GEOTECHNICAL ENGINEERING

Computational Modeling of Multiphase Geomaterials discusses how

numerical methods play a very important role in geotechnical engineering
and in the related activity of computational geotechnics. It shows how
numerical methods and constitutive modeling can help predict the behavior of
geomaterials such as soil and rock.
After presenting the fundamentals of continuum mechanics, the book explores
recent advances in the use of modeling and numerical methods for multiphase
geomaterial applications. The authors describe the constitutive modeling of
soils for rate-dependent behavior, strain localization, multiphase theory,
and applications in the context of large deformations. They also emphasize
viscoplasticity and watersoil coupling.
Features
Explains how to predict the behavior of geomaterials
Contains the governing equations for multiphase geomaterials
Discusses the constitutive modeling of multiphase geomaterials,
including elastoplastic and elastoviscoplastic models
Presents numerical methods, such as the fnite element method,
for analyzing geomaterials
Covers the latest developments in geomechanics, including the
deformation-seepage fow coupled analysis of an unsaturated
river embankment
Drawing on the authors well-regarded work in the feld, this book provides
you with the knowledge and tools to tackle problems in geomechanics. It gives
you a comprehensive understanding of how to apply continuum mechanics,
constitutive modeling, fnite element analysis, and numerical methods to
predict the behavior of soil and rock.
ISBN: 978-0-415-80927-6
9 780415 809276
90000
w w w . s p o n p r e s s . c o m
Y132935
C
O
M
P
U
T
A
T
I
O
N
A
L

M
O
D
E
L
I
N
G

O
F
M
U
L
T
I
P
H
A
S
E

G
E
O
M
A
T
E
R
I
A
L
S
FUSAO OKA
SAYURI KIMOTO
COMPUTATIONAL
MODELING OF
MULTIPHASE
GEOMATERIALS
O
K
A
K
I
M
O
T
O
A S P ON PR E SS BOOK
Y132935 cvr mech.indd 1 6/7/12 10:11 AM
COMPUTATIONAL
MODELING OF
MULTIPHASE
GEOMATERIALS
A SPON PRESS BOOK
COMPUTATIONAL
MODELING OF
MULTIPHASE
GEOMATERIALS
FUSAO OKA
SAYURI KIMOTO
CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742
2013 by Taylor & Francis Group, LLC
CRC Press is an imprint of Taylor & Francis Group, an Informa business
No claim to original U.S. Government works
Version Date: 20120619
International Standard Book Number-13: 978-1-4665-7064-1 (eBook - PDF)
This book contains information obtained from authentic and highly regarded sources. Reasonable efforts
have been made to publish reliable data and information, but the author and publisher cannot assume
responsibility for the validity of all materials or the consequences of their use. The authors and publishers
have attempted to trace the copyright holders of all material reproduced in this publication and apologize to
copyright holders if permission to publish in this form has not been obtained. If any copyright material has
not been acknowledged please write and let us know so we may rectify in any future reprint.
Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced, transmit-
ted, or utilized in any form by any electronic, mechanical, or other means, now known or hereafter invented,
including photocopying, microfilming, and recording, or in any information storage or retrieval system,
without written permission from the publishers.
For permission to photocopy or use material electronically from this work, please access www.copyright.
com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood
Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization that provides licenses and
registration for a variety of users. For organizations that have been granted a photocopy license by the CCC,
a separate system of payment has been arranged.
Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used
only for identification and explanation without intent to infringe.
Visit the Taylor & Francis Web site at
http://www.taylorandfrancis.com
and the CRC Press Web site at
http://www.crcpress.com
v
Contents
Preface xv
Acknowledgments xvii
1 Fundamentals in continuum mechanics 1
1.1 Motion 1
1.2 Strain and strain rate 2
1.2.1 Strain tensor 2
1.2.2 Compatibility relation of strain 4
1.2.3 Shear strain and deviatoric strain 5
1.2.4 Volumetric strain 6
1.3 Changes in area 7
1.4 Deformation rate tensor 8
1.5 Stress and stress rate 10
1.5.1 Stress tensor 10
1.5.2 Principal stresses and the
invariants of the stress tensor 12
1.5.3 Stress rate tensor and objectivity 16
1.6 Conservation of mass 19
1.7 Balance of linear momentum 20
1.8 Balance of angular momentum and the
symmetry of the stress tensor 22
1.9 Balance of energy 23
1.10 Entropy production and ClausiusDuhem inequality 24
1.11 Constitutive equation and objectivity 26
1.11.1 Principle of objectivity and constitutive model 27
1.11.2 Time shift 28
1.11.3 Translational motion 28
1.11.4 Rotational motion 28
References 29
vi Contents
2 Governing equations for multiphase geomaterials 31
2.1 Governing equations for fuidsolid two-phase materials 31
2.1.1 Introduction 31
2.1.2 General setting 32
2.1.3 Density of mixture 33
2.1.4 Defnition of the effective and partial
stresses of the fuidsolid mixture theory 34
2.1.5 Displacementstrain relation 34
2.1.6 Constitutive model 35
2.1.7 Conservation of mass 35
2.1.8 Balance of linear momentum 35
2.1.9 Balance equations for the mixture 38
2.1.10 Continuity equation 38
2.2 Governing equations for gaswater
solid three-phase materials 41
2.2.1 Introduction 41
2.2.2 General setting 41
2.2.3 Partial stresses 42
2.2.4 Conservation of mass 43
2.2.5 Balance of momentum 45
2.2.6 Balance of energy 46
2.3 Governing equations for unsaturated soil 46
2.3.1 Partial stresses for the mixture 47
2.3.2 Conservation of mass 48
2.3.3 Balance of linear momentum for the three phases 49
2.3.4 Continuity equations 51
References 52
3 Fundamental constitutive equations 55
3.1 Elastic Body 55
3.2 Newtonian viscous fuid 57
3.3 Bingham body and viscoplastic body 58
3.4 von Mises plastic body 59
3.5 Viscoelastic constitutive models 59
3.5.1 Maxwell viscoelastic model 60
3.5.2 KelvinVoigt model 61
3.5.3 Characteristic time 62
Contents vii
3.6 Elastoplastic Model 63
3.6.1 Yield conditions 64
3.6.2 Additivity of the strain 65
3.6.3 Loading conditions 66
3.6.4 Stability of elastoplastic material 67
3.6.5 Maximum work theorem 69
3.6.6 Flow rule and normality (evolutional
equation of plastic strain) 71
3.6.7 Consistency conditions 73
3.7 Overstress type of elastoviscoplasticity 76
3.7.1 Perzynas model 76
3.7.2 Duvaut and Lions model 77
3.7.3 Phillips and Wus model 80
3.8 Elastoviscoplastic model based on stress history tensor 80
3.8.1 Stress history tensor and kernel function 81
3.8.2 Flow rule and yield function 81
3.9 Other viscoplastic and viscoelasticplastic theories 83
3.10 Cyclic plasticity and viscoplasticity 83
3.11 Dissipation and the yield functions 86
References 89
4 Failure conditions and the Cam-clay model 91
4.1 Introduction 91
4.2 Failure criteria for soils 92
4.2.1 Failure criterion by Coulomb 92
4.2.2 Failure criterion by Tresca 93
4.2.3 Failure criterion by von Mises 93
4.2.4 Failure criterion by Mohr 94
4.2.5 MohrCoulomb failure criterion 94
4.2.6 MatsuokaNakai failure criterion 94
4.2.7 Lade failure criterion 95
4.2.8 Failure criterion on plane 95
4.2.9 Lode angle and MohrCoulomb failure condition 98
4.3 Cam-clay model 102
4.3.1 Original Cam-clay model 102
4.3.2 Ohtas theory 108
4.3.3 Modifed Cam-clay model 110
4.3.4 Stressdilatancy relations 112
References 113
viii Contents
5 Elastoviscoplastic modeling of soil 115
5.1 Rate-dependent and time-dependent behavior of soil 115
5.1.1 Strain rate-dependent behavior of clayey soil 115
5.1.2 Creep deformation and failure 117
5.1.3 Stress relaxation behavior 120
5.1.4 Strain rate-dependent compression 120
5.1.5 Isotaches 123
5.2 Viscoelastic constitutive models 125
5.3 Elastoviscoplastic constitutive models 126
5.3.1 Overstress models 126
5.3.2 Time-dependent model 127
5.3.3 Viscoplastic models based on
the stress history tensor 127
5.4 Microrheology models for clay 128
5.5 Adachi and Okas viscoplastic model 128
5.5.1 Strain rate effect 137
5.5.2 Simulation by the Adachi and Okas model 138
5.5.2.1 Effect of secondary consolidation 138
5.5.2.2 Isotropic stress relaxation 140
5.5.3 Constitutive model for anisotropic
consolidated clay 141
5.6 Extended viscoplastic model considering
stress ratio-dependent softening 141
5.7 Elastoviscoplastic model for cohesive
soil considering degradation 142
5.7.1 Elastoviscoplastic model considering degradation 142
5.7.2 Determination of the material parameters 148
5.7.3 Strain-dependent elastic shear modulus 149
5.8 Application to natural clay 150
5.8.1 Osaka Pleistocene clay 150
5.8.2 Osaka Holocene clay 152
5.8.3 Elastoviscoplastic model based on
modifed Cam-clay model 153
5.9 Cyclic elastoviscoplastic model 156
5.9.1 Cyclic elastoviscoplastic model based on
nonlinear kinematical hardening rule 157
5.9.2 Cyclic elastoviscoplastic model
considering structural degradation 158
5.9.2.1 Static yield function 158
Contents ix
5.9.2.2 Viscoplastic potential function 159
5.9.2.3 Kinematic hardening rules 159
5.9.2.4 Strain-dependent shear modulus 161
5.9.2.5 Viscoplastic fow rule 162
References 164
6 Virtual work theorem and fnite element method 171
6.1 Virtual work theorem 171
6.1.1 Boundary value problem 171
6.1.2 Virtual work theorem 173
6.2 Finite element method 175
6.2.1 Discretization of equilibrium equation 175
6.2.2 Discretization of continuity equation 179
6.2.3 Interpolation function 180
6.2.4 Triangular element 181
6.2.5 Isoparametric elements 183
6.3 Dynamic Problem 190
6.3.1 Time discretization method 191
6.3.1.1 Linear acceleration method
and Wilson method 191
6.3.1.2 Newmark method 192
6.3.1.3 Central fnite difference scheme 193
6.3.2 Mass matrix 193
6.4 Dynamic analysis of water-saturated soil 193
6.4.1 Equation of motion 194
6.4.2 Continuity equation 201
6.4.2.1 Galerkin method 201
6.4.2.2 Finite volume method 202
6.4.3 Time discretization 206
6.4.3.1 Equation of motion 207
6.4.3.2 Continuity equation 207
6.5 Finite deformation analysis for fuid
solid two-phase mixtures 210
6.5.1 Effective stress and fuidsolid mixture theory 210
6.5.2 Equilibrium equation 211
6.5.3 Continuity equation 214
6.5.4 Discretization of the weak forms
for the equilibrium equation and
the continuity equation 216
x Contents
6.5.4.1 Discretization of the weak forms
for the equilibrium equation 216
6.5.4.2 Discretization of the weak form
for the continuity equation 219
References 220
7 Consolidation analysis 223
7.1 Consolidation behavior of clays 223
7.2 Consolidation analysis: small strain analysis 225
7.2.1 One-dimensional consolidation problem 225
7.2.2 Two-dimensional consolidation problem 230
7.2.3 Summary 234
7.3 Consolidation analysis with a model
considering structural degradation 234
7.3.1 Effect of sample thickness 235
7.3.2 Simulation of Aboshis experimental results 238
7.3.2.1 Determination of material parameters 238
7.3.2.2 Elastic parameters 238
7.3.2.3 Viscoplastic parameters 238
7.3.2.4 Consolidation analysis 239
7.3.3 Effect of degradation 241
7.4 Consolidation analysis of clay foundation 244
7.4.1 Introduction 244
7.4.2 Consolidation analysis of soft clay
beneath the embankment 244
7.4.2.1 Soil parameters 244
7.4.2.2 Soil response beneath embankment 245
7.5 Consolidation analysis considering
construction of the embankment 249
7.5.1 Numerical example 251
References 255
8 Strain localization 259
8.1 Strain localization problems in geomechanics 259
8.1.1 Angle of shear band 260
8.2 Localization analysis 261
8.3 Instability of geomaterials 264
8.4 Noncoaxiality 270
8.5 Current stress-dependent characteristics and anisotropy 271
Contents xi
8.6 Regularization of Ill-posedness 271
8.6.1 Nonlocal formulation of constitutive models 272
8.6.2 Fluidsolid two-phase formulation 273
8.6.3 Viscoplastic regularization 273
8.6.4 Dynamic formulation 273
8.6.5 Discrete model and fnite element
analysis with strong discontinuity 274
8.7 Instability and effects of the transport of pore water 274
8.7.1 Extended viscoplastic models for clay 276
8.7.2 Instability analysis of fuid-
saturated viscoplastic models 278
8.7.2.1 Instability under locally
undrained conditions 278
8.7.2.2 Instability analysis considering
the pore water fow 281
8.8 Two-dimensional fnite element analysis
using elastoviscoplastic model 282
8.8.1 Effects of permeability 282
8.8.2 Strain localization analysis by the gradient-
dependent elastoviscoplastic model 286
8.8.2.1 Finite element formulation
of the gradient-dependent
elastoviscoplastic model 286
8.8.2.2 Effect of the strain gradient parameter 287
8.8.2.3 Effect of the heterogeneity
of the soil properties 288
8.8.2.4 Mesh-size dependency 290
8.9 Three-dimensional strain localization
analysis of water-saturated clay 291
8.9.1 Undrained triaxial compression tests
for clay using rectangular specimens 292
8.9.1.1 Clay samples and the testing program 292
8.9.1.2 Image analysis 293
8.9.2 Three-dimensional soilwater coupled
fnite element analysis method 294
8.9.3 Numerical simulation of triaxial
tests for rectangular specimens 295
8.9.3.1 Determination of the
material parameters 295
8.9.3.2 Boundary conditions 296
xii Contents
8.9.3.3 Comparison between experimental
and simulation results 297
8.9.3.4 Three-dimensional shear bands 301
8.9.3.5 Effects of the strain rates 302
8.10 Application to bearing capacity and
earth pressure problems 305
8.11 Summary 306
References 307
9 Liquefaction analysis of sandy ground 317
9.1 Introduction 317
9.2 Cyclic constitutive models 317
9.3 Cyclic elastoplastic model for sand
with a generalized fow rule 319
9.3.1 Basic assumptions 319
9.3.2 Overconsolidation boundary surface 319
9.3.3 Fading memory of the initial anisotropy 321
9.3.4 Yield function 322
9.3.5 Plasticstrain dependence of the shear modulus 324
9.3.5.1 Method 1 324
9.3.5.2 Method 2 324
9.3.5.3 Method 3 325
9.3.5.4 Method 4 325
9.3.6 Plastic potential function 326
9.3.7 Stressstrain relation 328
9.4 Performance of the cyclic model 329
9.4.1 Determination of material parameters 329
9.5 Liquefaction analysis of a liquefable ground 332
9.5.1 Vertical array records on Port Island 334
9.5.2 Numerical models 334
9.5.3 Common parameters 335
9.5.4 Parameters for elastoplasticity model 338
9.5.5 Parameters for elastoviscoplasticity model 338
9.5.6 Parameters for RambergOsgood model 339
9.5.7 Finite element model and numerical parameters 340
9.5.8 Numerical results 340
Contents xiii
9.6 Numerical analysis of the dynamic behavior of
a pile foundation considering liquefaction 341
9.6.1 Simulation methods 344
9.6.2 Results and discussions 345
References 349
10 Recent advances in computational geomechanics 353
10.1 Thermo-hydro-mechanical coupled fnite element method 353
10.1.1 Temperature-dependent viscoplastic parameter 354
10.1.2 Elastic and temperature-dependent stretching 357
10.1.3 Weak form of the equilibrium
equation for watersoil mixture 358
10.1.4 Continuity equation 360
10.1.5 Balance of energy 363
10.1.6 Simulation of thermal consolidation 365
10.2 Seepagedeformation coupled analysis
of unsaturated river embankment using
multiphase elastoviscoplastic theory 370
10.2.1 Introduction 370
10.2.2 Governing equations and analysis method 371
10.2.3 Constitutive model for unsaturated soil 371
10.2.3.1 Overconsolidation boundary surface 371
10.2.3.2 Static yield function 372
10.2.3.3 Viscoplastic potential function 373
10.2.3.4 Viscoplastic fow rule 373
10.2.3.5 Constitutive model for pore water:
soilwater characteristic curve 374
10.2.4 Simulation of the behavior of unsaturated
soil by elastoviscoplastic model 375
10.2.5 Numerical analysis of seepage
deformation behavior of a levee 376
10.2.5.1 Analysis method 376
10.2.5.2 Deformation during the seepage fow 377
References 385
xv
Preface
Over the last three decades, studies on constitutive models and numeri-
cal analysis methods have been well developed. Nowadays, numerical
methods play a very important role in geotechnical engineering and in a
related activity called computational geotechnics. This book deals with
the constitutive modeling of multiphase geomaterials and numerical meth-
ods for predicting the behavior of geomaterials such as soil and rock. The
book provides fundamental knowledge of continuum mechanics, con-
stitutive modeling, numerical methods for multiphase geomaterials, and
their applications. In addition, the monograph includes recent advances
in this area, namely, the constitutive modeling of soils for rate-dependent
behavior, strain localization, the multiphase theory, and their applications
in the context of large deformations. The presentation is self-contained.
Much attention has been paid to viscoplasticity, watersoil coupling, and
strain localization.
Chapter 1 presents the fundamental concept and results in continuum
mechanics, such as motion, deformation, and stress, which are necessary
for understanding the following chapters. This chapter helps readers make
a self-consistent study of the contents of this book.
Chapter 2 deals with the governing equations for multiphase geomaterials
based on the theory of porous media, such as water-saturated and air
watersoil multiphase soils including soilwater characteristic curves. This
chapter is essential for the study of computational geomechanics.
Chapter 3 starts with the elastic constitutive model and reviews the fun-
damental constitutive models including plasticity and viscoplasticity. For the
plasticity theory, the stability concept in the sense of Lyapunov is discussed.
At the end of the chapter, cyclic plasticity and viscoplasticity models are
presented with kinematical hardening rules.
In Chapter 4, failure criteria and the Cam-clay model are reviewed. For the
failure criteria, many well-known criteria have been proposed in this chapter,
from Coulombs criterion to MatsuokaNakais criterion. Then, the Cam-
clay model is reviewed since the model includes a description of the basic
properties of soil behavior such as dilatancy and the critical state concept.
xvi Preface
Chapter 5 is devoted to the rate- and time-dependent behavior and
modeling of soils. At frst, typical rate- and time-dependent behaviors of
soils are reviewed based on the experimental measurements. Several rate-
dependent models are discussed and elastoviscoplastic models based on
the Cam-clay model and Perzynas viscoplasticity theory are presented.
Adachi and Okas model is frst described and then an elastoviscoplastic
model considering structural degradation is introduced. The chapter ends
with the calibration of these models using the experimental results.
In Chapter 6, the virtual work theorem is presented and then the fnite
element method for two-phase materials is described for quasi-static and
dynamic problems within the framework of the infnitesimal strain theory.
Chapter 7 deals with a typical multiphase phenomenon of soils; namely,
the consolidation problem. In particular, the effects of sample thickness on
consolidation, using Aboshis well-known data, and the anomalous behav-
ior of pore water development in the clay foundation beneath the embank-
ment, during loading and after the end of construction embankment, are
numerically analyzed.
Chapter 8 starts with a review of the study on the strain localization
behavior of soils. Several issues related to the strain localization are then
discussed for rate-independent and rate-dependent models. Finally, a
numerical analysis of the strain localization of water-saturated clay is pre-
sented for triaxial tests and practical problems.
In Chapter 9, a liquefaction analysis method is presented with a cyclic
elastoplastic model using the two-phase theory presented in Chapter 2 for
water-saturated soils. Applications of the liquefaction behavior to a man-
made island during an earthquake and of the soilpilestructure interac-
tion are shown.
Chapter 10 deals with recent advances in geomechanics. It includes the
temperature-dependent behavior of soils such as consolidation due to the
change in temperature, and the numerical analysis of airwatersoil cou-
pled problems; namely, the deformationseepage fow coupled analysis of
an unsaturated river embankment is presented.
xvii
Acknowledgments
During the writing and preparation of this book, the authors became
indebted to many researchers and students. In particular, we express our
sincere thanks to Dr. K. Akai and Dr. T. Adachi, Emeritus Professors
of Kyoto University; Dr. H. Aboshi, Emeritus Professor of Hiroshima
University, for giving us data on consolidation; Dr. S. Leroueil, Professor
of Laval University; Dr. A. Yashima of Gifu University; Dr. T. Kodaka of
Meijo University; Dr. R. Uzuoka of Tokushima University; Dr. Y. Higo
of Kyoto University; Dr. F. Zhang of Nagoya Institute of Technology;
Dr. K. Sekiguchi; Dr. A. Tateishi; Dr. Y. Taguchi; Dr. S. Sunami; Dr. M.
Kato; Dr. M. J. Jiang, Dr. C.-W. Lu; Y.-S. Kim; Dr. Garcia; Dr. Mojtaba
Mirjalili; Dr. R. Kato; Dr. Young Seok Kim; Dr. A. W. Karnawardena; Dr.
H. Feng; Dr. Md. R. Karim; Dr. B. Siribumrungwong; Mr. T. Takyu; Mr.
T. Satomura; Mr. N. Nishimatsu; Ms. T. Ichinose; Mr. Takada; and the
graduate students of the Geomechanics Laboratory of Kyoto University for
their contributions and discussions. We thank Ms. Chikako Itoh for her
daily assistance; Mr. Shahbodagh Khan Babak, a PhD student of Kyoto
University, for his assistance in preparing the fgures; and Ms. H. Griswold
for her English corrections. Finally, we dedicate this book to our families,
in particular, O. Keiko and K. Keiko.
Many thanks are also due to the following organizations and the researchers
for permission for use the indicated fgures: Professor H. Aboshi, Figure5.8;
Professor T. Adachi, Figure 5.7a,b; Professor Liam Finn, Figure 5.3a,b
and Figure 5.5; Gihodo Syuppan Co. Ltd. (Dr. M. Saito), Figure 5.6;
Professor G. Sllfore, Figure 5.9; American Society of Civil Engineers
(ASCE), Figure10.1 and Figure10.3; ASTM, Figure5.2 and Figure5.12a,b;
Institution of Civil Engineers (Gotechnique), Figure5.1a,b and Figure5.11;
and National Research Council of Canada (NRC) (Canadian Geotechnical
Journal), Figure5.10.
1
Chapter 1
Fundamentals in
continuum mechanics
In this book, we use vectors and tensors in components, and the direct
notations for these vectors and tensors are given without further explana-
tion. A dot denotes a contraction of the inner indices, for example, a b
i i


a b so that A B
ij ij
A:B.
1.1 MOTION
The position of the material point X i
i
( 1 2 3) = , , of a body at time t is
expressed by

x
x
X t
i i j
( ) = ,
(1.1)
Material point X
i
can be given by the position of x
i
at a time t = 0.
Equation (1.1) expresses the motion of the material point of the body.
The rectangular Cartesian coordinates used in this book are described by
o
e e e
( )
1 2 3
, , ,
with origin o and unit base vector ei

.
There are two methods for describing the motion of a particle. One is the
material description, in which the motion is expressed by material point X
i
,
and the other is the spatial description, in which the motion is expressed
by spatial coordinates x
i
. The material description is called the Lagrangian
description and the spatial description is called the Eulerian description.
The velocity vector of a particle is given by

v
x X t
t
i
i j
( )
=
,

(1.2)
In the material description, the acceleration of a particle in a body is
expressed by

a
v X t
t
i
i j
( )
=
,

(1.3)
2 Computational modeling of multiphase geomaterials
In the spatial description, on the other hand, the acceleration of a particle
is given by

a
v x t
t
v
v x t
x
i
i j
k
i j
k
( ) ( )
=
,

+
,

(1.4)
1.2 STRAIN AND STRAIN RATE
1.2.1 Strain tensor
Strain is the change in shape or the change in volume of a body during the
application of force to the body. We need an objective measure of strain
that can be derived through changes in the variation of the line element.
Let us consider the motion of the body shown in Figure 1.1. Material
points P and Q have moved to points P and Q after the deformation.
Points Q and Q are the points located in the vicinity of points P and P.
Distance, dS, between points P and Q, is given by

dS dX dX
a a
2
= (1.5)
and the distance between points P and Q after the deformation, ds, is
given by

ds dx dx
b b
2
= (1.6)
where the summation convention is used for a,b = 1,2,3.
u
i
+ du
i
u
i
dx
i
X
i
x
i
P
dX
i
Q
Q
P
x
1
, X
1
x
2
, X
2
x
3
, X
3
Figure 1.1 Motion.
Fundamentals in continuum mechanics 3
Displacement vector u i
i
( 1 2 3) = , , is given by

x X u
i i i
= + (1.7)
Taking the difference between Equations (1.5) and (1.6), we have

ds dS dx dx dX dX F F dX dX
x
X
x
X
dX dX E dX dX
k k k k ki kj ij i j
k
i
k
j
ij i j ij i j
( )
2
2 2
= =
=

=
(1.8)
where F
ij
x
X
i
j
=

is the deformation gradient and i j i j

ij
(1 0 ) = , = ; , is
Kroneckers delta. E
ij
in Equation (1.8) is called the Green strain tensor.
Substituting Equation (1.7) into Equation (1.8), we get

E
u
X
u
X
u
X
u
X
ij
i
j
j
i
k
i
k
j
1
2
=

(1.9)
For the case of infnitesimal strain, that is,
u
X
i
j
1 | |<<

, the strain tensor is

called an infnitesimal strain tensor. It is expressed as

u
x
u
x
ij
i
j
j
i
1
2
=

(1.10)
Engineering strain,
ij
, is defned for the shear components as

i j
i j
ij ij
ij ij
2 ( )
( )
=
= =
(1.11)
It is worth noting that the engineering strain is not a tensor.
Relative displacement, du
i
, is determined by the gradient of the dis-
placement,
u
x
i
j

, as

du
u
x
dx dx dx
i
i
j
j ij j ij j
=

= + (1.12)
4 Computational modeling of multiphase geomaterials
where
ij
is given by

u
x
u
x
ij
i
j
j
i
1
2
=

(1.13)
and represents the rotation of a small element, for example, the rotation is
not zero for the rigid body motion.
1.2.2 Compatibility relation of strain
The strain tensor has six components, although the displacement vector
has only three components. This indicates that we need three indepen-
dent equations to obtain the displacement vector from the strain tensor.
However, six compatibility equations exist among the strain compo-
nents. As for the compatibility equations, three of them are indepen-
dent, and compatibility equations are necessary and provide suffcient
conditions for single-value displacements in a simple connected body
(see Malvern 1969).
As for the differentiation of the displacementstrain relations with
respect to coordinates, we obtain the compatibility equations as

y x x y
xx yy xy
2
2
2
2
2

=


(1.14)

z y y z
yy zz yz
2
2
2
2
2

=


(1.15)

x z z x
zz xx zx
2
2
2
2
2

=


(1.16)

y z x x y z
xx yz xz xy
2
2


=

(1.17)

x z y x y z
yy yz xz xy
2
2


=

(1.18)

x y z x y z
zz yz xz xy
2
2


=

(1.19)
Fundamentals in continuum mechanics 5
Equations (1.14) to (1.19) can be expressed by a tensor notation as

ij kl kl ij ik jl jl ik
0 + =
, , , ,
(1.20)
where i,j,k,l = 1,2,3.
1.2.3 Shear strain and deviatoric strain
Let us consider the deformation shown in Figure 1.2. The displacement
vector is given by u c y u c x c c
x y
0
1 2 1 2
= , = , , > .
Then,

u
y
u
x
c c
xy xy
x y
2 ( )
1 2
= =

= + (1.21)
and the strain components,
xx yy
, , are zero.
Assuming a small deformation gradient,
u
y
x

is given by
1
and
u
x
y

is
given by
2
, namely,

u
y
u
x
x y
1 2

= + (1.22)
This indicates that
xy
expresses changes in the angle, in other words,
changes in the shape, that is, shearing deformation.

e
ij ij ij kk
1
3
= (1.23)
is defned as the deviatoric strain tensor.
x
O

1
u
x
u
x
y
Figure 1.2 Shear deformation.
6 Computational modeling of multiphase geomaterials
1.2.4 Volumetric strain
Setting V as the volume after the deformation and V
0
as the volume before
the deformation, volumetric strain
v
is expressed by V V V
v kk
( )
0 0
= = / .
If we express the volume before the deformation by V dX dX dX
0 1 2 3
= ,

dX dX dX dX dX dX dX dX dX
o
v
[(1 )(1 )(1 ) ]
( )
11 22 33 1 2 3 1 2 3 1 2 3
11 22 33
= + + + /
= + + +

(1.24)
where o() is the higher-order small term.
We can disregard the higher-order term for the small deformation case.
Next, we will consider the changes in volume for the fnite deformation case.
The volume of the small hexahedron after the deformation, dV, is given by

dV d d d dx dx dx
ijk i j k
x x x ( ) = = (1.25)
where
ijk
is a permutation (or alternating) symbol.
The volume of the small hexahedron before the deformation, dV
0
, is
given by

dV d d d dX dX dX
pqr p q r
X X X ( )
0
= = (1.26)
Using the deformation gradient, F
ij
x
X
i
j
=

, we get

J det F F F F
mn ijk pqr ip jq kr
( )
1
6
= (1.27)
Using the following relation:

ijk pqr
ip iq ir
jp jq jr
kp kq kr
=



(1.28)
we obtain

det F F F F
pqr mn ijk ip jq kr
( ) = (1.29)
As for Equation (1.29), it is worth noting that det F F F F
mn ijk i j k
( )
1 2 3
=
following the expansion of the determinant det F
mn
( ).
Fundamentals in continuum mechanics 7
If pqr is an even permutation of 1,2,3, we have the determinant and if
pqr is odd we have the negative one.
Consequently, we have

dV F F F dX dX dX
det F dX dX dX JdV dV
ijk ip jq kr p q r
pqr mn p q r
( )
0
0
0
=
= = =

(1.30)
where
0
and are the initial mass density and the current mass density,
respectively.
Disregarding the higher-order term leads to

J det F
u
X
ij
i
i
( ) 1 = +

(1.31)
Therefore, we obtain the following relation consistent to Equation (1.24) as

dV dV
dV
ii v
0
0

= = (1.32)
1.3 CHANGES IN AREA
The changes in area have been estimated by Nansons formula in contin-
uum mechanics (Malvern 1969) and are given by

ds J dS
T
n F N
0
=

(1.33)
where n is the unit normal to area ds in the current confguration, ds is an
area in the current confguration, N is the unit normal to the initial con-
fguration, F
1
is the inverse of the deformation gradient, and dS
0
is an area
in the initial confguration.
Surface vector dS N
0
at point X in the referential confguration is
expressed by

dS d d N X X
0
= (1.34)
where dX is an infnitesimal vector at point X.
Surface vector ds n at point x in the current confguration is expressed by

ds d d n ds dx dx
s spq p q
n x x, ( ) = = (1.35)
where dx is an infnitesimal vector at point x.
8 Computational modeling of multiphase geomaterials
From Equation (1.34), we obtain

N dS
X
x
X
x
dx dx
i ijk
j
p
k
q
p q 0
=

(1.36)
Then, multiplying both sides of Equation (1.36) by
X
x
i
s

and using
Equations (1.28) and (1.35), we get

J
X
x
N dS n ds
i
s
i s 0

= (1.37)
in which the relation
J
X
x
X
x
X
x
spq ijk
i
s
j
p
k
q
1
=

Consequently, we get Equation (1.33), called Nansons theorem, since

components of F
1
and
T
F

are

F
X
x
F
X
x
ij
i
j
ij
T j
i
,
1
=

1.4 DEFORMATION RATE TENSOR
When we deal with the large deformation of a body, the material confguration
changes each time, and the deformation rate tensor is useful for the analysis.
Taking a time derivative of Equation (1.8) leads to

d
dt
ds dS dx
d
dt
dx
k k
( ) 2 ( )
2 2
=
(1.38)

d
dt
dx
d
dt
x
X
dX
x
X
d
dt
dX
k
k
m
m
k
m
m
( ) =

v
X
dX
k
m
m
=

dv L dx
k km m
= = (1.39)

L
v
x
ij
i
j
=

(1.40)
where L
ij
is called the velocity gradient tensor.
Fundamentals in continuum mechanics 9
The velocity gradient tensor can be separated into symmetric part D
ij

and antisymmetric part W
ij
as

L D W
ij ij ij
= + (1.41)

D
v
x
v
x
ij
i
j
j
i
1
2
=

(1.42)

W
v
x
v
x
ij
i
j
j
i
1
2
=

(1.43)
Substituting the preceding equations into Equation (1.39) gives

d
dt
ds dS dx
d
dt
dx dx v dx dx L dx
dx D dx dxW dx
i i i i m m i im m
i im m i im m
( ) 2 ( ) 2 2
2 2
2 2
= = =
= +
,
(1.44)
Since W
ij
is skew symmetric, dxW dx
i im m
2 0 = .
Hence, we have

d
dt
ds dS
d
dt
ds dx D dx
i im m
( ) ( ) 2
2 2 2
= = (1.45)
Subsequently, D
ij
is used to express the measure of the deformation rate
at the current confguration, which is called the rate-of-deformation tensor
or the stretching tensor. In contrast, W
ij
denotes the rate of rotation and is
called the spin.
In a small deformation feld, we do not distinguish the deformation
between the current and the reference confgurations. Hence, we can use
strain rate ij
instead of the deformation rate tensor as:

u
x
u
x
ij
i
j
j
i
1
2

(1.46)
10 Computational modeling of multiphase geomaterials
1.5 STRESS AND STRESS RATE
1.5.1 Stress tensor
The forces acting on a body can be classifed into two forces: the body force
and the surface force. The body force is the force acting on the body remotely,
such as the gravitational force, which is proportional to the mass volume.
The surface force is the force acting on the body through the surface, which
is proportional to the area of the surface and is called the stress vector.
Let us consider the surface force t ds t x n ( , , ) , shown in Figure1.3(a), acting
on the small surface element ds of the cross section of the body at a position
x. t is a surface traction vector per unit area acting on side II from side I.
In contrast, the force acting on side I from side II has the same magnitude
of force as that from side I to side II, only in an opposite direction.

t ds t ds t x n t x n ( , ) ( , ) , = , (1.47)
in which n is the unit normal vector to the surface. The surface force per
unit area t is called the stress vector or the traction vector.
Consider the stress state of a tetrahedron, which is in equilibrium under
the surface, and the body forces shown in Figure1.3(b).
The area of ABC is S, the area of OBC is S
1
, the area of OCA is
S
2
, and the area of OAB is S
3
. Then,

S Sn n n n
i i
n ( )
1 2 3
= , = , , (1.48)
F
3
F
1
F
2
II
t(n)
t(n)
P
n
n
I

t(e
1
)
t(e
2
)
t(e
3
)
t(n)
e
1

2
e
2
e
3
C
P
n
x
2
O B
x
1
x
3
A
(a) (b)
Figure 1.3 (a) Force balance. (b) Traction vectors in equilibrium.
Fundamentals in continuum mechanics 11
The reason is as follows: When we set the intersection point between
the perpendicular line from point O and ABC as point P, and OP = h, the
volume of the tetrahedron is Sh/3.
Hence,

Sh S AO S BO S CO
1 2 3
= = = (1.49)
Then,

S S h AO n cos
1 1 1
/ = / =
(1.50)
where n
i
is the direction cosine cos
i
.
From the equilibrium of the forces acting on the tetrahedron, we have

S S S S Sh t n t e t e t e F a ( ) ( ) ( ) ( ) ( ) 3 0
1 1 2 2 3 3
+ + + + / = (1.51)
where F is the gravitational force and a is the inertial force.
As h0, Equation (1.51) becomes

n n n t n t e t e t e ( ) ( ) ( ) ( ) 0
1 1 2 2 3 3
+ + + = (1.52)
Equation (1.52) can be rewritten as

n
i
i i
t n t e ( ) ( ) 0
1
3

+ =
=
(1.53)
Then, if we use the following expression

n
k
k k
n e
1
3

=
=
(1.54)
the traction vector becomes

n n n
k k k k k k
t e t e t e ( ) ( ) ( ) = = (1.55)
where Einsteins summation convention n n
k
k k k k
e e
1
3

=
is used.
12 Computational modeling of multiphase geomaterials
The stress vector can be defned as

m
k
mk k
t e e ( )
1
3

=
(1.56)
From Equations (1.55) and (1.56), we get Cauchys fundamental theorem
of stress vector as

n
m k
mk k m
t n e ( )
1
3

=
, =
(1.57)
where
mk
is called the stress tensor.
The kth component of t is given by t
k
= n
mk m
.
mk
denotes the com-
ponent of the stress vector in the x
k
direction acting on the perpendicular
plane to the x
m
axis.
When we disregard the couple stress (see Section 1.8), the stress tensor
becomes symmetric from the equilibrium of the moment as

ij ji
= (1.58)
The Cauchy stress tensor is expressed by both
ij
and T
ij
in this book.
1.5.2 Principal stresses and the invariants
of the stress tensor
In general, the stress vector is not parallel to the normal vector of the sec-
tion, as shown in Figure1.3. In a certain direction, however, the stress vec-
tor is parallel to the direction of normal vector, n
i
, in which direction stress
vector, t
i
, can be expressed by

t n n
i ji j i
= =
(1.59)
where is the magnitude of the stress vector.
Since
ij
is symmetric, we have

n n
ji ij j ij ij i
( ) ( ) 0 = = (1.60)
Equation (1.60) is a set of linear homogeneous equations for n
i
and has a
nontrivial solution, that is, n
i
0 if and only if the following relation holds:

det
ij ij
0 | |=
(1.61)
Fundamentals in continuum mechanics 13
Equation (1.61) is an eigenvalue equation. When the stress tensor is
symmetric, Equation (1.61) has three real roots (real eigenvalues). These
three eigenvalues,
1 2 3
, , , are called principal stresses. The direction of
n
i
, which satisfes Equation (1.60), is called the principal stress direction.
Equation (1.61) can be written as

I I I 0
3
1
2
2 3
+ = (1.62)

I
1 11 22 33
= + + (1.63)

I ( )
2 11 22 22 33 33 11 12
2
23
2
31
2
= + + + + (1.64)

I 2 ( )
3 11 22 33 12 23 31 11 23
2
22 31
2
33 12
2
= + + + (1.65)
Since Equation (1.61) holds for the principal stress conditions, we have

I I I ( )( )( ) 0
3
1
2
2 3 1 2 3
+ = = (1.66)
Consequently, from the relation between roots and coeffcients, I I I
1 2 3
, ,
can be expressed as

I
1 1 2 3
= + + (1.67)

I
2 1 2 2 3 3 1
= + + (1.68)

I
3 1 2 3
= (1.69)
Since I I I
1 2 3
, , are invariants under the rotation of the coordinates, they are
called the frst, the second, and the third invariants, respectively.
Alternatively, these three invariants can be expressed by I I I
1 2 3
, , as

I
ii 1
= (1.70)

I
ij ij
1
2
2
= (1.71)

I
ij jk ki
1
3
3
= (1.72)
14 Computational modeling of multiphase geomaterials
The difference between the stress tensor and the mean value of stress ten-
sor,
m
, is called the deviatoric stress tensor, s
ij
, as

s
ij ij m ij
= (1.73)

m
1
3
( )
11 22 33
= + + (1.74)
For the deviatoric stress tensor, three stress invariants exist, J J J
1 2 3
, , , as

J J s s J s s s
ij ij ij jk ki
0
1
2
1
3
1 2 3
= , = , = (1.75)
The angle of the coordinates for specifying the principal stresses is
obtained by setting
xy
0 = in Equation (1.77) as

xy
xx yy
tan2
2
=


(1.76)

xy xy
xx yy
cos2
( )
2
sin2 =

(1.77)
where
xy
is a component of the stress tensor, which is transformed with
respect to the rotation of the coordinates, and is the angle between the
reference coordinates and corresponds to the principal stress directions.
PROBLEM
Show the principal stresses and their directions for the following stress tensor:

[ ]
ij
=
4 1 1
1 2 1
1 1 2
Answer: If we set as the principal stress,

det
4 1 1
1 2 1
1 1 2
0

Fundamentals in continuum mechanics 15

This yields ( ) ( ) ( ) ( ) 2 4 2 4 2 2 0
2
+ = . Then, the principal stresses are

1
5 = ,
2
2 = , and
3
1 = .
From Equation (1.60), the direction corresponding to 5
1
= is given
as n n n n n n n n n 0 3 0 3 0
1 2 3 1 2 3 1 2 3
+ + = , + = , + = . Then,
: : n n n
1 2 3
= : : 2 1 1, where n n n n
i
( )
1 2 3
= , , are the components of the unit principal
direction vector. Similarly, the principal direction for 2
2
= is obtained as
n n n n n n n 2 0 0 0
1 2 3 1 3 1 2
+ + = , + = , + = . Then, n n n 1 1 1
1 2 3
: : = : : . The
principal direction for 1
3
= is n n n n n n 3 0 0
1 2 3 1 2 3
+ + = , + + = ,

and then
n n n 0 1 1
1 2 3
: : = : : .
Let us consider the current force vector by the nominal stress vector that
is the stress vector with respect to surface area dS
0
in the reference confgu-
ration as

t T n
i ji j
=
(1.78)

s N
i ji j
= (1.79)
where
ij
is the nominal stress tensor or the frst PiolaKirchhoff stress tensor.
Since the force is in equilibrium,

t ds s dS
i i 0
= (1.80)
Nansons theorem, Equation (1.37), gives

JN dS
x
X
n ds
j
s
j
s 0
=

(1.81)
By substituting Equations (1.78) and (1.79) for Equation (1.80) and using
Equation (1.81), we obtain

JT
x
X
ki
k
j
ji
=

(1.82)
Then,

= =

J J
X
x
T
ij
i
k
kj
F T or
1
(1.83)
16 Computational modeling of multiphase geomaterials
1.5.3 Stress rate tensor and objectivity
When we consider the stress rate, we have to examine the objectivity
of it. The objectivity is defned as the independence of the motion with
respect to the observers. Since the physical law has to be objective, the
physical law including the constitutive equation of materials should sat-
isfy the objectivity. Here, we will discuss the objectivity of the stress and
the stress rate tensor. The objectivity for the constitutive equations will be
discussed in the next section. Herein, the objectivity follows mostly that
by Malvern (1969).
The observer for an event is called a reference frame. For the different
observers, there exists a transformation among them, which is expressed by
a Euclid transformation.
The Euclid transformation between two frames, t x ( , ) and t x ( , )
* *
, is
given by

t t x Q x c ( ) ( ) = +

(1.84)

t t t
0
=

(1.85)
where t Q( ) denotes an orthogonal tensor that expresses the rotation between
two frames, t c( ) is the relative motion of the origin, and t
0
expresses the
time difference.
For the change in reference frame from x to x
*
by Equations (1.84) and
(1.85), scalar C, vector u, and tensor E are transformed as

C C =

(1.86)

u Qu =

(1.87)

E QEQ
T
=

(1.88)
For physical laws to be objective, they have to be described by tensor quan-
tities. The differentiation of Equation (1.84), with respect to time, provides

t t t t v c Q v Q x ( ) ( ) ( ) ( )

= + +

t t t t t
T
c Q v Q Q x c ( ) ( ) ( ) ( )( ( ))

= + +

(1.89)
where the superimposed dot ( ) indicates time differentiation.
Fundamentals in continuum mechanics 17
If we set

T
A
QQ

=
(1.90)
A is the angular velocity tensor of the unstarred frame to the starred frame.
In the following, we will examine the deformation gradient (F x X
iJ i J
= / ),
velocity gradient (L v x
ij i j
L FF
1

= / , =

), rate of deformation (stretching)
tensor D, spin tensor W, and Cauchy stress tensor T.
If both the new and the old frames in the reference state are the same,
from dx Qdx QFdX = =

, we obtain

F QF =

(1.91)
Differentiating the preceding equation,

F QF QF

= +

(1.92)

F F F F Q QFF Q A
T T 1 1 1

= = +

(1.93)
Then, by setting L F F
1

=

, velocity gradient tensor L becomes

L QLQ A
T
= +

(1.94)
Consequently, L is not objective.
Since the stretching tensor (or the deformation rate tensor), D L L
T
( ),
1
2
= +
satisfes the transformation as

D QDQ
T
=

(1.95)
D is objective.
Moreover, the spin tensor, W
L L
T
( )
1
2
=

, transforms as

W QWQ A
T
= +

(1.96)
Hence, W is not objective. W is a continuum spin tensor and does not rep-
resent rigid body motion.
The Cauchy stress tensor follows the transformation as

T QTQ
T
=

(1.97)
18 Computational modeling of multiphase geomaterials
Taking a time derivative of Equation (1.97) gives

T
QTQ QTQ QT
Q
T T
T

= + +

(1.98)
Then, the time derivative of the stress tensor is not objective. Let us con-
sider the quantity as T WT TW

+ . Since

Q AQ

= (1.99)

QQ I
T
= (1.100)

T
Q
Q A
T

= (1.101)
we have

T W T T W Q T WT TW Q
T
[ ]

+ = +

(1.102)
Hence, T

is objective.

T T WT TW

= + (1.103)
where T

is called the Jaumann stress rate tensor.

The other stress rate, the Jaumann derivative of Kirchhoff stress ( JT),
o
T

,
is objective.

T T T L L

( ), ( )
o
ii
tr tr L = + = (1.104)
Differentiating Equation (1.82) with respect to time,

v
X
x
X
JT JT
k
j
ji
k
j
ji ki ki

= + (1.105)
Hence,

x
X
JT JT
v
X
J T
J
J
T
v
x
T
J T L T L T
k
j
ji ki ki
k
j
ji
ki ki
k
p
pi
ki pp ki kp pi

( )

= +

= +

= +
(1.106)
Fundamentals in continuum mechanics 19
Multiplying Equation (1.106) by

X
x
q
k
, we have

X
x
x
X
J
X
x
T L T L T
q
k
k
j
ji
q
k
ki pp ki kp pi

( )

+ (1.107)
Subsequently,

J
X
x
S
ji
j
k
ki

=

(1.108)

S T L T L T
ki ki pp ki kp pi

+ (1.109)
where S
ij

is the nominal stress rate with respect to the current confguration.

1.6 CONSERVATION OF MASS
The conservation of mass continuous medium V is expressed as the bal-
ance of mass of volume V holds if there is no mass infow into volume V and
no mass is produced in V. In this case, the balance of mass for an arbitrary
volume V is expressed by

D
Dt
dv
V
0

= (1.110)
where D/Dt denotes the material time derivative.
Equation (1.110) gives

=
D
Dt
v
x
dv
V
i
i
0
(1.111)
When the integrand is a continuous function, the local form for the bal-
ance of mass is

D
Dt
v
x
i
i
0

= (1.112)
In contrast, using Equation (1.29), the mass conservation law is expressed by

dV dV J = =
0 0 0
or (1.113)
20 Computational modeling of multiphase geomaterials
where is the mass density after the deformation and
0
is the density
before the deformation.
Equation (1.112) is an Eulerian description and Equation (1.113) is a
Lagrangian description of the mass conservation law.
1.7 BALANCE OF LINEAR MOMENTUM
The balance of linear momentum is given by the statement the change in
the linear momentum of the body occupying region R (= volume V + the
boundary S) is proportional to the force acting on the body. The balance
of momentum is expressed as

D
Dt
v dv t ds b dv
V
i
S
i
V
i

= + (1.114)
where D
Dt
is the material time derivative, v
i
is the velocity vector, t
i
is the
stress vector,
b
i
is the body force, and is the mass density.
The left-hand side of Equation (1.114) indicates the time change of the
linear momentum, and the frst and second terms on the right-hand side
express the surface force and the body force, respectively.
Using Cauchys theorem (Equation 1.57, t n
i ji j
= ) and the Gauss theo-
rem, and considering the balance of mass, we have

V
i i
ji
j
a b
x
dv

=

0
(1.115)
where a Dv Dt
i i
= / is the acceleration term.
If Equation (1.115) holds locally,

x
b a
ji
j
i i

+ = (1.116)
When disregarding the acceleration term, that is, for the quasi-static case,
Equation (1.116) is called the equilibrium equation.
Equation (1.116) can be expressed in component form for a two-
dimensional problem as x x x y x z ,
1 2 3
, , x,y components.

x y z
b
xx yx zx
x
0

+ = (1.117)
Fundamentals in continuum mechanics 21

x y z
b
xy yy zy
y
0

+ = (1.118)

x y z
b
xz yz zz
z
0

+ = (1.119)
were b b b
x y z
, , are components of the body force vector.
The balance of linear momentum in the reference confguration is given by

a dV dV b dV
i
V V
ji j i
V
0 0 , 0 0 0

= +
(1.120)
where a
i
is the acceleration vector and
ij
is the nominal stress tensor in
Equation (1.83).
Taking a time derivative of the frst term on the right-hand side of
Equation (1.120) and using Nansons theorem and J
0
= , we fnd

D
Dt
N dS JS
X
x
N dS
JS
X
x J
x
X
n ds
S n ds
ji
S
j ki
j
k
S
j
ki
j
k
S
p
j
p
ki
S
k
1
0 0

=
(1.121)
Hence, a rate type of balance of linear momentum, with respect to the
current confguration, is obtained as

a dv S n ds b dv
i ki k
S V
i
V

= + (1.122)
Under the static conditions with constant body force, the preceding equa-
tion becomes

S dv
ki k
V
0
,

=
(1.123)
The above rate equilibrium equation will be used for the updated Lagrangian
formulation of the boundary value problem.
22 Computational modeling of multiphase geomaterials
1.8 BALANCE OF ANGULAR MOMENTUM AND
THE SYMMETRY OF THE STRESS TENSOR
From the balance of angular momentum, the sum of the momentum is zero
in the case of a zero time rate for the angular momentum. Then,

s
ds dv dv
s
ds
v v
t n r b r a r M ( ) 0

+ +

=
(1.124)
where denotes the vector product, t n ( ) is the stress vector, b is the body
force vector, a is the inertia force vector, r is the position vector, and M is
the couple stress vector. Couple stress is called moment stress and cannot
be disregarded for materials with a signifcant rotation of the particles,
such as granular materials.
Equation (1.124) can be written in component form as

t x ds b x dv a x dv M ds
s
ijk k j
V
ijk k j
V
ijk k j
s
i
0

+ + = (1.125)
where
ijk
is the permutation symbol.
Using Cauchys theorem and the divergence theorem, we have

s
ijk k j
s
ijk j mk m
v
ijk j
mk
m
j
t x ds x n ds x
x

= =

kk
dv
(1.126)
Considering couple stress tensor
ij
, we obtain M n
i ji j
; = and Equation
(1.125) becomes

V
ijk j
mk
m
k k
V
ijk jk
x
x
b a dv

+ +

++

=
ji
j
x
dv 0
(1.127)
Upon substituting Equation (1.116), the frst term of Equation (1.127)
becomes zero.
Hence,

x
dv
V
ijk jk
ji
j
0

= (1.28)
Fundamentals in continuum mechanics 23
When the stress distribution is continuous, the local form for Equation
(1.128) is

x
ijk jk
ji
j
0 +

= (1.129)
when couple stress tensor
ji
is zero.
For i = 1,
123 23 132 32 23 32 23 32
0 + = = = , .
In general,

ij ji
= (1.130)
Consequently, the stress tensor is symmetric when couple stress tensor

ji
is zero.
1.9 BALANCE OF ENERGY
The energy conservation law is called the frst law of thermodynamics, and
it is described as follows: The time rate of total energy of the mass system
is equal to the sum of the external mechanical work rate done by the body
force and the surface force, heat infow through the surface of the body and
the other energy supply.

K E F Q

+ = + (1.131)

K
D
Dt
v v dv
V
i i
1
2

= (1.132)

F b v dv t v ds
V
i i
S
i i

= + (1.133)

E edv
V

= (1.134)

Q hdv q n ds
V S
i i

= (1.135)
where K

is the rate of the mechanical energy, E is the internal energy, e is

the internal energy density, F is the external work rate, Q is the heat infow
and other supplies of energy, h is the energy supply density such as radia-
tion, and q
i
is the heat fow vector.
24 Computational modeling of multiphase geomaterials
From the conservation of linear momentum, we have

F
D
Dt
v v dv D dv
V
i i
V
ij ij
1
2

= + (1.136)
Then,

E D dv hdv q n ds
V
ij ij
V S
i i

= + (1.137)
The local form for Equation (1.131) becomes

e D h q
ij ij i i
= +
,
(1.138)
1.10 ENTROPY PRODUCTION AND
CLAUSIUSDUHEM INEQUALITY
The second law of thermodynamics is described as follows: The time rate
of the entropy of the body is not less than the change in entropy associated
with the heat infow and the other supplies of energy.
In other words, the entropy production during the motion of a body is
not always negative.

N H

(1.139)

N
D
Dt
dv
V

= (1.140)
where is the entropy density.

H
h
dv
q
n ds
V S
i
i

(1.141)
where is the temperature.

D
Dt
h q
i
i

,
(1.142)
The local form for Equation (1.139), ClausiusDuhem Inequality, is
Fundamentals in continuum mechanics 25
If we set D Dt = / , using the frst law of thermodynamics, the preced-
ing equation becomes

D e q
x
ij ij i
i
1
0 +

(1.143)
If we set Helmholtzs free energy function as = e, Equation
(1.143) becomes

D q
x
ij ij i
i
( )
1
0

+ +

(1.144)
Let us consider the case with a small change in density and the follow-
ing relations:

ij
e

( ) = , (1.145)

D
ij ij
e
ij
vp
ij ij

=

+

, =

(1.146)
where ij
e
and
ij
vp
are the elastic strain rate and the inelastic strain rate,
respectively.
From Equation (1.144), we have

ij
e
ij
e
ij ij
1
qq
x
i
i

0
(1.147)

ij
ij
e
=

, =

(1.148)
Hence,

q
x
ij
ij
vp
i
i
1
0

(1.149)
Equation (1.149) indicates that the internal entropy production occurs due
to the inelastic strain and the heat fow.
26 Computational modeling of multiphase geomaterials
Truesdell and Noll (1965) defned internal entropy production as

D e q
x
ij ij i
i
1 1 1
0
2
= +

(1.150)
The strong suffcient conditions for Equation (1.150) to be true are given
by the following two inequalities:

D e
ij ij
1 1
0 = +

(1.151)

q
x
i
i
1
0
2

(1.152)
When Fouriers law of heat fow is expressed as

q
i i
, = :

heat conduction coeffcient (1.153)
Equation (1.152) becomes

x
i
1
0
2
2

(1.154)
Consequently,
0 (1.155)
Namely, is nonnegative.
An important thermodynamical framework for the plasticity theory
has been studied by Collins and Houlsby (1997) based on the Zieglers
theory of dissipation function. The related results will be presented in
Chapter 3.
1.11 CONSTITUTIVE EQUATION AND OBJECTIVITY
As was been mentioned earlier, there are nine fundamental laws in con-
tinuum mechanics, except for electromagnetic laws. These include the mass
conservation law (1), the conservation laws of linear momentum (3), the
conservation laws of angular momentum (3), the conservation of energy (1),
and the entropy production inequality (1, constraint condition).
Fundamentals in continuum mechanics 27
On the other hand, there are nineteen variables contained in the laws,
namely, the mass density (1), velocity components (3), the components of
the stress tensor (9), temperature (1), the components of the heat fow vector
(3), internal energy (1), and entropy (1).
Hence, eleven more equations are required to describe the response of
materials. These eleven equations are called constitutive equations in order
to specify the response characteristics of materials. The number of equa-
tions is eleven, that is, six for stressstrain relations, three for heat fux, one
for internal energy, and one for entropy.
Constitutive equations are not given a priori but are derived based
on experiments satisfying the fundamental laws and objectivity. The
well-known Hookes law is a typical constitutive equation for elastic
materials.
1.11.1 Principle of objectivity and constitutive model
The response of a material to external action is independent of the observer.
It indicates that the constitutive equation should be indifferent to changes
in the coordinate frame. In the following, we will discuss the objectivity of
the constitutive equations and how the principle of objectivity prescribes
constitutive equations.
Constitutive equations describe the materials inherent response to exter-
nal action and are expressed by a functional of the history of motion and
deformation. This functional is called the constitutive functional.
For example, when the stress tensor T at a material point X is determined
by the motion of material point X in the vicinity of X, constitutive func-
tional G is given by

t t T G x X X X [ ( ) ( ) ] = , , , , (1.156)
Constitutive tensor functional G has to be indifferent with respect to the
rigid rotation, the translational motion, and the time shift so that constitu-
tive functional G satisfes the principle of objectivity as

t t t t x X Q x X c ( ) ( ) ( ) ( ) , = , +

(1.157)

t t t
0
=

(1.158)

t t t t t t t G x X X X G x X X X [ ( ) ( ) ] [ ( ) ( ) ]
0
, , , , , = , , , , ,

(1.159)
where X is a point in the vicinity of point X.
28 Computational modeling of multiphase geomaterials
In the above, we assume that functional G depends on the motion of the
material points in the vicinity of point X and, in general, the constitutive
functional depends on the past history ( t t < ). Herein, however, we
disregard it for the sake of simplicity.
1.11.2 Time shift
When we set t I Q( ) = , t c( ) 0 = , t t
0
= , and = , =

t t t x X x X ( , ) ( , ) 0.
Then, for example, the stress tensor T can be expressed by functional G as

T G x X X X [ ( 0) ( 0) 0] = , , , , , (1.160)
This indicates that the functional does not explicitly depend on time. In
other words, the response functional, which explicitly depends on time, is
not objective.
1.11.3 Translational motion
When we set t I Q( ) = , t t c x X ( ) ( ) = , , and t 0
0
= ,

t t t t t x X x X x X ( ) ( ) ( ) , = , , , =

(1.161)

t t t t T G x X x X X X [ ( ) ( ) ( ) ] = , , , , , , (1.162)
1.11.4 Rotational motion
When t Q( ) is set to be arbitrary and t c( ) 0 = and t 0
0
= ,

t t t t t t t
T
Q T X Q G Q x X X X T ( ) ( ) ( ) [ ( ) ( ) ( ) ] , = , , , , , =

(1.163)
From Equation (1.162), we get

t t t t t t t t
T
Q G x X x X X X Q G Q x X x X X ( ) [ ( ) ( ) ( ) ] ( ) [ ( )( ( ) ) ( )) ] , , , , , = , , ,,

(1.164)
Taking the Taylor series of t x X ( ) , around t x X ( ) , , assuming the continu-
ity of the functional, we have

t t t d x X x X F X X ( ) ( ) ( ) , , = , (1.165)
Fundamentals in continuum mechanics 29
Hence,

t t t t
T
Q TQ G QF X X X ( ) ( ) [ ( ) ( ) ] = , , , , (1.166)
The dependence of the functional on the relative position leads to the
dependence on the deformation gradient, F
i j
x
X
i
j
=
,

, and the dependence on

the strain.
For example, consider the case in which the deformation rate tensor, D,
depends on the stress, namely,

D G T ( ) = (1.167)
The objective functional G satisfes

QDQ G QTQ
T T
( ) = (1.168)
for a rotation Q.
From a physical point of view, it can be seen that the deformation rate
tensor rotates according to the rotation of the loading system. Hence, the
principle of objectivity can be viewed the principle of the space isotropy
(Figure1.4).
REFERENCES
Belytschko, T., Liu, W.K., and Moran, B. 2000. Nonlinear Finite Elements for
Continua and Structures, John Wiley & Sons, New York.
Boehler, J.P. 1987. Applications of Tensor Functions in Solid Mechanics, CISM
Courses and Lectures, No. 292, Springer-Verlag, New York.
y T
T
D
ij
D
ij
x
x
O

y
Figure 1.4 Change of reference frame.
30 Computational modeling of multiphase geomaterials
Collins, I.F., and Houlsby, G.T. 1997. Application of thermomechanical prin-
ciples to the modeling of geotechnical materials, Proc. Roy. Soc. London A,
453:19752001.
Eringen, A.C. 1967. Mechanics of Continua, New York: John Wiley & Sons.
Fung, Y.C., and Tong, P. 2001. Classical and Computational Solid Mechanics,
World Scientifc.
Gurtin, M.E. 1982. An Introduction to Continuum Mechanics, New York:
Academic Press.
Malvern, L.E. 1969. Introduction to the Mechanics of a Continuous Media, New
York: Prentice-Hall.
Maugin, G.A. 1992. The Thermomechanics of Plasticity and Fracture, Cambridge
University Press.
Spencer, A.J.M. 1988. Continuum Mechanics, Longman Scientifc and Technical,
New York.
Truesdell, C., and Noll, W. 1965. The non-linear feld theories of mechanics, In
Encyclopedia of Physics, Vol. III, Part 3, ed. by S. Flgge, Berlin: Springer-Verlag.
Ziegler, H. 1983. An Introduction to Thermomechanics, 2nd ed., Elsevier Science,
North-Holland, Amsterdam.
31
Chapter 2
Governing equations for
multiphase geomaterials
One of the important characteristics of geomaterials is that the material
is composed of solid, liquid, and gas in general. In this chapter, gov-
erning equations for the analysis including balance laws and constitu-
tive equations are presented based on the theory of porous media, that
is, an immiscible mixture of solid and fuids. First, governing equations
for fuidsolid materials are shown, then in Section 2.2, the governing
equations for gasliquidsolid three-phase materials are formulated.
In Section 2.3 the equations for the unsaturated saturated soils are
presented.
2.1 GOVERNING EQUATIONS FOR FLUID
SOLID TWO-PHASE MATERIALS
2.1.1 Introduction
The governing equations for pore watersoil coupled problems can be
derived from Biots theory of water saturated porous media, which is
based on continuum mechanics (Biot 1941, 1955, 1956, 1962; Atkin
and Craine 1976; Bowen 1976). Before Biots work, Fillunger (1913)
proposed a theory of porous media flled with water. Historical devel-
opment of the theory of porous media has been well documented by
de Boer (2000a,b). Various methods are proposed for Biots two-phase
mixture theory depending on the method of approximation and the
choice of unknown variables (Coussy 1995; Lewis and Schrefer 1998;
Zienkiewicz et al. 1999; Ehlers, Graf, and Ammann 2004). In many
computer programs for liquefaction and consolidation analyses, a u-p
formulation is adopted in which the displacement (u) of the solid and
pore water pressure (p) are used as the unknown variables, because we
can reduce the degree of freedom although the u-p formulation provides a
different solution in the high frequency range for the higher permeability
32 Computational modeling of multiphase geomaterials
(Zienkiewicz et al., 1980, LIQCA Res. Development Group, 2005). This
u-p formulation can be easily applied to the consolidation problems
since the displacement of pore water is explicitly introduced.
2.1.2 General setting
The following assumptions are adopted in the u-p formulation:
1. An infnitesimal strain is used.
2. The relative acceleration of the fuid phase to that of the solid phase is
much smaller than the acceleration of the solid phase.
3. The grain particles in the soil are incompressible.
4. The effect of the temperature is disregarded.
The motion of the mixture for the multiphase medium is described by
the superposition of multiphases in the context of continuum theory as
shown in Figure2.1. For the fuidsolid two-phase mixture, we assume
that each point within the mixture is occupied simultaneously by two
constituents and described by the rectangular Cartesian coordinates
(Figure2.2).
In the followings, the material time derivative is given by

D
Dt t
v
x
a
i
a
i
a
( )
=

(2.1)
where superscript a indicates phase a and v
i
a
is the velocity of the material
in phase a, x
i
a
is the position of particle of a phase.
Water
Soil particle
Solid uid mixture
Solid phase Fluid phase
Figure 2.1 Superposition of solid and fuid phases.
Governing equations for multiphase geomaterials 33
At current state at time t,
x x
i
s
i
f ( ) ( )
=
for two phases; (s) stands for solid
and (f) for fuid phases in Figure2.2. Hence, material time derivative for
the multiphases can be described in the spatial coordinate x
i
and the super-
scripts can be neglected as:

D
Dt t
v
x
i
a
i
=

(2.2)
2.1.3 Density of mixture
The densities of the solid phase,
s
, and the fuid phase,
f
, are defned as

s
s
f
f
n n = = ( ) , 1
(2.3)
where n is the porosity,
s
is the density of the solid, and
f
is the density
of the fuid.
The density of the mixture is described using Equations (2.1) and (2.2) as

= + = +
s f
s f
n n ( ) 1 (2.4)
In Biots theory, the water-saturated soil is described by the superposition
of the solid phase and the fuid phase as shown in Figure2.1.
x
3
x
2
x
1
at time t
at current time t
x
i
X
i
( f )
x
i
(s)
X
i
(s)
( f )
= x
i
= x
i
Figure 2.2 Geometric arrangement.
34 Computational modeling of multiphase geomaterials
2.1.4 Definition of the effective and partial
stresses of the fluidsolid mixture theory
The total stress is given by the sum of the partial stresses acting on the
phases as


ij ij
s
ij
f
= + (2.5)
where
ij
s
is the partial stress tensor of the solid phase and
ij
f
is the partial
stress tensor of the fuid phase.
The partial stresses for the fuid phase and the solid phase are given by


ij
f
ij
np = (2.6)

n p
ij
s
ij ij
(1 ) =

(2.7)
where
ij
is the total stress tensor,
ij
is the effective stress tensor, p is the
pore water pressure, n is the porosity, and
ij
is the Kroneckers delta. In
the derivation, tension is positive but the pore water pressure is positive
for compression.
The total stress is described by the effective stress and the pore water
pressure as

= p
ij ij ij
(2.8)
2.1.5 Displacementstrain relation
From Assumption 1, the displacementstrain relations for the solid and the
fuid phases are defned as


ij
s i
s
j
j
s
i
ij
f i
f
j
j
u
x
u
x
u
x
u
=

+

1
2
1
2
,
ff
i
x
(2.9)
where
ij
s
is the strain tensor of the solid phase and u
i
s
is the displacement
vector of the solid phase,
ij
f
is the strain tensor of the fuid phase, and u
i
f

is the displacement vector of the fuid phase.
The strain rates are given by the time differentiation of strains as

ij
s i
s
j
j
s
i
u
x
u
x

1
2
,

ij
f i
f
j
j
f
i
u
x
u
x

1
2
(2.10)
where () denotes the time differentiation.
Governing equations for multiphase geomaterials 35
2.1.6 Constitutive model
The constitutive relations of the solid phase are given by the relations
between the incremental strains and effective stress increments as

= D
ij ijkl kl
s
(2.11)
where
ij
is the effective stress increment tensor, D
ijkl
is the modulus ten-
sor, and
kl
s
is the strain increment tensor of the solid phase.
In the case of the elastoplastic model, it becomes

D D
ijkl ijkl
ep
= (2.12)
When disregarding the viscous resistance of fuid phase, the constitutive
equation is given by

p K
f
ii
f
=
(2.13)
where K
f
is the elastic volumetric modulus of the pore fuid.
2.1.7 Conservation of mass
The mass conservation laws for the solid and the fuid phases are given by

=
s s
i
s
i
t
u
x
( )

0
(2.14)

=
f f
i
f
i
t
u
x
( )

0 (2.15)
2.1.8 Balance of linear momentum
The linear momentum conservation laws for the two phases are given as

s
i
s
i
ij
s
j
s
i u
R
x
b

+ (2.16)

f
i
f
i
ij
f
j
f
i u
R
x
b

+ =

+ (2.17)
36 Computational modeling of multiphase geomaterials
where b
i
is the body force vector, and R
i
is the term expressing the energy
dissipation due to the relative motion between the solid and the fuid phases
(Biot 1956).

R n
k
w i
w
i
=

(2.18)

i i
f
i
s
w
n
u u
= ( ) (2.19)
where k is the permeability coeffcient(assumed to be scalar because of isot-
ropy), and
i w is the relative velocity vector of the fuid phase to the solid
phase.
w
f
g = is the unit weight of the pore water with the gravitational
acceleration g.
When we describe R
i
by Equation (2.18), it is easily shown that the equa-
tion of motion for the fuid phase is a general description of Darcys law.
Considering Equation (2.19) and that the porosity is constant, we obtain
the following equation from the equation of motion for the fuid phase
Equation (2.17), after manipulation:

f
i
s
f
i i
ij
f
j
f
i u
n
w R
x
b



+ + =

+ (2.20)
If the relative acceleration is almost zero,

u w
i
s
i
(2.21)
Equation (2.20) can be approximated considering Equation (2.21):

f
i
s
i
ij
f
j
f
i u
R
x
b

+ =

+ (2.22)
Substituting Equations (2.2), (2.5), and (2.18) into Equation (2.22),
we obtain

n
u
n
k
w
np
x
n b
f
i
s w
i
i
f
i

+ =

+ (2.23)
When we assume that the spatial gradient of the porosity is suffciently
small, the following equation holds:

=
n
x
i
0 (2.24)
Governing equations for multiphase geomaterials 37
Substituting Equation (2.24) into Equation (2.23), we have

f
i
s w
i
i
f
i u
k
w
p
x
b

+ =

+ (2.25)
Then, when the second term on the right-hand side of Equation (2.25) is
a body force due to the gravitational force, and we disregard the dynamic
term, the following equation holds:

+ =
p
x
b
i
f
i
0 (2.26)
Hence,
p gx
f
=
1
in which x
1
is a coordinate in the direction of the gravita-
tional force and g is the gravitational acceleration; then p is then called the
hydrostatic pressure.
From Equation (2.25), we have the following equation:

i
w
f
i
s
i
f
i w
k
u
p
x
b

= +

(2.27)
Then, disregarding the acceleration term and setting the direction of x
1

for the direction of gravitational force g and b g
1
= , we have

1
1 1
1
w
k p
x
g k
x
p
x
w
f
w
=

k
h
x
1
(2.28)
in which
w
f
g = and h is the total head.
The total head is expressed by

h
p
x
p
z
w w
= = +

1
(2.29)
where z x =
1
, z is the elevation head, and
p
w

is the pressure head.

It is seen that Equation (2.28) is Darcys law. It should be noted that
the conservation law of the linear momentum of the fuid phase Equation
(2.17) is a general description of Darcys law. In the preceding, the funda-
mental equations of governing equations have been described. Next, we
will derive the balance equation of the whole mixture and the continuity
equation from the fundamental equations.
38 Computational modeling of multiphase geomaterials
2.1.9 Balance equations for the mixture
By adding Equation (2.16) and Equation (2.17), we obtain the balance of
linear momentum of the mixture as

s
i
s
f
i
f
ij
s
j
ij
f
j
s
i
f
u u
x x
b


+ =

+ + bb
i
(2.30)
Upon substitution of Equations (2.3), (2.4), and (2.19) into Equation
(2.30), the following equation is derived:

i
s f
i
ij
j
i u w
x
b

+ =

+ (2.31)
From Assumption 2, namely, in the case that the relative acceleration can
be neglected as Equation (2.21), Equation (2.31) becomes the balance of
linear momentum for the mixture as

i
s
ij
j
i u
x
b

+ (2.32)
For the balance of angular momentum of the multiphase materials, we
assume the balance of angular momentum for the phases as well as for the
whole mixture.
2.1.10 Continuity equation
Substitution of Equation (2.1) into the mass conservation equation of the
solid phase Equation (2.14) leads to the following equation:

( )
( ) {( ) }
( ) 1
1 1
1

+ n
t
n
t
n
u
x
n
s
s s i
s
i
i
s

u
x
s
i

0 (2.33)
In a similar way, substituting Equation (2.2) into the mass balance equa-
tion of the fuid phase Equation (2.15) gives

n
t
n
t
n
u
x
n
u
x
f
f f i
f
i
i
f
f
i

( )

0 (2.34)
Governing equations for multiphase geomaterials 39
Multiplying Equation (2.33) by
f s
/ and adding the result and Equation
(2.34), we have


f f i
f
i
s
i
f i
n
t
n
t
n
u u
x

+
( ) { ( )} 1

ss
i
f
i
f
f
i
f
s
s
u
x
n
t
u
x
n
t

( ) 1
ii
s
s
i
u
x

0
(2.35)
In Equation (2.35), the frst term is equal to zero. Taking into account
Assumption 1 and substituting Equation (2.10) and Equation (2.19) into
Equation (2.35), we have

+ +

+

i
i
ii
s
f
f
i
f
f
i
s
w
x
n
t
u
x
n

( ) 1

=

s
i
s
s
i
t
u
x

0 (2.36)
where ii
s

is the volumetric strain rate of the solid phase.

Considering the material time derivative, Equation (2.36) can be
expressed as

+ + +

=
i
i
ii
s
f
f
s
s
w
x
n n

( ) 1
0 (2.37)
Herein, when we assume the incompressibility of the soil constituents
such as soil particles (Assumption 3), the following equation holds:

s

= 0 (2.38)
Substitution of Equation (2.38) into Equation (2.37) leads to the follow-
ing equation:

+ + =
i
i
ii
s
f
f
w
x
n

0
(2.39)
The preceding equation is the continuity equation for the case of incom-
pressibility of the solid constituent. The third term of the left-hand side of
Equation (2.39) denotes a compressibility of the pore water.
40 Computational modeling of multiphase geomaterials
Neglecting the effect of temperature, the following equation is given
since the time derivative of mass of the pore fuid (
f f
V ) is zero:

D V
Dt
f f
( )
= 0 (2.40)
After the manipulation of this equation we have

f
f
f
f
ii
f V
V

= =
(2.41)
Substituting the constitutive equation of the fuid, Equation (2.13), into
Equation (2.41) gives

f
f f
p
K

= (2.42)
Upon substitution of Equation (2.42) into Equation (2.39), the continuity
equation becomes

+ + =
i
i
ii
s
f
w
x
n
K
p

0 (2.43)
In Equation (2.43), this equation includes the relative velocity of the fuid
phase to the solid phase.
If Assumption 2 of the u-p formulation is adopted, we can express w
i
by
the acceleration of the solid phase and the pore water pressure as has been
shown in Equation (2.27) and can be eliminated in Equation (2.43).
Substituting Equation (2.27) into Equation (2.43), we get

x
k
u
p
x
b
i w
f
i
s
i
f
i

+ + =
ii
s
f
n
K
p

0 (2.44)
If the body force b
i
is constant, and the spatial gradients of permeability
and the density of the fuid are suffciently small, considering Equation (2.8),
the fnal form of the continuity is given as

k p
x
n
K
p
w
f
ii
s
i
ii
s
f

+ + =

2
2
0 (2.45)
Governing equations for multiphase geomaterials 41
2.2 GOVERNING EQUATIONS FOR GASWATER
SOLID THREE-PHASE MATERIALS
2.2.1 Introduction
Geomaterials generally fall into the category of multiphase materials.
They are basically composed of soil particles, water, and air. The behavior
of multiphase materials can be described within the framework of a mac-
roscopic continuum mechanical approach through the use of the theory
of porous media (de Boer 2000b). The theory is considered to be a gener-
alization of Biots two-phase porous theory for saturated soil (Biot 1941,
1955, 1956).
Proceeding from the general geometrically nonlinear formulation, the
governing balance relations for multiphase materials can be obtained (de
Boer 2000b; Loret and Khalili 2000; Lewis and Schrefer 1998; Ehlers and
Graf, 2003; Ehlers et al., 2004). Mass conservation laws for the gas phase
as well as for the liquid phase are considered in those analyses. In the feld
of geotechnics, air pressure is assumed to be zero in many research works
(Sheng et al. 2003), since geomaterials usually exist in an unsaturated state
near the surface of the ground and we have not enough data on a develop-
ment of air pressure. Considering gas hydrate dissociation in the sea bed
ground, however, we have to deal with the high level of gas pressure that
exists deep in the ground (Kimoto et al. 2007), this means that the mass
balance for three phases must be considered. Oka et al. (2006) proposed
an airwatersoil coupled fnite element model in which the skeleton stress
is used as a stress variable, and the suction effect is introduced in the
constitutive equation for soil. Furthermore, the conservation of energy is
required when there is a considerable change in temperature during the
deformation process. Vardoulakis (2002) showed that the temperature
of saturated clay rises with plastic deformation. Oka et al. (2004) and
Kimoto et al. (2007) numerically simulated the thermal consolidation pro-
cess, which will be shown in Chapter 10.
2.2.2 General setting
The material to be modeled is composed of three phases, namely, solid
(S), water (W), and gas (G), which are continuously distributed throughout
space. Total volume (V) is obtained from the sum of the partial volumes of
the constituents, namely,

V V S W G

= = ( , , )
(2.46)
42 Computational modeling of multiphase geomaterials
The volume of void, V
v
, which is composed of water and gas, is given as
follows:

V V W G
v

= = ( , )
(2.47)
Volume fraction, n

, is defned as the local ratio of the volume element

with respect to the total volume, namely,

n
V
V

= (2.48)

n S W G

= = 1 ( , , )
(2.49)
The volume fraction of the void, that is, porosity, n, is written as

n n
V
V
V V
V
n W G
v S
S
= = =

= =

1 ( , ) (2.50)
The volume fraction of the fuid, n
F
, is given by

n n W G
F
= =

( , )
(2.51)
The volume fraction concept has been adopted to construct the theory
of mixture (Mills 1967; Morland 1972). The historical development of the
volume fraction theory has been well discussed by de Boer (2000b).
In addition, the water saturation is required in the model, namely,

s
V
V V
n
n n
n
n
r
W
W G
W
W G
W
F
=
+
=
+
=
(2.52)
2.2.3 Partial stresses
By analogy to the water-saturated soil, we assume that

= n P
ij
S
ij
S F
ij (2.53)


ij
W W W
ij
n P =
(2.54)


ij
G G G
ij
n P = (2.55)
Governing equations for multiphase geomaterials 43
where
P
F
is the average pressure of the fuids surrounding the solid skeleton
(Bolzon et al. 1996) given by

P s P s P
F
r
W
r
G
= +
( )
1 (2.56)
and
ij
is a skeleton stress.
The skeleton stress, which will be explained in the following, is reason-
able to describe the behavior of solid skeleton in the constitutive relation.
Total stress tensor,
ij
, is obtained from the sum of the partial stresses,

ij
, namely,

= =

S W G
ij ij
( , , )
(2.57)
and

P
ij ij
F
ij
= + (2.58)
2.2.4 Conservation of mass
The conservation of mass for the solid, water, and gas phases, ( = S,W,G),
is given in the following equation:

( )
= + =
( )
t
n q m S W G
Mi i


,
, ,
(2.59)
in which

is the material density, q

Mi

is the fux vector, and m

is the mass
change rate of phase per unit volume. The fux vector is expressed in terms
of the velocity of the fow as

q n v S W G
Mi i

= = ( , , ) (2.60)
where v
i

is the velocity of phase .

The relative velocity of the fow, V
i

, with respect to the solid phase is

V n v v W G
i i i
S
= = ( ) ( , ) (2.61)
44 Computational modeling of multiphase geomaterials
The conservation laws in Equation (2.59) for the solid, water, and gas
phases are expressed with water saturation, s
r
, and the volume fraction of
void, n
F
, as

+ =
s F F s
i i
s s
n n v m ( )
,
1 (2.62)

n s s n n s ns v m
W F
r
W
r
F F
r
W
r
W
i i
W W
,
+ + + =
(2.63)


n s n s n s n s v
F
r
G F
r
G F
r
G F
r
G
i i
G
( ) ( ) ( )
,
1 1 1 + + == m
G
(2.64)
where we assume the incompressibility of soil particles,

S
= 0, and n n
F
=
is the porosity.
Assuming that the spatial gradient of the volume fractions are zero, we
obtain following relations from Equations (2.62) to (2.64) as

s n s n V s v s
m m
r
F
W
W
r
F
i i
W
r i i
S
r
s
s
W
W

+ + + =
, ,
0 (2.65)

s n s n s v V s
m m
r
F
G
G
r
F
r i i
S
i i
G
r
s
s
G
G
(1 ) (1 ) (1 ) 0
, ,

+ +

=
(2.66)
As for describing changes in the gas density, the equation of ideal gases
can be used, that is,

G
G G
M P
R
= (2.67)

G
G G G
M
R
P P
=
2
(2.68)
in which M
G
is the molecular weight of gas, R is the gas constant, is the
temperature, and tension is positive in the equation.
Dividing Equation (2.68) by Equation (2.67) yields

G
G
G
G
P
P
= (2.69)
Governing equations for multiphase geomaterials 45
2.2.5 Balance of momentum
Momentum balance is required for each phase, namely,

n v n F P S W G
i ji j i i



= + =
,
( , , ) (2.70)
in which F
i
is the gravity force and

P
i

is related to the interaction term

given in

P D v v D D S W G
i i i

= = =

( ), ( , , , )
(2.71)
where D

are parameters that describe the interaction with each phase.

The momentum balance equation for each phase is obtained with the fol-
lowing equations when the acceleration is disregarded:

n P n F D v v D v v
ji j
S F
i
S S
i
SW
i
S
i
W SG
i
S
i
G
' ( ) ( ) ( ) 0
, ,
+ =
(2.72)

n P n F D v v D v v
W W
i
W W
i
WS
i
W
i
S WG
i
W
i
G
( ) ( ) ( ) 0
,
+ =
(2.73)

n P n F D v v D v v
G G
i
G G
i
GS
i
G
i
S GW
i
G
i
W
( ) ( ) ( ) 0
,
+ =
(2.74)
D W G

( , , ) = are given as

D
n g
k
D
n g
k
WS
W W
W
GS
G G
G
= =
( )
,
( )
2 2

(2.75)
in which k
W

and k
G
are the permeability coeffcients for the water phase
and the gas phase, respectively. We assume that the interaction between
water and gas phases D
GW
and D
WG
is zero.
When the space derivative of volume fraction n
i ,

is negligible, Darcys
law for the water phase and the gas phase is obtained from Equations (2.73)
and (2.74), respectively, as

V n v v
k
g
P F
i
W W
i
W
i
S
W
W
i
W W
i , ( ) ( )
= =

(2.76)

V n v v
k
g
P F
i
G G
i
G
i
S
G
G
i
G G
i , ( ) ( )
= =

(2.77)
46 Computational modeling of multiphase geomaterials
The sum of Equations (2.72) to (2.74) leads to

ji j
E
i
E
F n S W G
,
, ( , , ) + = = =

0 (2.78)
It is worth noting that Darcy-type laws such as Equations (2.76) and
(2.77) are not objective since they include velocity but are a good approx-
imation (Eringen 2003). In addition, as has been printed out in Section
2.2.1.9, it is assumed that the angular momentum is balanced for the
phases for the whole mixture.
2.2.6 Balance of energy
The following energy conservation equation is applied in order to consider
the heat conductivity:

c D h Q
E
ij
vp
ij i i ,

( )
= +
(2.79)

c n c S W G
E
( )
= =

( , , )
(2.80)
where c

is the specifc heat, is the temperature for all the phases, D

ij
vp
is
the viscoplastic stretching tensor, h
i
is the heat fux vector, and

Q is the
heat source.
Heat fux, h
i
, is given by

h
i
E
i
=
,
(2.81)

E
n =

( = S,W,G) (2.82)
in which

is the thermal conductivity.

2.3 GOVERNING EQUATIONS FOR
UNSATURATED SOIL
In the theory of porous media, the concept of the effective stress tensor
is related to the deformation of the soil skeleton and plays an important
role. The effective stress tensor has been defned by Terzaghi (1943) for
Governing equations for multiphase geomaterials 47
water-saturated soil. However, the effective stress needs to be redefned
if the fuid is made of compressible materials. In the present study, the
skeleton stress tensor,
ij
, is defned and then used for the stress variable
in the constitutive relation for the soil skeleton (Oka et al. 2006, 2008,
2010), which has been called average skeleton stress by Jommi (2000)
and Gallipoli et al. (2003).
2.3.1 Partial stresses for the mixture
The total stress tensor is assumed to be composed of three partial stress
values for each phase:


ij ij
s
ij
f
ij
a
= + + (2.83)
where
ij
is the total stress tensor, and
ij
s
,
ij
f
, and
ij
a
are the partial
stress tensors for solid, liquid, and air, respectively.
Considering the volume fraction (Ehlers et al. 2004), the partial stress
tensors for unsaturated soil can be given by


ij
f
r
f
ij
nS p = (2.84)


ij
a
r
a
ij
n S p = ( ) 1 (2.85)

= n P
ij
s
ij
F
ij
(1 )
(2.86)

P S p S p
F
r
f
r
a
= + ( ) 1
(2.87)
where p
f
and p
a
are the pore water pressure and the pore air pressure,
respectively, n is the porosity, S
r
is the degree of saturation,
ij
is the skel-
eton stress, and P
F
is the average pore pressure. The tension is positive in
this chapter.
For the unsaturated soil, we will use the skeleton stress (Oka et al. 2006,
2008; Kimoto et al. 2010) as the basic stress variable in the model along with
suction. The skeleton stress only applies to the soil skeleton and Equation
(2.86) comes from the analogy to the effective stress for the water-saturated
soil. By adding Equations (2.84), (2.85), and (2.86) we have skeleton stress as

P
ij ij
F
ij
= +
(2.88)
48 Computational modeling of multiphase geomaterials
The skeleton stress was frst advocated by Jommi (2000) as the average
skeleton stress, which was defned as the difference between the total stress
and the average fuid pressure. Ehlers et al. (2004) called it effective stress.
However, herein we called it skeleton stress to avoid confusing it with mean
value of the skeleton stress.
Adopting the skeleton stress provides a natural application of the mix-
ture theory to unsaturated soil. The defnition in Equation (2.88) is similar
to Bishops defnition for the effective stress of unsaturated soil. In addition
to Equation (2.88), the effect of suction on the constitutive model should
always be taken into account. This assumption leads to a reasonable con-
sideration of the collapse behavior of unsaturated soil, which has been
known as a behavior that cannot be described by Bishops defnition for
the effective stress of unsaturated soil. Introducing suction into the model,
however, makes it possible to formulate a model for unsaturated soil, start-
ing from a model for saturated soil, by using the skeleton stress instead of
the effective stress.
2.3.2 Conservation of mass
The mass conservation law for the three phases is given by

=
J J
i
J
i
t
u
x
( )

0 (2.89)
where
J
is the average density for the J phase and u
i
J
is the velocity vector
for the J phase.

s
s
n = ( ) 1 (2.90)

f
r
f
nS = (2.91)

a
r
a
n S = ( ) 1 (2.92)
where J = s, f, and a, in which the superscripts s, f, and a indicate the
solid, the liquid, and the air phases, respectively; n is the porosity; and
S
r
is the saturation.
J
is the mass bulk density of the solid, the liquid,
and the gas.
Governing equations for multiphase geomaterials 49
2.3.3 Balance of linear momentum
for the three phases
The conservation laws of linear momentum for the three phases are
given by

i
s
s
i i
ij
s
j
s
i
u
Q R
x
b

+
(2.93)

f
i
f
i
ij
f
j
f
i u
R
x
b

+ =

+ (2.94)

a
i
a
i
ij
a
j
a
i u
Q
x
b

+ =

+ (2.95)
where u J a f s
i
J
( , , ) = are the acceleration vectors for the three phases, b
i
is
the body force, Q
i
denotes the interaction between the solid and the air
phases, and R
i
denotes the interaction between the solid and the liquid
phases. These interaction terms, Q
i
and R
i
, can be described as

R nS
k
w i r
w
f
i
f
=

(2.96)

Q n S
g
k
w i r
a
a
i
a
= ( ) 1

(2.97)
where k
f
is the water permeability coeffcient, k
a
is the air permeability, i
f
w
is the average relative velocity vector of water with respect to the solid skel-
eton, and
i
a
w is the average relative velocity vector of air to the solid skeleton.
The relative velocity vectors are defned by

i
f
r i
f
i
s
w
nS
u u
=
( )
(2.98)

i
a
r i
a
i
s
w
n S
u u
=
( )
( ) 1
(2.99)
50 Computational modeling of multiphase geomaterials
Using Equation (2.98), Equation (2.94) becomes

f
i
s
r
i
f
i
ij
f
j
f
i u
nS
w
R
x
b

+ + =

+
1
(2.100)
We deal with the behavior of soil in which the difference between
accelerations of the soil skeleton and pore fuid is suffciently small. This
assumption is reasonable except for the high frequency problem and very
high permeability (Zienkiewicz et al. 1980). For this reason we assume that
w
i
f
0, in this case, using Equations (2.84), (2.91), and (2.96), Equation
(2.100) becomes

nS
u
nS
k
w
nS
p
x
nS b
r
f
i
s
r
w
f
i
f
r
f
i
r
f
i

+ =

+
(2.101)
in which we assume that the spatial gradients of porosity and saturation
are suffciently small. The same assumption will be taken in the following
derivations of the governing equations.
After manipulation, the average relative velocity vector of water to the
solid skeleton and the average relative velocity vector of air to the solid
skeleton are shown as

i
f
f
w
f
i
f
i
s f
i w
k p
x
u
b

=

(2.102)

i
a
a
a
a
i
a
i
s a
i w
k
g
p
x
u
b

=

(2.103)
in which w
i
a
0 is assumed due to the reason mentioned earlier.
Based on the aforementioned fundamental conservation laws, we can
derive equations of motion for the whole mixture. Substituting Equations
(2.90), (2.91), and (2.92) into the given equation and adding Equations
(2.93) to (2.95), we have


( ) ( )
+ + =

+
u
nS
u u
n S
u u
x
b
i
s
r
f
i
f
i
s
r
a
i
a
i
s
ij
j
i
(1 )
(2.104)
where is the mass density of the mixture as = + +
f a s
, and
ij
is the
total stress tensor.
Governing equations for multiphase geomaterials 51
From the following assumptions,

i
s
i
f
i
s
u u u
>>
( )
(2.105)

i
s
i
a
i
s
u u u
>>
( )
(2.106)
equations of motion for the whole mixture are derived as

i
s
ij
j
i u
x
b

+
(2.107)
2.3.4 Continuity equations
Using the mass conservation law for the solid and the liquid phases,
Equation (2.89) (J = s,f) and Equations (2.90) and (2.91), and assuming the
incompressibility of soil particles, we obtain


{ }

+ + + =
nS
u u
x
S nS nS
r i
f
i
s
i
r ii
s
r
f
f
r
( )

0 (2.108)
Incorporating Equation (2.102) and p K
f
ii
f
= (K
f
: volumetric elastic
coeffcient) into the previous equation leads to the following continuity
equation for the liquid phase:

x
k
u
p
x
b
i
f
w
f
i
s
f
i
f
i

+ + + = S nS nS
p
K
r ii
s
r r
f
f

0 (2.109)
Similarly, we can derive the continuity equation for the air phase by
assuming that the spatial gradients of porosity and saturation are suff-
ciently small:

x
k
u
p
x
b
i w
a
i
s
a
i
a
i
a

+ + = ( ) ( ) 1 1 0 S nS n S
r ii
s
r r
a
a

(2.110)
For the saturation we will use a constitutive equation called water char-
acteristic relation or water retention relation.
52 Computational modeling of multiphase geomaterials
Since saturation is a function of suction, that is, the pressure head, the
time rate for saturation is given by

nS n
dS
d
d
d
d
dp
p
d
d
p
r
r
c
c
w
c
1

(2.111)
where =
V
V
w
is the volumetric water content, p
c
is the matric suction
( p p p
c a f
= ( )), = p
c
w
/ is the pressure head for suction, and C
d
d
=

is the specifc water content.
And we need the constitutive equation for air phase such as ideal gas.
REFERENCES
Atkin, R.J., and Craine, R.E. 1976. Continuum theories of mixtures: basic theory
and historical developments, Q. J. Mech. Appl. Math., 29(2):209244.
Biot, M.A. 1941. General theory of three-dimensional consolidation, J. Appl.
Phys., 12(4):155-164.
Biot, M.A. 1955. Theory of elasticity and consolidation for a porous anisotropic
solid, J. Appl. Phys, 26(2):182185.
Biot, M.A. 1956. Theory of propagation of elastic waves in a fuid-saturated porous
solid, J. Acoust. Soc. Amer., 28(2):168191.
Biot, M.A. 1962. Mechanics of deformation and acoustic propagation in porous
media, J. Appl. Phys., 33(4):14821498.
Bolzon, G., Schrefer, B.A. and Zienkiewicz, O. C. 1996. Elastoplastic soil constitutive
laws generalized to partially saturated states. Gotechnique, 46(2):279289.
Bowen, R.M. 1976. Theory of mixtures, in Continuum Physics, Vol. III, A.C. Eringen,
ed., Academic Press, 1127.
Coussy, O. 1995. Mechanics of Porous Continua, New York: John Wiley & Sons.
de Boer, R. 2000a. Contemporary progress in porous media theory, Appl. Mech.
Rev., 53(12):323369.
de Boer, R. 2000b. Theory of Porous Media, New York: Springer.
Ehlers. W. and Graf, T. 2003. Adaptive computation of localization phenomena in
geotechnical applications, Bifurcation and instabilities in geomechanics, J.
Labuz and A. Drescher eds., Swets & Zeitlinger, Lisse, 247262.
Ehlers, W., Graf, T., and Ammann, M. 2004. Deformation and localization analysis
of partially saturated soil, Comp. Meth. Appl. Mech. Eng., 193:28852910.
Eringen, A.C. 2003. Note on Darcys law, J. Appl. Phys., 94(2):1282.
Fillunger, P. 1913. Der auftrieb in talsperren, sterreichische Wochenschrift fr den
ffentrichen baudienst, 19:532556, 567570.
Gallipoli, D., Gens, A., Sharma, R. and Vaunat, J. 2003. An elasto-plastic model for
unsaturated soil incoorporating the effects of suction and degree of saturation
on mechanical behavour, Gotechnique, 53(1):123135.
Governing equations for multiphase geomaterials 53
Jommi, C. 2000. Remarks on the constitutive modeling of unsaturated soils,
experimental evidence and theoretical approaches in unsaturated soil, Proc.
Int. workshop on unsaturated soils, Trento, Tarantino, A. and Mancuso, C.
eds., Rotterdam, Balkema: 139153.
Kimoto, S., Oka, F., Fushita, T., and Fujiwaki, M. 2007. A chemo-thermo-mechani-
cally coupled numerical simulation of the subsurface ground deformations due
to methane hydrate dissociation, Computers and Geotechnics, 34(4):216228.
Kimoto, S., Oka, F., and Fushita, T. 2010. A chemo-thermo-mechanically cou-
pled analysis of ground deformation induced by gas hydrate dissociation,
Int. J. Mech. Sci., 52:365376.
Lewis, R.W., and Schrefer, B.A. 1998. The Finite Element Method in the Static and
Dynamic Deformation and Consolidation of Porous Media, 2nd ed., New
York: John Wiley & Sons.
LIQCA Research and Development Group (Representative: F. Oka, Kyoto University).
2005. Users Manual for LIQCA2D04 (2005 released print), http://nakisuna2.
kuciv.kyoto-u.ac.jp/liqca.htm.
Loret, B., and Khalili, N. 2000. A three-phase model for unsaturated soils, Int.
J. Numer. Anal. Meth. Geomech., 24:893927.
Morland, L.W. 1972. A simple constitutive theory for a fuid-saturated porous solid,
J. Geophy. Res., 77:890900.
Mills, N. 1967. On a theory of multi-component mixture, Q. J. Mech. Appl. Math.,
20:449508.
Oka, F., Yashima, A., Shibata, T., Kato, M., and Uzuoka, R. 1994. FEM-FDM cou-
pled liquefaction analysis of a poroussoil using an elasto-plastic model, Appl.
Sci. Res., 52: 209245.
Oka, F., Kimoto, S., Takada, N., Gotoh, H., and Higo, Y. 2010. A seepage-deforma-
tion coupled analysis of an unsaturated river embankment using a multiphase
elasto-viscoplastic theory, Soils and Foundations, 50(4):483494.
Sheng, D., Sloan, W., Gens, A., and Smith, D.W. 2003.Finite element formulation
and algorithms for unsaturated soils, Part I: Theory. Int. J. Numer. Anal. Meth.
Geomech., 27:745765.
Terzaghi, K. 1943. Theoretical soil mechanics, John Wiley & Sons Inc., New York.
Vardoulakis, I. 2002. Dynamic thermo-poro-mechanical analysis of catastrophic
landslides, Gotechnique, 52(3):157171.
Zienkiewicz, O.C., Chan, A.H.C., Pastor, M., Schrefler, B.A., and Shiomi,
T. 1999. Computational Geomechanics with Special Reference to Earthquake
Engineering, John Wiley & Sons.
Zienkiewicz, O.C., Chang, C.T., and Bettes, P. 1980. Drained, undrained, consolidat-
ing and dynamic behavior assumptions in soils, limits of validity, Gotechnique,
30(4):385395.
55
Chapter 3
Fundamental constitutive
equations
As has been discussed in Chapter 1, Section 1.9, constitutive equations are
required to describe material responses to external actions. In this chapter,
we will review fundamental constitutive equations such as the elastic model
and the elastoplastic model.
3.1 ELASTIC BODY
The reversible deformation of materials is described by elastic constitutive
equations. The linear elastic constitutive equation, whose stressstrain is
linear, is a generalization of Hookes law. It is given by


ij ijkl kl
E = (3.1)

E
ijkl ij kl ik jl il jk ik jl il jk
= + + + ( ) ( ) (3.2)
where E
ijkl
is a fourth-order isotropic tensor, and , , are material constants.
Assuming a symmetry for the stress and strain tensors, the elastic coef-
fcient, E
ijkl
, can be expressed by

E
ijkl ij kl ik jl il jk
= + + ( ) (3.3)
and Equations (3.1) and (3.3) yield


ij kk ij ij
= + 2 (3.4)
where and are called Lams constants.
Since the constitutive model for an elastic body depends on the strain, that
is, the initial shape, the elastic model is considered to be a model for solids.
56 Computational modeling of multiphase geomaterials
The linear elastic constitutive equation can be written in components as

xx xx yy zz
yy yy zz xx
zz
E
E
=
=
=
1
1
1
( )
( )
EE
zz xx yy
( )

xy xy
yz yz
zx zx
E
E
E
=
+
=
+
=
+
1
1
1
(3.5)
where E is Youngs modulus and is Poissons ratio.
Under isotropic compressive stress conditions,
xx yy zz
c = = = , =
xy
= =
yz zx
0, and strain components are given by

xx yy zz xx
E E
c = = =

=
1 2 1 2
Subsequently,



v xx yy zz
E
c
K
c = + + =

=
31 2 1 ( )
where K is the elastic volumetric modulus, namely,

K
E
=
31 2 ( )
(3.6)
Under shear stress conditions, shear strains develop as




xy xy xy xy
E G
= =
+
=
1
2
1 1
2
Hence,

G
E
=
+ 2 1 ( )
(3.7)
where G is the elastic shear modulus.
Fundamental constitutive equations 57
Under one-dimensional compressive conditions, for example,
yy
c = ,

= = = =

=
E E
c
E
c
yy yy xx zz yy
1 1
,
(3.8)
where is Poissons ratio.
When elastic stored energy is positive, that is, W > 0,

W
ij ij kk ij ij
= = +
1
2
1
2
2

(3.9)
where and are positive. Hence,

< < 1
1
2
(3.10)
3.2 NEWTONIAN VISCOUS FLUID
Fluid is defned as a material whose stress tensor is isotropic at the state of
rest. If the material behavior depends on the strain rate, the material exhib-
its viscosity. The material model in which the shearing stress is linearly
proportional to the strain rate is known as a linear Newtonian fuid. The
model is expressed as


ij ij kk ij ij
p D D = + +

2 (3.11)
where p is the pressure, and
*
and
*
are viscous coeffcients.
For incompressible materials,

s D
ij ij
= 2
*
(3.12)
where D
ij
is the rate of deformation tensor, s
ij
is the deviatoric tensor, and
*
is the viscous coeffcient.
Equation (3.12) indicates that the rate of deformation tensor D
ij
(for
small deformation felds, it corresponds to strain rates
ij

) is dependent on
the present stress, and consequently, the deformation continues even under
small deviatoric stresses.
58 Computational modeling of multiphase geomaterials
3.3 BINGHAM BODY AND VISCOPLASTIC BODY
The simplest viscoplastic model is a fuid model with a yield stress limit; it is
called the Bingham body. For the Bingham model with yield shear stress, k,
viscous fow never occurs when the stress is less than the yield stress. The vis-
coplastic strain rate,
, is proportional to shear stress, , when the stress is more
than yield shear stress, k.

= F 2 (3.13)
F = (F + |F|)/2, F = 1k/ (3.14)
where is a viscous coeffcient.
Hohenemser and Prager (1932) have generalized the Bingham model and
have proposed a viscoplastic model, namely,

= D F S
ij ij
2 (3.15)
F = 0; F < 0, F = F; F 0
(3.16)

F
k
J
= 1
2
2
(3.17)
where J s s
ij ij 2
1
2
= is the second invariant of the deviatoric stress tensor.
From Equation (3.15), we obtain

4 2
2
2
2 2
2
I F J = (3.18)
where I
2
is defned as

I D D
ij ij 2
= (3.19)
Substituting Equation (3.17) into Equation (3.14) yields

= + = + s
k
I
D D
k
I
D
ij ij ij ij
2 2
2 2
(3.20)
for F > 0.
Fundamental constitutive equations 59
Since the second term in Equation (3.20) is the zeroth order homoge-
neous of degree one with respect to the rate of deformation, the term is rate
independent. Therefore, the model has a viscous term and a rate-independent
plastic term with yield stress, k S k
ij
D
I
ij

( )
2
corresponds to the overstress that
will be discussed in Section 3.6.
3.4 von MISES PLASTIC BODY
Von Mises (1928) proposed a rigid perfectly plastic model, and the yield
condition for the von Mises plastic model and the fow rule are given by
Yield condition:
s s
J k
ij ij
2
2
2
2
= = / (3.21)
Flow rule: D
s
J
ij
ij
=
2
2
(3.22)
Using Equation (3.18), from Equation (3.21) we obtain = I
2
Hence,

s
k
I
D
ij ij
=
2
(3.23)
This equation is given by replacing viscous coeffcient, , in the linear
Newtonian viscous fuid model with
k
I
2
.
The right-hand term in Equation (3.23) is homogeneous of degree one
with respect to D
ij
; the von Mises model is a rate-independent plastic
model.
Hohenemser and Pragers model is similar to the Maxwell type of visco-
elastic model that will be explained later, although shear fow never occurs
below the yield stress in Hohenemser and Pragers model.
3.5 VISCOELASTIC CONSTITUTIVE MODELS
In general, soils exhibit elasticity, viscosity, and plasticity. Materials with
viscosity are affected by the strain rate or the stress rate. In particular, mod-
els that can describe viscous and elastic behaviors are called viscoelastic
models. The representative viscoelastic models include the Maxwell model
and the KelvinVoigt model.
60 Computational modeling of multiphase geomaterials
3.5.1 Maxwell viscoelastic model
The one-dimensional Maxwell type of viscoelastic model can be illustrated
by the model with a series of elastic and viscous elements, as shown in
Figure3.1.
Elastic strain,
e
, and viscous strain rate,
v

, are given by

e
E
= (3.24)

v

=
(3.25)
where is the stress, E is the elastic modulus, and is the viscous coef-
fcient. The total strain rate is given by the sum of the each strain rate, as

= +
e v
.
Then, the constitutive equation for the Maxwell model is given by

= +
E
(3.26)
Equation (3.26) can be extended to a multidimensional form as



ij ijkl kl ijkl kl
C A = + (3.27)
Equation (3.27) shows that the strain rate is a function only of the quanti-
ties determined by the present states, that is, the stress and the stress rate.
This implies that the model is for liquid or fuid. For this reason, the model
is called the Maxwell fuid model.
E

Figure 3.1 Maxwell model.

Fundamental constitutive equations 61
3.5.2 KelvinVoigt model
The KelvinVoigt model consists of an elastic spring element and a viscous
dashpot element, which are connected in a parallel manner, as shown in
Figure3.2.
Since strain is for two elements and the total stress is broken into two
parts, as = +
e v
, the constitutive equation is given by

=
e
E
(3.28)

=
v
(3.29)

= + E

(3.30)
Equation (3.30) can be extended to a multidimensional form as


ij ijkl kl ijkl kl
E = +

(3.31)
For the KelvinVoigt model, the stress depends not only on the strain
rate, which is determined by the current states but also on the strain, which
is affected by the value at the reference state. This indicates why the Kelvin
Voigt model is called the Voigt solid model.
E
Figure 3.2 Voigt model.
62 Computational modeling of multiphase geomaterials
3.5.3 Characteristic time
When strain, , is applied to the Maxwell type of viscoelastic body at t = 0
instantaneously, we have

1
0
E

+ = (3.32)
Solving Equation (3.29) with the initial stress of
0
, we obtain the stress
time profle:

= /
0
exp t ( ) (3.33)
where =
0
/e at t = .
Then, = /E denotes the time when the stress becomes 1/e of the ini-
tial value and is called the relaxation time, where e is the natural logarithm
(Figure3.3).
Now let us consider the behavior of the KelvinVoigt model under con-
stant stress conditions, that is, the creep process. If the applied stress is
constant as
0
, Equation (3.30) yields


0
= + E

(3.34)
The solution for the preceding equation is given by

= /
0
1
E
exp t ( ( ))
(3.35)

=
0
/e

0
O
t
Figure 3.3 Stress relaxation.
Fundamental constitutive equations 63
in which the initial strain is zero and = /E, and just after the loading at
time t = 0, the strain approaches
0
/ E as t . is a parameter that indi-
cates the rate at which the strain exponentially attains equilibrium and is
called retardation time (see Figure3.4).
The two aforementioned viscoelastic models can be formulated in an
integral form, namely, a convolution integral, which uses the relaxation
and creep functions as

t f t t
d
dt
dt
ij
t
( ) ( )

(3.36)

t g t t t dt
ij
t
( ) ( ) ( )

(3.37)
where f(t) and g(t) are relaxation and creep functions.
Exercise
Derive the constitutive equation for the three-element model consisting of the elastic
element and the Voigt element.
3.6 ELASTOPLASTIC MODEL
As has been described, the features of materials such as mechanical, ther-
mal, and electromagnetic characters are expressed by constitutive equa-
tions. When we apply loads to the materials, the materials deform. If the
strain induced by the loading is small, the materials behave elastically.
However, part of the deformation cannot be recovered if the strain level is
O

/e

t
t
0
Figure 3.4 Strain during the creep.
64 Computational modeling of multiphase geomaterials
high (Figure3.5). This type of behavior is elastoplastic and it is modeled
by an elastoplastic constitutive equation. In the elastoplastic model, time-
dependent behavior such as creep is not considered. Classical elastoplastic
models are generalizations of the von Mises plasticity model and have
been developed by Hill (1948), Drucker (1951), Prager (1949), and Mandel
(1964). In this section, elastoplastic models will be explained within the
framework of a small strain feld with the yield function, loading criteria,
the plastic potential, and hardening rules.
3.6.1 Yield conditions
The yield conditions prescribe the elastic limit and can be classifed as
initial yield conditions and subsequent yield conditions. The yield condi-
tions prescribe the elastic limit under which materials behave elastically.
The initial yield conditions are for the initial yield, and the subsequent
yield conditions are for the yield conditions after the initial yielding
(Figure3.5).
For a perfectly elastoplastic body, deformation increases without the need
for any change in stress after the initial yielding, while for the hardening
(i.e., work-hardening) materials, it is necessary for the stress to be increased
for the deformation to continuously increase after the initial yielding. The
yield conditions after the material has been initially yielded are called the
subsequent yield conditions (Figure 3.6). For one-dimensional problems,
the yield conditions correspond to the simple yield limit stress. For multi-
dimensional problems, on the other hand, the yield conditions have to be
described within the multidimensional stress space.

O b
a
d
c e
Reloading
Unloading
and
reloading
Figure 3.5 Stressstrain relation for elastoplastic material.
Fundamental constitutive equations 65
The yield function in the multidimensional stress space is written with
hardening, that is, an increase in the yield limit as

f k
ij ij
( ) , = 0 (3.38)
where
ij
is the stress tensor and k
ij
is a hardening-softening parameter.
When we take account of the strain hardening, it is usual to assume that
k
ij
is a function of the plastic strain or the plastic work.
In the case of plastic strain hardening,

f k
ij ij kl
p
( ( )) , = 0
(3.39)
in which
ij
p
ij
p
d = is the plastic strain tensor.
For isotropic materials, the yield function is expressed by the scalar func-
tion of the stress and the plastic strain as

f I I I k
ij
p
( ( ))
1 2 3
0 , , , =
(3.40)
where k
ij
p
( ) is the scalar function, namely, I
1
, I
2
, and I
3
are the frst, the
second, and the third invariant of the stress tensor, respectively.
3.6.2 Additivity of the strain
The total strain increment is assumed to be the sum of plastic strain incre-
ment, d
ij
p
, and elastic strain increment tensor, d
ij
e
:

d d d
ij ij
e
ij
p
= + (3.41)
Initial yield surface Subsequent yield surface
f = 0
O
Figure 3.6 Yield surfaces.
66 Computational modeling of multiphase geomaterials
3.6.3 Loading conditions
The loading indicates the changes in stress that induce the plastic strain.
For the strain-hardening case, the yield surface expands, that is, the stress
point moves to a point outside the current yield surface. Then,

f d k
ij ij ij
p
( ( )) + , > 0 (3.42)
Since the current stress point is on the yield surface,

f k
ij ij
p
( ( )) , = 0
(3.43)
Hence,

>
= .
f
d
ij
ij
k
ij
p
const

( )
0 (3.44)
or

df
k const
ij
p
( ) = .
> 0 (3.45)
Next, we consider the stress change, d
ij

, parallel to the tangent of the

yield surface. Then,

= .
f
d
ij
ij
k
ij
p
const

( )
0 (3.46)
Hence,

df
ij
p
const = .
= 0 (3.47)
In this case, the yield surface does not expand and the plastic strain does
not yield, and the stress change is called neutral loading. When the direc-
tion of the stress change is toward the inside of the yield surface, plastic
strain does not occur:

<
= .
f
d
ij
ij
k
ij
p
const

( )
0
(3.48)
Fundamental constitutive equations 67
Then, the plastic strain does not occur and the process is elastic. This pro-
cess is called unloading:

df
k const
ij
p
( ) = .
< 0
(3.49)
3.6.4 Stability of elastoplastic material
Drucker (1951) defned stable material by the following remark called sta-
bility postulates: The plastic work done by an external agency during the
application of additional stress is positive.

d d
ij ij
> 0
(3.50)
This condition is called the local stability or the stability in the small.
This condition is satisfed in the strain-hardening range, but d d
ij ij
< 0
in the strain-softening range (Figure 3.7). Equation (3.50) is called the
positiveness of the second-order work. In fact, it has been pointed out
that Druckers stability postulate is a suffcient condition for the stability
in the sense of Lyapunov, which was derived by Hill (1958). This means
that if the second order work by Equation (3.50) is violated, the mate-
rial is potentially unstable. The violation of the Lyapunov stability causes
the materials to experience unstable behavior such as strain localization,
that is, shear band formation and diffuse failure. Strain localization indi-
cates the instability with preferred orientation. Diffuse failure denotes the
instability with no preferred orientation and manifests the barreled type
of failure mode. Darve et al. (2001) pointed out that the diffuse failure
Strain softening
Strain hardening
B
E F
D
C
A
O

Figure 3.7 Strain hardening and strain softening.

68 Computational modeling of multiphase geomaterials
model for loose sand is due to the violation of Hills stability criteria, that
is, second-order work criteria. Nova (1994) showed that Hills condition is
equivalent to the controllability condition.
Hill (1957, 1958) proposed a stability condition for elastoplastic bodies at
the equilibrium state in the sense of Lyapunov. Hills stability condition is a
suffcient condition for stability and has been explained by Nakamura and
Uetani (1981) as the stability of the body system in the sense of Lyapunov
is that the displacement is small when the disturbance is small.
The stability of the solution of differential equation in the sense of
Lyapunov is defned as follows: A solution f(x, t) at point x is stable in the
sense of Lyapunov if for > 0. There exists > 0 such that if | | x y < ,
then || ( , ) ( , )|| f x t f y t < for all t, where t is a time and |||| is a norm for the
solution.
Hills theorem is described as follows: Let state S
II
be near equilibrium state
S
I
. If the following inequality is satisfed, S
I
is stable in the sense of Lyapunov:

U W
ex
> 0 (3.51)
where U is the sum of the stored energy in the body and the energy con-
sumed for the plastic deformation during the deformation from S
I
to S
II
,
and W
ex
is the work to the body by external energy (Figure3.8).
During the process from S
I
to S
II
, the following energy balance holds:

K U I W D
ex
+ = + (3.52)
where K is the kinetic energy at S
II
, I is the input energy in the body sys-
tem, and D is the consumed energy caused by viscous damping.
From Equation (3.51), there exists a positive number that satisfes
( )
min
U W
ex
> . For that we take I as I<. Since D is nonnegative,
the following inequality holds:

K U W I D I
ex
= + + < ( ) 0 (3.53)
S
II
S
I
(u + du, K)
(u, 0)
D
I
W
ex
Figure 3.8 Energy balance.
Fundamental constitutive equations 69
Since kinetic energy, K, is nonnegative, Equation (3.53) does not
hold. It means the state S
II
cannot be reached due to small disturbance
, which is smaller than input energy, I. The previous discussions indi-
cate that the elastoplastic body system is always stable in the sense of
Lyapunov if the body system satisfes Equation (3.51). Hence, if the body
does not satisfy Equation (3.51), the body system is potentially unstable.
In other words, a small disturbance may grow. Let us explain Equation
(3.51) by the principle of virtual work within the framework of a small
displacement feld:

U u dv
ik
v
ik i k
= +

( )
,
*
(3.54)
where

ik is the change in stress due to the small disturbance.

W t u ds g u dv
ex i
s
i i
v
i
= +


* *
(3.55)
Since


ik
v
i k i
s
i i
v
i
u t u ds g u dv

= +
,
* * *
(3.56)
and Equation (3.51) holds, we obtain

= > W dv
ij
v
ij
0
2
(3.57)
It is worth noting that this condition is a suffcient condition for stability
and is called the stability in the sense of second-order work.
3.6.5 Maximum work theorem
The maximum work theorem is given by the following proposition by
Drucker (1951): The net work done by the external agency during the
cycle of adding and removing stress is non-negative (Figure3.9).
Let us consider a stress cycle where the stress is at point a at the initial
state, then the stress changes to point b on the yield surface, then the stress
changes further to point c outside the yield surface, then the stress changes
to point d on the yield surface, and fnally the stress returns to the initial
point by the unloading process (Figure3.9).
70 Computational modeling of multiphase geomaterials
Since the process from point a to point b and from point d to point a is
elastic, the total work during the cycle is given by

w d
ij ij
cycle
=

(3.58)

w d d d d
d
a
b
ij ij
e
b
c
ij ij
e
ij
p
c
a
ij
e
= + + +
=



( )
iij
e
cycle
b
c
ij ij
p
d

+
(3.59)
Since the elastic deformation is reversible, the frst term on the right-hand
side is zero. Hence, the total work done by the stress is

w d D d
b
c
ij ij
p
ij
p
= =

( ) (3.60)
In Equation (3.60), D d
ij
p
( ) is called the internal dissipation.
From the above, the work done by the stress that causes the change in
the system becomes

b
c
ij ij
a
ij
p
d

( )
(3.61)
In addition, for the preceding equation to be held to the loading from b
to c during the arbitrary short time,

( )
ij ij
a
ij
p
d 0 (3.62)
c
d
b

c
ij

b
ij

d
ij

a
ij
a
Figure 3.9 Stress cycle and work.
Fundamental constitutive equations 71
where we have equality if and only if the process is in the elastic range.
Equation (3.62) is known as the maximum plastic work theorem.
Equation(3.62) can be written as


ij ij
p
ij
a
ij
p
d d (3.63)
Accordingly, the maximum work theorem indicates that the work dur-
ing the plastic deformation is the maximum compared with the work done
by the stress that does not violate the yield conditions. In other words,
internal dissipation, D d
ij
p
( ) , is the maximum during the plastic process. It
is worth noting that stress,
ij
, causes plastic deformation and the maxi-
mum work theorem holds in the strain-softening process (Figure3.10).
3.6.6 Flow rule and normality (evolutional
equation of plastic strain)
When the yield surface is convex with no cusp and it is continuous, and the
direction of the plastic stress increment depends on current stress,
ij
, the
fow rule is given by

d h
f
ij
p
ij

(3.64)
h > 0 (3.65)
Network
b
c
c
b
O

a d a d
Figure 3.10 Network during the stress cycles.
72 Computational modeling of multiphase geomaterials
where f is the plastic potential and h is the scalar parameter. f is called the
plastic potential simply because of its similarity to the potential in classi-
cal mechanics and f = const. surface is the equipotential surface. Now let
us consider the yield surface and the fow rule that satisfes the maximum
plastic work theorem.
The suffcient conditions for the yield surface and the fow rule to sat-
isfy the maximum work theorem have been derived by Drucker (1951) as
follows:
1. Convexity of the yield surface
2. Plastic potential is given by the yield function
In this case, the fow rule is called the associated fow rule. This condi-
tion is also called the normality of the strain rate tensor onto the yield
surface.
It can be shown that the two aforementioned conditions are suffcient
conditions for the maximum work theorem, as follows: As shown in
Figure3.11, if the yield surface is not convex in the six-dimensional stress
space, the angle between ( )
ij
p
ij
o
and d
ij
p
is obtuse when we consider the
strain space and the stress space to be the same.
Then, the inner products of the vectors show that ( )
ij
p
ij
o
ij
p
d < 0; this
violates the maximum work theorem. Moreover, even if the yield surface
is convex, the angle between
ij
p
ij
o
and d
ij
p
is obtuse if the plastic
strain increment vector is not perpendicular to the yield surface. When the
two aforementioned conditions are satisfed, the maximum work theorem
always holds. Hence, the two conditions are suffcient conditions.
The normality rule can be directly obtained from the maximum
work theorem (Maugin 1992). In addition, Collins and Houlsby (1997)
presented that the normal fow rule is derived based on the thermo-
dynamical consideration by Ziegler (1983), which will be shown in
Equation(3.130).
O
p
d
ij
p
P

ij
O

ij
Figure 3.11 Convexity of yield surface.
Fundamental constitutive equations 73
3.6.7 Consistency conditions
When stress,

ij
, on the yield surface moves onto the new stress,

ij ij
d + , the following equation must hold for the case of strain hardening
(see Figure3.12) as:

f d k dk
ij ij ij
p
ij
p
( ( ) ( )) + , + = 0 (3.66)

f k
ij ij
p
( ( )) , = 0 (3.67)
Then,

f d k dk f k
ij ij ij
p
ij
p
ij ij
p
( ( ) ( )) ( ( )) + , + , = 0 (3.68)
Taking a frst-degree Taylor series expansion of the yield function, with
respect to the plastic strain, the stress to the frst-order term yields

f d k dk f
f
ij ij ij
p
ij
p
ij ij
p
ij
( ( ) ( )) ( )

+ , + = , +

dd
f
k
k
d
ij
ij
p
ij
p

= 0 (3.69)
Stress space
f(
ij
, k) = 0
f (
ij
+ d
ij
, k + dk) = 0
O
Figure 3.12 Subsequent yield surface.
74 Computational modeling of multiphase geomaterials
Hence,

= = =
f
d
f
k
k
d df or f
ij
ij
ij
p
ij
p

0 0

(3.70)
This equation is called Pragers consistency condition, and it indicates that
the stress points during the loading process must always be on the yield
surface.
From this Pragers condition, we can determine parameter h in the fow
rule as

h
d
f
ij
f
k
k
f
ij
ij
p
ij
=

(3.71)
These loading and unloading conditions can be alternatively expressed
by the following KuhnTucker complementary condition:

h f k hf k
kl kl
= 0 0 0 , ( , ) , ( , )
(3.72)
h0 is determined by the following consistency condition:
Consistency condition: hf k
kl

( , ) = 0 (3.73)
Strain hardening is given by

k k
ij
p
= ( ) or

k
k
ij
p
ij
p
=

(3.74)
In other expressions, the loading and unloading conditions are expressed by
f < 0 h = 0 (elastic)

= = > f f h
kl
p
0, 0 and 0 0, 0
(loading)

= = = = = f f h
kl
p
0, 0 and 0 0, 0

(neutral loading) (3.75)

f f h
kl
p
= < = = = 0 0 0 0 0 , ,

(unloading)
Fundamental constitutive equations 75
It is worth noting that

f > 0 is excluded since f 0 for a subsequent

short increment.

f
f f f
C
f
ij
ij
ij
ijkl kl kl
p
=

( )

f
C
f
C h
f f
ij
ijkl kl
ij
ijkl
kl

=

ij
p
ij
p
ij
ijkl kl
ij
ijkl
kl
f
C
f
C h
f

f
h
f
f
C h
f
C
ij
p
ij
ij
ijkl kl
ij
ijk

ll
kl
ij
p
ij
f f f

0
(3.76)
Let us assume that the following inequality holds:

>
f
C
f f f
ij
ijkl
kl
ij
p
ij

0
(3.77)
From

f = 0,

h
f
C
f
C
f f
ij
ijkl kl
ij
ijkl
kl
ij
p
=

ff
ij

(3.78)
Tensors of tangent elastoplastic moduli are obtained by

ij ijkl kl kl
p
ijkl
ep
kl
C
C
=
=
( )
(3.79)

= = C C h
ijkl
ep
ijkl
e
if 0
(3.80)
If plastic potential function g is not equal to yield function f, that is,

d h
g
ij
p
ij

(3.81)
76 Computational modeling of multiphase geomaterials
we obtain non-associated fow rule, and

C C
C
g
C
f
f
C
ijkl
ep
ijkl
e
ijmn
e
mn
klpq
e
pq
ij
i
=

jjkl
kl
ij
p
ij
g f g
if h

>

0
(3.82)
For the case in which g = f, the theory is called the associated fow rule.
3.7 OVERSTRESS TYPE OF ELASTOVISCOPLASTICITY
3.7.1 Perzynas model
Perzyna (1963) has formulated a viscoplastic model called the overstress
model in order to generalize the linear viscoplastic model by Hohenemser
and Prager (1932). Perzyna adopted a nonlinear functional of F and proposed
the following fow rule. He has tried to take the nonlinear form of function
F in Equation (3.84) and has proposed the following viscoplastic fow rule:


=

F
f
ij
vp
ij
( )
(3.83)

F
f
=

(3.84)
where
ij
vp

is the viscoplastic strain rate tensor, f is a dynamic loading function

or dynamic viscoplastic potential, is a viscosity coeffcient, is a functional
of F that denotes a material viscoplasticity, and is a hardening parameter.
(F) = (F); F > 0, (F) = 0; F 0 (3.85)
For the work-hardening case,

= =

ij ij
vp
ij ij
vp
vp
d ,
0
(3.86)
and for the strain-hardening case,

= =

A A d
kl kl
vp
kl kl
vp
vp
,
0
(3.87)
where A
kl
is a material tensor.
Fundamental constitutive equations 77
Since F = 0 denotes a static yield function, dynamic loading function f
takes the same form as the static yield function (Figure3.13).
If we solve Equation (3.83) with respect to f, we have

= +

/
f
I f f
ij
ij
vp
vp
ij ij
( , ) 1
1
2
1 2
(3.88)
Equation (3.88) indicates that the yield function depends on the viscoplas-
tic strain rate;
( )
= I
vp
ij
vp
ij
vp
2
1/2
.
In addition, the dynamic yield function coincides with the static yield
function as , that is, the coeffcient of viscosity becomes infnite, .
Total strain rate tensor,

ij
, is composed of viscoplastic strain rate tensor,
vp
ij

, and elastic strain rate tensor,

ij
e
, as

ij ij
e
ij
vp


= +
(3.89)
The graphical presentation of the Perzynas model leads to Figure3.14.
3.7.2 Duvaut and Lions model
Duvaut and Lions (1972) proposed a viscoplastic model as follows:



ij ij
e
ij
vp
= + (3.90)
Dynamic loading surface
Static yield surface
O
Figure 3.13 Dynamic loading surface.
78 Computational modeling of multiphase geomaterials
When F < 0,

ij
vp

= 0
(3.91)
When F 0,

ij
vp
ij K ij
P

=
1
2
[ ( ) ]
(3.92)
where is a viscosity coeffcient and ( ) P
K ij
is the stress obtained by the
orthogonal projection of the current stress on the closed convex set K.

K F
ij ij
= , { ( ) } 0
(3.93)
Figure 3.14 Perzyna-type viscoplastic model.
Fundamental constitutive equations 79
where F is a continuous a convex function and corresponds to the static
yield surface.
Rewriting Equation (3.92) yields


ij K ij ij
vp
P = + ( ) 2
(3.94)
In the model, the stress is broken into viscous and plastic elements, as
shown in Figure 3.15. The feature of the model is that the viscoplastic
model can be constructed only by introducing a viscous coeffcient, not
by introducing a nonlinear functional. The limit of the model leads to the
linear viscoplastic model.
Figure 3.15 Duvaut and Lion-type viscoplastic model.
80 Computational modeling of multiphase geomaterials
3.7.3 Phillips and Wus model
Phillips and Wu (1973) proposed a viscoplastic model that is similar to
Duvaut and Lions model and at almost the same time. In Philips and Wus
theory, an excess stress is defned by h, which is obtained as the perpendicu-
lar distance from the current stress point to the quasi-static yield stress in
the deviatoric stress space. k
ij
is given by the stress at the intersection of the
perpendicular line to the quasi-static yield surface from current stress, s
ij
,
and the quasi-static yield surface.

h hn s k
ij ij ij ij
= =
(3.95)
The direction of the viscoplastic strain rate is parallel to the direction of h
ij
.
Subsequently, the deviatoric fow rule is given by

ij
vp
rs rs
vp vp
ij e
k e I h n h

= , , : > ( ) [ ]
2
0 (3.96)

ij
vp
e
h

= : = 0 0 (3.97)
where ( ) k e I
rs rs
vp vp
, ,
2
is a viscosity function that depends on position k
rs
, the
viscoplastic deviatoric strain e
rs
vp
, and its second invariant of viscoplastic
deviatoric strain rate, I
vp
2
. Function [h] is a scalar function of h.
If we obtain stress point k
ij
by Duvaut and Lions theory as ( ) P
K ij
, it
is not always possible to take a stress point on the quasi-static yield sur-
face because the line from the current stress point is perpendicular to the
dynamic yield function.
3.8 ELASTOVISCOPLASTIC MODEL BASED
ON STRESS HISTORY TENSOR
Oka (1985) has proposed a model that can describe the continuous transi-
tion from the viscoplastic theory to the plastic theory. He constructed an
elastoviscoplastic theory based on the concept of memory and internal vari-
ables by Perzyna (1971) using stress history tensor,
ij

. Using this theory, it

is possible to express strain softening and a continuous transition between
plastic and viscoplastic models.
It is assumed that total strain rate,
ij

, can be composed of two parts:

elastic component,
ij
e

; and viscoplastic or plastic component,

ij
vp

.

ij ij
e
ij
vp


= +
(3.98)
Fundamental constitutive equations 81
3.8.1 Stress history tensor and kernel function
Total stress history,
ij
z
, is a union of the two sets of the present stress and
reduced stress history,
rij
z
, which is defned by the history of the stress
without the present stress. Following Eringens manner (1967), the total
history tensor and the reduced stress history tensor are given by


ij
z
ij rij
z
z z z = , , < [ ( ) ] 0 (3.99)


rij
z
ij
z z z z = , < [ ( ) ] 0 (3.100)
where time measure z is defned by

dz g dt
ij
= ( )

(3.101)
Taking account of the integral, in a wider sense, the total stress history
tensor can be expressed by the functional of the reduced stress history as

ij ij rij
z
=

( )
(3.102)
The total stress history tensor can be expressed by the convolution inte-
gral, with respect to the generalized strain measure, as


ij
z
ij
K z z z dz

0
( ) ( )
(3.103)
where K is a continuous kernel function.
It is assumed that K/z < 0. This assumption provides that the stress
history tensor satisfes the principle of fading memory with respect to the
stress history. Namely, the effect of the older stress on the stress history
tensor becomes small (Figure3.16).
Based on the theorem of the integral, in a wider sense, the stress history
is defned in the closed interval as z z 0 .
3.8.2 Flow rule and yield function

d H
f
df
ij
p
p
ij
y

(3.104)
where f
p
is the plastic potential function,
f
y
is the yield function, and H is
the hardeningsoftening function.
82 Computational modeling of multiphase geomaterials
The yield function is a function of the stress history tensor and hardening
and softening parameter, .

f
y ij
( )

, = 0 (3.105)
Loading conditions are given by
When f df d
y y
f
ij
y
ij
= , = >

0 0 [ ( ) ]

, then
d
ij
p
0
(3.106)
When f df
y y
= , = 0 0, then d
ij
p
= 0 (3.107)
When f df
y y
= , < 0 0, then d
ij
p
= 0 (3.108)
Strain Measure z
S
t
r
e
s
s

z
z
O

(0 z z)
z
z O
K

(
z

)

=

e
x
p

(
z

)
/

(a) (b)
z
O
z
K

(
z

(
z

)

(0 z z)
z
z
O

*

=

z

K

(
z

(
z

)

d
z

0
(c) (d)
Figure 3.16 (ad) Stress and stress history tensor.
Fundamental constitutive equations 83
The plastic potential function is a function of the current stress and the
other history parameter, L.

f L
p ij
( ) , = 0 (3.109)
3.9 OTHER VISCOPLASTIC AND
VISCOELASTICPLASTIC THEORIES
Naghdi and Murch (1963) proposed a viscoelasticplastic model consider-
ing the yielding of viscoelastic materials. Tateishi and Miyamoto (1974)
also constructed a viscoelastic yield theory. Kremple (1986) proposed an
overstress theory using total strain rates, not using the concept of visco-
plastic strain for metallic materials. On the other hand, Chaboche and
Rousellier (1983a,b) derived a more general kinematic viscoplasticity the-
ory based on the nonlinear hardening theory advocated by Armstrong and
Frederick (1966) and Perzynas type of viscoplasticity theory. Chaboches
type of viscoplasticity theory will be explained in Chapter 6. Valanis
(1971) has initiated an integral type of viscoplasticity theory without the
yield surface in which he defned an inherent time measure. In his theory,
Valanis has used a convolution integral of strain with respect to an inter-
nal time; hence, his theory is called the endochronic theory. Watanabe and
Alturi (1986) generalized it as an internal time theory.
3.10 CYCLIC PLASTICITY AND VISCOPLASTICITY
Cyclic constitutive models for materials are of great importance in the
numerical simulation of behavior during cyclic loading or dynamic load-
ing, such as for vibrational problems.
Prager (1949) is the frst to have proposed a cyclic plasticity model with
a linear kinematical hardening rule. Pragers linear kinematical harden-
ing model can reproduce the Bauschinger effect, cyclic hardening, and the
cyclic hysteresis loop for the stressstrain curves.
The Bauschinger effect indicates the lowering of the yield stress upon
reverse loading behavior, such as compression after loading in tension, as
shown in Figure3.17. The Bauschinger effect is named after the German
engineer Johann Bauschinger (18601934).
Pragers yield function and kinematical hardening rule are given by

f R
ij ij
( ) = 0 (3.110)

d cd
ij ij
p
=
(3.111)
84 Computational modeling of multiphase geomaterials
where f R
ij ij
( ) = 0 is the yield function and

ij is the kinematical
hardening parameter, indicating the size of an initial yield surface, R is the
plastic strain increment, and c is a material parameter. Schematic fgure for
the kinematical hardening is shown in Figure3.18.
Pragers model has a shortcoming regarding the reduction in dimen-
sion of the stress space, such as in plane stress problems. To improve this
point, Ziegler (1959) proposed another type of kinematical hardening
(Strain)
(Stress)

Yi

Yi

Yi
2
Yi
Figure 3.17 Bauschinger effect.
(Stress space)
f = 0
f = 0
Figure 3.18 Kinematic hardening.
Fundamental constitutive equations 85
for the cyclic plasticity model. Zieglers kinematical hardening rule is
given by

d cd
ij ij
= (3.112)
where c is a constant hardening modulus.
Both Pragers and Zieglers models are restricted to linear hardening, although
experiments denote nonlinear hardening. To solve this problem, many other
models have been proposed. Mroz (1981) proposed a cyclic plasticity model
based on the feld of work-hardening moduli to describe the nonlinear hard-
ening shown in the experiments using a large number of yield surfaces.
Dafalias and Popov (1976) proposed a nonlinear cyclic plasticity model
introducing a continuously varying plastic modulus in which the hardening
modulus depends on the distance between the present stress state and the
stress state on the bounding surface, which is called the bounding surface
model or the two-surface model.
Chaboche (1977) newly proposed a nonlinear hardening plasticity model
based on the theory by Armstrong and Fredericks (1966) nonlinear kine-
matical hardening rule (Figure3.19). In the nonlinear hardening model, the
hardening parameter follows the differential equations as

dX cd X dp
ij ij
p
ij
=
2
3

(3.113)

dp d d
ij
p
ij
p
=
2
3
1 2

/
(3.114)
where X
ij
is a kinematical hardening parameter.
Bounding
surface
Yield surface

k
k
k
k
k
X
X
X
O

2
J () = k +
J ( X) = k
c

p
Figure 3.19 Nonlinear kinematic hardening rule (Chaboche and Rousselier 1983).
86 Computational modeling of multiphase geomaterials
The plastic fow rule is given by

d d
f
ij
p
ij

(3.115)

dp d
h
f
d
h
X
R k
d
ij
ij
ij ij
ij
= =

=

+

1 1 3
2
(3.116)

h c
f f
X
f f f
ij ij
ij
ij kl kl
=

2
3
2
3


1 2 /
(3.117)
Chaboche and Rousselier developed a viscoplastic cyclic model based on
the nonlinear kinematical hardening rule (1983).
For the case where the von Mises criterion is used, the viscoplastic fow
rule is given by

d
J X R k
K
f
ij
vp
ij ij
n
ij

( )
(3.118)

= dX C ad X dp
ij
ij
vp
ij
2
3
(3.119)
where < > is Macauleys bracket and n,K,R,k,C, and a are material constants.
During cyclic plastic loading, progressive deformation is, in general,
observed. However, if there is no progressive plastic deformation, we can
describe the cyclic hysteresis loop by the following type of one-dimensional
stressstrain relation as

f
r r
(3.120)
where is a scaling factor and
r
and
r

are the values at the stress reversal
point.
When is 2 in Equation (3.110), it is called the Masing rule (1927).
3.11 DISSIPATION AND THE YIELD FUNCTIONS
The energy dissipation has been dealt with with the ClausiusDuhem
inequality as shown in Chapter 1. For the dissipation, the following impor-
tant results have been obtained. For the existence of the yield functions,
Fundamental constitutive equations 87
Collins and Houlsby (1997) adopted Zieglers thermomechanical approach
(Ziegler 1983) and presented a thermodynamical background. They are
based on the several thermodynamical potentials such as Helmholtz free
energy and a dissipation function, Gibbs free energy function, and entropy
function. Collins and Houlsby (1997) obtained a new potential function
through Legendre transformations and then they derived a yield function
through a singular transformation for the case where the dissipation func-
tion is homogeneous of degree one in the rate of internal variables. Houlsby
and Puzrin (2000) presented constitutive models for geomaterials based on
a thermomechanical approach by Collins and Houlsby (1997).
From the frst law of thermodynamics Equations (1.138) and (1.150) and
the defnition of the free energy, the energy dissipation, D, in an isothermal
process is given by

ij ij
D

= + (3.121)
where

is a rate of free energy.

Following the Collins and Houlsby (1997), the internal dissipation func-
tion, D, is a rate at which energy dissipated.
Based on the work by Ziegler (1983), Collins and Houlsby (1997)
assumed that the dissipation function, D, arises from the internal variables
and corresponding stresslike variables

ij as:

D
ij ij
=

0 (3.122)
The dissipation function, D, corresponds to the mechanical dissipation by
Equation (1.149).
In the following, we will derive the yield function. Ziegler (1983)
assumed that the dissipation, D, is a function of the internal variable,
ij
,
and its time rate,

ij
.
When D is homogeneous function of degree one in

ij
, the dissipation
function can be described by the orthogonality condition; the difference
between

ij
and

D
ij

is orthogonal to

ij
, or if it is zero, we have

D
D
ij ij
ij
ij
= =

(3.123)

ij
ij ij
ij
D
=

( , )

(3.124)
where

ij
is a stress-like variable.
88 Computational modeling of multiphase geomaterials
Through Legendre transformation, we obtain the dual Legendre func-
tion, , as

D
ij ij ij ij ij ij
( , ) ( , )

+ = (3.125)
It is worth noting that in general, Legendre transformation is defned as
F p px f x
x
( ) max ( ( )) = ; p is a gradient of f(x)and max
x

indicates that maxi-
mization in x, and f(x) is convex downward.
When D is a homogeneous function of degree one in

ij
,

( , )
ij ij
= 0 (3.126)
In addition from Equation (3.123), Equation (3.125), and Equation(3.126),
we have,

d d d
ij
ij
ij
ij
=

0 (3.127)
and

dD d
D
d
D
d d d
ij
ij
ij
ij ij ij ij ij
+ =

= +

(3.128)
From Equation (3.124),

=
D
d d
ij
ij ij ij

(3.129)
From Equations (3.127) and (3.129), the following relations are derived:

ij
ij
=

(3.130)
and

D
ij ij

(3.131)
where

is a multiplier. This function corresponds to the yield function for

plasticity theory and Equation (3.130) is called a fow rule in plasticity theory.
Fundamental constitutive equations 89
REFERENCES
Armstrong, P.J., and Frederick, C.O. 1966. A mathematical representation of the
multiaxial Baushinger effect, C.E.G.B., Report RD/B/N 731.
Bingham, E.C. 1922. Fluidity and Plasticity, New York: McGraw-Hill.
Chaboche, J. I. 1977. Viscoplastic constitutive equations for the description of cyclic
and anisotropic behavior of material, bulletin de lAcademie Polonaise des
Sciences, Srie Sc. Et tech., 251(1):3342.
Chaboche, J.L. 1986. Time-independent constitutive theories for cyclic plasticity, Int.
J. Plasticity, 2(2):149188.
Chaboche, J.L., and Rousselier, G. 1983a. On the plastic and viscoplastic constitutive
equations, Part I: Rules developed with internal variable concept, J. Pressure
Vessel Tech., ASME, 105:153158.
Chaboche, J.L., and Rousselier, G. 1983b. On the plastic and viscoplastic constitu-
tive equations, Part II: Application of internal variable concepts to the 316
stainless steel, J. Pressure Vessel Tech., ASME, 105:159164.
Christensen, R.M. 1982. Theory of Viscoelasticity: An Introduction, 2nd ed., New
York: Academic Press.
Collins, I.F., and Houlsby, G.T. 1997. Application of thermomechanical principles to the
modeling of geotechnical materials, Proc. Roy. Soc. London A, 453:19752001.
Dafalias, Y.F., and Popov, E.P. 1976. Plastic internal variables formalism of cyclic
plasticity, J. Appl. Mech., 98:645651.
Darve, F., Laouafa, F., and Servant, G. 2001. Mode of rupture in geomaterials,
Proceedings of International Workshop on Deformation of Earth Materials,
TC34 of ISSMGE, May 23, 2001, Sendai, Japan, 3950.
Drucker, D.C. 1951. A more fundamental approach to plastic stress-strain relations,
Proc. First U.S. Natl. Congr. Appl. Mech., 487491.
Duvaut, G., and Lions, J.L. 1976. Inequalities in Mechanics and Physics, Springer-
Verlag, Berlin (Les inequations en mechanique et en physique, Paris: Dunod,
1972).
Eringen, A.C. 1967. Mechanics of Continua, New York: John Wiley & Sons.
Hill, R. 1948. A variational principle of maximum plastic work in classical plasticity,
Q. J. Mech. Appl. Math., 1:1828.
Hill, R. 1957. On uniqueness and stability in the theory of fnite elastic strain, J. Mech.
Phys. Solids, 5:229241.
Hill, R. 1958. A general theory of uniqueness and stability in elastic-plastic solids,
J. Mech. Phys. Solids, 6:236249.
Hohenemser, K., and Prager, W. 1932. ber die anstze der mechanik isotroper kon-
tinua, ZAMM, 12:216226.
Houlsby, G.T., and Puzrin, A.M. 2000. A mechanical framework for constitu-
tive models for rate independent dissipative materials, Int. J. Plasticity,
16:10171047.
Ionescu, I.R., and Sofonea, M. 1993. Functional and Numerical Methods in
Viscoplasticity, Oxford University Press.
Lubliner, J. 1990. Plasticity Theory, New York: Macmillan.
Kremple, E. 1986. Viscoplasticity based on overstress with a differential growth law
for the equilibrium stress, Mech. Mater., 5:3548.
90 Computational modeling of multiphase geomaterials
Lyapunov, A.M. 1907. Problme gnral de la stabilit des mouvements, Ann. Fac.
Sci. Toulouse, 9:203274.
Malvern, L.E. 1969. Introduction to the Mechanics of a Continuous Media, New
York: Prentice-Hall.
Mandel, J. 1964. Conditions de Stabilite et Postulat de Drucker, In Proceedings of
IUTAM Symposium on Rheology and Soil Mechanics, J. Kravtchenko, ed.,
Springer-Verlag, 5868.
Martin, J.B. 1927. Plasticity: Fundamentals and General Results, Cambridge, MA:
The MIT Press, 1975.
Masing, G., Veroff Siemens-Konzen, III Band.
Maugin, G.A. 1992. The Thermodynamics of Plasticity and Fracture, Cambridge
University Press.
Mroz, Z. 1981. On generalized kinematic hardening rule with memory of maximal
prestress, Journal de Mcanique Applique, 5:241259.
Naghdi, P.M., and Murch, S.A. 1963. On the mechanical behavior of viscoelastic/
plastic solids, J. Appl. Mech., 321328.
Nakamura, T., and Uetani, K. 1981.Critical behaviours of elasto-plastic struc-
tures, Review paper, Materials, Japanese Soc. Mater. Sci., 30(333):535548
(InJapanese).
Nova, R. 1994. Controllability of the incremental response of soil specimens sub-
jected to arbitrary loading programs. J. Mech. Behav. Mater., 5(2):193201.
Oka, F. 1985. Elasto/viscoplastic constitutive equations with memory and internal
variables, Comput. Geotech., 1:5969.
Perzyna, P. 1963. The constitutive equations for work-hardening and rate sensitive
plastic materials, Proc. Vibrational Probl., Warsaw, 4(3):281290.
Perzyna, P. 1971. Memory effects and internal changes of a material, Int. J. Non-
linear Mech., 6:707746.
Phillips, A., and Wu, H.-C. 1973. A theory of viscoplasticity, Int. J. Solids Struct.,
9:1530.
Prager, W. 1949. Recent developments in the mathematical theory of plasticity, J. Appl.
Phys., 20:235241.
Prager, W. 1961. Introduction to Mechanics of Continua, Boston: Ginn and Company.
Simo, J.C., and Hughes, T.J.R. 1998. Computational Inelasticity (Interdisciplinary
Applied Mathematics), New York: Springer. (Corrected 2nd printing, 2006)
Tateishi, T., and Miyamoto, H. 1974. On the yield conditions of viscoelastic body,
J. JSME, 40(339):30053016 (In Japanese).
Valanis, K.C. 1971. A theory of viscoplasticity without a yield surface, Arch. Mech.
Stos., 23(4):517533.
von Mises, R. 1928. Mechanik der plastischen formanderung von kristallen,
Zeitschrift fr Angewandte Mathematik und Mechanik, 8:161185.
Watanabe, O., and Alturi, S.N. 1986. Internal time, general internal variable, and
multi-yield-surface theories of plasticity and creep: a unifcation of concepts,
Int. J. Plasticity, 2:107134.
Ziegler, H. 1959. A modifcation of Pragers hardening rule, Q. Appl. Math.,
17:5565.
Ziegler, H. 1983. An Introduction to Thermomechanics, 2nd ed., North-Holland.
91
Chapter 4
Failure conditions and
the Cam-clay model
4.1 INTRODUCTION
Various constitutive models for soil have been developed over the last four
decades. In particular, many elastoplastic models have been proposed since
Cam-clay models were established by Roscoe et al. (1963; Roscoe and
Burland 1968). The constitutive model for soil should be able to describe
all types of soil behavior. The behavior of soil is complex, however, due to
its nature, that is, its granularity, its multiphase structure, and its inhomo-
geneity. The typical characteristics of soil can be listed as follows:
1. Multiphase mixture of soil particles, pore water and pore air, satu-
rated and unsaturated soil, and effective stress
2. Elasticity and hypoelasticity
3. Plasticity, hypoplasticity, and dilatancy characteristics
4. Rate sensitivity, viscoelasticity and viscoplasticity
5. Density dependency and confning pressure dependency
6. Strain-hardening and strain-softening characteristics
7. Cyclic deformation characteristics
8. Structural and induced anisotropy
9. Noncoaxiality
10. Deformation localization, bifurcation, and instability
11. Discontinuity
12. Degradation and the growth of microstructures
13. Inhomogeneity and nonlocality
14. Temperature dependency
15. Electromagnetic characteristics, the dieletric constant, and con du ctivity
Some of the characteristics have been included in the formulation of the
constitutive models. In particular, elastoplasticity and dilatancy are now
included in almost all of the models. However, some of them are not well
incorporated into the models.
92 Computational modeling of multiphase geomaterials
In this chapter, we begin with failure criterion and the Cam-clay model.
In Chapters 5 and 9 the characteristics of soil, such as rate dependency,
structural degradation, and cyclic plasticity will be discussed with regard
to constitutive modeling and its use in numerical analyses.
4.2 FAILURE CRITERIA FOR SOILS
When we apply shear stress to soil, the shear stress increases, exceeds
the elastic limit (initial yield stress), and then reaches the peak stress, as
shown in Figure4.1. The state at the peak stress is called the failure state.
Thereafter, when strain has been successively applied to the soil, thestress
decreases and reaches the stress state at a large strain, which may be called
the residual strength. The behavior of soil is strongly affected by the confn-
ing pressure and the stress components. Based on the many experimental
works and measurements in failure tests, several failure hypotheses have
been proposed. The following hypotheses are representative of them.
4.2.1 Failure criterion by Coulomb
Leonardo da Vinci found that the friction between the contact areas of two
bodies does not depend on the contact areas themselves but on the normal
stress applied to the contact areas. Thereafter, Amonton (1699) re-found it.
O

Elastic limit
Unloading
Failure

v
Figure4.1 Schematic stressstrain relations for inelastic geomaterials.
Failure conditions and the Cam-clay model 93
Following these studies, Charles-Augustin de Coulomb (1773) proposed a
failure criterion, namely,

c
N
tan = + (4.1)
where is the shear stress,
N
is the normal stress, c is the cohesive strength
that is independent of the normal stress, and is the frictional angle.
4.2.2 Failure criterion by Tresca
Trescas criterion (1884) is called the maximum shear stress hypothesis

c
2
1 3
=

= (4.2)
where c is the maximum shear strength.
When we consider the effect of the confning pressure, Equation (4.2)
becomes the extended Trescas criterion, namely,

= c
m
2
1 3
(4.3)
where c is a constant and
m
is the mean stress. For saturated soil,
m
is
replaced by the mean effective stress.
4.2.3 Failure criterion by von Mises
The von Mises (1913) yield criterion is given by

c ( ) ( ) ( )
1 3
2
2 3
2
1 2
2
+ + = (4.4)
where c is a material constant.
Elastic shear energy, W, is given by

W
G
1
12
( ) ( ) ( )
1 3
2
2 3
2
1 2
2
= + + (4.5)
where G is the elastic shear modulus.
Equation (4.5) indicates that failure occurs when the elastic energy
reaches the specifc value of the material. In addition, the features of the
von Mises criterion include the intermediate principal stress.
The extended von Mises criterion considers the effect of the mean effec-
tive stress and is given by

c
m
( ) ( ) ( )
1 2
2
2 3
2
3 1
2
+ + = (4.6)
where c is a material constant.
94 Computational modeling of multiphase geomaterials
4.2.4 Failure criterion by Mohr
Otto Mohr (1900) assumed that the shear strength, , of the mobilized slip
plane is a function of the normal stress perpendicular to the slip plane, namely,

F
N
( ) = (4.7)
Then, the shear strength criterion is given by the envelope of Mohrs circles
at failure since the stress point on the slip plane is on Mohrs circle at failure.
4.2.5 MohrCoulomb failure criterion
When the envelope in Mohr failure criterion is expressed by a straight line, the
criterion is called MohrCoulomb failure criterion. The criterion is given by

c
2 2
sin cos
1 3 1 3

=
+
+ (4.8)
where
1
,
3
are the maximum and the minimum principal stresses, and c
is the cohesive strength.
In cases where the cohesive strength can be disregarded, Equation(4.8)
becomes

1 sin
1 sin
1
3

=
+

(4.9)
where is the angle of the internal friction.
In MohrCoulomb failure criterion, the intermediate principal stress is
not taken into account but has been widely applied to soil.
4.2.6 MatsuokaNakai failure criterion
Matsuoka and Nakai (1974) generalized MohrCoulomb criterion consid-
ering the intermediate stress as

c
3
1
1
3
2
3
3
2
2
1
1
2

= (4.10)
where c is a material constant.
Using the stress invariants, Equation (4.10) can be written as

I I
I
c 3
1 2
3
= (4.11)
where I
1
, I
2
, I
3
are the invariants of the stress tensor (Equations 1.67, 1.68,
and 1.69) as

I I I .
1 1 2 3 2 1 2 2 3 3 1 3 1 2 3
= + + , = + + , =
Failure conditions and the Cam-clay model 95
When we can ignore the cohesive strength, Equation (4.11) can be rewrit-
ten as

I I
I
(9 8tan ) 0
1 2
3
2
+ = (4.12)
where is the angle of the internal friction.
4.2.7 Lade failure criterion
Lade (1977) proposed a failure criterion for geomaterials using the frst and
the third stress invariants as

I
I
const. (1977)
1
3
3
= (4.13)
Then, Lade and Kim (1988) generalized it as

I
I
I
p
const
a
m
27 . (2003)
1
3
3
1
= (4.14)
where p
a
is the atmospheric pressure.
4.2.8 Failure criterion on plane
Let us illustrate the shape of the failure criterion on the plane. Considering
the coordinates on the plane shown in Figure 4.2 and the fact that the
z axis is perpendicular to the plane, the following relation holds between
two coordinates.
The angle between the hydrostatic axis (the space diagonal) and any of
the coordinates of the principal stress space (
1
,
2
,
3
) is cos (1/ 3)
0
1
=

,
as shown in Figure4.2, and the plane perpendicular to the hydrostatic
axis is called the plane (or the octahedral plane).

x
y
z
0 2 6 1 3
1 2 1 6 1 3
1 2 1 6 1 3
1
2
3

=
/ /
/ / /
/ / /
,

x
y
z
0 1 2 1 2
2 6 1 6 1 6
1 3 1 3 1 3
1
2
3
=
/ /
/ / /
/ / /

(4.15)
96 Computational modeling of multiphase geomaterials
From Equation (4.15), the distance between the point (x,y) and the origin
(0,0) is given by x y
2 2
+ as

R x y
1
3
( ) ( ) ( )
2 2
1 2
2
2 3
2
3 1
2
= + = + + (4.16)
On the other hand, the component of the stress vector by the projection
of the stress vector to the hydrostatic axis (z axis),
oct
(octahedral normal
stress), and the components parallel to the plane,
oct
(octahedral shear
stress), are given as

oct m
3
1 2 3
=
+ +
= (4.17)

oct
1
3
( ) ( ) ( )
1 2
2
2 3
2
3 1
2
= + + (4.18)
Let us determine the shape of MohrCoulombs failure criterion on the
plane in the case of c = 0.
If we set
1
as the maximal stress and
3
as the minimal stress,

( )sin
1 3 1 3
= + (4.19)
O
A
B
plane
BA = OA OB
= (
1
,
2
,
3
) (p, p, p) = (s
1
, s
2
, s
3
)

0
0

1
z
Figure4.2 plane.
Failure conditions and the Cam-clay model 97
Using Equation (4.13), we have

y x 3 6 2
1 3
= / / (4.20)

x y z 2 6 2 3
1 3
+ = / + / + / (4.21)
Hence, from Equation (4.19),

y ax z
2 2 sin
3 sin
= +


(4.22)

a
3(1 sin )
3 sin
=
+

(4.23)
The magnitude of a is

a
1
3
3 (4.24)
As shown in Figure 4.3, the slope of the failure line through point P
is larger than that of line PQ, which is
1
3
parallel to the
2
axis on the
z plane. Consequently, the shape of the failure conditions is deformative
hexagonal.
In Figure 4.4, four failure criteria are shown on the plane. The
relationship between Tresca and von Mises criteria corresponds to the
relationship between MohrCoulomb and MatsuokaNakai criteria.
Figure4.5 illustrates the failure criteria in the principal stress space.
P
P
Q
O

1

3

2

1

2

3

3

1

2

3

2

1

2

3

1

2

2

3
Figure4.3 MohrCoulomb failure criterion on plane.
98 Computational modeling of multiphase geomaterials
4.2.9 Lode angle and MohrCoulomb
failure condition
In the following, the invariants of the deviatoric stress tensor are used as

J s s s s
ii
0
1 1 2 3
= = + + = (4.25)

J s s s s s s s s s s s
ij ij
1
2
1
2
( )
2 1
2
2
2
3
2
1 2 2 3 3 1
( ) = = + + = + + (4.26)

J s s s s s s s
ij jk ki ij
1
3
det[ ]
3 1 2 3
= = = (4.27)
where s
i
is the principal value of deviatoric stress tensor s
ij
.
von Misess criterion
Trescas criterion

3
(a)
MohrCoulomb
criteria
MatsuokaNakais
criteria

2
3
(b)
Figure4.4 (a), (b) Failure criteria on plane.
Failure conditions and the Cam-clay model 99
Expressing the mean stress by p, the coordinates of point B in Figure4.2
are (p, p, p).
Let an arbitrary point in the principal stress space be (
1
,
2
,
3
). The
vector BA is given by

p p p s s s ( , , ) ( , , )
1 2 3 1 2 3
= (4.28)
where p =
m
.
The magnitude of vector OB is given by the inner product as

p ( , , )
1
3
,
1
3
,
1
3
3
1 2 3
= (4.29)
Hence, the magnitude of vector BA is given by R in Equation (4.14) as

J R
1
3
( ) ( ) ( ) 2
1 2
2
2 3
2
3 1
2
2
+ + = = (4.30)
Since the unit vector on the y axis is expressed by

n
y
1
6
(2, 1, 1) = (4.31)
Extended MohrCoulomb
criteria
MatsuokaNakais
criteria

1
Figure4.5 Extended MohrCoulomb and MatsuokaNakai failure criteria in the prin-
cipal stress space.
100 Computational modeling of multiphase geomaterials
OC in Figure4.6 is given by

OC s s s R ( , , )
1
6
(2, 1, 1) cos
1 2 3
= = (4.32)
Consequently,

R s s s cos
1
6
(2 )
1 2 3
= (4.33)
In view of s s s 0
1 2 3
+ + = , R s cos 3/2
1
= . Since R J 2
2
= ,
s
J
cos
3
2
2
1
2
= .
Using the formula for trigonometrical functions, cos3 4cos 3cos
3
= ,
we obtain

J
s s J cos3
3 3
2
1
2
3/2
1
3
1 2
( ) = (4.34)
In view of J s s s s s s ( )
2 1 2 2 3 3 1
= + + , Equation (4.34) becomes

J
J
cos3
3 3
2
3
2
3/2
= (4.35)
where is called Lodes angle that indicates the angle on the plane.
A
x
y
O
C
n
y
z
plane

Figure4.6 plane.
Failure conditions and the Cam-clay model 101
For example, 0
3


for the region
1 2 3
. Lodes angle can be
expressed by the principal stresses, , ,
1 2 3
,
1 2 3
, as

tan
3( )
2
2 3
1 2 3
=


(4.36)
or

b b b
b b
cos3
(2 )(1 2 )(1 )
2( 1)
2 3/2
=
+
+
(4.37)
where b
2 3
1 3
=


,
1 2 3
. b = 0 corresponds to the triaxial compres-
sion conditions and b = 1 corresponds to the triaxial extension conditions.
Using Lodes angle, MohrCoulombs failure criterion is expressed in the
following. Using the xyz coordinate system stress, the point on the plane
is expressed by

x R
y R
z p
sin
cos
3
=
=
=
(4.38)
where R J 2
2
= .
Using the expression in Equation (4.15), we can write down the principal
stresses for the case
1 2 3
as

R p
2
6
cos
1
= + (4.39)

R R p
1
2
sin
1
6
cos
2
= +

R R p
1
2
sin
1
6
cos
3
= +
Substituting Equation (4.39) into Equation (4.19), we have a Mohr
Coulomb failure criterion expressed by the stress invariants and Lodes
angle as

p
R
c sin
2 6
(3 sin )cos 3(1 sin )sin cos 0
( )
+ + + = (4.40)
where c is the cohesion and is the internal friction angle.
102 Computational modeling of multiphase geomaterials
When c = 0,

J
p
2
6 2 sin
3(1 sin )sin 3(3 sin )cos
2
=

+ +
(4.41)
4.3 CAM-CLAY MODEL
As has been previously described, the features of materials such as
mechanical, thermal, and electromagnetic characters are expressed by
constitutive equations. When we apply loads to these materials, the mate-
rials deform. If the strain induced by the loading is small, the materials
behave elastically, that is, the deformation can be recovered. However,
part of the deformation cannot be recovered if the strain level is high as
shown in Figure4.1. This type of behavior is elastoplastic, and the model
that can describe such elastoplastic behavior is called an elastoplastic con-
stitutive equation.
In addition, if the model can account for time-dependent behavior, such
as creep, it is called an elastoviscoplastic model. Herein, we will disregard
the time-dependent behavior.
4.3.1 Original Cam-clay model
Roscoe and the research group at Cambridge University (1963) developed
an elastoplastic constitutive model for clay, which is called the Cam-clay
model after the name of the river fowing through the Cambridge University
campus. In the derivation of the Cam-clay model, the following assump-
tions have been used with the concept of critical state:
1. Existence of yield function
2. Drucker type of elastoplastic theory; fow rule and normality rule for
strain rate
3. Internal dissipation (or stressdilatancy relation)
4. Hardening rule; plastic void ratio e
p
is used as a hardening parameter
According to the work by Roscoe et al. (1963), we will consider the axi-
symmetric stress conditions and use the following quantities and symbols:

i
: effective stress tensor, 1 2 3
2 3
i u
i i w
= , , , = , =
u
w
: pore water pressure

i
: principal stress
q: deviator stress, q
1 3 1 3
= , >
p: mean total stress, p ( 2 ) 3
1 3
= + /
Failure conditions and the Cam-clay model 103
p: mean effective stress, p ( 2 ) 3
1 3
= + /
M: slope of the critical state line on pq plane or value of q/p at critical state
Under triaxial compression conditions,
1 2 3
> =
The stress ratio at the critical state, that is, the failure state, is

q p
( 2 ) 3
3( 1)
( 2)
1 3
1 3
1 3
1 3
/ =

+ /
=
/
/ +
(4.42)
If we set the internal friction angle =, (1 sin ) (1 sin )
1 3
/ = + / . Then,

M q p ( )
6sin
3 sin
= / =


(4.43)
Under triaxial extension conditions,
1 2 3
= > , then

q p
(2 ) 3
1 3
1 3
/ =

+ /
(4.44)
and

M
6sin
3 sin
=

+
(4.45)
where is the internal friction angle.

d d d d
v
p
v
p p p
: plastic volumetric strain increment, 2
11 33
= + (4.46)

d d d d d d
d
p
d
p p
v
p p p
: plastic shear strain increment,
1
3
2
3
11 11 33
( ) = =

(4.47)
The yield function of the Cam-clay model is derived in the following.
From the plastic fow rule in Equation (3.64),

d h
f
p
v
p
=

(4.48)

d h
f
q
d
p
=

(4.49)

d
d
v
p
d
p
f
p
f
q

(4.50)
104 Computational modeling of multiphase geomaterials
From the normality of the strain increment vector to the yield surface
given by assumption 2, as Equations (4.48) and (4.49),

d d dp dq
v
p
d
p
( ) 0 ( ) , , = (4.51)
where denotes the inner product.
Then,

=
d
d
dq
dp
v
p
d
p
(4.52)
From Equations (4.50) and (4.52), we have

dq
dp
f
p
f
q
=

(4.53)
Integrating Equation (4.53) provides the yield surface.
The internal dissipation energy increment, dW
p
, has been assumed to be
given by

dW d d d p d qd Mp d
p
ij ij
p p p
v
p
d
p
d
p
2
11 11 33
33
= = + = + = (4.54)
This dissipation function is related to the dissipation function shown in
Chapter 3.
By rewriting Equation (4.54), we obtain the following stressdilatancy
relation:

d
d
M
q
p
v
p
d
p

= (4.55)
Integrating Equation (4.50) with Equations (4.53) and (4.55) leads to a
yield function as

f
q
Mp
ln p p ln p p
y
( ) ( ) 0
0 0
= + / / = (4.56)
where p
y
is the hardening parameter and p
0
is the initial mean effective stress.
It is worth noting that deviator stress, q, corresponds to plastic deviatoric
strain,
d
p
, under axisymmetric stress conditions, as shown in the defnition
for plastic work increment, dW
p
. This plastic work increment corresponds
to the dissipation function shown in Chapter 3.
The yield function is illustrated in the (p,q) space shown in Figure4.7.
Failure conditions and the Cam-clay model 105
When we assume that the hardening parameter is given by plastic void
ratio, e
p
, and its evolution equation is given by Equation (4.57), parameter
h in the fow rule is obtained.

de
dp
p
p
y
y
( ) = (4.57)
where is the compression index, the slope of the compression curve
(elnp); is the swelling index, the slope of the swelling curve (elnp);
and ln indicates the natural logarithm.
The relationship between the plastic volumetric strain and the plastic
specifc volume (v
p
) is expressed by

d
dv
v
de
e
v
p
p p
1
= =
+
(4.58)
where v is the specifc volume (= 1 + e).
Upon substitution of Equation (4.57) into Equation (4.58) we obtain

=
+


dp
p
e
d
y
y
v
p
1
(4.59)
If we assume that the initial plastic volumetric strain is zero at p p
y 0
= ,
the yield function is rewritten as

f
q
Mp
ln p p
e
v
p
( )
1
0
0
= + /
+

= (4.60)
p
q
O
p
y
p
y
/e
f = 0
dq
dp
= 0
q = Mp
Figure4.7 Yield surface for Cam-clay model.
106 Computational modeling of multiphase geomaterials
Then,

f
p
q
Mp p
1
2

= + (4.61)

f
q Mp
1

= (4.62)

f e
v
p
1

=
+

(4.63)
Subsequently, using Pragers consistency condition, we have h as

h
dp dq dq M dp
M
f
p
f
q
f f
p
q
p
e
q
p
v
p
( )
( )
1
=

=
+

(4.64)
For the Cam-clay model, the strain increments under triaxial stress con-
ditions are given by

d
e Mp
dq M dp
v
p
(1 )
[ ( ) ] =

+
+ (4.65)

d
e Mp
dq M dp
M
q
p
(1 )
( )
=

+
+

(4.66)
where q p = / , d
v
p
is the plastic volumetric strain, and d
q
p
is the plastic
shear strain.
Let us obtain the state boundary surface using the Cam-clay model dis-
cussed earlier. Since the total volumetric strain increment is zero under
undrained conditions,

dq
dp
q
p
M
=


(4.67)
The elastic void ratio increment, de
e
, is given by

de
dp
p
e
= (4.68)
Then, the plastic volumetric increment, d
v
p
, is obtained by

d
de de
e e
de
dp
p
v
p
e
1
1
1
=

+
=
+
+ (4.69)
Failure conditions and the Cam-clay model 107
By substituting Equation (4.69) into the total differentiation of Equation
(4.60) and integrating it, we obtain

q
Mp
ln p p e e ln p p ( )
1
[ ( ) ]
0 0 0
+ / =

/ + (4.70)
where we set e e
0
= and p p
0
= when q/p = M.
If we set the specifc volume at v 1 + e = , when p kN m 1( )
2
= / ,
and q p M / = at the critical state, we have the following equation from
Equation(4.70):

v lnp
0 0
= + (4.71)
Then, we get the state boundary surface from Equation (4.70) as

q
Mp
v lnp [ ] =

+ (4.72)
The state boundary surface is shown in Figures4.8 and 4.9.
From the state boundary surface, we obtain the undrained stress path
under the undrained condition of e = const. and the yield surface by the
condition of
v
p
=

constant.
p
O
q
p
y
f = 0
p
O
ln p
O
p
B
B
C
= q/p
= M
C.S.L.
v = 1 + e
B
C
V
B
V
C
C.S.L.
v
0
C
B
V
B
V
C
C.S.L.
N.C.L.
v = 1 + e
Figure4.8 Yield surface and critical state line.
108 Computational modeling of multiphase geomaterials
From the void ratio constant condition, that is, the undrained condi-
tion, we have the undrained stress path. And from the plastic volumetric
strain constant condition,
v
p
=

constant, we obtain the yield surface in the
epq space.
It is worth noting that there is a unique relationship between the stress states
and the void ratio for the critical state from q p M / = and Equation(4.72). In
addition, if this unique relation is observed when we perform triaxial tests on
geomaterials, we can verify that the Cam-clay model concept is applicable to
geomaterials. For sand, Been and Jeffries (1985) found that the steady state
concept, which is equivalent to the critical state concept, holds and proposed
state parameter that controls dilatancy characteristics.
The projection of the state boundary surface on the p-v plane is

v lnp = (4.73)
and it is illustrated in Figure4.10.
From Equations (4.65) and (4.66), d
v
p
0 = at the critical state, q Mp = ,
and d
d
p
.
The normally consolidated line is given by v N lnp = ; it is parallel to
the critical state line in Equation (4.73), as shown in Figures4.10 and 4.11.
4.3.2 Ohtas theory
Ohta and Hata (1971) derived a change in void under general stress condi-
tions by the stressdilatancy relation (Shibata, 1963) and the elnp rela-
tion during the compression process.
q
p
v
State boundary surface
Elastic wall
Yield surface
q = Mp
Undrained stress path
Figure4.9 State surface.
Failure conditions and the Cam-clay model 109
The stressdilatancy relation is given by

+
=
de
e
d
q
p 1
0
(4.74)
where is a material constant and e
0
is the initial void ratio.
From isotropic compression tests, we have

de
dp
p
= (4.75)
ln p
v = 1 + e
C.S.L.
N.C.L.
p = 1 kN/m
2
N

Figure4.11 Critical state line and normally consolidated line.

q
p
v
v = ln p
(Critical state line)
v = N ln p
(Normal consolidation line)
Figure4.10 Normally consolidated line and critical state line.
110 Computational modeling of multiphase geomaterials
Adding two equations, we obtain

= + de
dp
p
e d
q
p
(1 )
0 (4.76)
Integrating the preceding equations on the normally consolidated curve,
under the conditions of p p e e q p 0
0 0
= , = , / = , produces

e e ln p p e
q
p
( ) (1 )
0 0 0
= / + + (4.77)
If we set

M e ( ) (1 )
0
/ = + , (4.78)
we obtain

e e ln p p M
q
p
( ) ( )
0 0
= / + / (4.79)
This equation is equal to the equation obtained by substituting Equation
(4.70) into Equation (4.60) and integration under the conditions of
p p e e q p 0
0 0
= , = , / = on the normally consolidated curve. Namely, the
preceding equation coincides with Equation (4.79).
Ohta and Hata (1971) obtained the yield surface by the projection of the
intersection of Equation (4.79) and the elastic wall e e ln p p ( ( ))
0 0
= / onto
the pq plane.
Consequently, Ohta and Hata (1971) derived an elastoplastic theory
equivalent to the Cam-clay model.
Equation (4.56) can be generalized under the general three-dimensional
stress conditions as

f
J
M
m
m
my
2
ln 0
2
*
=

= (4.80)
where J
2
is the second invariant of the deviatoric stress tensor,
m
is the
mean effective stress,
my
is the hardening parameter, and M
*
is a value
of J
m
2 /
2
at critical state.
4.3.3 Modified Cam-clay model
Using the Cam-clay model yields shear strain during isotropic compression
due to the shape of the yield surface. In order to modify this point, Roscoe
Failure conditions and the Cam-clay model 111
and Burland (1968) proposed an elliptic-shaped type of yield function by
generalizing the energy equation.
In the original Cam-clay theory, the increment of work dissipated per unit
bulk volume of an isotropic continuum during deformation was expressed as

W p q
p p
vol q
p
= + (4.81)
where p and q are the mean effective stress and the shear stress, respec-
tively, and
vol
and
q
are the corresponding strain increments associ-
ated with p and q.
In the derivation of the modifed Cam-clay theory, Roscoe and Burland
(1968) suggested that it was reasonable to assume that under an isotropic
stress of q = 0, there is no distortion (
q
= 0) in normally consolidated clay.
Since normally consolidated clay can deform irrecoverably under isotropic
stress, Equation (4.81) becomes

W p
dissipated
vol
p
( )
q 0
=
=
(4.82)
At the critical state condition, q/p = M and
vol p
0
,
= , so that
Equation(4.81) becomes

W Mp
dissipated q Mp q
p
( ) =
=
(4.83)
A generalization of the conditions in Equations (4.82) and (4.83) is

W p M
dissipated
vol
p
q
p
( ) ( )
2 2
= +
(4.84)
This is the expression for work dissipation used in the modifed Cam-
clay theory.
Employing Equation (4.84) and the normality condition (Roscoe and
Burland, 1968), the equation for the yield locus of the modifed Cam-clay
can be obtained as

f
p
p
M
M
y
y
0
2
2 2
=
+
= (4.85)
where is the value of the stress ratio (q/p) and M is the value when the stress
ratio reaches the critical state. p p
y
= at = 0 (isotropic normal compression).
Equation (4.85) represents an ellipse in the q, p space with its center at
p
y
/2 shown in Figure4.12.
From the fow rule, if we set q p = / , the stressdilatancy equation is
obtained as

d
d
M
p
p
d
p
2
2 2
(4.86)
112 Computational modeling of multiphase geomaterials
This equation indicates that the shear strain does not occur when the stress
ratio is zero, that is, = 0.
As is similar to Equation (4.80), Equation (4.85) can be generalized
under the general three-dimensional stress conditions as

f
M
J M
m
my
2
0
*2
2
*2
=
+

= (4.87)
The Cam-clay model has been used as a basic model for geomaterials.
Then, other types of inelastic models for describing plastic behaviors of
geomaterials have been developed over the last 20 years such as nonlinear
incremental model (Chambon 1986; Darve 1990). The hypoplastic model
is also a nonlinear incremental model (Kolymbas 1991; Gudehus 1996).
As for the cyclic plasticity theory, subloading surface models have been
developed by Hashiguchi and Ueno (1977), and Hashiguchi (1989) consid-
ering the expansion and contraction of loading surface. Oka et al. (1999)
developed a cyclic model based on the nonlinear kinematical hardening
theory. A rate-dependent model such as an elastoviscoplastic model will be
presented in Chapter 5.
4.3.4 Stressdilatancy relations
The stressdilatancy equations such as Equations (4.52) and (4.86) can
be brought back to Rows stressdilatancy theory. Row (1962) derived a
stressdilatancy relation by minimizing the ratio of the rate of work done
on an assembly of particles by major principal stress to the rate of work
done by the minor principal stress as

KD K D
d
d
v
p
P
; tan 45
2
, 1
1
3
2
1

= = + =

(4.88)
p
q
O
p
y
p
y
/2
Figure4.12 Yield surface for modifed Cam-clay model.
Failure conditions and the Cam-clay model 113
where is the interparticle friction angle, d
v
p
is the plastic volumetric
strain increment, d
p
1
is the plastic axial strain increment, and ,
1 3
is
the major and minor principal stress respectively.
Equations (4.52) and (4.58) are the variants of Rows relation and it has
been pointed out that the dilatancy is a function of fabric, density of pack-
ing and stress level, and the void ratio (Wan and Guo 1999).
REFERENCES
Amontons, G. 1699. Method of substituting the force of fre for horse and man power
to move machines, Histoire et Mmoires de lAcadmie Royale des Sciences.
Atkinson, J.H., and Bransby, P.L. 1981. The Mechanics of Soils, In An Introduction
to Critical State Soil Mechanics, London: McGraw-Hill.
Been, K., and Jeffries, M. 1985. A state parameter for sands, Gotechnique,
35(2):99112.
Chambon, R. 1986. Bifurcation en bande de cisaillement incrementalement non-
linearire, J. Mec. Theor. Appl., 5(2):277298.
Chen, W.F., and Mizuno, E. 1990. Nonlinear analysis in soil mechanics, Elsevier
Science Publishers B.V.
Coulomb, C.A. 1773. Essai sur application des regles des maiximas et minimis a
quelques problems de statique relatifs a larchitecture, Mem. Acad. Roy. Pres.
Divers Savants, 7:342382.
Darve, F. 1990. The expression of rheological laws in incremental form and the main
classes of constitutive equations, In Geomaterials Constitutive Equations and
Modelling, F. Darve, ed., Elsevier Appl. Science, London, 123148.
Darve, F., and Larouafa, F. 2000. Instabilities in granular materials and application
to land slides, Mech. Cohesive-Frictional Mater., 5(8):627652.
Drucker, D.C. 1951. A more fundamental approach to plastic stress-strain relations,
Proc. 1st U.S. Natl. Congr. Appl. Mech., 487.
Gudehus, G. 1996. A comprehensive constitutive equation for granular materials,
Soils and Foundations, 36(1):112.
Hashiguchi, K. 1989. Subloading surface model in unconventional plasticity,
Int.J.Solids Struct., 25:917945.
Hashiguchi, K., and Ueno, M. 1977. Elasto-plastic constitutive laws of granular
materials, In Constitutive Equations of Soils, S. Murayama and A.N. Schofeld,
eds., Proc. 9th Int. Conf. SMFE Spec. Session 9, Tokyo, JSSMFE, 7382.
Hill, R. 1957. On uniqueness and stability in the theory of fnite elastic strain,
J.Mech. Phys. Solids, 5:229241.
Hill, R.1958. A general theory of uniqueness and stability in elastic-plastic solids,
J.Mech. Phys. Solids, 6:239249.
Kolymbas, D. 1991. An outline of hypoplasticity, Arch. Appl. Mech., 61:143154.
Lade, P. 1977. Elasto-plastic stress-strain theory for cohesionless soil with curved
yield surfaces, Int. J. Solids Struct., 13:10191035.
Lade, P. V. and Kim, M. K. 1988. Single hardening constitutive model for fric-
tional mateials II. Yield criterion and plastic work contours, Computers and
Geotechnics, 6:1329.
114 Computational modeling of multiphase geomaterials
Lode, W. 1925. Versuche uber den Einfu der mittleren Hauptspannung auf die
Fliegrenze, ZAMM, 5:142-144.
Lubliner, J. 1990. Plasticity Theory, New York: Macmillan.
Malvern, L. 1969. Introduction to the Mechanics of a Continuous Medium,
Prentice-Hall.
Martin, J.B. 1975. Plasticity: Fundamentals and General Results, Cambridge, MA:
The MIT Press.
Matsuoka, H., and Nakai, T. 1974. Stress-deformation and strength characteristics
of soil under three different principal stresses, Proc. JSCE, 232:5970.
Mohr, O. 1900. Welche Umstande bedingen die Elastizitatsgrenze und den Bruch
eines Materials?, Zeitschrift des vereines deutscher Ingenieure, 44:15241530.
Muir Wood, D. 1991. Soil Behaviour and Critical State Soil Mechanics, Cambridge
University Press.
Nova, R. 1994. Controllability of the incremental response of soil specimens sub-
jected to arbitrary loading programmes, J. Mech. Behav. Mater., 5(2):193201.
Ohta, H., and Hata, S. 1971. A theoretical study of the stress-strain relation for
clays, Soils and Foundations, 11(3):4570.
Oka, F., Yashima, A., Tateishi, A., Taguchi, Y.. and Yamashita, S. 1999. A cyclic
elasto-plastic constitutive model for sand considering a plastic-strain depen-
dence of the shear modulus, Gotechnique, 49(5):661680.
Prager, W. 1961. Introduction to Mechanics of Continua, Boston: Ginn and Company.
Roscoe, K.H., and Burland, J.B. 1968. On the generalized stress-strain behaviour
of wet clay, in Engineering Plasticity, J. Heyman and F.A. Leckie, eds.,
Cambridge University Press, 535609.
Roscoe, K.H., Schofeld, A.N., and Wroth, C.P. 1963. Yielding of clays in states wet-
ter than critical, Gotechnique, 13(3):211240.
Row, P.W. 1962. The stress-dilatancy relation for static equilibrium of an assembly
of particle in contact, Proc. Royal Soc. London, 269(A):500527.
Schofeld, A.N., and Wroth, C.P. 1968. Critical State Soil Mechanics, London:
McGraw-Hill.
Shibata, T. 1963. On the dilatancy, Bulletin of the DPRI of Kyoto University, 6:
128134 (In Japanese).
Tresca, H. 1864. Memoir sur lecoulement des corps solides soumis a de fortes
pression, Comptes Rendus Acad. Sci. Paris, 59:754758.
von Mises, R. 1913. Mechanik der festen Korper im plastisch-deformablen Zustand,
nachrichten von der koniglichten Gesellschaft der Wissenschaft zu Gottingen,
582592.
Wan, R., and Guo, P.J. 1999. A pressure and density dependent dilatancy model for
granular materials, Soils and Foundations, 39(6):112.
115
Chapter 5
Elastoviscoplastic
modeling of soil
It is well known that the behavior of soil, in particular cohesive soil, is
affected by the loading time and the strain rate (e.g., Adachi et al. 1996).
Hence, it is necessary to include the viscous effect in the derivation of the
constitutive equations. Studies on the effect of the viscosity of soil have been
performed with viscoelastic or viscoplastic models. Over the last 30years,
viscoplastic models have been proposed following the development of elas-
toplastic models. In this chapter, we will discuss both the viscous nature
and the plastic nature of soil.
5.1 RATE-DEPENDENT AND TIME-DEPENDENT
BEHAVIOR OF SOIL
5.1.1 Strain rate-dependent behavior of clayey soil
Richardson and Whitman (1963) studied the effect of the strain rate on
remolded and reconstituted Mississippi River Valley clay by conducting
undrained triaxial compression tests. Figure 5.1 shows the results of tri-
axial compression tests that were done under a strain rate of about 1%/min
for high strain rate tests and under a strain rate of

2 10 %/ min
3
for low
strain rate tests. It can be seen that the stress ratio is higher in the case of
higher strain rates in the early stage of loading and that the decrease in
mean effective stress is larger in the case of lower strain rates. The slope of
the stress ratio-axial strain curve is larger in the case of higher strain rates.
Yong and Japp (1969) performed triaxial tests on water-saturated kaolin
under the high strain rate of 100%2000%/min, as shown in Figure5.2,
and they derived the following phenomenological relation:


, = , + / log ( ) ( ) ( )
0 0 0
(5.1)
where =
11 33
is the deviator stress, is the axial strain, and

is the
axial strain rate. Subscript 0 indicates the values at the referential state,
116 Computational modeling of multiphase geomaterials
0 1 2 3 4 5
Strain (%)
6 7 8 9 10
1
2
3

1
/

3
Fast test
Slow test
(a)
10 0
0
10
20
30
20 30 40 50 60
Slow test
Fast test
lb/in
2
= 6.9 kPa

1
+
3
2
(lb/in
2
) ( )

1

+

3
2
(
l
b
/
i
n
2
)
(
)
(b)
Figure 5.1 (a) Stressstrain relations with different strain rates. (b) Stress paths with
different strain rates. (After Richardson, A.M. Jr., and Whitman, R., 1963,
Gotechnique, 13(4):310324.)
Elastoviscoplastic modeling of soil 117
that is, the referential strain rate,
0
is the stress at a strain by the reference
strain rate, and is a parameter.
5.1.2 Creep deformation and failure
Creep deformation is a typical time-dependent behavior of soil. The succes-
sive deformation of soil, called creep, can be observed during constant stress
conditions. Finn and Shead (1973) carried out undrained creep tests for nor-
mally consolidated and overconsolidated clay in which the deviator stress
was kept constant, but the mean effective stress varied. Figure5.3 shows the
undrained triaxial creep test results for an isotropically consolidated speci-
men. Figure5.4 illustrates the deformationtime relationship during creep. In
the early stage of creep, the strain rate decreases; this is called the preliminary
creep process. Then, the strain rate becomes a constant minimum strain rate;
this is called the secondary creep process. Finally, the strain rate increases
with time; this is called acceleration or tertiary creep, and the soil reaches the
failure state. Finn and Shead (1973) found the following relationship between
the minimum strain rate and the failure time, as shown in Figure5.5:

= Timetofailure constant
min (5.2)
Saito and Uezawa (1969) found the following empirical relationship
based on many observed results:

= t t
f
( ) constant
(5.3)
= 3000%/sec
= 2000
= 1000
= 500
= 100

3
= 45 psi

3

(
p
s
i
)
0
0
5
10
15
20
25
30
35
Kaolinite-Water
Axial Strain (%)
5 10 15 20
Figure 5.2 Stressstrain relations for different strain rates. (After Yong, R.N., and Japp,
R.D., 1969, ASTM, STP, 450:233262.)
118 Computational modeling of multiphase geomaterials

1

3
= 43.7 psi
A
x
i
a
l

S
t
r
a
i
n

(
%
)
0
0
5
10
15
20
500 1000 1500
Time (min)
2000 2500 3000
(a)

1

3
= 43.7 psi
Normally consolidated clay test no. C-6
10
0
10
1
10
2
10
3
10
4
Time (min)
10
0
10
1
10
2
10
3
10
4
A
x
i
a
l

S
t
r
a
i
n

R
a
t
e

(
%
/
m
i
n
)
(b)
Figure 5.3 (a) Straintime profle during the creep. (b) Strain rate-time profle for the
creep test. (After Finn, L., and Shead, D., 1973, 8th ICSMFE, Moscow,
1(1):135142.)
Elastoviscoplastic modeling of soil 119
where t
f
is the failure time and

is the strain rate at time t during the

acceleration creep process. Figure5.6 is a measured example; it is seen that
Equation (5.3) is clearly effective.
Since it is diffcult to determine the time of the onset of acceleration creep
in practical cases, Equation (5.3) is proposed. It is interesting to note that
Equation (5.3) is similar to Equation (5.2).
Time (t)
S
t
r
a
i
n

(

)
O
A B C
Failure
A: Primary creep
B: Secondary creep
(Steady creep)
C: Tertiary creep
(Acceleration creep)
Figure 5.4 Schematic stressstrain relationship during the creep.
q = 51.3 psi
q = 42.8 psi
48.3 psi 46.3 psi 44.9 psi
43.7 psi
43.4 psi
Minimum strain rate line
log
10
t
m
= 0.142 1.15 log
10

m
10
0
10
1
10
2
10
3
10
4
Time (min)
10
5
10
4
10
3
10
2
10
1
10
0
A
x
i
a
l

S
t
r
a
i
n

R
a
t
e

(
%
/
m
i
n
)
Figure 5.5 Strain ratestime profle during the creep. (After Finn, L., and Shead, D.,
1973, 8th ICSMFE, Moscow, 1(1):135142.)
120 Computational modeling of multiphase geomaterials
5.1.3 Stress relaxation behavior
It is observed that the stress decreases when the strain is fxed. This behav-
ior is called the stress relaxation phenomenon. Figure5.7 indicates stress
relaxation during undrained triaxial compression tests (Akai et al. 1974).
In this fgure, it is seen that the stress decreases, but the pore water pressure
remains almost constant.
5.1.4 Strain rate-dependent compression
Figure5.8 shows the straintime relationship during consolidation tests on
samples with different thicknesses (Aboshi 1973). The strain does not con-
verge with time and it increases in proportion to log t. However, the strain
does converge during consolidation for elastic material. This type of behav-
ior is called secondary consolidation. It is a typical type of time-dependent
behavior for soil. Figure5.9 illustrates the results of one-dimensional con-
solidation tests with different strain rates. CRS tests, that is, constant rate
of strain tests, are shown in Figure5.9 (Sllfors 1975), in which the strain
rates (%/min) are as follows: C7-1(0.003), C7-2(0.006), C7-3A(0.012),
C7-4(0.020), C7-5(0.045), and C7-6(0.100). In the case of higher strain
log t
f
= 2.33 0.916 log ( 10
4
) 0.59
Dashed line indicates the region of 95% of measured results
[Experiments]
[From the reference]
Tonegawa loam
Kawasaki clay
Tonegawa silt
Fukiage clay
t
f
: Failure time from
the onset of secondary
(steady) creep
S. Murayama
A. Casagrande
M. Goldstein
10
7
10
0
10
1
C
r
e
e
p

F
a
i
l
u
r
e

T
i
m
e

t

f

(
m
i
n
)

10
2
10
3
10
4
10
5
10
6
10
5
10
4
Steady Strain Ratio (1/min)
10
3
10
2
10
1
Figure 5.6 Creep failure time versus strain rate at steady state. (After Saito, M., 1992,
Jissho Doshitsu Kogaku, Gihodo Syuppan Co. Ltd., 156.)
Elastoviscoplastic modeling of soil 121
10
1
0.0
0.5
1.0
1.5
2.0
10
0
10
1
10
2
10
3
10
4
Time t (min)
Axial strain
1
(%)
: 2.08
: 0.995
: 0.831
: 0.422
D
e
v
i
a
t
o
r

S
t
r
e
s
s

3

(
k
g
f
/
c
m
2
)
(a)
10
1
0.0
0.5
1.0
1.5
2.0
10
0
10
1
10
2
10
3
10
4
Time t (min)
Axial strain
1
(%)
: 2.08
: 0.995
: 0.831
: 0.442
P
o
r
e

W
a
t
e
r

P
r
e
s
s
u
r
e

u

(
k
g
f
/
c
m
2
)
(b)
Figure 5.7 (a) Deviator stresstime profle during the stress relaxation. (b) Pore
water pressuretime profle during the stress relaxation. (After Akai, K.,
Adachi,T., and Ando, N., 1974, Proc. JSCE, 225:5361.) ( 98 kPa).
122 Computational modeling of multiphase geomaterials
10
1
0
2
4
6
8
10
12
10
0
10
1
10
2
10
3
10
4
10
5
10
6
t (min)
V
e
r
t
i
c
a
l

S
t
r
a
i
n

v

(
%
)
Experimental results
H = 2 cm
20 cm
4.8 cm
40 cm
100 cm
100%
End of primary
consolidation
Figure 5.8 Settlementtime curves during the one-dimensional consolidation with dif-
ferent thickness. (After Aboshi, H., 1973, 8th ICSMFE, Moscow, 4(3):88.)
Efective Vertical Stress (kPa)
10
0
5
10
15
20
20 40 80 160
C74
C71
C72
C73A
C75
C76
320
C
o
m
p
r
e
s
s
i
v
e

S
t
r
a
i
n

(
%
)
Figure 5.9 Vertical effective stresscompression strain relations for different strain
rates. (After Sllfors, G., 1975, Preconsolidation pressure of soft high-plastic
clays, PhD thesis, Chalmers University of Technology, Gothenburg, Sweden.)
Elastoviscoplastic modeling of soil 123
rates, the strain is smaller at the same level of stress and the infection
point of the stressstrain curve, that is, the consolidation yield stress or
the compression yield stress is larger. From this fgure, we can observe the
time-dependency of the compression characteristics. Figure5.10 illustrates
the relationship between the preconsolidation pressure, that is, the consoli-
dation yield stress, and the strain rate by Leroueil et al. (1983). This fgure
clearly shows the strainrate dependency of the consolidation yield stress.
From earlier, it is concluded that clayey soil exhibits time-dependent behav-
ior such as rate sensitivity, creep, and relaxation.
5.1.5 Isotaches
The stressstrain characteristics, in which the stressstrain curves are the
same for the equi-void ratio and the equi-strain rates, are called isotaches
(Suklje 1957, 1969). Iso means equivalent and tache indicates the rate.
Isotaches include a unique relation among the stress, the strain, and the
strain rate, that is, the existence of stressstrainstrain rate relations.
Experimental results by Graham et al. (1983) and one-dimensional consoli-
dation test results with different strain rates shown by Leroueil et al. (1985)
are shown in Figures5.11 and 5.12, respectively. These results support the
concept of isotaches.
10
0
10
1
10
2
10
3
10
4
Strain Rate
v
(%/min)
Depth: 3.45~3.90 m
Core size: 70 mm diameter
Depth: 4.08~4.15 m
Core size: 300 mm diameter
In situ from the end of
construction to 1.5 years
10
5
10
6
0
20
40
60
80
100
120
C
o
n
s
o
l
i
d
a
t
i
o
n

Y
i
e
l
d

S
t
r
e
s
s

v
c

(
k
P
a
)
(
vc
) lab
(
vc
) feld
Figure 5.10 Preconsolidation pressurestrain rate relationship in laboratory and in situ.
(After Leroueil, S., Samson, L., and Bozozuk, M., 1983, Can. Geotech. J.,
20(3):477490.)
124 Computational modeling of multiphase geomaterials
Strain Rate
v
(s
1
)
P
r
e
c
o
n
s
o
l
i
d
a
t
i
o
n

P
r
e
s
s
u
r
e

p

(
k
P
a
)
10
8
160
140
120
100
80
60
10
7
10
6
: CRS test
: Creep test
Batiscan
10
5
10
4
(a)
Figure 5.12 (a) Variation in preconsolidation pressure with strain rate for Batiscan clay.
(b) Normalized effective stressstrain relationship deduced from CRS
odometer tests on Batiscan clay. (After Leroueil, S., Kabbaj, M., Tavenas, F.,
and Bouchard, R., 1985, Gotechnique, 36(2):288290.)
0.6
0.5
0.4
0.3
(

3
)
/
2

1
c
0.2
0.1
0.0
0 4 8 12
Axial Strain (%)
1%/h
R
R
R
5%/h
0.5%/h
0.05%/h
16%/h
0.25%/h
Belfast clay a depth of 4 m
1c
=
v0
Winipeg clay 11.5 m

1c
>
vc
CAU: Triaxial compression test
R: Stress relaxation test
16 20
Figure 5.11 Stressstrain curves for triaxial compression tests with step changed strain
rates and relaxation procedure. (After Graham, J., Crooks, J.H.A., and Bell,
A.L., 1983, Gotechnique, 33(3):327340.)
Elastoviscoplastic modeling of soil 125
5.2 VISCOELASTIC CONSTITUTIVE MODELS
The modeling of many kinds of materials has been carried out within the
framework of viscoelasticity for polymers, metals, concrete, soil, and rock.
The well-known linear models are the Maxwell model, the KelvinVoigt
model, and the springVoigt three-parameter model. It has been reported
that the linear springVoigt model can describe the dynamic nature of soil
(Kondner and Ho 1965; Hori 1974). By introducing the concept of the
distribution of relaxation time into the linear model, it is possible to model
the wide range of time-dependent behavior of soil. Murayama and Shibata
(1964) have proven the time-dependency of clay in high-frequency regions
by considering the distribution of relaxation time. Murayama (1983) pro-
posed nonlinear viscoelastic and viscoplastic models based on the original
model (Murayama and Shibata 1964).
Di Benedetto et al. (1997) proposed a simple asymptotic body (SAB) for
the simplifcation of the viscoelastic model for soil that can be classifed into
three-parameter models. In the range of small strain, the linear viscoelastic
approach is valid. In the range of large strain, however, the features include
both viscoelasticity and visoplasticity. Oka, Kodaka, and Kim (2004) have
succeeded in describing the behavior by a viscoelasticviscoplastic model
for clay, which can explain the dynamic behavior of clay for a wide range
of strain levels.
Batiscan
CRS (14)
Creep (9)
Normalized Preconsolidation Pressure

p
(
v
)
0.4
0
4
8
12
16
20
24
1 1.7 3
S
t
r
a
i
n

v

(
%
)
(b)
Figure 5.12 (Continued)
126 Computational modeling of multiphase geomaterials
5.3 ELASTOVISCOPLASTIC CONSTITUTIVE MODELS
5.3.1 Overstress models
To describe both the viscous nature and the plastic nature of soil, viscoplas-
tic modeling is necessary. Perzyna (1963) proposed a viscoplastic theory that
generalizes the linear theory for the viscoplasticity theory by Hohenemser
and Prager (1932). Prager and Hohenemsers model, a linearly extended
viscoplastic model, is based on Binghams fuid and plasticity model.
Yong and Japp (1969) indicated the possibility of applying the vis-
coplasticity theory to the dynamic behavior of clay. Then, Adachi and
Okano (1974) frst proposed an elastoviscoplastic theory for clay based on
Perzynas theory and the original Cam-clay model (1963). They assumed
that the hardening parameter was axial strain. They showed that visco-
plasticity was an applicable theory to the rate-dependent behavior of water-
saturated clay. However, no quantitative description had been successively
given for the model. Okas (1981), Adachi and Okas (1982) newly pro-
posed elastoviscoplastic model was based on both the Perzyna model and
the Cam-clay model. It incorporates the assumption that in the stress state,
after the completion of consolidation, the soil has not yet reached the equi-
librium state but is still in a nonequilibrium state with the strain-hardening
parameter for the inelastic volumetric strain, although the inelastic void
ratio has been taken as a hardening parameter in the Cam-clay model. The
model is capable of describing the rate sensitivity, the creep, and the relax-
ation of cohesive soil, and in particular, the volumetric relaxation behavior
reported by Arulanandan et al. (1971).
The model is a rigorous combination of two theories, namely, the Cam-
clay model and Perzyna model. However, the model has a shortcoming;
namely, it cannot describe conventional accelerated creep behavior, that is,
creep failure. Professor S. Sture of the University of Colorado pointed out
this shortcoming at the International Workshop in Grenoble (Oka 1982).
Aubry et al. (1985) experimentally showed that the critical state line is not
rate sensitive. It can be understood that rate dependency fades out at the
critical state. Giving consideration to the rate independency at the criti-
cal state leads to the fact that the viscosity asymptotically becomes zero
when approaching the critical state. Following the above point, Adachi,
Oka, and Mimura (1987) constructed an improved viscoplastic model by
considering the variations in viscosity. The derived model is very capable
of describing creep failure, that is, accelerating creep behavior. The predic-
tion obtained through this model indicates that the drop in stress is rather
small in comparison to the experimental evidence for sensitive clay and
natural soil during strain softening. During the strain-softening behavior
of natural clay, it is observed that strain softening follows a rather large
decrease in mean effective stress. This indicates that the soil exhibits both
Elastoviscoplastic modeling of soil 127
shear and volumetric softening. To incorporate these features, a new model
has been developed considering the degradation of soil structures and rate
dependency by Kimoto (2002) and Kimoto and Oka (2005). This new vis-
coplastic model will be introduced in the following section.
Many other models have been proposed to describe the time-dependent
behavior of soil. For the overstress models, Dafalias (1982), Katona (1984),
Baladi and Rohani (1984), and Zienkiewicz et al. (1975) have proposed
elastoviscoplastic models within the framework of an overstress type of
theory. Another type of overstress model has been proposed by Duvaut and
Lions (1976). Although their model is a linear overstress type of model, the
DuvautLions model is advantageous in that the plasticity model can eas-
ily be transferred into a viscoplastic one using the projection rule. Phillips
and Wus (1973) model is a nonlinear viscoplastic model using a similar
projection technique to obtain the overstress. Sawada et al. (2001) pro-
posed a Cosserat viscoplasticity model for clay. Yin and Karstunen (2008)
proposed an anisotropic elastoviscoplastic model based on the modifed
Cam-clay model and Perzynas type of viscoplasticity for the analysis of a
clay foundation beneath an embankment.
5.3.2 Time-dependent model
Sekiguchi (1977) proposed an elastoviscoplastic model that clearly includes
real time. Sekiguchis model was originally proposed as a creep model that
included failure. Nova (1982), Dragon and Mroz (1979), and Matsui and
Abe (1985) derived time-dependent models that are called nonstationary
models. It should be pointed out that these models include time explicitly
and violate the principle of objectivity. Yin and Graham (1999) proposed
an elastoviscoplastic model based on the modifed Cam-clay model and
the fow surface. Kutter and Sathialingam (1992) proposed an elastovisco-
plastic model for clay with a reference surface and discussed various rate-
dependent properties and overconsolidation ratios, for which they used a
time-dependent hardening law. The aforementioned models are consistent
with the delayed compression concept by Bjerrum (1973) in which equi-time
lines are incorporated since the equi-time line includes the solution for sec-
ondary compression.
5.3.3 Viscoplastic models based on
the stress history tensor
Oka (1985) proposed a viscoplastic model with the stress history ten-
sor, which is based on the assumption that the state of materials depends
on the stress and the stress history. He assumed that the yield function
depends on the stress history tensor and not on just the current stress or
128 Computational modeling of multiphase geomaterials
the internal variables. The stress history tensor is given by the convolution
integral of the stress tensor with respect to the generalized time measure,
which is inherent to the materials. Oka and Adachi (1985) developed an
elastoviscoplastic model using a stress history tensor for the analysis of the
strain-softening behavior of soft rock and frozen sand (Adachi et al. 1990;
Okaetal. 1994), and generalized it as the viscoplastic model (Adachi and
Oka 1995; Adachi et al. 2003, 2005). This type of model can be applied
to the rate-independent behavior by adopting a special time measure for
defning the stress history tensor.
5.4 MICRORHEOLOGY MODELS FOR CLAY
The viscous behavior of clay has been analyzed in the feld of microrhe-
ology. Murayama and Shibata (1964) applied the rate process theory by
Eyring (1936) to clay and derived a rheological model. Then, Singh and
Mitchell (1968, 1969) and Mitchell et al. (1968) successfully described the
creep behavior of clay based on the rate process theory. Using the rate pro-
cess theory, an exponential type of nonlinear fow law, between the shear
force acting on each fow unit and the strain rate when the shear force is
found to be larger than the thermal energy, was created.
5.5 ADACHI AND OKAS VISCOPLASTIC MODEL
Oka (1981) and Adachi and Oka (1982) developed an elastoviscoplastic
constitutive model for clay based on the Cam-clay model and an overstress
type of viscoplastic theory (Perzyna 1963). The important assumption
taken in the derivation of the model is that at the end of consolidation, the
state of the clay does not reach the static equilibrium state, but is in a non-
equilibrium state. In the following, Terzaghis effective stress is used as

= u
ij ij w ij (5.4)
where
ij
is the effective stress and u
w
is the pore water pressure.
It is assumed that the strain rate tensor consists of elastic strain rate ten-
sor,

ij
e
, and viscoplastic strain rate tensor,
ij
vp
, such that


= +
ij ij
e
ij
vp
(5.5)
The elastic strain rate is given by a generalized Hooke type of law as


= +

+

G
s
e
ij
e
ij
m
m ij
1
2 3(1 )
(5.6)
Elastoviscoplastic modeling of soil 129
where s
ij
is the deviatoric stress tensor,

m is the mean effective stress, G is

the elastic shear modulus, and e is the void ratio; in the infnitesimal theory,
initial void ratio, e
0
, is used, is the swelling index, and the superimposed
dot denotes the time differentiation.
In Equation (5.6), the elastic shear modulus, G, is assumed to be propor-
tional to the square root of the mean effective stress and the initial void ratio
can be replaced by the current void ratio for the fnite deformation analysis.
As has been presented in Chapter 3, the viscoplastic fow rule by Perzyna
is given by

F
f
ij
vp
ij
( ) (5.7)

=
>

F
F F
F
( )
( ) : 0
0 : 0
(5.8)

=

F
f
s
s
(5.9)
where

ij
vp
is the viscoplastic strain rate tensor, is the viscosity parameter,
is a rate-sensitivity function, < > are Macaulays brackets, f is the loading func-
tion, F = 0 is a static yield function, and
s
is the strain-hardening parameter.
To construct a viscoplastic constitutive model for soil, we have to present
the yield function. Herein, we assume that the mechanical behavior of soil
at the static equilibrium state can be described by the original critical state
energy theory developed by Roscoe et al. (1963). According to the Cam-
clay model, the following yield function is assumed to be valid for isotropi-
cally consolidated clay as

=

+

= f
M
s
m
m
my
s
ln 0
*
* ( )
(5.10)
in which
my
s ( )
is the hardening parameter and M
m
*
is the value of

= = = J J s s
ij ij m ij ij
2 / ( /2)
* * *
2 2 (5.11)

= s
ij ij m
/
*
(5.12)
at the critical state.
f in Equation (5.7) is a function that depends on the viscoplastic strain
and the stress; it can be called a dynamic yield function and it takes the
same form as the static yield function.
130 Computational modeling of multiphase geomaterials
Figure5.13 is a schematic diagram of the static and the dynamic load-
ing (or yield) surface. In the fgure, P
i
d ( )
is a dynamic state at the end of
one day of isotropic consolidation under prescribed pressure,
myi
d ( )
. On the
other hand, P
i
s ( )

is the corresponding static state to P
i
d ( )
with the same strain
hardening, namely, in the same inelastic volumetric strain state, and it lies
on the static isotropic consolidation line that is attained by the infnite
time duration of isotropic consolidation. The state path, P P
i
d d ( ) ( )
, repre-
sents a shear deformation process with an inelastic strain rate. State path
P P
i
d
u
s ( ) ( )
shows the increase in pore water pressure when returning to the
undrained condition after the end of one day of consolidation, and path

P P
i
d s ( ) ( )
represents secondary consolidation (delayed compression).
Since the static yield function is a nondimensional function, we can use
(F) = (f
s
) in Equation (5.7). Based on the experimental results, Oka
(1981) and Adachi and Oka (1982) assumed the functional form as

=

f C M m
J
M
y m m
m m
m
my
s
( ) exp
2
ln
0
*
2
* ( )
(5.13)
In the following section, it will be explained why Equation (5.13) is effective.
The derivative of Equation (5.10), with respect to the stress tensor, is
given by

f
M
s
J
M
J
ij m m
ij
m
m
ij
1
2
2 1
3
*
2
* 2
(5.14)
2J
2
f
di
= k
di
f
si
= k
si
f
d
= k
d
f
s
= k
s
Dynamic stress path
Critical state line
Delayed compression
1 Day isotropic
consolidation line
Isotropic relaxation

(s)
myi

(d)
myi

(s)
my

(d)

m
my
v
vp
v
i
vp
p
i
(s)
p
i
(d)
p

(s)
p
u
(s)
p
(d)
Figure 5.13 Schematic diagram of the yield surface and the loading path.
Elastoviscoplastic modeling of soil 131
The introduction of Equations (5.13) and (5.14) into Equation (5.7) yields

C m
J
M
s
J
M
J
ij
vp
m m
m
my
ij
m
m
ij
exp
2
ln
2
2 1
3
0
2
*
2
* 2
(5.15)
In Equation (5.15), we omit superscript s for simplicity.
Although the plastic void ratio is adopted as a hardening parameter in
the Cam-clay model, the strain-hardening parameter,
my
, herein follows
the following hardening rule as

=

+

dv
e
d
p my
my
1
(5.16)
where dv
p
is the viscoplastic volumetric strain increment.
Integrating Equation (5.16) under the initial conditions of = v v
p
i
p
and
=
my myi
, we have

=

+

v v
e
p
i
p my
myi
1
ln (5.17)
Line I in Figure 5.14 shows this relation and how the strain-hardening
parameter,
my
s
,
( )
increases with an increase in inelastic volumetric strain, v
p
.
It is found that the strain-hardening parameter,
my
s
,
( )
namely, yield function
line - I
ln
m

myi
(d)

myi
(s)
p
i
(s)
p
i
(d)
p
(s)
v
vp
v
vp
v
p
v
i
p
=
v
i
vp

my
(s)

myi
(s)

my
(s)

1 + e
ln
/
Figure 5.14 Volumetric hardening rule.
132 Computational modeling of multiphase geomaterials
F, can be determined only if the values of v
p
and

myi
s ( )
are given. In prin-
ciple, it is impossible to obtain the value of
my
s ( )
, since line I, which is the
infnite time isotropic consolidation line, cannot be obtained. Fortunately,
however, it is not necessary to know the exact value of
myi
s ( )
.
Using the consolidation stress at the end of consolidation,
me
, we have

+
+


e
v v
my
me
myi
me
p
i
p
ln ln
1
( ) (5.18)
Then, if = v
i
p
0, Equation (5.15) becomes

+

+

C m
J
M
e
v
s
J
M
J
ij
vp
m m
m
me
p ij
m
m
ij
exp
2
ln
1
2
2 1
3
2
*
2
* 2
(5.19)

=

C C m
myi
me
exp ln
0 (5.20)
When the stress ratio is zero, namely, at the end of the isotropic con-
solidation, from Equation (5.19), the viscoplastic volumetric strain rate is
given by

+

C m
e
v M
kk
vp m
me
p
m
exp ln
1
*
(5.21)
Let us explain viscoplastic parameters m and C. All material param-
eters can be determined from the results of consolidation, swelling, and
strain-rate controlled undrained triaxial compression tests. Secondary con-
solidation is the well-known time-dependent behavior of clay, and we will
examine the interrelation between secondary compression rate parameter
and strain rate parameter m in the derived constitutive equations. Under
isotropic consolidation, the inelastic volumetric strain rate (secondary
consolidation rate) is expressed by Equation (5.21) under the condition of
=
m me
(= constant) as

=
+

v CM m
e
v
p
m
p
exp
1
*
(5.22)
Elastoviscoplastic modeling of soil 133
If we set

=

+e m (1 )
(5.23)
then

v CM v
p
m
p
exp
1
*
(5.24)
When we assume that C is constant, Equation (5.24) can be integrated
under the condition of = v v
p p
0
at = t t
0
as

= + v t t v
p p
ln( / )
0
0
(5.25)
and

= v CM t
p
m
ln{ /( )}
0
*
0
(5.26)
From this explanation, m is determined by a secondary compression
rate, , and the compression and swelling indices. On the other hand, the
viscoplastic parameter, C, is related to the initial viscoplastic volumetric
strain rate as


= = v CM
vp p
m
0 0
*
(5.27)
Using secondary compression coeffcient = + C e C t t e
a
p
a
p
( log( / ) )
0
0
,

+
=

C
e 2.303(1 )
(5.28)
since

= =
+
v
e
e kk
p p
p
1
.
Finally, we can obtain

=

m
C C
C
c s
(5.29)
where C
c
and C
s
are the compression index and the swelling index in terms
of the void ratio. During secondary compression, the void ratio changes
viscoplastically under constant stress.
134 Computational modeling of multiphase geomaterials
When we assume that ratio k between C
c
and C
s
is constant, we have

=

m
k C
C
c
(1 )
(5.30)
Mesri et al. (1995) experimentally found that

C C
c
/ is a material con-
stant for geomaterials. It is worth noting that, based on Mesris work, we
could say that m is a material constant. Leroueil and Hight (2003, p. 126)
have shown that, based on Mesris work,

C C
c
/ is between 17 and 35.
For these values, m is larger for inorganic clay than for organic clay. For
example, Mesri et al. (1995) have experimentally obtained experimental
evidence for ratio =

C C
c
/ 0.03 for Batiscan clay and Saint-Hilaire clay.
Leroueil and Hight indicated that

C C
c
/ remains in the narrow range as
shown in Table5.1.
The other method for determining m, based on undrained triaxial
tests with different strain rates, will be explained in the following section.
Viscoplastic parameter m can be determined from undrained triaxial com-
pression tests with different strain rates.
Since the volume change is negligible under undrained conditions, that
is, =
kk
0, we obtain the next relationship between mean effective stress,

m
, and inelastic volumetric strain, v
p
, from Equation (5.6) as
d
e
d dv
kk m
p
1
0 =

+
+ = (5.31)
Integrating Equation (5.31) under the condition of = v v
p
i
p
at =
m me

results in
=

+
+ v
e
v
p
m me
i
p
1
ln( / )
(5.32)
where
me
denotes the stress state at the end of consolidation.
Table5.1 Values of C

/C
c
for several geomaterials
Material C

/C
c
Granular soils including rockfll 0.02 0.01
Shale and mudstone 0.03 0.01
Inorganic clays and silts 0.04 0.01
Organic clays and silts 0.05 0.01
Fibrous and amorphous peats 0.06 0.01
Source:

Leroueil, S., and Hight, D.W., 2003, Behaviour and
properties of natural soils and soft rocks, In Characterization
and Engineering Properties of Natural Soils, T.S. Tan, K.K. Phoon,
D.W. Hight, and S. Leroueil, eds., Swets and Zeitlinger, 29254.
Elastoviscoplastic modeling of soil 135
The meaning of Equation (5.32) is that the same inelastic volumetric
strain takes place at two different stress states with different strain rates,
provided the two states have the same mean effective stress and the same
initial conditions.
Figure 5.15 shows this fact schematically. In other words, the inelastic
volumetric strain is the same at both stress states, represented by P
1
and P
2

and lying on two different stress paths that correspond to strain rates

11
(1)

and

11
(2)
, shown in Figure5.15. The same inelastic volumetric strain (same
strain-hardening) states on different stress paths are shown in Figure5.15.
Under undrained conditions, total strain rate component

ij
is equivalent to
deviatoric strain rate component

e
ij
, because Equation (5.31) is always satisfed.
We discuss the problem in a simple case of conventional axisymmetric
triaxial undrained compression, > =
1 2 3
. Under this specifc condi-
tion, the following relations are reduced:

= = = = s J e
2
3
( ), 2 2/3( ),
2
3
( )
11 1 3 2 1 3 11 11 1 3

(5.33)
Inserting these relations in Equation (5.19) results in


= +
G
s F
1
2
( )
2
3
11 11
(5.34)

=

+
+

F C m
q
M
e
v
m
m m
p myi
s
m
( ) exp ln( / )
1
ln
0 0
( )
0
(5.35)
where = q
1 3
, = M M 3/2
*
, and G is the elastic shear modulus.
2J
2
O
2J
2

11
p
1
p
2

*(1)

*(2)

*
=
(1)

11
(2)
/
m

me

m

m0
Figure 5.15 Same inelastic volumetric strain states on the different stress paths.
136 Computational modeling of multiphase geomaterials
Assuming


p
11
11
, namely, the elastic deviatoric strain is negligible, the
next relation is obtained from Equations (5.34) and (5.35) by comparing
the states at P
1
and P
2
, as shown in Figure5.15.


( )
=
m
M
q q
m m
ln( / ) / /
11
(1)
11
(2) (1) (1) (2) (2)
or

m
M
q q
m m
exp
11
(1)
11
(2)
(1) (2)
(5.36)


( ) ( )
=

m M
q q
m
m
ln ln
11
(1)
11
(2)
(1)
(2)
(5.37)
where = q( )
11 33
is the deviator stress,
m
is the mean effective stress,
and superscripts (1) and (2) denote two stress states on the stress paths with
different strain rates and the same mean effective stress.
Figure5.16 is prepared to evaluate the validity of Equation (5.37). It is
clearly shown in the fgure that a linear relationship exists between the
logarithm of strain rate,

11
, and stress ratio, q
m
/ , as an equi-inelastic
volumetric strain line. Thus, the validity of Equation (5.34) is evaluated
and the parameter can be determined from the slope of the equi-inelastic
volumetric strain line.
10
3
2.0

me
= 196 kN/m
2
, e = 0.76

m
/
me
1.5
1.0
q
/

m
0.5
0.0
10
2

11
(%/min)
10
1
0.9
0.8
0.7
Figure 5.16 Relationship between the stress ratio and the logarithm of the strain rate.
Elastoviscoplastic modeling of soil 137
The next task is to determine the remaining parameters, namely, C
0

and
myi
s ( )
. Taking into account only the inelastic strain rate component in
Equation (5.19), under undrained conditions, we have

+

e C m
J
M
e
v
s
J
ij
vp
m m
m
me
p ij
exp
2
ln
1
2
2
*
2
(5.38)

=

C C m
myi
s
me
exp ln
0
( )
(5.39)
The preceding equation indicates that it is not necessary to determine
parameters C
0
and
myi
s ( )
separately; we only need parameter C. Substituting
the strain rate, the mean effective stress, the deviator stress, and the stress
ratio at the same stress point into Equation (5.38), we can determine
parameter C.
5.5.1 Strain rate effect
Figures 5.17 and 5.18 show the simulated and the experimental stress
strain relations and stress paths of one-day consolidated clay under strain-
rate controlled undrained triaxial conditions. The clay is reconstituted
Fukakusa clay preconsolidated under a pressure of 49 kPa. The physical
0 2 4 6 8

11
(%)
10 12
1.0

me
= 392 kN/m
2
, e
0
= 0.72
C = 1.96 10
9
(1/sec)
0.5
0.0
q
/

m
e

11
(%/min)
0.0835
0.00817
Experiment Simulation
Figure 5.17 Stressstrain relations for constant strain rate triaxial tests.
138 Computational modeling of multiphase geomaterials
properties are as follows: specifc gravity = 2.613; LL = 48.5%; PL = 26.7%;
IP = 21.8%; Fines content:clay and silt content (75m) = 67.7%; and sand
content ( 420m) = 5.9%. The fxed parameters used for the calculation
are = 0.1, = 0.02, = M M
m
3
2
*
= 1.5, m = 28.8, and G
0
= 363 kPa.
The theoretical curves were obtained using Equations (5.34) and (5.35).
It can be concluded that the strain-rate effect on the stress path and on the
stressstrain relation is reasonably assessed by the proposed theory. With
respect to elastic shear modulus, G, it is assumed that G is proportional
to the square root of the mean effective stress, and it is worth noting that
Youngs modulus, E, is equal to three times G under undrained conditions.
5.5.2 Simulation by the Adachi and Okas model
5.5.2.1 Effect of secondary consolidation
Figure5.19 shows the calculated and the measured effective stress paths of
one-day and seven-day consolidated samples, and Figure 5.20 shows the
effect of secondary consolidation on the stressstrain relations. The clay
sample, consolidated for a long period, increases in undrained strength and
initial tangent modulus. These phenomena are in general due to aging.
The test results correspond to Shen et al. (1973).
1.0
0.5
0.0
0.0 0.5 1.0

m
/
me

11
q
/

m
e
1.5
Critical state line

me
= 392 kN/m
2
, e
0
= 0.72
C = 1.96 10
9
(1/sec)

11
(%/min)
0.0835
0.00817
Experiment Simulation
Figure 5.18 Stress paths for constant strain rate triaxial tests.
Elastoviscoplastic modeling of soil 139
Critical state line
1.0
1 Day consolidation
Experiment
7 Days consolidation
Experiment
C = 1.44 10
8
(1/sec)

me
= 196 kN/m
2

11
= 0.0835 (%/min)
0.5
q
/

m
e
0.0 0.5 1.0
0.0

m
/
me
Simulation e
0
= 0.75
Simulation e
0
= 0.77
Figure 5.19 Calculated and measured effective stress paths for one-day consolidated and
seven-days consolidated clays.
0.0
0.0
0.5
1.0
0.5 1.0 1.5

11
(%)
2.0 2.5
q
/

m
e

11
= 0.0835 (%/min)
1 Day consolidation
7 Days consolidation
Experiment
C = 1.44 10
8
(1/sec)
Simulation e
0
= 0.75
Experiment Simulation e
0
= 0.77
Figure 5.20 Calculated and measured stressstrain relations for one-day consolidated
and seven-days consolidated clays.
140 Computational modeling of multiphase geomaterials
5.5.2.2 Isotropic stress relaxation
Figures 5.21 and 5.22 show the increase in pore water pressure during
isotropic stress relaxation. After the frst relaxation test, the sample was
reconsolidated and returned to undrained conditions, namely, the second
isotropic relaxation process. The results given in Figure 5.22 correspond
to the second isotropic stress relaxation process, which follows the frst
10
1
0.25
0.00
10
2
10
3
Time (min)
10
4
10
5
u
/

m
e

me
= 196 kN/m
2
, e
0
= 0.834
: Experiment
: Simulation
(C = 1.2 10
5
1/sec, v
i
= 0)
vp
Figure 5.21 Induced pore water pressure during the undrained isotropic stress relaxation.
10
1
0.25
0.00
10
2
10
3
Time (min)
10
4
10
5
u
/

m
e

me
= 196 kN/m
2
, e
0
= 0.829
: Experiment
: Simulation
(C = 1.2 10
5
, v
i
= 5.81 10
3
)
vp
Figure 5.22 Induced pore water pressure during the undrained isotropic stress relax-
ation after the drainage of the pore water pressure during the stress
relaxation test shown in Figure 5.21.
Elastoviscoplastic modeling of soil 141
relaxation test results shown in Figure5.21. Therefore, the calculated ini-
tial inelastic volumetric strain is different in the two cases. The void ratio
was determined from the volume change in the clay specimen during the
consolidation process. Similar experimental results have been reported by
Arulanandan et al. (1971).
5.5.3 Constitutive model for anisotropic
consolidated clay
For anisotropic consolidated clay, we have extended the static yield func-
tion using a second invariant of relative stress ratio,
(0)
*
, as

= +

= f M
m
my
s
ln 0
(0)
* *
( )
(5.40)

( )( )
=
ij ij ij ij (0)
* *
(0)
* *
(0)
*
(5.41)
where = s
ij ij m
/
*
and the subscript (0) denotes the state at the end of con-
solidation, namely, at the initial state before deformation.
Using the extended yield function, we have obtained the viscoplastic
strain rate as

=

+

C m
M
M
ij
vp
m
m
my
ij ij
m
kl kl kl
ij
exp ln
( ) 1
3
0
(0)
*
*
*
(0)
*
(0)
*
*
* *
(0)
*
(0)
*
(5.42)
5.6 EXTENDED VISCOPLASTIC MODEL CONSIDERING
STRESS RATIO-DEPENDENT SOFTENING
As mentioned earlier, Adachi et al. (1987a) extended the original model to
describe the acceleration creep behavior of clay by introducing a second
material function into the model. Oka et al. (1994, 1995) studied the insta-
bility of the extended model during the undrained conventional creep pro-
cess and strain localization analysis.

, =

F
f
F
f
ij
vp
ij
s
s
( ) ( )
1 2
(5.43)
where
2
is the second material function.
142 Computational modeling of multiphase geomaterials
The second material function is introduced to explain that the rate-
dependency of clay vanishes at the failure state. In other words, the stress
ratio at the failure state does not depend on the strain rate. In the study by
Adachi et al. (1987a), the following form for the second material function
is adopted:

= + 1
2 (5.44)
Internal variable expresses the deterioration of the materials and obeys
the following evolutional equation as:

M
G M
f
f
( )
2
2
2
(5.45)
where

M
f
is the value of stress ratio

at the failure state and

G
2
is a
material parameter.
We can rewrite Equation (5.10) in an alternative form as

=

/
f
I f f
s s
vp
ij ij
1
1
2
2
1 2
(5.46)
where I
vp
2
is the second invariant of the viscoplastic strain rate tensor. From
Equation (5.46), we can see that the yield function depends implicitly on
both the hardening parameter and the strain rate.
5.7 ELASTOVISCOPLASTIC MODEL FOR COHESIVE
SOIL CONSIDERING DEGRADATION
5.7.1 Elastoviscoplastic model
considering degradation
The prediction by the extended model with the second material function
(Adachi et al. 1987a) indicates that the drop in stress is rather small com-
pared to the experimental evidence for sensitive clay and natural soil during
strain softening. During the strain-softening behavior of natural clay, it
is observed that strain softening follows a rather large decrease in mean
Elastoviscoplastic modeling of soil 143
effective stress. This indicates that the soil exhibits both shear and volumet-
ric softening. In order to incorporate these features, a new model has been
developed by Kimoto (2002), Kimoto et al. (2004), and Kimoto and Oka
(2005), considering the degradation of soil structures and rate dependency.
The model with the structural degradation can be applicable to the soft
rock as well as soft and hard soils (Oka et al. 2011).
We assume an overconsolidation boundary surface that delineates the
overconsolidated (OC) region < f
b
( 0) from the normally consolidated
(NC) region f
b
( 0) , namely,
= +

= f M
b m
m
mb
ln 0
(0)
* *
(5.47)
in which M
m
*
is the value of =
ij ij
* * *
when the volumetric strain incre-
ment changes from compression to swelling and
mb
is the hardening
parameter.
Originally, the hardening rule for the

mb surface was defned with

respect to the viscoplastic volumetric strain. In order to describe the degra-
dation of the material caused by structural changes, strain softening with
viscoplastic strain is introduced in addition to strain hardening with visco-
plastic volumetric strain as
( )
= =
+

A A
e
mb ma v
vp
exp ,
1
3 3
0
(5.48)
where e
0
is the initial value of the void ratio and the current void ratio can
be used for the fnite deformation analysis.
( ) ( )
= + z
ma maf mai maf
exp
(5.49)
where
ma
is assumed to decrease with an increasing viscoplastic strain
in Equations (5.48) and (5.49), and
mai
and
maf
are the initial and the
fnal values for
ma
, respectively.
mai
corresponds to the consolidation
yield stress and
maf
is determined from the difference between the peak
stress and the residual stress. Material parameter, , controls the rate of
structural changes and z is an accumulation of the second invariant of the
viscoplastic strain rate as

= = z zdt z
t
ij
vp
ij
vp
, ,
0

(5.50)
144 Computational modeling of multiphase geomaterials
The mechanical behavior of clay at its static equilibrium state is assumed
to be described by the original Cam-clay model (Adachi and Oka 1982).
The following static yield function is used:

= +

= f M
y
m
my
s
ln 0
(0)
* *
( )
(5.51)
In a similar way for the OC boundary surface, f
b
, strain softening is
defned in order to express the effect of a structural collapse through
changes in
my
s ( )
with the viscoplastic strain, namely,
{ }
= +
+

n n z
e
my
s
myi
s
v
vp
(1 )exp( ) exp
1
( ) ( ) 0
(5.52)
where = n
maf mai
/ describes the degree of structure at the initial state
and is the rate of degradation.
According to the structural collapse, the decrease in
my
s ( )
leads to the
shrinking of the static yield function.
The viscoplastic potential function is given as

= +

= f M
p
m
mp
ln 0
(0)
* *
(5.53)
where the dilatancy parameter

M
*
is assumed to be constant in the NC
region.
mp
is determined automatically from the stress state in the NC
region, and it coincides with
mb
in the OC region shown in Figure5.23.
The value varies with the current stress in the OC region as

( )
=

M
M NC region
OC region
m
ij ij
m mc
:
ln /
:
*
*
* *
(5.54)
where
mc
denotes the mean effective stress at the intersection of the over-
consolidation boundary surface and the
m
axis as

=

M
mc mb
ij ij
m
exp
(0)
*
(0)
*
*
(5.55)
Elastoviscoplastic modeling of soil 145
According to the earlier defnition, the value of dilatancy coeffcient

M
*

becomes zero when the stress path coincides with the mean effective stress
axis during cyclic loading. Therefore, a new defnition for

M
*
(Kimoto
et al. 2007) is introduced, as shown in Figure5.24 which is for the isotropic
consolidated one as

( )
( )
=

M
M
M
m
m mb m
: consolidated NC region
( / ) : one as OC region
*
*
* *
(5.56)
where
m
*
denotes the mean effective stress at the intersection of the sur-
face, which has the same shape as f
b
, and is given by

=

M
m m
m
exp
( )
* (0)
*
*
(5.57)
S
ij
S
ij
S
ij
S
ij
OC region
(a) OC Region
M
*
m
f
b
= 0
f
p
= 0

mb
=
mc

my
f
y
= 0
M
*
~
Current stress state
NC region
(b) NC Region
M
*
m
f
b
= 0
f
p
= 0

mb

mp

my
(s)
f
y
= 0
Current stress state
(s) 0
Figure 5.23 Yield surface and potential surface for isotropically consolidated clay.
146 Computational modeling of multiphase geomaterials
In Equations (5.56) and (5.57),

M
m
, in general, depends on the Lodes
angle (Equation (4.35)). The viscoplastic strain rate tensor is given as
the following equation based on the overstress type of viscoplastic theory
(Perzyna 1963):

f
f
ij
vp
y
p
ij
( )
1
(5.58)
where are Macaulays brackets.
S
ij
S
ij
OC region
(a) OC Region
M
*
m
()
M
*
m
()
M
*
f
b
= 0
f
p
= 0
f
y
= 0

mb
*
m

my
Current stress
~
(s)
S
ij
S
ij
NC region
f
b
= 0
f
y
= 0
f
p
= 0
Current stress
(b) NC Region

mp

mb
my
(s)
Figure 5.24 Overconsolidation boundary surface, static yield function, and viscoplastic
potential function.
Elastoviscoplastic modeling of soil 147
Based on the experimental results of strainrate constant triaxial tests,
material function
1
is defned as

= +

= +

+
= +


= +

= +

f C m M
C m M
n n z A
C m M
A
C m M
C m M
y m
m
my
s
m
m
myi
s
v
vp
m
m
ma v
vp
m
mai
myi
s
m
mb
m
m
mb
myi
s
mai
( ) exp ln
exp ln
{ (1 )exp( )}exp( )
exp ln
exp( )
exp ln
exp ln
1 (0)
* *
( )
(0)
* *
3
(0)
* *
3
(0)
* *
( )
(0)
* *
(5.59)

C C mM
myi
s
mai
exp ln
*
( )
(5.60)

= +

f
f
C m M
M
ij
vp
y
p
ij
m
mb
ij ij kl kl kl
ij
( ) exp ln
( ) 1
3
1 (0)
* *
*
(0)
*
(0)
*
*
* *
(0)
*
(0)
*
(5.61)
When = =
mi mbi mai
and the shear stress is zero at the initial state
after consolidation,

CM
kki
vp *

=
(5.62)
where
mi
,
mai
, and
mbi
are the initial values for
m
,
ma
, and
mb
,
respectively. C is the viscoplastic parameter corresponding to the viscoplas-
tic volumetric strain rate at the initial stress state with zero shear stress.
148 Computational modeling of multiphase geomaterials
In the preceding formulation, we assumed that the viscoplastic param-
eter, C, was scalar. However, we can generalize this assumption with a
tensorial value for C such as the fourth-order isotropic tensor C
ijkl
=
+ + = = + a b C b C a b
ij kl ik jl il jk
( ), 2 , 3 2
1 2
, as has been adopted by Oka
(1982, 1992) and Oka et al. (2003a).

= +

C m M
M
ij
vp
ijkl
m
mb
kl kl mn mn mn
kl
exp ln
( ) 1
3
(0)
* *
*
(0)
*
(0)
*
*
* *
(0)
*
(0)
*
(5.63)
5.7.2 Determination of the material parameters
There are ten material parameters for the proposed constitutive model. The
procedure for determining these parameters is as follows. The initial void
ratio, e
0
, can be obtained from tests for physical properties. The compres-
sion index, , and swelling index, , are given by the slope of the isotro-
pic consolidation and swelling tests, respectively. Compression yield stress,

mbi
, is assumed to be determined from the yield point of the isotropic
consolidation tests. Elastic shear modulus, G
0
,

can be determined from the
initial slope of the triaxial compression tests.
The stress ratio at maximum compression, M
m
*
, is defned as the stress
ratio whereby maximum compression occurs in the drained compression
tests. For clay, however, it has been assumed to equal the stress ratio at the
critical state. Herein, M
m
*
is determined from the stress ratio at the residual
state in the undrained triaxial compression tests.
The viscoplastic parameter, m, can be determined from undrained tri-
axial compression tests with different strain rates as

m
q q
m m
exp
2
3
11
(1)
11
(2)
(1) (2)
(5.64)


( ) ( )
=

m
q q
m m
3
2
ln ln
11
(1)
11
(2)
(1) (2)
(5.65)
where = q ( )
11 33
is the deviator stress,
m
is the mean effective stress,
and superscripts (1) and (2) denote two stress states on the stress paths with
different strain rates and the same mean effective stress.
When m is determined, the viscoplastic parameter, C, is obtained from
the deviatoric strain rate by the constitutive equation.
Elastoviscoplastic modeling of soil 149
Alternatively, we can determine parameters m and C through the sec-
ondary compression rate and the initial volumetric strain rate in a similar
way, as shown in Equation (5.23), although the expression is a bit different
since the form for the rate sensitivity function, shown in Equation (5.59), is
different from that in Equation (5.13).

=

+
m
e M (1 )
*
(5.66)

= C M
kk
vp
/
(0)
*
(5.67)
where is a secondary compression rate in a natural logarithm, namely,

= + = v t t v v
p p p
kk
vp
ln / ,
0
0

(5.68)
The relationship given by Equation (5.66) can be rewritten by the sec-
ondary compression index as

m
C C
C M
c s
*
(5.69)
where C
c
is the compression index, C
s
is the swelling index, C

is the sec-
ondary compression rate, and M
m
*
is the stress ratio at the maximum com-
pression or the stress ratio at a large strain.
Structural parameter
maf
is determined by the decrease from the peak
stress to the residual stress in the undrained tests. Structural parameter is
determined by curve ftting for the strain-softening process in the undrained
tests. The other possible model for the elastoviscoplastic model is a constitu-
tive equation in which the degradation method is adopted for Equation (5.19).
5.7.3 Strain-dependent elastic shear modulus
The nonlinearity of soil stiffness has been studied extensively for materials
such as sand, clay, and gravel, and has been summarized well by Ishihara
(1996). For cohesive soil, several empirical equations have been proposed
by considering the dependency of the shear modulus on the effective confn-
ing stress (Kokusho et al. 1982). In the original model by Adachi and Oka
(1982) and Kimoto and Oka (2005), the change in the elastic shear modu-
lus of the elastoviscoplastic model is given by the square root function of
the normalized mean effective stress, namely,

=

G G e
m
m
( )
0
0
(5.70)
150 Computational modeling of multiphase geomaterials
in which G
0
is the value for G when
=
m m0 and the function of the void
ratio e.
Equation (5.70) considers only the effect of the confning pressure, which
can accurately approximate the variation in shear modulus at very small lev-
els of strain. In regions with large levels of strain, however, as demonstrated
by the experimental results, the strain dependency of the shear modulus
should be considered as well. Various empirical formulations have been
provided from the laboratory test results to express the strain dependency
of the shear modulus (Hardin and Drnevich 1972; Fahey 1992; Wang and
Kuwano 1999). Ogisako et al. (2007) have introduced a normalized shear
modulus reduction function based on the viscoplastic shear strain in soft
clay specimens and have proposed a hyperbolic equation for that expres-
sion, namely,
=
+
G G
vp r
1
(1 ( ) )
0 (5.71)
where and r are the experimental constants, which can be defned from
the laboratory test results, and
vp
is the accumulated viscoplastic shear
strain given by the accumulation of the viscoplastic deviatoric strain incre-
ment, de
ij
vp
, as

= de de
vp
ij
vp
ij
vp
(5.72)
5.8 APPLICATION TO NATURAL CLAY
5.8.1 Osaka Pleistocene clay
The model is applied to Osaka Pleistocene clay, namely, Kyuhoji clay. It was
sampled from the upper Pleistocene layer called Ma12, which is distributed
widely in the western and eastern parts of Osaka, Japan, at a depth of 20 to
40 meters. This is marine clay containing diatoms, and it exhibits sensitive
behavior due to the effect of the structures formed during the sedimentation
process. Figure5.25 compares the undrained compression test results between
the undisturbed and the reconstituted samples of Kyuhoji clay (Yashima et
al. 1999; Shigematsu 2002). Both the undisturbed clay and the reconstituted
clay were sheared with an axial strain rate of 0.005%/min after isotropic
consolidation at a confning pressure of 392 kPa, which is a little larger than
the compression yield stress of 340 kPa. The initial void ratio of the undis-
turbed clay is larger than that of the reconstituted clay, specifcally, 1.41 for
the undisturbed clay and 1.02 for the reconstituted clay. The undisturbed
Elastoviscoplastic modeling of soil 151
clay exhibits larger strength and the deviator stress decreases after the peak
stress, as seen in Figure5.25. Figure5.26 shows the results of simulations
that give compressive strain rates under the triaxial stress state. An axial
strain rate of

11
= 0.005%/min is provided for the calculations. The material
parameters used in the simulations are shown in Table5.2. The structural
parameter,
maf
, is set to be 280 kPa for the undisturbed clay, and is set to
be 10 for the undisturbed clay and 0 for reconstituted clay. = 0 provides the
original model, which does not describe structural changes. The values for
= 0.005%/min
OCR = 1.0
= 0.005%/min
OCR = 1.0
400
300
200
100
0
0 2 4 6 8 10
Axial Strain
11
(%)
Mean Efective Stress p (kPa)
12 14 16
D
e
v
i
a
t
o
r

S
t
r
e
s
s

q

(
k
P
a
)
D
e
v
i
a
t
o
r

S
t
r
e
s
s

q

(
k
P
a
)
Undisturbed clay
Reconstituted clay
Undisturbed clay
Reconstituted clay
0
400
300
200
100
0
100 200 300 400
Figure 5.25 Experiments of undrained triaxial tests for Kyuhoji clay. (After Yashima,
A., Shigematsu, H., Oka, F., and Nagaya, J., 1999, J. Geotech. Eng., JSCE,
624 (III-47): 217229.)
152 Computational modeling of multiphase geomaterials
C contain
myi
s ( )
concerning the degree of initial structures in the derivation.
Since the degree of the initial structure of the reconstituted clay is thought
to be lower than that of the undisturbed clay, a larger value for C is given
for the reconstituted clay. Figures5.25 and 5.26 confrm that the proposed
model can describe the difference in behavior between the highly structured
soil and the lowly structured soil.
5.8.2 Osaka Holocene clay
The model of Equation (5.61) has been used for various types of clay.
Figures5.27 and 5.28 show the stressstrain relations and the stress paths
of Osaka Holocene clay in triaxial compression tests under undrained
400
300
200
100
0
0 2 4 6
Axial Strain (%)
8 10 12 14 16
D
e
v
i
a
t
o
r

S
t
r
e
s
s

(
k
P
a
)
Undisturbed clay
Reconstituted clay
OCR = 1.0
Mean Efective Stress (kPa)
400
300
200
100
0
0 100 200 300 400
D
e
v
i
a
t
o
r

S
t
r
e
s
s

(
k
P
a
)
OCR = 1.0
M
m
Undisturbed clay
Reconstituted clay
Figure 5.26 Simulations of undrained triaxial tests for Kyuhoji clay.
Elastoviscoplastic modeling of soil 153
conditions (Mirjalili 2010; Mirjalili et al. 2011). The material parameters
used in the analysis are listed in Table 5.3. The fgures indicate that the
behavior is rate dependent and the stressstrain relations exhibit extensive
strain softening in the overconsolidated range. Figures5.27 and 5.28 illus-
trate a comparison between the simulated results by the proposed model and
the experimental results. From the fgures, it is seen that the model can well
simulate the rate-dependent behavior of soft clay as well as strain softening.
5.8.3 Elastoviscoplastic model based on
modified Cam-clay model
When we use the modifed Cam-clay model (Roscoe and Burland 1968),
the following yield and plastic potential functions are recommended, since
if we use the original yield functions by Roscoe and Burland, we encounter
complexity in determining the viscoplastic parameter, m (Karunawardena
2007; Kimoto et al. 2011):
=
+

+

f
M
M
y
m
my
s
ln ln 0
*2 *2 *2
*2 *2 ( )
(5.73)
=
+

+

f
M
M
p
m
mp
ln ln 0
*2 *2 *2
*2 *2
(5.74)
= =
x ij ij ij ij ij ij
( )( ), ,
* * * * * * * *
(5.75)
where
ij
*
is a kinematical hardening parameter.
Table5.2 Material parameters for Kyuhoji clay
Undisturbed Reconstituted
Elastic shear modulus G
0
8333 (kPa) 6330 (kPa)
Compression index 0.327
Swelling index 0.028
Initial void ratio e
0
1.41 1.02
Compression yield stress
mbi
=
mai
= 392 (kPa)
Stress ratio at failure M
*
m
1.22
Viscoplastic parameter m 21.5
Viscoplastic parameter C 4.5 10
11
(1/s) 2.5 10
8
(1/s)
Structural parameter
maf
280 (MPa) 392 (MPa)
Structural parameter 10.0 0
154 Computational modeling of multiphase geomaterials
500
400
300
200
D
e
v
i
a
t
o
r

S
t
r
e
s
s
,

q

(
k
P
a
)
100
0
500
400
300
200
D
e
v
i
a
t
o
r

S
t
r
e
s
s
,

q

(
k
P
a
)
100
0
0 5 10
Axial Strain,
11
(%)
15 20
500
400
300
200
D
e
v
i
a
t
o
r

S
t
r
e
s
s
,

q

(
k
P
a
)
100
0
0 5 10
Axial Strain,
11
(%)
15 20
0 5 10
Axial Strain,
11
(%)
15 20
Tw 2-1-2; 0.05%/min
Exp.
3
= 100 kPa
Exp.
3
= 200 kPa
Exp.
3
= 400 kPa
Simulated results
Tw 2-2-1; 0.05%/min
Exp.
3
= 100 kPa
Exp.
3
= 200 kPa
Exp.
3
= 400 kPa
Simulated results
500
400
300
200
D
e
v
i
a
t
o
r

S
t
r
e
s
s
,

q

(
k
P
a
)
100
0
0 5 10
Axial Strain,
11
(%)
15 20
Tw 2-2-2; 0.005%/min
Exp.
3
= 100 kPa
Exp.
3
= 200 kPa
Exp.
3
= 400 kPa
Simulated results
500
400
300
200
D
e
v
i
a
t
o
r

S
t
r
e
s
s
,

q

(
k
P
a
)
100
0
0 5 10
Axial Strain,
11
(%)
15 20
Tw 2-3-1; 0.05%/min
Exp.
3
= 100 kPa
Exp.
3
= 200 kPa
Exp.
3
= 400 kPa
Simulated results
500
400
300
200
D
e
v
i
a
t
o
r

S
t
r
e
s
s
,

q

(
k
P
a
)
100
0
0 5 10
Axial Strain,
11
(%)
15 20
Tw 2-3-2; 0.005%/min
Exp.
3
= 100 kPa
Exp.
3
= 200 kPa
Exp.
3
= 400 kPa
Simulated results
Tw 2-1-2b; 0.005%/min
Exp.
3
= 100 kPa
Exp.
3
= 200 kPa
Exp.
3
= 400 kPa
Simulated results
Figure 5.27 Experimental and simulated stressstrain relations of Torishima clay.
Elastoviscoplastic modeling of soil 155
500
400
300
200
D
e
v
i
a
t
o
r

S
t
r
e
s
s
,

q

(
k
P
a
)
100
0
0 100 200
Mean Efective Stress,
m
(kPa)
300 500 400
Tw 2-1-2; 0.05%/min
M = 1.52
500
400
300
200
D
e
v
i
a
t
o
r

S
t
r
e
s
s
,

q

(
k
P
a
)
100
0
0 100 200
Mean Efective Stress,
m
(kPa)
300 500 400
Tw 2-2-1; 0.05%/min
M = 1.43
500
400
300
200
D
e
v
i
a
t
o
r

S
t
r
e
s
s
,

q

(
k
P
a
)
100
0
0 100 200
Mean Efective Stress,
m
(kPa)
300 500 400
Tw 2-3-1; 0.05%/min
M = 1.36
Exp.
3
= 100 kPa
Exp.
3
= 200 kPa
Exp.
3
= 400 kPa
Simulated results
500
400
300
200
D
e
v
i
a
t
o
r

S
t
r
e
s
s
,

q

(
k
P
a
)
100
0
0 100 200
Mean Efective Stress,
m
(kPa)
300 500 400
Tw 2-3-2; 0.005%/min
M = 1.36
500
400
300
200
D
e
v
i
a
t
o
r

S
t
r
e
s
s
,

q

(
k
P
a
)
100
0
0 100 200
Mean Efective Stress,
m
(kPa)
300 500 400
Tw 2-2-2; 0.005%/min
M = 1.43
Exp.
3
= 100 kPa
Exp.
3
= 200 kPa
Exp.
3
= 400 kPa
Simulated results
500
400
300
200
D
e
v
i
a
t
o
r

S
t
r
e
s
s
,

q

(
k
P
a
)
100
0
0 100 200
Mean Efective Stress,
m
(kPa)
300 500 400
Tw 2-1-2b; 0.005%/min
M = 1.52
Exp.
3
= 100 kPa
Exp.
3
= 200 kPa
Exp.
3
= 400 kPa
Simulated results
Exp.
3
= 100 kPa
Exp.
3
= 200 kPa
Exp.
3
= 400 kPa
Simulated results
Exp.
3
= 100 kPa
Exp.
3
= 200 kPa
Exp.
3
= 400 kPa
Simulated results
Exp.
3
= 100 kPa
Exp.
3
= 200 kPa
Exp.
3
= 400 kPa
Simulated results
Figure 5.28 Experimental and simulated stress paths of Torishima clay.
156 Computational modeling of multiphase geomaterials
It is interesting to note that the shape of the yield surface and the vis-
coplastic potential are the same as the original modifed Cam-clay model.
5.9 CYCLIC ELASTOVISCOPLASTIC MODEL
Several constitutive models have been proposed to describe the rheological
behavior of clay under static loading conditions, as mentioned in Section
5.3. However, few viscoplastic constitutive models are available for analysis
under dynamic loading conditions (e.g., Oka 1992; Modaressi and Laloui
1997; Oka et al. 2004; Maleki and Cambou 2009). Oka (1992) developed
a cyclic elastoviscoplastic constitutive model for clay based on nonlinear
kinematic hardening rules (Armstrong and Frederick 1966; Chaboche and
Rousselier 1983). Later, Oka, Kodaka, et al. (2003) proposed a cyclic vis-
coelasticviscoplastic model by incorporating the viscoelastic feature into
the constitutive equations, in which the behavior of clay can be described
not only in the range of middle to high levels of strain, but also in the
range of low levels of strain. The models have been successively applied
to the dynamic analysis of the ground during earthquakes, considering
liquefaction (Oka, Uzuoka, et al. 2003), as shown in Chapter 9.
Table5.3 Material parameters of soft clay layer Ac2
Ac2-U Ac2-M Ac2-L
Depth (m) 11.0 15.0 15.0 19.0 19.0 23.0
Tw 2-1-2 Tw 2-2-1 Tw 2-3-1
Test no. Tw 2-1-2b Tw 2-2-2 Tw 2-3-2
Initial void ratio, e
0
1.25 1.65 1.42
Poissons ratio, 0.3 0.3 0.3
Initial elastic shear modulus
(1)
G
0
(kPa) 3759 3927 5993
Compression index, 0.341 0.593 0.652
Swelling index, 0.019 0.027 0.014
Stress ratio at failure, M
*
mc
1.24 1.18 1.12
Viscoplastic parameter, m 24.68 28.2 21.15
Viscoplastic parameter, C
1
(1/s) 3.83 10
11
1.85 10
11
8.99 10
11
Viscoplastic parameter, C
2
(1/s) 3.83 10
11
1.85 10
11
8.99 10
11
Structural parameter, = n /
maf mai
0.83 0.67 0.60
Structural parameter, 10 15 20
Strain dependent parameter, a 20 10 10
Strain dependent parameter, r 0.4 0.4 0.4
(1)
Based on the mean effective stress at the depth of the specimen.
Elastoviscoplastic modeling of soil 157
5.9.1 Cyclic elastoviscoplastic model based on
nonlinear kinematical hardening rule
Oka (1992) proposed a cyclic elastoviscoplastic model by adopting the
nonlinear kinematical hardening rule by Armstrong and Frederick (1966).
The static yield function is given by

= +

=

f
M
y
x
m
my
s
ln 0
( )
( )
(5.76)
= x x
x ij ij ij ij
( )( )
( )
* * * * *
(5.77)
where

ij
*
is a hardening parameter described by

= = dx B A de x d d de de
ij
ij
vp
ij
vp vp
ij
vp
ij
vp
( )
* * * *

(5.78)
in which B* and = A M
f
* *
are kinematical hardening parameters.
The plastic potential function is given by

= +

=

f
M
p
x
m
mp
ln 0
( )
(5.79)
where

M is constant in normally consolidated clay and varies in the over-
consolidated region as

=
:

: <

M
M f
f
m b
ij ij
m mc
b
0
ln( )
0
(5.80)
in which mc is the value at the intersection point between the overconsoli-
dated boundary surface and mean effective stress axis.

M
mc mb
ij ij
m
exp
(0) (0)
(5.81)
For the isotropic stress conditions,
mc
corresponds to
mb
.
The fow rule is now generalized as

C f
f
ij
vp
ijkl y
p
kl
( )
1 (5.82)

= + + C a b
ijkl ij kl ik jl jk il
( )
158 Computational modeling of multiphase geomaterials
Finally, the deviatoric viscoplastic strain rate tensor,

e
ij
vp
, and the visco-
plastic volumetric strain rate,

kk
vp
, are obtained as

e C m
ij
vp ij ij
exp( )
01 0
* *
(5.83)

C m M
kk
vp mn mn mn
exp( )
( )
02 0
*
* * *
(5.84)

= = + C b C a b 2 , 3 2
01 02
5.9.2 Cyclic elastoviscoplastic model
considering structural degradation
Despite the ability of the above model (Oka, 1992) to explain the deforma-
tion characteristics under cyclic loading conditions, the effect of the struc-
tural degradation of clay particles was disregarded. Taking into account
structural degradation and microstructural changes, a cyclic elastovisco-
plastic model is developed based on the nonlinear kinematic hardening
rules for changes in both the stress ratio and the mean effective stress. In
addition, the kinematic hardening rule for changes in viscoplastic volumet-
ric strain is generalized to predict the behavior during the cyclic loading
process (Shahbodagh 2011).
5.9.2.1 Static yield function
The static yield function is obtained by considering the nonlinear kinematic
hardening rule for changes in the stress ratio, the mean effective stress, and
the viscoplastic volumetric strain, as

f M y
y
mk
my
s
m
mk
m
ln ln 0
* *
( )
*

= +

=
(5.85)

( )( ) { }
=
ij ij ij ij
* * * * *
1
2
(5.86)
in which
mk
is the unit value of the mean effective stress, y
m
*
is the sca-
lar kinematic hardening parameter, and
my
s ( )
denotes the static hardening
parameter.

ij
*
is the so-called back stress parameter, which has the same
dimensions as the stress ratio,
ij
*
.
Elastoviscoplastic modeling of soil 159
Incorporating strain softening into the structural degradation, the hard-
ening rule for
my
s ( )
can be expressed as
=
+

z
my
s maf mai maf
mai
myi
s
( )exp( )

( ) ( )
(5.87)
5.9.2.2 Viscoplastic potential function
In the same manner as for the static yield function, the viscoplastic poten-
tial function, f
p
, is given by
f M y
p
mk
mp
m
mk
m
ln ln 0
* * *

= +

(5.88)
The dilatancy coeffcient,

M
*
, is defned separately for the NC region
and the OC region as

=

M
M
M
m
m mb m
: NC region
( / ) : OC region
*
*
* *
(5.89)
where
m
is the mean effective stress at the intersection of the surface and
the axis of anisotropic consolidation
+

M
m
m
m
ln 0
(0)
(5.90)
and
m
*
is given by
=

M
m m
m
exp
* (0)
*
*
(5.91)
Figure5.29 illustrates the overconsolidation boundary surface (f
b
= 0), the
static yield surface (f
y
= 0), and the viscoplastic potential function (f
p
= 0) for
the isotropically consolidated soil in the OC and NC regions, respectively.
5.9.2.3 Kinematic hardening rules
The evolution equation for the nonlinear kinematic hardening parameter,

ij
*
, is given by
( )
= d B A de d
ij
ij
vp
ij
vp * * * *
(5.92)
160 Computational modeling of multiphase geomaterials
where A
*
and B
*
are material parameters, de
ij
vp
is the viscoplastic deviatoric
strain increment tensor, and = d de de
vp
ij
vp
ij
vp

is the viscoplastic shear strain

increment tensor. A
*
is related to the stress ratio at failure, namely, = A M
f
* *
,
and B
*
is proposed to be dependent on the viscoplastic shear strain as
( )
= + B B B C B
f
n
vp
( )exp
*
max
*
1
*
( )
*
1
*
(5.93)
S
ij
S
ij
OC region
M
*
f
M
*
m
M
*
*
f
b
= 0
f
p
= 0
f
y
= 0

mb
*
m
(a) OC Region

my
Current
stress
~
(s)
(a)
M
*
f
*
M
*
m
NC region
(b) NC Region
f
b
= 0
f
y
= 0
f
p
= 0
Current stress

mp

mb

my
(s)
S
ij
S
ij
(b)
Figure 5.29 (a) Overconsolidation boundary surface, static yield function, and visco-
plastic potential function for isotropically consolidated state in the (a) OC
region and (b) NC region.
Elastoviscoplastic modeling of soil 161
in which B
*
1
is the lower boundary of B
*
, C
f
is the parameter controlling the
amount of reduction, and
n
vp
( )
*
is the accumulated value of viscoplastic shear
strain between two sequential stress reversal points in the previous cycle.
B
max
*
is the maximum value of parameter B
*
, which is defned following the
proposed method by Oka et al. (1999) as

=
+
B
B
B
n
vp
n r
vp
:Before reaching failure line
1 /
:After reaching failure line
max
*
0
*
0
*
( )max
*
( )
*
(5.94)
where B
0
*
is the initial value of B
*
,
n
vp
( )max
*
is the maximum value of
n
vp
( )
*
in
past cycles, and

n r
vp
( )
*
is the viscoplastic reference strain.
In order to improve the predicted results under cyclic loading conditions,
a scalar nonlinear kinematic hardening parameter, y
m
*
, can be decomposed
into two parts as

= +

y y y
m m m 1 2
(5.95)
Then, two hardening parameters follow the evolutional equations as

( )
= dy B A d y d
m v
vp
m v
vp
1
*
2
*
2
*
1
*
(5.96)

=

y B d
m v
vp
2 3
(5.97)
where A
2
*
, B
2
*
and

B
3
are material parameters, and d
v
vp
is the increment
in the viscoplastic volumetric strain tensor. The values for A
2
*
, B
2
*
and

B
3

are determined by a data-adjusting method from the laboratory test data.
5.9.2.4 Strain-dependent shear modulus
The degradation of the elastic shear modulus from the beginning of load-
ing can be expressed by its dependency on accumulated viscoplastic shear
strain
vp
as
( )
( )
=
+

G
G
vp
r
m
m
1
0
0
(5.98)
where r and are the strain-dependent parameters, which can be determined
from the laboratory test results. In this study, based on the experimental
results, r = 0.4 is chosen, and
( )
*
( )
*
d
n
vp
n
vp
= is the accumulated plastic shear
strain between two sequential stress reversal points in the nth cycle.
162 Computational modeling of multiphase geomaterials
5.9.2.5 Viscoplastic flow rule
Based on the overstress type of viscoplastic theory frst adopted by Perzyna
(1963), viscoplastic strain rate tensor

ij
vp
is defned as

C
f
f
ij
vp p
kl
ijkl y
( ) (5.99)
=
>

f f
f
f
y y
y
y

( ) : 0
0: 0
) (5.100)
where are Macaulays brackets, and (f
y
) is the rate-sensitive material
function and is determined as

= +

f C m M
m M y
C m M y
ijkl y ijkl m
mai
myi
s
mk
ma
m
mk
m
ijkl m
mk
ma
m
mk
m
C ( ) exp ln
exp ln ln
exp ln ln

*
( )
* *
1
*

* *
1
*
(5.101)
in which m is the viscoplastic parameter and

= + + C a b
ij kl
ijkl ik jl il jk
( )
(5.102)
where C
ijkl
is a fourth-order isotropic tensor. a and b in Equation (5.99) are
material constants.
Finally, by combining Equations (5.99), (5.101), and (5.102), the visco-
plastic deviatoric strain rate,

e
ij
vp
, and the viscoplastic volumetric strain
rate,

ij
vp
, can be expressed as

= +

e C m M y
ij
vp mk
ma
m
mk
m
ij ij
exp ln ln
1
* *
1
*
* *
*
(5.103)
Elastoviscoplastic modeling of soil 163

exp ln ln
ln
ln
( )
2
* *
1
*
*
1
*
1
*
* * *
*

C m M y
M
y
y
kk
vp mk
ma
m
mk
m
m
m
mn mn mn
m
mk
m
mk

= +

(5.104)
where

m M C b
mai
myi
s
exp{ ln } 2
*
1
( )
and

= +

m M C a b
mai
myi
s
exp{ ln } (3 2 )
*
2
( )
are the
viscoplastic parameters for the deviatoric and the volumetric strain compo-
nents, respectively.
To verify the performance of the proposed model, the modeling of the
soft clay samples is carried out by integration of the constitutive equa-
tions under undrained triaxial conditions. Figure5.30 shows the simulated
results for Nakanoshima soft clay from Osaka, Japan in which

B
3
is zero.
The material parameters are listed in Table 5.4. It is seen that the mean
effective stress decreases with an increase in loading cycles.
As for the shape of cyclic stressstrain curve, we can improve it if we use
the following formula for the elastic shear modulus:

( )
( )
=
+

G
G e
n
vp
r
m
m
( )
1
0
( )
*
0
(5.105)
where =

d
n
vp
n
vp
( )
( )
is the accumulated plastic shear strain between two
sequential stress reversal points in the nth cycle.
200
100
0
100
D
e
v
i
a
t
o
r

S
t
r
e
s
s

(
k
P
a
)
200
M
me
= 1.3
M
me
= 1.4
200
100
0
100
D
e
v
i
a
t
o
r

S
t
r
e
s
s

(
k
P
a
)
200
6 0 100 200 300 400 5 4 3 2 1 0 1 2 3 4
Axial Strain (%) Mean Efective Stress (kPa)
N = 38.5
Nakanoshima clay
C.S.R. = 0.30
Figure 5.30 Simulated results of cyclic behavior of an undrained triaxial test.
164 Computational modeling of multiphase geomaterials
REFERENCES
Aboshi, H. 1973. An experimental investigation on the similitude in the consoli-
dation of a soft clay including the secondary creep settlement, 8th ICSMFE,
Moscow, 4(3):88.
Adachi, T., and Oka, F. 1982. Constitutive equations for normally consolidated clay
based on elasto-viscoplasticity, Soils and Foundations, 22 (4):5570.
Adachi, T., and Oka, F. 1984. Constitutive equations for sands and overconsolidated
clays and assigned works for sand, Constitutive Relations for soils, Results of the
International Workshop on Constitutive Relations for Soils, G. Gudehus, F. Darve,
and I. Vardoulakis, eds., 68 September 1982, Grenoble, Balkema, 141157.
Adachi, T., and Oka, F. 1995. An elasto-plastic constitutive model for soft rock with
strain softening, Int. J. Numerical Anal. Methods in Geomechanics, 19: 233247.
Adachi, T., Oka, F., Hirata, T., Hashimoto, T., Nagaya, J., Mimura, M., and Pradhan,
T.B.S. 1995. Stress-strain and yielding characteristics of eastern Osaka clay,
Soils and Foundations, 35(3):113.
Adachi, T., and Oka, F., and Koike, M. 2003. An elasto-viscoplastic constitu-
tive model with strain softening for a sedimentary soft rock, Proceedings
of International Workshop on Prediction and Simulation Methods in
Geomechanics, Athens, F. Oka, I. Vardoulakis, A. Murakami, and T. Kodaka
eds., JGS, 58.
Table5.4 Material parameters of Nakanoshima clay
Initial void ratio, e
0
1.373 1.573
Compression index, 0.2173
Swelling index, 0.0344
Initial elastic shear modulus (kPa), G
0
22670
Stress ratio at compression, M
*
mc
1.143
Stress ratio at extension, M
*
me
1.061
Viscoplastic parameter, m 22.7
Viscoplastic parameter (1/s), C
1
1.00 10
5
Viscoplastic parameter (1/s), C
2
3.30 10
6
Structural parameter, = n /
maf mai
1
0.325
Structural parameter, 3.7 5.7
Hardening parameter, B
*
0
105
Hardening parameter, B
*
1
1.0
Hardening parameter, C
f
5 75
Reference value of viscoplastic strain (%),
n r
vp
( )
*
3.5
Strain-dependent parameter, a 10 1
Scalar hardening parameter, A
*
2
5.1
Scalar hardening parameter, B
*
2
2.6
Overconsolidation ratio, OCR 1.0
1
= OCR
mai m0
Elastoviscoplastic modeling of soil 165
Adachi, T., Oka, F., and Koike, M. 2005. An elasto-viscoplastic constitutive model
with strain-softening for soft sedimentary rocks, Soils and Foundations,
45(2):125133.
Adachi, T., Oka, F., and Mimura, M. 1987a. An elasto-viscoplastic theory for clay fail-
ure, Proc. 8th Asian Reg. Conf. Soil Mech. Found. Eng., Kyoto, 1, JSSMFE, 58.
Adachi, T., Oka, F., and Mimura, M. 1987b. Mathematical structure of an overstress
elasto-viscoplastic model for clay, Soils and Foundations, 27(3):3142.
Adachi, T., Oka, F., and Mimura, M. 1996. Modeling aspects associated with time
dependent behavior of soils, Measuring and modeling time dependent soil
behavior, Geotechnical Special Publication No. 61 ASCE, T.C. Sheahan and V.
Kaliakin, eds., 6195.
Adachi, T., Oka, F., and Poorooshasb, H.B. 1990. A constitutive model for frozen
sand, J. Energy Resour. Tech., ASME, 112:208212.
Adachi, T., and Okano, M. 1974. A constitutive equation for normally consolidated
clay, Soils and Foundations, 14(4):5573.
Akai, K., Adachi, T., and Ando, N. 1974. Stress-strain-time relationship of saturated
clay, Proc. JSCE, 225:5361 (In Japanese).
Armstrong, P.J., and Frederick, C.O. 1966. A mathematical representation of the
multiaxial Bauschinger effect, C. E. G. B. Report RD/B/N 731.
Arulanandan, K.-Shen, C.-K., and Young, R. B. 1971. Undrained creep behavior of a
coastal organic clay, Gotechnique, 21(4): 359375.
Asaoka, A., Nakano, M., Noda, T., and Kaneda, K. 2000. Delayed compression/
consolidation of natural clay due to degradation of soil structure, Soils and
Foundations, 40(3):7585.
Aubry, D., Kodaissi, E., and Meimon, Y. 1985. A viscoplastic constitutive equation
for clays including a damage law, Proceedings 5th International Conference on
Numerical Methods in Geomechanics, Nagoya, T. Kawamoto and Y. Ichikawa,
eds., Balkema, 1:421428.
Baladi, G.Y., and Rohani, B. 1984. Development of an elasto-viscoplastic constitu-
tive relationship for earth materials, In Mechanics of Engineering, C.S. Desai
and R.H. Gallerger, eds., John Wiley & Sons, New York, 2343.
Bjerrum, L. 1967. Engineering geology of Norwegian normally consolidated marine
clays as related to settlements of buildings, Gotechnique, 17(2):81118.
Bjerrum, L. 1973. Problems of soil mechanics and construction on soft clays,
state-of-the-art report to Session IV, Proc. 8th ICSMFE, Moscow, 3:111159.
Chaboche, J.L., and Rousselier, G. 1983. On the plastic and viscoplastic constitu-
tive equations Part I and Part II. J. Pressure Vessel Technol., Trans. ASME,
105:153164.
Dafalias, Y. 1982. Bounding surface elasto-viscoplasticity for particulate cohesive
media, Proceedings. of IUTAM Conference on Deformation and Failure of
Granular Materials, P.A. Vermeer and H.J. Luger, eds., Balkema, Rotterdam,
97107.
Di Benedetto, H., Tatsuoka, F. and Ishihara, M. 1997. Time-dependent deformation
characteristics of sand and their constitutive modeling, Soils and Foundations,
42(2):122.
Dragon, A. and Mroz, Z. 1979. A model for plastic creep of rock-like materi-
als accounting for the kinematics of fracture, Int. J. Rock Mech. Min. Sci.
Geomech., Abstr., 16:253259.
166 Computational modeling of multiphase geomaterials
Drucker, D.C. 1959. A defnition of stable inelastic materials, J. Appl. Mech., Trans.
ASME, 26:101106.
Duvaut, G., and Lions, J.L. 1976. Inequalities in mechanics and physics, Springer-
Verlag, Berlin (Les inequations en mechanique et en physique, Dunod, Paris).
Eyring, H. 1936. Viscosity, plasticity, and diffusion as example of absolute reaction
rates, J. Chem. Phys., 4(4):283291.
Fahey, M. 1992. Shear modulus of cohesionless soil:variation with stress-strain level,
Can. Geotech. J., 29(1):157161.
Febro-Cordero, E., and Mesri, G. 1974. Infuence of testing conditions on creep
behavior of clay, I Report No. FRA-ORD&D-75-29, UILU-ENG-74-2031.
Finn, L., and Shead, D. 1973. Creep and creep rupture of a sensitive clay, 8th
ICSMFE, Moscow, 1(1):135142.
Graham, J., Crooks, J.H.A., and Bell, A.L. 1983. Time effects on the stress-strain
behavior of soft marine clays, Gotechnique, 33(3):327340.
Hardin, B.O., and Drnevich, V.P. 1972. Shear modulus and damping in soils: design
equation and curves. J. Soil Mech. Foundations, ASCE, 98(SM7):667692.
Hirst, T.J. 1968. The infuences of compositional factors on the stress-strain-time
behavior of soils, PhD thesis, University of California, Berkeley.
Hohenemser, K., and Prager, W. 1932. ber die anstze der mechanik isotroper kon-
tinua, ZAMM, 12:216226.
Hori, M. 1974. Fundamental study of wave propagation characteristics through
soils, PhD thesis, Kyoto University.
Ishihara, K. 1996. Soil Behavior in Earthquake Geotechnics. Oxford University
Press, 85152.
Kabbaj, M., Oka, F., Leroueil, S., and Tavenas, F. 1985. Consolidation of natural
clays and laboratory testing, Consolidation of Soils: Testing and Evaluation,
ASTM Special Technical Publication 892, R.N. Yong and F.C. Townsend, eds.,
378404.
Kabbaj, M., Tavenas, F., and Leroueil, S. 1988. In situ and laboratory stress-strain
relationships, Gotechnique, 38(1):83100.
Karunawardena, A. 2007. Consolidation analysis of Sri Lankan peaty clay using
elasto-viscoplastic theory, PhD thesis, Kyoto University.
Katona, M.G. 1984. Evaluation of viscoplastic cap model, J. Geotech. Eng., ASCE,
110(8):11061125.
Kimoto, S. 2002. Constitutive models for geomaterials considering structural
changes and anisotoropy, PhD thesis, Kyoto University, Japan.
Kimoto, S., and Oka, F. 2005. An elasto-viscoplastic model for clay considering
destructuralization and consolidation analysis of unstable behavior, Soils and
Foundations, 45(2):2942.
Kimoto, S., Oka, F., and Higo, Y. 2004. Strain localization analysis of elasto-
viscoplastic soil considering structural degradation, Comput. Meth. Appl. Mech.
Eng., 193:28452866.
Kimoto, S., Oka, F., and Karunawardena, A. 2011. An elasto-viscoplastic model
based on the modifed Cam-clay model, Proceedings of Annual Meeting of
JSCE, 307308 (In Japanese).
Kimoto, S., Oka, F., Watanabe, T., and Sawada, M. 2007. Improvement of the visco-
plastic potential and static yield function for the elasto-viscoplastic constitutive
model. Proceedings of 62nd Annual Meeting of JSCE, Hiroshima, 637638.
Elastoviscoplastic modeling of soil 167
Kokusho, T., Yoshida, Y., and Esashi, Y. 1982. Dynamic properties of soft clays for
wide strain range, Soils and Foundations, 22:118.
Kondner, R.L., and Ho, M.M.K. 1965. Viscoelastic response of a cohesive soil in the
frequency domain, Trans. Soc. Rheology, 9(2):329342.
Kovacs, W.D., Seed, H.B., and Chan, C.K. 1971. Dynamic modulus and damping
ratio for a soft clay, J. Soil Mech. Foundations, ASCE, 97(SM1):5975.
Kutter, B.L., and Sathialinggam, N. 1992. Elastic-viscoplastic modeling of the rate-
dependent behaviour of clays, Gotechnique, 42(3):427441.
Leroueil, S. 1988. Recent developments in consolidation of natural clays, Tenth
Canadian Geotechnical Colloquium, Can. Geotech. J., 25:85107.
Leroueil, S., and Hight, D.W. 2003. Behaviour and properties of natural soils and soft
rocks, In Characterization and Engineering Properties of Natural Soils, T.S. Tan,
K.K. Phoon, D.W. Hight, and S. Leroeuil, eds., Swets and Zeitlinger, Lisse, 29254.
Leroueil, S., Kabbaj, M., Tavenas, F., and Bouchard, R. 1985. Stress-strain-strain
rate relation for the compressibility of sensitive natural clays, Gotechnique,
36(2):288290.
Leroueil, S., Samson, L., and Bozozuk, M. 1983. Laboratory and feld determination
of preconsolidation pressures at Gloucester, Can. Geotech. J., 20(3):477490.
Leroueil, S., Tavenas, F., Brucy, F., La Rochelle, P., and Roy, M. 1979. Behavior of
destructed natural clays, J. Geotech. Eng., ASCE, 115(6):759778.
Maleki, M. and Cambou, B., 2009. A cyclic elastoplastic-viscoplastic constitutive
model for soils, Geomech. Geoeng.: Int. J., 4(3):209220.
Matsui, T., and Abe, N. 1985. Elasto/viscoplastic constitutive equation of nor-
mally consolidated clays based on fow surface theory, Proceedings of 5th
International Conference on Numerical Methods in Geomechanics, Nagoya,
T. Kawamoto and Y. Ichikawa, eds., Balkema, Rotterdam, 1:407413.
Mesri, G., and Choi, Y.K. 1979. Excess pore water pressure during consolidation,
Proc. 6th Asian Regional Conf. on SMFE, 1:151154.
Mesri, G., Shahien, M., and Feng, T.W. 1995. Compressiblity parameters during
primary consolidation, Proceedings International Symposium on Compression
and Consolidation of Clayey Soils, IS-Hiroshima, H. Yoshikuni and O.
Kusakabe, eds., Balkema, Rotterdam, 2:10211037.
Mirjalili, M. 2010. Numerical analysis of a large-scale levee on soft soil deposits
using two-phase fnite deformation theory, PhD thesis, Kyoto University.
Mirjalili, M., Shahbodagh Khan, B., Oka, F. and Kimoto, S. 2011. Dynamic Strain
Localization Analysis of Elasto-Viscoplastic Soil Using Updated Lagrangian
Finite Element Formulation, Proceedings of the 2nd International Symposium
on Comutational Geomechanics, S. Pietruzczak and G.N. Pande eds.,
International Center for Computational Engineering, 203210.
Mitchell, J.K. 1986. Practical problems from surprising soil behavior, 20th Terzaghi
Lecture, J. Geotech. Eng., ASCE, 112(3):259289.
Mitchell, J.K. Campanella, R.G., and Singh, A. 1968. Soil creep as a rate process,
ASCE J. SMFD, 94(1):231254.
Modaressi, H., and Laloui, L., 1997. A thermo-viscoplastic constitutive model for
clay, Int. J. Numer. Anal. Meth. Geomech., 21:313335.
Murayama, S., and Shibata, T. 1964. Flow and stress relaxation of clays, Proceedings
of IUTAM Symposium on Rheology and Soil Mech., J. Kravtchenko and P.M.
Sirieys, eds., Springer-Verlag, Berlin, 99129.
168 Computational modeling of multiphase geomaterials
Murayama, S. 1983. Formation of stress-strain-time behavior of soils under devia-
toric stress condition, Soils and Foundations, 23(2): 4357.
Nova, R. 1982. A viscoplastic constitutive model for normally consolidated clay.
Proceedings of IUTAM Conference on Deformation and Failure of Granular
Materials, P.A. Vermeer and H.J. Luger, eds., Blakema, Rotterdam, 287295.
Ogisako, E., Nishio, S., Denda, A., Oka, F., and Kimoto, S. 2007. Simulation of tri-
axial compression tests on soil samples obtained from seabed ground in deep
sea by elasto-viscoplastic constitutive equation, Proceedings of the Seventh
ISOPE Ocean Mining & Gas Hydrates Symposium, Lisbon, Portugal, Chung
and Komai, eds., ISOPE, 6368.
Oka, F. 1979. Constitutive theory for solid-fuid mixture and its application to stress
wave propagation through cohesive soil, Trans. JSCE, 272:117130.
Oka, F. 1981. Prediction of time-dependent behaviour of clay, Proceedings of 10th
International Conference on Soil Mechanics and Foundation Engineering,
Stockholm, Balkema, 1:215218.
Oka, F. 1982. Constitutive equations for granular materials in cyclic loadings,
Proceedings of IUTAM Conference on Deformation and Failure of Granular
Materials, Delft, P.A. Vermeer and H.J. Luger eds., Balkema, 297306.
Oka, F. 1985. Elasto/viscoplastic constitutive equations with memory and internal
variables, Computers and Geotechnics, 1:5969.
Oka, F. 1992. A cyclic elasto-viscoplastic constitutive model for clay based on the
non-linear-hardening rule, Proceedings of 4th International Symposium on
Numerical Models in Geomechanics, G.N. Pande and S. Pietruszczak, eds.,
Swansea, UK, Balkema, Rotterdam, 1:105114.
Oka, F., and Adachi, T. 1985. A constitutive equation of geologic materials with
memory, Proceedings of 5th International Conference on Numerical Methods
in Geomechanics, Balkema, 1:293300.
Oka, F., Adachi, T., and Yashima, A. 1994. Instability of an elasto-viscoplastic consti-
tutive model for clay and strain localization, Mech. Mater., 18:119129.
Oka, F., Adachi, T., and Yashima, A. 1995. A strain localization analysis of clay using
a strain softening viscoplastic model, Int. J. Plasticity, 11, 5:523545.
Oka, F., Feng, H., Kimoto, S., Kodaka, T., and Suzuki, H. 2008. A numerical simu-
lation of triaxial test of unsaturated soil at constant water and air content
by using an elasto-viscoplastic model, In Unsaturated Soils: Advances in Geo-
Engineering, D.G. Toll, C.E. Augrade, D. Gallipoli, and S.J. Wheeler, eds.,
Taylor and Francis Group, London, 735741.
Oka, F., Higo, Y., and Kimoto, S. 2002. Effect of dilatancy on the strain localization
of water-saturated elasto-viscoplastic soil, Int. J. Solids Struct., 39:36253647.
Oka, F., Kimoto, S., and Nguyen, Q.H. 2011. Elasto-viscoplastic constitutive mod-
eling of soft sedimentary rocks, Proceedings of 15th European Conference
on Soil Mechanics and Geotechnical Engineering, A. Anagnostopoulos, M.
Pachakis, and C. Tsatsanifos, eds., IOS press, Amsterdam, 1:563568.
Oka, F., Kodaka, T., and Kim, Y.-S. 2003a. A cyclic viscoelastic-viscoplastic consti-
tutive model for clay and liquefaction analysis of multi-layered ground, Int.
J.Numer. Analy. Meth. Geomech., 28:131179.
Oka, F., Kodaka, T., and Kim, Y.-S. 2004. A cyclic viscoelastic-viscoplastic consti-
tutive model for clay and liquefaction analysis of multi-layered ground, Int.
J.Numer. Anal. Meth. Geomech., 28(2):131179.
Elastoviscoplastic modeling of soil 169
Oka, F., Kodaka, T., Kimoto, S., Ichinose, T., and Higo, Y. 2005. Strain localization of
rectangular clay specimens under undrained triaxial compression conditions.
Proc. 16th ICSMGE, 2:841844.
Oka, F., Tavenas, F., and Leroueil, S. 1991. An elasto-viscoplastic FEM analysis of sen-
sitive clay foundation beneath embankment, Proceedings of 7th International
Conference on Computer Methods and Advanced in Geomechanics, Cairns,
G.Beer, J.R. Booker, and J.P. Carter, eds., Balkema, 2:10231028.
Oka, F., Uzuoka, R. Tateishi, A., and Yashima, A. 2003b. A cyclic elasto-plastic model
for sand and its application to liquefaction analysis, Constitutive Modeling of
Geomaterials, selected contributions from the Frank L. DiMaggio Symposium,
Inelastic Behavior Committee Engineering Mechanics Division, ASCE, CRC
Press, 7599.
Oka, F., Yashima, A., Tateishi, A., Taguchi, Y., and Yamashita, S. 1999. A cyclic
elasto-plastic constitutive model for sand considering a plastic-strain depen-
dence of the shear modulus, Gotechnique, 49(5):661680.
Peirce, D., Shih, C.F., and Needleman, A. 1984. A tangent modulus method for rate
dependent solids, Comput. Struct., 18(5):845887.
Perzyna, P. 1963. The constitutive equations for work hardening and rate sensitive
plastic materials, Proc. Vibrational Problems, Warsaw, 4(3):281290.
Phillips, A., and Wu, H.-C. 1973. A theory of viscoplasticity, Int. J. Solids Struct.,
9:1530.
Potts, D.M., and Zdravkovic, L. 1999. Finite Element Analysis in Geotechnical
Engineering: Theory, Thomas Telford, London.
Richardson, A.M. Jr., and Whitman, R. 1963. Effect of the strain rate upon und-
rained shear resistance, Gotechnique, 13(4):310324.
Roscoe, K.H., and Burland, J.B. 1968. On the generalised stress-strain behaviour
of wet clay, In Engineering Plasticity, J. Heyman and F.A. Leckie, eds.,
Cambridge University Press, 535609.
Roscoe, K.H., Schofeld, A.N., and Thurairajah, A. 1963. Yielding of clays in states
wetter than critical, Gotechnique, 13(3):211240.
Saito, M. 1992. Jissho Doshitsu Kogaku, Gihodo Syuppan Co. Ltd., 156 (In
Japanese).
Saito, M., and Uezawa, H. 1961. Failure of soil due to creep, 5th ICSMFE, Paris,
1:315318.
Sllfors, G. 1975. Preconsolidation pressure of soft high-plastic clays, PhD thesis,
Chalmers University of Technology, Gothenburg, Sweden.
Sawada, K., Yashima, A., and Oka, F. 2001. Numerical analysis of deformation of
saturated clay based on Cosserat type elasto-viscoplastic model, J. Soc. Mater.
Sci. Japan, 50(6):585592 (In Japanese).
Seed, H.B., Wong, R.T., Idriss, I.M., and Tokimatsu, K. 1986. Moduli and damp-
ing factors for dynamics analyses of cohesionless soils, J. Geotech. Eng.,
112(11):10161032.
Sekiguchi, H. 1977. Rheological characteristics of clays, Proc. 9th ICSMFE, Tokyo,
1:289292.
Sekiguchi, H., and Ohta, H. 1977. Induced anisotropy and time dependency in Clays,
Proc. Speciality Session 9 9th ICSMFE, Tokyo, JSSMFE: 229238.
Shahbodagh, K. B. 2011. Large deformation dynamic analysis method for partially
saturated elasto-viscoplastic soils, PhD thesis, Kyoto University.
170 Computational modeling of multiphase geomaterials
Shen, C.-K., Arulanandan, K., and Smith, W.S. 1973. Secondary consolidation
and strength of a clay, J. Soil Mech. Foundation Div., Proc. ASCE, 99 (SM
1):95110.
Shigematsu, H. 2002. Study on the microstructure and mechanical behavior of natu-
ral sedimentary soils with aging effect, Doctoral thesis, Gifu University, Japan
(In Japanese).
Singh, A., and Mitchell, J.K. 1968. General stress-strain-time function for soils,
J.Soil Mech. Foundat. Eng., ASCE, 94, SM1:2146.
Singh, A., and Mitchell, J.K. 1969. Creep potential and creep rupture of soils, Proc.
7th ICSMFE, Mexico, 379384.
Suklje, L. 1957. The analysis of the consolidation process by the isotaches method,
Proc. 4th ICSMFE, London, 1:200206.
Suklje, L. 1969. Rheological Aspects of Soil Mechanics, Wiley-Interscience, London.
Terzaghi, K. 1944. Ends and means in soil mechanics, Eng. J. Can., 27:608613.
Vaid, Y. P., and Campanella, M. 1977. Time-dependent behavior of undisturbed clay,
J. Geotech. Eng. Div., Proc. ASCE, 103 (GT7):693709.
Walker, L.K. 1969. Secondary compression in the shear of clays, J. Soil Mech.
Foundat. Div., Proc. ASCE, 95 (SM1):167188.
Wang, G.X., and Kuwano, J. 1999. Modeling of strain dependency of shear modulus
and damping of clayey sand, Soil Dynamics Earthquake Eng., 18:463471.
Yashima, A., Shigematsu, H., Oka, F., and Nagaya, J. 1999. Mechanical behavior
and micro-structure of Osaka upper-most Pleistocene marine clay, J. Geotech.
Eng., JSCE, 624 (III-47):217229 (In Japanese).
Yin, J.-H., and Graham, J. 1999. Elasto visco-plastic modelling of the time-dependent
stress-strain behavior of soils. Can. Geotech. J., 36(4):736745.
Yin, Z.Y., and Karstunen, M. 2008. Infuence of anisotropy, destructuration and vis-
cosity on the behavior of an embankment on soft clay, Proceedings of the 12th
IACMAG, 1Goa, India, D.N. Singh, ed., 47294735.
Yong, R.N., and Japp, R.D. 1969. Stress-strain behavior of clays in dynamic com-
pression, Vibrational effects on earthquakes on soil and foundations, ASTM,
STP, 450:233262.
Zienkiewicz, O.C., Humpheson, C., and Lewis, R.W. 1975. Associated and non-
associated viscoplasticity and plasticity in soil mechanics, Gotechnique,
25(4):671689.
171
Chapter 6
Virtual work theorem and
finite element method
In the present chapter, the virtual work theorem is reviewed and the fnite
element method for multiphase geomaterials is presented.
6.1 VIRTUAL WORK THEOREM
6.1.1 Boundary value problem
Finding the displacement and the stress felds that satisfy the constitutive
equations, such as the elastoplastic model, the straindisplacement rela-
tions, the equations of motion, and the initial and the boundary conditions,
is called an elasticplastic boundary value problem.
When governing equations such as the equations of motion are formu-
lated in an incremental form, the boundary value problem is called an
incremental boundary value problem and the incremental formulation is
usually applied to nonlinear problems.
Equilibrium equationsIn body B, neglecting the acceleration term,
the following equations hold:

ji j i
F
,
+ = 0
(6.1)
where
ij
i j ( , , , ) = 1 2 3 are the components of the symmetric stress ten-
sor assuming the equilibrium of the moment (
ij ji
= ). i F is a body
force and subscript i denotes

x
i
.
Compatibility conditionsIn body B,

ij i j j i
u u = +
, ,
1
2
( )
(6.2)
where
ij
is the strain tensor and u
i
is the displacement vector.
172 Computational modeling of multiphase geomaterials
Boundary conditionsAs shown in Figure 6.1, boundary S can be
divided into two parts, namely, stress traction boundary, S
t
; and dis-
placement boundary, S
u
.

S S S
t u
= +
(6.3)
Stress traction boundary conditionsFor S
t

T n
T i ji j i
= =
(6.4)
where n
i
is a component of the unit outward normal vector on the
boundary and T
i
is the component of the stress traction vector.
Displacement boundary conditionsFor S
u

u
u i i
=
(6.5)
Constitutive equations (stress-strain relations)For linear elastic
bodies, the model can be expressed as


ij ijkl kl
E =
(6.6)
where E
ijkl
is a fourth-order elastic coeffcient tensor.
The displacements and the displacement rates that satisfy the bound-
ary conditions are called kinematically admissible displacements or dis-
placement rate felds, and the strain and the strain rate felds that satisfy
S
t
T
i
VV
S
u
Figure 6.1 Body B and boundary conditions.
Virtual work theorem and finite element method 173
the compatibility conditions and the aforementioned boundary conditions
are called kinematically admissible strain and strain rate felds. The stress
felds that satisfy the stress traction boundary conditions are called stati-
cally admissible.
When the equilibrium Equation (6.1) at time t holds and Equations (6.1)
to (6.4) hold between time t and time t + dt, the boundary value problem is
called an incremental boundary problem.
For incremental boundary value problems, Equations (6.1) to (6.6)
become as follows:

ji j
i F ,
+ =

0
(6.7)
where the super dot denotes the time differentiation and
ij
i j

( ) , = , , 1 2 3
is
the stress rate tensor.

ij i j j i u u

= +
, ,
1
2
( ) (6.8)
where
ij

is the stain rate tensor and i u is the displacement rate vector.

On stress traction boundary, S
t
,

ji j
i
n
T

=
(6.9)
where n
i
is the unit outward normal vector on the boundary and
i T

is the
stress rate vector.
On displacement rate boundary, S
u
,

i i u u
=
(6.10)

ij ijkl kl
E

= (6.11)
6.1.2 Virtual work theorem
The weak form of the equilibrium equations, with respect to the kine-
matically admissible displacements and strain felds for continuously
differentiable body B, are derived next. On displacement boundary, S
u
,
the displacement is set to be u
i
. Herein, for such arbitrary kinematically
admissible strain and displacement felds as

and u
i

, respectively, we
consider the weak form and assume
ij
and u
i

are single-valued and con-

tinuous in B.
174 Computational modeling of multiphase geomaterials
Since boundary S is divided into two parts, namely, stress boundary, S
p
,
and displacement boundary, S
u
, we have

S S S
F
u dv
T
n u ds
u p
V
ij j i i
Sp
i ij j i
= +
+ + =

,

( ) ( ) 00
(6.12)
where is an arbitrary scalar.
Using the Gauss theorem, we obtain

V
ij i j
S
ij j i
v
ij j i
u dv u n ds u dv

,

,

= (6.13)
Considering the symmetry of the stress tensor,

F
u dv u n ds u dv
T
u ds
n u ds
V
i
i
Su
ij j i
V
ij i j
Sp
j
j
St
ij j i
(1 ) 0
**

+ +
+ =

(6.14)
If we set = 1 and u u
j j

= on S
u
, Equation (6.14) becomes

dv
T
u ds u n ds
F
u dv
v
ij ij
St
i
i
Su
ij j i
v
i
i

= + +

(6.15)
Conversely, if Equation (6.15) holds, the boundary conditions and the
equilibrium equations can be derived. Hence, we can solve Equation (6.15)
instead of solving the equilibrium equations with the boundary conditions.
Now for the other kinematically admissible strain feld,
***
, and the
displacement feld, u
i
***
, we can consider a similar weak form. Taking
the difference in the weak form and Equation (6.15), we get a new weak
form, namely,

dv
T u
ds
F u
dv
v
ij ij
St
i i
v
i i

= +

(6.16)
where
i i i u
u u

=
and u
u i i
= on S
u
are the additional conditions.
Equation (6.1) includes the frst-order spatial derivative of the stress, but
Equation (6.15) does not.
Virtual work theorem and finite element method 175
As mentioned earlier, Equation (6.10) is called a weak form since the
order of the differentiability required has been weakened. It is worth
noting that the virtual work theorem does not depend on the constitutive
model.
When the displacement rate may be discontinuous on , the virtual work
theorem for incremental boundary value problems is expressed by

dv
F
u
dv
T
u
ds
u
n ds n
u
ds
v
ij ij
v
i
i
St
i
i
Su
ij i
j ij j j
[ ]




= + +

+

(6.17)
where ij

is the strain rate tensor and i u

is the velocity. The brackets [ ]

denote the discontinuity across .
The equation equivalent to Equation (6.16) is given by

[ ]







dv
F u
dv
T u
ds n
u
ds
v
ij
ij
v
i i
St
i i
ij j
j

= + +

(6.18)
When discontinuity does not exist, we have







dv
T u
ds
F u
dv
v
ij
ij
St
i i
v
i i

= +

(6.19)
where
i i i u u u

=

and the supplementary conditions are

i i u u
=
on S
u
.
6.2 FINITE ELEMENT METHOD
6.2.1 Discretization of equilibrium equation
The fnite element method is an approximation analysis method used
to solve boundary value problems based on variational principles, the
weighted residual method, the Galerkin method, and so on. In the fnite
element method, the body is divided into a fnite number of elements. If the
compatibility conditions and the completeness conditions are satisfed in
the formulation of the fnite element method, the solution converges to an
accurate solution when the size of the element becomes small.
The formulation given later is for a fnite element method; it is based on
the virtual work theorem and is called the displacement method, in which
the displacements are taken as unknowns. In the displacement method, the
displacements or displacement increments in the element are interpolated
176 Computational modeling of multiphase geomaterials
by the values at nodal points. By solving the equations for all of the ele-
ments under the boundary conditions, the displacement increments at the
nodal points are obtained for the incremental problems. The stain incre-
ments are then calculated by the displacement increments, and the stress is
calculated using constitutive equations. Thereafter, the calculation moves
to the next step.
Let

ij
denote the strain increment tensor, u
i
the displacement increment
vector,
ij

the virtual strain increment tensor,

i u the virtual displacement
increment vector, u
w
the pore water pressure in the element, and the
effective stress increment tensor.
The virtual work theorem for a rate type of formulation is expressed in
matrix form as

V
T
V
T
V
T
W
dV dV u dV

= + {

} { } ( {

} { } {

} { } ))

= +

V
T
b
S
T
s
u F dV u T dS { } { } { } { }
(6.20)
In the preceding equation, the strain increment vector, the total stress
vector, the effective stress vector, the displacement increment vector, the
pore pressure vector, the body force vector, and the surface force vector for
two-dimensional problems are as follows:
Total increment stress vector: { } { } { } ; = , ,
T
xx yy xy
Effective increment stress vector:
T
xx yy xy
{ } { } { } ; = , ,
Displacement increment vector: u u u u
T
x y
{ } { } { } ; = , , { } { } ; = u u
T
{

} ,
x y u u
Strain increment vector: { } { } { } ; = , ,
T
xx yy xy
Pore pressure increment vector: { } { } { } u u u u
W W
T
W W
; = , ,0
Body force increment vector: { } F
b
Surface force increment vector: { } T
s
The displacement increment {u} at a point in the element is approxi-
mated by the interpolation function as

{ } [ ]{ }, { } [ ]{

} = =

u N u u N
u
(6.21)
where [N] is the interpolation function, and { } [ ]{

} =

u N
u
and { }
*
u are
the nodal increment displacement vectors.
Virtual work theorem and finite element method 177
The strain increment vector, {}, and volumetric strain,
v
, of the ele-
ment are given by

B u B
u
{ } [ ]{ }, { } [ ]{

} = =

(6.22)

B
u
B
u v v
T
v v
T
{ } [ ] {

}, [ ] {

} = =

(6.23)
where [B] is a strain-nodal displacement matrix and [ ] B
v
T
is a volumetric
strain-nodal displacement matrix.
Using Equations (6.21) and (6.22), Equation (6.20) becomes

V
T T
V
T
W u
B dV u dV

+

{

} [ ] { } {

} { }

= +


V
T T
b
S
T T
s u
N F dV
u
N T dS {

} [ ] { } {

} [ ] { } (6.24)
In view of Equation (6.23), the second term on the right-hand side of the
preceding equation becomes

V
T
W
V
xx yy
xy
W
W
u dV
u
u

= , , {

} { } {
}


0
dV


u dV u dV
u B u dV
W
V
xx yy w v
V
w v
T
V
{ }
[ ] { }
*

=

+

=

=
(6.25)
Let the constitutive model be written as

D D
vp
{ } [ ]{ } { } { } [ ]{ } = , =

(6.26)
where [D] is the elastic coeffcient matrix, { }

is the viscoplastic relax-

ation stress vector, and { }
vp

is the viscoplastic strain increment vector.
As to the estimation method for the viscoplastic increment vector, there
are two methods, namely, the explicit scheme and the implicit scheme.
178 Computational modeling of multiphase geomaterials
Substituting Equation (6.26) into Equation (6.24) yields

V
T T
W
V
T
v u
B D dV u
u
B

+

{

} [ ] ([ ]{ } { }) {

} [ ] ddV
u
N F dV
u
N T dS
V
T T
b
S
T T
S
= +


{

} [ ] { } {

} [ ] { }
(6.27)
Since
{

}

u
T
is arbitrary, except for the boundary where the displace-
ments are prescribed, by substituting Equation (6.22), we have

B D B u dV K u B dV
N F dV N T dS
V
T
V W
V
T
V
T
b
S
T
s
[ ] [ ][ ]{ } { } [ ] { }
[ ] { } [ ] { }


+ =
+ +


K B dV
v
V
v
{ } [ ]

=
(6.29)
For solving Equation (6.28), we need to discretize the equations of motion
for the liquid phase.
For simplicity, let us assume zero pore water pressure, = u
W
0. Thus,
we have

[ ]{ } { } K u Q =

(6.30)

[ ] [ ] [ ][ ] K B D B dV
V
T
=

(6.31)

{ } [ ] { } [ ] { } [ ] { } = + +


Q N F dV N T dS B dV
V
T
b
S
T
s
V
T
(6.32)
where {Q} is the force vector and [K] is the stiffness matrix. [K] is in gen-
eral nonlinear, and Equation (6.30) has been solved by an iteration method
such as NewtonRaphson method (Zienkiewicz 1977).
Virtual work theorem and finite element method 179
6.2.2 Discretization of continuity equation
When we solve the solidwater saturated problem, u
w
0 in Equation (6.28),
we need to simultaneously solve the equilibrium equations for the pore water.
In the following section, the standard formulation for the continuity equation
will be presented based on the weighted residual method, subdomain colloca-
tion method, and Galerkin method. Herein, the simplest method based on the
fnite difference method is presented using a fnite element grid.
As is described by Equation (2.45), a continuity equation can be derived
from the equation of motion disregarding the acceleration term, the com-
pressibility of water, and the mass conservation law as

k u
x
t
w
w
i
kk

2
2
(6.33)
where k is the permeability coeffcient and
w
is the unit weight of water

f
g.
In view of Figure6.2, discretizing Equation (6.33) gives

{ } { } ( ) ( ) K u u t t u t t
v
T
w
i
i wi
= + + +

1
4
(6.34)

k tb
s
i
i
w i
i
i
1
4

, =
=
(6.35)
x
1
u
w1
u
w4
u
w2
u
w3
S
1
S
4
S
3
S
2 b
2
,

b
4
b
1
, b
3
x
2
Figure 6.2 Discretization of pore water pressure.
180 Computational modeling of multiphase geomaterials
In the derivation of Equation (6.34) from Equation (6.33), it is worth noting
that the left-hand side of Equation (6.34) is a change in volume of the element.
Equations (6.28) and (6.34) can be written in matrix form as

K K
K
u
u t t
K u t Q
u t t
v
v
T
w
v w
i wi
i
[ ] { }
{ }
{ }
( )
{ } ( ) { }
( )
1
4

,
, +
=
+
+

=
(6.36)
where u u t dt u t
* * *
( ) ( ) = + .
Solving the aforementioned equations under the appropriate boundary
and initial conditions yields the solution of consolidation problems.
6.2.3 Interpolation function
In the fnite element method, the body is divided into elements, as shown in
Figure6.3, in which triangular elements are adopted.
The state variable Q in the element is interpolated by the value of Q
i
(i is
number of nodes) at the nodes of the element:

Q QN
i
n
i i
=
=

1
(6.37)
When the interpolation function of the element is continuous and differ-
entiable to the necessary degree, the element is compatible and this condi-
tion is called compatibility. The compatibility condition ensures that the
displacement in the elements and the displacements along the edge of the
elements are continuous (Bathe 1996; Belytschko et al. 2000).
Node
Element
Figure 6.3 Division of the body into fnite elements.
Virtual work theorem and finite element method 181
The fnite element solution converges to the exact solution if the fnite
element and the mesh are compatible and complete. The completeness is
defned by the basis functions, that is, the interpolation functions are com-
plete in the solution space and the reproduction of the constant strain and
the rigid body motions. For example, the displacement in the element is
approximated by

u N u
i
n
i i
=
=

1
(6.38)
If u
i
is constant, we have

i
n
i
N
=

=
1
1
(6.39)
This condition ensures the completeness and is called the partition of unity.
If the interpolation function or the shape function satisfes the conditions
of compatibility and completeness, it is known that the approximated solu-
tion converges to an accurate solution.
In the following, we will consider the two-dimensional plane strain prob-
lem as an example, namely, a triangular element and a four-node quadri-
lateral element.
6.2.4 Triangular element
Figure 6.4 shows a triangular element with three nodes, i, j, and k. For
plane stress and strain problems, the displacement at a node has two com-
ponents, namely, u
i
is the component in the x direction and v
i
is the compo-
nent in the y direction of node i.
The components of the displacement vector at the three nodes constitute
a displacement vector {} as

{ } [ ]
T
i i j j k k
u v u v u v = , , , , , (6.40)
When we assume the linear variation in the displacements of the element,
the displacements are expressed by the linear polynomials as

u x y = + +
1 2 3 (6.41)

v x y = + +
1 2 3 (6.42)
182 Computational modeling of multiphase geomaterials
If the displacements at nodes, i,j,k are expressed by ( ), ( ), u u u v v v
i j k i j k
, , , ,
we have

u x y v x y
u x y v x y
u x y v x y
i i i i i i
j j j j j j
k k k k k k
,
,
,
1 2 3 1 2 3
1 2 3 1 2 3
1 2 3 1 2 3
= + + = + +
= + + = + +
= + + = + +
(6.43)
in which x
i
are the nodal coordinates.
We can determine six parameters from Equation (6.43) and then we have
the following displacement in the x direction by inserting the six parameters:

u a b x c y u a b x c u a b x c y
i i i i j j j j k k k
= + + + + + + + +
1
2
{( ) ( ) ( )uu
k
} (6.44)

a x y x y b y y c x x
i j k k j i j k i k j
= , = , = (6.45)
where

2
1
1
1
= det
x y
x y
x y
i i
j j
k k
(6.46)
where is the area of the triangle (i, j, k).
O
x
i
y
i
i
k
j
x
y
Figure 6.4 Triangular element.
Virtual work theorem and finite element method 183
In a similar manner, we have displacement v in the y direction as

v a b x c y v a b x c y v a b x c y
i i i i j j j j k k k
= + + + + + + + +
1
2
{( ) ( ) ( )) } v
k
(6.47)
The strain vector {} for the plane problem is given by

{ }

= =

+

xx
yy
xy
u
x
v
y
u
y
v
x
(6.48)
where
xy xy
= 2 and
xy
is the engineering strain.

x xx y yy
, . = =

{ } =
1
2
0 0 0
0 0 0
b b b
c c c
c b c b c b
i j k
i j k
i i j j k k
u
v
u
v
u
v
i
i
j
j
k
k
= [ ]{ } B
(6.49)
where [B] is called the [B] matrix.
6.2.5 Isoparametric elements
A shape function of the fnite element is defned by the transformation
between the master element and the real elements as

x N x
i n in
= ( , ) (6.50)
where x
in
is the ith coordinate of the node, x
i
is the ith coordinate of the
point in the element, N
i
( , ) is a one-to-one mapping of a vector of the
nodal coordinates of the parent element onto that of the nodal coordinates
of the real element shown, and (, ) are the natural coordinates shown in
Figure6.5.
184 Computational modeling of multiphase geomaterials
We can use shape functions as the interpolation functions in the ele-
ments. In that case, the elements are called isoparametric elements.
Hence, the displacement vector is expressed by the displacement at the
nodes.

u N u
i
n
i i
=
=

1
(6.51)
= 1
= 1

= 1 = 1

2
3 4
1
(a)
O
y
x
1 (x
1
, y
1
)
2 (x
2
, y
2
)
3 (x
3
, y
3
)
4 (x
4
, y
4
)
(b)
Figure 6.5 (a) Parent element in local coordinates. (b) Four-noded quadrilateral
element.
Virtual work theorem and finite element method 185
Let us consider the parent square element and the real quadrilateral ele-
ment shown in Equation (6.51), which can be expressed by the components as

x N N N N
x
x
x
x
y N N N N
y
y
y
y
[ ] , [ ]
1 2 3 4
1
2
3
4
1 2 3 4
1
2
3
4
= , , , = , , ,
(6.52)

x N x y N y
i
i i
i
i i
,
1
4
1
4

= =
= =
(6.53)
In the fnite element method, the variables, such as the displacement, are
approximated by the basic interpolation functions as

u p a(x) =
T
, u() = {P()}{a} (6.54)
where {P()} is a vector composed of interpolation functions and {a} is a
coeffcient vector.
In general, polynomial functions are adopted for the interpolation functions.
For the four-node element, we will use the following interpolation functions:
{P()} = {1,,,} (6.55)
For the parent element shown in Figure6.5, the vectors are described as

P P ( , ) ( , , , ) = = 1 1 1 1 1 1
1

for node No. 1, (6.56)
The displacement at node n is given by

{ } [ ]{ } u P a
n n
=
(6.57)

u
u
u
u
1
2
3
4
1 1 1 1
1 1 1 1
1 1 1 1
1 1 1 1
=



a
a
a
a
1
2
3
4
(6.58)
186 Computational modeling of multiphase geomaterials
Hence, we have the coeffcient vector as

{ } [ ] { } a P u
n n
=
1
(6.59)
Finally, we have obtained the interpolation functions {N()}.

u P P u
N u
n n
n
( ) { ( )}[ ] { }
{ ( )}{ }

=
=
1
(6.60)
where { ( )} { ( )}[ ] N P P
n
=
1
.

P
n
=



1 1
4
1 1 1 1
1 1 1 1
1 1 1 1
1 1 1 1
(6.61)

{ ( )} { , , , } N =



1
1
4
1 1 1 1
1 1 1 1
1 1 1 1
1 1 1 1
(6.62)
Hence,

N N ( )
1
4
(1 )(1 ), ( )
1
4
(1 )(1 )
1 2
= = +

N N ( )
1
4
(1 )(1 ), ( )
1
4
(1 )(1 )
3 4
= + + = +
(6.63)
From the preceding, we have the interpolation functions for the isopara-
metric elements, also called the shape functions, in Equation (6.51) by

N
i i i
= + +
1
4
1 1 ( )( )

i
i
i
= , , , = , , ,
= , , ,
1 2 3 4 1 1 1 1
1 1 1 1
,

(6.64)
Let us consider the quadrilateral isoparametric elements. The dis-
placements in the element are described by the nodal displacements u
i

Virtual work theorem and finite element method 187
(displacements in the x direction at node i) and v
i
(displacements in the y
direction at node i)

u N u v N v
i
i i
i
i i
,
1
4
1
4

= =
= =
(6.65)
where u and v are the displacements in the x and y directions, respectively.
It is seen that the isoparametric elements satisfy the condition of com-
pleteness from Equation (6.39). The isoparametric elements satisfy the
aforementioned conditions because the interpolation function is the same
as the shape function. If the displacement mode is rigid, displacement u is
constant in the element. Equation (6.65) means that the isoparametric ele-
ments can describe the rigid body motion.
The other condition for the fnite element solutions, which converges to
the true solutions, is the condition of compatibility. The compatibility con-
dition is satisfed when the shape function is continuous, the frst derivative
of that is continuous in the element, and the function is continuous at the
boundaries of the elements.
Substituting Equation (6.65) into Equation (6.48), the strain vector is
obtained for the plane problem as

{ }
[ ]
[ ]

= =

xx
yy
xy
N
x
N
y
0
0
[[ ] [ ]
{ }
{ }
N
y
N
x
u
v

= [ ]{ } B
(6.66)
where
xy xy
2 , =

xy
is the engineering strain,

N N N N N u u u u u v v v v v
T T
[ ] [ ], { } { }, { } { }
1 2 3 4 1 2 3 4 1 2 3 4
= , , , = , , , = , , , (6.67)
We need

,
N
x
N
y
i i
to obtain the [B] matrix.
Since N
i
is the function of the natural coordinates (,) (Figure6.5), from
Equation (6.64), we have

N
N
x y
x y
i
i

N
x
N
y
J
N
x
N
y
i
i
i
i
[ ]
(6.68)
188 Computational modeling of multiphase geomaterials
From Equations (6.64) and (6.68), Jacobian matrix [J] is given by
J
N
x
N
y
N
x
N
y
x y
x y
j
j
j
i
i
j
j
j
j
j
j
j
j j j
j
j j j
j
j j j
j
j j j
j
[ ]
1
4
(1 ) (1 )
(1 ) (1 )
1
4
1
4
1
4
1
4
1
4
1
4
1
4
1
4

=
+ +
+ +
= =
= =
=
=
= =

(6.69)
Then,

N
x
N
y
J
N
N
i
i
i
i
[ ]
1

(6.70)
Substituting Equation (6.70) into Equation (6.66), we have
N
x
N
x
N
x
N
x
N
y
N
y
N
y
N
y
N
y
N
x
N
y
N
x
N
y
N
x
N
y
N
x
u
v
u
v
u
v
u
v
B { }
0 0 0 0
0 0 0 0
[ ]{ }
1 2 3 4
1 2 3 4
1 1 2 2 3 3 4 4
1
1
2
2
3
3
4
4
=

, ,

, ,

, ,

,
,

, ,

, ,

, ,

=

(6.71)
Using the [B] matrix, we have the [K] matrix shown in Equation (6.31) as

[ ] [ ] [ ][ ] [ ] [ ][ ] [ ] K B D B dV B D B det J d d
V
T T
= =

1
1
1
1

(6.72)
[B] = [B(,)] (6.73)
where (, ) are the natural coordinates used for the shape functions.
If we set [ ( )] [ ] [ ][ ] [ ] f B D B det J
T
, = , we have

K f d d [ ] ( , )
1
1
1
1

=

(6.74)
Virtual work theorem and finite element method 189
In order to perform the integration of Equation (6.74), we can use Gauss
Legendres numerical as

[ ] ( ) [ ( K f d d H H f
p q
p q p
= , =

= =


1
1
1
1
2 1
2
2 1
2
2 2 2
,,
p2
)] (6.75)
For the two-point Gauss quadrature rule for integration, integration
points ( ),
p q
p q
2 2 2 2
1 2 , , = , are given in Figure (6.6) as

( )
p q 2 2
1
3
, =
(6.76)
where the weighting coeffcients are ( ) ( . , . ) H H
p q 2 2
1 0 1 0 , = .
The volumetric strain is

N
x
N
y
N
x
N
y
N
x
N
y
N
x
N
y
B
v xx yy v
T
{ } [ ] { }
1 1 2 2 3 3 4 4
= + =

=

(6.77)
Then,

[ ] [ ] [ ] [ ] K B dV B det J d d
v
T
V
v
T
v
T
= =

1
1
1
1
(6.78)
and

V
T T
B dV B det J d d

=

[ ] { } [ ] { } [ ]
1
1
1
1
(6.79)
(1, 1) (1, 1)
(1, 1) (1, 1)

Figure 6.6 Integration points.

190 Computational modeling of multiphase geomaterials
If we use bilinear quadrilateral elements as

u x y a a x a y a xy
v x y b b x b y b xy
( , )
( , )
= + + +
= + + +
1 2 3 4
1 2 3 4
(6.80)
we may have incompatible deformation shown in Figure6.7 (Terada 2008)
although the isoparametric elements satisfy the compatibility.
6.3 DYNAMIC PROBLEM
Based on the principle of dAlembert, the virtual work theorem for a one-
phase material becomes

V
T
S
T
s
V
T
b
dV u T ds u F u

= + {

} { } { } { } { } ({ } { }) ddV (6.81)
where { } u is the acceleration vector, { } F
b
is the body force, {

} is the virtual
strain vector, and { } u is the virtual displacement vector.
Then, from Equations (6.26) and (6.28), we have
V
T
B D B u dV

[ ] [ ][ ]{ }
B dV N F dV N N
u
dV N T dV
T
V
T
b
V
T
S
T
s
[ ] { } [ ] { } [ ] [ ]{ } [ ] { }

= + +




(6.82)
y y
3
3
4
2
Bilinear Quadrilateral Element
2
1 1
4
x x
Deformation
Figure 6.7 Incompatible formation of trapezoidal four-noded quadrilateral element.
Virtual work theorem and finite element method 191
and

[ ]{ } [ ]{ } { } M
u
K u F
e

+ =
(6.83)

[ ] [ ] [ ] M N N dV
V
T
=

(6.84)

[ ] [ ] [ ][ ] K B D B dV
V
T
=

(6.85)

{ } [ ] { } [ ] { }
[ ] { }
( )
F N F dV N T dS
B dV
e
V
T
b
S
T
s
T
t
V V
= +
+

[ ] { } B dV
T

(6.86)
where { }
*
u is the acceleration vector at a node and
t
{ }
{ }
is a stress at a pre-
vious time.
Considering the viscous term, [ ]{ } C
u

, due to the material damping and

the energy dispersion, Equation (6.83) becomes

[ ]{ } [ ]{ } [ ]{ } { } M
u
C
u
K u F
e

+ + =

(6.87)
6.3.1 Time discretization method
For integrating Equation (6.87), we can use the Newmark method, linear
acceleration method, or Wilson method.
6.3.1.1 Linear acceleration method and Wilson method
In the Wilson method, the velocity and the displacement are approxi-
mated by

{ } { } [( ){ } { }]
t t t t u u
a
u
a
u + +
= + +

1
(6.88)

{ } { } { } (( ){ } { }) u u
u
b
u
b
u t t t t t + +
= + + +

2
2
1 (6.89)
where ,a,b are parameters.
We assume = t, > 1 and that the equations of motion hold at time
t + t as

[ ]{ ( )} [ ]{ ( )} [ ]{ ( ) M
u
t t C
u
t t K u t t

+ + + + +

}} { ( )} = + F t t
e
(6.90)
192 Computational modeling of multiphase geomaterials
From Equations (6.88), (6.89), and (6.90), we have
t t u +
and then t t u +

is obtained by the interpolation as

{ } { } ({ } { }) (( )
t t t t t t u
u
u u + +
= + =

1 1
1

{{ } { })
t t t u u
+ ,
+
0 1
(6.91)
The velocity and the displacement are obtained as

{ } { } (( ){ } { })
t t t t t t u u
t a
u
a
u + +
= + +

1
(6.92)

{ } { } { } (( ){ } { u u t
u
t
b
u
b
u t t t t t t t + +
= + + +

2
2
1 }}) (6.93)

= . , = , = 1 4
1
2
1
3
a b
It is known that the Wilson method is unconditionally stable when
>1.37. However, it may bring about the artifcial damping effect. When
= . , = , = 1 0
1
2
1
3
a b , the Wilson method becomes the linear acceleration
method.
6.3.1.2 Newmark method
The extended linear acceleration method is the Newmark method. Using
the constants and , we can calculate the displacement and the velocity as

{ } { } { } ( ) { } ( ) ({ u u t
u
t
u
t
t t t t t t t + +
= + + +

1
2
2 2

u ut
} { })
(6.94)

t t
t t t t t
u
u
t
u
t
u u
+
+
= + +
{
} { } { } ({ { })

(6.95)

[ ]{ ( )} [ ]{ ( )} [ ]{ ( )} { M
u
t t C
u
t t K u t t

+ + + + + =

FF t t
e
( )} +
(6.96)
= 0.5 is frequently used and this method is unconditionally stable when
0.25. This method is unconditionally stable if 20.5 for the zero
damping system. Even if 0.5 and 1/4, the system is conditionally stable
when t 1 2 / ( / )
max
;
max
is the maximum angular frequency.
By simultaneously solving Equations (6.94), (6.95), and (6.96),
{ }
t t u +
,
{ }
t t u +
, and { } u
t t +
are obtained.
Virtual work theorem and finite element method 193
Rewriting Equation (6.96) by the Newmark method, we have

([ ] [ ] ( ) [ ]) { } M t c t K
u t t t t
+ +
+ +

2

= + +
+
{ } [ ]({ } { } { }) [ ]({ } F C u t u t u K u
e t t t t
+ t u t u
t t
{ } ( )( ) { } )
1
2
2

(6.97)
6.3.1.3 Central finite difference scheme
Using the central fnite difference scheme, we can get the acceleration and
the velocity as

{ }
( )
({ } { } { })
t t t t t t u
t
u u u

= +
+
1
2
2
(6.98)

{ } ({ } { })
t t t t t u
t
u u

=
+
1
2
(6.99)
6.3.2 Mass matrix
As has been shown in Equation (6.83), we need mass matrix [M]. There
are two types of mass matrices, namely, (1) consistent mass matrix method
and (2) lumped mass matrix method. The consistent mass matrix, shown
in Equation (6.84), is derived rigorously, but the mass matrix includes a
nondiagonal component. From a physical point of view, the mass matrix
has to be symmetric. For the mass matrix to be symmetric, we can take the
average of the masses of the nodes.
The consistent mass is given by Equation (6.84) as M N N dV
V
T
[ ] [ ] [ ] = .
The lumped mass matrix can be obtained by equally dividing the mass of
the element and distributing it symmetrically. The lumped mass matrix has
only diagonal components.
6.4 DYNAMIC ANALYSIS OF WATER-
SATURATED SOIL
Let us explain the discretization of the governing equations. In the discreti-
zation of the governing equations in space, both the fnite element method
(FEM) and fnite difference method (FDM) are used. For the discretiza-
tion of the equations of motion of the mixture, FEM is used. On the other
hand, for the discretization of the continuity equation of pore fuids, FDM
is used in this section for simplifed practical use. In the method, the pore
194 Computational modeling of multiphase geomaterials
water pressure is defned at the center of the element, and the stress and
the strain are also calculated at the center of the element with the reduced
integration technique. Using this method, it is possible to avoid shear lock-
ing under undrained conditions and to reduce the degree of freedom in the
computation (Hisada and Noguchi 1995). The Newmark method is used
in the discretization in time. For the accuracy of the numerical analysis,
the method has been verifed by comparing its results with the analytical
solution of BVP for water-saturated elastic media (Oka et al. 1994). Of
course we can use FEM for the discretization of the continuity equation. In
particular, we need the fnite element formulation for the fnite deformation
analysis (Mirjalili et al. 2011).
6.4.1 Equation of motion
Multiplying Equation (2.32), the arbitrary weighted function u
i
s
(vir-
tual displacement) that satisfes u
i
s
= 0 on the displacement boundary, we
obtain the weak form after integration in domain V:

V
i
s
ij
j
i i
s
u
x
b u dV

0 (6.100)
On the traction boundary S
t
, we have

T n u dS
i j i i
s
S
t
( ) 0

=
(6.101)
The integration of the second term on the left-hand side of Equation
(6.100) leads to the following equation considering the displacement
boundary conditions:

V
i
s
i i
s
S
ij j i
s
V
ij u
b u dV n u dS
t

+

uu
x
dV
i
s
j

= 0 (6.102)
where n
j
is the outward unit normal vector of surface S
t
.
Virtual strain tensor
ij
s
can be expressed in a similar manner to
Equation (2.8):

ij
s i
s
j
j
s
i
u
x
u
x
=

1
2
(6.103)
Virtual work theorem and finite element method 195
Considering the symmetry of stress tensor.
ij ji
= , we have

ij
i
s
j
ij ij
s
ii
s
u
x
p

= ( )
(6.104)
Upon substitution of Equations (6.101) and (6.104) into Equation (6.101),
we get the following equation after manipulation:

u
u dV dV p dV b u dV
T
u dS
V
i
s
i
s
V
ij ij
s
V
ii
s
V
i i
s
S
i
i
s
t

+

= +
(6.105)
This equation is a weak form of the equation of motion in domain V.
Here, the vector expression of Equation (6.105) for the formulation by
FEM becomes

V
s T s
V
s T
V
v
s
u
u
dV dV p dV

+ { } { } { } { }

= +

V
s T
S
s T
u b dV u T dS
t
{ } { } { } { } (6.106)
where variables without subscripts correspond to the vectors in Equation
(6.105) and the pore water pressure p(= u
w
) is positive in compression. In
order to use the incremental nonlinear constitutive law shown in Equation
(2.11), the effective stress vector, { } , is expressed as

{ } { } { } = +
|

t
(6.107)
where { }
|

t
is an effective stress vector at a previous time t when the current
time is t + t.
The substitution of Equation (6.107) into Equation (6.106) gives

V
s T s
V
s T
t
V
u
u
dV dV p

+ +
|
{ } { } { } ({ } { })

v
s
dV

= +

V
s T
S
s T
u b dV u T dS
t
{ } { } { } { }
(6.108)
In the following, the displacement vector, { } u
N
, at the nodes and pore
water pressure, p
E
, at the gravitational center of the element (pore pressure
196 Computational modeling of multiphase geomaterials
increment during earthquake in the fnal setting p
dE
) are taken as unknown
variables in the discretized equation of motion.
The displacement vector of the solid phase, { } u
s
, is given by the displace-
ment vector at the nodes { } u
N
in terms of the shape function matrix [N] as

{ } [ ]{ } u N u
s
N
= (6.109)
The pore water pressure is assumed to be constant in the element (namely,
the 0th order shape function is used) based on Christians method (1970).
Using the pore water pressure at the gravitational center of the element,
the pore water pressure is given by

p N p p
p E E
= = (6.110)
where N
p
is the zeroth-order shape function, that is, equal to 1.
From Equation (2.9), the strain vector of the element { }
s
is given as

{ } [ ]{ }
s s
L u =
(6.111)
where [L] is a matrix that transforms displacements into strains.
Upon substitution of Equation (6.109) into Equation (6.111), we have the
following equation:

{ } [ ][ ]{ } [ ]{ }
s
N N
L N u B u = = (6.112)
The virtual volumetric strain,
ii
s
, of the element is expressed using the
virtual displacement vector, { } u
N
, as


ii
s
v
s
v
T
N N
T
v
B u u B = = = [ ] { } { } [ ] (6.113)
where { } B
v
is a vector that transforms the nodal displacements of the ele-
ment into the volumetric strain of the element.
Finally, Equation (2.11) can be written in matrix form as

{ } [ ]{ } [ ][ ]{ } = = D D B u
s
N
(6.114)
where { } u
N
is the displacement increment vector.
Virtual work theorem and finite element method 197
Substituting Equations (6.109), (6.110), (6.112), (6.113), and (6.114) into
Equation (6.108), we obtain the following equation:

u N N dV
u
u B D B dV u u
B dVp
N
T
V
T
N N
T
V
T
N N
T
V
v E
{ } [ ] [ ] { } { } [ ] [ ][ ] { } { }
[ ]

+

u N b dV u N T dS u B dV
N
T
V
T
N
T
S
T
N
T
V
T
t
t
{ } [ ] { } { } [ ] { } { } [ ] { }

= +
|


(6.115)
Rearranging the terms with respect to the virtual displacement vector,
{ } u
N
, we have

u N N dV
u
B D B dV u B dVp
N
T
V
T
N
V
T
N
V
v E
{ } [ ] [ ] { } [ ] [ ][ ] { } [ ]

+

u N b dV N T dS B dV
N
T
V
T
S
T
V
T
t
t
{ } [ ] { } [ ] { } [ ] { }

= +
|
(6.116)
Since this equation holds for the arbitrary unrestricted virtual displace-
ment vector, { } u
N
, the stiffness equation of the element is given by

N N dV
u
B D B dV u B dVp
V
T
N
V
T
N
V
v E
[ ] [ ] { } [ ] [ ][ ] { } [ ]

+

N b dV N T dS B dV
V
T
S
T
V
T
t
t
[ ] { } [ ] { } [ ] { }

= +
|
(6.117)
Herein, we set the following terms as

M N N dV
V
T
[ ] [ ] [ ]

=
(6.118)

K B D B dV
V
T
[ ] [ ] [ ][ ]

=
(6.119)

K B dV
v
V
v
{ } [ ]

=
(6.120)
198 Computational modeling of multiphase geomaterials

F N b dV N T dS
V
T
S
T
t
{ } [ ] { } [ ] { }

= +
(6.121)

R B dV
t
V
T
t
{ } [ ] { }

=
| |
(6.122)
These matrices are calculated for the element. Then, assembling the stiff-
ness matrices of all the elements, the total stiffness matrix is derived.
In the calculation of the integration included in the stiffness equation of
the element, the Gauss numerical integration method is used.
For Equations (6.119) and (6.122), the stresses are calculated at the
Gauss integration points. Although four integration points are, in general,
used for the square element, there is a tendency for the shear stiffness to
be overestimated under the undrained conditions and the deformation is
underestimated (shear locking).
To avoid shear locking under undrained conditions, the reduced integra-
tion technique is adopted, namely, the stress levels are calculated at the
gravitational center of the element. Due to this method, the pore water
pressure defned at the center of gravitational center corresponds one-to-
one with the stress.
The mass matrix [M] of Equation (6.118) is a consistent mass matrix.
In the program LIQCA (Oka et al. 1994, 2004; LIQCA Research and
Development Group 2005), a lumped mass matrix is used as

[ ] [ ] M
V
n
I
e
e
e
=

(6.123)
where V
e
is the volume of the element, n
e
is the number of nodal points, and
[ ] I
e
is a unit matrix that corresponds to the degree of freedom of the nodal
points of the element ( 2 2 n n
e e
).
In Equation (6.117), the damping of the system is given by the hysteretic
damping of the constitutive model. However, the material damping of the
system is diffcult to reproduce only by the hysteretic damping of the model
in both the small vibration and in the high frequency region.
Hence, we use the Rayleigh damping matrix [C] that is proportional to
the velocity vector of the nodal points. Rayleigh damping is described by
the linear combination of the mass matrix [M] and the stiffness matrix
[K] as

[ ] [ ] [ ] C M K = +
0 1
(6.124)
where
0
and
1
are constants.
Virtual work theorem and finite element method 199
These two constants can be determined by solving the following simulta-
neous equation derived by specifying damping for two particular frequen-
cies (the ith and the jth order characteristic frequencies)

2
2
0 1
0 1
h
h
i i i
j j j
/ + =
/ + =
(6.125)
where
i
and
j
are the ith and the jth natural circular frequencies, and
h
i
and h
j
are damping for the ith and the jth modes. For the joint ele-
ment, Rayleigh damping is not considered, which is defned in the entire
domain.
In addition, the damping matrix [C] is used for setting the viscous
boundary. At the viscous boundary, the following terms are given at the
nodal points:
For the degree of freedom vertical to the boundary,
b p
V / 2
For the degree of freedom horizontal to the boundary,
b s
V /2
where
b
is the density of the semi-infnite ground, V
p
and V
s
are the veloc-
ity of the P wave and the velocity of the S wave of the semi-infnite ground,
and is the length of the side of the element at the boundary.
From the above, the substitution of Equations (6.118) to (6.124) into
Equation (6.117) leads to

[ ]{ } [ ]{ } [ ]{ } { } { } { } M
u
C
u
K u K p F R
N N N v E t

+ + + =
|
(6.126)
Next, we will consider the external force vector. The external forces
applied to the system can be derived into two parts: the force that has
been applied before earthquakes and the force that is applied during earth-
quakes. The former is a gravitational force and the latter is the inertia force
due to earthquakes.
The external force expressed by Equation (6.121) can be divided into the
force having been applied before earthquakes and the incremental force
during earthquakes, as follows:

{ } { } { } F F F
s d
= +
N b dV N
T
dS N b dV N
T
dS
V
T
s
S
T
s
V
T
d
S
T
d
t t
[ ] { } [ ] { } [ ] { } [ ] { }

= + + +

(6.127)
200 Computational modeling of multiphase geomaterials
where {F
s
} is the self-weight force vector due to gravity and {F
d
} is the load-
ing vector due to earthquakes.
Similarly, the body force vector, {b}, can be divided into { } b
s
and { } b
d
,
and the surface traction vector, { } T , can be divided into
{ }
s T and
{ }
d T .
Now, we consider the static analysis (i.e., initial stress analysis) due to
the self-weight force.
The equilibrium equation is obtained by subtracting the dynamic term
from Equation (6.126) as

[ ]{ } { } { } { } K u K p F R
N v Et s t
=
| = | 0
(6.128)
where p
Et | =0
is the initial pore water pressure before an earthquake. In
this equation, the nonlinearity of the stiffness matrix is taken into
consideration.
Using the initial effective stress vector before earthquakes, { }
| =

t 0
, the
equilibrium equation becomes

B dV K p F
V
T
t v Et s
[ ] { } { } { }
0 0

=
| = | =
(6.129)
Upon substitution of Equations (6.127) and (6.129) into Equation (6.126),
M
u
C
u
K u K p B dV K p F R
N N N v E
V
T
t v E t d d t
e
[ ]{ } [ ]{ } [ ]{ } { } [ ] { } { } { } { }
0 0

+ + = +
| = | = |

(6.130)
The residual vector, { } R
d t |
, can be rewritten considering Equation (6.122) as

R B dV
d t
V
T
t t
{ } [ ] ({ } { } )
0

=
| | | =
(6.131)
Substituting Equations (6.131) into Equation (6.130), after manipulation
we obtain

[ ]{ } [ ]{ } [ ]{ } { }( ) {
|
M
u
C
u
K u K p p F
N N N v E Et d

+ + =
=0
}} { }
|
R
d t
(6.132)
As noted earlier, we will use Equation (6.132) considering the initial stress
due to the self-weight force before the earthquake.
Virtual work theorem and finite element method 201
6.4.2 Continuity equation
In the general formulation of the fnite element method, the general
weighted function has been used to obtain the weak form of the govern-
ing equations. For the selected point collocation method, the unit value
function at several selected points is used; for the subdomain collocation
method, weighted function, w, is used in which W = 1 in the domain and
W = 0 outside of the domain; for the Bubnov Galerkin method, the shape
function is used as a weighted function. The weak form of the governing
equation can be discretized based on the appropriated weighting function,
w, which is locally defned.
6.4.2.1 Galerkin method
We can employ the shape function, which is used as the interpolation
function for the isoparametric elements as W N
h
= [ ]. Using the Galerkin
method, we obtain the weak form of the continuity equation (Equation
2.45) as

[ ] N
k k p
x
n
K
p d
h
T
w
f
ii
s
w
i
ii
s
f

+ +

2
2
VV
V

= 0
(6.133)
Using the Gauss theorem, we have

+ +

[ ] [ ] { } [ ] N
k
dv
k
N p dv N
h
T
w
V
f
ii
s
w
h
V
T
h
V
T

ii
s
f
V
h
T
w
S
h
T
dv
n
K
N pdv
k
N p n ds
+
=

[ ]
[ ] { }{ }

(6.134)
When permeability, porosity, and the volumetric modulus of water are
constant in the element, we can discretize the equation as

+ + +
k
K a K v
k
K p
n
K
K
w
f
v
T
N v
T
N
w
h N
f
p

[ ] { } [ ] { } [ ]{ } [ ]{

pp
k
N p n ds
N
w
h
S
}
[ ]{ }{ } =

(6.135)

p N p
h N
= [ ]{ }
,
{ } [ ]{ } [ ]{ } = = p N p B p
h N h N
(6.136)
202 Computational modeling of multiphase geomaterials

ii v
T
N
B v = [ ] { }
,

ii v
T
N
B a = [ ] { } (6.137)

[ ] [ ] [ ] K N B dv
v
T
h
V
T
v
T
=

(6.138)

[ ] [ ] [ ] K B B dv
h h
V
T
h
=

(6.139)

[ ] [ ] [ ] K N N dv
p h
V
T
h
=

(6.140)
where { } v
N
is the velocity vector, { } a
N
is the acceleration vector, and { } p
N

is the pore pressure vector.
6.4.2.2 Finite volume method
When the pore water pressure is constant in the element, it is useful to apply
a variant of the fnite volume method, which is similar to the subdomain
collaboration method. In the fnite volume method, the node is defned at
the center of the element.
The continuity equation is discretized by a weight function as

W in the element
W outside the element
n
n
=
=
1
0
( )
( )
(6.141)
When we consider the water pressure boundary, we will have the same
results even if disregarding the initial hydrostatic pressure, that is, using the
excess pressure that is obtained by subtracting the initial hydrostatic pres-
sure from the total water pressure.
Multiplying the continuity equation, Equation (2.45), by the weighted
function, W
n
, and integrating it over domain V, we have the following weak
form as

V
w
f
ii
s
w
d
i
ii
s
f d
n
k k p
x
n
K
p W

+ +

2
2
ddV = 0 (6.142)
where p
d
is the pore pressure.
Using the Gauss theorem, we have

V
w
f
ii
s
ii
s
f d
n
S
w
d
k n
K
p dV
k p

+

+
W
xx
n dS
i
i n
W = 0 (6.143)
Virtual work theorem and finite element method 203
In the following, we will develop the equations with the unknowns of the
nodal displacement vector of the element, { } u
N
, and pore water pressure
increment, p
dE
, during earthquakes.
As is the case with Equation (6.113), the term including the volumetric
strain of solid phase,
ii
s
, in Equation (6.143) can be written using the nodal
displacement vector as Equation (6.23):
ii
s
v
T
N
B u = [ ] { }.
Expressing the frst term on the left-hand side of Equation (6.143) in a
vector form and substituting Equation (6.137) into it, we have the follow-
ing equation:
k
B dV
u
B dV
u
n
K
dVp
k p
x
n dS
V
w
f
v
T
N
V
v
T
N
V
f dE
S
w
dE
i
i
[ ] { } [ ] { } 0

+ +

=
(6.144)
Herein, using Equation (6.120), Equation (6.144) becomes

+ +

f
w
v
T
N v
T
N
dE
S
w
dE
i
k
K
u
K
u
Ap
k p
x
{ } { } { } { }


nn dS
i
= 0
(6.145)
where coeffcient A is given by

A
n
K
dV
V
f
=

(6.146)
It is worth noting that Equation (6.145) can be applicable to any shape of
quadrilateral element.
For rectangular elements, the second gradients of the pore water pressure
can be approximated by the fnite difference scheme as

=
+ +
2
2 2
2 p
x
p x x p x p x x
x
d d d d
( ) ( ) ( )
(6.147)
However, we can use the fnite volume method for the continuity equa-
tions. For the fnite volume method, we can use a polygon, such as a quad-
rilateral element, and the nodes are located in the center of the volume, that
is, elements. First, we consider the two adjacent square elements shown in
Figure6.8. The pore water pressure at the gravitational center of the element
concerned is p
dE
, and the pore water pressure at the gravitational center of
the adjacent element i is p
dEi
. The permeability coeffcient of the element con-
cerned is k, and the permeability of the adjacent element i is k
i
. The length of
the boundary adjacent to the neighboring element is b
i
, and its components
in the x and y directions are b
xi
and b
yi
, respectively. Thedistance between
204 Computational modeling of multiphase geomaterials
the gravitational center of the element and adjacent element i is s
i
, and its
components in the x and y directions are s
xi
and s
yi
, respectively. s
i
is given
by adding distance s
1i
to distance s
1i
, as shown in Figure6.8.
Since the permeability of the element may be different from that of
the adjacent element, we use the average permeability coeffcient, i k .
The average permeability coeffcient between elements, i k , is given by
Darcys law as

s
k
s
k
s
k
i
i
i i
i
= +
1 2
(6.148)
Using the unit normal vector to the side of element n
i
in the coordi-
nate system shown in Figure 6.8, the third term on the left-hand side of
Equation(6.145) is expressed as

S
w
dE
i
i
i
b
xi xi yi yi
k p
x
n dS v n v n db
i

= +
=

1
4
(6.149)
Note
p
dE
, p
dEi
k, k
i
b
i
s
i
: Pore water pressure
: Coefcient of permeability
: Width of element boundary
: Distance between two elements
y
x
s
i
(x
1i
, y
1i
)
(x
g
, y
g
)
(x
2i
, y
2i
)
(x
gi
, y
gi
)
s
1i
s
xi
p
dE
b
yi
b
xi
n
i
b
i
n
xi
s
yi
k
i
p
dEi

k
s
2i
n
yi
Figure 6.8 Discretization of pore water fow equation.
Virtual work theorem and finite element method 205
where n
xi
is the component of the unit normal n
i
in the x direction and n
yi

is the component of the unit normal n
i
in the y direction.
The velocity of the in-fow of pore water into the center element from
element i is given by

v
k p p
s
i
w
dEi dE
i
=

(6.150)
The components in the x and y directions are

v
k p p
s
k p p
s
S
s
xi
w
dEi dE
i w
dEi dE
i
xi
i
cos =

(6.151)

v
k p p
s
k p p
s
S
s
yi
w
dEi dE
i w
dEi dE
i
yi
i
sin =

(6.152)
The components of the unit normal n
i
to the boundary are expressed by

n
b
b
n
b
b
xi
yi
i
yi
xi
i
= , = (6.153)
The in-fow per unit time into the element from the adjacent elements is
calculated as

v n v n db
k
p p
s
s
b
k
p p
s
s
b
k
p p
s b s b
s
b
xi xi yi yi
w
dEi dE
xi
i
yi
w
dEi dE
yi
i
xi
w
dEi dE
xi yi yi xi
i
i
( ) ( ) ( )
( )
2 2
2

+ =

+
(6.154)
Finally, the total in-fow of the pore water into the elements becomes

k
p p
s b s b
s
p p
i
w
dEi dE
xi yi yi xi
i
dE
i
i dEi
( )
1
4
2
1
4

+
= +
= =
(6.155)
where

i
w
xi yi yi xi
i
k
b s b s
s
=

2
,

=
=

i
i
1
4
(6.156)

+ +
=

f
w
v
T
N v
T
N
dE
de
i
i
k
K
u
K
u
Ap P { } { } { } { }


1
4
PP
dEi
= 0
(6.157)
206 Computational modeling of multiphase geomaterials
6.4.3 Time discretization
The discretized equation of motion, Equation (6.132), and the continuity
equation, Equation (6.157), at current time t + t are given by
M
u
C
u
K u K p F R
N t t t t N t t t t N t t v dEt t d t t d t
[ ]{ } [ ] { } [ ] { } { } { } { }

+ + =
| + | + | + | + | + | + | + |

(6.158)

k
K
u
K
u
Ap p p
f
w
v
T
N t t v
T
N t t
dEt t
dEt t
i
i dEi t t
{ } { } { } { } 0
1
4

+ + =
| + | +
| +
| +
=
| +

(6.159)
For the time discretization, the Newmark method is used. The
nodal displacement vector and the velocity vector at the current time are
described as

{ } { } { }
( )
{ } ( ) u u t
u
t
u
t
N t t N t N t N t | + | | |
= + + +

2
2

22
({ } { } )
N t t N t u u
| + |
(6.160)

{ } { } { } ({ } {
N t t N t N t N t t u u
t
u
t
u
| + | | | +
= + +
NN t u
} )
|
(6.161)
where t is the time increment, and and are parameters. In many
cases, 0.3025 and 0.6 are taken for and , respectively. With these
values and a time increment of 0.001, the present method gives a stable
solution for a sandy ground with a smaller permeability than 0.1 cm/s
since has to satisfy the following relations based on the stability of the
u-p formulation (Kato 1995; LIQCA Research and Development Group
2005).

+
k
g t
1
2
(6.162)
where t is the time increment, k is the permeability coeffcient and g is the
gravitational acceleration.
Using these relations, we have the equations with unknown variables of
nodal acceleration vector
{ }
N t t u
| +
and the excess pore water pressure of
element p
dEt t | +
.
Virtual work theorem and finite element method 207
6.4.3.1 Equation of motion
Upon substitution of Equations (6.160) and (6.161) into Equation (6.158),
we obtain

[ ]{ } [ ] { } { } ({ M
u
C
u
t
u
t
N t t t t N t N t

| + | + | |
+ + +
NN t t N t u u
} { } )
| + |

( )

+ + +
| + | |
[ ] { }
( )
{ } ( ) ({ } K t
u
t
u
t
u t t N t N t N

2
2
2

|| + |

t t N t u
{ } )

K p F R
v dEt t d t t d t
{ } { } { } =
| + | + |
(6.163)
After the transposition of unknown nodal acceleration vector
{ }
N t t u
| +
and the excess pore water pressure of element p
dEt t | +
to the left-hand side
and the manipulation, we have the following equation:

M t C t K
u
K p
t t t t N t t v dEt t
[ ] [ ] ( ) [ ] { } { }
2

+ +
| + | + | + | +

= +
| + | | + |
{ } { } [ ] { } ( ) { } F R C
u
t
u d t t d t t t N t N

1
||t

+
| + | |
[ ] { } ( ) { } K t
u
t
u t t N t N t

1
2
2

(6.164)
This is the discretized equation of motion.
6.4.3.2 Continuity equation
The substitution of Equation (6.161) into Equation (6.159) gives

k
K
u
K
u
t
u
t
u u
f
w
v
T
N t t v
T
N t N t N t t N t
{ } { } { } { } { } ({ } { } )
( )

+ + +
| + | | | + |

+
| +
| +
=
| +

Ap p p
dEt t
dEt t
i
i dEi t t

1
4
= 0
(6.165)
208 Computational modeling of multiphase geomaterials
Here, we apply the backward Euler fnite difference scheme to the third
term on the left-hand side of the equation:

dEt t
dEt t dEt
p
p p
t
| +
| + |
=

(6.166)
Substituting
w
f
g = and Equation (6.166) into Equation (6.165) yields

k
g
K
u
K
u
t
u
t
u u v
T
N t t v
T
N t N t N t t N t
{ } { } { } { } { } ({ } { } )
( )
+ + +
| + | | | + |

+
| + | | +
=
| +

A
t
p p p p
dEt t dEt dEt t
i
i dEi t t
( )
1
4
= 0
(6.167)
Rearranging the unknown nodal acceleration vector
{ }
N t t u
| +
and the
excess pore water pressure of element p
dEt t | +
in Equation (6.167), we have

k
g
t K
u
A
t
p p
v
T
N t t dEt t
i
i dEi t t
{ } { }
1
4

+ +
| + | +
=
| +

K
u
t
u
A
t
p
v
T
N t N t dEt
{ } { } (1 ) { }

= + +
( ) | | |
(6.168)
As noted earlier, the continuity equation can be discretized. Combining
Equations (6.164) and (6.168), we will have an antisymmetric total matrix
because the coeffcients of the pore water pressure in Equations (6.164) and
(6.168) at time t+t are different.
Hence, we will rearrange Equation (6.168) to be symmetric although
stiffness matrix K is not symmetric, because the elastoplastic matrix is
derived based on the nonassociated fow rule.
In Equation (6.168), the coeffcients are replaced as

=

( )

1
k
g
t
(6.169)

i
k
g
i
t
=

( )

1
(6.170)

=

( )
A
t t
A
k
g
1

(6.171)
Virtual work theorem and finite element method 209
The substitution of Equations (6.169), (6.170), and (6.171) into Equation
(6.168) gives

K
u
A p p
v
T
N t t dEt t
i
i dEi t t
{ } { } ( )
1
4

+ +
| + | +
=
| +

t
K
u
t
u
Ap
k
g
v
T
N t N t dEt
1
{ } { } (1 ) { }

( )
=


+ +
( ) | | |
(6.172)
Finally, combining the equation of motion, Equation (6.164), and the
continuity equation, Equation (6.172), we have
M t C t K K
K A
u
p
p
t t t t v
v
T
N t t
dEt t
i
i dEi t t
[ ] [ ] ( ) [ ] { }
{ }
{ }
0
2
1
4

+ +

+

| + | + | +
| +
=
| +

F R C
u
t
u
K t
u
t
u
t
K
u
t
u
Ap
d t t d t t t N t N t
t t N t N t
k
g
v
T
N t N t dEt
{ } { } [ ] { } (1 ) { }
[ ] { }
1
2
( ) { }
1
{ } { } (1 ) { }
2



( )
=
+
+

+ +
( )
{ } | + | | + | |
| + | |
| | |
(6.173)
Since the pore water pressure of the two adjacent elements is unknown,
after the rearrangements we have

M t C t K K
K A
u
p
p
t t t t v
v
T
i
T
N t t
dEt t
dEi t t
[ ] [ ] ( ) [ ] { } 0
{ } { }
{ }
{ }
2

+ +


| + | +
| +
| +
| +

F R C
u
t
u
K t
u
t
u
t
K
u
t
u
Ap
d t t d t t t N t N t
t t N t N t
k
g
v
T
N t N t dEt
{ } { } [ ] { } (1 ) { }
[ ] { }
1
2
( ) { }
1
{ } { } (1 ) { }
2



( )
=
+
+

+ +
( )
{ } | + | | + | |
| + | |
| | |
(6.174)
210 Computational modeling of multiphase geomaterials
Although this equation seems to be nonsymmetric, the total matrix
becomes globally symmetric, because the adjacent elements have the same
coeffcient vector
i
when we construct the total matrix. It should be noted,
however, that the stiffness matrix is not symmetric because of the nonas-
sociated fow rule.
6.5 FINITE DEFORMATION ANALYSIS FOR
FLUIDSOLID TWO-PHASE MIXTURES
In this chapter, we formulate a three-dimensional fnite element method
based on Biots two-phase mixture theory (1956) in the framework of the
fnite deformation theory. A fnite deformation analysis is necessary to sim-
ulate large deformation problems, such as strain localization problems. The
strain localization phenomenon is a geometrically nonlinear problem since
the deformation in shear bands is large. The constitutive equation for clay
used in this study is a nonlinear elastoviscoplastic model and it is formu-
lated in an incremental form. For simplicity, both the grain particles and
the fuid are assumed to be incompressible.
In order to deal with such a large nonlinear deformation problem,
applying an incremental constitutive model, an updated Lagrangian
method is used. There are several methods for the fnite deformation
theory, namely, the total Lagrangian method is suitable for solids and
the Eulerian method has been used for fuids. Granular materials, such
as soil, are characterized as materials between solids and fuids. Hence,
the updated Lagrangian method is appropriate for describing soils. Since
the deformation is large and the reference confguration is updated at
each step, it is necessary to use an objective stress rate. Hence, in
this section, the objective Jaumann rate of Cauchy stress is adopted. Of
course, we can use another type of objective rate, such as Nagdhis rate.
The Jaumann stress rate is used since the strain level is moderate and less
than 100% (Johnson and Bammann 1984). In addition, the Jaumann
rate of Cauchy stress has been derived based on the double slip theory for
granular materials (Anand 1983).
6.5.1 Effective stress and fluidsolid mixture theory
Using Terzaghis concept of effective stress, the total stress tensor and the
time rate of stress are given as

T T u
ij ij w ij
= + (6.175)

T T u
ij ij w ij

= +
(6.176)
Virtual work theorem and finite element method 211
in which T
ij
denotes the total Cauchy stress tensor,

T
ij
denotes the effective
Cauchy stress tensor, u
w
denotes the pore water pressure, tension is posi-
tive (=
u p
w
=
; p is pressure and compression is positive for p),
ij
is the
second-order identity tensor, and the superimposed dots indicate the time
differentiation.
6.5.2 Equilibrium equation
When we consider an arbitrary domain V with a boundary S, the con-
servation of linear momentum for the whole fuidsolid mixture in the
current confguration is given by the following equation, as presented in
Chapter 1:

D
Dt
v dv t ds b dv
i
V
i
S
i
V


= + (6.177)
in which D/Dt is the material time derivative, is the mass density, v
i

is the velocity vector, t
i
is the surface traction vector, and b
i
is the body
force vector.
When we deal with quasi-static problems, the acceleration can be
assumed to be zero. Consequently, this assumption provides the equilib-
rium equation resulting from Equation (6.177) as

t ds b dv
i
S
i
V

+ = 0
(6.178)
When the body force is constant, the rate type of equilibrium equation is
expressed as follows:

D
Dt
t ds
i
S

= 0
(6.179)
Now let us discuss the equilibrium equation with respect to the reference
confguration. The balance of linear momentum with respect to the refer-
ence confguration,

D
Dt
v dV N dS b dV
i
V
ji j
S
i
V
0 0

= +
(6.180)
212 Computational modeling of multiphase geomaterials
where
ij
is the nominal stress tensor, N
ij
is the outward normal with
respect to the reference confguration,
0
is the mass density at the reference
confguration, and b
i
the body force vector.
For the quasi-static problems, taking a time derivative of Equation
(6.180) gives

ji j
S
i
V
N dS b dV

+ =
0
0
(6.181)
When we take the current confguration as the reference confguration
and use the Gauss theorem, Equation (6.181) becomes


S dv b dv
ji j
V
i
V
,

+ = 0
(6.182)
where the following relations presented in Chapter 1 are used:

ji
j
k
ki
J
X
x
S =

(6.183)


S T T L L T
ki ki kp ip pp ki
+ (6.184)
in which

S
ki
is the nominal stress rate tensor with respect to the current
confguration.
Without any body force increment, the material time derivative of the
equilibrium equations becomes

S dv
ji j
V
,

= 0
(6.185)
The effective nominal stress rate tensor,

S
ij
'
, is given by the following
equation:

S T T L L T
ij ij ip jp pp ij
'

= +
(6.186)
To obtain the relation between

S
ki
and

S
ij
'
, we substitute the defnition for
the effective Cauchy stress and the effective Cauchy stress rate, Equations
(6.175) and (6.176), respectively, into Equation (6.184), namely,


S S u L u u L
ij ij w ij pp w ij w ji
= + +
'
(6.187)
Virtual work theorem and finite element method 213
By letting

U L u u L
ij pp w ij w ji
= (6.188)
Equation (6.187) becomes


S S u U
ij ij w ij ij
= + +
'

(6.189)
When we consider closed domain V, the weak form of the rate type of
equilibrium equation is given as follows:

S v dv
ji j
V
i ,

= 0
(6.190)
in which v
i
is the virtual velocity vector.
The boundary of domain V is composed of a velocity boundary and
a traction boundary. The velocity boundary is denoted by S
u
if the
velocity is prescribed; S
t
denotes the traction boundary if traction is
prescribed.

v v
i i
=

on S
u
(6.191)

s S n s
i ji j i
= =

on S
t
(6.192)
in which v
i
is the velocity vector, n
i
indicates the unit normal to the body,
s
i
is the nominal traction rate vector, and the specifed values are desig-
nated by a superposed bar.
By taking the Gauss theorem and the compatibility condition, Equation
(6.190) can be written as


S v n ds S L dv
ji
S
i j ji ij
V

= 0
(6.193)
Considering the weak form of the boundary conditions on S
t
and sub-
stituting Equations (6.186) and (6.189) into the second term of Equation
(6.193), and transforming the frst term by Equation (6.192) yields

T D dv D T L dv T L L dv u D dv
U L dv s v ds
ij
V
ij kk
V
ij ij jp
V
ip ij w
V
kk
ji
V
ij i
S
i
0
' ' '

+ +
+ =
(6.194)
214 Computational modeling of multiphase geomaterials
in which D
ij
is the stretching tensor and the symmetry of the effective
Cauchy stress is used.
An updated Lagrangian method is used in this formulation based
on the finite deformation theory. Thus, the Jaumann rate of effective
Cauchy stress tensor,

'
T
ij
, is employed as an objective stress rate, which
leads to

' ' '

T T W T T W
ij ij ik kj ik kj
= +

(6.195)
where W
ij
is the spin tensor.
The elastoviscoplastic constitutive model can be written as

'
T C D C D
ij ijkl kl ijkl kl
vp
=
(6.196)
where D
ij
is the total stretching tensor, D
ij
vp
is the viscoplastic stretching
tensor, and C
ijkl
is the elastic stiffness tensor of the fourth order.
6.5.3 Continuity equation
For describing the motion of pore water, Biots two-phase mixture theory
(1956) is used in the analysis with a v
i
f
(velocity) u
w

(pore pressure; ten-
sion is positive) formulation. When the soil particles and the pore water are
incompressible, Darcys law and the conservation of mass for the mixture
yield the continuity equation as

k
u D
w
w kk

+
2
=0
(6.197)
where k is the coeffcient of permeability and spatially constant, and
w
is
the unit weight of the pore water.
The boundary for the pore water pressure is decomposed into two parts,
S
p
and S
q
. S
p
is the boundary at which the pore pressure is specifed and S
q

is the boundary at which the fow of pore water is specifed as

u
u
S
w w p
on = (6.198)

q
k
u q S
i
w
w i
f
q
on =

= (6.199)
Virtual work theorem and finite element method 215
in which the specifed values are designated by a superposed bar and q
i
f
is
the fow velocity of the pore water through the boundary surface.
Weak forms of the continuity equation and boundary conditions are
given by

k
u
u
dv D
u
dv
w
w w
V
kk w
V

+ =

2
0

(6.200)

( ) u u u ds
w w w
S
p
=

0 (6.201)

( ) u q q ds
w i
f
i
f
S
q
=

0 (6.202)
In order to derive a weak form of the continuity equation, we will use
a test function that is zero
w u

= 0 at the boundaries, S
p
where the pore
pressure is prescribed because the positive defniteness of the resultant coef-
fcient matrix (Atluri 2004) can be assured and the solution exists, and in
practice we will consider the prescribed value at the boundaries during the
construction of the fnal matrix. Of course we can use trial functions, that
is, approximation functions that satisfy the prescribed boundary conditions
such as S
p
(Zienkiewicz and Taylor 1997; Lewis and Schrefer 1998).
Considering the aforementioned test function,
w u , we can obtain the
weak form of the continuity equation with the boundary conditions as

k
u
u
dv D
u
dv u q q n ds
w
w w
V
kk w
V
w i
f
i
f
S
i
q

+ =

2
0

( ) (6.203)
By applying the Gauss theorem, the following equation can be obtained:

+ + =

k
u
u dv
u
D dv
u
q n ds
w
w w
V
w kk
V
w i
f
i
S
q

0
(6.204)
It is worth noting that if test function u
w
= 0 on S
p
or the trial function
satisfes the boundary condition on S
p
and zero on other boundary, the
positive defniteness of the resultant matrix is assured, that is, the solution
exists.
216 Computational modeling of multiphase geomaterials
6.5.4 Discretization of the weak forms
for the equilibrium equation and
the continuity equation
6.5.4.1 Discretization of the weak forms
for the equilibrium equation
In the following, we will use the Galerkin method in which the interpola-
tion function such as the shape function is used as a test function. For the
discretization of the weak form of the equilibrium equation, the following
relations are used:

v N v [ ]
{ }
= ,

(6.205)
in which {v} is the velocity vector in an element,
v

is the nodal velocity vec-

tor, and [N] is the shape function of the element. It should be noted that the
square brackets [ ] denote the matrix and the braces { } denote the vector.

D B v
{ }
= ,

[ ]
(6.206)
in which [B] is the matrix which transforms the nodal velocity vector to the
vector form of the stretching tensor {D}.

L B v
M
{ }
= ,

[ ] L B v
M
{ }
=

[ ] (6.207)
where [ ] B
M
is the matrix that transforms the nodal velocity vector into the
vector form of the velocity gradient vector {L}.

D B v
kk
T
v
[ ] =

(6.208)
where [ ] B
v
is the vector that transforms the nodal velocity into the trace
of D
ij
.

w h w u
N
u
=
{ }

[ ]
(6.209)
in which uw
{ }
represents the pore pressure rate, uw
{ }

represents the nodal

pore pressure rate vector, and [ ] N
h
represents the shape function.
In the analysis, the tangent modulus method for the rate-dependent
material (Peirce et al. 1984) is used in order to evaluate the viscoplastic
stretching tensor,
vp
D . This method can be classifed as a forward gradient
method and has been well adopted in viscoplastic analyses (see Oka et al.
1995; Kimoto et al. 2004; Higo et al. 2006). It is known that the method
Virtual work theorem and finite element method 217
is stable and yields accurate results for time steps larger than those for the
Euler explicit method (Peirce et al. 1984). This method is identical to the
tangent modulus of the semi-implicit backward Euler scheme on the frst
iteration (Belytschko et al. 2000). In the formulation, the trapezoidal rule
can be used to evaluate the value of the viscoplastic strain, in which the
value of trapezoidal parameter (Peirce et al. 1984) is taken as 0.5 in the
present analysis. = 1 corresponds to the usual tangent modulus.
The relation between the rate of effective stress and the stretching tensor
for the elastoviscoplastic model presented in Chapter 5 can be written in
matrix form, as shown in the following equation:

{

} [ ] =
{ }

{ }
T C D Q (6.210)
where [C] is the elastoviscoplastic tangential stiffness matrix and {Q} can
be called the relaxation stress vector.
Substituting Equation (6.195) into Equation (6.210) yields

{ } [ ] { }
*

=
{ }

{ }
+ T C D Q W
(6.211)
where { }
*
W is the column vector related to the spin tensor and the compo-
nents are computed via W W T T W
ij ik kj ik kj
*
= .
By all the matrix and vector relations obtained previously, and based on
the theory of virtual displacement, we have obtained the following relation
considering the arbitrariness of the unconstrained virtual nodal velocities:

K v B Q dv B W dv K v K
u
F
V V
L v w
[ ] [ ] [ ] [ ] [ ] { }
T T

{ } { } + + + =
{ } { } { }

(6.212)
in which

[ ] [ ] [ ][ ] K B C B dv
V
=

T
(6.213)

[ ] [ ] [ ][ ] [ ] [ ][ ] [ ] K B
D
B dv B U B dv B
L
V
M s M
V
M M
V
M
= + +

T T TT
T
{ }{ }
T B dv
v

(6.214)

[ ] [ ][ ] K B N dv
v
V
v h
=

(6.215)

{ } [ ] { }

F N s ds
S
t
=

T
(6.216)
where {s} is the traction vector at the boundary.
218 Computational modeling of multiphase geomaterials
In the preceding equations, [ ]{ }
'
s ik jk D
L T L = , [ ]{ } U L L u u L
kk w ij w ik jk
= .
And, these matrix forms are given by

=

T L
D
B v
ik jk s M
'
[ ][ ]
(6.217)

L u u L U B v
kk w ij w ik jk M
=

[ ][ ] (6.218)
where [ ] B
M
is the nodal velocity- L
T
matrix and D
s
[ ] is the matrix that
expresses the matrix form of T L
ik jk
by L
T
.
The relation between nodal velocity vector { } v

and nodal displacement

increment vector { }

u can be obtained by using Eulers approximation as

{ }
{ }
v
u
t

(6.219)
Similarly, the pore water pressure can be obtained as

w
t t
w
t
w
u
u u
t
+

{ }

(6.220)
Substituting Equations (6.219) and (6.220) into Equation (6.212), the
weak form of the equilibrium equations is obtained, that is,

[[ ] [ ]] [ ] { } [ K K u K u t F K
L v
t t
w
+ + = +

vv
t
w
u ]


+
{ }



t B Q dv t B W dv
V V
[ ] [ ]
T T
(6.221)
For the discretization of the continuity equation, Equation (6.204), the
following vectors and matrices are used:

u N u
u
N
u w h w w h w
= , =
{ }

[ ]

[ ]
(6.222)
where u
w

is the nodal pore pressure vector and N

h
{ } is the shape function
of the element.
The spatial gradient of the pore water pressure, u
w
, is discretized by

= =

u N u B u
w h w h w
[ ] [ ] (6.223)
Virtual work theorem and finite element method 219
in which [B
h
] is the matrix that transforms the nodal pore pressure into the
spatial derivative of the pore pressure.
6.5.4.2 Discretization of the weak form
for the continuity equation
Substituting Equations (6.222), (6.223), and (6.199) into Equation (6.204),
the following relation is given considering the arbitrariness of the uncon-
strained virtual nodal pore water pressures:
+


k
B B dv u N B dv v
w
V
h h w
V
h v

[ ] [ ] [ ] [ ]
T T T
+
{ } { }
=

p
N q n ds
h i
f
[ ]
T
T
0

(6.224)
Using Equation (6.219), the discretization of the continuity equation is
obtained as follows:

[ ] [ ] [ ] K u t K u t V
v h
t t
w
T
=

(6.225)
where

[ ] [ ] [ ] K
k
B B dv
h
w
V
h h
=

T
(6.226)

[ ] [ ] [ ] K N B dv
v
V
h v
T T T
=

(6.227)

[ ] [ ] V N q n ds
q
h i
f
=
{ } { }

T
T
(6.228)
Finally, the weak form of the rate type of equilibrium equation, Equation
(6.221), and the weak form of the continuity equation, Equation (6.225),
are given in the following matrix form:

[ ] [ ]
[ ]
[ ]
{ }
{ }
[ ]
K K
K
K t
u
u
L
v
v
T
K
w t t h
+

+
=
+ +
{ }

t F K u t B Q dv t
v
t
w
V V
{ }

T T

{ }

T
B W dv
t V
(6.229)
220 Computational modeling of multiphase geomaterials
In the formulation of Equation (6.229), the incremental governing equa-
tions are evaluated at t+t. As shown in Equation (6.225), the pore water
pressure is implicitly evaluated at t+t.
For the element type, for example, in the three-dimensional analysis of
Chapter 8, a twenty-node quadrilateral isoparametric element with a reduced
Gaussian eight-point integration is used to eliminate shear locking as well as
to reduce the appearance of a spurious hourglass mode. The pore water pres-
sure is defned by an eight-node quadrilateral isoparametric element.
REFERENCES
Anand, L. 1983. Plane deformations of ideal granular materials, J. Mech. Phys.
Solids, 31(2):105122.
Atluri, S.N. 2004. The Meshless Method for Domain and BIE Discretizations,
Techno Science Press, Chapter 1.
Bathe, K.J. 1996. The Finite Element Procedure, Prentice Hall.
Belytschko, T., Liu, W.K., and Moran, B. 2000. Nonlinear Finite Elements for conti-
nua and Structures, John Wiley & Sons, New York 290291.
Biot, M.A. 1956. Theory of propagation of elastic waves in a fuid-saturated porous
solid, J. Acoust. Soc. Am., 28(2):168178.
Christian, J.T., and Boehmer, J.W. 1970. Plane strain consolidation by fnite elements,
Proc. ASCE, 96(SM4):14351457.
Dhatt, G., and Touzot, G. 1984. The Finite Element Method Displayed, New York:
John Wiley & Sons.
Higo, Y., Oka, F., Jiang, M., and Fujita, Y. 2005. Effect of transport of water and
material heterogeneity on strain localization analysis of fuid-saturated gradi-
ent-dependent viscoplastic geomaterial, Int. J. Numer. Anal. Meth. Geomech.
29:495523.
Higo, Y., Oka, F., Kodaka, T., and Kimoto, S. 2006. Three-dimensional strain local-
ization of water-saturated clay and numerical simulation using an elasto-visco-
plastic model, Philos. Mag., 86(2122):32053240.
Hisada, T., and Noguchi, H. 2002. Fundamentals and Application of Non-Linear
Finite Element Method, Maruzen, Tokyo (In Japanese).
Johnson, G.C., and Bammann, D. 1984. A discussion of stress rates in fnite deforma-
tion problems, Int. J. Solids Struct., 20(8):725737.
Kato, M. 1995. Study on multi-dimensional liquefaction analysis method and its
application, Doctoral thesis, Gifu University (In Japanese).
Kimoto, S., Oka, F., and Higo, Y. 2004. Strain localization analysis of elasto-viscoplas-
tic soil considering structural degradation, Int. J. Comput. Meth. Appl. Mech.
Eng., 193(2729):28452866.
Lewis, R.W., and Schrefer, B.A. 1998. The Finite Element Method in the Static and
Dynamic Deformation and Consolidation of Porous Media, Chichester, UK: Wiley.
LIQCA Research and Development Group (Representative Fusao Oka, Kyoto
University). 2005. Users manual for LIQCA2D04 (2005 released print), http//
nakisuna2.kuciv.kyoto-u.ac.jp/liqca.htm
Virtual work theorem and finite element method 221
Mirjalili, M., Kimoto, S., Oka, F., and Higo, Y. 2011. Elasto-visoplastic modeling of
Osaka soft clay considering destructuration and its effect on the consolidation
analysis of an embankment, Geomech. Geoeng.: Int. J., 6(2):6989.
Oka, F. 2000. Elasto-Viscoplastic Constitutive Model for Geomaterials, Morikita
Syuppan (In Japanese).
Oka, F., Adachi, T., and Okano, Y. 1986. Two-dimensional consolidation analysis
using an elasto-viscoplastic constitutive equation, Int. J. Numer. Anal. Meth.
Geomech., 10:116.
Oka, F., Adachi, T., and Yashima, A. 1995. A strain localization analysis using a vis-
coplastic softening model for clay, Int. J. Plasticity, 11(5):523545.
Oka, F., Higo, Y., and Kimoto, S. 2002. Effect of dilatancy on the strain local-
ization of water-saturated elasto-viscoplastic soil, Int. J. Solids Struct.,
39(1314):36253647.
Oka, F., Kodaka, T., and Kim, Y.-S. 2004. A cyclic viscoelastic-viscoplastic consti-
tutive model for clay and liquefaction analysis of multi-layered ground, Int.
J.Numerical Anal. Methods Geomech., 28(2):131179.
Oka, F., Yashima, A., and Kohara, I. 1992. A fnite element analysis of clay foun-
dation based on fnite elasto-viscoplasticity, Proceedings of 4th International
Symposium on Numerical Models in Geomechanics, Swansea, G.N. Pande,
and S. Pietruszczak, eds., Balkema, 2:915922.
Oka, F., Yashima, A., Shibata, T., Kato, M., and Uzuoka, R. 1994. FEMFDM cou-
pled liquefaction analysis of a porous soil using an elasto-plastic model, Appl.
Sci. Res., 52:209245.
Peirce, D., Shih, C., and Needleman, A. 1984. A tangent modulus method for rate
dependent solids, Comput. Struct., 18(5):845887.
Schrefer, B.A., Majorana, C.E., and Sanavia, L. 1995. Shear band localization in
saturated porous media, Arch. Mech., 3:577599.
Terada, K. 2008. Common knowledge of computational mechanics, JSCE, Maruzen,
4445 (In Japanese).
Togawa, H. 1975. Vibration Analysis by Finite Element Method, Saiennsu Sya. (In
Japanese)
Vardoulakis, I., and Sulem, J. 1995. Bifurcation Analysis in Geomechanics, Blackie
Academic & Professional.
Washizu, H. 1982. Variational Methods in Elasticity and Plasticity, 3rd ed.,
Pergamon Press, Oxford.
Zienkiewicz, O.C. 1977. The Finite Element Method, 3rd ed., McGraw-Hill,
London.
Zienkiewicz, O.C., and Taylor, R.L. 1997. The Finite Element Method, 4th ed.,
McGraw-Hill, London.
223
Chapter 7
Consolidation analysis
It is well known that there are two types of time-dependent behavior for
soil. One is consolidation and the other is brought about by the inherent
viscous nature of the soil skeleton. The interaction between the pore water
and the soil skeleton results in consolidation. The viscous properties of the
soil skeleton are related to the microstructure of the soil particles. Although
many problems due to the consolidation of various types of soil have been
solved, some problems still exist. One of them is the interaction between
the viscosity and the changes in the soil structure. In the following, two
problems will be discussed. One is the infuence of the soil specimen thick-
ness on consolidation and the other is the interaction between the visco-
plastic properties and the strain softening due to structural changes.
7.1 CONSOLIDATION BEHAVIOR OF CLAYS
It has been reported that the infuence of specimen thickness on consolida-
tion plays an important role in the prediction of the actual settlements due
to the consolidation (Aboshi 1973; Ladd et al. 1977; Aboshi and Matsuda
1981; Oka et al. 1986; Leroueil 1995; Mesri et al. 1995; Tang and Imai 1995;
Oka 2005). As is well known, in the general report for the 9th ICSMFE,
Ladd et al. (1977) showed two hypotheses for consolidation behavior by
compiling the previous results (Figure7.1). Curve A is supported by Ladd et
al. (1977) and Mesri and Rokhsar (1974). Curve B is based on the hypoth-
esis that there is a unique stressstraintime relationship with respect to
time-dependent characteristics and that creep deformation occurs from the
beginning of the consolidation. Curve C is between curves A and B, and
appears to correspond to the experimental results of Aboshi (1973, 1995,
2004), shown in Figure7.2.
Many researchers have reported that the anomalous pore pressure
behavior, pore pressure stagnation, or a continuous increase after all the
fll placement, which appears to be associated with the collapse of the soil
structure, can be recognized during the consolidation process. Bishop and
224 Computational modeling of multiphase geomaterials
Lovenbury (1969) conducted constant stress creep tests under drained con-
ditions on undisturbed clay and observed a sudden increase in the strain
rate. Concerning feld cases, the anomalous pore pressure behavior dur-
ing the consolidation of soft clay has been reported by many researchers.
Mesri and Choi (1979) frst pointed out that this kind of problem was based
on the in situ measured results of the pore water pressure. Mitchell (1986)
reported this behavior as a surprising behavior in soil mechanics due to
a structural breakdown during compression. Furthermore, Leroueil (1988)
observed increases in the pore water after the completion of the construction
of test embankments, refecting the fact that the effective stress temporarily
diminished in the stressstrain curve. The prediction of these phenomena by
Time in Min
0.1
0
2
4
6
8
10
1 10 10
2
10
3
10
4
1

d
a
y
1

y
e
a
r
1
0

y
e
a
r
s
10
5
10
6
S
e
t
t
l
e
m
e
n
t

i
n

S
t
r
a
i
n

(
%
)
c

= 0.84%
Time min
0
0.01
0.02
0.03
3.89 4.07 4.19
10
6
10
6

(
%
)
10
6
Test No. 1 2 3 4 5
C
v
(cm
2
/min)

100
(%)
t
t 7.3
8.25
5.110
2
4.210
2
7.4
6.110
2
4.710
2
8.2
8.3
4.710
2
4.710
2
7.8
8.35
8.210
2
7.910
2
8.4
8.9
6.410
2
5.610
2
log t
log t
8.2
c

= 0.34%
Figure7.2 Comparison of settlement curves. (From Aboshi, H., 1973, Proc. 8th ICSMFE,
Moscow, 4(3):88.)

log t
Tick sample
Tin sample
Curve A
Curve C
Curve B
Figure7.1 Schematic diagram of the average strain to time.
Consolidation analysis 225
numerical methods has been studied since the 1980s. Kabbaj et al. (1986)
analyzed one-dimensional creep tests by the fnite difference method using
an elastoviscoplastic constitutive model (Oka 1981). They showed that the
strain rate remained momentarily constant during creep simulations around
the preconsolidation pressure.
7.2 CONSOLIDATION ANALYSIS:
SMALL STRAIN ANALYSIS
7.2.1 One-dimensional consolidation problem
To examine the effect of a sample thickness of clay on consolidation phe-
nomena, we used the constitutive equations introduced in the previous sec-
tion, which were developed by Adachi and Oka (1982), for the consolidation
analysis of clay strata of different heights. The fnite element mesh is shown
in Figure 7.3. Four-node isoparametric elements are used. The material
B
1
1 cm
1

c
m
x
3
x
1
x
2
B
2
B
2
B
2
: Impervious boundary
B
1
: Drainage boundary
Figure7.3 Finite element mesh.
226 Computational modeling of multiphase geomaterials
parameters are listed in Table7.l. The coeffcient of permeability is assumed
to be determined by

= k k e e C
k
exp[( )/ ]
0 0
*
(7.1)
where k
0
is the permeability coeffcient, C
*
k
is a material constant, and the
subscript 0 indicates the initial value.
For a one-dimensional consolidation problem, the fully implicit scheme
is used to calculate the viscoplastic strain increment.
Vertical settlementtime curves for different heights are shown in
Figure7.4. The effect of the length of the drainage path has been well inves-
tigated. Various hypotheses, which can explain the previously mentioned
scale effect, are illustrated schematically in Figure7.1. Curve A is supported
by Ladd (1973) and by Mesri and Rokhsar (1974); curve B is supported by
Barden (1969); curve C appears to correspond to the experimental results
of Aboshi. The calculated results denoted by the solid lines in Figure7.4 are
nearly equal to those of curve B. In this case, the initial viscoplastic strain
H = 10 cm
H = 25 cm
H = 25 cm
H = 5 cm
C = 1.0 10
9
(1/sec)
C = 4.0 10
11
(1/sec)
0
1
2
3
O
v
e
r
a
l
l

S
t
r
a
i
n

(
%
)
10
1
10
2
10
3
Time (min)
10
4
10
5
Figure7.4 Settlementtime curve.
Table7.1 Material parameters
M
*
m
m C
*
k
k
0
(m/sec)
0.231 0.05 0.865 25.0 0.1 1.16 10
9
G
0
(kPa) e
0
K
0

22(0)
(kPa)
u
w(0)
(kPa)
3675.0 1.5 0.5 196.0 98.0
Consolidation analysis 227
rates are assumed to be equal. In the experimental study by Aboshi (1973),
howeve, the initial strain on the four clay specimens of different heights is
almost equal, but the initial volumetric strain rate for the thick sample is
lower than that for the thin sample.
As has been pointed out by Sekiguchi and Toriihara (1976), the value
of the initial volumetric strain rate, v CM
p
m
( ),
(0)
*
= plays an important role
in the consolidation analysis. Predictably, the values of

v
p
(0)
for the thick
clay deposits in the feld may differ from those for the thin samples used in
laboratory tests. If we hypothesize that

v
p
(0)
is proportional to H
2
(H is the
height of the specimen), the broken line in Figure 7.4 is obtained. In this
case, the tendency is for the settlementtime curve to approach the value of
curve A in Figure7.1. The calculated results are very sensitive to the value of

= C v M
p
m
/
(0)
*
, as seen in Figure7.4. Clearly, it is then necessary to accurately
determine the value of C, in other words, the initial viscoplastic strain rate,
when we want to obtain good predictions of the settlementtime curve. If a
unique relation among stress, strain, and the strain rate exists, that is, if the
concept of isotaches (Suklje 1969) holds true, theoretically we can predict
that the effect of sample thickness is described by curve B in Figure7.1.
Parameter C also refects the effect of the duration of the preceding sec-
ondary compression, that is, quasi-overconsolidation. A heavily aged clay
stratum has a large value for the initial volumetric strain in comparison to
the value for normally consolidated clay, that is, young clay. Conversely,
if we assume that the value of the initial volumetric strain is zero for both
types of clay at the beginning of consolidation, then the value of C for aged
clay is smaller than that for young clay.
The theoretical effect of the value of C on the excess pore water pressure
time profle is shown in Figure7.5. The calculated results are supported by
the experimental results of Barden (1969) and of Mesri and Choi (1979). The
curve that corresponds to the smaller value of C is S-shaped and has an infec-
tion point. Leroueil, Lebihan, and Tavenas (1980) reported similar results in
one-dimensional tests on undisturbed samples of Champlain clay in relation
to the determination of the preconsolidation pressure. They took into account
the effective stress at the infection point of the u t
W
curve of the preconsoli-
dation pressure. The e
m
ln relation, in which the value of
m
at point A,
corresponding to point A in Figure7.5, is given in Figure7.6. From the above
numerical study, we conclude that the proposed theory can also predict the
one-dimensional consolidation behavior of undisturbed aged clay.
Next, we discuss the stress path during the one-dimensional consolida-
tion of a clay stratum. Akai and Adachi (1965) and Moore and Spencer
(1972) have shown that the stress ratio of vertical to horizontal effec-
tive stress is nearly constant during the one-dimensional consolidation of
normally consolidated clay. The solid line in Figure 7.7 shows the stress
path when H is 5 cm. In the early stages of consolidation, the stress ratio
decreases, after which it approaches the line defned by the initial stress
228 Computational modeling of multiphase geomaterials
ratio, the so-called K
0
-line. Comparing the two curves, we conclude that
the separation of the stress path from the K
0
-line is caused by the elasticity
of the soil and that the stress path approaches the K
0
-line because of the
viscoplasticity of the soil. Similar experimental results have been obtained
by Akai and Adachi (1965). A detailed interpretation of these results, based
on Equation (5.15) follows.
H = 25 cm
A
1.5
1.4
1 1.33

m
(kgf/cm
2
)
2
V
o
i
d

R
a
t
i
o

e
Figure7.6 Void ratiomean effective stress relation.
H = 25 cm
A
C = 1.0 10
9
(1/sec)
C = 0.2 10
7
(1/sec)
1.0
0.5
0.0
10
2
10
3
Time (min)
10
4
U
w
/
U
w
(
0
)
Figure7.5 Pore water pressuretime relations.
Consolidation analysis 229
When the lateral deformation is constrained, that is,
11
= 0 under plane
strain conditions, the stress increments are given by

=

K G GCC q q t
KCC M
s
t
m m
m
kl kl kl
m
2
3
2
3
1
{ / ( / ) }
( )
11 22 3
*
(0)
3
*
*
(0)
*
*
(7.2)

= +

K G GCC q q t
KCC M
s
t
m m
m
kl kl kl
m
4
3
2
3
2
{ / ( / ) }
( )
22 22 3
*
(0)
3
*
*
(0)
*
* '
(7.3)

=
33 11
(7.4)
in which q =
22

11
and subscript (0) denotes the values at the end of the
anisotropic consolidation,namely,
=
+

= + + K
e
C m M e v
m
m m me
p
(1 )
and exp[ { / ln( / ) (1 ) /( )}].
3 (0)
* *

In this case, the shear stress is negligible. When
q q
m m
[ / ( / ) ]
(0) is posi-
tive, the second term on the right-hand side of Equation (7.2) is positive and
0 2

v
(=
22
) (kgf/cm
2
)

h

(
=

1
1
)

(
k
g
f
/
c
m
2
)
3
2.0
1.5
1.0
H = 5 cm
C = 1.0 10
9
(1/sec)
Figure7.7 Effective stress path.
230 Computational modeling of multiphase geomaterials
that of Equation (7.3) is negative. Therefore, the difference between
22
and

11
decreases and the stress path asymptotically approaches the state defned
by the K
0
-line. The stress state of clay element returns to its original stress
ratio in all cases. If K
0
is 0.5, the assumption that the effective stress ratio is
constant during the one-dimensional consolidation is true only when the elas-
tic Poissons ratio is l/3. This means that the memory of the initial stress ratio
does not fade in soil defned by the proposed constitutive equations for clay.
7.2.2 Two-dimensional consolidation problem
The response of clay foundations during the construction of embankments
has been explained in detail on the basis of feld data collected by Tavenas
and Leroueil (1980). They reported that none of the existing analytical
methods can universally describe feld data. In this section, we have tried
to account theoretically for the real behavior of clay foundations during the
construction of embankments.
The behavior of clay foundations under plane strain conditions during
the construction of embankments can be analyzed numerically using elas-
toviscoplastic constitutive equations and Biots consolidation theory. The
fnite element mesh and the boundary conditions are shown in Figure7.8.
The material parameters and the initial conditions are listed in Table7.2.
The time increment in the calculation is 8 h. Coeffcient of permeability,
k, and elastic shear modulus, G, are determined as in the one-dimensional
Number of
Nodal point
Number of
element
q
2
5
9
15
24
(m)
(m) 34 24 15 9 5 2 0
30 29 28 27 26 25
24 23 22 21 20 19
18 17 16 15 14 13
12 11 10 9 8 7
6 5 4 3 2 1
6 7 5 4 3 2 1
0
Figure7.8 Finite element mesh.
Consolidation analysis 231
problem. Eulers time integration method and the fully implicit scheme are
used to calculate the strain increment.
The following type of loading rate, dq/dt, is also used:
dq/dt = 0.2 ton/m/day (calculation being stopped at 100 days)
The relation between the excess pore water pressure, u
w
, and the total ver-
tical stress increment,
v
, of element 1 in case A is shown in Figure7.9. The
ratio of excess pore water pressure to the vertical stress increment, u
w v
/ ,
is almost 0.75 during the frst 33 days after construction of the embankment
has begun; thereafter, u
w v
/ increases and becomes equal to 1.0.
After compiling many feld observations, Tavenas and Leroueil (1980)
found the typical pattern for the ratio of the excess pore water pressure
increment under the center of fll to the applied vertical stress on the clay
foundation during the construction of an embankment. It is similar to the
calculated result given earlier.
They explained this phenomenon as follows: as natural clay is initially
characterized by the typical properties of overconsolidated clay (high
1
1
1.5
1.0
0.5
0.0
0.5 1.0

v
(
v

v0
) (kgf/cm
2
)
1.5
U
w

(
k
g
f
/
c
m
2
)
B
u
(33 day)
Explicit scheme
Implicit scheme
Figure7.9 Excess pore water pressuretotal vertical stress relations.
Table7.2 Material parameters
M
*
m
m C
*
k
k
0
(m/sec)
0.231 0.05 0.865 25.0 0.1 1.16 1.0
10
G
0
(kPa) e
0
K
0

22(0)
(kPa) C (1/s)
1960.0 1.5 0.5 98.0 1.0 1.0
12
232 Computational modeling of multiphase geomaterials
stiffness), the rate of pore water pressure dissipation is high at the begin-
ning of construction. At that time, u
w v
/ , is smaller than unity. As the
construction advances, however, the pore water pressure increment becomes
equal to the embankment load increment, u
w v
/ , being almost equal to
unity due to the passage of the clay to the state of normally consolidated
clay. One possible reason why the proposed theory can explain the phenom-
enon described is that the proposed constitutive equations can describe the
quasi-overconsolidated effect of natural clays due to aging, as discussed in
the previous section on one-dimensional consolidation.
Tavenas and Leroueil (1980) suggested that it is appropriate to describe
the lateral displacement behavior by the relation of maximum lateral dis-
placement, y
m
, below the toe of the embankment, to the settlement s of the
center line. It can been seen in Figure7.11 that dy
m
/ds is small in the early
stage of loading, but it increases after 33 days. Tavenas and Leroueil (1980)
reported that infection point B
u
in Figure7.9 corresponds to point B
s
in
Figure7.11. In this case, the two points do agree.
The difference between the settlementlateral displacement curves
obtained by Eulers integration method and by the fully implicit scheme
increases with the increase in settlement. The settlementtime profle and the
excess pore water pressuretime profle of element l are given in Figure7.10.
Point C corresponds to the infection points B
u
and B
s
in Figures 7.9 and
7.11. The distributions with depth of the lateral displacement and the excess
pore water pressure build-up during construction are shown in Figure7.12.
1.0 10.0
Time (day)
100.0
50
0
100
S
e
t
t
l
e
m
e
n
t

s

(
c
m
)
U
w

(
k
g
f
/
c
m
2
)
2
1
0
C (33 day)
Settlement node 1
U
w
element 1
Implicit
Explicit
Figure7.10 Excess pore water pressuretime profle and settlementtime profle.
Consolidation analysis 233
20
10
0
5
50
10
100
Settlement s (cm)
15
150
(Node 3)
(Node 1)
B
s
: 33 day
B
s
B
s
Node
Implicit integration
Explicit integration
1 3
L
a
t
e
r
a
l

D
i
s
p
l
a
c
e
m
e
n
t

y
m

(
c
m
)
Figure7.11 Lateral displacementsettlement relations.
Lateral Displacement y
m
(cm)
14 12 10 8 6 4 2 0
2
5
9
15
24
D
e
p
t
h

x
2

(
m
)
y
m
Node 3
u
w
Element (2~26)
Implicit
Explicit
0.0 0.5
Excess Pore Water Pressure u
w
(kgf/cm
2
)
40 day
60 day
20 day
Figure7.12 Distributions of lateral displacement and excess pore water pressure with
depth.
234 Computational modeling of multiphase geomaterials
Both lateral displacement and excess pore water pressure are at maxima
between 2 and 5 m.
7.2.3 Summary
One-dimensional consolidation
1. The theory can describe the effect of the length of the drainage
path on the settlementtime curve. We also confrmed that the
initial volumetric inelastic strain rate has an important function
in the prediction of the settlementtime curve.
2. Parameter C in Equation (5.19) refects how long the clay stratum
has been aged in situ. Heavily aged clay has a smaller C value than
young clay.
3. The excess pore water pressuretime profle, which is peculiar to
aged clay, can be simulated by the proposed constitutive equations.
4. According to our proposed theory, the assumption that the ratio
of horizontal to vertical effective stress is nearly constant is true
only when the elastic material constants take special values.
Two-dimensional consolidation
1. The trend in the calculated relations between the lateral displace-
ment at the toe of the embankment and the settlement under the
center of the embankment during construction is in good agree-
ment with the empirical relations given by Tavenas and Leroueil
(1980) based on their feld observations.
2. The typical empirical relation between the applied vertical stress
increment and the excess pore water pressure increment in a
clay foundation during embankment construction, proposed by
Tavenas and Leroueil (1980), is very similar to the calculated
relations between the total vertical stress increment and the excess
pore water pressure increment.
3. With the theory, it is possible to simulate the behavior of a clay
foundation during the construction of an embankment.
7.3 CONSOLIDATION ANALYSIS WITH A MODEL
CONSIDERING STRUCTURAL DEGRADATION
In this section of the present study, one-dimensional consolidation behav-
ior is simulated using an elastoviscoplastic constitutive model taking into
account the effect of structural degradation. We will discuss the infuence of
sample thickness with respect to the soil structure and the initial strain rate.
One-dimensional consolidation has been numerically examined by a fnite
element analysis. For boundary value problems related to the soilwater
coupled consolidation problem, Biots two-phase mixture theory is adopted.
Consolidation analysis 235
The infnitesimal strain is valid for these problems since large deforma-
tions are not expected. A four-node quadrilateral element with a reduced
Gaussian integration is used for the displacement, and the pore water pres-
sure is defned at the center of each element. The top of the specimen is set to
be permeable while the bottom and the sides are set to be impermeable. The
size of each element is 0.4 cm 0.4 cm for all the calculations.
7.3.1 Effect of sample thickness
Simulations have been performed for normally consolidated clay. The ini-
tial stress conditions of the calculations and the material parameters are
shown in Table7.3. The elastic modulus, G
0
, was set at 36100 kPa in the
previous calculations (Kimoto 2002), and the settlement during the pri-
mary consolidation was much smaller than that obtained during the sec-
ondary consolidation. For comparison, G
0
is supposed to 360 kPa in the
present study. Viscoplastic parameter C, which describes the initial visco-
plastic strain rate, is set to be 1.0 10
13
(1/s) (=C
0
) at frst. An excess pore-
pressure level of 1160 kPa, which is twice as large as the compression yield
stress, is applied as the initial loading for all the analyses.
In the frst calculation, viscoplastic parameter C is assumed to be C
0

for samples with different heights (H = 2, 20 cm). Vertical strain reaches
almost the same value at 6 10
6
seconds (= 70 days). The results correspond
to curve B as shown in Figure7.1.
In the next calculation, an initial viscoplastic strain rate C is assumed to
be inversely proportional to H
2
, that is, C = C
0
(H
0
/H)
2
(H is the height of
the specimen and H
0
is set at 2 cm). The results are shown in Figure7.13.
Table7.3 Initial conditions and material parameters
Initial mean effective stress
m(0)
= 580 kPa
Coeffcient of earth pressure K
0
= 0.5
Coeffcient of permeability k
0
= 0.8 10
9
m/s
Permeability change index C
*
k
= 0.1
Elastic shear modulus G
0
= 360 kPa
Compression index = 0.508
Swelling index = 0.0261
Initial void ratio e
0
= 1.70
Compression yield stress
mai
(=
mai
) = 580 kPa
Stress ratio at maximum compression M
*
m
= 1.09
Viscoplastic parameter m = 18.5
Viscoplastic parameter C = 1.3 10
13
1/s (=C
0
)
Structural parameter
maf
= 290 kPa
Structural parameter = 0, 5, 20, 40
236 Computational modeling of multiphase geomaterials
The two lines (C = C
0
for H
0
= 2 and C= 0.01C
0
for H = 20) are parallel to
each other; this corresponds to curve A.
Aboshi (1973) experimentally observed that the initial strain rate for the
thick sample is lower than that for the thin one. The pre-compression as a pre-
paratory consolidation was performed with the stress change between 0 and
19.6 kPa. The data of the compression by Aboshi are shown as follows (from
private communication with Aboshi in 1983, presented in Oka 2000).
The initial strains,
v(0)
, are just after the preparatory consolidation. The
initial strain rates,

v(0)
, are average values between consolidation stresses of
9.819.6 kPa, except No. 5. For No. 5, the strain rate is calculated between
consolidation pressures of 019.6 kPa. The average initial water content is
80%; and then 58.8kPa has been applied in the consolidation.
Specimen No. 1 (H = 2 cm)
Pressure 09.8 kPa :
v
= 1.3% (duration time: 1440 min)
Pressure 9.819.6 kPa:
v
=3.4% (1440 min)
Average strain rate

v(0)
= 2.36 10
3
%/min (9.819.6 kPa)
Total strain after preparatory consolidation =
v
4.7%
(0)
Specimen No. 2 (H = 4.8 cm)
Pressure 09.8 kPa:
v
= 2.8% (1440 min)
1 10 100 1000 10000
Time (s)
100000 1000000 1E7
= 0
1200
1000
800
600
400
200
0
E
x
c
e
s
s

P
o
r
e

W
a
t
e
r

P
r
e
s
s
u
r
e
a
t

t
h
e

B
a
s
e

(
k
P
a
)
H = 2 cm C = C
0
H = 20 cm C = C
0
H = 20 cm C = 0.1 C
0
H = 20 cm C = 0.01 C
0
1 10 100 1000 10000
Time (s)
100000 1000000 1E7
1 day 1 year
1 year
= 0
0
2
4
6
8
10
12
14
16
18
A
v
e
r
a
g
e

V
e
r
t
i
c
a
l

S
t
r
a
i
n

(
%
)
H = 2 cm C = C
0
H = 20 cm C = C
0
H = 20 cm C = 0.1 C
0
H = 20 cm C = 0.01 C
0
1 day
Figure7.13 Effects of parameter C (without structure = 0, C
0
= 1.3 10
13
).
Consolidation analysis 237
Pressure 9.819.6 kPa:
v
= 2.5% (1440 min)
Average strain rate

v(0)
= 1.736 10
3
%/min (9.819.6 kPa)
Total strain after preparatory consolidation =
v
5.3%
(0)
Specimen No. 3 (H = 20 cm)
Pressure 09.8 kPa:
v
= 1.9% (11000 min)
Pressure 9.819.6 kPa:
v
= 2.2% (21000 min)
Average strain rate

v(0)
= 1.05

10
4
%/min (9.819.6 kPa)
Total strain after preparatory consolidation =
v
4.1%
(0)
Specimen No. 4 (H = 40 cm)
Pressure 09.8 kPa:
v
= 2.3% (20000 min)
Pressure 9.819.6 kPa:
v
= 2.7% (42000 min)
Average strain rate

v(0)
= 6.43 10
5
%/min (9.819.6 kPa)
Total strain after preparatory consolidation =
v
5.0%
(0)
Specimen No. 5 (H = 100 cm)
Pressure 019.6 kPa:
v
= 4.7% (62000 min)

v(0)
= 7.58 10
5
%/min (019.6 kPa)
Total strain after preparatory consolidation =
v
4.7%
(0)
As shown, the initial strain rate of the thick sample just before consolida-
tion is smaller than that of the thin sample, although the strain just after
the preparatory consolidation is almost equal. The reason for the difference
in strain rates is that the periods of preconsolidation are different, namely,
from one day for the thinner sample to four months for the thicker one.
In the calculation in Figure7.13, we assumed that C
0
(H
0
/H)
2
< C = C
0

< H H C ( / )
0 0
, namely, C = 0.1 C
0
for H = 20 cm. Considering Equations (5.27)
and (5.67), the assumption leads to viscoplastic parameter C depending
on the initial viscoplastic strain rates and

( ) ( )
< = < C C C C
i i i i
/ / ,
0 0
2
0 0 0

where

i
is the initial strain rate before consolidation and

i0
is the initial
strain rate before consolidation of the sample with a height of H
0
, although
the strain rate in Equations (5.27) and (5.67) is the viscoplastic volumetric
strain rate.
In this case, curve C is obtained when the value of C is larger than

( )
C
i i
/
0 0
2
and smaller than C
0
, which is consistent with Aboshis experi-
ments. Consequently, it was found that the effect of the sample thickness
depends on the value of C, that is,


( )
= C C
i i
A / (for curve )
0 0
2

C C C C
i i i i
C ( / ) ( / ) (for curve )
0 0
2
0 0 0
< = <

C = C
0
(for curve B)
The effect of sample thickness on the timesettlement curve is mainly due
to the difference in strain rates before consolidation. This explanation was
238 Computational modeling of multiphase geomaterials
frst pointed by Oka, Adachi, and Okano (1986), and was confrmed by
Tang and Imai (1995). Of course there might be other reasons such as sample
disturbance and natural variability (Mesri and Choi 1985; Leroueil 1995).
7.3.2 Simulation of Aboshis experimental results
In this section, Aboshis consolidation test results with different thicknesses
have been numerically simulated.
7.3.2.1 Determination of material parameters
Aboshi (1973) performed consolidation tests on clays with different thick-
nesses. The specimens were obtained from Hiroshima Bay and reconsti-
tuted in a trench, 10 m 15 m with a depth of 1.5 m, for more than 6 years
with a thin sand layer on it. The material contains 27% clay, 68% silt, and
5% sand, and the plasticity index, I
p
, is 42% (Aboshi 1973). There are
two elastic parameters, six viscoplastic parameters, and two parameters
for hydraulic conductivity. Some assumptions are needed since there are not
enough data, such as triaxial test results. The determination of the param-
eters is shown in the following.
7.3.2.2 Elastic parameters
The swelling index, , is assumed as 0.2 of compression index . is
obtained as 0.304 from the e-logp curve in Matsuda and Aboshi (1983).
Initial shear modulus G
0
is determined by assuming that Poissons ratio is
equal to 0.3 at the initial state of the main loading.
7.3.2.3 Viscoplastic parameters
The consolidation yield stress,
mbi
, is equal to the initial mean effective
stress in the normally consolidation region. The stress ratio at maxi-
mum compression, M
m
*
, which is equal to the failure stress ratio in the
model, can be assumed to be around 0.73 to 0.98 from the relationship
between the plasticity index and the friction angle for the clays shown in
Leroueil and Hight (2003, p. 196). We take the mean value 0.85 for M
m
*
.
Viscoplastic parameter m is calculated from Equation (5.66) related to the
secondary compression rate =

C de d t / log . Aboshi and Matsuda (1981)

mentioned that the average value of the secondary compression coeffcient
=

C d d t / log is about 0.95%. However,

C differs for the different thick-

nesses of the samples and also gradually becomes smaller during the sec-
ondary consolidation process. We take the mean value of

C for H = 1 and
Consolidation analysis 239
50 cm from the experimental results by Aboshi and Matsuda (1981). We
performed simulations for the value of m from 22.0 to 33.0, which cor-
respond to

C from 0.98% and 0.63%. Finally, we used 31.0 of m. The

viscoplastic parameter, C
0
, is determined by ftting the settlement curve
for the specimen since we do not have the triaxial test data. For the void
ratio dependency of the permeability coeffcient, k, following Tang and
Imai (1995), C
k
*
is assumed to be 0.478.
7.3.2.4 Consolidation analysis
Simulations of one-dimensional consolidation have been conducted by
the fnite element analysis. The simulation method, that is, the soilwater
coupled consolidation analysis method based on Biots two-phase mixture
theory presented in Chapter 6, is used. A four-node quadrilateral element
with a reduced Gaussian one-point integration is used for the displacement,
and the pore water pressure is defned at the center of each element. The
infnitesimal strain is assumed. The fnite element meshes and boundary
conditions are shown in Figure7.14. The top of the specimen is set to be
permeable, while the bottom and the side are impermeable. The sample
thickness is assumed to be 1.0, 2.4, 10, 20, and 50 cm. First, the loading
is simulated from 9.8 kPa to 19.6 kPa as a preparatory consolidation. The
material parameters are listed in Table7.4. From the results of the prepa-
ratory consolidation, the strain rates at the end of the primary consolida-
tion have been obtained. Figure7.15 indicates the relationship between the
strain rate at the end of the primary consolidation and the height of the
Permeable boundary
Element number Impermeable
boundary
H = 1, 2.4, 10,
20, 50 cm
25
2
1
Figure7.14 Finite element meshes and boundary conditions.
240 Computational modeling of multiphase geomaterials
specimen for the preparatory consolidation in which the stress increment of
9.8 kPa has been applied to the specimen with an initial stress of 9.8 kPa,
that is, the fnal stress level is 19.6 kPa.
From Figure7.15, the average strain rate at the end of the primary con-
solidation is proportional to

H
1.1
. Hence, in the following analysis of
the main consolidation, we assumed that the initial strain rate is propor-
tional to

1.1
, that is,

H
1.1
. This is consistent with the fact that Aboshi
(1973) observed that it takes a much longer time for the preparatory
1/H
1.1
, H = Drainage Length (cm)
0.0 0.2 0.4 0.6 0.8 1.0
H = 2.4
H = 10
H = 20
H = 1
2.0 10
7
1.8 10
7
1.6 10
7
1.4 10
7
1.2 10
7
1.0 10
7
8.0 10
8
6.0 10
8
4.0 10
8
2.0 10
8
0.0
A
v
e
r
a
g
e

S
t
r
a
i
n

R
a
t
e

(
1
/
s
)
H = 50
Figure7.15 Relationship between strain rates at the end of primary consolidation and
H
1.1
during the preparatory consolidation.
Table7.4 Material parameters for the preparatory consolidation (9.819.6 kPa)
Coeffcient of static earth pressure, K
0
0.5
Initial mean effective stress,
mo
(kPa) 6.53
Initial void ratio 2.22
Initial permeability, k (m/s) 1.9 10
09
Permeability changing index, C
*
k
0.478
Compression index, 0.304
Swelling index, 0.0608
Initial shear modulus, G
0
(kPa) 221
Compression yield stress,
mbi
/
m0
1.0
Stress ratio at maximum compression, M
*
m
0.85
Viscoplastic parameter, m 31.0
Viscoplastic parameter, C (1/s) 4.00 10
13
Consolidation analysis 241
consolidation of a thicker sample, although the strain levels are almost
equal. The material parameters used in the analysis of the main con-
solidation are listed in Table 7.5. The applied stress increment is 58.8
kPa, and the fnal stress level is 78.4 kPa. Figure7.16 shows the simula-
tion of the consolidation with different thicknesses. This fgure indicates
the assumption that the viscoplastic parameter C is proportional to

1.1
,
that is,

H
1.1
, which is taken above is effective. The simulation results
obtained with the above assumption reproduce Aboshis experiments
very well, whereas the simulation results based on the other assumptions
do not.
7.3.3 Effect of degradation
In the calculations in Figure7.16, is set at zero. Next, let us discuss the
effect of the parameter on structural degradation . Calculations are per-
formed for different values of parameter (= 0, 5, 20, 40). The case of = 0
corresponds to the original model without considering structural changes.
The vertical settlement with time and the pore water generation with time
are shown in Figure 7.17. The excess pore pressure shows a temporary
increase when the rate of structural changes takes a high value, in other
words, = 40 (Figure 7.17b). The average void ratio e-log
m
relations
for = 0, 20, 40 are shown in Figure7.18. It shows a temporary decrease
in the mean effective stress for the e-log
m
relations for = 20 and 40. A
similar tendency is obtained from the test embankment (Leroueil 1988).
The infexion points in the curve correspond to points where the pore water
pressure begins to be stagnant or to increase in Figure7.17b, at about 40 s
for = 0 and 66 s for = 40.
Table7.5 Material parameters for the main consolidation
Coeffcient of static earth pressure, K
0
0.5
Initial mean effective stress,
m0
(kPa) 13.1
Initial void ratio 2.15
Initial permeability, k (m/s) 1.64 10
09
Permeability changing index, C
*
k
0.478
Compression index, 0.304
Swelling index, 0.0608
Initial shear modulus, G
0
(kPa) 313
Compression yield stress,
mbi
/
m0
1.0
Stress ratio at maximum compression, M
m
* 0.85
Viscoplastic parameter, m 31.0
Viscoplastic parameter C (1/s) (for H = 1 cm) 1.00 10
14
Height dependency of C (for H = 2.4, 10, 20, 50) 0, 1.1, 2.0
242 Computational modeling of multiphase geomaterials
Experiment
H = 1 cm
H = 2.4 cm
H = 10 cm
H = 20 cm
H = 50 cm
Experiment
H = 1 cm
H = 2.4 cm
H = 10 cm
H = 20 cm
H = 50 cm
Experiment
H = 1 cm
H = 2.4 cm
H = 10 cm
H = 20 cm
H = 50 cm
1 10 100 1000 10000
Time (min)
100000 1000000
0
2
4
6
8
10
12
S
e
t
t
l
e
m
e
n
t

(
%
)
1 10 100 1000 10000
Time (min)
100000 1000000
0
2
4
6
8
10
12
S
e
t
t
l
e
m
e
n
t

(
%
)
1 10 100 1000 10000
Time (min)
100000 1000000
0
2
4
6
8
10
12
S
e
t
t
l
e
m
e
n
t

(
%
)
Simulation
C = C
0
Simulation
C = C
0
= C
0
H
H
0
1.1

i0
1.1

i
Simulation
C = C
0
= C
0
H
H
0
2

i0
2

i
Figure7.16 Settlementtime relations during the consolidation with different viscoplas-
tic parameters.
Consolidation analysis 243
(a) Average vertical strain with time
1 10 100 1000 10000
Time (s)
100000 1000000 1E7
0
4
8
12
16
20
24
28
A
v
e
r
a
g
e

V
e
r
t
i
c
a
l

S
t
r
a
i
n

(
%
)
(b) Generated pore water pressure with time
1 10 100 1000 10000
Time (s)
100000 1000000 1E7
1200
1000
800
600
400
200
0
E
x
c
e
s
s

P
o
r
e

W
a
t
e
r

P
r
e
s
s
u
r
e
a
t

t
h
e

B
a
s
e

(
k
P
a
)
H = 2 cm
1 day 1 year
1 year
= 0
= 5
= 20
= 40
= 0
= 5
= 20
= 40
1 day
H = 2 cm
Time = 66 s
Time = 40 s
Figure7.17 Effects of structural parameter . (a) Average vertical strain with time. (b)
Generated pore water pressure with time.
1.7
1.6
1.5
1.4
1.3
V
o
i
d

R
a
t
i
o
1.2
1.1
100 1000
Mean Efective Stress (kPa)
= 0
= 20
= 40
Time = 66 s
Time
= 40 s
Figure7.18 Average e-log
m
curve.
244 Computational modeling of multiphase geomaterials
7.4 CONSOLIDATION ANALYSIS
OF CLAY FOUNDATION
7.4.1 Introduction
Stagnation or an increase in the pore water pressure after loading and dur-
ing the consolidation of soft clay is called anomalous pore pressure. It has
been observed after loading and during consolidation by J.K. Mitchell in
his 20th Terzaghi Lecture (1986). This problem has hitherto been studied,
but it has not yet been fully solved. The main reason is the rate-dependent
structural degradation of soft clay, as discussed in the previous section.
Oka et al. (1991) numerically analyzed such a phenomenon observed in
the clay foundation at St. Albans test embankment D (Leroueil et al. 1978;
Tavenas et al. 1974) using an elastoviscoplastic model. They used a model
with volumetric strain softening and a comparatively better solution
than the conventional model. However, the stagnation or the temporary
increase in pore water pressure after the construction of the embankment
could not be reproduced. In this section, we have analyzed the same prob-
lem using an elastoviscoplastic model with the aforementioned structural
degradation.
7.4.2 Consolidation analysis of soft clay
beneath the embankment
A comprehensive numerical investigation has been carried out for the con-
solidation behavior of soft sensitive clay in order to reproduce the feld
response. The soil parameters and the experimental results are related to
the clay foundation beneath the test embankments, constructed in 1972 at
Saint Alban, west of Quebec City, Canada (Tavenas et al. 1974; Leroueil
et al. 1978; Oka et al. 1994). The current simulation method is based on
the elastoviscoplastic constitutive model with structural changes (Kimoto
and Oka 2005), as described in Chapter 5. Based on the fnite deforma-
tion theory presented in Chapter 6, a two-dimensional fnite element code
(COMVI3D) has been used. The results of the analysis are compared with
the experimental ones. It is noted here that the embankment stiffness and
foundation roughness are not considered in the present study for simplicity
of the numerical implementation.
7.4.2.1 Soil parameters
The soil parameters and the model parameters for the Champlain clay
beneath the test embankments at St. Alban are listed in Table 7.6.
Consolidation analysis 245
Viscoplastic parameter C is chosen based on the viscoplastic strain, as
shown in Table7.7.
Instead of Equation (5.49), the degradation rule for the soil structure is
governed by = + z
ma maf mai maf
a
( )exp( ) and a = 0.5 is assumed. The
value of is taken as 5.856, and the other structural degradation param-
eters are chosen based on the following relations: = OCR
mai m mean (0)
and
=
maf m(0)
. We assumed that the top layer is modeled as being elastic with
a Poisson ratio = 0.25 and = 0.01125(1+e) (2.5 times higher than the
other layers). The basis for the selection of the values for the parameters is
to reproduce the feld responses as closely as possible.
7.4.2.2 Soil response beneath embankment
Figure 7.19a shows the optimal fnite element mesh confguration for the
fnite deformation analysis for the soil response beneath embankment D.
The construction history of embankment D is shown in Figure 7.19b. As
for the boundary conditions, the bottom boundary of the clay foundation is
Table7.6 Soil and material parameters
Layer
Depth
(m) e
0

22(0)

(kPa)

p

(kPa) G
0
(kPa) Others
1 0~0.66 0.100 1.1 981.6 72.5 880.0 = 0.0045(1+e)
2 0.66~1.5 0.363 1.7 1280.0 57.1 1500.9
= C e 0.5
k
*
3 1.5~3.0 0.495 2.3 1770.7 46.0 2079.7 m=27.24
4 3.0~4.8 0.411 1.8 2550.3 70.6 3011.7
M 0.98
m
*
=
5 4.8~6.7 0.282 1.8 3436.6 88.3 4041.2
= K 0.8
0
6 6.7~9.6 0.175 1.4 4417.2 13.7 5601.5
=

k 1.05 10 m/s
0
8
7 9.6~13.5 0.100 1.4 6871.3 17.7 10987.2
= kN m 9.81 /
w
3
Table7.7 Viscoplastic parameter C
Volumetric viscoplastic strain
vp
(%) C (1/sec)

vp
< 0.01 1.2 10
12
0.01
vp
22.2 5.9 10
11

vp
> 22.2 5.9 10
11
246 Computational modeling of multiphase geomaterials
0
1
2
3
4
5
6
7
8
9
1
0
1
1
1
2
1
3
1
4
1
5
1
6
1
7
1
8
1
0
9
1
1
0
1
1
1
1
1
2
1
1
3
1
1
4
1
1
5
1
1
6
1
1
7
1
1
8
1
1
9
1
2
0
1
2
1
1
2
2
1
2
3
1
2
4
1
2
5
1
2
6
9
1
9
2
9
3
9
4
9
5
9
6
9
7
9
8
9
9
1
0
0
1
0
1
1
0
2
1
0
3
1
0
4
1
0
5
1
0
6
1
0
7
1
0
8
7
3
7
4
7
5
7
6
7
7
7
8
7
9
8
0
8
1
8
2
8
3
8
4
8
5
8
6
8
7
8
8
8
9
9
0
5
5
5
6
5
7
5
8
5
9
6
0
6
1
6
2
6
3
6
4
6
5
6
6
6
7
6
8
6
9
7
0
7
1
7
2
3
7
3
8
3
9
4
0
4
1
4
2
4
3
4
4
4
5
4
6
4
7
4
8
4
9
5
0
5
1
5
2
5
3
5
4
1
9
2
0
2
1
2
2
2
3
2
4
2
5
2
6
2
7
2
8
2
9
3
0
3
1
3
2
3
3
3
4
3
5
3
6
3
.
4
5
6
.
9
1
0
.
3
5
D e p t h , H ( m )
1
3
.
8
0
9
1
8
2
7
3
6
4
5
H
o
r
i
z
o
n
t
a
l

D
i
s
t
a
n
c
e
,

L

(
m
)
5
4
6
3
7
2
C
L
E
m
b
a
n
k
m
e
n
t
C
l
a
y

f
o
u
n
d
a
t
i
o
n
1
4
.
0
4

3
.
2
8

m
1
3
.
4

m
3
.
8

m
F
i
g
u
r
e

7
.
1
9

(
a
)

S
h
a
p
e

o
f

t
h
e

e
m
b
a
n
k
m
e
n
t

D

a
t

S
t
.

A
l
b
a
n

a
n
d

F
E
M

m
e
s
h
.

(
b
)

C
o
n
s
t
r
u
c
t
i
o
n

p
r
o
c
e
s
s

f
o
r

e
m
b
a
n
k
m
e
n
t

D
.

(
A
d
a
p
t
e
d

f
r
o
m

L
e
r
o
u
e
i
l
,

S
.
,

T
a
v
e
n
a
s
,

F
.
,

T
r
a
k
,

B
.
,

R
o
c
h
e
l
l
e
,

P
.

L
.
,

a
n
d

R
o
y
,

M
.
,

1
9
7
8
,

C
a
n
.

G
e
o
t
e
c
h
.

J
.
,

1
5
:
5
4

6
5
.
)
Consolidation analysis 247
assumed to be completely rigid in both directions, while the vertical bound-
aries are only horizontally restrained. The top surface is free and permeable,
while the other boundaries are impermeable because the hydraulic condi-
tions are not clear although it has been reported that the sandy soil layer
exists below the depth of 13.8 m (Leroueil et al., 1978). The effect of the
boundary conditions have to be examined more in detail in the future. The
automatic time increment selection scheme by Karim and Oka (2010) was
used throughout the analysis.
The elastoviscoplastic fnite element method (FEM) model provides rea-
sonable predictions for the soil responses during and after the construction
of the embankment. Figure7.20 shows the time history of the vertical set-
tlements at node 1 (at the center line of embankment D and on the surface),
which depicts an excellent agreement between the FEM model and the test
results. However, during the initial period of time, the predicted values
by the FEM model are slightly smaller than the feld responses. Lateral
displacements at the toe of the embankment are shown in Figure 7.21.
Although just before the end of the construction, the predicted lateral dis-
placements are higher than the feld ones, the overall agreement is quite
acceptable. Figure7.22 shows the distribution of pore water pressure with
time at a depth of 2.25 m. Some time after the end of the construction, the
pore water pressure becomes stagnant and then slowly increases with time.
The distribution of pore water pressure also complies reasonably well with
the feld response.
0.663
0.612
0.561
0.510
0.459
0.408
0.357
0.306
0.255
0.204
0.153
0.102
0.051
0.000
0.0 2.5 5.0 7.5 10.0
Time, t (days)
12.5 15.0 17.5 20.0
E
m
b
a
n
k
m
e
n
t

L
o
a
d

(
k
g
f
/
c
m
2
)
Figure 7.19 (Continued)
248 Computational modeling of multiphase geomaterials
0.012
0.014
At toe of the embankment
0.008
0.01
FEM simulation
Field result
= 5.856
0.002
0.004
0.006
L
a
t
e
r
a
l

D
i
s
p
l
a
c
e
m
e
n
t
,

u
x
x

(
m
)

mai
=
m(0)
OCR
mean

maf
=
m(0)
0 20 40 60 80 100
0
Time (day)
Figure7.21 Observed and calculated lateral displacement at the toe of the slope.
0.2
0.18
0.16
0.14
0.12
0.1
0.08
0.06
0.04
0.02
0
0 20 40 60
Time (day)
At center of the embankment (Node 1)
FEM simulation
Field result
80 100
V
e
r
t
i
c
a
l

D
i
s
p
l
a
c
e
m
e
n
t
,

u
z
z

(
m
)

mai
=
m(o)
OCR
mean

maf
=
m(o)
= 5.856
Figure7.20 Observed and calculated displacement below the centerline of the embank-
ment (node 1).
Consolidation analysis 249
7.5 CONSOLIDATION ANALYSIS CONSIDERING
CONSTRUCTION OF THE EMBANKMENT
The simplest approximation for the simulation of the new material placing
in a fnite element analysis is by applying the increment of the weight of the
fll as the external load on the nodes of the embankment foundation inter-
face. However, for a more accurate analysis, the embankment construction
procedure should be implemented, in particular for the large deformation
of soft clay deposits, by which the stiffness and the consolidation of the
embankment can be considered in addition to the embankment loading.
For the construction of the embankment, the following procedure can be
used and is recommended (Potts and Zdravkovic 1999):
1. Divide the analysis into a set of increments. For a particular incre-
ment, the element to be constructed is inserted and given a constitu-
tive model appropriate to the material behavior during placing. This
often means that the material has low stiffness. In the current study,
the elastic behavior with low stiffness equal to 75% of the original
stiffness of the material is assumed for the material during placing.
40
45
At depth 2.25 m (Element 37)
20
25
30
35
FEM simulation
Field result
5
10
15
20
P
o
r
e

W
a
t
e
r

P
r
e
s
s
u
r
e
,

U
w

(
k
N
/
m
2
)
0 20 40 60 80 100
0
Time (day)
= 5.856

mai
=
m(0)
OCR
mean

maf
=
m(0)
Figure7.22 Observed and calculated porewater pressure levels of element 37 beneath
the center of the embankment at a depth of 2.25 m.
250 Computational modeling of multiphase geomaterials
2. Nodal forces due to the self-body forces of the constructed materials
are calculated and applied on the corresponding nodes.
3. The global stiffness matrix and all the other boundary conditions are
assembled for the increment. The analysis is performed to obtain the
incremental changes in displacements, strains, and stresses.
4. Before applying the next increment, the constitutive model for the ele-
ments just constructed is changed to represent the behavior of the fll
material once placed. The incremental displacements of any nodes that
are only connected to the constructed elements (i.e., not connected to
elements that were active at the previous increment) are zeroed.
5. Apply the next increment of the analysis.
The incremental procedure is schematically depicted in Figure7.23.
Layer 3
Layer 2
Layer 1
Figure7.23 The procedure for implementing the embankment construction.
Consolidation analysis 251
7.5.1 Numerical example
The linearized construction sequence of the embankment is schematically
shown in Figure 7.24. As shown, it was assumed that the embankment
would be constructed in four layers within 40 consecutive days. A con-
solidation analysis was performed until 1000 days after the end of the
construction.
The geometry and fnite element mesh of the problem are presented in
Figure 7.25. Table 7.8 gives the material parameters that are determined
considering the data of Osaka soft clay (Mirjalili et al. 2011) used in the
fnite element analysis. In order to discuss the effects on two factors, struc-
tural degradation and strain dependent shear modulus, analyses for three
cases have been performed (see Table7.9). In addition, M
m

is a function of
Lodes angle in the analysis. The calculated results of vertical settlements
time profle at the ground level are presented in Figure7.26a. The results
clearly indicate that the effect of both the structural degradation and the
strain dependency of the elastic shear modulus. The settlement profles at
the ground level at several times for case3 are illustrated in Figure7.26b.
Despite the increase in ground settlement beneath the embankment, during
and after construction, the surface heave around the toe decreases during
consolidation. Figure7.26c indicates the overall deformation pattern that
shows the distorted mesh beneath the embankment. In Figure7.27 we can
see the pore water pressuretime profle at several points. The secondary
generated pore water pressure around 40 day at points C and D, located in
the area with strain localization are higher than those in points A and B.
40 days 1000 days
Construction Time (days)
4
3.2
2.4
1.6
F
i
l
l

H
e
i
g
h
t

(
m
)
0.8
0
Figure7.24 Loading profle based on the construction stages.
252 Computational modeling of multiphase geomaterials
9
.
5

m
4
5
.
5

m
1
:
2
3
.
2
5

m
I
m
p
e
r
v
i
o
u
s

b
o
u
n
d
a
r
y
I
m
p
e
r
v
i
o
u
s

b
o
u
n
d
a
r
y
D
r
a
i
n
a
g
e

b
o
u
n
d
a
r
y
1 0 m 3 . 2 m
F
i
g
u
r
e

7
.
2
5

F
i
n
i
t
e

e
l
e
m
e
n
t

m
e
s
h

a
n
d

b
o
u
n
d
a
r
y

c
o
n
d
i
t
i
o
n
s

f
o
r

t
h
e

e
m
b
a
n
k
m
e
n
t

c
o
n
s
t
r
u
c
t
i
o
n
.
Consolidation analysis 253
T
a
b
l
e

7
.
8

M
a
t
e
r
i
a
l

p
a
r
a
m
e
t
e
r
s

f
o
r

t
h
e

e
m
b
a
n
k
m
e
n
t

a
n
d

t
h
e

g
r
o
u
n
d

l
a
y
e
r
P
a
r
a
m
e
t
e
r
s
k

(
m
/
s
)

t
(
k
N
/
m
3
)
e
0
G
0

(
k
P
a
)
O
C
R

M
*
m
c
m

C
1
,

C
2

(
1
/
s
)
n

E
m
b
a
n
k
m
e
n
t
1
.
0
0

1
0

5
1
9
.
8
0
.
8
4
3
0
0
G
r
o
u
n
d
3
.
8
5

1
0

1
0
1
6
.
0
1
.
6
5
3
9
3
0
1
.
1
0
0
.
5
9
3
0
.
0
2
7
1
.
1
8
2
8
.
2
1
.
8
5

1
0

1
1
0
.
6
7
1
0
,
1
5
(
1
)
0
,
1
0
*

E
m
b
a
n
k
m
e
n
t

i
s

m
o
d
e
l
e
d

a
s

a
n

e
l
a
s
t
i
c

m
a
t
e
r
i
a
l
.
254 Computational modeling of multiphase geomaterials
Table7.9 Specifcations for each case in the 2D numerical analysis
Structural degradation
Strain dependency
of shear modulus
Case 1 (No structural degradation: = 0, = 0)
Case 2 (Structural degradation: = 10, a = 0)
Case 3 (Structural degradation + Strain
dependency of G: = 15, a = 10)

Note: , Considered, , Not considered.
0
2
4
6
8
10
0 200 400 600
Time (day)
800 1000
10 days
20 days
30 days
40 days (E.O.C)
100 days
500 days
1000 days
S
e
t
t
l
e
m
e
n
t

(
c
m
)
S
e
t
t
l
e
m
e
n
t

(
c
m
)
Case 1
Case 2
Case 3
(a) Settlements-time prole at the ground level
0
2
4
10
0 5 10 15 20 25 30
Distance from Symmetry Axis (m)
(b) Settlement proles at the ground level at several times for case 3
(c) Overall deformation pattern after 1000 days of consolidation (deformations have been enlarged 10 times)
35 40 45 50 55
8
6
Case 3
Figure7.26 Deformation of the embankment. (a) Settlementstime profle at the ground
level. (b) Settlement profles at the ground level at several times for Case 3. (c)
Overall deformation pattern after 1000 days of consolidation (deformations
have been enlarged 10 times).
Consolidation analysis 255
REFERENCES
Aboshi, H. 1973. An experimental investigation on the similitude of consolidation
of clay including secondary creep settlement, Proc. 8th ICSMFE, Moscow,
4(3):88.
Aboshi, H., and Matsuda, H. 1981.Secondary compression of clays and its effect on
settlement analysis, Tsuchi-to-Kiso JSSMFE, 29(3):1924 (In Japanese).
Aboshi, H. 1995. Creep records of long-term measurement of consolidation settle-
ment and their predictions, Compression and consolidation of clayey soils, H.
Yoshikuni and O. Kusakabe eds., Rotterdam, Balkema, 2:847872.
Aboshi, H. 2004. Long-term effect of secondary consolidation on consolidation set-
tlement of marine clays, Advances in geotechnical engineering, The skempton
conference, ICE, Thomas Telford, 1:345355.
Adachi, T., and Oka, F. 1982. Constitutive equations for normally consolidated clay
based on elasto-viscoplasticity, Soils and Foundations, 22(4):5770.
Adachi, T., Oka, F., and Tange, Y. 1982. Finite element analysis of two-dimen-
sional consolidation using an elasto-viscoplastic constitutive equation,
40
35
30
25
20
E
x
c
e
s
s

P
o
r
e

W
a
t
e
r

P
r
e
s
s
u
r
e

(
k
P
a
)
15
10
5
0
0.01 0.1 1 10
Time (days)
100 1000 10000
Case 3
= 15, = 10
Point A
Point B
Point C
Point D
Y (m)
X (m)
D (X = 2.1; Y = 9.8)
C (X = 2.1; Y = 9.4)
B (X = 1.05; Y = 8.95)
A (X = 1.05; Y = 7.9)
Figure7.27 Excess pore water pressuretime profle.
256 Computational modeling of multiphase geomaterials
Proceedings of 4th International Conference on Numerical Methods in
Geomechanics, Z. Zisenstein, ed., Canada Edmonton, Balkema Rotterdam,
2:287296.
Akai, K., and Adachi, T. 1965. Study on the one-dimensional consolidation and the
shear strength characteristics of fully saturated clay, in terms of effective stress,
Proc. 6th ICSMFE, Montreal, 1:146150.
Akai, K., and Tamura, T. 1979. Study of two-dimensional consolidation accompanied
by an elasto-plastic constitutive equation, Proc. JSCE, 269:95104 (In Japanese).
Barden, L. 1969. Time dependent deformation of normally consolidated clays and
peats, Proc. ASCE, 95(SMI):131.
Berre, T., and Iversen, K. 1972. Oedometer tests with different specimen heights on a
clay exhibiting large secondary compression, Gotechnique, 22(1):5370.
Bishop, A.W., and Lovenbury, H.T. 1969. Creep characteristics of two undisturbed
clays, Proc. 7th Int. Conf. Soil Mech. Found. Eng., Mexico City, 1:2937.
Christian, J.T., and Boehmer, J.W. 1970. Plane strain consolidation by fnite elements,
Proc. ASCE, 96(SM4):14351457.
Garlanger, J.E. 1972. The consolidation of soils exhibiting creep under constant
effective stress, Gotechnique, 22(1):7178.
Kabbaj, M., Oka, F., Leroueil, S., and Tavenas, F. 1986. Consolidation of natural
clays and laboratory testing, Consolidation of soils: testing and evaluation,
ASTM STP 892, R.N. Yong and F.C. Townsend, eds., 378404.
Kabbaj, M., Tavenas, F., and Leroueil, S. 1988. In situ and laboratory stress-strain
relationships, Gotechnique, 38(1):83100.
Karim, Md Rezaul. 2006. Simulation of long term consolidation behavior of soft sensitive
clay using an elasto-viscoplastic constitutive model, PhD thesis, Kyoto University.
Karim, M. R., and Oka, F. 2010. An automatic time increment selection scheme
for simulation of elasto-viscoplastic consolidation of clayey soils, Geomech.
Geoeng., 5(3):153177.
Kimoto, S. 2002. Constitutive models for geomaterials considering structural
changes and anisotropy, PhD thesis, Kyoto University, Japan.
Kimoto, S., Oka, F., and Higo, Y. 2004. Strain localization analysis of elasto-
viscoplastic soil considering structural degradation, Comput. Meth. Appl.
Mech. Eng., 193:28452866.
Kimoto, S. and Oka, F. 2005. An elasto-viscoplastic model for clay considering
destructuralization and consolidation analysis of unstable behavior, Soils
and Foundations, 45(2):2942.
Ladd, C.C. 1973. Settlement analysis for cohesive soils, Research Report R71-2(272),
Department of Civil Engineering, Massachusetts Institute of Technology, 115.
Leroueil, S., and Hight, D.W. 2003. Behavior and properties of natural soils and soft
rocks, In Characterisation and Engineering Properties of Natural Soils, T.S.
Tan, K.K. Phoon, D.W. Hight, and S. Leroueil, eds., Balkema, 1:29254.
Leroueil, S. 1988. Recent developments in consolidation of natural clays, Canadian
Geotechnical Journal, 25(1):85107.
Leroueil, S. 1995. Could it be that clays have no unique way of behaving during
consolidation? Compression and consolidation of clayey soils, H. Yoshikuni
and O. Kusakabe eds., Rotterdam, Balkema, 2:10491048.
Leroueil, S., Lebihan, J.P., and Tavenas, F. 1980. An approach for determination of the
preconsolidation pressure in sensitive clays, Can. Geotech. J., 17(3):446453.
Consolidation analysis 257
Leroueil, S., Tavenas, F., Trak, B., Rochelle, P. L., and Roy, M. 1978. Construction
pore water pressures in clay foundations under embankments. Part I: the Saint-
Alban test flls, Can. Geotech. J., 15:5465.
Matsuda, H. and Aboshi, H. 1983. A study on preconsolidation by separate-type
consolidometer, Research report of the Department of Engineering, Yamaguchi
University, 33(2):3948.
Mesri, G., and Choi, Y.K. 1979. Excess pore water pressure during consolidation,
Proc. 6th Asian Reg. Conf. on Soils Mech. Found. Eng., 1:151154.
Mesri, G., and Choi, Y.K. 1980. (Discussion) Excess pore water pressure and pre-
consolidation effect developed in normally consolidated clay of same age, Soils
and Foundations, 20(4):143148.
Mesri, G., and Rokhsar, A. 1974. Theory of consolidation for days, J. Geotech. Eng.
Div., ASCE, 100(GT8):889904.
Mesri, G. and Choi, Y.K. 1985. The uniqueness of the end-of-primary (EOP) void
ratio-effective stress relationship, Proc. 11th ICSMFE, San Francisco, ISSMFE,
2:587590.
Mesri, G., Shahien, M. and Feng, T.W. 1995. Compressibility parameters during
primary consolidation, Compression and consolidation of clayey soils, H.
Yoshikuni and O. Kusakabe eds., Rotterdam, Balkema, 2:10211037.
Mirjalili, M. 2010. Numerical analysis of a large scale levee on soft soil deposits
using two-phase fnite deformation theory, PhD thesis, Kyoto University.
Mirjalili, M., Kimoto, S., Oka, F., and Higo, Y. 2011. Elasto-viscoplastic modeling of
Osaka soft clay considering destructuration and its effect on the consolidation
analysis of an embankment, Geomech. Geoeng., Int. J., 6(2), 6989.
Mirjalili, M., Kimoto, S., Oka, F., and Hattori, T. In press. Long-term consolidation
analysis of a large-scale embankment construction soft clay deposits using an
elasto-viscoplastic model, Soils and Foundations.
Mitchell, J.K. 1986. Practical problems from surprising soil behavior, 20th Terzaghi
Lecture, ASCE, J. Geotech. Engrg., 112(3):259289.
Moore, P.J., and Spencer, G.K. 1972. Lateral pressures from soft clay, Proc. ASCE,
98 (SM11):12251244.
Oka, F. 1981. Prediction of time dependent behaviour of clay, Proc. 10th ICSMFE,
Stockholm, 1:215218.
Oka, F. 2000. Elasto-viscoplastic constitutive models of geomaterials, Morikita
Shuppan, (in Japanese).
Oka, F. 2005.Computational modeling of large deformations and the failure of geo-
materials, Proc. 16th ICSMGE, Osaka, 1:95122.
Oka, F., Adachi, T., and Okano, Y. 1986. Two-dimensional consolidation analysis
using an elasto-viscoplastic constitutive equation, Int. J. Numer. Anal. Meth.
Geomech, 10:116.
Oka, F., Tavenas, F., and Leroueil, S. 1991. An elasto-viscoplastic FEM analysis of sen-
sitive clay foundation beneath embankment, Proceedings of 7th International
Conference on Computer Method and Advances in Geomechanics, G. Beer,
J.R. Booker, and J.P. Carter, eds., Cairns, Balkema, 2:10231028.
Perzyna, P. 1963. The constitutive equations for work-hardening and rate sensitive
plastic materials, Proc. Vibrational Probl., Warsaw, 4(3):281290.
Potts, D.M., and Zdravkovic, L. 1999. Finite Element Analysis in Geotechnical
Engineering Theory, Thomas Telford.
258 Computational modeling of multiphase geomaterials
Roscoe, K.H., Schofeld, A.N., and Thurairlljah, A. 1963. Yielding of clays in states
wetter than critical, Gotechnique, 13(3):211240.
Sekiguchi, H., and Ohta, H. 1977. Induced anisotropy and time dependency in clays,
Proc. Speciality Session 9, 9th ICSMFE, Tokyo, 2292381.
Sekiguchi, H., and Shibata, T. 1979. Undrained behavior of soft clay under embank-
ment loading, Proc. 3rd International Conference on Numerical Methods in
Geomechanics, Achen, W. Wittke, ed., Balkema, 2:717724.
Sekiguchi, H., and Toriihara, M. 1976.Theory of one-dimensional consolidation of
clay with consideration of their rheological properties, Soils and Foundations,
16(1):2744.
Suklje, L. 1969. Rheological Aspects of Soil Mechanics, New York: Wiley-Interscience.
Tang, Y.X., and Imai, G. 1995. A constitutive relation with creep and its application
to numerical analysis of one-dimensional consolidation, In Compression and
Consolidation of Clayey Soils, H. Yoshikuni and O. Kusakabe, eds., Balkema,
Rotterdam, 465472.
Tavenas, F., Chapeau, C., Rochelle, P. L., and Roy, M. 1974. Immediate settlements of
three test embankments on Champlain clay, Can. Geotech. J., 11(4):109141.
Tavenas, F., and Leroueil, S. 1980. The behavior of embankments on clay founda-
tions, Can. Geotech. J., 17(2):236260.
Zienkiewicz, O.C. 1977. The Finite Element Method, 3rd ed., New York:
McGraw-Hill, 450499.
259
Chapter 8
Strain localization
It is well known that the strain localization of geomaterials causes such
important problems as slope failure. In slope failure, deformation occurs
in a narrow zone. In other words, strain localization is closely related to
the onset of failure (Scott 1987). In the present chapter, strain localization
problems are reviewed, and the instability and the numerical simulation of
strain localization will be presented.
8.1 STRAIN LOCALIZATION PROBLEMS
IN GEOMECHANICS
In order to analyze failure, we have to deal with strain localization near
and/or after failure. The development of a failure surface is a classical issue
in soil mechanics. Coulomb (1773) considered a failure surface in order to
determine the collapse load in his famous work. Sokolovsky (1942) ana-
lyzed a slip plane as a stress characteristic at the limit equilibrium. As has
been pointed out by Taylor (1948), the failure phenomenon of geomaterials
is progressive. Hence, strain localization is a precursor to the development
of a failure surface and is a very important subject to investigate. Over
the last three decades, the problems of strain localization in geomaterials,
that is, the formation of shear band of soil and rock, have been extensively
studied within the context of experimental, theoretical, and numerical
approaches. Many researchers (Hill 1962; Rice 1975, 1976; Rudnicki and
Rice 1975; Vardoulakis 1980; Vermeer 1982; Mhlhaus and Vardoulakis
1987; Wang et al. 1990; Oka et al. 1994; Muir Wood 2002; Borja 2004;
Gudehus and Nbel 2004) have been working in this area from both exper-
imental and analytical points of view. Rice (1976) and Rudnicki and Rice
(1975) pointed out that the nature of this problem can be solved under the
general framework of bifurcation problems and that localization problems
should be studied within a wider framework of mechanics, including the
rapid degradation of the material strength.
260 Computational modeling of multiphase geomaterials
8.1.1 Angle of shear band
The classical MohrCoulomb law gives the angle of the shear band (failure
surface) as

M
= +
4 2
(8.1)
where is the internal friction angle.
On the other hand, as is well known, Roscoe (1970) reported in his
Rankine lecture that the rupture (failure) surface does not coincide with the
stress characteristics, but with the zero-extension lines. This means that the
angle of a shear band to the major principle stress plane (see Figure8.1)
is given by

R
= +
4 2
(8.2)
where is the dilatancy angle.
In addition, Roscoe (1970) pointed out that the thickness of a shear band
is approximately ten times the average grain diameter.
For the angle of a shear band, Arthur et al. (1977) proposed a shear band
angle based on experiments as

A
= + +
4
1
4
( ) (8.3)
x
2
x
1

3
Figure8.1 Angle of shear band (
1 3
> ).
Strain localization 261
since < and
R A M
< < . The angles experimentally obtained by
Arthur et al. (1977) were supported by theoretical works (Vardoulakis
1980; Vermeer 1982). Vardoulakis (1980) reported that the experimen-
tal results coincide with Equation (8.3). However, the experimental
results for sand, obtained by Desrues (1990) and Lade (2003), are close
to
M
. Vardoulakis (1977) illustrated that the shear band angle depends
on the particle diameter; in other words, the larger the soil particle size,
the smaller the shear band angle. Tatsuoka et al. (1990) showed that the
shear band angle depends on the b-value (see Equation 4.35) and the
anisotropy. These indicate the other effects on the angle of the shear
band. For clay, Oka et al. (2002) and Higo et al. (2006) numerically
indicated that the angle of a shear band depends on the strain rates and
the permeability, as shown in Sections 8.8 and 8.9.
8.2 LOCALIZATION ANALYSIS
It has been theoretically found that the onset conditions for strain localiza-
tion can be captured by a bifurcation analysis (Hill 1962; Rice 1976). Rice
(1976) showed that shear band formation is a problem because a shear band
is the result of bifurcation from a homogeneous deformation. The onset
conditions for localization are given by linear and nonlinear incremental
constitutive models. For static problems, the conditions involve a loss of
the elliptic onset condition, the thickness, the angle, and the distribution.
Postlocalization behavior and the loss of ellipticity correspond to the loss
of uniqueness, namely, the instability discussed within the framework of
bifurcation and material instability.
Within the framework of a linear incremental model, the onset condi-
tions of bifurcation are derived as the condition of equilibrium, which
requires the time rate of change in stress traction

ij j
n at the shear band to
be zero, namely,

[ ]

ij j
n = 0 (8.4)
where n
j
is the unit normal to the shear band. The brackets [ ] of a quantity
denote the difference in values of that quantity across the shear band.
The velocity is continuous, but the gradient of velocity is discontinuous.

[ ] L n
ij i j
=
(8.5)
where L v
ij i j

,
is the velocity gradient.
262 Computational modeling of multiphase geomaterials
The incremental constitutive model is given by

ij ijkl kl
D L =
(8.6)
From Equations (8.4), (8.5), and (8.6),

H H D n n
ik k ik ijkl j l
= = 0, (8.7)
where D
ijkl
is the stiffness tensor.
The onset conditions for nonzero
k
are

det[ ] H
ik
= 0
(8.8)
From these conditions, we can obtain shear band angle in a two-
dimensional case, in which n
1
= cos and n
2
= sin in Equation (8.7) and
is the angle between the unit normal to the shear band and the major
principle stress plane.
Equation (8.8) can be written as

det[ ] H a n a n n a n n a n n a n
ik
= + + + + =
0 1
4
1 1
3
2 2 1
2
2
2
3 1 2
3
4 2
4
0 (8.9)
where
a L L L L a L L L L
0 1111 1212 1112 1211 1 1111 1222 1111 22
= = + ,
112 1112 2211 1122 1211
L L L L ,
a L L L L L L L L
2 1111 2222 1112 1222 1211 2212 1122 1212
= + + LL L L L
1122 2211 1212 2211

a L L L L L L L L
3 1112 2222 1211 2222 1122 2212 1222 2211
= +
,
a L L L L
4 1212 2222 2212 1222
=
If we set n n
1 2
= = cos , sin , we have to consider the symmetry with
respect to .

g a a a ( ) tan tan = + + =
4
4
2
2
0
0 (8.10)
For a two-dimensional infnitesimal strain model,



11 11 11 12 22 22 21 11 22 22 12
= + = + = D D D D , , 22
12
4 11 2 11 22 12 21 12 21 0
G
a GD a D D D D GD GD a G

,
, , = = = DD
22
(8.11)
Strain localization 263
The character of Equation (8.9) is classifed as elliptic, parabolic, and
hyperbolic regions depending on the roots of Equation (8.10); the ellip-
tic region for the four imaginary or four complex roots, the parabolic
region for the two real and two imaginary roots, and the hyperbolic
region for four real roots (Hill and Hutchinson 1975). It is worth not-
ing that bifurcation never occurs in the elliptic region. On the contrary,
bifurcation, such as shear bands, could develop in the parabolic and
hyperbolic regions.
It is seen that Rices bifurcation condition is the same as the singular-
ity of the acoustic tensor (Mandel 1964). This means that the occur-
rence of a shear band coincides with the accelerated wave trapped in
the narrow zone. For the propagation of an acceleration wave, that is,
the propagation of the discontinuity of the acceleration, it is necessary
for the eigenvalues to be positive and real. The acceleration is discon-
tinuous across the wave front for the acceleration wave, although the
displacement and the velocity are continuous. When the eigenvalues are
negative and zero, discontinuity cannot propagate and it is referred to
as stationary discontinuity, namely, shear bands or complex (flutter)
instability.
The zero determinant of the acoustic tensor by Equation (8.7) cor-
responds to a change in the type of partial differential governing equa-
tions from elliptic to hyperbolic (Hill and Hutchinson 1975). Bigoni
and Hueckel (1991) found that the loss of strong ellipticity coincides
with the nullity of the determinant of the symmetric part of the acoustic
tensor.
Strain localization has been well recognized as the development of
shear bands. On the other hand, another type of strain localization has
also been observed, namely, compaction bands (Mollema and Antonellini
1996; Olsson 1999). Compaction bands were experimentally found for
sandstone as a tabular zone that exhibits closure, but no shear offset
(Mollema and Antonellini 1996). The compaction bands comprise a kind
of volumetric strain localization due to pore collapse or particle crushing.
The compaction bands of porous sandstone exhibit a sharp reduction in
permeability that causes diffculties in oil production. Olsson (1999) has
shown that compaction bands can be described by the strain localiza-
tion theory of Rudnicki and Rice (1975) as well as by shear localization
problems. The re-examination of Rudnicki and Rices theory comes from
the correction of the Rudnick and Rice theory by Perrin and Leblond
(1993). They showed that the possible range for the sum of the dilatancy
factor and the coeffcient of internal friction is wider than that shown
in Rudnicki and Rices paper; in other words, + 3 2 / should be
+ 3 3 .
264 Computational modeling of multiphase geomaterials
Rudnicki and Olsson (1998) re-examined the fault angles predicted by
the shear localization theory and obtained the following convenient form:

= +
4
1
2
arcsin ,

=
+ +

( / )( )( ) ( ) 2 3 1 1 2
4 3
2
N
N
(8.12)

N
II
/ =
(8.13)
where = s s
ij ij
2 , s
ij
is the deviatoric stress tensor, is Poissons ratio, is
the gradient of the plastic potential surface, '
II
is the intermediate princi-
pal deviatoric stress, and is the gradient of the yield surface.
Using Equation (8.12), it has been realized that a smaller angle for the
compaction bands can be described. Issen and Rudnicki (2000) obtained
the conditions for compaction bands in porous rock.
In order to observe compaction bands for soil, Castellanza and Nova
(2003) reported a compaction band forming in the cemented granular
soil during odometer tests. They observed a stressstrain relation with a
plateau that occurred after the peak stress had been reached, which is a
typical stressstrain relationship for compaction bands. Oka and Kimoto
(Kimoto and Oka 2004, 2005) pointed out that the large compression of
structured clay is due to the development of compaction bands, and they
numerically analyzed the unstable consolidation of clay with degradation.
This compaction band phenomenon for a soft rock has been successfully
simulated by an elastoviscoplastic model with volumetric strain softening
by Okaetal. (2011).
8.3 INSTABILITY OF GEOMATERIALS
As mentioned earlier, strain localization can be described as bifurcation
conditions by Equation (8.8), namely, a zero determinant of the acous-
tic tensor (Rice 1976; the loss of the positive defniteness of constitutive
matrix D). The loci on the plane for which the determinant of the acoustic
tensor is zero, namely, the loci of the localization conditions, never inter-
sect the stress axes. This means that shear banding cannot occur in con-
ventional triaxial compression tests before the failure state. On the other
hand, shear bands may easily develop in plane strain tests. The stress state
reaches the surface given by the zero determinant of the acoustic tensor
before it reaches the limit state. The limit state is characterized by the con-
ditions under which the determinant of the constitutive matrix equals zero
Strain localization 265
(Tokuoka 1971) and it is called the failure condition. The failure condition
is the limit of allowable stress space and is given by
det[D] = 0 (8.14)
where [D] is the constitutive stiffness matrix and { } [ ]{ }

= D . Since { } { }

= 0
at the failure state, we have [ ]{ } { } D

= 0 .
The basic defnition of the stability is the stability in the sense of
Lyapunov, as shown in Chapter 3, where the positive second-order work is
a suffcient condition for the stability. The local form of the second-order
work can be written in matrix form as

W D
ij ij
T
{ } [ ]{ } 0; 0
2
= = >
(8.15)
From the fundamental theorem in linear algebra, the aforementioned
condition is called positive defniteness. If the positive defniteness holds,
det[D] > 0 if and only if for the real symmetric [D] is positive defnite. Then,
all minor determinants are positive.
On the contrary, if the [D] is positive semidefnite, that is, det[D] = 0,
many nonzero solutions for strain rate { }

can exist.
[D] is positive defnite if and only if its symmetric part is positive def-
nite, since for the skew-symmetric part of [D], Equation (8.15) becomes
zero. Nova (1994) pointed out that based on the theorem by Ostrowski and
Taussky (1951):

det[ ] det[ ] D D
s
> 0
(8.16)
Nova (1994) studied the other bifurcation criteria and found that the
conditions for the loss of controllability are obtained as a zero determinant
of the symmetric part of stiffness tensor [ ] D
s
.
In general, for the nonassociated plasticity model, det[ ] D
s
= 0 even when
det[D]0.
The constitutive matrix [D] is decomposed into the symmetric part and
the skew-symmetric part, namely,

[ ] [ ] [ ] D D D
s a
= + (8.17)
And if det[ ] D
s
= 0 and [ ]{ } { } D
s

= 0 for a particular strain rate { }

, which
is proportional to the eigenvector of [ ] D
s
, then

D D D
D
s a
a
{ } [ ]{ } [ ]{ } [ ]{ }
[ ]{ }

= = +
=
(8.18)
266 Computational modeling of multiphase geomaterials
In this case, the second-order work becomes zero as

2
0
W D
D
T T
T
a
= =
= =
{ } { } { } [ ]{ }
{ } [ ]{ }




(8.19)
This indicates the loss of Hills stability condition (Equation 3.57). Nova
(1994) indicated and discussed that the loss of controllability may occur
at the same time as the loss of Hills stability. The loss of controllability is
explained in the following. The same discussion holds for the expression by
compliance matrix [ ]; { } [ ]{ '} C C

= .
In the experiments on soil, we can choose the variables for controlling
the system; for example, we control the axial strain rates and maintain the
lateral stress constant in the strain control triaxial tests. When the condi-
tions are satisfed, we cannot continue the experiments. For instance, we
cannot execute triaxial tests for strain hardeningsoftening material after
the peak stress by the axial stress control. Nova found that the loss of con-
trollability is equal to the violation of the positive second-order work. This
indicates that if the stress states satisfy the zero determinant of the sym-
metric part of the material stiffness matrix, det[ ] D
s
= 0, instability (bifur-
cation) may occur. Lades instability condition (1992) is equivalent to the
conditions under which the volumetric compliance is zero. This state comes
before the failure state in the triaxial tests.
The loss of controllability is explained using compliance matrix [C]
based on an example following Imposimato and Nova (1998).
Under axisymmetric triaxial conditions, we can write the constitutive
model as

C C
C C
p
q
v
d
pp pq
qp qq

= (8.20)
where

v
is the volumetric strain rate, and

d
is the shear strain rate,
namely,


d a r
= 2 3 ( ) / ;

a
is the axial strain rate and

r
is the radial
strain rate.

p' is the mean effective stress rate, and q is the deviator stress
rate;
q =
11 33
.
This expression indicates that the control parameters are the shear strain
and the volumetric strain. For

v
and

d
, see Chapter 4. In addition, we
assume det[ ] C C C C C
pp qq pq qp
= > 0.
Consider the mixed control case in which we change the control param-
eters from two components of strain rates to the other set of control param-
eters, that is, shear strain rate

d
and deviator stress rate
q.
Strain localization 267
In this case, we have

v
qq
pp pq pq qp pq
qp
q C
C C C C C
C
=

1
1

p
d
'

(8.21)
When the determent of the matrix is zero in the preceding equation,

C C
pp pq
= 0 (8.22)
This condition holds when

C
pp
= 0
(8.23)
since C
pq
0 for the existence of dilatancy. It is worth noting that the deter-
minant of Equation (8.21) is zero, but that of Equation (8.20) is not zero.
Since

v
= 0 under undrained conditions, the constitutive model can be
written as

C
C C C C C
C
p
q
qq
pp pq pq qp pq
qp d
1
1
' 0

= (8.24)
When the stress rate is at the maximum q, q = 0 under load-control
conditions,

det
C C C C C
C
pp pq pq qp pq
qp

=
1
0 (8.25)
From Equation (8.25), we have C
pp
= 0 (Equation 8.23), and in this case
many solutions may exist for Equation (8.24). However, the solution indi-
cates the ratio of the pore water pressure rate and the deviatoric strain rate
such that

p
C
d qp

=
1
(8.26)
for C
pp
= 0.
268 Computational modeling of multiphase geomaterials
It is worth noting that only the ratio can be determined for the zero
eigenvalue problem, but not for each value. Hence, at the peak value for the
deviator stress, the pore water pressure and the deviatoric strain rate can
be arbitrary for the constant total isotropic pressure, p p u
w
0

= + = , that
is, u p arbitrary

= = and C u
d qp w
= arbitrary = . These results indicate
the spontaneous pore water pressure development during the undrained
test with stress control test (Imposimato and Nova 1998), schematically
shown in Figure8.2. Since the compliance matrix in Equation (8.20) is pos-
itive defniteness, Lades instability condition (1992) corresponds, C
pp
= 0,
under undrained conditions, which manifests at the peak of the deviator
stress shown in Figure8.3.
p
A
I
Time
(a) q Constant test after drained test (b) Spontaneous pore water pressure
development after instability line
P
o
r
e

P
r
e
s
s
u
r
e
A
Failure line
q
I
Instability line
Figure8.2 Schematic fgure for the deviator stress constant undrained test. (a) q constant
test after drained test. (b) Spontaneous pore water pressure development
after instability line.
p
Efective stress path
Peak of the deviator stress q
Instability line
Failure line
q
Figure8.3 Effective stress path during the undrained triaxial test for loose sand.
Strain localization 269
Since the nonassociative fow rule of plasticity is not symmetric, a variety
of unstable conditions may arise. They are the strain localization condition
(the loss of the determinant of the acoustic tensor; Rice 1976), the zero
second-order work condition (Hill 1957, 1958), and the loss of controllabil-
ity condition (Nova 1994, 2009). The loss of positiveness of the secondary
work condition and the loss of the controllability condition may lead to
diffuse (homogeneous) bifurcation or instability that manifests as a barrel-
ing or a bulging type of phenomenon (Nova 1989; Darve 2001; Daouadji
et al. 2010).
The surfaces that defne these types of instability are inside the failure
surface (Imposimato and Nova 2001). Benallal and Comi (2003) conducted
a study using a perturbation approach and found that the failure mode of
nonassociated material exhibits either a diffuse mode (long wavelength) or
a localization mode (short wavelength).
Hill (1958) derived a suffcient condition for the uniqueness of the solu-
tion to boundary value problems using the virtual work theorem as



ij
V
ij
dv

> 0 (8.27)
where


ij ij
, are the differences between the stress rates and the strain
rates for two different solutions and zero at the appropriate boundary of
the body in motion.
Equation (8.24) indicates that if the material is stable, that is, the
second-order work is positive, and the uniqueness of the solution holds.
This means that no localization solution exists. However, it is worth
noting that if



ij
V
ij
dv

= 0 (8.28)
bifurcation, multiple solutions may occur, but bifurcation solutions include
unstable and stable solutions.
For the incrementally strong nonlinear law, the onset conditions for a
shear band have been obtained by Chambon et al. (2000) and Kolymbas
and Rombach (1989). The incrementally nonlinear model has been pro-
posed as a hypoplastic model by Kolymbas and Rombach (1989) and
Chambon et al. (2000).
For the hypoplastic models,

ij ijkl kl ij kl
M F b D = + (8.29)
270 Computational modeling of multiphase geomaterials
where M
ijkl
is the constitutive fourth order tensor, b
ij
is the constitutive
second-order tensor,

F
kl
is the rate of the deformation gradient, D
ij
is the
stretching tensor, and || || denotes the Euclidian norm.
For the hypoplastic models, several criteria for bifurcation have been
obtained based on the invertibility of M n n
ijkl k l
. Namely, three types of
bifurcation criteria were given as (1) the norm criteria, (2) the determinant
criteria, and (3) the tangential criteria. The norm criteria are frst met, while
the determinant criteria are never met (Chambon et al. 2000; Tamaganini
et al. 2001).
8.4 NONCOAXIALITY
It is known that the direction of the principle strain rates does not need
to coincide with the direction of the stress rates for anisotropic materi-
als. And even for isotropic materials, the direction of the principle strain
rates does not need to coincide with the direction of the stress rates; this
is called noncoaxiality. The yield vertex that comes from the microme-
chanical consideration leads to noncoaxiality. Rudnicki and Rice (1975)
introduced this vertex effect into the model by a noncoaxial term called
the MandelSpencer modulus term. The noncoaxial term was naturally
derived into the planer model from the double slip theory by de Josselin de
Jong (1971). Spencer (1964) proposed a double shearing model by assum-
ing a micromechanical slip along the stress characteristic lines obtained
from the equilibrium equations and MohrCoulomb failure criteria
(Sokolovsky 1942). The double shearing model was generalized to be a
model with dilatancy by Mehrabadi and Cowin (1978). The model has
resulted in the noncoaxial term that is is workless, for example, purely
deviatoric. Anand (1983) generalized it into the isotropic hardening model.
This noncoaxial model has accounted for the vertex of the plastic potential
surfaces (Rudnicki and Rice 1975). Yatomi et al. (1989) used the nonco-
axial term in a Cam-clay model and showed that the noncoaxial term
easily led to instability as shear banding or as accessibility to bifurcation
from an elliptic to a hyperbolic type of governing equation. Papamichos
and Vardoulakis (1995) developed a different type of noncoaxial theory by
introducing a kinematical concept. Khojastehpour and Hashiguchi (2004)
developed a tangential plasticity theory accounting for noncoaxiality. All
of the models are rate-independent plasticity theories. Oka (1993b) devel-
oped a rate-dependent noncoaxial viscoplastic model using a transformed
stress tensor with the current stress-induced quasi-anisotropy shown in
the next paragraph. The other source of noncoaxiality is the anisotropy of
the materials (Nemat-Nasser 1983).
Strain localization 271
8.5 CURRENT STRESS-DEPENDENT
CHARACTERISTICS AND ANISOTROPY
As mentioned, the noncoaxial term, called the MandelSpencer modulus
term, is derived from the characteristic plane. There are several constitutive
models based on the characteristic plane. Matsuoka and Nakai (1974) pro-
posed a model in which the failure criteria depend on the interim principal
stress, although Mohr-Coulombs criteria depend on the maximum and the
minimum principal stresses. Oka (1993a, 1993b) proposed a transformed
stress tensor to describe the current stress-dependent behavior of soil, by
which Matsuoka and Nakais failure criteria (1985), shown in Chapter 4,
can be derived. In addition, the deviatoric fow rule can be reproduced in
such a way that the direction of the strain rate is different from the direc-
tion of the stress path. The transformed stress tensor


ij
A
ij ij
F A = ( , ) (8.30)
is derived from Wangs representation theorem as a function of the stress
tensor and the structural tensor (Spencer 1987). The structural tensor was
frst adopted by Boehler and Sawczuk (1977) as

A m m
ij i j
= (8.31)
where m
i
is a component of unit normal vector to the characteristic plane.
Considering Matsuoka and Nakais well-known failure criteria (1974),
m
i
can be taken as the unit normal to the spatial mobilized plane. If we
take the sedimentation plane as the characteristic plane, we can construct a
transversely isotropic theory (Oka 1993a, 1993b; Oka et al. 2002b). Nakais
t
ij
theory can be classifed as being similar to the transformed stress tensor
(Nakai and Mihara 1984; Chowdhury and Nakai 2001), although they
originally proposed the theory based on physical considerations.
8.6 REGULARIZATION OF ILL-POSEDNESS
The onset conditions for a shear band can be obtained analytically or
numerically, but it is also necessary to predict the postlocalization behav-
ior. After a shear band has occurred, the boundary value problem becomes
ill-posed. Hence, ellipticity (well-posedness) must be retrieved. Well-
posedness is the uniqueness of a solution and the continuous dependence
on the boundary conditions. In the fnite element analysis, this instability
272 Computational modeling of multiphase geomaterials
leads to a strong mesh-size dependency and, in turn, the thickness of the
shear band becomes zero, although the thickness of shear bands is fnite in
the experiments. There are several methods for the regularization of the
ill-posedness of governing equations.
8.6.1 Nonlocal formulation of constitutive models
Nonlocal models, such as the higher-order strain gradient-dependent model
(Aifantis 1984; Mhlhaus and Aifantis 1991; Vardoulakis and Aifantis
1991), and micropolar models, such as the Cosserat model (Mhlhaus
1986; Teichman and Wu 1993; Swada et al. 1999, 2001; Bauer and Huang
2001), have been used for the postlocalization analysis. An integral type of
nonlocal model has been developed to account for the nonlocal effect. The
feature of nonlocal models is that they contain the material length scale.
Mhlhaus and Oka (1996) and Oka et al. (1998) clarifed that the higher
strain gradient term naturally comes from the fact that the material is dis-
crete and has an inherent length scale.
We will discuss the performance of the second-order strain gradient
dependent model using the one-dimensional viscoelastic model. Let us
consider a nonlocal model, for example, a viscoelastic model with a second
order strain gradient as

= + a b c
t xx
(8.32)
where is the stress, is the strain,
t
is the strain rate,
xx
is the second-order
gradient with respect to the space coordinate, and a,b,c are material parameters.
Incorporating Equation (8.32) into the equilibrium equation,
x
= 0, we
have

a b c
x tx xxx
+ = 0 (8.33)
Then, we consider the fuctuation in strain as
= Aexp(ikx + t)
(8.34)
in which A is a constant, k is the wave number, and is the growth rate
of fuctuation. Introducing Equation (8.34) into Equation (8.33), we have

= ( )/ a ck b
2
(8.35)
When a is positive, that is, the strain-hardening case, < 0 for the
usual viscous effect of b > 0 even if c = 0. On the other hand, when a is
negative, that is, the strain-softening case, can be negative if and only
Strain localization 273
if c>0. This means that the material is stable and the boundary value
problem becomes well-posed even if the material exhibits strain softening.
Otherwise, instability occurs in a fnite time.
8.6.2 Fluidsolid two-phase formulation
For water-saturated soil, the Biot two-phase mixture theory can be applied
to an elastoplastic model for the soil skeleton. In this type of formula-
tion, Zhang and Schrefer (2001) found that the critical hardening moduli
become smaller than those for a single material; the permeability is large.
This indicates that the localization is delayed for the case of water-
saturated soil. They also pointed out that there exists a domain of perme-
ability where stability is lost, but the hyperbolicity remains in the dynamic
analysis of water-saturated soil.
Vardoulakis (1996a, 1996b) reported that ill-posedness was encountered
with the boundary value problem even when the two-phase formulation
was adopted. However, from the numerical studies, it was found that the
strain localization is depressed in the case of low permeability (Loret and
Prvost 1991; Oka et al. 1995; Asaoka et al. 1997).
8.6.3 Viscoplastic regularization
An elastoviscoplastic formulation can retrieve the ill-posedness of the gov-
erning equations through the instantaneous elastic response. For example,
Loret and Prvost (1991) adopted a viscoplastic regularization technique
using the DuvautLions theory (1976), presented in Chapter 3, by which
the inviscid model can be transformed into a viscoplastic model with one
viscous parameter.
The elastoviscoplastic model, which is transformed from an elastoplastic
model, can maintain the well-posedness; namely, it can avoid the strong
mesh-size dependency in the fnite element analysis if the growth of the
viscoplastic strain is bounded. The numerical example for the mesh-size
dependency of the numerical analysis by an elastoviscoplastic model will
be shown later in this chapter. It is worth noting that the elastoviscoplastic
models can delay the instability, for example, shear bands, but cannot fully
stabilize the catastrophic instability.
8.6.4 Dynamic formulation
The dynamic formulation of boundary value problems has been success-
fully used to solve static boundary value problems. This may be due to the
fact that the type of governing equations remains hyperbolic. Vardoulakis
(1996a, 1996b) proposed a regularization method that introduced
274 Computational modeling of multiphase geomaterials
microinertial and strain gradients into the nonassociated plasticity model
called the second grade elastoplasticity model.
8.6.5 Discrete model and finite element
analysis with strong discontinuity
Since discrete analyses such as the discrete element method (DEM; Cundall
and Strack 1979; Kishino 1988; Thornton 1998) include an internal length
scale, the instability could be avoided, although it needs many degrees of
freedom. Oda and Kazama (1998) clearly showed that the shear band was
able to be reproduced by DEM through the weak restriction of the moment
between the particles. The fnite element method with strong discontinuity
has been developed to capture the slip surface without mesh-size depen-
dency (Oliver et al. 1999; Regueiro and Borja 2001).
As mentioned earlier, many studies have been done on the strain local-
ization problems of geomaterials, both experimentally and theoretically
(Darve 1994; Vardoulakis and Sulem 1995). However, many of them are
for cohesionless soils. On the other hand, comparatively fewer studies
have been performed for cohesive soil, such as clay, although Palmer and
Rice (1973) pointed out the strain localization of overconsolidated clay
as a slip surface.
It is well known that cohesive soil, such as clay, exhibits strainrate sen-
sitivity. Hence, in the modeling of clayey soil, it is more natural to adopt
elastoviscoplastic models instead of elastoplastic ones, although the elasto-
plastic models are extreme models for such materials. As was mentioned
earlier, elastoviscoplastic models have been developed for clay to take into
account the rate sensitivity.
Strain localization has been found to be the change in the type of partial
differential equations of the governing equations of boundary value prob-
lems for the rate-independent modeling of materials such as sand. In other
words, the type changes from elliptic to hyperbolic for static problems. On
the other hand, for elastoviscoplastic modeling, such instability as strain
localization can be treated as the exponential growth of fuctuation, that is,
Lyapunov instability, and as the growth of kinetic energy and momentum
for dynamic cases (see Table8.1).
8.7 INSTABILITY AND EFFECTS OF THE
TRANSPORT OF PORE WATER
In this section, we will discuss the instability of the viscoplastic water-
saturated soil, and clarify the effects of permeability and initial heterogene-
ity on the strain localization of fuid-saturated cohesive soil modeled by a
strain gradient-dependent poroviscoplastic constitutive model.
Strain localization 275
The effects of permeability and gradient parameters on the growth rate
of the fuctuation were obtained by a linear instability analysis. The defor-
mation behavior of the clay specimens modeled with a viscoplastic model
with a second-order strain gradient during shear was numerically analyzed
by a soilwater coupled FEM under both globally undrained and partially
drained conditions. We found that the deformation pattern and the stress
strain curve greatly depend on the permeability, the drainage conditions,
and the initial nonhomogeneous properties.
Rice (1976) and Rudnicki and Rice (1975) pointed out that the nature
of this problem can be solved within the general framework of bifurcation
problems and that localization problems should be studied within the wider
framework of mechanics, including the rapid degradation of the mate-
rial strength. In addition, Rice (1975) indicated the importance of local
inhomogeneity and the behavior of pore fuid. The effect of pore fuid on
localization problems has been analyzed by several researchers within the
context of a two-phase mixture theory such as Biots theory (1956) and de
Boer (1996). Loret and Prvost (1991), Schrefer et al. (1995, 1996), and
Ehlers and Volk (1998) numerically studied the localization problems of
water-saturated geomaterials with the rate-independent constitutive model.
Vardoulakis (1996a, 1996b) found that boundary value problems with a
nonassociated rate-independent plastic model become mathematically ill-
posed even if the pore water fow is included.
Oka, Adachi, and Yashima (1994) have been dealing with the localiza-
tion problem of water-saturated clay through the use of viscoplastic consti-
tutive equations due to the rate-dependent nature of cohesive soil. Zhang
et al. (1999) and Zhang and Schrefer (2000) investigated the interaction
between permeability and a gradient-dependent parameter with a one-
dimensional instability analysis and a numerical simulation in the context
of the dynamic strain localization of saturated and partially saturated
porous media. As for the experimental study, Finno et al. (1998) discussed
the effects of drainage conditions on strain localization in sand specimens.
In these studies, many points have been clarifed, such as the effect of
Table8.1 Classifcation of plastic and viscoplastic instability
Rate-Independent Model Rate-Dependent Model
Body Elastoplastic body Elastoviscoplastic body
Static conditions
Loss of ellipticity
Loss of controllability
Zero secondary order work
Loss of Lyapunov stability
Exponential growth of fuctuation
Loss of Lyapunov stability
Dynamic conditions
Loss of hyperbolicity
Loss of Lyapunov stability
Growth of kinetic energy and
momentum
Loss of Lyapunov stability
276 Computational modeling of multiphase geomaterials
dilatancy, permeability, and strain rates, for particular constitutive mod-
els. Loret and Prvost (1991) and Schrefer et al. (1995) showed that strain
localizes in a narrow zone in the case of higher permeability levels. On the
other hand, Oka, Adachi, and Yashima (1995) reported different results
in which deformation was more localized in the case of low permeability
levels compared with a material with absolutely very high permeability.
Several problems remain to be studied. One of them is to clarify the roles
of permeability and drainage conditions in the instability of the governing
equations and the deformation patterning of nonlocal viscoplastic mate-
rials, such as a higher-order strain gradient-dependent model. The other
problem is to clarify the effect of the initial heterogeneity.
8.7.1 Extended viscoplastic models for clay
Herein, we adopt the elastoviscoplastic model by Oka (1981), Adachi and
Oka (1982), Adachi et al. (1987a), and Higo (2003), discussed in Chapter 5,
and its generalized viscoplastic model with a higher-order straingradient.
The following formulation is based on the model presented in Chapter 5,
which is an extended model of the original model (Adachi and Oka 1982)
and takes shear softening into account.
The viscoplastic fow rule is given by

( ) ( )
1 2
D F
f
F
f
ij
vp
ij
s
s
=

, =

(8.36)
where D
ij
vp
is the viscoplastic stretching, is the viscosity parameter,
ij
is
the total stress tensor, and
ij
( =
ij ij w ij
u ) is Terzaghis effective stress
tensor.

2
1 = + (8.37)
The internal variable expresses the deterioration of the materials and
obeys the following evolutional equation:

2
2
2
M
G M
f
f
( )
(8.38)
where M
f

is the value of stress ratio

at the failure state,

G
2
is a material
parameter, and

is the stress invariant ratio defned by J

m
2 /
2
=

,
where J
2
is the second invariant of deviatoric stress tensor s
ij
and
m
is the
mean effective stress.
Strain localization 277
It is worth noting that a more rigorous viscoplastic model with strain
softening has been developed by Oka and Kimoto (2005) and was presented
in Chapter 5. It has been experimentally found that the shear strength and
the deformation characteristics of clay depend on the volumetric strain.
The volumetric plastic strain is used as a hardening parameter in the well-
known Cam-clay model (Roscoe et al. 1963). The volumetric inelastic
strain associated with both consolidation and dilatancy is a measure of
the deterioration of the granular materials. On the other hand, Mhlhaus
and Oka (1995, 1996) demonstrated that the higher-order gradients may
be attributed to the fact that the soil is discrete. Frantziskonis (1993) also
showed that the material inhomogeneity can be described by the con-
stitutive model with higher-order strain gradients. Thus, in the present
paper, we have introduced the second-order gradient of the viscoplastic
volumetric strain into the constitutive model to more accurately and more
suffciently describe the deformation of clay by considering the nonlocal
and the viscoplastic effects of the material. In practice, the yield function
includes the Laplacian of the viscoplastic volumetric strain and it is pro-
posed as follows:

f
J
M
a v
s
m
m
my
p
= + =

2
0
2
3
2
ln (8.39)
in which f
s
= 0 indicates the static state ( f f
s
= ),
s
is the hardening
parameter, v
p
is the viscoplastic volumetric strain (= D dt
kk
vp
), a v
p
3
2
is
the gradient term with a
3
, defned as a material constant, J
2
is the second
invariant of deviatoric stress tensor S
ij
, and
m
is the mean effective stress.

my
is a hardening parameter. The evolutional equation for
my
is

d
e
dv
my
my
p
=
+

1
(8.40)
where dv
p
is an increment in v
p
.
We assume that the dynamic yield function is the same as the static yield
function. Following the experimental results,
1
( ) F in Equation (8.36) is
given by

1
2
3
2
2
1
( ) exp
'
ln F C m
J
M
e
v v
f m
m
me
p
= +
+

pp
(8.41)

C C m
me
my
=
0
0
exp ln

(8.42)
278 Computational modeling of multiphase geomaterials
where m and C are viscoplastic parameters, and the gradient coeffcient
a
3
is assumed to be constant.
me
is the initial value of
m
,
my0
is the
initial value of the hardening parameter, is the consolidation index,
is the swelling index, and e is the void ratio.
Elastic stretching D
ij
e
(or strain rate tensor
ij
e

) is given by an isotropic
Hookes law, that is,

D
G
s
e
ij
e
ij
m
ij m
= +
+
1
2 3 1

( )

(8.43)
where G is the elastic shear modulus,
ij s is the deviatoric part of the
Jaumann rate of effective Cauchy stress rate tensor,
m

is the Jaumann rate

of mean effective stress, and
ij
is Kroneckers delta.
The Jaumann rate of Cauchys effective stress tensor is given by

W W
ij ij ik kj ik kj

=

+ (8.44)
where W
ij
is the spin tensor.
Total stretching D
ij
(or the strain rate tensor for the small strain case of
ij

) can be broken down into elastic and viscoplastic parts as

D D D
ij ij
e
ij
vp
ij ij
e
ij
vp
= + = + or

(8.45)
8.7.2 Instability analysis of fluid-
saturated viscoplastic models
8.7.2.1 Instability under locally undrained conditions
We can easily discuss the instability of the model given by Equation(8.36)
to study the conventional undrained creep behavior under locally
undrained conditions, that is, the permeability coeffcient is zero. Oka,
Adachi, and Yashima (1995) obtained the time rate of the second invariant
of viscoplastic strain rate, I
vp
2
, under the undrained creep conditions where
the deviatoric stress levels are constant as

I a I
vp vp
2 2
2
= ( )[ ]
*
(8.46)
where I
vp
ij
vp
ij
vp
2
=

,
*
= 2
2
J
m
.
When a( )
*
0,

I
vp
2
is negative or zero, the system is stable. On the other
hand, when a( )
*
> 0, the system becomes unstable because the small fuc-
tuation in the viscoplastic strain rates will grow. It is worth noting that
Strain localization 279

c
M
* *
< (M
*
is the value of
*
at the critical state and
c
*
is the value of
*

when a( )
*
= 0). This indicates that the clay will be unstable before the criti-
cal state under the locally undrained conditions in which the permeability
coeffcient is zero in the normally consolidated region.
In addition, Oka, Adachi, and Yashima (1995) obtained the growth rate
of small fuctuation, , under plane strain locally undrained conditions
using the linear stability analysis as

=

+
Z
Z H
2
1
2
1
2 cot
(8.47)
where is the angle of the shear band and Z
1
0 > , > 0, and H
1
0 > .
Hence if Z
2
0 < , the material becomes unstable and the fuctuation will
grow in the orientation of = 45. This means that the angle of the shear
band is 45. These results coincide with the prediction by the bifurcation
analysis for an elastoplastic model (Vardoulakis 1996). The results of the
aforementioned instability are limited in the normally consolidated region
(Figure 8.4). However, the instability analysis for the viscoplastic model
with degradation proposed by Kimoto (2002) and Kimoto and Oka (2005)
(see Equation 5.87) showed that the material is more unstable in the over-
consolidated region (Figure8.5).
For the purpose of evaluating the instability when q equals the constant
undrained creep condition, the numerical simulations of undrained creep
tests under a triaxial compression stress state are conducted. The second
Mean Efective Stress (kPa)
0 100 200 300 400
Stress path
= 80
= 40
= 20
M
Unstable region
I
2
> 0
700
600
500
400
D
e
v
i
a
t
o
r

S
t
r
e
s
s

(
k
P
a
)
300
200
100
0
500 600 700
Figure8.4 Unstable regions (
* *
< M ).
280 Computational modeling of multiphase geomaterials
invariant of the deviatoric viscoplastic strain, I
vp
2
, for the viscoplastic model
with degradation, described in Chapter 5, is given as

I e e C m M y
vp
ij
vp
ij
vp m
ma
m 2
= = + exp ' ln
* * *

(8.48)

dy
e
d
m v
vp *
=
+

1
0

(8.49)
The differentiation of I
vp
2
, with respect to time under the q equals con-
stant undrained creep condition, is obtained as

I m M y I
m
e
M M
z
M M I
m a z I
vp m
m
ma
ma
m
vp
mai maf
ma
vp
vp
1
exp( )
1 ( )
( , )
2
* * *
2
0 * * * *
* * * 2
2
2
1
*
2
2

( )
( )
= +

=
+

+
=
(8.50)
in which
z
ij
vp
ij
vp
= and z zdt
t
=
0

.
Mean Efective Stress (kPa)
0 100 200 300 400
Stress path
f
b
= 0
500 600
Unstable region
in the original model = 0
Unstable
region I
2
> 0
M = (3/2 M
*
)
M/( )
= 20
= 40
= 80
600
500
400
D
e
v
i
a
t
o
r

S
t
r
e
s
s

(
k
P
a
)
300
200
100
0
Figure8.5 Unstable regions (
* *
> M ).
Strain localization 281
In the numerical simulations of undrained creep tests, the deviator stress
levels are applied in a second for each case and kept constant until the sign for
index

I
vp
2
becomes positive. Figures8.4 and 8.5 show the unstable regions in
which the sign for

I
vp
2
becomes positive during undrained creep for normally
consolidated and overconsolidated clay, respectively. In the case of the original
model with the material parameter of = 0,

I
vp
2
becomes positive only in the
region of M M
* * *
/( ) < < , whereas in the case of the model with degra-
dation, the unstable region changes with the value of the material parameter ,
which controls the degradation rate and expands around the failure line. This
point will be numerically discussed in Sections 8.8 and 8.9.
8.7.2.2 Instability analysis considering the pore water flow
Let us consider the instability of the pore water fow in soil. Loret and
Prvost (1991) and Schrefer et al. (1995) studied the effects of permeability
on the dynamic strain localization analysis using a MohrCoulomb law
with the associated fow rule. Loret and Prvost (1991) stated that in the
case of low permeability levels, the instability may develop more slowly
than in the case of high permeability levels. On the other hand, Oka etal.
(1995) conducted a numerical analysis of strain localization under quasi-
static conditions using an elastoviscoplastic model. From the numeri-
cal results, they pointed out that the distribution of pore water pressure
is moderate with higher permeability levels and the strain localization is
weaker. In their paper, the two results were compared and they suggested
that the difference between the two results might be due to the different
dilatancy characteristics. They pointed out that they took different loading
conditions, such as quasi-static and dynamic conditions, and indicated the
difference between rate-dependent and rate-independent models. However,
the effects of permeability on the strain localization for a viscoplastic mate-
rial under quasi-static conditions have not yet been fully studied.
Oka et al. (1995) stated only the effect of permeability on the strain
localization obtained through the distribution of pore water pressure in
their conclusions. Hence, it should be pointed out that the effect of perme-
ability on the strain localization using an elastoviscoplastic model under
quasi-static conditions has to be studied by both a numerical simulation
and an instability analysis. To more clearly discuss the effects of perme-
ability on strain localization, an instability analysis was carried out under
two-dimensional conditions within the context of a small strain theory for
simplicity (Oka et al. 1999; Higo et al. 2005). An instability analysis has
been conducted by Oka et al. (1999) in which a simplifed linear rigid-visco-
plastic model was used. It was confrmed that the simplifed rigid viscoplas-
tic model neglecting the dilatancy model leads to the more instability with
larger permeability in the strain softening. From the instability analysis, the
effect of the permeability depends on the strain softening characteristics.
282 Computational modeling of multiphase geomaterials
Itis worth noting that the stability related to the permeability also depends
on the dilatancy characteristics of the materials.
8.8 TWO-DIMENSIONAL FINITE ELEMENT ANALYSIS
USING ELASTOVISCOPLASTIC MODEL
Using the elastoviscoplastic constitutive model by Oka et al. (1995) and
the fnite deformation numerical analysis method described in Chapters 5
and 6, a two-dimensional fnite element analysis has been performed under
plane strain conditions (Higo et al. 2005). From the numerical results,
strain localization characters are discussed.
8.8.1 Effects of permeability
In Figure8.6, the boundary conditions are shown for the plane strain prob-
lem used in the numerical analysis, while the parameters used in the com-
putation are shown in Table8.2. The gradient parameter, in principle, can
be determined by the width of the shear band, namely, the wavelength of
the localized pattern. The strain rate of the compression is 1.0%/min. The
horizontal displacement at the top and the bottom of the specimen was
fxed as a trigger of the localization. All of the boundaries were assumed to
be impermeable, while the pore fuid was allowed to fow in the specimen.
The average vertical stress versus strain relations with different coeffcients
of permeability are shown in Figure8.7. In the early stage of loading, that is,
in the strain-hardening range, little difference can be seen among the three
40 cm
Compelled displacement
1%/min
Top and bottom edges
are fxed as a trigger of
strain localization
20 cm
20 20 =
400 elements
Half of the specimen was
calculated for symmetry
20 cm
20 cm
: Horizontally fxed
: Vertically fxed
: Fixed
Impermeable
boundary
Impermeable
boundary
Figure8.6 Boundary conditions and the size of the specimen.
Strain localization 283
cases. On the other hand, in the strain-softening range, the material with a
low permeability level of 1.54

10
12
(m/s) is less unstable, that is, it is rela-
tively stable because the average stress is larger than materials with higher
permeability levels of

10
6
(m/s) and

10
8
(m/s) before an average axial
strain of 5%. This behavior is consistent with the theoretical consideration
mentioned in Section 8.7. After an axial strain of 5%, however, the average
stress in the case of

10
8
(m/s) is smaller than that in the case of

10
6
(m/s).
These results indicate that a material with a small permeability coeffcient is
not necessarily more stable than one with a larger permeability coeffcient.
Table8.2 Material parameters for the strain localization analysis with
different permeability coeffcients
Compression index, 0.172
Swelling index, 0.054
Initial void ratio, e
0
1.28
Initial mean effective stress,
me
200 (kPa)
Coeffcient of earth pressure at rest, K
0
1.0
Viscoplastic parameter, m 21.5
Viscoplastic parameter, C 4.5 10
8
(1/s)
Stress ratio at failure, M
f
*
1.05
Elastic shear modulus, G 5500 (kPa)
Softening parameter, G
2

100
Gradient parameter, a
3
0.0, 5.0, 30.0 (10
4
m
2
)
Coeffcient of permeability, k
x
= k
y
(m/s) 1.54 10
6
, 10
8
, 10
12
Average Axial Strain (%)
0
100
200
A
v
e
r
a
g
e

V
e
r
t
i
c
a
l

S
t
r
e
s
s

(
k
P
a
)
k = 1.54 10
6
(m/s)
k = 1.54 10
8
(m/s)
k = 1.54 10
6
(m/s)
k = 1.54 10
2
(m/s)
k = 1.54 10
8
(m/s)
k = 1.54 10
12
(m/s)
300
0 5 10
Figure8.7 Average vertical stressstrain relations with different coeffcients of
permeability.
284 Computational modeling of multiphase geomaterials
The calculations with coeffcients of permeability of k = .

1 54 10
8
(m/s)
and 10
12
(m/s) diverged around 8% of the axial strain. The calculations
with k = .

1 54 10
6
(m/s) also diverged around 11% of the axial strain.
This may be because the constraint conditions, that is, no lateral displace-
ments at either the top or the bottom plates, induce numerical instabil-
ity near the top and the bottom of the specimen. Figure 8.8 shows the
deformed mesh at an average axial strain of 7% with different coeffcients
of permeabilityk. It is shown that the pore fuid has an apparent infuence
on the formation of shear bands. It is found that a symmetrical deforma-
tion can be seen in all cases, in particular, a clear shear band formed in the
case of k = .

1 54 10
6
(m/s). In Figure 8.9, velocity vectors are presented
for half of the specimens at an average axial strain of 7% with different
coeffcients of permeabilityk. Discontinuous distributions of velocity felds
are found in all cases due to the formation of shear bands. The patterns of
distributions of the velocity vectors are consistent with the deformed mesh.
Figure 8.10 shows the distributions of accumulated viscoplastic shear
strain,
p
, at average axial strain levels of 3% and 7% with different coef-
fcients of permeability k.
p
is defned as follows:


p p p
ij
p
ij
p
d d de de = =

, ( )
/ 1 2
(8.51)
k = 1.54 10
6
m/s k = 1.54 10
8
m/s k = 1.54 10
12
m/s
Figure8.8 Deformed mesh at an average axial strain of 7%.
k = 1.54 10
6
m/s k = 1.54 10
8
m/s k = 1.54 10
12
m/s
Figure8.9 Distribution of the velocity vector at an average axial strain of 7%.
Strain localization 285
where de
ij
p
is the viscoplastic deviatoric strain-increment tensor. The local-
ized patterns of the fgures for 3% are very similar to each other, but the
maximum value for
p
becomes larger as coeffcient of permeability k
decreases. On the other hand, at an average axial strain of 7%, the maxi-
mum value for
p
is larger in the case of higher coeffcients of permeability
than that of lower coeffcients of permeability. It can be said that materials
with larger accumulated shear strain are more unstable than those with
smaller shear strain. Following this point of view, when the average axial
strain is small in the viscoplastic-hardening area, materials with lower per-
meability levels are rather unstable. On the other hand, when the average
axial strain becomes large in the viscoplastic-softening area, materials with
higher permeability levels are relatively unstable. This tendency is also con-
sistent with the results obtained in Section 8.7. In addition, a larger differ-
ence between the maximum and the minimum values for
p
is also seen in
the case of higher permeability levels. This suggests that the strain localizes
prominently when materials have high permeability levels. It is interesting,
however, that two shear bands appear in the case of k = .

1 54 10
6
(m/s),
while the other cases have four shear bands and the distance between two
shear bands is larger in the case of small permeability compared with the
case of large permeability.
The inclination angles of the shear bands for all cases are 45 at the small
axial strain of 3%. When the axial strain becomes 7%, the angles of the
shear bands with higher permeability become larger than those with lower
permeability. Oka et al. (1995) demonstrated that the preferred orientation
Axial Strain of 7%
k = 1.54 10
6
m/s k = 1.54 10
8
m/s k = 1.54 10
12
m/s
k = 1.54 10
6
m/s k = 1.54 10
8
m/s k = 1.54 10
12
m/s
Axial Strain of 3%
56 56
56 56
49 49 47 47
45 45 45 45 45 45
47 47
0.08
0.05
0.03
0.00
0.99
0.66
0.33
0.00
0.48
0.32
0.16
0.00
0.37
0.25
0.12
0.00
0.08
0.06
0.03
0.00
0.15
0.10
0.05
0.00
Figure8.10 Distribution of
p
at an average axial strain of 3% and 7%.
286 Computational modeling of multiphase geomaterials
for the shear bands is 45 under plane strain locally undrained conditions,
that is, k=0, for Adachi and Okas viscoplastic model introduced in Section
5.5. The reasons why the angles of the shear bands with lower strain are
more proximate to 45 than materials with lower permeability levels are
similar to the reasons for those under locally undrained conditions. In
Figure 8.11, the distributions of pore water pressure are shown with dif-
ferent coeffcients of permeability k. When permeability k is smaller, the
pore water pressure is more localized. This trend is similar to the results
obtained by Oka et al. (1995).
The distributions of mean effective stress, the second invariant of devi-
atoric stress, and the volumetric viscoplastic strain are all affected by the
formation of shear bands and are inhomogeneous. The mean effective
stress inside the shear band becomes smaller than that outside the shear
band. The maximum values for the deviatoric stress and the volumetric
viscoplastic strain become larger with higher permeability levels.
8.8.2 Strain localization analysis by the gradient-
dependent elastoviscoplastic model
8.8.2.1 Finite element formulation of the gradient-
dependent elastoviscoplastic model
For higher-order gradient-dependent constitutive equations, a second-order
gradient of viscoplastic volumetric strain, v
p
, is used in the constitutive
equation. In order to obtain the second-order gradient by the fnite element
method, the discretization of v
p
as an independent variable by the eight-
node quadrilateral element is needed. Hence, we assume the weak form of
the yield function and defne viscoplastic volumetric strain, v
p
, at each node
in the same manner as that by Aifantis et al. (1999).
To obtain the weak form of the yield function, we adopt a Taylor series
expansion around the current state and consider the frst term. Introducing
the second gradient of the viscoplastic volumetric strain, we have

p
ij
p p
v
G v v

= , , ( ' )
2
(8.52)
160.00
159.49
158.99
158.49
186.60
173.53
160.47
147.41
445.29
341.18
237.07
132.95
k = 1.54 10
6
m/s k = 1.54 10
8
m/s k = 1.54 10
12
m/s
Figure8.11 Distribution of pore water pressure at an average axial strain of 7%
(unit: kPa).
Strain localization 287
Expanding the viscoplastic volumetric strain rate in a Taylor series and
disregarding the second- and higher-order terms, we obtain a linearized
evolutional equation in the form

p p
v
p p
v v
G G
v
G
v p


= + + +
( ) 0
2
[ ][ '] ( )

(8.53)
where

( )
2
G
G
G
G
v
G
G
v
ij
v p p
p =

, =

, =

(8.54)
0
p
v denotes the value of the volumetric strain rate at the current state.
Using a Taylor series expansion and truncating the frst-order term, we
obtain the following expression for the total strain rate tensor:

[ ] [ ] [ ] [ ] [ ] [ ][ ] [ ]
* * **
= + + + +

L A v A A
p p 1
0
2
vv
p
(8.55)
where [ ]
*
L
1
denotes the inversion of the elastic tensor and

A
v
A A
v
ij
vp
p
ij
vp
kl
ij
vp
p
=

, ,
( )
* **
2
From Equation (8.55), the stress rate tensor,
[ ]

, is obtained as

[ ] [ ][ ] [ ][ ] [ ][ ] [ ][ ] ,




=

L L L A
v
L A
v
p p p
0
2
[[ ] [ ] [ ] L L A

= +
1 1
(8.56)
where [ ]

is the total strain rate tensor and

[ ]
0
p

is the viscoplastic strain rate

tensor at the current state. In addition, for the two-phase material we use
the effective stress tensor for the stress.
8.8.2.2 Effect of the strain gradient parameter
Strain gradients, in principle, can describe the thickness of shear bands.
In addition, it is found in the instability analysis of Section 8.7 that strain
gradients act as stabilizers. In this section, the effects of the strain gra-
dient parameter on the strain localization analysis are investigated. The
boundary conditions and the material parameters are the same as those
288 Computational modeling of multiphase geomaterials
mentioned in the last section. Figure8.12 depicts the distributions of accu-
mulated viscoplastic shear strain,
p
, at an average axial strain of 7%
with different gradient parameters. In this case, k = .

1 54 10
10
(m/s). It
can be found from the fgure that the thickness, the spacing of the shear
bands, and the extent of the strain localization also depend on the gra-
dient parameter, a
3
. The accumulated strain is more localized when the
gradient parameter, a
3
, is rather small. This suggests that the gradient
term makes the system more stable. The distance between shear bands
will decrease with the gradient parameter, a
3
, while the angles of the shear
bands are consistently 48.
8.8.2.3 Effect of the heterogeneity of the soil properties
The boundary conditions are shown in Figure8.13, and Table8.3 shows
the material parameters used in the analysis in this section. We assumed
three cases of distribution for the stress ratio at critical state, M
m

, as shown
in Figure8.14. The perturbation of M
m

was obtained using a pseudo-ran-

dom number by the linear congruential method.
In Figure8.15, the stressstrain relations for cases 1 to 3 are obtained
by the different ranges in perturbation of M
m

, R; R = 0.5%, 1.0%, and

3.0%. R = 0% means a homogeneous clay sample. The applied strain rate
is 1.0/min. The effects of heterogeneity on the stressstrain relations are
dependent on the initial distribution of M
m

. It can be seen in case 1 that

the average vertical stress of the nonhomogeneous clay is a little larger than
that of the homogeneous clay, but that it becomes smaller in the failure
state. On the other hand, the heterogeneous clay in case 3 shows softening
behavior and the average stress is smaller than that of the homogeneous
clay. The axial strain at the failure state is consistently smaller as the range
of perturbation, R, is larger. Figure8.16 shows the deformed mesh and the
distributions of
p
for a homogeneous one and for case 1 with different R
at an average axial strain of 10%. Regarding the homogeneous case, the
deformation and the distribution of
p
are uniform in the specimen. On the
48
0.43
0.28
0.14
0.00
0.35
0.24
0.12
0.00
0.25
0.18
0.09
0.00
48 48 48 48 48
a
3
= 0.0 a
3
= 5.0 a
3
= 30.0
(Gradient independent)
Figure8.12 Distribution of
p
at an average axial strain of 7% (k = 1.54 10
8
cm/s).
Strain localization 289
contrary, localized deformations are seen in the nonhomogeneous clay and
the shear band of R = 3% is clearer than the others. Similar results have
been obtained by Ehlers and Volk (1998). They showed that a random dis-
tribution of the Lam constants within local deviations of 0.5% provided
clear shear bands under plane strain conditions, although the homogeneous
specimens do not.
Compelled displacement
1%/min
Impermeable
boundaries
20 cm
20 cm
: Fixed
: Vertically fxed
20 20 =
400 elements
Figure8.13 Boundary conditions and the size of the specimen (undrained plane strain
condition).
Table8.3 Material parameters for the strain localization analysis
of homogeneous and heterogeneous types of clay
Compression index, 0.372
Swelling index, 0.054
Initial void ratio, e
0
1.28
Initial mean effective stress,
me
600 (kPa)
Coeffcient of earth pressure at rest, K
0
1.0
Viscoplastic parameter, m 21.5
Viscoplastic parameter, C 4.5 10
8
(1/s)
Stress ratio at critical state,
*
M
m
1.05
Elastic shear modulus, G 13210 (kPa)
Softening parameter, G
2

100
Gradient parameter, a
3
0.0 (m
2
)
Coeffcient of permeability, k
x
= k
y
1.16 10
14
(m/s)
290 Computational modeling of multiphase geomaterials
8.8.2.4 Mesh-size dependency
We should discuss the mesh-size dependency of the present analysis. It is
well known that the fnite element analysis has an inherent mesh-size depen-
dency and that many researchers have studied regularization methods. The
frst method is to introduce the rate dependency of the material through
Case 1 Case 2 Case 3
1.071
1.057
1.043
1.029
1.071
1.057
1.043
1.029
1.071
1.057
1.043
1.029
Figure8.14 Initial distribution of M
f

(R = 2.0%).
R = 3.0%
R = 1.0%
700
600
500
400
300
200
100
0
0 5 10 15
A
v
e
r
a
g
e

V
e
r
t
i
c
a
l

S
t
r
e
s
s

(
k
P
a
)
Average Axial Strain (%)
Case 1
Homogeneous
R = 0.5%
R = 1.0%
R = 3.0%
R = 0.5%
R = 3.0%
R = 1.0%
700
600
500
400
300
200
100
0
0 5 10 15
A
v
e
r
a
g
e

V
e
r
t
i
c
a
l

S
t
r
e
s
s

(
k
P
a
)
Average Axial Strain (%)
Case 3
Homogeneous
R = 0.5%
R = 1.0%
R = 3.0%
R = 0.5%
Figure8.15 Stressstrain curves for Case 1 and Case 3 obtained by different R.
Strain localization 291
the use of an elastoviscoplastic model or regularization in the numerical
analysis (Hughes and Taylor 1978). The second method is to introduce
higher-order strain gradients into the constitutive model (Aifantis 1984).
The third approach is to incorporate Darcys law of soilfuid interaction,
which can alleviate the problem of instability by delaying the onset of mate-
rial instability (Rice 1975). Herein, the mesh-size dependency is small since
an elastoviscoplastic model with the second-order strain gradient and a
solidfuid mixture theory was applied. Oka et al. (2002b) found that the
analysis method has no signifcant mesh-size dependency. It is worth noting
that Zhang and Schrefer (2000) have shown that gradient dependence and
permeability can regularize the fnite element solution and that the second
sometimes prevails over the frst.
8.9 THREE-DIMENSIONAL STRAIN LOCALIZATION
ANALYSIS OF WATER-SATURATED CLAY
Strain localization has been numerically analyzed for geomaterials by many
researchers. Many of them, however, were treated as two-dimensional
problems although the phenomena are generally three dimensional. To
investigate the strain localization behavior of geomaterials under three-
dimensional conditions, undrained triaxial compression tests using rectan-
gular specimens and their simulation by a fnite element analysis using an
elastoviscoplastic model by Kimoto and Oka (2005) have been conducted
(a) Deformed mesh at an average axial strain of 10%
(b) Distribution of
p
at an average axial strain of 10%
0.08
0.17
0.25
0.34
0.07
0.15
0.23
0.30
0.07
0.14
0.22
0.29
Case 2 Case 3
Case 1
Case 1
Case 2 Case 3
Figure8.16 Deformed mesh and the distribution of
p
for cases 1, 2, and 3 (R = 2.0%). (a)
Deformed mesh at an average axial strain of 10%. (b) Distribution of
p
at an
average axial strain of 10%.
292 Computational modeling of multiphase geomaterials
(Higo et al. 2006). In the experiments, both normally consolidated and over-
consolidated clay samples were tested with different strain rates (Kodaka
et al. 2001; Oka et al. 2005; Higo et al. 2006). Using the distribution of
shear strain obtained by the image analysis of digital photographs taken
during deformation, the effects of the strain rates, dilatancy, and overcon-
solidation on strain localization have been studied in detail. The method
of numerical simulation is a soilwater coupled fnite element method that
is based on the fnite deformation theory described in Chapter 6, using
an elastoviscoplastic model for water-saturated clay considering structural
changes, as shown in Chapter 5.
8.9.1 Undrained triaxial compression tests
for clay using rectangular specimens
8.9.1.1 Clay samples and the testing program
The clay used in the experiment is Fukakusa clay, which is a Pleistocene
marine clay produced in the southeastern part of the Kyoto Basin. The
liquid limit was w
L
= 62%, plasticity index I
p
= 33, and the density of soil
solid
s
= . 2 69 g/cm. Reconstituted clay samples were prepared by remold-
ing them in slurry and then preconsolidating them. The specimens were
consolidated one dimensionally at a preconsolidation pressure of 98 kPa.
The preconsolidated specimens were covered with paraffn and not to be
disturbed. The scale of the transverse section is 4 cm 4 cm and the height
is 8 cm (see Figure 8.17). The test cases are listed in Table 8.4. All the
8 cm
4

c
m
4 cm
f s
f : Front surface
s: Side surface
Figure8.17 Sizes of the specimens.
Strain localization 293
specimens used were saturated by the double vacuum method and were
acted upon by 200 kPa of back pressure. The normally consolidated clay
specimens were isotropically consolidated to 200 kPa. The overconsoli-
dated clay specimens were isotropically consolidated to 300 kPa, and then
isotropically swelled to 50 kPa. Therefore, the overconsolidation ratio
(OCR) is 6.
After the consolidation or the swelling procedure, axial pressure was
applied under undrained conditions by an axial loading device with an axial
strain or displacement control system. The three axial strain rates mono-
tonically applied in the tests were 1%/min, 0.1%/min, and 0.01%/min. The
tests were stopped at an axial strain of 20%.
8.9.1.2 Image analysis
We drew 2 mm square meshes on the rubber membranes covering the speci-
men. A digital camera was used to take photographs of two surfaces of
the specimens during the tests. The following photograph and Figure8.18
show a sample of the digital photographs taken through the triaxial cell
and a schematic fgure of the photography, respectively. After correct-
ing the effects of the refraction, we digitized the nodal coordinates of the
meshes. Using the coordinates at the initial state, that is, before the und-
rained loading, and those of each axial strain level, the nodal displacements
were calculated. Adopting the B matrix for the four-node isoparametric
Table8.4 Test cases
No. Sizes (c
m
)
Strain rate
(%/min)
Normally Consolidated Clay

m0
200 = kPa (Consolidation pressure)
NC-1 4 4 8 1
NC-2 4 4 8 0.1
NC-3 4 4 8 0.01
Overconsolidated clay

mc
=300 kPa (Pre-consolidation pressure)

m0
0 =5 kPa (Swelling pressure), OCR = 6
OC-1 4 4 8 1
OC-2 4 4 8 0.1
OC-3 4 4 8 0.01
294 Computational modeling of multiphase geomaterials
fnite elements provided the strain for each element (Kodaka et al. 2001;
Oka et al. 2005).
8.9.2 Three-dimensional soilwater coupled
finite element analysis method
We used the three-dimensional fnite element method presented in
Chapter6, based on Biots two-phase mixture theory and the fnite defor-
mation theory, to simulate the three-dimensional strain localization tests
for rectangular-shaped clay specimens. The strain localization phenomenon
is a geometrically nonlinear problem since the deformation in a shear band
is large. In addition, the constitutive equation for clay used in this study is
nonlinear and is defned in an incremental form. In order to deal with such
a nonlinear large deformation problem using an incremental constitutive
model, an updated Lagrangian method with the objective Jaumann rate
of Cauchy stress is used for the weak form of the rate type of equilibrium
equations. As for the element type, a 20-node quadrilateral isoparamet-
ric element with a reduced Gaussian four-point integration is used for the
displacement to eliminate shear locking as well as to reduce the appear-
ance of a spurious hourglass mode. The pore water pressure is defned by
an eight-node quadrilateral isoparametric element. The formulation of the
three-dimensional soilwater coupled fnite element analysis method is the
same as those presented in the last section. Detailed formulations are in
the references (Oka et al. 2002b; Higo 2003; Higo et al. 2005, 2006).
Triaxial cell
Support pillar
Specimen
f
(Front)
(Side)
S
2 mm mesh
(a) (b)
Figure8.18 (a) Schematic fgure of the triaxial cell. (b) An Example of the photographs
taken through the acrylic cell.
Strain localization 295
8.9.3 Numerical simulation of triaxial
tests for rectangular specimens
8.9.3.1 Determination of the material parameters
The material parameters required by the constitutive model introduced in
the last section are listed in Table8.5. We determined to be 0.191 and
to be 0.043 using the isotropic consolidation and the swelling test results
for Fukakusa clay. For initial void ratio, e
0
, we used the average obtained in
each test, that is, 1.10 for normally consolidated clay and 1.11 for overcon-
solidated clay, since calculating with different void ratios is not appropriate
for a comparison of the simulation results.
The initial elastic shear modulus, G
0
, is determined by the initial slope
of the undrained triaxial compression tests, namely, G q
0 11
3 = /( ) , in
which q is the increment in deviator stress and
11
is the increment in
axial strain. In this study,
11
was determined to be 0.1%. G
0
is depen-
dent on the strain rate because of the viscoelastic properties. The com-
pression yield stress,
mbi
, is assumed to be the preconsolidation stress.
Therefore, that of normally consolidated clay is 200 kPa and that of over-
consolidated clay is 300 kPa. The stress ratio M
m

, is defned as the stress

ratio whereby maximum compression occurs in the drained compression
tests. Herein, M
m

is assumed to be determined from the stress ratio at the

critical state in the undrained triaxial compression tests.
Table8.5 Material parameters used in the numerical simulation
Parameter NC Clay OC Clay
Compression index, 0.191 0.191
Swelling index, 0.043 0.043
Initial void ratio, e
0
1.10 1.11
Initial elastic shear
modulus, G
0
0.01%/min 16300 (kPa) 9190 (kPa)
0.1%/min 17700 (kPa) 9920 (kPa)
1%/min 23400 (kPa) 13080 (kPa)
Initial mean effective stress,
m0
200 (kPa) 50 (kPa)
Compression yield stress,
mbi
200 (kPa) 300 (kPa)
Coeffcient of earth pressure at rest, K
0
1.0 1.0
Stress ratio at maximum compression, M
m

1.14 1.14
Viscoplastic parameter, m 24.3 20.5
Viscoplastic parameter, C 5.8 10
10
(1/s) 2.7 10
9
(1/s)
Structural parameter,
maf
170 (kPa) 270 (kPa)
Structural parameter, 10 5
Coeffcient of permeability, k 1.63 10
9
(m/s) 2.86 10
9
(m/s)
296 Computational modeling of multiphase geomaterials
The viscoplastic parameters m and C can be determined from undrained
triaxial tests with different strain rates. Adachi and Oka (1982) noted that
viscoplastic parameter m is estimated from the slope of the relation between
the stress ratio and the logarithm of the strain rate. In principle, material
parameters are determined from elemental tests. In this study, however,
undrained triaxial compression tests using cylindrical specimens were con-
ducted only for a strain rate of 1.0%/min, since we could not prepare a suf-
fcient number of specimens for the tests using cylindrical specimens that
had been consolidated under the completely same conditions. Hence, we
applied the test results using rectangular specimens (4 cm 4 cm 8 cm) to
estimate the viscoplastic parameter, m (Higo 2003).
Figure 8.19 shows the relations between the applied strain rates and
stress ratio q
m
/

in which q is the deviator stress and

m
is the mean
effective stress. For normally consolidated clay, we plotted the stress
ratios at
m m
= .

0 7
0
, and estimated m to be 24.3. As for the over-
consolidated clay, m is estimated to be 20.5 from the stress ratios at
= .
m m
2 0
0
. After m is fxed, the viscoplastic parameter C is determined
by the peak stress. The structural parameter,
maf
, can be obtained by
the deviator stress at the residual stress state, while dominates the
decreasing rate of deviator stress.
8.9.3.2 Boundary conditions
Figure 8.20 shows the boundary conditions, which are set up accord-
ing to the same boundary conditions as those of the undrained triaxial
10
0
1%/min
10
1
0.1%/min
0.9 1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9
10
2
NC clay
OC clay
0.01%/min
S
t
r
a
i
n

R
a
t
e

(
%
/
m
i
n
)
q/
m
Figure8.19 Relations between the logarithm of the strain rate and the stress ratio (NC
clay:
m
= 0.7
m0
, OC clay:
m
= 2.0
m0
).
Strain localization 297
compression tests with displacement control. All the boundaries are assumed
to be impermeable, however, the transport of pore water between each
element is allowed. Constant displacement rates (z-direction) of 0.01%/
min, 0.1%/min, and 1%/min are applied to the nodes on the bottom sur-
face. The time increment is determined by the increment of average strain
= 0.05%. As for the top and the bottom surfaces, frictional force occurs
between the top and the bottom surfaces and the top cap and the pedestal.
Hence, the top and the bottom surfaces deform. However, it is diffcult to
accurately estimate the friction force. In addition, the horizontal displace-
ments of the top and the bottom surfaces, which were measured after the
tests (see Higo 2003), are rather small. Consequently, we assumed that the
horizontal (x-direction and y-direction) displacement of the nodes on both
the top and the bottom surfaces is constrained.
8.9.3.3 Comparison between experimental
and simulation results
Undrained triaxial compression tests for normally consolidated clay
and overconsolidated clay with different axial strain rates have been
simulated. Figures8.21 and 8.22 show the experimental results and the
simulation results for normally consolidated clay and overconsolidated
clay, respectively.
8.9.3.3.1 Stressstrain relations and effective stress paths
Figures 8.21a and 8.22a illustrate the stressstrain relations for both the
simulation and the experiment. The deviator stress, the mean effective stress,
and the pore water pressure used in the stressstrain curves and the effective
X
Displacement control with
0.01%/min, 0.1%/min, and 1%/min
All boundaries are impermeable
(permeable between each element)
Fixed
Horizontally fxed
Y
Z
Figure8.20 Boundary conditions for three-dimensional fnite element analysis.
298 Computational modeling of multiphase geomaterials
stress paths for simulation results are calculated in the same manner as
the experimental data. The deviator stress and the pore water pressure are
obtained using the average of those nodal values of the top surface.
We can see that the stressstrain relations for both the experiment
and the simulation are greatly dependent on the strain rate and the
dilatancy characteristics. In the experimental results, we can also observe
250
200
150
D
e
v
i
a
t
o
r

S
t
r
e
s
s

(
k
P
a
)
100
50
0
250
200
150
D
e
v
i
a
t
o
r

S
t
r
e
s
s

(
k
P
a
)
100
50
0
0 5 10 15 20 250 200 150 100 50 0
0.50 0.25
f s
(d) Picture of the specimen
0.00
0.50 0.25
(f ) Deformed mesh
0.00
(e) Distribution of shear strain (simulation)
(c) Distribution of shear strain (experiment)
Axial Strain (%)
(a) Stressstrain relationships
Test Simulation
20%
20% Simulation
NC-2
Axial Strain
f
s
0% 4% 8% 12% 16% 20%
Experiment
NC-2
f
s
0% 4% 8% 12% 16% 20%
f s
Y
X
Z
Mean Efective Stress (kPa)
(b) Efective stress paths
1%/min
0.1%/min
0.01%/min
1%/min
0.1%/min
0.01%/min
Failure line
M
m
= 1.4
Test
Simulation
1%/min
0.1%/min
0.01%/min
1%/min
0.1%/min
0.01%/min
Figure8.21 Comparison between the experimental results and simulation results (NC
clay, 0.1%/min).
Strain localization 299
gradual strain-softening behavior for both normally consolidated clay
and overconsolidated clay. On the other hand, in the case of the simula-
tion, it is seen that the stressstrain relations for normally consolidated
clay consistently show strain-hardening behavior, while those for over-
consolidated clay show strain-softening behavior just after the peak
stress around an axial strain of 2%, and then they show a gradual
hardening.
250
200
150
D
e
v
i
a
t
o
r

S
t
r
e
s
s

(
k
P
a
)
100
50
0
250
200
150
D
e
v
i
a
t
o
r

S
t
r
e
s
s

(
k
P
a
)
100
50
0
0 5 10 15 20
250 200 150 100 50 0
0.50 0.25
f s
(d) Picture of the specimen
0.00
0.50 0.25
(f ) Deformed mesh
0.00
(e) Distribution of shear strain (simulation)
(c) Distribution of shear strain (experiment)
(a) Stress strain relationships
Test
Test
Simulation
Simulation
Axial Strain (%)
20%
Simulation
OC-2
Axial Strain
Axial Strain
f
s
0% 4% 8% 12% 16% 20%
Experiment
OC-2
f
s
0% 4% 8% 12% 16% 20%
Y
X
Z
Mean Efective Stress (kPa)
(b) Efective stress paths
1%/min
0.1%/min
0.01%/min
1%/min
0.1%/min
0.01%/min
1%/min
0.1%/min
0.01%/min
1%/min
0.1%/min
0.01%/min
Failure line
M
m
= 1.4
Figure8.22 Comparison between the experimental results and simulation results (OC
clay, 0.1%/min).
300 Computational modeling of multiphase geomaterials
It is seen in Figures 8.21b and 8.22b that the effective stress paths for
normally consolidated clay exhibit negative dilatancy, that is, a decrease in
mean effective stress, whereas those for overconsolidated clay exhibit posi-
tive dilatancy, that is, an increase in mean effective stress.
We can say that the stressstrain curves and the effective stress paths
for normally consolidated clay are well reproduced by the presented analy-
sis method. However, there are some differences between the experiment
and the simulation for both cases. We would say that the differences can
be improved by the inverse analysis technique to accurately determine the
input parameters.
8.9.3.3.2 Distribution of shear strain
Figures 8.21c and 8.22c show the distributions of shear strain, , for the
experiment, whereas Figures 8.21e and 8.22e are those for the simulation.
Figures8.21d and 8.22d are the pictures taken after the tests, and Figures8.21f
and 8.22f indicate the deformed meshes at an axial strain of 20%. In these
fgures, s and f indicate the side surface and the front surface of the specimens,
respectively (see Figure 8.17). To obtain the distributions of shear strain, ,
on the surfaces of the specimens for the simulation results, we used a special
method, which is the same method as that for the experiment.
We can see in the experimental results (Figures 8.21c and 8.22c) that
strain localization starts at an axial strain of 8% and that shear bands are
clearly seen at an axial strain of 12%. In addition, shear bands develop
from the edges of the top and the bottom of the specimens since the friction
force generated between the specimen and the top cap or the pedestal acts
as a trigger of strain localization. As the axial strain becomes large, clear
shear bands appear on the side surface and the front surface, and develop
with increases in the thickness of the shear bands.
As shown in Figures 8.21e and 8.22e, the simulation results can well
reproduce the strain localization behavior observed in the experiment.
Although homogeneous deformations can be seen until an axial strain of
4% is reached, the strain starts to localize at an axial strain of 8%, and
then four or two shear bands appear at an axial strain of 12% and develop
with an increased thickness on both surfaces. The generating process of the
shear bands is simulated well in both cases for normally consolidated clay
and overconsolidated clay.
8.9.3.3.3 Strain localization pattern
In the experiment and the simulation, we can see a deformation pattern
in which two or four shear bands develop from the edges of the top and
the bottom of the specimens. This mode is due to the material instability
induced by the frictional boundary conditions between the clay specimens
Strain localization 301
and the top caps and the pedestals. The two shear bands intercrossing each
other are just like an X, thus, we call it the X mode. Figure8.23 shows
the schematics of the estimated process generating the X mode. The four
generated shear bands fnally develop two clear and thick shear bands. In
the case of the fnite element analysis, we depict the distribution of the
second invariant of accumulated viscoplastic deviatoric strain,
p
, for all
cases. In the distributions, by disregarding smaller values of
p
, we can see
the localized strain, that is, three-dimensional shear bands. Note that
p

is obtained at the Gaussian integration points of the fnite element method
and is different from the shear strain used in the previous fgures.
8.9.3.4 Three-dimensional shear bands
Estimated shear bands for case NC-2 (normally consolidated clay, 0.1%/
min), tested and simulated, are depicted in Figure8.24. In the contour for
the simulation results, the accumulation of the second invariant of visco-
plastic deviatoric strain,
p
, is illustrated if
p
is more than 0.32. Since the
shear bands observed on the front surface are clearer than those on the side
surface, two shear planes are estimated. It is found, however, that the X
mode appears just on the surface and that higher levels of shear strain are
distributed in the center of the specimen.
Experiment
Axial Strain
Simulation
Axial Strain
0.00
0.03
0.06
0.09
x
y
z
0.01
0.18
0.35
0.52
x
y
z
Distributions of Shear Strain
4 Shear bands appear on each surface
Te Upper Shear Bands and the Lower Shear Bands
((1 and 4) and (2 and 3))
Interact with Each Other and
Two Shear Bands Develop
1 2 1 2 1 2
3 4 3 4 3 4
Distribution of the Second Invariant of
Accumulated Viscoplastic Deviatoric Strain
8% 12% 16%
3% 16%
20%
8% 16% 20%
f
Frictional force
2 Shear bands develop
X mode
Figure8.23 Schematics of the estimated process of the X mode for the experiment
and simulation (OC-1, 1%/min).
302 Computational modeling of multiphase geomaterials
8.9.3.5 Effects of the strain rates
8.9.3.5.1 Strain rate sensitivity
It is well known that clay exhibits strain rate sensitivity. Oka et al. (2003)
reported the strain rate sensitivity of Fukakusa clay under undrained triax-
ial compression conditions. Figure8.21a,b and Figure8.22a,b show the rate
sensitivity of the stressstrain relations and the stress paths in both cases
of normally consolidated clay and overconsolidated clay can be observed.
It can be seen that the strain rate sensitivity observed in the experimental
results is smaller than that in the simulation results.
8.9.3.5.2 Strain localization pattern
The shear-strain distributions and the inclination angles of the shear
bands for both the simulation and the experiment with different strain
rates at an axial strain of 20% are shown in Figure8.25. The numerical
simulation very well reproduces the experimental results with respect
not only to the X mode but also to the effects of the strain rates on the
strain localization pattern. Shear bands develop from the top and the
bottom edges in the case of higher strain rates, while those with lower
strain rates develop beneath the top and the bottom edges. Due to this
tendency, the angles of the shear bands become smaller as the strain rate
decreases. If we could assume that the effect of the strain rate is equal
to the effect of permeability, that is, the higher strain rates correspond
to the lower permeability levels, and vice versa, we would then obtain
a different trend from the one in the present study, that is, angles with
higher strain rates would be smaller than those with lower strain rates.
f
0.65
Z
X Y
0.43
0.32
0.00
Simulation
Axial Strain : 20%
p
> 0.32
s
f
Experiment
s
Figure8.24 Estimated three-dimensional shear bands for experiment and simulation
(NC-2, 0.1%/min).
Strain localization 303
Note that the constitutive equation used in Oka, Kimoto, et al. (2002) is
slightly different than the one used in the present study. In addition, the
maximum thickness of the shear bands with lower strain rates is larger
than that of the shear bands with higher strain rates. It should be noted
that these types of behavior are more clearly seen in the case of overcon-
solidated clay.
8.9.3.5.3 Effect of the sample shape
The effect of the sample shape has been studied by Kodaka et al. (2001)
for normally consolidated clay. For a tall specimen with a high height
over width ratio of H/B = 3.0, the bucking type of deformation mode is
predominant for normally consolidated clay. On the contrary, for over-
consolidated clay with H/B = 3.0, the X type of shear bands was experi-
mentally observed and well simulated by the model (Kodaka et al. 2001;
Oka, Kodaka, et al. 2005).
Slow
Fast
(a)
f f
50
53
53
(b)
50
(c)
49
(a)
53
(b)
51
(c)
47
Simulation Experiment
NC-1(1%/min)
NC-2 (0.1%/min)
NC-3 (0.01%/min)
(a)
f f
53
52
52
(b)
46
(c)
44
(a)
53
(b)
(c)
50
Simulation Experiment
OC-1(1%/min)
OC-2 (0.1%/min)
OC-3 (0.01%/min)
Figure8.25 Comparison of distributions of shear strain and inclination angles of shear
bands for specimen B with different strain rates between simulation and
experimental results (axial strain: 20%, front surface). (a) Shear bands develop
from the top edge. (bc) Shear bands develop beneath the top edge.
304 Computational modeling of multiphase geomaterials
8.9.3.5.4 Mesh-size dependency
Figure8.26 indicates the numerical results for the deformation of a quarter
of the specimen using different mesh sizes under the globally undrained
conditions using the same technique for the three-dimensional analysis
mentioned earlier. As is shown in Figure8.26, the effect of the mesh-size
dependency is very weak in these numerical examples.
0.298
0.208
0.119
0.029
0.332
0.231
0.129
0.027
20 40 =
800 elements
10 20 =
200 elements
0.269
0.190
0.111
0.032
400
360
320
280
240
200
A
v
e
r
a
g
e

V
e
r
t
i
c
a
l

S
t
r
e
s
s

(
k
P
a
)
160
120
80
40
0
0 5 10
Average Axial Strain (%)
k = 1.54 10
8
(m/s)
Coefcient of friction: 0.00001
50 elements
200 elements
800 elements
15 20
5 10 =
50 elements
Figure8.26 Average stressstrain relations for different numbers of fnite element mesh.
Strain localization 305
8.10 APPLICATION TO BEARING CAPACITY
AND EARTH PRESSURE PROBLEMS
A series of analyses for the footing of a clay deposit with different micro-
structure parameters was carried out (Siribumrungwong et al. 2004).
The results of the analyses show that strain localization can be predicted
during the loading of a footing on highly structured soil. This strain
localization acts as a slip line and it affects the bearing behavior of the
strip footing. The effects of footing roughness on the failure mechanism
were also discussed.
For a smooth strip footing on clay soil with a constant level of und-
rained shear strength, both the Hill and the Prandtl failure mechanisms
are theoretically possible. For a rough footing, however, the Hill mecha-
nism is not appropriate as it implies horizontal soil movement at the soil-
footing interface. It can be seen in Figures 8.27 and 8.28 that the Hill
mechanism (Hill 1950) is predicted for a smooth footing and the Prandtl
mechanism (1920) is predicted for a rough footing. Through a plasticity
solution, both mechanisms yield the same value of NC = 5.14. However,
a comparison shows that with a large footing displacement, the reaction
= 0, rough = 20, rough = 20, smooth
0.00 0.12 0.24 0.36 0.48 0.60 0.00 0.12 0.24 0.36 0.48 0.60 0.00 0.12 0.24 0.36 0.48 0.60
4 m
6 m
1 m
Figure8.27 Distribution of viscoplastic deviatoric strain at a footing displacement of 10
cm for homogeneous soil cases.
= 0, rough = 20, rough = 20, smooth
0.00 0.02 0.04 0.06 0.08 0.10 0.00 0.02 0.04 0.06 0.08 0.10 0.00 0.02 0.04 0.06 0.08 0.10
4 m
6 m
1 m
Figure8.28 Vectors of incremental nodal displacements at a footing displacement of 10
cm for homogeneous soil cases.
306 Computational modeling of multiphase geomaterials
force predicted for the smooth footing is lower than that for the rough
footing.Figure 8.29 illustrates the behavior of backfll due to the move-
ment of the wall analyzed by the same elastoviscoplastic model for clay
as that mentioned earlier. It is well simulated that the strain localization
depends on the friction of the wall.
8.11 SUMMARY
In the present chapter, the strain localization problem in geomateri-
als was described. First, strain localization phenomenon and related
theoretical problems were reviewed and discussed. Then, the instability
of viscoplastic geomaterials was discussed. The instability of viscoplas-
tic geomaterials is rather complex compared with the rate-independent
case. Namely, the effect of strain rate, permeability, rate sensitivity and
strain softening interrelates each other depending on the boundary con-
ditions. Finally, numerical simulations of viscoplastic geomaterials were
performed and it was shown that strain localization features, such as
the formation of shear bands in the material, can be reproduced well by
the viscoplastic softening model. In addition, it has been proven that the
method is applicable to practical problems such as bearing capacity and
earth pressure.
Smooth wall (A-A2)
Very rough wall (A-B1)
48
44
6 m
12 m
4 m
0.00 0.05 0.11 0.16 0.22 0.27
0.00 0.13 0.25 0.38 0.50 0.63
Figure8.29 Comparison of deformed mesh and the accumulated viscoplastic shear
strain from soil with different friction conditions between the soil and the
wall in the case of active earth pressure.
Strain localization 307
REFERENCES
Adachi, T., and Oka, F. 1982. Constitutive equations for normally consolidated clay
based on elasto-viscoplasticity, Soils and Foundations, 22(4):5570.
Adachi, T., Oka, F., and Mimura, M. 1987a. An elasto-viscoplastic theory for clay fail-
ure, Proc. 8th Asian Reg. Conf. Soil Mech. Found. Eng., Kyoto, 1, JSSMFE:58.
Adachi, T., Oka, F., and Mimura, M. 1987b. Mathematical structure of an overstress
elasto-viscoplastic model for clay, Soils and Foundations, 27(3):3142.
Aifantis, E.C. 1984. On the microstructural origin of certain inelastic models, J. Eng.
Mater. Tech., ASME, 106:326330.
Aifantis, E.C., Oka, F., Yashima, A. and Adachi, T. 1999. Instability of gradient dependent
elasto-viscoplasticity for clay. Int. J. Numer. Anal. Meth. Geomech., 23:973994.
Anand, L. 1983. Plane deformations of ideal granular materials, J. Mech. Phys.
Solids, 31(2):105122.
Anandarajah, A., Rashidi, H., and Arulanandan, K. 1995. Elasto-plastic fnite ele-
ment analysis of a soil-structure system under earthquake excitation, Comput.
Geotech., 17:301325.
Arthur, J.R.F., Dustan, T., Al-Ani, Q.A.J. and Assadi, A. 1977. Plastic deformation
and failure of granular media, Gotechnique, 27(1):5374.
Asaoka, A., Nakano, M., and Noda, T. 1997. Soil-water coupled behavior of heavily
overconsolidated clay near/at critical state, Soils and Foundations, 37(1):1328.
Bauer, E., and Huang, W. 2001. Evolution of polar quantities in a granular Cosserat
material under shearing, Proceedings of 5th International Workshop on
Bifurcation, Localization Theory in Geomechanics, H.B. Mhlhaus, A.V.
Dyskin, and E. Pasternak, eds., Balkema, Rotterdam, 227238.
Benallal, A., and Comi, C. 2002. Material instabilities in inelastic saturated porous
media under dynamic loadings, Int. J. Solids Struct., 39:36933716.
Benallal, A., and Comi, C. 2003. Perturbation growth and localization in fuid-
saturated inelastic porous media under quai-static loadings, J. Mech. Phys.
Solids, 51:851899.
Bigoni, D., and Hueckel, T. 1991. Uniqueness and localizationI, Associative and
non-associative plasticity, Int. J. Solids Struct., 28:197213.
Biot, M.A. 1956. Theory of propagation of elastic waves in a fuid-saturated porous
solid, J. Acoust. Soc. Am., 28(2):168178.
Boehler, J.P., and Sawczuk, A. 1977. On yielding of oriented solids, Acta Mechanica,
27:185206.
Bolzon, G., Schrefer, B.A., and Zienkiewicz, O.C. 1996. Elasto-plastic soil constitutive
laws generalized to partially saturated states, Gotechnique, 46(2):279289.
Borja, R.I. 2004. Computational modeling of deformation bands in granular media,
II Numerical simulations, Meth. Appl. Mech. Eng., 193:26992718.
Borja, R.I., and Aydin, A. 2004. Computational modeling of deformation bands in
granular media, I Geological and mathematical framework, Comput. Meth.
Appl. Mech. Eng., 193:26672698.
Castellanza, R., and Nova, R. 2003. Compaction bands in oedometeric tests on
cemented granular soils, Procedings of International Workshop on Prediction
and Simulation Methods in Geomechanics, Athens, F. Oka, I. Vardoulakis, A.
Murakami, and T. Kodaka, eds., Japanese Geotechnical Society, 3740.
308 Computational modeling of multiphase geomaterials
Chambon, R., Crochepeyre, S., and Desrues, J. 2000. Localization criteria for
non-linear constitutive equations of geomaterials, Mech. Cohesive-Frictional
Mater., 5:6182.
Chowdhury, E.Q., and Nakai, T. 2001. An unconventional approach of elas-
toplastic modeling of soils, In Bifurcation and Localization Theory in
Geomechanics, H.-B. Mhlhaus, A.V. Dyskin, and E. Pasternak, eds.,
Balkema, 193200.
Cleall, P.J., Thomas, H.R., Seetharam, S.C., and Melhuish, T.A. 2004. Proceedings of
9th Numerical Models in Geomechanics, G.N. Pande and S. Pietruszczak, eds.,
Balkema, 311317.
Coulomb, C.A. 1773. Essai sur application des regles des maiximas et minimis a
quelques problems de statique relatifs a larchitecture, Memoires de lAcademie
Royale pres Divers Savants, 7:342382.
Cundall, P.A., and Strack, O.D.L. 1979. A discrete numerical model for granular
assemblies, Gotechnique, 29(1):4765.
Dafalias, Y. 1982. Bounding surface elasto-viscoplasticity for particulate cohesive
media, Proceedings of IUTAM Conference on Deformation and Failure of
Granular Materials, P.A. Vermeer and H.J. Luger, eds., Balkema, 97107.
Daouadji, A., Darve, F., Al Gali, H., Hicher, P.Y., Laouafa, F., Lignon, S., Nico, F.,
etal. 2010. Diffuse failure in geomaterials: Experiments, theory and modeling,
Int. J. Numer. Anal. Meth. Geomech. DOI:10.1002/nag.975.
Darve, F. 1994. Stability and uniqueness in geomaterials constitutive modeling,
Localization and bifurcation theory for soils and rocks, R. Chambon, J.
Desrues and I. Vardoulakis eds., Balkema, Rotterdam: 7388.
Darve, F. 2000. Instabilities in granular materials and application to landslides,
Mech. Cohesive-Frictional Mater., 5(8):627652.
Darve, F. 2001. Modes of rupture in geomaterials, Proceedings of International
Workshop on Deformation of Earth Materials, TC34, ISSMGE, Sendai,
F.Oka,ed., 3950.
de Boer, R. 1996. Theory of Porous Media, Highlight in the Historical Development
and Current State, Springer, Berlin.
De Josselin de Jong, G. 1971. The double sliding, free rotating model for granular
assemblies, Gotechnique, 21(3):155162.
De Josselin de Jong, G. 1977. Mathematical elaboration of the double-sliding, free-
rotating model, Archs Mech., 29:561591.
Desrues, J. 1990. Shear band initiation in granular materials: experiments and the-
ory, Geomaterials Constitutive Equations and Modeling, F. Darve, ed., Elsevier
Applied Science, Oxford, 283310.
Desrues, J., Chambon, R., Mokni, M., and Mazerolle, F. 1996. Void ratio evolution
inside the shear band in triaxial sand specimen studied by computer tomogra-
phy, Gotechnique, 46(3):529546.
Duvaut, G., and Lions, J.L. 1976. Inequalities in Mechanics and Physics,
Springer-Verlag, Berlin.
Earthquake Engineering Committee, JSCE, 2003. Soil liquefaction under level II
earthquake motion subcommittee, Report on the Soil Liquefaction under
Level II Earthquake Motion (by F. Oka, K. Furuya, and R. Uzuoka), 157201
(in Japanese).
Strain localization 309
Ehlers, W. and Volk, W. 1998. On the theoretical and numerical methods in the the-
ory of porous media based on polar and non-polar elasto-plastic solid materi-
als, Int. J. Solids and Structures, 35:45974617.
Finno, R.J., Viggiani, G., Harris, W.W., and Mooney, M.A. 1998. Localization of
strains in plane strain compression of sand, Proceedings of 4th International
Workshop on Localization and Bifurcation Theory for Soils and Rocks, Gifu,
T. Adachi, F. Oka and A. Yashima, eds., Balkema, Rotterdam, 249258.
Frantziskonis, G. 1993. Crack pattern related universal constants, Probabilities and
Materials, Tests, Models and Applications, D. Breysse, ed., Kluwer Academic,
New York, 361376.
Gudehus, G., and Nbel, K. 2004. Evolution of shear bands in sand, Gotechnique,
54(3):187201.
Hashiguchi, K. 1980. Constitutive equations of elastoplastic materials with elastic-
plastic transition, J. Appl. Mech., ASME, 47:266272.
Hashiguchi, K. 1989. Subloading surface model in unconventional plasticity, Int.
J. Solids Struct., 25:917945.
Hashiguchi, K., and Tsutsumi, S. 2003. Shear band formation analysis in soils by the
subloading surface model with tangential stress rate effect, Int. J. Plasticity,
19:16511677.
Higo, Y. 2003. Instability and strain localization analysis of water-saturated clay by
elasto-viscoplastic constitutive models, PhD thesis, Kyoto University.
Higo, Y., Oka, F., Jiang, M., and Fujita, Y. 2005. Effects of transport of pore water
and material heterogeneity on strain localization of fuid-saturated gradient-
dependent viscoplastic geomaterial, Int. J. Numer. Anal. Meth. Geomech.,
29:495523.
Higo, Y., Oka, F., Kodaka, T., and Kimoto, S. 2006. Three-dimensional strain local-
ization of water-saturated clay and numerical simulation using an elasto-
viscoplastic model, Philosoph. Mag., 86(21)(22):32053240.
Hill, R. 1950. The Mathematical Theory of Plasticity, Oxford Clarendon Press.
Hill, R. 1957. On uniqueness and stability in the theory of fnite elastic strain,
J.Mech. Phys. Solids, 5:229241.
Hill, R. 1958. A general theory of uniqueness and stability in elastic-plastic solids,
J.Mech. Phys. Solids, 6:239249.
Hill, R. 1962. Acceleration waves in solids. J. Mech. Phys. Solids, 10:116.
Hill, R., and Hutchinson, J.W. 1975. Bifurcation phenomena in the plane strain test,
J. Mech. Phys. Solids, 23:239264.
Hughes, T.J.R., and Taylor, R.J. 1978. Unconditionally stable algorithms for quasi-
static elasto-viscoplastic fnite element analysis. Comput. Struct., 8:169173.
Imposimato, S., and Nova, R. 1998. Instability of loose sand specimens in undrained
tests, Proceedings of International Workshop on Localization and Bifurcation
Theory for Soils and Rocks, T. Adachi, F. Oka, and A. Yashima, eds., Balkema,
313322.
Issen, K.A., and Rudnicki, J.W. 2000. Conditions for compaction bands in porous
rock, J. Geophys. Res., B9, 105(21):529521, 536.
Khojastehpour, M., and Hashiguci, K. 2004. Plane strain bifurcation analysis
of soils by the tangential-subloading surface model, Int. J. Solids Struct.,
41:55415563.
310 Computational modeling of multiphase geomaterials
Khojastehpour, M. and Hashiguchi, K. 2004. Axisymmetric bifurcation analysis
in soils by the tangential-subloading surface model, J. Mech. Phys. Solids,
52:22352261.
Kimoto, S. 2002. Constitutive models for geomaterials considering structural
changes and anisotropy, PhD thesis, Kyoto University.
Kimoto, S., and Oka, F. 2004. Numerical study on the effects of strain rate and
structure on the consolidation process of soft clay deposits, Proceedings of
International Symposium on Engineering Practice and Performance of Soft
Deposits, T. Matsui, Y. Tanaka, and M. Mimura, eds., IS-Osaka, Japanese
Geotechnical Society, 105108.
Kimoto, S., and Oka, F. 2005. An elasto-viscoplastic model for clay considering
destructuration and consolidation analysis of unstable behavior, Soils and
Foundations, 45(2):2942.
Kimoto, S., Oka, F., and Higo, Y. 2004. Strain localization analysis of elasto-vis-
coplastic soil considering structural degradation, Int. J. Comput. Meth. Appl.
Mech. Eng., 193(2729):28452866.
Kishino, Y. 1988. Discrete model analysis of granular media, in Poromechanics and
Granular Media, M. Satake and J.T. Jenkins, eds., Elsevier, 285292.
Kodaka, T., Higo, Y., and Takyu, T. 2001. Deformation and failure characteristics
of rectangular clay specimens under three-dimensional condition, Proc. 15th
ICSMGE, Istanbul, 1:167170.
Kolymbas, D., and Rombach, G. 1989. Shear band formation in generalized hypo-
plasticity, Ingenieur-Archive, 59:177186.
Lade, P.V. 1992. Static instability and liquefaction on loose fne sandy slopes,
J.Geotech. Eng., ASCE, 118(1):5171.
Lade, P.V. 2003. Analysis and prediction of shear banding under 3D conditions in
granular materials, Soils and Foundations, 43(4):161172.
Leroy, Y. 1991. Linear stability analysis of rate dependent discrete systems, Int.
J.Solids Struct., 27(6):783808.
Loret, B., and Prvost, J.H. 1991. Dynamic strain localization in fuid-saturated
porous media. J. Eng. Mech., ASCE, 117(4):907922.
Mandel, J. 1964. Conditions de stabilite et postulat de Drucker, Proceedings of
IUTAM Symposium on Rheology and Soil Mechanics, J. Kravtchenko and
P.M. Sirieys, eds., Springer, Berlin: 5867.
Matsuoka, H., and Nakai, T. 1974. Stress-deformation and strength characteristics
of soil under three different principal stresses, Proc. JSCE, 232:5970.
Matsuoka, H., and Nakai, T. 1985. Relationship among Tresca, Mieses, Mohr-
Coulomb and Matsuoka-Nakai failure criterion, Soils and Foundations,
25(4):123128.
Mehrabadi, M.M., and Cowin, S.C. 1978. Initial planar deformation of dilatant
granular materials, J. Mech. Phys. Solids, 26:269284.
Mollema, P.M., and Antonellini, M.A. 1996. Compaction bands: A structural analog
for anti-mode I cracks in aeolian sandstone, Techtonophysics, 267:209228.
Mhlhaus H.-B. 1986. Shear band analysis in granular materials by Cosserat theory,
Ing. Archiv., 56:389399.
Mhlhaus, H.-B., and Aifantis, E.C. 1991. A variational principle for gradient plas-
ticity, Int. J. Solids Struct., 28(7):845857.
Strain localization 311
Mhlhaus, H.-B., and Oka, F. 1995. A continuum theory for random packings of
elastic spheres, Fracture of Brittle and Disordered Materials, Proceedings of
IUTAM Symposium on Fracture of Brittle and Disordered Materials, Concrete,
Rocks and Ceramics, Brisbane, Australia, G. Baker and B.L. Karihaloo eds.,
E& FN Spon, 285298.
Mhlhaus, H.-B., and Oka, F. 1996. Dispersion and wave propagation in dis-
crete and continuous models for granular materials. Int. J. Solids Struct.,
33(19):28412858.
Mhlhaus, H.-B., and Vardoulakis, I. 1987. The thickness of shear bands in granular
materials, Gotechnique, 37:271283.
Muir Wood, D. 2002. Some observations of volumetric instabilities in soils, Int.
J.Solids Struct., 39:34293449.
Nakai, T., and Mihara, Y. 1984. A new mechanical quantity for soils and its
application to elasto-plastic constitutive models, Soils and Foundations,
24(2):8294.
Nemat-Nasser, S. 1983. On fnite plastic fow of crystalline solids and geomaterials,
Trans. ASME, J. Appl. Mech., 50:11141126.
Nova, R. 1989. Liquefaction, stability, bifurcation of soil via strain-hardening plas-
ticity, Numerical Methods for the Liquefaction and Bifurcation of Granular
bodies: Proceedings of the International Workshop, Gdansk, E. Dembicki,
G.Gudehus, and Z. Sikora, eds., 117132.
Nova, R. 1994. Controllability of the incremental response of soil specimens sub-
jected to arbitrary loading programmes, J. Mech. Behav. Mater., 5(2):193201.
Nova, R. 2009. Controllability of geotechnical tests and their relationship to the
instability of soils, in Micromechanics of Failure in Granular Geomaterials,
F.Nicot and R. Wan, eds., ISTE Wiley, London, 134.
Oda, M. and Kazama, H. 1998. Micro-structure of shear band and its relation
to the mechanism of dilatancy and failure of granular soils, Gotechnique,
48(1):117.
Oka, F. 1981. Prediction of time-dependent behaviour of clay, Proc. 10th Int. Conf.
Soil Mech. Found. Eng., Stockholm, Balkema, Rotterdam, 1:215218.
Oka, F. 1993a. An elasto-viscoplastic constitutive model for clay using a transformed
stress tensor, Mech. of Mater., 16:4753.
Oka, F. 1993b. Anisotropic and pseudo-anisotropic elasto-viscoplastic constitutive
models for clay, Modern Approaches to Plasticity, Proceedings of a Workshop
held in Horton, D. Kolymbas, ed., Elsevier Science, 505526.
Oka, F. 2000. Elasto-viscoplastic constitutive models of geomaterials, Morikita
Shuppan, Japan 103 (in Japanese).
Oka, F. 2005. Computational modeling of large deformations and the failure of
geomaterials, Proc. 16th ICSMGE, Millpress Science, Rotterdam, 1:95122.
Oka, F., Adachi, T., and Yashima, A. 1994. Instability of an elasto-viscoplastic consti-
tutive model for clay and strain localization, Mech. Mater., 18:119129.
Oka, F., Adachi, T., and Yashima, A. 1995. A strain localization analysis using a vis-
coplastic softening model for clay, Int. J. Plasticity, 11(5):523545.
Oka, F., Adachi, T., Yashima, A., and Chu, L.L. 1994. Strain localization analysis by
elasto-viscoplastic softening model for frozen sand, Int. J. Numer. Anal. Meth.
Geomech., 18:813832.
312 Computational modeling of multiphase geomaterials
Oka, F., Higo, Y., and Kimoto, S. 2002a. Effect of dilatancy on the strain local-
ization of water-saturated elasto-viscoplastic soil, Int. J. Solids Struct.,
39(1314):36253647.
Oka, F., Jiang, M., and Higo, Y. 1999. Effect of transport of water on strain local-
ization analysis of fuid-saturated strain gradient dependent viscoplastic
geomaterial, Proceedings of 5th International Workshop on Bifurcation and
Localization, Perth, Australia, H.-B. Mhlhaus, A.V. Dyskin, and E. Pasternak,
eds., Swetst Zeitlinger, B.V. Lisse, 7783.
Oka, F., Kimoto, S., and Adachi, T. 2005. Calibration of elasto-viscoplastic models
for cohesive soils, Proceedings of 11th IACMAG, G. Barla and M. Barla, eds.,
AGI, Patron Editore, Bologna, 1:449456.
Oka, F., Kimoto, S., Higo, Y., Ohta, H. Sanagawa, T., and Kodaka, T. 2011. An
elasto-viscoplastic model for diatomaceous mudstone and a numerical
simulation of compaction bands, Int. J. Numer. Anal. Meth. Geomech.,
35(2):244263.
Oka, F., Kimoto, S., Kobayashi, H., and Adachi, T. 2002b. Anisotropic behavior
of soft sedimentary rock and a constitutive model, Soils and Foundations,
42(5):5970.
Oka, F., Kodaka, T., Kimoto, S., Ichinose, T., and Higo, Y. 2005. Strain localization
of rectangular clay specimen under undrained triaxial compression conditions,
Proc. 16th ICSMGE, Osaka, Millpress Science, Rotterdam, 2:841844.
Oka, F., Kodaka, T., Kimoto, S., Ishigaki, S., and Tsuji, C. 2003. Step-changed strain
rate effect on the stress-strain relations of clay and a constitutive modeling,
Soils and Foundations, 43(4):189202.
Oka, F., Muhlhaus, H.B., Yashima, A., and Sawada, M. 1998. Quasi-static and
dynamic characteristics of strain gradient dependent non-local constitutive
models, in Material Instabilities in Solids, R. de Borst and E. Van der Giessen,
eds., John Wiley & Sons, Chichester, 387404.
Oka, F., Yashima, A., and Kohara, I. 1992. A fnite element analysis of clay foun-
dation based on fnite elasto-viscoplasticity, Proceedings of 4th International
Symposium on Numerical Models in Geomechanics, Swansea, UK, G.N.
Pande, and S. Pietruszczak, eds., Balkema, Rotterdam, 915922.
Oka, F., Yashima, A., Sawada, K., and Aifantis, E.C. 2000. Instability of gradient-
dependent elastoviscoplastic model for clay and strain localization, Comput.
Meth. Appl. Mechan. Eng., 183:6786.
Oliver, J., Cervera, M., and Manzoli, O. 1999. Strong discontinuities and con-
tinuum plasticity models: the strong discontinuity approach. Int. J. Plasticity,
15:319351.
Olsson, W.A. 1999. Theoretical and experimental investigation of compaction bands
in porous rock, J. Geophys. Res., 104, B4:72197228.
Ostowski, A. and Taussky, O. 1951. On the variation of the determinant of a positive
defnite matrix, Neder. Akadem. Wet. Proc. A54:333351.
Palmer, A.C., and Rice, J. 1973. The growth of slip surfaces in the progressive failure
of overconsolidated clay, Proc. Roy. Soc. London, A269:500527.
Papamichos, E., and Vardoulakis, I. 1995. Shear band formation in sand according
to non-coaxial plasticity model, Gotechnique, 45(4):649661.
Peirce, D., Shih, C., and Needleman, A. 1984. A tangent modulus method for rate
dependent solids, Comput. Struct., 18(5):845887.
Strain localization 313
Perrin, G., and Leblond, J.B. 1993. Rudnicki and Rices analysis of strain localization
revisited, J. Appl. Mech., 60:842846.
Perzyna, P. 1963. The constitutive equations for work-hardening and rate sensitive
plastic materials, Proc. Vib. Probl., Warsaw, 4(3):281290.
Prandtl, L. 1920. ber die Harte plashecher Krper, Nachr. Math-phys. KL.
Gottingen, 74.
Regueiro, R., and Borja, R. 2001. Plane strain fnite element analysis of pres-
sure sensitive plasticity with strong discontinuity, Int. J. Solids Struct.,
38:36473672.
Rice, J.R. 1975. On the stability of dilatant hardening for saturated rock masses,
J.Geophys. Res., 80(11):15311536.
Rice, J.R. 1976. The localization of plastic deformation, Proceedings of 14th
International Congress on Theoretical and Applied Mechanics, W.T. Koiter,
ed., North-Holland, 207220.
Roscoe, K.H. 1970. The infuence of strains in soil mechanics, Gotechnique,
20(2):129170.
Roscoe, K.H., and Burland, J.B. 1968. On the generalised stress-strain behaviour
of wet clay, In Engineering Plasticity, J. Heyman and F.A. Leckie, eds.,
Cambridge, 535609.
Roscoe, K.H., Schofeld, A.N., and Thrairajah, A. 1963. Yielding of clays in states
wetter than critical, Gotechnique, 13(3):211240.
Rudnicki, J.W, and Rice, J.R. 1975. Condition for the localization of deformation in
pressure-sensitive dilatant material, J. Mech. Phys. Solids, 23:371394.
Rudnicki, J.W., and Olsson, W.A. 1998. Reexamination of fault angles predicted by
shear localization theory, Int. J. Rock Mech. Min. Sci., 35(4/5):512513.
Sawada, K., Yashima, A. and Oka, F. 1999. Numerical analysis of deformation of
saturated clay based on elasto-viscoplastic Cosserat type model, Proc. 5th Int.
Workshop on Bifurcation and Localization, Perth, Australia, H.-B. Mhlhaus,
A.V. Dyskin, and E. Pasternak eds., Swets & Zeitlinger B.V., Lisse: 155160.
Sawada, K., Yashima, A., and Oka, F. 2001. Numerical analysis of deformation of
saturated clay based on Cosserat type elasto-viscoplastic model, J. Soc. Mater.
Sci. Japan, 50, 6:585592 (in Japanese).
Schrefer, B.A., Majorana, C.E., and Sanavia, L. 1995. Shear band localization in
saturated porous media. Arch. Mech., 47(3):577599.
Schrefer, B.A., Sanavia, L., and Majorana, C.E. 1996. A multiphase medium
model for localization and postlocalization simulation in geomaterials. Mech.
Cohesive-Frictional Mater., 1:95114.
Scott, R.F. 1987. Failure, Gotechnique, 37(4):423466.
Siribumrungwong, B. 2005. Elasto-viscoplastic numerical analysis of soil-founda-
tion interaction in soft clay ground, PhD thesis, Kyoto University.
Siribumrungwong, B., Oka, F., and Kimoto, S. 2004. Numerical study of the effect
of degradation on bearing capacity of clay using an elasto-viscoplastic model,
Proc. 17th KKCNN Symp. Civil Eng., Thailand, 589594.
Siribumrungwong, B., Oka, F., Kimoto, S., Kodaka, T., and Higo, Y. 2007. Elasto-
viscoplastic fnite element study of the effect of degradation on bearing capac-
ity of footing on clay ground, Geomech. Geoeng., 2(4):235251.
Sokolovsky, V.V. 1942. Statics of earthy media (Russian), Moscow, Izdatelstiv
Akademii Nauk.
314 Computational modeling of multiphase geomaterials
Spencer, A.J.M. 1964. A theory of the kinematics of ideal soils under plane strain
conditions, J. Mech. Phys. Solids, 12:337351.
Spencer, A.J.M. 1987. Isotropic polynominal invariants and tensor functions,
Applications of tensor functions in solid mechanics, J.P. Boehler, ed., Springer,
141164.
Tamaganini, R., Viggiani, G., and Chambon, R. 2001. Some remarks on shear
band analysis in Hypoplasticity, Bifurcation and localization theory in geo-
mechanics, H.B. Mhlhaus, A.V. Dyskin, and E. Pasternak eds., Balkema,
Rotterdam, 8593.
Tatsuoka, F., Nakamura, S., Huang, C.C., and Tani, K. 1990. Strength anisotropy
and shear band direction in plane strain tests of sands, Soils and Foundations,
30(1):3554.
Taylor, D.W. 1948. Fundamentals of Soil Mechanics, New York: John Wiley & Sons.
TC34 of ISSMGE. 2005. The state-of-the-art report, Prediction methods in large
strain geomechanics, Proceedings of 16th ICSMGE, Osaka, JGS, Tokyo.
Teichman, J., and Wu, W. 1993. Numerical study of patterning of shear bands in a
Cosserat continuum, Acta Mechanica, 69:6174.
Thornton, C. 1998. Numerical simulations of deviatoric shear deformation of gran-
ular media, Gotechnique, 50(2):4353.
Tokuoka, T. 1971. Yield conditions and fow rules derived from hypo-elasticity,
Arch. Rat. Mech. Anal., 42:239252.
Valdoulakis, I. 1977. Scherfugenbildung in Sandkrpern als Verzweiguns problem,
Dissertation, Universitt Karlsruhe, Verffentlichungen IBF, Heft Nr. 70.
Vardoulakis, I. 1980. Shear band inclination and shear modulus of sand in biaxial
tests, Int. J. Numer. Anal. Meth. Geomech., 4:103119.
Vardoulakis, I. 1996a. Deformation of water-saturated sand: I uniform undrained
deformation and shear banding, Gotechnique, 46(3):441456.
Vardoulakis, I. 1996b. Deformation of water-saturated sand: II effect of pore water
fow and shear banding, Gotechnique, 46(3):457472.
Vardoulakis, I. 2002. Dynamic thermo-poro-mechanical analysis of catastrophic
landslides, Gotechnique, 52(3):157171.
Vardoulakis, I., and Aifantis, E.C. 1991. A gradient fow theory of plasticity for
granular materials, Acta. Mech., 87:197217.
Vardoulakis, I., and Sulem, J. 1995. Bifurcation Analysis in Geomechanics,
Glasgow, Blackie.
Vermeer, P. 1982. A simple shear band analysis using compliance, Proceedings of
IUTAM Conference on Deformation and Failure of Granular Materials, P.A.
Vermeer and H.J. Luger, eds., Balkema, Rotterdam, 493499.
Wan, R.G., Chan, D.H., and Morgenstern, N.R. 1990. A fnite element method for
the analysis of shear bands in geomaterials, Finite Elements Anal. Design,
7(2):129143.
Wang, R., and Guo, P.J. 1997. Strain localization of granular materials with spe-
cial reference to Barotropy and Pyknotropy, Proceedings 4th Interanational
Symposium on Deformation and Progressive Failure in Geomechanics, A.
Asaoka, T. Adachi, and F. Oka, eds, Pergamon, 4550.
Yashima, A., Leroueil, S., Oka, F., and Guntoro, I. 1998. Modelling temperature and
strain rate dependent behavior of clays: One-dimensional consolidation, Soils
and Foundations, 38(2):6373.
Strain localization 315
Yatomi, C., Yashima, A., Iizuka, A., and Sano, I. 1989. General theory of shear
bands formation by a non-coaxial Cam-clay model. Soils and Foundations,
29(3):4153.
Zhang, H.W., Sanavia, L., and Schrefer, B.A. 1999. An internal length scale in
dynamic strain localization of multiphase porous media. Mech. Cohesive-
Frictional Mater., 4:443460.
Zhang, H.W., and Schrefer, B.A. 2000. Gradient-dependent plasticity model
and dynamic strain localization analysis of saturated and partially satu-
rated porous media: One dimensional model, Eur. J. Mechs. A/Solids,
19(3):503524.
Zhang, H.W., and Schrefer, B.A. 2001. Uniqueness and localization analysis of
elasto-plastic saturated porous media, Int. J. Num. Anal. Meth. Geomech.,
25:2948.
Zienkiewicz, O.C., Humpheson, C., and Lewis, R.W. 1975. Associated and non-
associated viscoplasticity and plasticity in soil mechanics, Gotechnique,
25(4):671689.
317
Chapter 9
Liquefaction analysis
of sandy ground
9.1 INTRODUCTION
The aim of this chapter is to present a cyclic elastoplastic constitutive model
for sand and its application to an effective-stress-based fully coupled lique-
faction analysis. The model is a cyclic elastoplastic model for sand based on
the nonlinear kinematical hardening theory considering the strain depen-
dency of the plastic modulus (Oka et al. 1999). The model has already
been applied to liquefaction analyses, in particular, after the 1995 Hyogo-
ken Nambu earthquake in Japan. First, an outline of the model for sand
is described. Second, a possible modifcation for the strain dependency of
the plastic modulus is presented. Finally, the numerical application of the
model to a liquefaction analysis of a ground and a groundstructure system
is shown.
9.2 CYCLIC CONSTITUTIVE MODELS
During the past three decades, many elastoplastic constitutive models have
been proposed. The elastoplastic theory is known as a representative tool
for modeling the nonlinear and hysteretic behavior of soil. In the realm of
soil mechanics, an associated fow rule has been used to model the behavior
of soil, in particular, soil with negative dilatancy such as normally consoli-
dated clay. In contrast, the nonassociated fow rule is frequently adopted to
model sand with positivenegative dilatancy. One of the well-known mod-
els for soil is the DruckerPrager model (1952). The DruckerPrager model
includes a dependency of the mean effective stress. There are, however,
several shortcomings, namely, an unreasonable amount of dilatancy. In the
1970s, DiMaggio and Sandler (1971) proposed an elastoplastic model for
sand called the cap model to generalize the DruckerPrager model. The
model contains two yield functions, the frst function being a modifed
DruckerPrager model and the second function intersecting the mean
318 Computational modeling of multiphase geomaterials
effective stress axis in the space of a second invariant of the deviatoric
stress tensor and the mean effective stress. This model improved the behav-
ior of the soil in the process with an increase in the mean effective stress.
Although this cap model is a simple model, it has been used for a vari-
ety of geological materials, including rocks. Nishi and Esashi (1978) and
Vermeer (1978) proposed a double-hardening model based on the nonasso-
ciated fow rule. These models are consistent with the experimental results
obtained by Poorooshasb et al. (1966).
A number of constitutive models for sand have been proposed and then
applied to design practice with appropriate numerical methods, such as
the fnite element method. In order to analyze the dynamic behavior of a
liquefable ground, it is necessary to predict the cyclic behavior of satu-
rated sand.
For the cyclic behavior of sand, many models have been proposed, includ-
ing those from Ghaboussi and Momen (1979), Pastor and Zienkiewicz
(1986), Nishi and Kanatani (1990), Ishihara and Kabilamany (1990), and
Prvost and Keane (1990). Others have proposed cyclic elastoplastic mod-
els for sand. They have tried to incorporate their models into computer
programs for two-dimensional liquefaction analyses. Hashiguchi and Ueno
(1977) and Hashiguchi (1980) proposed a cyclic plasticity model called a
subloading surface model, which considers the expansion and the contrac-
tion of the loading surface.
One of the authors proposed a constitutive model based on the rota-
tional hardening rule (Adachi and Oka 1982; Oka 1982) that can incor-
porate the Masing rule (Masing 1926). This model was incorporated into
a two-dimensional liquefaction analysis and some practical problems were
solved by it (Oka, Yashima, Kato, et al. 1994). Furthermore, the nonlinear
kinematic hardening rule originally proposed by Armstrong and Frederick
(1966) for metal, and later modifed by Chaboche and Rousselier (1983),
was introduced (Oka et al. 1992) into the rotational hardening model
(Oka 1982) for generalizing kinematic hardening. This nonlinear kine-
matic hardening rule is described by an evolutional differential equation
that is a generalized Pragers kinematic hardening rule and can take into
account initial anisotropy as well as the Bauschinger effect. The model by
Oka et al. (1992) is based on the nonlinear kinematic hardening rule and
has shown a better performance for describing the behavior of sand under
cyclic loading than the model based on the simple rotational hardening
model (Oka and Washizu 1981). It has already been applied to several
types of liquefaction problems.
The effectiveness of the cement mixing column method and the gravel
drain method as countermeasures against liquefaction has been clarifed
by numerical work and a two-dimensional liquefaction analysis (Shibata
et al. 1992; Kato et al. 1994). The liquefaction of a seabed due to ocean
Liquefaction analysis of sandy ground 319
waves (Oka et al. 1993) and the seepage failure of loose sand layers (Oka,
Yashima, Shibata, et al. 1994) were also studied with the original kine-
matic hardening model. In the model proposed by Taguchi et al. (1995,
1997a,b) and Oka et al. (1999) a generalized fow rule was incorporated
to accurately describe the cyclic behavior of sand and the liquefaction
strength curve that relates the number of loading cycles to the shear stress
amplitude.
9.3 CYCLIC ELASTOPLASTIC MODEL FOR SAND
WITH A GENERALIZED FLOW RULE
Oka et al. (1999) developed an elastoplastic constitutive model based on
the nonlinear kinematic hardening rule. Several modifcations to the origi-
nal kinematic hardening model (Oka et al. 1992) were proposed based on
a comparison between the experimental results and the numerical predic-
tions by the model. The frst feature of the model is the fow rule and the
second feature is the introduction of the cumulative plastic strain depen-
dence of the plastic shear modulus. The third feature is the introduction
of a fading memory of the initial anisotropy. The nonlinear kinematic
hardening variables are used in both the yield function and the plastic
potential function.
9.3.1 Basic assumptions
The basic assumptions are taken as follows:
Infnitesimal strain theory
Elastoplastic theory
Nonassociated fow rule
Overconsolidated boundary surface
Nonlinear kinematic hardening rule
9.3.2 Overconsolidation boundary surface
An overconsolidation boundary surface is adopted that distinguishes the
overconsolidated region from the normally consolidated region. The over-
consolidation boundary surface, f
b
= 0, is defned as

f M ln
b m
m
mb
0
(0)
=

+

(9.1)
320 Computational modeling of multiphase geomaterials

( )
( ) ( )
0
1 2
0 0

/

=

ij ij ij ij
(9.2)


ij ij m
s

= / (9.3)
where
m
is the mean effective stress, s
ij
is the deviatoric stress tensor,
M
m

is the value of the stress ratio expressed by

ij ij

when the maximum
volumetric strain during shearing takes place and that can be called the
phase transformation stress ratio, and
ij( ) 0

denotes the value of

ij

at the
end of consolidation.
The condition f
b
< 0 means that the stress state stays in an overconsoli-
dated (OC) region, whereas f
b
0 means that the stress state stays in a
normally consolidated (NC) region. Herein,
mb
in Equation (9.1) is given
as follows:

e
v
mb mbi
P
exp
1
=
+

(9.4)
where
mbi
is the initial value of
mb
, is the swelling index, is the com-
pression index, e is the void ratio, and v
P
is the plastic volumetric strain.

mbi
is determined by considering the volume change characteristics of
sand, although it is usually thought that for isotropically consolidated soils,

mbi
is equal to the preconsolidation pressure in the context of the conven-
tional concept of overconsolidation.
mbi
of the sand samples is not always
equal to
m0
(mean effective stress at the end of consolidation) due to the
material anisotropy, the method of sample preparation, the degree of com-
paction, aging, and so forth. OCR
mbi m

= /
0
should be called the quasi-
overconsolidation ratio (quasi-OCR). Furthermore,
mc
, which is the mean
effective stress at the intersection of the overconsolidated boundary surface
and the
m
axis, is defned as

M
mc mb
m
exp
(0)
=

.

(9.5)
The overconsolidated boundary surface, that is, the surface of
isotropically consolidated sand in the 2
2
J
m
plane, is shown in
Figure9.1. J
2
is the second invariant of deviatoric stress tensor, s
ij
, that
is, J s s
ij ij 2
1
2
= .
Liquefaction analysis of sandy ground 321
9.3.3 Fading memory of the initial anisotropy
During cyclic loading in soil, the effect of the initial anisotropy decreases.
In the original model, the overconsolidation boundary surface depends on
the initial anisotropy of the soil, as shown in Equation (9.5), so that the
existence of the initial anisotropy infuences the shape of the overconsoli-
dation boundary surface. Herein, the initial anisotropy is assumed to fade
during cyclic loading in soil. To take this into account, the following coef-
fcient,
=

( )
P
, is used in Equation (9.5), that is,

M
mc mb
m
exp
(0)
=

(9.6)

C C d
d
P
d
P*

= =

(9.7)

d de de
P
ij
P
ij
P

/
=
1 2
(9.8)
in which,
P
is the cumulative or accumulated plastic shear strain from
the initial condition, de
ij
P
is the plastic deviatoric strain increment tensor,
and C
d
is a constant that controls the rate of disappearance of anisotropy.
When sand is initially at the isotropic state, Equation (9.6) is equivalent to
Equation (9.5) because the value of
( ) 0

is zero.
M*
f
b
= 0
f
b
= 0
g = 0
g = 0
f
M*
m
M*
*
(0)
~
~
2
3
x
x
2
3
q

ma

mb

mc

m
x + M*
m
ln (
m
/
mb
) = 0
2
3
+ M*
m
ln (
m
/
mb
) = 0
q

m
q
0

m0
2
3
q
q =
1

3
x = kinematic hardening parameter
under triaxial conditions
Figure 9.1 Plastic potential and overconsolidation boundary surface.
322 Computational modeling of multiphase geomaterials
9.3.4 Yield function
The yield function for changes in stress ratio
f
y1
is denoted as

f
y ij ij ij ij 1
1 2
=
/

= k 0 (9.9)
where k is a numerical parameter that controls the size of the elastic region
and

ij

is the nonlinear kinematic hardening parameter.

ij

has the same

dimensions as the stress ratio
ij

and is the so-called back stress parameter.

The evolution equation for the hardening parameter is defned by Equation
(9.10), namely,

d B A de d
ij ij
P
ij
P


= (9.10)
in which A

and B

are material parameters. A

and B

are related to the

stress ratio at failure, M
f

, and the initial plastic shear modulus normalized

with the mean effective stress, G
P
, respectively, as follows:

A M B
G
M
f
P
f
= , = .

(9.11)
In general, parameter B

can be considered to follow the evolution

equation as

dB C B B d
f
P
= ( )
1
(9.12)
in which C
f
and B
1

are material parameters.

ij
P
d

in Equation (9.10)
is the nonlinear term that depends on the magnitude of the plastic
shear strain increment. If
ij
P
d

is negligible, Equation (9.10) will lead
to the well-known Pragers linear kinematical hardening rule given by
d B A de G de
ij ij
P P
ij
P

= = .
In the following Equation (9.13), the accumulated plastic shear strain
between two sequential stress reversal points in the previous cycle is given
by
( ) ( ) n
P
n
P
d

= .
Cycle n can be easily indebted. In reality, a half cycle is defned as
the process between two sequential stress reversal points. The stress
reversal point is judged by the change in the sign of df
y1
, increment of
f
y1
in Equation (9.9), in the general stress cognition. This automatic
Liquefaction analysis of sandy ground 323
identifcation of a cycle is an advantage of using the nonlinear kinematic
hardening rule. G
P
decreases with an increase in
( ) n
P
(Oka et al. 1999).
After the stress state earnest cyclic mobility, the rate of decrease in G
P
,
due to the accumulation of plastic strain, is accelerated. Therefore, a
decrease in G
P
with an increase in
( ) n
P
is introduced as

G G G C G
P P P
f n
P P
= +

( )exp( )
max min ( ) min

(9.13)


( )
( )
n
P
n
P
d

=

(9.14)
where G
P
max
is the initial value for G
P
, G
P
min
is the lower limit of G
P
, and
C
f
is a constant. The regression curves are found to be in good agreement
with the experimental results shown by the symbols. Using Equation (9.13),
three parameters, namely, G G
P P
max min
, , and C
f
, can be identifed in each cycle.

( )max n
P
is the maximum accumulated plastic shear strain between sequential
stress reversal points in the previous nth cycle,
( )max n
P
, after entering into a
condition of cyclic mobility. Herein, the onset of the cyclic mobility condi-
tion is judged when stress ratio

reaches M
m

. It is found that during cyclic

mobility, the value of G
P
max
decreases with an increase in
( )max n
P
. The remain-
ing parameters, G
P
min
and C
f
, are also generally found to be dependent on

( )max n
P
. In the modeling, however, only G
P
max
is assumed to be dependent on

( )max n
P
. The two remaining parameters are assumed not to be dependent
on

( )max n
P
, because G
P
max
was found to have a stronger effect on the change
in G
P
than the other two parameters. The relationship between G
P
max
and

( )max n
P
is assumed as follows:

G
G
P
P
n
P
n r
P
max
max
( )max ( )
=
+ /

0
1
(9.15)
where G
P
max0
is the initial value of G
P
max
and
( ) n r
P
is the reference value of

( )max
*
n
P
when the value of G
P
max
is the half value of G
P
max0
. The lower limit
of G
P
max
is G
P
min
. The same type of equation is also applied to model the
reduction in the elastic shear modulus in which
( ) n r
E
is used as a reference
value instead of
( ) n r
P
.

G
G
E
E
n
E
n r
E
=
+ /

0
1
( )max ( )
(9.16)
324 Computational modeling of multiphase geomaterials
where
( )max n
E
is the maximum accumulated elastic shear between sequen-
tial stress reversal points in past cycles. G
E
0
is the initial value of G
E
, the
elastic shear modulus normalized with the mean effective stress.
Changes in yield function f
y1
in Equation (9.9) are governed by changes
in the stress ratio. As for the analysis under the general stress conditions,
the second yield function, f
y2
, which is described based on changes in the
mean effective stress, should be taken into account (Oka 1992).
9.3.5 Plasticstrain dependence
of the shear modulus
So far, three methods have been proposed by which the plasticstrain
dependence of the plastic shear modulus is described as follows.
9.3.5.1 Method 1
Oka et al. (1993) proposed the plasticstrain dependency of B

as

B B B exp C B
f
P
= + ( ) ( )
0 1 0 1
(9.17)
in which B
1

is the lower bound of B

and
0
P
is the accumulated value of
the second invariant of the deviatoric plastic strain from the initial state.
In method 1, the elastic modulus is assumed to not be dependent on the
plastic strain.
9.3.5.2 Method 2
Oka et al. (1999) proposed a method using Equations (9.13), (9.15), and
(9.16). Method 2 is the most general and the most complicated method
among the three methods. In method 2, B G M
P
f

= / is updated with the
accumulated plastic shear strain between the two sequential stress reversal
points in the previous nth cycle when the stress reverses. Equation (9.13)
can be rewritten with B
*
as

B G M B B C B
P
f f n
P
= / = + ( )exp( )
max ( ) 1 1
(9.18)
Moreover, B
max

is reduced with the maximum accumulated plastic shear

strain between the sequential stress reversal points in the previous nth cycle
after the stress ratio

has reached M
m

. Equation (9.15) can be rewritten

with B
*
as

B G M
B
P
f
n
P
n r
P
max max
( )max ( )

= / =
+ /
0
1
(9.19)
Liquefaction analysis of sandy ground 325
The elastic modulus is also reduced by using Equation (9.16) in a similar
manner as the plastic modulus.
9.3.5.3 Method 3
Uzuoka (2000) proposed a simplifed method 3 in which B

is determined
by the following equation only when the stress ratio has reached the phase
transformation line:

B
B
ap
P
r
P

=
+ /
0
1
(9.20)
in which

ap
p
is the accumulated value of the second invariant of the devia-
toric plastic strain tensor after it has reached the phase transformation line
and
r
P
is the reference strain. The lower limit of B

is given by B
1

. As for
elastic shear modulus G
E
, a similar relation is adopted for updating it in
method 3.
9.3.5.4 Method 4
Oka et al. (2004) proposed a method 4 that is similar to method 2, except
for using Equation (9.20) in place of Equation (9.19). B
*
is updated with
the accumulated plastic shear strain between the two sequential stress
reversal points with Equation (9.18). After the stress path reaches the
phase transformation line, B
*
and the elastic shear modulus decrease with
Equation (9.20).
In general, methods 2 and 4 have a high potential to reproduce the cyclic
behavior of various soils. However, methods 2 and 4 require more parame-
ters than the other methods. From a practical point of view, it is not always
easy to determine many parameters. In contrast, when we adopt method1
or method 3, the soil parameters can be determined more simply. With the
knowledge gained from numerical studies (Oka et al. 2003), it has been
found that method 3 is effective for modeling the plasticstrain depen-
dence of the shear modulus even though the accuracy of the method is low;
method 4 provides an effective method for determining both the plastic
modulus and the elastic modulus.
Changes in yield function f
y1
in Equation (9.9) are governed by changes
in the stress ratio. As for the analysis under general stress conditions, the
second yield function, f
y2
, which is defned on the basis of changes in the
mean effective stress, should be taken into account (Oka 1992), namely,

f M y R
y m
m
m
m d
ln 0
2
0
=

=

(9.21)
326 Computational modeling of multiphase geomaterials
where y
m

is the scalar kinematic hardening parameter,

m0
is the unit value
of the mean effective stress, and R
d
is the scalar variable. Since the strains
brought about by changes in the mean effective stress are small in the over-
consolidated region, it can be assumed that the second static yield function
can be disregarded in the overconsolidated region.
The kinematic hardening parameter y
m
is broken down into y
m1

and
y
m2

. The evolutional equations for these kinematic hardening parameters

are given by

dy dy dy
m m m

= +
1 2
(9.22)

dy B A dv y dv
m
p
m
p
1 2 2 1

= | | ( ) (9.23)

dy H dv
m
p
2 2

= (9.24)
where v
P
kk
P
= .
9.3.6 Plastic potential function
The nonlinear kinematic hardening variable is used in the plastic potential
function as well as in the yield function. The plastic potential function, g,
is denoted as follows:

g
M
ln
ij ij ij ij
m
mk
0
1 2

= +

=
/

(9.25)
where
mk
is a constant and

M is defned by

M
ln
f
M f
m mc
b
m b
( )
(OC region in which 0)
(NC region in which 0)

/
<
.

(9.26)
M
m

is the stress ratio when a maximum contraction of the material

takes place.

M , which is a variable depending on the stress state, con-

trols the direction of the plastic strain increment. When the stress state is
inside the overconsolidated region,

M takes a value that is less than that

of M
m

. In addition, it is assumed that

M becomes equal to M
m

once

M
has reached M
m

. The plastic potential function and the overconsolidated

boundary surface are shown in Figure9.1.
Liquefaction analysis of sandy ground 327
The stressdilatancy characteristic, which is normally derived from
Equation (9.25), sometimes gives a rather steeper slope to the liquefaction
strength curve than that obtained from laboratory tests. To counteract this
shortcoming in the original model, the fow rule is generalized using the
fourth rank isotropic tensor, H
ijkl
(Naghdi and Trapp 1975), as

d H
g
ij
P
ijkl
kl

(9.27)

H a b
ijkl ij kl ik jl il jk
= + + ( )
(9.28)
Namely,

de b
g
s
d a b
g
ij
P
ij
kk
P
m
2 (3 2 ) =

, = +

(9.29)
Desai and Siriwardane (1980) adopted a similar approach to the gen-
eral description of d
ij
P
by introducing a correction factor into the fow
rule. Coeffcients a and b in Equation (9.28) generally depend on the
state parameters, for example, stress and strain. From Equations (9.25),
(9.27), and (9.28), a stressdilatancy characteristic relation of the gener-
alized model can be derived by the generalized fow rule, Equation (9.27),
as follows:

dv
d
D
M
P
P
x

= ( )
(9.30)
where v
P
kk
P
= and

D
a
b

= +
3
2
1 (9.31)
D

is a so-called coeffcient of dilatancy, which controls the proportion

of the plastic deviatoric strain increment to the plastic volumetric strain
increment. In cases where coeffcient a is zero, D

is equal to one and the

stress-dilatancy relation is equivalent to that of the original model. As for
the functional form of D

, two types of functions are used in this study.

The frst one is as follows:

D D const
0
= = .

(9.32)
328 Computational modeling of multiphase geomaterials
A constant coeffcient of dilatancy has been often used, for example, by
Pradhan and Tatsuoka (1989). In the model, D

is the nonlinear function:

D D D
M ln
n
r
n
m mc m

= =
/
=
0 0

( )
DD
M
M
n
m
0

(9.33)
where
r

is the value of the stress ratio on the overconsolidation bound-

ary surface with the current mean effective stress. When

is larger
than
r

is kept to the value of

r

. In cases where the stress state is

inside the overconsolidated region, Equation (9.33) is the nth power of
the stress ratio invariant. When the value of n is larger than one and the
stress conditions are inside the overconsolidated region, the value of the
coeffcient of dilatancy drastically changes due to changes in the stress
ratio.
9.3.7 Stressstrain relation
The total strain increment tensor, d
kl
, is given by adding the elastic strain
increment tensor, d
kl
E
, and the plastic strain increment tensor, d
kl
P
, as

d d d
kl kl
E
kl
P
= + (9.34)
The plastic part of the deviatoric strain increment is derived from
Equation (9.29) and the consistency condition df = 0. The elastic parts of
the deviatoric strain increment are given as

de
G
ds
ij
E
E
m
ij
1
2( )
=

(9.35)
where G
E
is the elastic shear modulus normalized to the mean effective
stress,
m
. The elastic volumetric strain increment is also obtained with the
swelling index, , and void ratio, e, as

d
e
d
kk
E m
m
1
=

+

(9.36)
In the proposed constitutive model, the independent parameters are as
follows: void ratio, e; compression index, ; swelling index, ; elastic shear
modulus, G
E
; the stress ratio at the phase transformation, M
m

; the stress
ratio at failure, M
f

; anisotropic consolidation ratio, K

0
; fading memory
Liquefaction analysis of sandy ground 329
parameter, C
d
; kinematic hardening parameters, B
0

, B
1

, and C
f
; dilatancy
parameters D
0

and n; and strain parameters

( ) n r
P
and
( ) n r
E
.
All parameters can be theoretically determined by conventional tests, phys-
ical properties tests, or triaxial compression tests. As for parameters D
0

, n,
B
0

, B
1

, C
f
,
( ) n r
P
, and
( ) n r
E
, the curve-adjusting technique is available for easier
identifcation. Through the internal friction angle,
f
, and the phase trans-
formation angle,
m
, for drained triaxial tests, M
f

and M
m

are expressed as

M M
f
f
f
m
m
f
2 6 sin
3 sin
2 6 sin
3 sin
=


, =



(9.37)
9.4 PERFORMANCE OF THE CYCLIC MODEL
A numerical simulation of the sand behavior in undrained cyclic shear tests
has been conducted using a torsional hollow cylinder test apparatus (Oka
et al. 2003). The specimen was consolidated under an isotropic pressure of
98 kPa. The minimum void ratio was 0.605 and the maximum void ratio
was 0.977. The material parameters are listed on Table 9.1. Figures 9.2
through 9.9 show the numerical and the experimental results. From these
fgures, it is seen that the proposed model can reproduce the cyclic behavior
of sandy soil.
9.4.1 Determination of material parameters
Parameters e
0
, , , M
m

, and M
f

were determined directly from physi-

cal property tests and undrained monotonic shear tests. can be deter-
mined by the slope of the compression curve (e ln
m
) and by the slope
of the swelling curve (e ln
m
). The quasi-overconsolidation ratio, OCR

,
was set to be 1.0 based on the experimental conditions. M
m

and M
f

were
calculated by Equation (9.37) through monotonic shear tests. Laboratory
tests were not performed to determine the initial shear modulus. The initial
shear modulus of the specimen was estimated using the following equation
(Iwasaki and Tatsuoka 1977):

G
e
e
kPa
m
14092
(2 17 )
1
( )
0
2
0 4
=
.
+

.
(9.38)
The normalized elastic shear modulus, G
E
, was calculated by G
m 0
/ .
Although, in principle, the remaining parameters could be determined by
physical property tests and undrained monotonic and cyclic shear tests, the
330 Computational modeling of multiphase geomaterials
Table9.1 Material parameters
Parameters Dr = 60% Dr = 70%
Initial void ratio,e
0
0.754 0.716
Swelling index, 0.00052 0.00052
Compression index, 0.0091 0.0091
G
m 0 0
/
2023.6 1980.0
OCR* 1.2 1.2
M
m
*
0.707 0.707
M
f
*
0.990 0.990
B
0
*
4089 4001
B
1
*
54.5 61.5
C
f
0 0
Referential strain,
r
P*
0.002 0.003
Referential strain,
r
E*
0.012 0.015
D
0
*
0.60 0.60
n 5.1 7.0
Parameters Dr = 80% Dr = 90%
Initial void ratio, e
0
0.683 0.672
Swelling index, 0.00052 0.00052
Compression index, 0.0091 0.0091
G
m 0 0
/
1941.0 1877.3
OCR* 1.2 1.2
M
m
*
0.707 0.707
M
f
*
0.990 0.990
B
0
*
3924.0 3793.0
B
1
*
65.4 108.4
C
f
0 400
Referential strain,
r
P*
0.005 0.030
Referential strain,
r
E*
0.025 0.36
D
0
*
0.52 0.22
n 8.5 10.0
Liquefaction analysis of sandy ground 331
data-adjusting method is more practical for determining the soil param-
eters. In this method, the values of the material parameters are selected
to provide a good description of the stressstrain relations under mono-
tonic and cyclic loading conditions. The data-adjusting method was usedto
determine the dilatancy parameters, D
0

and n, the hardening parameters,

and the reference strain parameters. These parameters basically infu-
ence the effective stress path, the stressstrain curve, and the liquefaction
strength curve in the undrained tests. The dilatancy parameters, D
0

and n,
control the decrease in the mean effective stress before the cyclic mobility
condition and adjust the slope of the liquefaction strength curve. The hard-
ening parameters and the reference strain parameters, on the other hand,
10 5 0
Shear Strain (%)
LIQCA (Dr = 60%) EXPERIMENT (Dr = 60%)
S
h
e
a
r

S
t
r
e
s
s

(
k
P
a
)
5 10
30
20
10
10
0
20
30
10 5 0
Shear Strain (%)
S
h
e
a
r

S
t
r
e
s
s

(
k
P
a
)
5 10
30
20
10
10
0
20
30
Figure 9.2 (a) Stressstrain relation (Dr = 60%). (b) Simulated stressstrain relation
(Dr = 60%).
40
30
20
10
0
S
h
e
a
r

S
t
r
e
s
s

(
k
P
a
)
10
20
30
40
10 5 0
Shear Strain (%)
LIQCA (Dr = 70%) EXPERIMENT (Dr = 70%)
5 10
40
30
20
10
0
S
h
e
a
r

S
t
r
e
s
s

(
k
P
a
)
10
20
30
40
10 5 0
Shear Strain (%)
5 10
Figure 9.3 (a) Stressstrain relation (Dr = 70%). (b) Simulated stressstrain relation
(Dr = 70%).
332 Computational modeling of multiphase geomaterials
mainly control the behavior under cyclic mobility. They have an effect on
the effective stress path and stressstrain relations in the range of lower
mean effective stress.
9.5 LIQUEFACTION ANALYSIS OF A
LIQUEFIABLE GROUND
In this section, an application of the proposed elastoplastic model to a liq-
uefaction analysis (Oka et al. 2003) is presented. An effective-stress-based
fully coupled method for the dynamic analysis of a ground considering soil
liquefaction has been developed by many researchers. Herein, we adopt
S
h
e
a
r

S
t
r
e
s
s

(
k
P
a
)
Shear Strain (%)
EXPERIMENT (Dr = 90%)
60
40
20
0
20
40
60
4 2 0 2 4
S
h
e
a
r

S
t
r
e
s
s

(
k
P
a
)
Shear Strain (%)
LIQCA (Dr = 90%)
60
40
20
0
20
40
60
4 2 0 2 4
Figure 9.5 (a) Stressstrain relation (Dr = 90%). (b) Simulated stressstrain relation
(Dr = 90%).
40
30
20
10
S
h
e
a
r

S
t
r
e
s
s

(
k
P
a
)
0
10
20
30
40
10 5 0
Shear Strain (%)
EXPERIMENT (DR = 80%)
5 10
S
h
e
a
r

S
t
r
e
s
s

(
k
P
a
)
40
30
20
10
0
10
20
30
40
Shear Strain (%)
LIQCA (DR = 80%)
10 5 0 5 10
Figure 9.4 (a) Stressstrain relation (Dr = 80%). (b) Simulated stressstrain relation
(Dr = 80%).
Liquefaction analysis of sandy ground 333
30
20
10
0
10
20
30
30
20
10
0
S
h
e
a
r

S
t
r
e
s
s

(
k
P
a
)
S
h
e
a
r

S
t
r
e
s
s

(
k
P
a
)
10
20
30
Efective Mean Stress (kPa)
LIQCA (Dr = 60%)
120 100 80 60 40 20 0 0 120 100 80 60
Efective Mean Stress (kPa)
EXPERIMENT (Dr = 60%)
40 20
Figure 9.6 (a) Stress path (Dr = 60%). (b) Simulated stress path (Dr = 60%).
20
10
30
40
0
10
20
30
40
0 20 40 60 80 100 120
Efective Mean Stress (kPa)
LIQCA (Dr = 70%)
EXPERIMENT (Dr = 70%)
S
h
e
a
r

S
t
r
e
s
s

(
k
P
a
)
20
10
30
40
0
10
20
30
40
0 20 40 60 80 100
Efective Mean Stress (kPa)
S
h
e
a
r

S
t
r
e
s
s

(
k
P
a
)
Figure 9.7 (a) Stress path (Dr = 70%). (b) Simulated stress path (Dr = 70%).
20
10
30
40
0
10
20
30
40
0 20 40 60 80 100 120
Efective Mean Stress (kPa)
LIQCA (Dr = 80%) EXPERIMENT (Dr = 80%)
S
h
e
a
r

S
t
r
e
s
s

(
k
P
a
)
20
10
30
40
0
10
20
30
40
0 20 40 60 80 100 120
Efective Mean Stress (kPa)
S
h
e
a
r

S
t
r
e
s
s

(
k
P
a
)
Figure 9.8 (a) Stress path (Dr = 80%). (b) Simulated stress path (Dr = 80%).
334 Computational modeling of multiphase geomaterials
the method presented in Chapter 6 based on Biots mixture theory. The
numerical analysis of a three-dimensional liquefaction analysis of Port
Island in Kobe City, Japan, during the 1995 Hyogo-ken Nambu earth-
quake is presented, and we discuss the applicability of the proposed consti-
tutive model through a comparison of the simulated results and the vertical
array records.
9.5.1 Vertical array records on Port Island
Borehole strong motion vertical array observations were conducted on
Port Island by the Development Bureau of Kobe City. Three components of
accelerations, consisting of N-S, E-W, and U-D directions, were recorded
at four levels and at depths of GL-83 m, GL-32 m, GL-16 m, and GL-0m.
The depth of GL-83 m corresponds to the base layer that is on the top of
the diluvial dense sand: Ds. The depth of GL-32 m is under the alluvial
medium-dense sand: As. The depth of GL-16 m is under the reclaimed
gravelly sand: B3.
The acceleration histories of the main shock during the 1995 Hyogo-ken
Nambu earthquake of January 17, 1995, were obtained. The epicenter was
the northern edge of Awaji Island and the Japanese Meteorological Agencys
(JMA) magnitude was 7.2 (later modifed to 7.3 by JMA). The obtained
acceleration histories will be shown later with the simulated results.
9.5.2 Numerical models
In the analysis, a computer code called LIQCA3D was used. This program
was developed based on the cyclic elastoplasticity model for sand, men-
tioned earlier, and a cyclic elastoviscoplasticity model developed by Oka
and Yashima (1995). The cyclic elastoplasticity model for sand was applied
EXPERIMENT (Dr = 90%)
LIQCA (Dr = 90%)
60
40
20
0
20
40
60
0 20 40 60
Efective Mean Stress (kPa)
80 100 120
0 20 40 60
Efective Mean Stress (kPa)
S
h
e
a
r

S
t
r
e
s
s

(
k
P
a
)
S
h
e
a
r

S
t
r
e
s
s

(
k
P
a
)
80 100
60
40
20
0
20
40
60
Figure 9.9 (a) Stress path (Dr =90%). (b) Simulated stress path (Dr = 90%).
Liquefaction analysis of sandy ground 335
to the B1, B2, B3, and As layers shown in Figure9.10. In this simulation,
we used method 3 for the shear modulus reduction depending on the plastic
shear strain history. The cyclic elastoviscoplasticity model for clay was
applied to the Ac layer in Figure9.10. To model the stiff sandy ground and
sandy gravel, a generalized RambergOsgood model was used in which
the second invariant of the deviatoric stress tensor is incorporated. The
RambergOsgood model was applied to the Ds/c, Dc, and Ds layers in
Figure9.10
In Table9.2, the material parameters used in the analysis are listed. These
parameters are determined by the existing feld and laboratory testing data.
9.5.3 Common parameters
The soil profle and the groundwater table were determined from Figure9.10,
which was obtained by the boring at the observation site. The density and the
void ratio were determined from the density measurement of the undisturbed
Soil SPT-N
0 10 20 30 40 50
Shear Velocity (m/s)
0 100 200 300 400
B1
B2
B3
AC
(Ma 13)
As
Ds/c
Dc
(Ma 12)
Ds
D
e
p
t
h

(
m
)
100
90
80
70
60
50
40
30
20
10
0
Model
Port Island
Strong motion station
: Strong motion record
83 m
32 m
16 m
0 m
1995
1999
1995
1999
Figure 9.10 Soil profle at the observation site.
336 Computational modeling of multiphase geomaterials
T
a
b
l
e

9
.
2

M
a
t
e
r
i
a
l

p
a
r
a
m
e
t
e
r
s
N
a
m
e

o
f

S
o
i
l

P
r
o
f
l
e
B
1
B
2
B
3
A
c
A
s
D
s
/
c
D
c
D
s
S
o
i
l

t
y
p
e
G
r
a
v
e
l
l
y

s
a
n
d
G
r
a
v
e
l
l
y

s
a
n
d
G
r
a
v
e
l
l
y

s
a
n
d
M
a
r
i
n
e

c
l
a
y
S
a
n
d
S
a
n
d

w
i
t
h

c
l
a
y
M
a
r
i
n
e

c
l
a
y
S
a
n
d
M
o
d
e
l

t
y
p
e
E
-
P
E
-
P
E
-
P
E
-
V
P
E
-
P
R
-
O
R
-
O
R
-
O
D
e
n
s
i
t
y

(
t
/
m
3
)
1
.
9
0
2
.
1
0
2
.
1
0
1
.
6
7
2
.
0
0
2
.
0
0
2
.
0
0
2
.
0
0
I
n
i
t
i
a
l

v
o
i
d

r
a
t
i
o
e
0
0
.
4
2
0
.
4
2
0
.
4
2
1
.
4
1
0
.
5
0
0
.
5
0
1
.
2
0
0
.
5
0
C
o
e
f
f
c
i
e
n
t

o
f

p
e
r
m
e
a
b
i
l
i
t
y
k

(
c
m
/
s
)

0
.
0
0
3
0
.
0
0
3
0
.
0
0
0
2
0
.
0
0
2
0
.
0
0
2
0
.
0
0
0
1
0
.
0
0
2
C
o
m
p
r
e
s
s
i
o
n

i
n
d
e
x

0
.
0
1
0
.
0
1
0
.
0
1
0
.
3
3
1
0
.
0
1

S
w
e
l
l
i
n
g

i
n
d
e
x
k
0
.
0
0
1
0
.
0
0
1
0
.
0
0
1
0
.
0
4
2
5
0
.
0
0
1

P
o
i
s
s
o
n

s

r
a
t
i
o

0
.
3
5
0
.
3
5
0
.
3
5
I
n
i
t
i
a
l

s
h
e
a
r

v
e
l
o
c
i
t
y
V
s

(
m
/
s
)
1
4
0
1
4
0
2
3
0
1
8
0
2
3
0
3
3
0
2
8
0
4
5
0
I
n
i
t
i
a
l

s
h
e
a
r

m
o
d
u
l
u
s

r
a
t
i
o
G
o
/

m
2
0
0
2
7
3
0
1
0
1
9
3
2
8
5
1
6

F
a
i
l
u
r
e

s
t
r
e
s
s

r
a
t
i
o
M
*
f
1
.
3
4
1
.
3
4
1
.
3
4
1
.
2
3
1
.
2
6

P
h
a
s
e

t
r
a
n
s
f
o
r
m
a
t
i
o
n

s
t
r
e
s
s

r
a
t
i
o
M
*
m
0
.
9
1
0
.
9
1
0
.
9
1
1
.
0
3
0
.
9
1

I
n
t
e
r
n
a
l

f
r
i
c
t
i
o
n

a
n
g
l
e

(
d
e
g
)

4
5
4
5
4
5
R
-
O

p
a
r
a
m
e
t
e
r

1
.
0
0
1
.
1
7
0
.
8
7
r

2
.
1
3
2
.
0
6
2
.
1
3
a

1
2
4
1
4
6
4
6
6
1
7
9
7
1
b

0
.
5
0
.
5
0
.
5
H
a
r
d
e
n
i
n
g

p
a
r
a
m
e
t
e
r
B
*0
6
0
0
0
1
5
0
0
2
1
0
0

5
5
5
0
0
0

B
*1
0
1
5
0
1
4
0

1
0
0
C
o
n
t
r
o
l

p
a
r
a
m
e
t
e
r

o
f

a
n
i
s
o
t
r
o
p
y
C
d
2
0
0
0
2
0
0
0
2
0
0
0

2
0
0
0

Liquefaction analysis of sandy ground 337

R
e
f
e
r
e
n
c
e

s
t
r
a
i
n

p
a
r
a
m
e
t
e
r

r p

1
0
0
0

0
.
0
0
5
0
.
0
0
4

0
.
0
1

r
E

1
0
0
0
0
.
0
0
5
0
.
0
0
4

0
.
1

D
i
l
a
t
a
n
c
y

p
a
r
a
m
e
t
e
r
D
*0
0
.
0
1
.
0
1
.
0

0
.
0

n
0
.
0
4
.
0
4
.
0

0
.
0

V
i
s
c
o
p
l
a
s
t
i
c

p
a
r
a
m
e
t
e
r
m
0

1
4
.
0

C
1

(
1
/
s
)

5
.
5
4
E

0
6

C
2

(
1
/
s
)

7
.
7
6
E

0
7

338 Computational modeling of multiphase geomaterials

sample obtained after the earthquake. The coeffcient of permeability was
determined from laboratory permeability tests on the undisturbed sample
obtained after the earthquake. The shear velocity was estimated from the
PS logging tests at the observation site in 1995 and 1999.
9.5.4 Parameters for elastoplasticity model
It is important to estimate an appropriate shear modulus in an earthquake
response analysis. The initial shear modulus of the elastoplastic model is
given by the elastoplastic shear modulus, G
EP
, in the analysis. When there
exists an elastic region, the initial shear modulus G G
EP E
0 0
= and can be
obtained as

G V
E
s 0
2
= (9.39)
where G
E
0
is the initial elastic shear modulus, is the density, and V
s
is the
shear velocity. It is worth mentioning that the elastic modulus is normalized
by the mean effective stress.
The failure stress ratio was determined by undrained triaxial tests using
the undisturbed samples. The phase transformation stress ratio was deter-
mined from past laboratory tests by Pradhan et al. (1989).
The remaining parameters were determined by the data-adjusting
method for the undrained cyclic shear tests. In this method, the values for
the material parameters were selected in order to provide a good descrip-
tion of the stressstrain relations under cyclic loading conditions and lique-
faction strength curves. The undrained cyclic shear tests for the reclaimed
soil were conducted with the frozen in situ sample by Suzuki et al. (1997).
The experimental and the simulated liquefaction strength, which is the
cyclic shear stress that produces a double amplitude strain of 7.5% with
a particular number of cycles, is shown in Figure9.11. The simulated liq-
uefaction strength for the reclaimed B2 layer agrees well with the experi-
mental one.
9.5.5 Parameters for elastoviscoplasticity model
For clay layers we have used a cyclic elastoviscoplastic model presented
in Section 5.9.1. All remaining parameters were determined by un -
drained triaxial tests using the clay samples obtained at the observa-
tion site. In particular, the viscoplastic parameters were determined from
monotonic shear tests with two different loading rates. We confrmed the
applicability of the determined parameters through a simulation of the
Liquefaction analysis of sandy ground 339
dynamic strength obtained by the cyclic shear tests (Oka 1992; Okaetal.
2004).
9.5.6 Parameters for RambergOsgood model
The remaining parameters were determined by the strain-dependent shear
modulus and the damping ratio, which were obtained by the undrained
cyclic shear tests with multi-step loading. The Ds/c, Dc, and Ds layers
modeled by the RambergOsgood model have no dilatancy; hence, these
layers generate no buildup of excess pore water pressure (LIQCA Res.
Development Group 2005).
The elastic shear modulus is given by

G a
m
b
max
' =
0
(unit: kPa) (9.40)
The skeleton curve of the stressstrain relation is given by

= +

G
f
r
max
1
1
(9.41)
The shear strength follows MohrCoulomb failure criterion:
f
=
c
m
sin cos
0
+ .
Experiment
Number of Cycles N
10 1 100
Model
0.5
0.4
0.3
0.2
0.1
0.0
C
y
c
l
i
c

S
h
e
a
r

S
t
r
e
s
s

R
a
t
i
o

d
/

m
0
Undrained shear test for B2 Layer
DA = 7.5%
Figure 9.11 Liquefaction strength for reclaimed soil.
340 Computational modeling of multiphase geomaterials
9.5.7 Finite element model and
numerical parameters
The single-column ground model, composed of three-dimensional solid
fnite elements, was used for the analysis. The number of fnite elements is
31 and the number of nodes is 128. The model displacements are fxed at
the base. Four nodes at the same depth were assumed to move coinciden-
tally. The lateral and the bottom boundaries were assumed to be imperme-
able, while the ground water surface was assumed to be permeable. As an
input earthquake motion at the base rock, the three components (NS, EW,
and UD) of the acceleration records obtained at the depth of 83 m at the
vertical array recording site on Port Island were used.
Rayleigh damping proportional to the initial stiffness, which was deter-
mined by assuming that the damping factor was 1%, was used as the
convenient method in this study. A time integration step of 0.002 seconds
was adopted to obtain suffcient accuracy. and in the Newmark
method were set to be 0.3025 and 0.6, respectively, to ensure numerical
stability.
9.5.8 Numerical results
A comparison of the simulated and the observed absolute acceleration his-
tories is presented in Figure9.12. The amplitudes as well as the phases were
reproduced in the simulation results, although the peak accelerations at the
depths of GL-16 m and GL-32 m were underestimated. The simulated peak
acceleration was affected by the numerical parameters related to damping.
In this simulation, we overestimated the damping of the lower layers more
than the reclaimed layer. A comparison of the simulated and the observed
relative velocity histories is presented in Figure 9.13. The relative veloc-
ity is given by subtracting the base velocity at the depth of GL-83 m. The
observed velocity was obtained by integrating the acceleration histories in
the time domain. The simulation reproduced the observed velocity very
well for the amplitudes and the phases. The simulated time histories of the
effective stress decreasing ratio (ESDR), which is given by 1 0
0
. /
m m
, in
reclaimed layers B2 and B3 are shown in Figure9.14. Complete liquefac-
tion occurred after a few strong motion cycles of about 8 seconds in the
reclaimed layer. The changes in the frequency property of the accelera-
tion and the velocity at the ground surface are due to the liquefaction of
the reclaimed layer. It is well known that extensive soil liquefaction was
observed on Port Island, a man-made island in Kobe City (Shibata et al.
1996). It is seen that the proposed liquefaction method, based on the cyclic
elastoplasticity model, can simulate the dynamic response of liquefable
grounds well.
Liquefaction analysis of sandy ground 341
9.6 NUMERICAL ANALYSIS OF THE DYNAMIC
BEHAVIOR OF A PILE FOUNDATION
CONSIDERING LIQUEFACTION
Many structures were damaged during the 1995 Hyogo-ken Nambu
earthquake. It was found from the feld investigations after the earth-
quake that not only the pile heads but also the lower parts of the piles
had cracked or failed. This phenomenon indicates that both the inertia
force from the upper structures and the kinematic interaction between
A
c
c
e
l
e
r
a
t
i
o
n

(
m
/
s
2
)
6
4
2
0
2
4
6
0 5 10 15 20 25
GL-0 m
NS direction
30
A
c
c
e
l
e
r
a
t
i
o
n

(
m
/
s
2
)
6
4
2
0
2
4
6
0 5 10 15 20 25
GL-16 m
NS direction
GL-32 m
NS direction
30
A
c
c
e
l
e
r
a
t
i
o
n

(
m
/
s
2
)
6
4
2
0
2
4
6
0 5 10 15 20 25 30
GL-83 m
NS direction
Observed Simulated
A
c
c
e
l
e
r
a
t
i
o
n

(
m
/
s
2
)
6
4
2
0
2
4
6
0 5 10 15 20 25 30
6
4
2
0
2
4
6
0 5 10 15 20 25
GL-0 m
EW direction
30
6
4
2
0
2
4
6
0 5 10 15 20 25
GL- 16 m
EW direction
30
6
4
2
0
2
4
6
0 5 10 15 20 25
GL-32 m
EW direction
30
6
4
2
0
2
4
6
GL-83 m
EW direction
0 5 10 15 20 25 30
Time (s) Time (s)
Figure 9.12 Time histories of observed and simulated horizontal acceleration.
342 Computational modeling of multiphase geomaterials
the piles and the ground play important roles in the mechanical behav-
ior of piles. In particular, when the ground surrounding a structure
liquefes due to seismic excitations, the behavior of the piles is more
complicated. Damage related to liquefaction may involve cases in which
the pile foundation is damaged due to the lateral fow of liquefed soils,
or the piles fail at the boundary between two different soil layers, of
which one liquefes while the other does not. In this study, a series of
numerical simulations were conducted to study the dynamic behavior
V
e
l
o
c
i
t
y

(
m
/
s
)
1.0
0.5
0.0
0.5
1.0
1.0
0.5
0.0
0.5
1.0
1.0
0.5
0.0
0.5
1.0
1.0
0.5
0.0
0.5
1.0
1.0
0.5
0.0
0.5
1.0
1.0
0.5
0.0
0.5
1.0
1.0
0.5
0.0
0.5
1.0
1.0
0.5
0.0
0.5
1.0
0 5 10 15 20 25
GL-0 m
NS direction
30
V
e
l
o
c
i
t
y

(
m
/
s
)
0 5 10 15 20 25
GL-16 m
NS direction
GL-32 m
NS direction
30
V
e
l
o
c
i
t
y

(
m
/
s
)
0 5 10 15 20 25 30
GL-83 m
NS direction
Observed Simulated
V
e
l
o
c
i
t
y

(
m
/
s
)
Time (s) Time (s)
0 5 10 15 20 25 30
0 5 10 15 20 25
GL-0 m
EW direction
30
0 5 10 15 20 25
GL-16 m
EW direction
30
0 5 10 15 20 25
GL-32 m
EW direction
30
GL-83 m
EW direction
0 5 10 15 20 25 30
Figure 9.13 Time histories of observed and simulated relative horizontal velocity.
Liquefaction analysis of sandy ground 343
of a single-pile foundation constructed in a two-layer ground, whose
upper layer is flled with sandy soil that is dense sand, reclaimed soil,
medium dense sand or loose sand, and whose lower layer is flled with
clayey soil employing a three-dimensional liquefaction analysis method
(code name: LIQCA3D) to clarify the mechanism of the interactions
among the soilpilestructure (Lu 2002; Oka, Lu, Uzuoka, and Zhang
2004; Lu, Oka, and Zhang 2008).
1.0
0.8
0.6
1

m
/

m
0
0.4
0.2
0.0
Time (s)
0 5 10 15 20
GL-13.8 m
(B3 layer)
25 30
1.0
0.8
0.6
1

m
/

m
0
0.4
0.2
0.0
Time (s)
0 5 10 15 20
GL-6.4 m
(B2 layer)
25 30
Figure 9.14 Time histories of simulated effective stress decreasing ratio.
344 Computational modeling of multiphase geomaterials
9.6.1 Simulation methods
The two-layer ground considered in this chapter is a typical one near the
shore of a major Japanese urban city, such as Kobe. In order to study the
infuence of the soil characteristics, four different sandy materials are con-
sidered for an upper sandy ground: dense sand, medium dense sand, loose
sand, and reclaimed soil. Table 9.3 shows the parameters involved in the
constitutive models for the different types of soil (Lu et al. 2008). The con-
stitutive model for sand is described in detail in Section 9.2. On the other
hand, an axial force-dependent (AFD) model (Zhang and Kimura 2002),
in which the nonlinear behavior of steel and concrete is properly described
and shown in Figure9.15, is used to describe the dynamic behavior of the
RC pile that is 1.5 m in diameter. The parameters of the RC pile are shown
in Table9.4 (Lu et al. 2008).
The governing equations for the coupling problems between the soil
skeleton and the pore water pressure are obtained based on the two-phase
mixture theory. The liquefaction analysis is formulated using u (displace-
ment) p (pore pressure). The side boundaries of the simulated system are
C: the position where peak strength is reached
U: the position where residual stress state is reached
P: a post peak position where the maximum compressive strain
is experienced before the residual stress state is reached
v
s
= 0.8,
s
= 0.8,
= 0.001
v
c
= 0.5,
c
= 0.5
= 4, = 4

tc

m
0

E
r
E
r E
r
E
r
E
c
E
r
Crack-closing eect

r
= 0.5

mb

r
= 0.5

tc

m
: Maximum tensile strain
Er = (
c
+
t
)/(
c
+
t
)
v
s

y
v
c

c

c

c

t
T
v
s

y

y
(a) Steel (a) Concrete

s
E
s

s
E
s
Es
Es
Es

y
P
U
Figure 9.15 Stressstrain relations of steel and concrete adopted in the AFD model for
RC material.
Table9.3 Main material parameters for the soil
Soils Density M
*
f
M
*
m
G
0
/
m0
B
*
0
, B
*
l
, C
f
D
0
,n
Loose 2.0 0.80 0.70 500.0 2500,25,0 1.0,1.0
Medium 2.0 1.00 0.80 1060.0 4000,40,0 1.0,2.0
Reclaimed 2.0 1.19 0.91 2140.0 5500,55,0 1.0,4.0
Dense 2.0 1.10 0.85 1980.0 8500,85,0 1.0,2.5
Clay 1.7 1.31 1.28 300.0 500,50,0
Liquefaction analysis of sandy ground 345
assumed to be equal-displacement boundaries, the bottom of the system
is fxed, and the boundaries are impermeable except for the surface of the
ground. In this dynamic analysis, a stiffness-matrix-dependent Rayleigh
type damping is adopted and the direct integration method of Newmark
is used in this dynamic analysis with a time interval of 0.01 sec. The
groundwater table is 1.5 m beneath the ground surface. The mass of the
superstructure is 80,000kg and the height of pier is 8 m. Figure9.16 shows
the seismic wave used in this study, which is an NS component of the earth-
quake recorded on Port Island during the 1995 Hyogo-Ken Nambu earth-
quake, and Figure9.17 shows the confguration of the single-pile system and
the fnite element mesh used in the calculation.
9.6.2 Results and discussions
Figure9.18 shows the history of the effective stress decreasing ratio (ESDR
( )/
m m m 0 0
);
m0
is the initial mean effective stress) of the soil in the
middle of different types of sandy layers. Liquefaction occurs when the
8
4
0
4
8
0 2 4
Time (sec)
6 8 10
A
c
c
e
l
e
r
a
t
i
o
n

(
m
\
s
e
c
2
)
Figure 9.16 Input wave.
Table9.4 Main material parameters for the pile
Youngs modulus of concrete E
c
(kN/m
2
) 2.5 10
7
Diameter of pile D (m) 1.5
Compressive strength of concrete f
c
(kN/m
2
) 36000.00
Tensile strength of concrete f
t
(kN/m
2
) 3000.00
Degrading parameter of concrete
c
0.20690
Youngs Modulus of steel E (kN/m
2
) 2.1 10
8
Diameter of reinforcement d (m) 0.029
Number of reinforcement N 24
Yielding strength of steel Y
s
(kN/m
2
) 3.8 10
5
346 Computational modeling of multiphase geomaterials
ESDR is equal to 1. It can be seen that loose sand easily liquefes entirely,
while medium sand and reclaimed soil almost liquefy at the end of the
major seismic event (t = 10 sec). The effective stress of dense sand does
not decrease much at all. Figure 9.19 shows the histories of the bending
moments at the pile head and in the pile segment at the boundary between
the soil layers. Since the dense sand layer does not liquefy at all, the earth-
quake wave motion does not deamplify and the largest bending moment
occurs at the pile head among the cases. On the other hand, the larger
bending moments occur in the pile at the boundary between the layers
20 m
78 m
b7
b15
10 m
Clay layer
10 m
Sand layer
11 m
80 ton
8 m
Figure 9.17 Finite element mesh.
0 2
E
S
D
R

(
1

m
/

m
0
)
4
Time (sec)
6 8 10
1
0.5
0
0.5
1
Loose sand
Medium sand
Dense sand
Reclaimed sand
Figure 9.18 Effective stress decreasing ratiotime profle.
Liquefaction analysis of sandy ground 347
at t = 4 sec and t = 7 sec for loose sand, and medium dense sand and
reclaimed soil, respectively, when the effective stress of the sand layers
decreases signifcantly. Figure9.20 shows the distribution of the bending
moments when the maximum bending moment takes place in each case
and Figure9.21 shows the distribution of the bending moments at the end
of the seismic event. The fgures show that although the maximum bending
(b) In the pile segment at the soillayer boundary
Time (sec)
0 2 4 6 8 10
9000
9000
4500
0
4500
B
e
n
d
i
n
g

M
o
m
e
n
t

(
K
N
*
m
)
(a) At the pile head
Time (sec)
0 2 4 6 8 10
9000
9000
4500
4500
0
B
e
n
d
i
n
g

M
o
m
e
n
t

(
K
N
*
m
)
Dense sand
Medium sand
Loose sand
Reclaimed soils
Dense sand
Medium sand
Loose sand
Reclaimed soils
Figure 9.19 Histories of the bending moments in the pile.
348 Computational modeling of multiphase geomaterials
D
e
p
t
h

(
m
)
0
5
10
15
20
Sand layer
Clay layer
Bending Moment (KNm)
5000 5000 1 10
4
0
Loose Dense
Medium Reclaimed
Figure 9.21 Distribution of the bending moment at the time when the maximum bending
moment occurred at the bottom of the pier.
Loose Dense
Medium
Sand layer
Clay layer
Bending Moment (KNm)
4500 2250 0 2250 4500
Reclaimed
0
5
D
e
p
t
h

(
m
)
10
15
20
Figure 9.20 Distribution of the bending moment when the maximum bending moment
at the end of the seismic event (t = 10 sec).
Liquefaction analysis of sandy ground 349
moment takes place at the pile head (b15) in every case, the development
of bending moments in the ground varies due to the features of the soil.
A large bending moment takes place in the lower pile segment (b7) in the
cases of liquefable soil, but at the upper pile segment (b15) in the case of
dense sand at the end of the seismic event.
REFERENCES
Adachi, T., and Oka, F. 1982. Constitutive equations for sands and overconsolidated
clays and assigned works for sand. Constitutive relations for soils; Results of
the International Workshop on Constitutive Relations for Soils, Grenoble,
G.Gudehus, F. Darve, and I. Vardoulakis, eds., Balkema, Rotterdam, 141157.
Armstrong, P.J., and Frederick, C.O. 1966. A mathematical representation of the
multiaxial Bauschinger effect, C. E. G. B. Report RD/B/N 731.
Chaboche, J.L., and Rousselier, G. 1983. On the plastic and viscoplastic consti-
tutive equations Part I and Part II. J. Pressure Vessel Tech., Trans. ASME,
105:153164.
Desai, C.S., and Siriwardane, H.J. 1980. A concept of correction functions to
account for non-associative characteristics of geologic media, Int. J. Numer.
Anal. Meth. Geomech., 4:377387.
DiMaggio, F.L., and Sandler, I.S. 1971. Material model for granular soils, J. Eng.
Mech., ASCE, 97(EM 3):935950.
Drucker, D.C., and Prager, W. 1952. Soil mechanics and plastic analysis or limit
design, Quart. Appl. Math., 10:157164.
Earthquake Engineering Committee (JSCE). 2003. Soil liquefaction under level II
earthquake motion subcommittee, 2003. Report on the Soil Liquefaction under
Level II Earthquake Motion (written by F. Oka, K. Furuya, and R. Uzuoka),
157201 (In Japanese).
Ghaboussi, J., and Momen, H. 1979. Plasticity model for cyclic behaviour of sands,
Proceedings of 3rd International Conference on Numerical Methods in
Geomechanics, Aachen, W. Wittke, ed., Balkema, Rotterdam, 423434.
Hashiguchi, K. 1980. Constitutive equations of elastoplastic materials with elastic-
plastic transition, J. Appl. Mech., ASME, 47:266272.
Hashiguchi, K., and Ueno, M. 1977. Elastoplastic constitutive law of granular mate-
rials, In Constitutive Equations of Soils, S. Murayama and A.N. Schofeld, eds.,
Proceedings of 9th ICSMFE, Special session 9, Tokyoto, JSSMFE, 7382.
Ishihara, K., and Kabilamany, K. 1990. Stress dilatancy and hardening laws for rigid
granular model of sand, Soil Dynamics Earthquake Eng., 9(2):6677.
Iwasaki, T., and Tatsuoka, F. 1977. Effects of grain size and grading on dynamic
shear moduli of sands, Soils and Foundations, 17(3):1935.
Kato, M., Oka, F., Yashima, A., and Tanaka, Y. 1994. Dissipation of excess pore
water pressure by gravel drain and its analysis, Tsuchi-to-Kiso, JSSMFE,
42(4):3944 (in Japanese).
LIQCA Research and Development Group (Representative: F. Oka). 2005. Users
manual for LIQCA2D04 (2005 released print), http://www.nakisuna2.kuciv.
kyoto-u.ac.jp/liqca.htm.
350 Computational modeling of multiphase geomaterials
Lu, C.W. 2002. Numerical study of soil-pile interaction during earthquakes consid-
ering liquefaction, PhD thesis, Kyoto University.
Lu, C.W., Oka, F. and Zhang, F. 2008. Analysis of soilpilestructure interaction in a
two -layer ground during earthquakes considering liquefaction, Int. J. Numer.
Anal. Meth. Geomech., 32:863895.
Masing, G. 1926. Eigenspannungen und Verfestigung beim Messing.Proc. 2nd Int.
Congr. Appl. Mech., Zurich, 332335.
Naghdi, P.M., and Trapp, J.A. 1975. Restrictions on constitutive equations of fnitely
deformed elastic-plastic materials. Q. J. Mech. Appl. Math., 28(1):2546.
Nishi, K., and Esashi, Y. 1978. Stress-strain relationships of sand, Proc. JSCE,
280:111122.
Nishi, K., and Kanatani, M. 1990. Constitutive relations for sand under
cyclic loading based on elasto-plasticity theory, Soils and Foundations,
30(2):4359.
Oka, F. 1982. Constitutive equations for granular materials in cyclic loadings.
Proceedings of IUTAM Conference on Deformation and Failure of Granular
Materials, Delft, P.A. Vermeer and H.J. Luger, eds., Balkema, Rotterdam,
297306.
Oka, F. 1992. A cyclic elasto-viscoplastic constitutive model for clay based on the
non-linear-hardening rule, Proceedings of 4th International Symposium on
Numerical Models in Geomechanics, Swansea, G.N. Pande and S. Pietruszczak,
eds., Balkema, Rotterdam, 1:105114.
Oka, F., Furuya, K., and Uzuoka, R. 2004. Numerical simulation of cyclic behavior
of dense sand using a cyclic elasto-plastic model, Proceedings of International
Symposium on Cyclic Behavior of Soils and Liquefaction Phenomena,
Th.Triantafyllidis, ed., Balkema, Rotterdam, 8590.
Oka, F., and Kimoto, S. 2008. An elasto-viscoplastic constitutive model and its
application to the sample obtained from the seabed ground at Nankai trough,
Zairyo, Japanese Soc. Mater. Sci., 57(3):237242.
Oka, F., Kodaka, T., Koizumi, T., and Sunami, S. 2001. An effective stress based
liquefaction analysis based on fnite deformation theory, Proceedings of 10th
IACMAG, Tucson Arizona, C.S. Desai, T. Kundu, S. Harpalani, D. Contractor,
and J. Kemeny, eds., Balkema, Rotterdam, 11131116.
Oka, F., Kodaka, T., Kim, Y.-S. 2004. A cyclic viscoelastic-viscoplastic constitu-
tive model for clay and liquefaction analysis of multi-layered ground, Int. J.
Numerical and Analytical Methods in Geomechanics, 28(2):131179.
Oka, F., Lu, C.-W., Uzuoka, R. and Zhang, F. 2004. Numerical study of structure-
group pile foundations using an effective stress based liquefaction analysis
method, Proceedings of 13th WCEE, Vancouver, Canada, Paper No. 3338.
Oka, F., Sugito, M., Yashima, A., Furumoto, Y. and Yamada, K. 2000. Time-dependent
ground motion amplifcation at reclaimed land after the 1995 Hyogo-Ken-
Nambu Earthquake, Proceedings of 12th WCEE: No. 2046.
Oka, F., Uzuoka, R. Tateishi, A., and Yashima, A. 2003. A cyclic elasto-plastic model
for sand ant its application to liquefaction analysis, Constitutive Modeling of
Geomaterials, selected contributions from the Frank L. DiMaggio Symposium,
Inelastic Behavior Committee Engineering Mechanics Division, ASCE, CRC
Press, 7599.
Liquefaction analysis of sandy ground 351
Oka, F., and Washizu, H. 1981. Constitutive equations for sand and overconsoli-
dated clays, Proceedings of International Conference on Recent Advances
in Earthquake Engineering and Soil Dynamics, St. Louis, S. Prakash, ed.,
1:7174.
Oka, F., and Yashima, A. 1995. A cyclic elasto-viscoplastic model for cohesive soil,
Proceedings of XI European Conference on Soil Mechanics and Foundation
Engineering, Copenhagen, The Danish Geotechnical Society:6.1456.150.
Oka, F., Yashima, A., and Kato, M. 1993. Numerical analysis of wave-induced lique-
faction in seabed, Proc. 3rd Int. Offshore Polar Eng. Conf., 591598.
Oka, F., Yashima, A., Kato, M., and Nakajima, Y. 1994. An analysis of seepage fail-
ure using an elasto-plastic constitutive equation and its application, J. JSCE,
493, III-27:127135 (in Japanese).
Oka, F., Yashima, A., Kato, M., and Sekiguchi, K. 1992. A constitutive model for
sand based on the non-linear kinematic hardening rule and its application.
Proc. 10th WCEE:25292534.
Oka, F., Yashima, A., Shibata, T., Kato, M., and Uzuoka, R. 1994. FEM-FDM cou-
pled liquefaction analysis of a porous soil using an elasto-plastic model, Appl.
Sci. Res., 52:209245.
Oka, F., Yashima, A., Tateishi, A., Taguchi, Y., and Yamashita, S. 1999. A cyclic
elasto-plastic constitutive model for sand considering a plastic-strain depen-
dence of the shear modulus, Gotechnique, 49(5):661680.
Pastor, M., and Zienkiewicz, O.C. 1986. A generalized plasticity, hierarchical model
for sand under monotonic and cyclic loading, Proceedings of 2nd International.
Symposium on Numerical Models in Geomechanics Ghent, G.N. Pande and
W.F. Van Impe, eds., M. Jackson & Son, Cornwall, 131150.
Poorooshasb, H.B., Holubec, I., and Scherbourne, A.N. 1966. Yielding and fow of
sand in triaxial compression: Part I, Can. Geotech. J., 3(4):179190.
Poorooshasb, H.B., Holubec, I., and Scherbourne, A.N. 1967. Yielding and
fow of sand in triaxial compression: Part II and Part III, Can. Geotech. J.,
4(4):376397.
Pradhan, T.B.S., and Tatsuoka, F. 1989. On stress-dilatancy equations of sand sub-
jected to cyclic loading, Soils and Foundations, 29(1):6581.
Pradhan, T.B.S., Tatsuoka, F., and Sato, Y. 1989. Experimental stress-dilatancy
relations of sand subjected to cyclic loading, Soils and Foundations,
29(1):4564.
Prvost, J.H., and Keane, C.M. 1990. Multimechanism elasto-plastic model for soils.
J. Eng. Mech., ASCE, 116(9):19241944.
Shibata, T., Oka, F., and Ozawa, Y. 1996. Characteristics of ground deformation due
to liquefaction, Soils and Foundations, Special issue on Geotechnical Aspects
of the Jan. 17 1995 Hyogo-ken-Nambu Earthquake, 6579.
Shibata, T., Oka, F., Yashima, A., Goto, H., Goto, K., and Takezawa, K. 1992. A
numerical simulation of shaking table test of coal fy ash deposit with cement
mixing column, In Numerical Models in Geomechanics, G.N. Pande and
S.Pietruszczak, eds., Balkema, Rotterdam, 411420.
Sugito, M., Oka, F., Yashima, A., Furumoto, Y., and Yamada, K. 2000. Time-
dependent ground motion amplifcation characteristics at reclaimed land after
the 1995 Hyogoken Nambu Earthquake, Eng. Geol., 56:137150.
352 Computational modeling of multiphase geomaterials
Suzuki, Y., Hatanaka, M., and Uchida, A. 1997. Drained and undrained shear
strengths of a gravelly fll of weathered granite from Kobe Port Island, J. Struct.
Constr. Eng., AIJ, 49:867873 (in Japanese).
Taguchi, Y., Tateishi, A., Oka, F., and Yashima, A. 1995. A cyclic elastoplastic
model based on the generalized fow rule and its application, Proceedings of
5th International Symposium on Numerical Models in Geomechanics, Davos,
Switzerland, G.N. Pande and S. Pietruszczak, eds., Balkema, 5762.
Taguchi, Y., Tateishi, A., Oka, F., and Yashima, A. 1997a. Numerical simulation of
soil-foundation interaction behavior during subsoil liquefaction, Proceedings
of 6th International Symposium on Numerical Models in Geomechanics,
S.Pietruszczak and G.N. Pande, eds., Blakema, Rotterdam, 561566.
Taguchi, Y., Tateishi, A., Oka, F., and Yashima, A. 1997b. Three-dimensional liq-
uefaction analysis method and array record simulation in Great Hanshin
Earthquake, Proceedings of 11th World Conference on Earthquake
Engineering, Acapulco, Sociedad Mexicana de Ingenieria Sismica, Balkema
(CD-ROM), Rotterdam, 1042.
Uzuoka, R. 2000. Analytical study on the mechanical behavior and prediction of soil
liquefaction and fow, PhD thesis, Gifu University (in Japanese).
Vermeer, P.A. 1978. A double hardening model for sand, Gotechnique,
28(4):413433.
Yashima, A., Oka, F., and Kanami, H. 2000. 3-D analysis to evaluate of soil improve-
ment on liquefaction of man-made island, Proceedings of 12th WCEE, No. 2116.
Zhang, F., and Kimura, M. 2002. Numerical prediction of the dynamic behaviors of
an RC group-pile foundation, Soils and Foundations, 42(3):7792.
353
Chapter 10
Recent advances in
computational geomechanics
In this chapter, two selected subjects on recent advances in geomechanics
are presented: the thermo-hydro-mechanical coupled problem and the seep-
agedeformation coupled analysis of unsaturated soil.
10.1 THERMO-HYDRO-MECHANICAL
COUPLED FINITE ELEMENT METHOD
In this section, we develop a thermo-hydro-mechanical coupled fnite ele-
ment method by considering the conservation of energy in addition to the
soil-water coupled fnite element method. Biots (1956) theory of porous
media is adopted to give the governing equations for the soilwater cou-
pling problem presented in Chapter 2.
Thermal consolidation problems have been studied by many researchers
(e.g., Campanella and Mitchell 1968; Baldi et al. 1988; Akagi and Komiya
1995; Delage et al. 2000; Sultan et al. 2002). Boudali et al. (1994) and
Leroueil and Marques (1996) demonstrated the temperature dependency of
clay through experiments on natural clay, and Cekerevac and Laloui (2004)
reported the thermal effects on the mechanical behavior of clay in experi-
ments. Thermo-hydro-mechanical (THM) coupling problems are very
important in the feld of geomechanics. Thermo-hydro-mechanics has been
applied to several problems such as nuclear waste disposal, ground heat-
ing, thermal consolidation, cleanup techniques for contaminated grounds,
rapid landslides, and earthquake faulting (e.g., Komine and Ogata 2002;
Cleall et al. 2006; Laloui et al. 2006; Vardoulakis 2002; Sulem and Famin
2009). Vardoulakis (2002) studied the effect of thermal softening on
catastrophic landslides. Kimoto et al. (2010) developed a chemo-thermo-
mechanical coupled numerical method to analyze the behavior of ground
during the dissociation of methane hydrate.
The temperature-dependent viscoplastic behavior was incorporated into
the elastoviscoplastic model by Yashima et al. (1998) based on the experi-
mental results by Boudali et al. (1994). They showed that the viscoplastic
354 Computational modeling of multiphase geomaterials
parameter, m, in Equation (5.19) is not dependent on the temperature,
but the viscoplastic parameter C in Equation (5.19) is dependent on the
temperature. The model was then incorporated into the fnite element pro-
gram and successfully applied to the thermal consolidation analysis using
the fnite element method with the energy balance law (Oka et al. 2005).
Theother application of the temperature-dependent model is the analysis
of the ground deformation due to the dissociation of a methane hydrate
(Kimoto et al. 2010).
10.1.1 Temperature-dependent
viscoplastic parameter
Yashima et al. (1998) introduced the temperature dependency of the visco-
plastic parameter based on the experimental data by Boudali et al. (1994).
Figure10.1 shows the relations between the logarithm of the strain rate and
the preconsolidation pressure. Parameter m seems to be independent of the
temperature. This means that viscoplastic parameter, C, may depend on the
temperature. As a result, Yashima et al. (1998) showed the following relation
between consolidation yield stress,
p
, and temperature, (see Figure10.1):

p
pr
r
(10.1)
CRS
Creep
MSL
P
MSL
24
20 5 35
100
90
80
70
60
50
P
r
e
c
o
n
s
o
l
i
d
a
t
i
o
n

P
r
e
s
s
u
r
e

p

(
k
P
a
)
Vertical Strain Rate
1
(s
1
)
40
30
10
9
10
8
10
7
10
6
10
5
10
4
Figure 10.1 Relation between logarithm of strain rate and consolidation yield stress.
(After Leroueil, S., and Marques, M.E.S. 1996, Geotechnical Special
Publication No. 61 ASCE, T.C. Sheahan and V. Kaliakin, eds., 160.)
Recent advances in computational geomechanics 355
in which
pr
is the value of
p
at referential temperature
r
and is the
gradient of line

p
log log .
If the stress ratio during one-dimensional compression is assumed to be
constant and the initial hardening parameter
myi
s ( )
in Equation (5.60) cor-
responds to consolidation yield stress,
p
, namely,

mai
myi
s
p
( )
0
(10.2)
where
0
is a consolidation stress. Temperature-dependent viscoplastic
parameter C() is rewritten from Equation (5.60) as

C C m
M
p
( ) exp ln
0
(10.3)
Substituting Equation (10.1) into Equation (10.3) yields

C C m
M
r
r
( ) ( )exp ln (10.4)

C C m
M
r
pr
( ) exp ln
0
(10.5)
From Equations (10.4) and (10.5), the temperature dependency of visco-
plastic parameter C is obtained as follows:

C
C
m
M
r r
( )
( )

, =

(10.6)
in which the coeffcient of dilatancy,

M , is a function of the stress ratio
in the overconsolidated (OC) region (see Equation 5.51). However, the
thermoviscoplastic parameter is independently determined as a mate-
rial constant. Adopting the temperature-dependent viscoplastic parameter,
Equation (10.6), into the viscoplastic fow rule, Equation (5.56), an elasto-
thermo-viscoplastic model is derived.
Figure 10.2 shows simulated results for constant rate-of-strain (CRS)
odometer tests using the above-mentioned model (Yashima et al. 1998)
with the experimental results at strain rates of

s 1 10
5 1
and

s 1.6 10
7 1

and temperatures of 5C and 35C from tests performed by Boudali et al.
(1994). It is seen that the simulated results reproduce all the compression
curves well.
356 Computational modeling of multiphase geomaterials
Average Vertical Efective Stress (kPa)
V
e
r
t
i
c
a
l

S
t
r
a
i
n

(
%
)
0
0
5
10
15
20
25
50 100 150 200 250 300
T(C)
35
5
35
5
Sim. Exp.
1
(s
1
)
1.0 10
5
1.0 10
5
1.6 10
7
1.6 10
7
(a)
0 50 100 150
Average Vertical Efective Stress (kPa)
200 250 300
P
o
r
e

W
a
t
e
r

P
r
e
s
s
u
r
e

a
t

t
h
e

B
o
t
t
o
m

o
f

t
h
e

S
p
e
c
i
m
e
n

(
K
p
a
)
40
35
30
25
20
15
10
5
0
(b)
Figure 10.2 (a) Simulated and experimental results of CRS tests with different tempera-
tures. (b) Simulated and experimental results of CRS tests with different
temperatures. (From Yashima, A., Leroueil, S., Oka, F., and Guntoro, I.,
1998, Soils and Foundations, 38(2):6373.)
Recent advances in computational geomechanics 357
10.1.2 Elastic and temperature-
dependent stretching
An additive decomposition of the total stretching tensor, D
ij
, into elastic
stretching D
ij
e
and viscoplastic stretching D
ij
vp
is assumed such that

= + D D D
ij ij
e
ij
vp
(10.7)
Elastic stretching, D
ij
e
, is given by a generalized Hookes law, namely,

= +

+


D
G
S
e
ij
e
ij
m
m ij
1
2

3(1 )
(10.8)
where

S
ij

is the Jaumman rate of the deviatoric Cauchy stress tensor, m is

the mean effective stress, G is the elastic shear modulus, e is the void ratio, is
the swelling index, and the superimposed dot denotes the time differentiation.
In the thermo-hydro-mechanical coupled analysis, we added a new term
to the constitutive equations, which is contributed by the change in tem-
perature. Total stretching tensor, D
ij
, is assumed as

= + +

D D D
ij ij
e
ij
vp
ij
1
3
(10.9)
where

is the coeffcient of thermal expansion.

The relation between the elastic stretching tensor, D
ij
e
, and the Jaumann
rate of Cauchy stress tensor, T ij

, can be obtained as

=
T
C D
ij
ijkl
e
kl
e

(10.10)
which leads to the relation

T
C D D
ij
ijkl
e
kl
kl
vp
kl
(
1
3
)
(10.11)
From the fow rule, the viscoplastic stretching tensor, D
ij
vp
, which cor-
responds to Equation (5.58), can be obtained as

= < >

D f
f
T
ij
vp
y
p
ij
( )
1
(10.12)
In the analysis, the tangent modulus method (Peirce et al. 1984) is
adopted in order to evaluate the viscoplastic stretching tensor, D
ij
vp
. Next,
358 Computational modeling of multiphase geomaterials
the relation between the rate of effective stress and the stretching tensor can
be written in matrix form, as shown in the following equation:

T C D Q C I {

} [ ]{ } { }
1
3
[ ]{ } (10.13)
Substituting Equation (6.195) into Equation (10.13) yields


= +

T C D Q W C I { } [ ]{ } { } { }
1
3
[ ]{ } (10.14)
where [C] is the tangential stiffness matrix, {Q} is the relaxation stress vec-
tor, and {W*} = {WT TW} is the vector of the spin tensor.
10.1.3 Weak form of the equilibrium
equation for watersoil mixture
We will use the following weak form of the equilibrium equation from
Equation (6.194):

T D dv T L L dv T D L dv
u
D dv
ij ij
V
jp pi ij
V
ij kk ij
V
w kk
V

+ +

U L dv
s
v d
ji ij
V
i i
t

+ =

(10.15)
in which tension is positive for the pore water pressure, u
w
, as seen by
Equation (6.175) in Chapter 6.
By all the matrix and vector relations obtained previously, by discretizing
Equation (10.15) based on the theory of virtual work, and by considering
the arbitrariness of the unconstrained virtual nodal velocity, we obtain
K v B Q dV B W dV K v K
u
K F
T
V
T
V
L v w
[ ]{ } [ ] [ ] { } [ ]{ } [ ]{ } [ ]{ } { }
*

{ }
+ + + + =

(10.16)
in which

K B C B dV
T
V
[ ] [ ] [ ][ ]

=
(10.17)
K B
D
B dV B U B dV B T B dV
L
V
M
T
s M
V
M
T
M
V
M
T
T
v
[ ] [ ] [ ][ ] [ ] [ ][ ] [ ]

{ } = + + { }
(10.18)
Recent advances in computational geomechanics 359

K B N dV
v
V
v
T
h
[ ] { } { }

=
(10.19)

F N s d
T
t
[ ] { }

{ }
=

(10.20)

K C B N dV
V
v
T
[ ] [ ]{ }
1
3
[ ]

=
(10.21)
where

s { } is the rate of the surface force vector, [B] is the matrix that trans-
forms the nodal velocity vector to the stretching vector,

v { } is the nodal
velocity vector, [N] is a shape function of the eight-node quadrilateral ele-
ment, B
M
[ ] is the matrix that transforms the nodal velocity vector to the
velocity gradient vector, { }
B
v
is the vector that transforms the nodal veloc-
ity into the trace of D
ij
,

uw
{ }
and

{ }
are the nodal pore pressure rate
vector and the nodal temperature rate vector, respectively, and N
h
[ ] and

N [ ] represent the four-node quadrilateral element shape functions for

pore water pressure and temperature, respectively.
The relation between the nodal velocity vector and the nodal displace-
ment vector can be obtained using Eulers approximation as

=

v
u
t
u { }
{ }
{ }: nodal displacement increment vector (10.22)
Similarly, the pore water pressure rate and the temperature rate can be
obtained as

=

+

u
u u
t
w
t t
w
t
w
{ }
{ } { }
(10.23)

+

t
t t t
{ }
{ } { }
(10.24)
Substituting Equations (10.22) and (10.23) into Equation (10.16) yields

{ }
+ +

+



+ =

+

K
u
t
K
u
t
K
u u
t
K
t
B Q dv B W dv F
L v
t t
w
t
w
t t t
T
V
T
V
[ ]
{ }
[ ]
{ }
[ ]
{ } { }
[ ]
{ } { }
[ ] [ ] { } { } (10.25)
360 Computational modeling of multiphase geomaterials
Finally, by transforming the preceding equation, the weak form of the
equilibrium equation is obtained:
K K u K
u
K t F K
u
K
t B Q dv t B W dv
L v
t t
w
t t v
t
w
t
T
V
T
V

{ }
+ + + = + +
+

[[ ] [ ]]{ } [ ]
{ }
[ ]{ } { } [ ]
{ }
[ ]{ }
[ ] [ ] { }
* *
(10.26)
10.1.4 Continuity equation
For the behavior of the pore water pressure, the boundary surface of a
closed domain V can be broken into two components, namely,

= +
p q
(10.27)
where
p
is the pore pressure boundary and
q
is the pore water fow
boundary.
On boundary
p
, = u
u w w
(10.28)
On boundary
q
, q q
w w
= (10.29)
Considering the change in temperature, the continuity equation of the
fuid phase can be written as

+ + =
x
k u
x
b D
n
K
u
i w
w
i
f
i ii
f
w w
0 (10.30)
where k is the coeffcient of permeability,
w
is the pore water density, K
f
is
the volumetric modulus of pore water,
w
is the thermal expansion coef-
fcient of the pore water, the soil particle is incompressible, and tension is
positive for pore water pressure.
If the body force and the mass density of the pore water are constant,
Equation (10.30) becomes

+ =
x
k u
x
D
n
K
u
i w
w
i
ii
f
w w
0
(10.31)
Considering a test function of uw
, which is an arbitrary function, except
that uw
= 0 at the pore pressure prescribed boundary, a weak form of the
continuity equation can be obtained as

u
x
k u
x
dv
u
D dv
n
K
u dv
u
dv
u
q q d
u
w
i w
w
i
V
w ii
V
f
w
V
w w
V
w wi wi
w
q

( ) 0

+
=

(10.32)
Recent advances in computational geomechanics 361
By applying Gausss theorem to the previous equation and disregard-
ing the spatial dependency of the permeability, the following equation is
obtained:

k
u
u
x
n d
k
u
u
x
dv
u
D dv
n
K
u u dv
u dv u q n d
w
w
w
i
i
w
w i
w
i
V
w ii
V
f
V
w w
w w
V
w wi
q
i
p


0

+
+ =

(10.33)
Since the frst term can be omitted, because = u
w
0 on
p
or the trial
function satisfes the pore pressure prescribed boundary
p
, Equation
(10.33) becomes
k
u
u
x
dV
u
D dV
n
K
u u dV u dV
u q n d
w
V
w i
w
i
V
w ii
f
V
w w w w
V
w wi
q
i


0

+
+ =
,

(10.34)
Equation (10.34) is discretized as
k
u
B B u dV
u
N B v dV
u
n
K
N N u dV
u
N N dV
u N q n d
w
V
T
w
h
T
h w
V
T
w
h
T
v
V
T
w
f
h
T
h w
V
T
w
w h
T
w
T
h
T
wi
q
i
{

} [ ] [ ] {

} [ ] { }{ }
{

} [ ] [ ]{ } {

} [ ] [ ]{ }
{ } [ ] 0
* *
*

+

+ =
{ }

(10.35)
where

=

u N u
w h w
[ ]{ } (10.36)

= =
, ,

u N u B u
w i h i w h w
{ } [ ]{ } [ ]{ } (10.37)

=
,
N N
h i h
[ ] { } (10.38)

= =

D trD B v
ii
T
v
{ } { } (10.39)

=

N [ ]{ }
*
(10.40)
362 Computational modeling of multiphase geomaterials
By taking the arbitration of the test function, considering the arbitrari-
ness of the test function, unconstrained nodal pore pressure

u
T
w
{

}
, we have
k
B B dV u N B dV v
n
K
N N u dV
N
N dV N q n d
w
V
h
T
h w
V
h
T
v
V
f
h
T
h w
V
w h
T
h
T
i
f
q
i
[ ] [ ] [ ] { } { }
[ ] [ ]{ }
[ ]
[ ]{ } [ ] 0
* *

+


=
{ }

(10.41)
From Equation (10.40), we defne the following terms:

K
k
B B dV
h
w
V
h
T
h
[ ] [ ] [ ]

(10.42)

K N B dV
v
T
V
h
T
v
[ ] [ ] [ ]

=
(10.43)

K
n
K
N N dV
c
V
f
h
T
h
[ ] [ ] [ ]

= (10.44)

K N N dV
w
V
w h
T
[ ] [ ] [ ]

=
(10.45)

V N q n d
h
T
wi i
q
{ } [ ]

(10.46)
Substituting these matrices into Equation (10.41) yields

K v K u K u K V
v
T
h w c w w

+ + =

[ ] { } [ ] [ ]{ } [ ]{ } { }
* *
(10.47)
By using Eulers approximation, the following relation between the nodal
velocity vector and the nodal displacement vector can be obtained:

=

v
u
t
{ }
{ }
(10.48)
where

u { } is the nodal displacement.

For the pore water pressure rates at the nodes, the temperature rates at
the nodes are approximated by the fnite difference scheme, as shown by
Equations (10.23) and (10.24).
Recent advances in computational geomechanics 363
Substituting Equations (10.48), (10.23), and (10.24) into Equation (10.47)
gives

+ +
= + +

+ +

K u t K u K K u
K u K t V
v
T
h
t t
w w t t c w t t
c w t w t
[ ] [ ] [ ]{ } [ ]{ }
[ ]{ } [ ]{ } [ ]
* *
* *
(10.49)
10.1.5 Balance of energy
The frst law of thermodynamics presented in Chapter 1, namely, the con-
servation of energy (Equation 1.138), is given as

=
,
e
T
D h
ij ij i i
(10.50)
where

e is the rate of internal energy, T ij is the effective Cauchy stress

tensor, and h
i
is the heat fux vector.
As for the formulation of h
i
, we can use the Cattaneo equation (Mller
and Ruggeri 1998), namely,

+ =
,
h
h
k
i
i
i
(10.51)
in which

k is the coeffcient of heat conductivity, is the temperature, and

is the heat fux parameter regarded as the relaxation time for the station-
ary state. If we use the Cattaneo equation we have a hyperbolic type of heat
conduction equation instead of a parabolic equation. When is equal to
zero, Equation (10.51) becomes Fouriers law.
The constitutive relation between the elastic stretching tensor, D
ij
e
, and
the rate of internal energy,

e , is assumed as follows:

= + e c
T
D
ij ij
e
(10.52)
where is the density of soil and c is the specifc heat capacity.
By taking Equations (10.50) and (10.52), the conservation of energy is
written using the viscoplastic stretching tensor, D
ij
vp
, as

=
,
c
T
D h
ij ij
vp
i i
(10.53)
Using the boundary condition for the heat fow, the boundary can be
divided into two parts, namely,

and
T
as = +
T
.
364 Computational modeling of multiphase geomaterials
On boundary

=
On boundary
T
= h h
i i
where the upper bar denotes the prescribed values.
Adopting the boundary conditions, the weak form of the conservation of
energy can be written as follows:

c
T
D h dv h
h
n d
ij ij
vp
i i
V
i
i
i
T
( )

( )

+ + =
,

(10.54)
in which

is the test function that is zero on

and hi
is the speci-
fed value of the heat fux. Since is arbitrary, we can take = 1. Using
Fouriers law in Equation (10.51) and adopting the Gauss theorem into
Equation (10.54) provides

c
T
D dv k dv
h
n d
ij ij
vp
V
i i
V
i
i
( )

0
, ,

+ + =

(10.55)
For the discretization, the shape function

N [ ] is used

= , = =

, ,

N N B
i i
{ } [ ]{ } { } [ ]{ } [ ]{ }
(10.56)
where superscript asterisk (*) indicates the nodal value.
Eulers approximation is used for the rate of temperature,

{ } , and the
rate of heat fux,

hi
{ }
, and considering arbitrariness,

T
{ }
, we have from
Equation (10.55)

+

=
+

K K F H
T T
t t
T
([ ] [ ])
{ }
{ } { }
1 2 (10.57)

K c N N dv K t k B B dv
T
T
V
T
T
V
[ ] [ ] [ ] [ ] [ ] [ ]
1 2

= , =
(10.58)

F K t N
T
D dv
T T
t
T
ij ij
vp
V
{ } [ ] [ ]
1

= +
{ }

(10.59)

H t N
h
n d
T
i
i
T
[ ] { }{ }

{ } =

(10.60)
in which subscript t + t denotes an unknown value, and subscript t indi-
cates the latest known value.
The element for the temperature is a four-node quadrilateral isoparamet-
ric element with four Gaussian integration points. Combining Equations
Recent advances in computational geomechanics 365
(10.26), (10.49), and (10.58) gives the governing equation for the fnite ele-
ment formulation as
K K K K
K t K K K
K K
u
u
t F K
u
K t B Q dV t B W dV
K u K t V
F H
L v
v
T
h c w
T T
t t
w
t t
v
t
w
t
V
T
V
T
c w t w t
T
[ ] [ ] [ ] [ ]
[ ] [ ] [ ] [ ]
0 0 [ ] [ ]
{ }
{ }
{ }
{ } [ ]
{ }
[ ]{ } { } [ ] { }
[ ]{ } [ ]{ } { }
{ }
1 2
*
* *

[ ]
{ }
+
+
+

=
+ + +
+ +

(10.61)
10.1.6 Simulation of thermal consolidation
Heating water-saturated soil induces volume expansion of the soil particles
and the pore water. Even if the heated soil is under drained conditions,
excess pore water pressure is generated. Dissipation of the heat-generated
pore pressure induces consolidation, which is called thermal consolidation
(Campanella and Mitchell 1968; Baldi et al. 1988; Delage et al. 2000; Sultan
et al. 2002). Figure10.3 shows the experimental results by Campanella and
Mitchell (1968) during the increase and decrease in temperature under a
constant confning pressure. It can be seen that the consolidation and swelling
occur during the increase and decrease in temperature. In this section, we
conducted a simulation of thermal consolidation due to the dissipation of
pore water pressure induced by heat and viscoplastic thermal softening.
For simplifed and practical formulations, both the grain particles and
the fuid are assumed to be mechanically incompressible. Based on the fnite
deformation theory presented in Chapter 6, an updated Lagrangian method
with the objective Jaumann rate of Cauchy stress is used for the weak form
of the rate type of equilibrium equations for the whole soilwater mixture.
As for the element type, an eight-node quadrilateral isoparametric element
with a reduced Gaussian two-point Gauss quadrature rule for integration is
used for the displacement, while the pore water pressure and the tempera-
ture are defned by a four-node quadrilateral isoparametric element.
The heat-generated pore pressure is simulated using the model by
Campanella and Mitchell (1968), which is given as

=
+
u
n
m
w
s w st
v
( )
(10.62)

=

+
m
e
v
m
(1 )
(10.63)
366 Computational modeling of multiphase geomaterials
in which u
w
is the heat-generated pore pressure, is the change in tem-
perature, n is the porosity, m
v
is the coeffcient of volume compressibility,

s
is the thermal coeffcient of expansion for the soil particles, and
st
is
a physicochemical coeffcient of the change in structural volume. In this
analysis,
s
= 0.35 10
4
(1/C) and
st
= 0.50 10
4
(1/C), which are the
same values used by Campanella and Mitchell (1968). As for the thermal
coeffcient of expansion for water,
w
, we employed an empirical formula
by Baldi et al. (1988):

= +

w
(139.0 6.1 ) 10
6
(10.64)
In addition, the temperature dependency of the coeffcient of permeabil-
ity, k, is also considered, namely,

=

k
K
w
( )
( )
(10.65)
S
h
o
r
t
e
n
i
n
g
S
a
m
p
l
e

H
e
i
g
h
t

C
h
a
n
g
e

(
i
n
c
h
)
0.000
0
3
4
5
13
14
6
7
1
15
9
16
8
2
12
10
17
11
0.002
0.004
0.006
0.008
0.010
0.012
0.014
0.016
0.018
40 50 60 70 80 90 100
Temperature (degree in fahrenheit)
110 120 130 140 150
Figure 10.3 Effect of temperature variations on height and volume change. (After
Campanella, R.G., and Mitchell, J.K., 1968, J. SMFE, ASCE, 94, 3:709734.)
Recent advances in computational geomechanics 367
where K is the intrinsic permeability and
w
( ) is the unit weight of water,
in which the temperature dependency of
w
( ) is assumed to be negligible.
As for the temperature dependency of the viscosity of water, , an empiri-
cal formula, = 0.00046575ln()+0.00239138 (Pas), is used (Delage et
al. 2000).
Other material parameters that were used in the simulation of thermal
consolidation are listed in Table 10.1. Figure 10.4 depicts the boundary
conditions and the size of the specimen for the simulation of thermal con-
solidation. An isothermal boundary is imposed at the left side of the speci-
men as the installation of a heat source, for which temperature increases
gradually from its initial value of 20C at t = 0 (hr) to 60C at t = 4 (hr) (see
Figure 10.5). The upper boundary is also an isothermal boundary of the
atmospheric temperature of 20C.
The settlement at node A is shown in Figure10.6. We can observe around
1.8 cm of settlement at t = 150 (hr) and it is seen that the deformation does
not converge.
The distribution of temperature, the pore water pressure, the mean effec-
tive stress, and the accumulated viscoplastic volumetric strain are demon-
strated in Figure10.7. It can be observed from these fgures that thermal
consolidation due to the dissipation of the pore water pressure induced by
Table10.1 Material parameters
Compression index 0.172
Swelling index 0.054
Initial void ratio e
0
0.72
Initial mean effective stress (kPa)
me
392
Compression yield stress (kPa)
mbi
392
Coeffcient of earth pressure at rest K
0
1.0
Viscoplastic parameter m 21.5
Viscoplastic parameter (1/s) C(
r
) 4.510
-8
Stress ratio at critical state M
*
m
1.05
Initial elastic shear modulus (kPa) G 5500
Coeffcient of permeability (m/s) k 1.5410
-8
Referential temperature (C)
r
20.0
Soil density (t/m
3
) 1.96
Specifc heat capacity (J/kgC) c 0.87
Coeffcient of heat conductivity (J/kgC) k

0.14
Heat fux parameter (s) 0.0
Thermo-viscoplastic parameter a 0.15
Structural parameter (kPa)
maf
350
Structural parameter 10
368 Computational modeling of multiphase geomaterials
Permeable boundary
Isothermal boundary
(atmospheric temperature: 20C)
2 m
2 m
: Horizontally xed
Impermeable
and adiabatic
boundaries
: Fixed
Isothermal
boundary
60C
Pore water and heat
are allowed to ow
10 10 = 100 elms
Figure 10.4 Finite element mesh and boundary conditions for pore water pressure and
temperature.
Temperaure
60C
10C increase
per 1 hour
20C
0 4 hours Time
148 hours
Figure 10.5 Increase in temperature at the heat source.
Recent advances in computational geomechanics 369
4 hours
0.0005
0.0000
0.0005
D
i
s
p
l
a
c
e
m
e
n
t

(
m
)
0.0 1.0 2.0 3.0
Time (s) ( 10
5
)
4.0 5.0 6.0
0.0010
0.0015
0.0020
1 hour
40 hours
150 hours
Figure 10.6 Displacementtime profle at point A.
48.87
Initial 1 hr 4 hr 10 hr
(a) Temperature (C)
40 hr 80 hr 150 hr
Initial 1 hr 4 hr 10 hr
(b) Excess Pore Water Pressure (kPa)
40 hr 80 hr 150 hr
Initial 1 hr 4 hr 10 hr
(c) Mean Efective Stress (kPa)
40 hr 80 hr 150 hr
Initial 1 hr 4 hr 10 hr
(d) Viscoplastic Volumetric Strain
(

= 0.35 10
4
)
40 hr 80 hr 150 hr
37.58
26.29
15.00
27.28
19.10
8.19
0.00
201.51
198.11
193.58
190.19
0.0053
0.0037
0.0016
0.0000
Figure 10.7 Distributions of temperature, excess pore water pressure, mean effective
stress, and viscoplastic volumetric strain.
370 Computational modeling of multiphase geomaterials
the change in temperature is reproduced. At the beginning of heating, the
temperature gradually increases near the heat source. Corresponding to
the temperature elevation, pore water pressure is generated near the heat
source. The maximum displacement in Figure 10.4 corresponds to the
development of large pore water pressure.
Since the viscosity of clay is dependent on temperature and the plastic
stress power changes the temperature of clay, it is necessary for the analy-
sis method of clay to consider the variation in temperature. In this chapter,
to address the temperature dependency of saturated clay and to deal with
the change in temperature induced by the plastic stress power and the
heat conduction, we proposed an elasto-thermo-viscoplastic constitutive
model for clay and developed a thermo-hydro-mechanical coupled fnite
element method with the constitutive model through the fnite element
formulation of the energy balance equation. Using the newly developed
analysis method, simulations of thermal consolidation were conducted.
We can simulate the thermal consolidation phenomena by use of the pro-
posed method.
10.2 SEEPAGEDEFORMATION COUPLED ANALYSIS
OF UNSATURATED RIVER EMBANKMENT USING
MULTIPHASE ELASTOVISCOPLASTIC THEORY
10.2.1 Introduction
In recent years, many natural disasters due to foods associated with tor-
rential rains, typhoons, hurricanes, and so forth have occurred around the
world. In many cases, river embankments have failed due to seepage and
overfow. A multiphase deformation analysis of a river embankment has
recently been carried out using an airsoilwater coupled fnite element
method considering the unsaturated seepage fow. A numerical model for
unsaturated soil has been constructed based on the porous media theory
and the elastoviscoplastic constitutive model shown in Chapters 2 and 5.
As for the stress variables in the formulation of unsaturated soil, we use
the skeleton stress and suction simultaneously. The skeleton stress has
been validated by experimental results (Oka, Kodaka, et al. 2010; Kimoto
et al. 2011). The skeleton stress is used in the constitutive model instead
of Terzaghis effective stress for saturated soil, and suction is incorporated
through the constitutive parameters of the model. An airsoilwater cou-
pled fnite element method is developed using the governing equations for
the three-phase soil based on the nonlinear fnite deformation theory, that
is, the updated Lagrangian method. Two-dimensional numerical analyses
of the river embankment under seepage conditions have been conducted for
the high river water level.
Recent advances in computational geomechanics 371
10.2.2 Governing equations and analysis method
For the governing equations, such as the equilibrium equation and the
mass balance equation for unsaturated soil, we use the method presented
in Chapter 2. In the numerical analysis, we use the updated Lagrangian
method described in Chapter 6.
10.2.3 Constitutive model for unsaturated soil
In the analysis, the saturated elastoviscoplastic model for the overstress-type
of viscoplasticity with soil structure degradation presented in Chapter 5
has been extended to unsaturated soil using the skeleton stress and includ-
ing the effects of suction (Oka, Kimoto, et al. 2010; Kimoto et al. 2011).
The collapse behavior of unsaturated soil is macroscopic evidence of the
structural instability of the soil skeleton and it is independent of the chosen
stress variables (Oka et al. 2008). In the present model, the collapse behav-
ior is described by the shrinkage of both the overconsolidated boundary
surface and the static yield surface due to the decrease in suction.
It is assumed that the strain rate tensor consists of an elastic stretching
tensor, D
ij
e
, and a viscoplastic stretching tensor, D
ij
vp
, as

= + D D D
ij ij
e
ij
vp
(10.66)
The elastic stretching tensor is given by a generalized Hookes law, namely,

= +

+

D
G
S
e
ij
e
ij
m
m
ij
1
2

3(1 )
(10.67)
where

S
ij

is the Jaumann rate of the deviatoric Cauchy stress tensor,

m

is the mean skeleton stress, G is the elastic shear modulus, e is the initial
void ratio, is the swelling index, and the superimposed dot denotes the
time differentiation.
10.2.3.1 Overconsolidation boundary surface
The overconsolidation boundary surface separates the normal consolidated
(NC) region, f
b
0, from the overconsolidated region, < f
b
0, as follows:

= +

= f M
b m
m
mb
ln 0
(0)
* *
(10.68)

( )( ) { }
=
ij ij ij ij (0)
* *
(0)
* *
(0)
*
1
2
(10.69)
372 Computational modeling of multiphase geomaterials
where
ij
*
is the stress ratio tensor = s
ij ij m
( / )
*
and (0) denotes the state
at the end of the consolidation, in other words, the initial state before the
shear test. M
m
*
is the value of =
ij ij
* * *
when the volumetric strain incre-
ment changes from negative to positive dilatancy, which is equal to ratio M
f

at the critical state.
mb
is the strain-hardening parameter, which controls
the size of the boundary surface.
The suction effect is introduced into the value of
mb
as

e
S s
P
P
mb ma
kk
vp
I d
i
c
c
exp
1
1 exp 1 =
+

+
(10.70)
where
kk
vp
is the viscoplastic volumetric strain, P
c
is the present suction
value, P
i
c
is a reference suction, and S
I
denotes the increase in the yield
stress when the suction increases from zero to reference value P
i
c
. S
d
con-
trols the rate of increase or decrease in
mb
with suction and
ma
is a
strain-softening parameter used to describe the degradation caused by
structural changes, namely,

( )
= + z
ma maf mai maf
( )exp
(10.71)

z zdt z
t
ij
vp
ij
vp
with
0

= =

(10.72)
in which
mai
and
maf
are the initial and the fnal values of
ma
, respec-
tively, while controls the rate of degradation with viscoplastic strain and

ij
vp
is the viscoplastic strain rate.
10.2.3.2 Static yield function
To describe the mechanical behavior of the soil at its static equilibrium
state, a Cam-clay type of static yield function is assumed as

= +

= f M
y
m
my
s
ln 0
(0)
* *
( )
(10.73)
where

M
*
is assumed to be constant in the NC region, but varies with the
current stress in the OC region (see Chapter 5).
The static strain-hardening parameter

my
s ( )
controls the size of the
static yield surface. In the same way as for the overconsolidation bound-
ary surface, parameter
my
s ( )
varies with the changes in suction as well as
Recent advances in computational geomechanics 373
with the changes in viscoplastic volumetric strain and structural degrada-
tion as

e
s s
P
P
my
s myi
s
mai
ma
kk
vp
I d
i
c
c
exp
1
1 exp 1
( )
( )
=

+

+ (10.74)
where
myi
s ( )
is the initial value of

my
s ( )
.
10.2.3.3 Viscoplastic potential function
The viscoplastic potential function is given by

= +

= f M
p
m
mp
ln 0
(0)
* *
(10.75)
where
mp
denotes the mean skeleton stress at the intersection of the visco-
plastic potential function surface and the

m axis.
10.2.3.4 Viscoplastic flow rule
The viscoplastic stretching tensor is an extention of Perzynas viscoplastic
theory and is given as

( )
=

D f
f
ij
vp
ijkl y
p
kl
(10.76)
where the symbol is defned as

( )
( )
=

>

f
f
f
f
ijkl y
ijkl y
y
y
0
; 0
; 0
(10.77)
in which
ijkl
denotes a material function for rate sensitivity. Herein, the
value of f
y
is assumed to be positive for any stress state in this model; in
other words, the stress state always exists outside the static yield func-
tion, so that viscoplastic deformation always occurs. Based on the experi-
mental results of constant strain-rate triaxial tests, the material function

ijkl
is defned by an exponential function (Kimoto and Oka 2005):

f C m M
ijkl y ijkl m
m
my
s

= +

( ) exp ln
(0)
* *
( )
(10.78)
374 Computational modeling of multiphase geomaterials
where m is the viscoplastic parameter that controls the rate sensitivity and
viscoplastic parameter C
ijkl
is a fourth-rank isotropic tensor given by

= + + C a b
ijkl ij kl ik jl il jk
( ),

C b C a b 2 , 3 2 ,
1 2
= = + (10.79)
where a and b are material parameters, which are related to the deviatoric
component C
1
and volumetric component C
2
of the viscoplastic parameter.
10.2.3.5 Constitutive model for pore water:
soilwater characteristic curve
Since saturation is a function of the suction, that is, the pressure head, the
time rate for saturation is given by

nS n
dS
d
d
d
d
dp
p
d
d
p
r
r
c
c
w
c
1
(10.80)
where =
V
V
w
is the volumetric water content, p
c
is the matrix suction
= p p p
c a f
( )
, = p
c
w
/ is the pressure head for the suction, and =

C
d
d

is the specifc water content. It is worth noting that the saturation is a
decreasing function of the suction.
The soilwater characteristic model proposed by Van Genuchten (1980)
is used

to describe the unsaturated seepage characteristics for which effec-
tive saturation, S
e
, is adopted as

=


=


S
nS
e
r
s r
r r
s r
(10.81)
where is the volumetric water content,
s
is the volumetric water content
in the saturated state, which is equal to porosity n, and
r
is the residual
volumetric water content retained by the soil at a large value of suction
head, which is a disconnected pendular water meniscus. For relatively large
and uniform sand particles, such as those of Toyoura sand,
r
becomes
zero, which is equal to common saturation.
In order to determine the soilwater characteristics, effective saturation,
S
e
, can be related to negative pressure head, , through the following relation:

( )
= +

S
e
n
m
1
(10.82)
Recent advances in computational geomechanics 375
where is a scaling parameter that has the dimensions of the inverse of ,
and n and m determine the shape of the soilwater characteristic curve.
The relation between n and m leads to an S-shaped type of soilwater
characteristic curve, namely,

= m
n
1
1
(10.83)
In the present analysis, we follow the Guide for Structural Investigations
of River Embankments (Japan Institute of Construction Engineering 2002),
and set = 2 and = n 4.
Specifc water content, C, used in Equation (10.80), can be calculated as

= C
d
d
n S S
s r e
m
e
m m
( ) ( 1)( ) (1 )
1/ 1/
(10.84)
The specifc permeability coeffcient, k
r
, which is a ratio of the unsatu-
rated to the saturated permeability, is defned by

= k S S
r e e
m m
{ 1 (1 ) }
1/2 1/ 2
(10.85)
Applying the aforementioned relations, we can describe the unsaturated
seepage characteristics. In the analysis, the unsaturated region is treated in
the following manner. In the embankment, the initial suction, that is, the
initial negative pore water pressure is assumed to be constant. Below the
water level, the pore water pressure is given by the hydrostatic pressure.
In the transition region between the water level and the suction constant
region, we assume that the distribution of the pore water pressure is lin-
early interpolated as in Figure 10.7. When the pressure head is negative,
the increase in the soil modulus due to suction is considered. The effec-
tive saturation and the saturation are calculated with Equations (10.81)
and (10.82) using the negative pressure head, that is, suction. Applying the
obtained effective saturation, specifc water content, C, is then calculated
by Equation (10.84) and specifc permeability, k
r
, is calculated by Equation
(10.85).
10.2.4 Simulation of the behavior of unsaturated
soil by elastoviscoplastic model
For unsaturated soil, we have applied the viscoplastic model to the experi-
mental results of drained triaxial tests on DL clay (silt). Figure10.8 indicates
the simulated results and the material parameters used in the simulation are
376 Computational modeling of multiphase geomaterials
listed in Table10.2. From this fgure, the extended model can describe the
effect of suction on the triaxial behavior.
10.2.5 Numerical analysis of seepage
deformation behavior of a levee
10.2.5.1 Analysis method
Weak forms of the governing equations are discretized in space and solved
by the fnite element method. In this formulation, an updated Lagrangian
method with the objective Jaumann rate of Cauchy stress is adopted, as pre-
sented in Chapter 6. The independent variables are the pore water pressure,
Table10.2 Material parameters
Initial suction s (kPa) 0 10 50 100
Initial void ratio, e 1.01 1.03 1.07 1.04
Compression index, 0.095 0.095 0.114 0.114
Swelling index, 0.0086 0.0086 0.0102 0.0102
Initial mean skeleton stress,
m0
(kPa) 200 205 217 222
Viscoplastic parameter, m 52.0
Viscoplastic parameter, C
1
(1/s) 1.0 10
11
Viscoplastic parameter, C
2
(1/s) 1.5 10
11
Stress ratio at critical state, M
*
m
1.00 1.00 1.00 1.00
Elastic shear modulus, G(kPa) 34800 36600 45100 46800
Compression yield stress,
mbi
(kPa) 200 205 217 222
Suction effect parameter, S
I
0.5
Suction effect parameter, s
d
0.25
Slope of stress path, q/
m
2.99 2.98 3.01 3.01
s = 10 kPa
s = 0 kPa
s = 50 kPa
s = 100 kPa
s = 50
s = 100
Experiment
Simulation
D
e
v
i
a
t
o
r

S
t
r
e
s
s

(
k
P
a
)
V
o
l
u
m
e
t
r
i
c

S
t
r
a
i
n

(
%
)
500
400
300
200
100
0
0 5 10 15
Axial Strain (%)
0
6
5
4
3
2
1
0
5 10 15
Axial Strain (%)
s = 10
s = 0
s = 100
s = 50
s = 10
s = 0
Figure 10.8 Stressstrain relations and volume change characteristics.
Recent advances in computational geomechanics 377
the pore air pressure, and the nodal velocity. In the fnite element formula-
tion, an eight-node quadrilateral element with a reduced Gaussian integra-
tion is used for the displacement, and four nodes are used for the pore water
pressure and the pore air pressure. The backward fnite difference method
is used for the time discretization.
10.2.5.2 Deformation during the seepage flow
Figure 10.9 shows the model of the river embankment and the fnite ele-
ment mesh used in the analysis with boundary conditions. The boundary
of the embankment is defned as follows. The boundary that contacts with
water or air is a drainage boundary if the pore fuid pressure is positive,
but the boundary between the embankment and the air is assumed to be
a no-water-fow boundary if the pore fuid pressure is negative to avoid
the fctitious water fow-in. The soil parameters used in the simulation
are listed on Tables10.3 and 10.4. The parameters are determined for DL
clay obtained by Oka, Kodaka, et al. (2010). The water level of the river
has been increased for 18 hours and then remains constant, as shown in
Water
fow/
Drainage
boundary
Suction from initial degree of
saturation
Embankment
Flow-controlled
boundary
Impermeable boundary
Foundation
Hydrostatic pressure
Water pressure = 0
Figure 10.9 Initial pressure head and fnite element mesh and boundary conditions.
378 Computational modeling of multiphase geomaterials
Figure10.10. Figure10.11 shows the distribution of the saturation with dif-
ferent times. The phreatic surface proceeds from the river side to the land
side of the embankment. Figures10.12 and 10.13 indicate the distributions
of the pore water pressure and the mean skeleton stress during the seepage
fow. The increases in pore water pressure associated with the advancement
of the seepage surface, and correspondingly with the mean skeleton stress,
decrease with the advancing seepage. Figure 10.14 indicates the develop-
ment of the air pressure with time. The magnitude of the air pressure is
relatively small but well simulated in the analysis.
Table10.3 Soil parameters for DL clay used in the analysis
Compression index, 0.136
Swelling index, 0.0175
Initial void ratio, e
0
1.05
Initial elastic shear modulus, G
0
32400 (kPa)
Initial mean skeleton stress,
mi
205 (kPa)
Stress ratio at critical state, M
*
m
1.01
Viscoplastic parameter, m 23.0
Viscoplastic parameter, C
1
1.0 10
11
(1/s)
Viscoplastic parameter, C
2
1.5 10
11
(1/s)
Structural parameter, 0.0
Suction parameter, S
I
0.2
Suction parameter, s
d
0.25
Water permeability for full saturation, k
s
W
1.0 10
5
(m/s)
Air Permeability for full saturation, k
s
G
1.0 10
3
(m/s)
Table10.4 Soil parameters for hydraulic properties
Parameter Value
Shape parameter for water permeability, a 3.0
Shape parameter for air permeability, b 2.3
Maximum saturation, S
r max
1.0
Minimum saturation, S
r min
0.0
Van Genuchten parameter, 0.2(1/kPa)
Van Genuchten parameter, n 0.0
Parameter for initial stress analysis Value
Youngs modulus, E 7900 (kPa)
Poisson ratio, 0.33
Internal friction angle, 30.0
Cohesion, c 0.0
Recent advances in computational geomechanics 379
The development of the accumulated plastic shear strain = d
p p
,
= d de de de
p
ij
vp
ij
vp
ij
vp
( ) ,
1/2
(viscoplastic deviatoric strain increment) during the
seepage fow is presented in Figure 10.15. The accumulated viscoplastic
shear strain is an indicator of the inelastic deformation history. From this
fgure, in an early stage of seepage fow, the strain develops near the surface
of the slope on the river side of the embankment, then a larger strain of more
than 1% occurs at the toe of the embankment on the land side of the levee.
Figure10.16 shows the distribution of the horizontal local hydraulic gra-
dient. In this simulation, the maximum value of 1.26 has been reached
inside the embankment on the river side after 18 hours. After 250 hours,
the maximum value of 0.66 of the horizontal gradient is observed at the toe
of the embankment.
Two-dimensional numerical deformation analyses of a river embankment
under seepage conditions have been conducted based on the mixture theory.
In modeling the unsaturated soil, we have adopted the skeleton stress and
the suction as stress variables. The use of the skeleton stress is consistent
with the defnition of partial stresses in the mixture theory. In the analysis,
we have adopted the fnite element method with an updated Lagrangian
scheme. For the constitutive model for soil, we used the viscoplastic model
that can take the degradation into account. We have numerically analyzed
the seepagedeformation coupled behavior of the river embankment with
an increase in the river water level. Using the proposed seepagedeformation
coupled three-phase analysis method, we can predict the behavior of levees
during the increase in the river water level. The numerical results indicate
that the large value of the horizontal hydraulic gradient corresponds to the
large deformation at the toe of the embankment.
Time
18 hr
R
i
v
e
r

W
a
t
e
r

L
e
v
e
l
6.0 m
3.0 m
9 hr 0
Figure 10.10 River water leveltime profle.
380 Computational modeling of multiphase geomaterials
Initial state
After 18 hrs
After 40 hrs
After 100 hrs
After 250 hrs
0.56 0.71 0.85 1.00
Figure 10.11 Distribution of the saturation with different times.
Recent advances in computational geomechanics 381
Initial state
After 18 hrs
After 40 hrs
After 100 hrs
After 250 hrs
15.30 24.10` 63.50 102.90
(kPa)
Figure 10.12 Distribution of the pore water pressure during the seepage fow.
382 Computational modeling of multiphase geomaterials
Initial state
After 18 hrs
After 40 hrs
After 100 hrs
After 250 hrs
4.73 29.63 54.52 79.42
(kPa)
Figure 10.13 Distribution of the mean skeleton stress during the seepage fow.
Recent advances in computational geomechanics 383
After 18 hrs
After 100 hrs
After 250 hrs
0.00000
0.00000
0.00000
0.00049
0.00247
0.00000
0.00098
0.00495
0.00001
0.00147
(kPa)
0.00742
(kPa)
0.00001
(kPa)
Figure 10.14 Distribution of air pressure with time.
384 Computational modeling of multiphase geomaterials
After 18 hrs
After 100 hrs
After 250 hrs
0.0000 0.0064 0.0127 0.0191
Figure 10.15 Distribution of the accumulated viscoplastic deviatoric strain.
Recent advances in computational geomechanics 385
REFERENCES
Adachi, T., and Oka, F. 1982. Constitutive equations for normally consolidated clay
based on elasto-viscoplasticity, Soils and Foundations, 22(4):5770.
Akagi, H., and Komiya, K. 1995. Constant rate of strain consolidation properties of
clayey soil at high temperature, Proc. Int. Symp. Compression Consolidation
of Clayey Soils, Hiroshima, H. Yoshikuni and O. Kusakabe eds., Balkema,
Rotterdam, 1:38.
Baldi, G., Hueckel, T., and Pellegrini, R. 1988. Thermal volume change of the mineral-
water system in low-porosity clay soils, Can. Geotech. J., 25:807825.
After 18 hrs
After 40 hrs
After 100 hrs
After 250 hrs
0.00 0.42 0.84 1.26
Figure 10.16 Distribution of the horizontal local hydraulic gradient.
386 Computational modeling of multiphase geomaterials
Biot, M. A. 1956. Theory of propagation of elastic waves in a fuid-saturated porous
solid, J. Acoust. Soc. Am., 28(2):168178.
Boudali, M., Leroueil, S., and Srinivasa Murthy, B. R. 1994. Viscous behaviour of
natural clays, Proc. 13th Int. Conf. on SMFE, New Delhi, India, 411416.
Campanella, R.G., and Mitchell, J.K. 1968. Infuence of temperature variations on
soil behavior, J. SMFE, ASCE, 94, 3:709734.
Cattaneo, C. 1958. A form of heat conduction equation which eliminates the para-
dox of instantaneous propagation, Comp. Rend, 247:431433.
Cekerevac, C., and Laloui, L. 2004. Experimental study of thermal effects on
the mechanical behaviour of a clay, Int. J. Numer. Anal. Meth. Geomech.,
28:209228.
Cleall, P.J., Melhuish, T.A. and Thomas, H.R. 2006. Modelling the three-dimensional
behaviour of a prototype nuclear waste repository, Eng. Geo., 85:212220.
Delage, P., Sultan, N., and Cui, Y.J. 2000. On the thermal consolidation of Boom
clay, Can. Geotech. J., 37:343354.
Ehlers, W., Graf, T., and Ammann, M. 2004. Deformation and localiza-
tion analysis of partially saturated soil, Comp. Meth. Appl. Mech. Eng.,
193:28852910.
Higo, Y. 2003. Instability and strain localization analysis of water-saturated clay by
elasto-viscoplastic constitutive models, PhD thesis, Kyoto University, Japan.
Higo, Y., Oka, F., Kodaka, T., and Kimoto, S. 2006. Three-dimensional strain local-
ization of water-saturated clay and numerical simulation using an elasto-
viscoplastic model, Philos. Mag., 86(2122):32053240.
Japan Institute of Construction Engineering. 2002. Guide for structure investiga-
tions of river embankments (in Japanese).
Jommi, C. 2000. Remarks on the constitutive modelling of unsaturated soils, In
Experimental Evidence and Theoretical Approaches in Unsaturated Soils,
A.Tarantio and C. Mancuso, eds., Balkema, Rotterdam, 139153.
Kim, Young Seok. 2004. Elasto-viscoplastic modeling and analysis for cohe-
sive soil considering suction and temperature effects, PhD thesis, Kyoto
University.
Kimoto, S., and Oka, F. 2003. An elasto-viscoplastic model for clay consider-
ing destructuralization and prediction of compaction bands, Proceedings
of International Workshop on Prediction and Simulation Methods in
Geomechanics, F. Oka, I. Vardoulakis, A. Murakami, and T. Kodaka, eds.,
Athens, Greece, TC34 of ISSMGE, 6568.
Kimoto, S., and Oka, F. 2005. An elasto-viscoplastic model for clay considering
destructuralization and consolidation analysis of unstable behavior, Soils and
Foundations, 45(2):2942.
Kimoto, S., Oka, F., Fukutani, J., Yabuki, T., and Nakashima, K. 2011. Monotonic
and cyclic behavior of unsaturated sandy soil under drained and fully un drained
conditions, Soils and Foundations, 51(4):663681.
Kimoto, S., Oka, F. and Fushita, T. 2010. A chemo-thermo-mechanically coupled
analysis of ground deformation induced by gas hydrate dissociation, Int. J.
Mech. Sci., 52:365376.
Kimoto, S., Oka, F., Fushita, T., and Fujiwaki, M. 2007. A chemo-thermo-mechani-
cally coupled numerical simulation of the subsurface ground deformations due
to methane hydrate dissociation, Comput. Geotech, 34(4):216228.
Recent advances in computational geomechanics 387
Kimoto, S., Oka, F., and Higo, Y. 2004. Strain localization analysis of elasto-
viscoplastic soil considering structural degradation, Comput. Meth. Appl.
Mech. Eng., 193:28452866.
Kimoto, S., Oka, F., Fukutani, J., Yubuki, T. and Nakashima, K. 2011. Monotonic
and cyclic behavior of unsaturated sandy soil under drained and fully und-
rained conditions, Soils and Foundations, 51(4):663681.
Komine, H., and Ogata, N. 2002. Hydraulic properties of bentonite buffer and
backfll materials and simplifed evaluation, J. Geotech. Eng., JSCE, 708/III-
59:133144 (in Japanese).
Laloui, L., Nuth, M., and Vulliet, L. 2006. Experimental and numerical investi-
gations of the behaviour of a heat exchanger pile, Int. J. Num. Anal. Meth.
Geomech, 30(8):763781.
Leroueil, S., Kabbaj, M., Tavenas, F., and Bouchard, R. 1985. Stress-strain-strain
rate relation for the compressibility of sensitive natural clays, Gotechnique,
35(2):159180.
Leroueil, S., and Marques, M.E.S. 1996. State of the art: Importance of strain rate
and temperature effects in geotechnical engineering, Geotechnical Special
Publication No. 61 ASCE, T.C. Sheahan and V. Kaliakin, eds., 160.
Mller, I., and Ruggeri, T. 1998. Rational Extended Thermodynamics, 2nd ed.,
Springer, New York, 914.
Oka, F., Feng, H., Kimoto, S., Kodaka, T., and Suzuki, H. 2008. A numerical simu-
lation of triaxial tests of unsaturated soil at constant water and constant air
content by using an elasto-viscoplastic model, Proceedings of 1st European
Conference on Unsaturated Soils, D. Toll and S.J. Wheeler, eds., Taylor &
Francis/CRC Press, 735741.
Oka, F., Higo, Y., and Kimoto, S. 2002. Effect of dilatancy on the strain localization
of water-saturated elasto-viscoplastic soil, Int. J. Solids Struct., 39:36253647.
Oka, F., Kimoto, S., Takada, N., Gotoh, H., and Higo, Y. 2010. A seepage-deforma-
tion coupled analysis of an unsaturated river embankment using a multiphase
elasto-viscoplastic theory, Soils and Foundations, 50(4):483494.
Oka, F., Kodaka, T., Kimoto, S., Kato, R., and Sunami, S. 2007. Hydro-mechanical
coupled analysis of unsaturated river embankment due to seepage fow, Key
Eng. Mater., 340341:12231230.
Oka, F., Kodaka, T., Suzuki, H., Kim, Y., Nishimatsu, N., and Kimoto, S. 2010.
Experimental study on the behavior of unsaturated compacted silt under tri-
axial compression, Soils and Foundations, 50(1):2744.
Oka, F., Yashima, A., and Leroueil, S. 1997. An elasto-thermo-viscoplastic model for
natural clay and its application, Proceedings of 6th International Symposium
on Numerical Models in Geomechanics, S. Pietruszczak and G. Pande, eds.,
Balkema, Rotterdam, 105110.
Oka, F., Yashima, A., Sawada, K., and Aifantis, E.C. 2000. Instability of gradient-
dependent elastoviscoplastic model for clay and strain localization, Comput.
Meth. Appl. Mech. Eng., 183:6786.
Peirce, D., Shih, F., and Needleman, A. 1984. A tangent modulus method for rate
dependent solids, Comput. Struct., 18(5):845887.
Schrefler, B. A. 1995. F.E. in environmental engineering-coupled thermo-
hydro-mechanical processes in porous media including pollutant trans-
port, Arch. Comput. Meth. Eng., 2(3):154.
388 Computational modeling of multiphase geomaterials
Suga, G., Hashimoto, H., and Ishikawa, T. 1984. Final report on the research of over-
fow levee (Interpretation edition), Report of the Civil Engineering Research
Institute, 2074 (in Japanese).
Sulem, J., and Famin, V. 2009. Thermal decomposition of carbonates in fault
zone:slip-weaking and temperature-limiting effects, J. Geophys. Res., 114,
B03309, doi10.1029/2008JB006004.
Sultan, N., Delage, P., and Cui, Y.J. 2002. Temperature effects on the volume change
behaviour of boom clay, Eng. Geol., 64:135145.
Van Genuchten, M. T. 1980. A closed-form equation for predicting the hydraulic
conductivity of unsaturated soils, Soil Sci. Soc. Am. J., 44:892898.
Vardoulakis, I. 2002. Dynamic thermo-poro-mechanical analysis of catastrophic
landslides, Gotechnique, 52(3):157171.
Yashima, A., Leroueil, S., Oka, F., and Guntoro, I. 1998. Modelling temperature and
strain rate dependent behavior of clays: One-dimensional consolidation, Soils
and Foundations, 38(2):6373.
GEOTECHNICAL ENGINEERING
Computational Modeling of Multiphase Geomaterials discusses how
numerical methods play a very important role in geotechnical engineering
and in the related activity of computational geotechnics. It shows how
numerical methods and constitutive modeling can help predict the behavior of
geomaterials such as soil and rock.
After presenting the fundamentals of continuum mechanics, the book explores
recent advances in the use of modeling and numerical methods for multiphase
geomaterial applications. The authors describe the constitutive modeling of
soils for rate-dependent behavior, strain localization, multiphase theory,
and applications in the context of large deformations. They also emphasize
viscoplasticity and watersoil coupling.
Features
Explains how to predict the behavior of geomaterials
Contains the governing equations for multiphase geomaterials
Discusses the constitutive modeling of multiphase geomaterials,
including elastoplastic and elastoviscoplastic models
Presents numerical methods, such as the fnite element method,
for analyzing geomaterials
Covers the latest developments in geomechanics, including the
deformation-seepage fow coupled analysis of an unsaturated
river embankment
Drawing on the authors well-regarded work in the feld, this book provides
you with the knowledge and tools to tackle problems in geomechanics. It gives
you a comprehensive understanding of how to apply continuum mechanics,
constitutive modeling, fnite element analysis, and numerical methods to
predict the behavior of soil and rock.
ISBN: 978-0-415-80927-6
9 780415 809276
90000
w w w . s p o n p r e s s . c o m
Y132935
C
O
M
P
U
T
A
T
I
O
N
A
L

M
O
D
E
L
I
N
G

O
F
M
U
L
T
I
P
H
A
S
E

G
E
O
M
A
T
E
R
I
A
L
S
FUSAO OKA
SAYURI KIMOTO
COMPUTATIONAL
MODELING OF
MULTIPHASE
GEOMATERIALS
O
K
A
K
I
M
O
T
O
A S P ON PR E SS BOOK
Y132935 cvr mech.indd 1 6/7/12 10:11 AM