Skalak, R., 93, 94, 246, 247, 248 Smit, D. C., 84, 94, 248, 250, 254 Soga, N., 162, 163, 166 Sonnad, J. R., 304 Spathis, A. P., 157 Spencer, J. W., 123, 171, 222, 227 Spencer, T. A., 304 Spencer, T. W., 190, 207, 213, 219 Spetzler, H., 195 Spinner, S., 166 Stacey, F. D., 155, 170
Staron, Ph., I~ Stegun, I. A., 125, 127 Stewart, R. G., 157 Stewart, R. R., 218 Stoll, R. D., 218 Strick, E., 124
Suhubi, S., 111 Suquet, P., 63, 64 Swanson, B. F., 31 Tanner. M. T., 304 Tarif. P., 154, 155, 157, 158. 160 Taylor. R. P., 26 Teft. W. E., 166 Timur. A .• 154, ISS, 194. 195, 197. 218. 243 Tittmann, B. R.• 166, 190. 207, 213, 219 Tok50z, M. N., 58,.154, 155,203,205,218,243,311,
-~
313
Tourenq, C., 177 Tosaya, C., 178, 180, 183, 186, 192, 195, 196, 197, 210, 211, 233, 236
Tra,ioUa, M. L., 227, 228 Truell, R., 115, 147, 151 Truesdel, A. H., 309 Tullos. F. N .• 241
...--...~
Van Nostrand, R. G., 241 Viktorov, I. A., 263 Vo-Thanh, D., 192, 195, 196. 197,210. ~II, 223,225 Walls. J. D., 187, 242 Walsh, J. B., 58, 94, 116, 180, 184, 195. 218, 2l6 Wang, lt.·F., t80, 233,236 Ward, R. W., 154 Wardlaw, N. C., 26 Waterman, P. C., 115 Waters, K. H., 302 White, J. E., 68, 141, 153, 161, 164, l~. 241,1171 Widess, M. 8., 302 Willis, D. G., 65, 67 Willis, M. E., 311 Willm, C., 304 Winbow, G. A., 311 Winkler, K., 58, 139, 152, 153, 165, 166. 169, IB7, 197,200,201,205,206.212,214,215,216,218, 222, 242 Wu, T. T., 58, 180
~-""""'
~~
Wyllie,M. R.J., 75,166,182.230,231,254,265,273
Ying, C. F., liS Zinsmer, B., 29, 80, IS9, 185, 188, 190. 204, 206, 208. 232, 243
~
~
...--...
-----:'1
....... ''--' '
.
'---' '----'
subject irdex '·.,____
'---'
'-' '''-"
------
Act.orbed water, 219, 225 Ac:ces radius (porous medita), ~6. 29, 87 Atlnl (material), I 00 Air)' ..... Ill A8qiiAollte, 216 Allltydrite, 235, 240 Allilotropy iaduced, 183, 184 intrinsic, 186 ~pasation,61, 74,96 transverse, 97 Aaale (eritical), 88 Aaortllollte, 220 Ane.atloa capillary forces, 220, 221 contrast, 306 deftnition, 15, 107, 1 t: extrinsic, 147, 303 frequency, 211-214, 2::2 in situ measurements. 240, 243 interface, 279-281 intrinsic, 147, 305 mechanisms, '63, 217, 229 measurements, 15 I ~.156
permeability, 275 "---" "--'
'---'
'-----' "--· '-'
'-"-'
attenuation contrast, 306 velocity contrast, 304
Caldera, 309 Capmary, Calilarity, 22 desorbtion, 24 equilibrium, 23 force, 191, 220, 221
pressure, 24 visualization, 29, 32 ear.a.Kozeay fonmda, 35,47 Casco (see Gruite) C...uty prillcifle, 103 Claalt (see u.e.toae)
a ...ca~._.
hydrosen, 219 Pray dellllfacatloll 15, 16 Oay (see also, Sllale), ll, 12, 34, 235, 236--240
a...-ue· ...
phase change, 211 viscosity, 207, 210 saturation, 2()4..207 methods, 207, 208 scatterin& 214, 215 strain, 215 temperature, 207, 209
Colli, 240 COCGidao (see Saadst011e) Celoralle(seeSIIIIes) eo.,.ctloa, 237 CotDctlfllle,23 COIIdaulty bldex, 179 ConloYa ereaaa (see I' eltDae) C....... l47, 153 inertial, 71, 83, 225
Barre (see Graaite) BMalt,216,240 Wfonl (see IU.IIt1M1e) Berea (see s..tst..) IJerrymaa's foranlla, 79. 265 Bimodal perGiimelric SJiftb ... 27,
8riPt ...... 304-308
breakage, 219, 221
pressure, 203, 204 saturatina fluid
-~
--------
Blot's dleory, 63-84. 141, 161, 166, 190. 225, 250, 259, 263 (experimental veriftcation of), 85 Bolle (see Saadltoae) ~.. prladple, 102 .............. 152
tbermomechaDica 65
viscous, 87
rT
COYMiuce, 18, 40 Creep fuctloa, 100, 124, 12S
'-'
_.. ..::£:;-
330
SUBJECT INDEX
Cracks, 179. 180, 186, 204 growth, 57 porous medium, 21 closure, 57 Cross-sectioa (scatterilll), 115
Damping (re4uced), 136
Dare:,·
Ia~·. 31, 69. 281 unit, 31 Decay (lopritluaic), 135, 161 Delayed behamr (coastlbltlve law), 54, 56, 100 Desorbtioa (capillary), 24 Diatomite, 231 Diffusion, 81 Diffushity (hydraulic), 81, 227 Dilatation (vollllllie), 52, 55 Diopsiclite, 216 Dileoatiauides aad wafti;24S..l49 Dislocation, 219, 221 Dispersion, 128 inverse, Ill normal, Ill Rayleigh waves, 265 Stonely waves, 267 Dissipation, 269 Diveraeace(spreadllll), 147, 153, t'5s, t60, 303 Dolomite, 16, 17, 32,230,234,235,240 Webatuck, 177, 179 Ih1Uaa,e,23, 29,32 Dry rock (defulitioa), 187 Duaite, 179, 216
Echo method, 151, 172, 173 ~. 132,273 llgeafrequency, 131, 136 Elastic constaats, 62 Elasticity (linear), 54 Eblltodynamfcs, 49 Eaeray average, 122, 138 balance, 287 elastic, 106, 121, 138 kinetic, 70 maximum, 122, 139
EVA tool, 159, 240, 310-312
Exteasional IROde, 162 Fadel variation, 304 Faust's fennula, 237 Feldspar, 17 Filtratioa velocity, 246 Flexural..-, 162 Flow (1Rultiphase), 38 Foataiaellleu (see s..lstoae) For~Ratiolt faetor, 71 Fort Ualoa (see S.Ddstoae)
~
-,,
Fourier.......,_, 104 Foxbills (see s..lstoae) Fraehuila ' (hydraulic), 310 (natural), 310 Free 'ribratloas, 134, 161
,,
~
F~
band, 146, 301 central, 228 ch~stic,
75,80 (eigen), 130, 131, 136 resonance, 162 sweep, 136, 162 thermal relaxation, 221 Frldloa. 216
c .........
fOI'IIIIIIa, 68, 82, 84, 190, 265 Gcentma .... Smlt ........ 84, 190 Geopltysia (petrale-), 301 Gneiss, 186, 240 Granite, 212, 216, 239, 240 Barre, 184,195,210 Casco, 179 Oklahoma, 222 Sierra White, 182, 188, 199, 200, 206. 242 Stone, 179 Troy, 177, 179 Westerly, 177, 179, 184, 193, 197, 211 Grain contact,9 size, 19,44
~-
-~~
~
---, ' ~
Halite, 235, 240 Hamilton's prilldple, 70, 111, 246
~
' ~
~
'"'""
-~~ ~. 16, 17,2?;28,32, 37
Haakel . . . . , _ , 270
attenuation, 243 Bedford attenuation, 203, 210 ve~ty, 171, 179, 195. 196 chalk, 27, 28, 39 attenuation, 207, 241 velocity, 178, 179,240 Indiana, 207 Oak Hall, 179 porosity/permeability relation., 34 relative permeability, 39 Solenhofen, 177, 179 Speraen, 222 ~ty, 179, 230, 234, 235, 231-ol40 ~ 146,247,255,2&1
Hannoaic:, 137, 167 lleteropaelty, 167 -~
''-
~
'--
'---
~ (minimum volume), 40, 4i, 42 Biot's theory, 63 Hooke's law, S4 HydrauUc fraetwiDI, 310 Hydrogea bollll, 190, 209 Hydroxyla, 190, 219, 225 ,:
- · ...., . , 18, 19 llllbiWdoa, 23, 29, 32
lncelapr! Dllitr ... '--'
I
I
S6
M....,. (see Sa~Mtstwe)
Impulse ........ 125 ladiua (SH I I !ROlle) bljecdoD (epoxy), 21, 29 laerdal coupling. 71, 83, 225 effects, 81 lntercryltallia (ponul ......), 18
Madlemadc:al ~. 18 MaxweiiiBCMiel, 119, 120 Mlcroeraeb, 9, 21, 179, 204, 301
MoWIIty, 72 MCNiel
Interface eff~(attenuation),29S
open or sealed, 247, 268-270
laterfereace, 108, 109, 133
IJiter&ruulM' (ponul . . . . .),
13, 16, f8, 35
~(paroal---), 16.'t« lrrotadoltal ...,_ellt's. 59
'-
331
Jn's.......,26
~
Kelrut-Veip ...... 119,"120 Kera RiYer (see SaillllsteM) Kr~~~~en-Kroal& .........., 122. 142
'
Constant Q, 124-128, ISO, 200, 287 Kelvin-Voigt, 119, 120 Maxwell, 119, 120 Nearly Constant Q, 123 standard, 117, 123 20,288 Zener, 117, 123
~
complex, 104, 112, 142. 285 delayed, 118, 143 effective, $1 instantaneous, 118, 143 reflection and transmission c:oefticieat, 263 relaxed, lOS static, 58 MoNk (porous me6Jiia), 16, 18 M .. cake, 247, 248 Myloaite, 186
(
---------~
-~
l..qrutiu, 70, Ill
Lame's ea•t•..., 54, 62, 61, 104 Laplace's ,......., 22
Lateral extealioll eofta:tiDM. 166
Na•ajo (SH Saadstone) N.--Stokes equadoa, 72 Neller (deflaldoa), 112 Nolle, 147, ISS, 161 NGIIIinearity, 168
-,_
_-,J
332
SUIUECt INDEX
Oak Hall (.rn I..ilaelltoae) Offset VSP, 308 Oalager's priadple, 69 Orthotropy' 97 Ottawa (see Sand)
Pressure confming, 176, 181 differential, 176 effective, 176, 181 hydrostatic, 56, 67 pore, 176, 181 Pyroxeaite, 216
"'·
'
\
Packing (spheres), 13, 14
.....Utude.
129, 130 Pendulum, 163-169, 172, 173 Peak
Q (see Quality faeter) Quadratic form, 66 Quartzite, 179, 212, 216 QaaUtyfaetor(defmitioa), 112, 117, 125, 140, 170, 288 attenuation, 113 kinetic energy, 113 2 0 equation. 286 in situ measurement. 241 Q•• 139 QK,, 139 Q_,, 298 Q,., 139 Q•• 265 Qs. 139,289 QSr, 267 Que• 298
Percolation theory, 15, 36 PenaeaWIIty absolute, 31, 7l bulk, 31 global, 31 hydraulic, 69, 72 interface wave, 247, 275 . ! matrix, 31 measurement. 33 monophasic, 31 porosity, 34-36 relative, 38, 39 Permeameter, 33 Phase chana~ (flllicl), 207, .211 Phase shift, 134, 261 stress-strain, 113, 170,.173
Pierre (see Shale) Polarity cbaage, 307
Racliolarite, 231 RQP,.20, 40
Polseuille's law, 33
Raylftala
Poissoa's ratio,. 54, SS, 62, 139, 181, 231, 239, 304. 306
..
Radius ac:ces·(poroas Blediam), 26, 29, 87 Reflectioa coefficient, 152, 250, 255-263, 29().:296 modulus, 263, 295 phase, 263, 295 interface, 249-258 multiple, 147, 303 Relaxation, 101, 11&, 142 time, 118 thermal, 221, 224 viscous, 226 Resouace peak, 162, 166
Poteatial strain, 56, 66 dissipation, 69 surface dissipation, 246
;
-t•
Resoaant bar, 137, 163. 169, Restam'ed .._. mediad. 24
I
,
approximation, I t6 (pseudo), 311 wave, 263
Pore cast, 12, 17, 21, 37, 45, 187
Porosimetry (mercury), 26 '· Porosity, 10 clay, II connected, II fiSSure, 21 measurement, 10 occluded, 11, 36 total, II, 46 trapped, II, 25
~,
~
P-.elllllddty, lS4 Po~BMtrics~,27,28,44
~.
··'
---.....-..,
17~ 173
Rheological lllOdel, 117
Rise time, 1~6, 148, 155-160
"' ~
~.
\.____
'-
'--
I
,
I
"-
!
'--
I
'-
I
'--
I
'-
I
~
'--~
I
'-
I
'~
I
\.____
I
'--
'-'-....-
'--'
I
I I
~
'- ,,__,
I.
1..~
I
'------
I
'~
I
3l3
Saiat Peter (.sw SaMIIeae)
~11,89,91
Salt. 235, l40
Skill, effect, 86 depth, &7
s.-......... fonladoal.238,240, 241,304 '-._,
;;
~IMJD;I
s..l. s..t.to.e, 234, 235,238.241, 243 ~.254,273,279
attenuation, 203, 209, 211, 212. 216.222 velocity, 181, 182, 194, 197, 200, 201 Boise, 194, 196, 207, 210 Coconino, 207 Fontainebleau, 43-47, 185 attenuation, 159, 204, 206, 208, 243 velocity, ISO, 186, 188, 232 Fort Union, 199 Foxhills, 2S4, 279 Kern River, 196 Massillon attenuation, 169, lOS, 212, 214, 216, :us, 219, 222, 22S, 242 velocity,l77,179,182,187, 18&,199,200, 201 Navajo, 205, 222 Ottawa, 216, 231 porosity/permeability relation., )4, 44 relative permeability, 39 Saint Peter, 182, 197, 211 shaly, 27, 28, 37 Teapot,2S4,273,279 Venezuela, 192, 196, 211 Vosges, 166-168 Wingate, 207, 213 Satwation
irreductib1e, 24, 39 methods, 207, 208 residual, 25, 39 Seale, IC8hl ......... 40, 94, 302 Scatteriac. 91, 115, 146, 147, I», 197, 211, 214,303 cross-section, II S Seia1ic p~upediag borehole, 308 full waveform ~. 3l0 , low frequency, 302 '' ·, reflection, 302 reservoir, 302, 308 statigraphic, 304 three-component, 306 well-to-well, 310 Sbale Colorado, 203 Pierre, 178, 180, 182, 186, 241 Shape factor, 116, 180 Sllear ..... 162 Sierra White (see Grufte)
Soleala.r. (see U.llhlle) Source (.a-lc:). 269 SpecifiC area, 22t gravity (rocks). 240 Spectral ratio, 1S3-l 56 s,erp. (.r« Umestoae)
s.at--.227 Sf(alllt flow, 227 Staq4*,128 ~ ............ 110
Stall rroat, 309 Stelle (see Graaite)
Stoaeley mode, 274 wave, 266, 311 Stylolidle, 16
StniD principal directions, S l small, Sl, S3 teasor, 49 ~eyde,10S
s......
compressional, SS effective, 81
'I
iI
shear," telliOr, S2 traction,SS Sweep (fnlpleaey). 136, 162
SJDdledc rrf
i
J ~\
......., 276; 282
I
Syttaa open, 75,79 closed, 75, 79
Teapot (see Saadstoae) Terupi'alaw, 81-84, 93, 176 Textwe ...aysls, 18, 19
l'henlapa••en\....,., 12 ~221,224
nta .etioll (rock). 12, 14, 32 'J'IIree.colllpoMDt reccrii..,
306
• >.-/ (
'~
_:J;I
Tor.Honal~, 166 Tortuosity, 35, 71, 80, 81, 83, 93, 191, 252; 254 Traasmission coefficient, 250, 255-263, 290-296 modulus, 263, phase, 263, 295 method, 153, 172, 173 Tripolite, 231 Troy (see Granite) Tube wave, 275
m
o t \.\,~as- C\' ~). Unsteady-state, 128 \..c:..\ Y"CE.J<.-4 Ch~"do "C
Vector
attenuation, 260, 284 propagation, 284 Venezuela (see Saad) Velocity average value rocks, 230, 240 clay content, 179, 233, 2l_6 compaction, 237-239 complex, 286 dispersion, 197 energy, Ill extensional, 128, 141, 166 frequency, 197 group, 108-112 in situ measurement, 23o;.240 interval, 146 measurement, 148 phase, 108,149,260,286 porosity, 23() pressure, 177-186 Rayleigh, 265 saturating fluid phase, 196 viscosity, 192-196 saturation, 187-192 specific gravity, 233-235 Stoneley, 267 strain, 198 temperature, 193 uniaxial stress, 182 Vibrations, 128 . forced, 134, 161 free, 134, 161 · _,quality factor, 135
Viscoelasticln· (liaear), 107-ll2 interface ~aves, 297 reflection and transmission, 283-297 v-...coupllaa,87 ~. 194. 204, 210 V~tlon
capillary properties. 29, 30 porous medium, 12 VSP (vertical seismic profdiag), 240, 308 Vugy (Jiorous IDellium), 15, 16 Vycor, 191, 206. 213, 219, 222, 228
'
.,
'
~·'
\
·----..
Water content, 66, 252 Waftl dilatational. 59 equation, 61 extensional. 63, 128, 132 homogeneous, 284 inhomogeneous, 258, 284 interface. 263-269 longitudinal. 60, 63 primary. 58. 73 propagative. 128 Rayleigh, 263 refracted, 274 secondary, 60 shear, 70, 74 slow (second kind), 73-82, 83, 88., 252 standing, 161-169 Stoneley, 266, 311 transverse, 60 tube, 275 wave guidt~ 110 WeiMtuel (SH Debaite) Well loallll (acoustic), 146, 247, 255, 281 Westerly (see Graaite) Wettallility, 23 Wettillc/IIOil •"ddlli .... 22 Wi~~&&te (see Saadstone) WOCMI's metal, 29 WyDie's formula, 230-233
.,
~-,
.,. .,
-,
·,
...__
-.
' -.
---...
·-..
' ----.,
~,
YoiiRI'S modulus, 54, 57, 170
'·
Zener model, 117' 123
......
"
' \
~,
J