Sie sind auf Seite 1von 16

BORJA, R. I.

& KAVAZANIIAN,
E. (1985). Giotechnique

A constitutive

35, No. 3, 283-298

model for the stress-strain-time


wet clays
R.

I. BORJA*

and

A constitutive
model is developed
to describe
the
stress-train-time
behaviour of wet clays subjected to
three-dimensional
states of stress and strain.
The
model is based on Bjerrums
concept of total strain
decomposition
into an immediate
(time-independent)
part and a delayed (time-dependent)
part generalized
to three-dimensional
situations. The classical theory of
plasticity
is employed
to characterize
the timeindependent
stress-train
behaviour
of cohesive soils
using the ellipsoidal yield surface of the modified Cam
Clay model presented
by Roscoe and Burland.
The
time-independent
strain is divided into an elastic part
and a plastic part. The plastic part is evaluated
using
the normality
condition
and the consistency
requirement on the yield surface. The time-dependent
(creep)
component
of the total strain is evaluated by employing the normality rule on the same yield surface as in
the time-independent
model and the consistency
requirement
which requires
that the creep strain rate
reduces to phenomenological
creep rate expressions
for isotropic
or undrained
triaxial stress conditions.
The mathematical
characterization
of the constitutive
model is given by the constitutive
equation expressed
in a form suitable for direct numerical implementation
(i.e. finite element
formulation).
The required
soil
parameters
are easily obtainable
from conventional
laboratory
tests. The constitutive
equation is shown to
predict accurately
the stress-train-time
behaviour
of
an undisturbed
wet clay in triaxial and plane strain
stress conditions.
Un modele constitutif
a CtC dtveloppt
pour dtcrire le
comportement
contrainte-dtformation
dans le temps
des argiles humides soumises a des Ctats tridimensionnels de contraintes
et de deformations.
Le modtle
est base
sur une generalisation
aux conditions
tridimensionnelles
du concept
de Bjerrum
de la
decomposition
de la deformation
en une partie
immediate
(indtpendante
du temps) et une partie
retardee
(dependante
du temps). On utilise la thtorie
classique de la plasticite pour caracteriser
le comportement contraintedtformation
indtpendant
du temps
des sols cohtrents,
en se servant
de la surface
decoulement
ellipsoidale
du modele modifie de largile de Cam present& par Roscoe
et Burland.
La
deformation
independante
du temps est subdivisee en

Discussion
on this Paper closes on 1 January
1986.
For further details see inside back cover.
* Department
of Civil Engineering,
Stanford University.
283

E.

KAVAZANJIAN,

behaviour

of

JR*

une partie tlastique et une partie plastique. On tvalue


la partie plastique dapris la condition de normalitt
et
la consistance
exigee sur la surface dtcoulement.
La
partie dependant
du temps (fluage) de la deformation
totale est &al&e daprts de la regle de normalite sur
la m&me surface decoulement
que dans le cas du
modble independant
du temps, en employant
en m&me
temps Iexigence de consistance
qui veut que la vitesse
de la deformation
de fluage se rtduise a des expressions phtnomenologiques
de vitesse de fluage pour des
conditions
de contrainte
triaxiales
isotropiques
ou
non-drainees.
La caracterisation
mathtmatique
du
modele constitutif
est donne par Itquation
constitutive exprimee dans une forme qui est utilisable pour
lapplication
numtrique
directe (cest-a-dire
pour la
formulation
delements
finis). Les paramttres
du sol
necessaires sobtiennent
facilement a partir dessais de
laboratoire
conventionnels.
On
dtmontre
que
lequation
constitutive
predit de facon precise le comportement
contraintedltformation
dependant
du
temps dune argile non-remaniee
dans des conditions
contraintedeformation
triaxiales et planes.

KEYWORDS:
clays; constitutive
relations;
deformation; finite elements; plasticity; time dependence.
INTRODUCTION
Consideration
of the time dependence
of the
stress-strain
behaviour
of cohesive
soils is important
in the
evaluation
of long-term
performance
of geotechnical
structures
such as

embankments,
tunnels
and excavations.
The
deformations
and pressures
that develop with
time are due to both hydrodynamic
lag and
creep effects.
Numerous
constitutive
models
have been
proposed
(Bonaparte,
1981; Kavazanjian
&
Mitchell, 1980; Pender, 1977; Roscoe, Schofield
& Thurairajah,
1963; Roscoe & Burland, 1968;
Tavenas & Leroueil, 1977) and numerically implemented
(Bonaparte,
198 1; Johnston,
198 1)
to characterize
yielding
of wet clays using
effective stress quantities and the classical theory
of plasticity.
The deformations
predicted
by
these models are analogous to those corresponding to initial undrained
loading and primary
consolidation
or to those which would immediately
develop at the instant the imposed

284

BOFUA

AND KAVAZANJIAN

NOTATION

hyperbolic stress-strain
parameter
Singh-Mitchell
creep parameter
hyperbolic stress-strain
parameter
stress-strain
tensor
virgin compression
index, log,,, scale
swelling-recompression
index, log,, scale
secondary
compression
coefficient,
log,,
scale
deviator stress level
void ratio, natural number 2.71828..
e I_= 1on the isotropic consolidation
curve
e\,=, on the critical state line
initial tangent modulus
elliptical yield surface
first invariant of ( 1
second invariant of ( )
(i, j) component
of k
permeability
tensor
bulk modulus
Singh-Mitchell
creep parameter
slope of the critical state line
volumetric stress, $akk
preconsolidation
pressure
equivalent preconsolidation
pressure, poG
ZLiatoric
stress, (3/v5)7,,,
undrained shear strength
Duncan and Chang corrector parameter
time
deviatoric age
instant deviatoric time, usually set to unity
time for 100% consolidation
volumetric age
instant
volumetric
time, usually set to
unity
ith (Cartesian) component
of u

loading
is applied,
in the absence
of hydrodynamic
lag. No deformation
is predicted
by
these models under constant effective stress conditions.
Soil materials,
however,
also exhibit timedependent
or creep deformations
under a state
of constant effective stress. Volumetric compression at a constant effective stress, for example,
may persist
indefinitely
beyond
the primary
phase in a typical isotropic or one-dimensional
consolidation
test. Plastic axial deformation
at a
constant deviator stress level may also develop
indefinitely
with time in a conventional
undrained triaxial creep test.
These
creep
deformations
can become
a
major contributor
to the total deformation
for
young clays (Bjerrum, 1967; Borja, 1984). De-

displacement
field vector
Singh-Mitchell
creep parameter
deviatoric strain
unit weight of water
octahedral
deviatoric strain
Kronecker
delta
strain, compression
positive
major principal strain
intermediate
principal strain
minor principal strain
axial strain
axial creep strain rate
volumetric strain
octahedral
normal strain, ~~~~
strain tensor
creep strain rate tensor
stress ratio q/p
recompression
index, In scale
virgin compression
index, In scale
elastic shear modulus
pore pressure field function
stress, compression
positive
major principal stress
intermediate
principal stress
minor principal stress
vertical stress
octahedral
normal strain
(i, j) component
of the (Cauchy) stress
tensor
stress relaxation rate
octahedral
shear stress
associative flow rule proportionality
factor
effective friction angle
factor
secondary
compression
coefficient,
In
scale

their significance, however, little work has


been done to incorporate
time-dependent
deformations into effective stress-based
constitutive
models for application in the design and analysis
of full-scale geotechnical
structures.
This Paper presents a recently developed constitutive model to characterize
the stress-straintime behaviour
of wet clays in general threedimensional
stress and strain conditions.
The
constitutive
equation is formulated
using tensor
notation and requires material constants obtainable from conventional
laboratory
tests, e.g.
triaxial
compression,
isotropic
or
onedimensional
consolidation
and undrained triaxial
creep tests. Since the stress condition in a triaxial test is seldom realized in practical field situations, the predictive capability of the constitutive

STRESS-STRAIN-TIME

BEHAVIOUR

model not only in axisymmetric


(torsionless)
conditions but also in plane strain applications is
investigated
using a finite element program.
THEORETICAL

BASES

Separation of immediate
and delayed
strains
The concept of total strain decomposition
into
an immediate
(time-independent)
part and a
delayed
(time-dependent)
part proposed
by
Bjerrum
(1967) for one-dimensional
compression is illustrated
schematically
in Fig. 1. The
plots of void ratio e and effective vertical stress*
u, versus time t show an immediate component,
occurring at the same instant that the external
load is applied, and a delayed component
that
persists indefinitely
with time. This decomposition scheme does not consider the influence of
hydrodynamic
lag.
Superimposed
in Fig. 1 is Taylors (1948)
description
of consolidation
using a curve consisting of two phases: a primary consolidation
phase for tst,
in which excess pore pressures
dissipate
and a secondary
compression
phase,

OF WET CLAYS

285

governed
by the secondary
compression
coefficient C, (=$ In 10, where $ is the secondary
compression
coefficient in the In t scale) for t 2
t,. The time t, corresponds
to end of pore
pressure dissipation,
or to 100% primary consolidation.
During secondary
compression,
the
soil continues to deform at a constant effective
stress.
Bjerrums
graphical
representation
of the
effect of delayed compression
on the void ratiolog vertical effective stress diagram for a onedimensional
consolidation
test is shown in Fig.
2. The diagram consists of contours of constant
time, each contour
representing
compression
after an equal duration of sustained loading. By
assuming a constant C,, the constant time lines
in Fig. 2 become equally spaced.
The effect of continued
volumetric
compression at a constant vertical effective stress for a
typical soil element portrayed in Fig. 2 is for the
soil to exhibit apparent stiffening and to develop
a quasi-preconsolidation
pressure (T, during subsequent loading.
Modified

Cam

Clay as a time-independent

plas-

ticity model
The Cam Clay theory postulates the existence
of a unique state boundary surface, XXYY in
Fig. 3, representing
the limit of all possible
states of a wet clay in the void ratio evolumetric stress p-deviatoric
stress 4 space, in
which the stress parameters
p and q are defined

Instant
compresslot

I-Consohdatmn-;P
t

(b)
Fig. 1. Definitions of primary and secondary consolidation and immediate and delayed compression
* All stress quantities used in this Paper are effective
stress quantities.
To simplify notation, the prime symbol commonly
used to differentiate
effective stresses
from total stresses will be omitted.

Fig. 2. Bjerrums
pression

model

for one-dimensional

com-

286

BOFUA AND KAVAZANJIAN

consolidation

line

e
q = MP
line
/

/
Undrained

recompresslon

line

-Gr~nn;sotroplc
consolidation

line

line

Fig. 3. State boundary surface


as follows
p =~(o,+uz+u3)

(1)

q =u1-r73

(2)

where ur, uZ and u3 are the major, intermediate


and the minor principal effective stresses respectively.
The state boundary surface intersects the e-p
plane along the virgin isotropic
consolidation
line YY defined by the strain-hardening
equation
e,-e-Alnp,=O
(3)
where p= is the preconsolidation
pressure,
A is
the virgin compression
index and e, = elP,=r is
the void ratio at p = pc= 1-O on the isotropic
consolidation
curve. The same surface intersects
the critical state line, XX in Fig. 3, defined by
the two equations
q=MP

(4)

e=e,-Alnp

where ec= e(,=, (a quantity which can be back


calculated by knowing e,) and M is a constant
which is related to the effective friction angle 4
by the expression
M = 6sin4
3-sin$
Any

state

point

on

the

critical

state

line

is

characterized
by plastic shear deformation
instability without volume change.
The modified Cam Clay model uses an ellipsoidal yield surface of the form
F = F(Uii, p,) = $+

p(p - p,) = 0

(6)

which has its centre on the (p, q) plane at


(ip,, 0). This yield surface is the projection
of
the arc in the e-p-q space defined by the intersection of the state boundary
surface and the
elastic wall XYX in Fig. 3.
The undrained stress path through pc is given
by the constant void ratio line
I--r/h

(7)

where K is the swelling-recompression


index.
Equation (7) reduces to equation (6) when K =
0.
Roscoe & Burland (1968) demonstrated
the
capability of the above model to represent
the
time-independent
plasticity behaviour
of wet
clays in triaxial compression,
triaxial extension
(provided that rupture in extension is considered
appropriately)
and plane strain stress conditions.
Overall, the model requires four material constants, K, A, M and e,, obtainable from isotropic
(or one-dimensional)
consolidation
and triaxial
shear tests.

STRESS-STRAIN-TIME

BEHAVIOUR

The deformations
predicted by this model are
analogous to those corresponding
to undrained
loading and primary consolidation,
or to those
which would immediately
develop at the instant
that the imposed load is applied, in the absence
of hydrodynamic
lag.
Inclusion

of creep deformations

Kavazanjian
& Mitchell
(1980) postulated
that time-dependent
soil deformations
can be
divided into distinct, but interdependent,
volumetric and deviatoric
contributions.
They considered
these
creep
deformations
using the
following phenomenological
volumetric and deviatoric expressions
for creep.
Volumetric creep. Volumetric
creep deformations were based on Taylors (1948) secondary
compression
equation.
The accumulated
volumetric creep E, in time period At in a typical
isotropic (or one-dimensional)
consolidation
test
is the integral (consult Fig. 2)
f+At

t-

E -

I
t

ICI
(l+e)t

dt

where t, is the volumetric


age, relative to an
initial reference
time (t.,)i, of the state point
associated-with
the constant time contour on the
e-ln p plane similar to that shown in Fig. 2. If
there is no primary loading or unloading,
the
soil ages linearly with natural time during this
period; hence At = At,.
Deviatoric
creep. Deviatoric
creep deformations were based on the Singh-Mitchell
creep
function
(1968). The accumulated
axial creep
strain E, m time period At in a typical undrained triaxial creep test is the integral
E,=,.,+AtAeG[~]m

dt

(9)

where A, di and m are the Singh-Mitchell


creep
parameters,
(f& is an initial reference time, td
is the deviatoric
age ,relative to (f& and D =
(a1 - as)/(ol - a&
is the deviator stress level. If
there is no deviatoric loading or unloading, the
soil ages linearly with natural time during this
period; hence At = At,.
In general, equ_ation (9) holds for stress levels
of about 0.2< D < 0.8, overestimates
cat for
stress states near the isotropic condition (D + 0)
and underestimates_
F, for stress states near the
failure condition (D + 1.0).
Kavazanjian
& Mitchell
(1980) further
assumed
that
the
time-independent
deviator
stress-axial
strain diagram corresponding
to the
reference
time (t& in equation
(9) is given by
Kondners (1963) hyperbola normalized with respect to the confining stress (Ladd & Foott,

OF WET CLAYS

287

1974) (T, as follows


lS-(T3=

E,fl,
a+be,

(10)

where a and b are hyperbolic


parameters
obtained from conventional
triaxial shear tests. A
third parameter
Rf, termed the failure ratio by
Duncan & Chang (1970), is introduced
to force
equation (10) to pass through the failure point at
an actual finite strain.
It will be shown subsequently
that the trace
of the Cam Clay yield surface corresponding
to
reference
time (tJi on the deviator stress-axial
strain plane for soils in triaxial compression
can
be described,
approximately,
by the hyperbolic
curve (equation (10)) when (t,&= (t,)i.
DEVELOPMENT
EQUATION

OF THE

CONSTITUTIVE

Definitions

of stress and strain parameters


general
threethe
Representation
of
dimensional
state of stress requires appropriate
definitions
of stress and strain parameters
to
encompass
all the components
of the stress and
strain tensors gli and E,+
The volumetric effective stress is defined as
P = coct = &

= +I,

(11)

where I, is the first invariant of the stress tensor


c,i
The deviatoric stress q is defined as
4=5

T0,t = [f(crd)ii(~d)i,]12= (311,)


(12)

where ILd is the second invariant


of the deviatoric stress tensor (uJii. In the triaxial stress
condition,
this definition for q reduces to ul(TV, equation (2). In the undrained
plane strain
condition,
where u2 = f(u, + u3), the definition
for q reduces to A((+,-u&2.
The volumetric strain is given by
E, = 3&,,, = Ekk = I,

(13)

where I, is the first invariant of the strain tensor


Eij.
The deviatoric strain y is defined as
r=&&..=

r%&(GJkJ

= (411EdY
(14)

where II,, is the second invariant


of the deviatoric part of ekl. In the undrained
triaxial
condition
where
the
principal
strains
are
the definition for y reduces to
(E,, -;Ea, -I &,
E,.
In the undrained
plane strain condition

288

BORJA

AND

KAVAZANJIAN

which is easily verified


struction in Fig. 4.
Solving for ApJp,

from

geometrical

con-

(19a)

f...

(19b)

where I/J is the secondary


compression
coefficient, while equation (19b) is the binomial series
expansion of equation (19a). Taking the limit of
ApJAt, as At,-+ 0
aPc _-- II, Pi
et,- h-K
t,
Fig. 4. Development

of quasi-preconsolidatioion

where e3 = -cl and F~ = 0, the definition


reduces to 2&J&.

for y

Growth of preconsolidation
pressure
Assume that the size pF of the yield surface
given by the function
Pc = Pc(.%> tv)

is

(15)

Equation
(15) states that the preconsolidation
pressure pc does not only grow because of timeindependent
strain hardening
but also expands
with time, resulting
in the development
of a
state of quasi-preconsolidation.
Consider
the consolidation
curve of Fig. 4.
Along any line of constant t,, say at tV= (tJi, the
time-independent
plastic volumetric strain increment is given by

(20)

Hence, the rate of growth of pc decreases as the


soil ages.
Equations
(17) and (20) are respectively
the
hardening
rules
that
describe
the
timeindependent
and the time-dependent
components of the rate of growth of the size p= of the
yield surface.
General formulation
Let the strain rate tensor Ekl be decomposed,
i.e.
Et, = &I + &, + &,
(21)
where the superscripts
e and p denote
the
time-independent
elastic and plastic parts respectively,
while the superscript
t denotes the
time-dependent
(creep) plastic part. In principle,
the above decomposition
employs
Bjerrums
scheme of separating
the total strains into immediate
and delayed
components
(Bjerrum,
1967).
Applying the associative flow rule on &,
dF
i,,P = l#laukI

in which equation
(16b) is the Taylor series
expansion of equation (16a). Taking the limit of
Ap,/h~, as As.,--+ 0,
aPC

a&,

where 4 is a proportionality

factor.

Setting

aF
FkkD= 4 ~

P=

(17)

I= k

GW

aakk

l+e
A--K

(22)

or
,,%,~

Again, consider the consolidation


curve of Fig.
4. If a soil element at a state point at A, with
volumetric
age t,, volumetrically
creeps
in
time period At = At, at a constant effective stress,
the soil would shrink by an amount

Rewriting

equation
& = (c;J1&,j

(23b)

(21) explicitly
aF

i-4 -+

afl,,

EL,l

(24)

or
cri, = CRk,

(25)

STRESS-STRAIN-TIME

BEHAVIOUR

where c$, is the fourth-rank


elastic stress-strain
tensor.
Consider the plastic potential F = F(p, q, p=) =
F(u,,, p,) given by equation (6). The consistency
requirement
on F demands that the time rate of
chance

289

OF WET CLAYS

by defining
riately.

the

tensors

ciikl and

Evaluation
of derivatives
The following are the derivatives
to eauation (6))
aF

(26)

of F (refer

(36)

-=2p-p,
ap

where

ai, approp-

(274

aF
-=2
aq

2q

(27b)

Gc=-p

(37)

aF

in which the symbol lo implies differentiation


with the quantity inside the parentheses
held
fixed, while the terms in equation
(27b) are
obtained on substitution
of equations
(17) and
(20) in equation (27a).
Substituting equations (25) and (27b) in equation (26) and solving for 4
aF

au,,

aF
-=
auzi

where
1

aF

Substituting
simplifying

aF
&Tkt

aa,,

in

(Tii= cy&

and

(T,, = c%,&lf+x

&,

(39)

(40)

aF ap
--+-ap au,,

aF aq
aq au,,

(414

aFaFl+e
aPc aP A -

PC

(29)

(25)

and

equation

(41b)

. t
- Ufj

(30)

where
cT[ is the fourth-rank
elasto-plastic
stress-strain
tensor and I?,, is the stress relaxation rate given respectively
by

c~,=c~kl-x

8P
_=1
aui,

1
a4
3
=ui, - - a!&sii
3
)
aa,
2q (

(28)
-=-_~k,-____

(11) and (12) for p and q re-

where IS,~is the Kronecker


delta.
Thus, the normal at any point on F is given by

aFPc 4
apt t, h - K 1

-&,(&-&y+---

By definitions
spectively

(38)

aF

aF

C:pq~gCrskl

e
(31)

p4 IS

@P,
9
--~
ap, t, A -

Gkl g

(32)

Elastic soil constants


The elastic stress-strain
tensor c& requires
at least two independent
elastic material properties for complete definition. Two properties
are
adequate
by assuming
homogeneity,
material
isotropy
and major/minor
symmetries
in c&.
The two elastic constants chosen herein are the
bulk modulus K and the shear modulus pe.
The elastic bulk modulus is obtained from the
volumetric Cam Clay model by noting that along
the swelling-recompression
line

(424
It can be seen that c$ has the major symmetry
if and only if czkl has the major symmetry.
When creep is ignored,
&t= 0, Ic,= 0 and
hence, ai, = 0. If the response
of the soil is
perfectly elastic
uij = cp&& - &
where

c
a:, = Ciik,Eklf

(33)
(34)

Thus, whatever
the behaviour
of the material,
the creep-inclusive
constitutive
equation
for
wet clay can always be written in the rate form

or

(42b)
Thus the bulk modulus increases linearly with
the volumetric
stress p, necessitating
that p be
always positive (i.e. always compressive).
Assuming that the trace of the modified Cam
Clay yield surface on the q-y plane (or the
deviator stress-axial
strain plane in triaxial stress
condition) is a hyperbola of the form (cf. equation (10))
YPC R
(43)
q=f
afby

BOWA

290

AND

KAVAZANJIAN

where cp is a proportionality
factor and F is the
equivalent
yield surface evaluated
using equation (6), whose size p0 is the equivalent preconsolidation pressure given by

lsotrop~cconsolidatw

which equals pc for normally consolidated


soils
(refer to Fig. 4).
The magnitude
of Fklf is obtained
by degenerating this tensor either to an isotropic tensor or a triaxial tensor and appropriately
scaling
cp(dF/duk,)using either the C, creep law for the
isotropic condition, or the Singh-Mitchell
creep
equation for the triaxial stress condition. These
two methods are herein called volumetric scaling
and deviatoric scaling respectively.

4 = MP
(4

Volumetric scaling
The magnitude
of the trace of Etlf along the
volumetric axis p is a measure of the volumetric
creep rate for the soil. The volumetric
part of
&, is
1
aF
(E,f)k, = f&Q Id=-(P-&d
(47)
3 aa,,

hyperbola

Recalling that the rate of secondary compression


is governed by the index C, (+ in the natural
logarithm scale)

y from hyperbola
Ibl

Fig. 5. Evafuation
ages

of (a) volumetric

and (b) deviatoric

or

t $E=
JI
aa,, (1-t e)t,

(48)

(49)
the elastic shear modulus is back calculated from
the initial tangent modulus of the hyperbolic
curve as follows

dq
G
or

I =--P&
a
-v=

_ 3 _=33CLe
dr,,t
2d Yoct

P&
P -3a
e

(44a)

(44b)

Thus, the shear modulus increases linearly with


the size pc of the yield surface, necessitating
that
the soil be initially preconsolidated
to develop a
non-zero elastic shear stiffness.
Creep strain rate
The quantities in equation (32) that remain to
be evaluated
are the volumetric
age t, and the
components
of the creep strain rate tensor &lf.
The direction
of &lf is obtained
from the
normality
rule applied to the equivalent
yield
surface associated with the stress state (p, q) as
follows
~-

Eklf=qE

dukl

(45)

On substitution
of cp in equation
strain tensor is obtained as

(45), the creep

1 + 1

ik'= (l+e)(2p-p,)t,

-dF
aok,

(50)

This expression
for ik, is singular when p =
p,/2 (i.e. when the point is on the critical state
line) because the normal to F at this point is
vertical and cannot be scaled in the (horizontal)
p direction. If 4 is assumed constant, equation
(50) will predict higher deviatoric
creep strain
rates at higher deviatoric stress levels as p approaches p,/2.
Volumetric age. The volumetric age of the soil
is obtained
by examining
its location
in the
e-p-q
space relative to the position it would
occupy if it were normally consolidated.
The
volumetric
age is back calculated
on the basis
of the secondary
compression
coefficient
C,
and the void ratio distance of the state point
from the state boundary surface.
Figure 5 shows an overconsolidated
soil element A with co-ordinates
(e,, p, q) beneath the

STRESS-STRAIN-TIME

BEHAVIOUR

state boundary
surface. This soil element
developed a preconsolidation
pressure pc, either by
unloading or by natural ageing. In this situation,
the quantities e,, p, q and pc are not all independent but satisfy the relationship
el=ea-(h-K)hp,-K

hp

The location of soil element


ratio axis would be given by

(51)

hp

(52)

if it were normally consolidated.


Assuming
that the secondary
compression
coefficient C, (or +) is constant, the volumetric
age of the soil is computed from equation (8) to
be
(

ez-el
J/

e2-e1
= (t,), antilog ( C, >
where (tJi is the reference
(often taken as 1.0) associated
consolidation
curve.
Deviatoric

scaling

Consider
given bv

the

deviatoric

291

ing for cp

(58)
Substituting

cp in equation

volumetric
time
with the virigin

component

of

Et,

-4-H aa,,(59)
1 aF 2 -r* dF

3 an

This expression for &lf is singular when p = p0


(the isotropic condition) because the normal to F
at this point is horizontal
and cannot be scaled
in the (vertical) q direction.
Further, equation
(59) strictly holds only for values of D of about
20-80%,
the same restriction
that governs the
validity of the Singh-Mitchell
equation. For values of D close to zero (the isotropic condition),
equation (59) would overpredict
the creep strain
rate; for values of D close to unity (the failure
condition),
equation
(59) would underpredict
the creep strain rate.
Deviatoric stress level _andfailure strength. The
deviatoric stress level D that appears in equation (57) is defined as
D=9
qurt

The magnitude
evaluated as

of this

aF

deviatoric

aF

creep

tensor

ap
(31

1 a

=q [ aa,,aa,,-3
In undrained
triaxial
Eklf would simplify to

(45)

A on the void

e2=e,-(A-K)hpcj-K

t, = (tJ, exp

OF WET CLAYS

tests,

li2

is

(55)

the tensor

which is purely deviatoric


because drainage is
suppressed.
The magnitude of this tensor is obtained from the Singh-Mitchell
equation to be

where td is the deviatoric age of the soil and (f&


is the immediate
deviatoric time (usually taken
as 1.0) associated with the immediate hyperbolic
stress-strain
curve (43).
Comparing
equations
(55) and (57) and solv-

(60)

where q is given by equation (12) and qun is the


ultimate or failure strength which is assumed to
be uniquely related to the void ratio.
The ultimate strength of the soil preconsolidated to a pressure pc and allowed to trace the
swelling curve (or the elastic wall) is
quit = ;MP,

(61)

This ultimate strength is the q co-ordinate


of the
intersection
point of the elliptical yield surface F
and the critical state line (4).
If the condition
is undrained
(e =constant),
the ultimate strength of the soil is given by
I-K/h
1
(62)
4Ult= MP,
02
This strength is the q co-ordinate
of the intersection point of the undrained
stress path (7)
through pc and the critical state line (refer to
Fig. 3).
The ultimate strength obtained from equation
(62) is higher than that obtained from equation
(61). If K = 0 (i.e. no swelling during volumetric unloading),
these two equations
would give
identical
strengths.
Since the Singh-Mitchell

BOFUA AND KAVAZANJIAN

292

creep parameters
are usually obtained from undrained triaxial creep tests, equation (62) will be
used to define the ultimate strength.
To be compatible
with equation
(62), the
hyperbolic
curve (43) should yield the same
failure strength at an infinite deviatoric
strain,
This requisite condition can be
i.e. qUlt= q/,-.
used to back calculate the relationship
between
the stressstrain
parameters
b and R, as follows
21-

_-

R,-

(63)

where K, A and M are the Cam Clay parameters.


Deuiatoric age. Consider the same point A in
Fig. 5 which is now given by the co-ordinates
(q, yr) on the q-y plane. If the soil is normally
consolidated,
the stress q would locate A on the
hyperbolic curve (43) at
-Y?_=-

PS-

(64
qb

The deviatoric age td is computed


tion (9) based on y2 as

(y,-Y&-m)
t,j

Ae(t&

( td),exp

from equa-

1(bmJif

mf

Remarks
This numerical approach of expressing creep
contributions
as artificial forces allows stationary
creep problems
(e.g. isotropic undrained
stress
relaxation
experiments
and undrained
triaxial
creep tests) to be numerically
simulated.
The
treatment
of hydrodynamic
lag also allows the
separation
of time-dependent
deformations
into
components
due to pore pressure
dissipation
and due to creep.

(654

if m=l

(65b)
Numerical implementation
The development
of equation (35) allows the
resulting effective stress-based
constitutive
equation to be incorporated
into Biots (1941) general three-dimensional
hydrodynamic
theory of
consolidation.
This consolidation
theory uses
Darcys law to characterize
the transient
pore
pressure dissipation
condition, i.e.

(66)
where
vector

work or a variational
formulation.
Such a
mixed type of formulation
has been numerically implemented
by Borja (1984) for solution
of two-dimensional
boundary value problems of
axisymmetric
(torsionless)
and plane strain configurations
using the above constitutive
equation. The program,
called SPIN 2D, incorporates creep contributions
by explicitly evaluating
the stress relaxation term hilt in equation (35) at
the start of each time increment.
These contributions
can be treated in a manner similar
to or along with temperature
stresses in the
finite element matrix equations,
giving rise to a
pseudo-force
term F which can be explicitly
evaluated and added to the applied nodal force
arrays (Borja, 1984).

ic; is the ith component


of the velocity
ic, p is the pore pressure field, k, is the
(i, j) component
of the permeability
tensor k,
and y_, is the unit weight of water (the negative
sign implies that the flow is in the direction of
decreasing gradient).
It can be seen from equation
(66) that the
displacement
field u and the pore pressure field
p are not independent,
but satisfy the relationship given. A finite element solution can then be
formulated
(Borja, 1984; Christian, 1977; Johnston, 1981) in which the nodal unknowns interpolating u and p are coupled using a virtual

Example
The following example is based on the result
of a drained triaxial compression
test on Weald
Clay reported by Bishop & Henkel (1962). The
test consisted of consolidating
a cylindrical soil
sample to an isotropic stress of p= = 207 kN/m*
and then shearing the sample very slowly while
maintaining
the radial stress constant. The test
results, shown in Fig. 6, are duplicated by SPIN
2D by suppressing the effects of creep and using
the following
material
properties:
K = 0.031,
A = 0.088,
M = 0.882, a = 0.023, Rf = 1.00 and
ea= 1.31.
To illustrate the influence of creep, the volumetric scaling option on creep strain was employed (4 = 0.22 artificially selected) using a fournode axisymmetric
finite element. The bold lines
in Fig. 6 illustrate what the deviator stress-axial
strain and the volumetric strain-axial
strain behaviour would have been if the soil specimen
had been allowed to undergo stress relaxation
increment bc and loaded to failure (cd). During
the stress relaxation
increment
bc, the yield
surface continually expands with time owing to
creep, resulting in the development
of a quasipreconsolidation
pressure p=. Concurrently,
the
*The same test results have been duplicated
by the
program
PEPCO developed
by Johnston
(1981) for
the case when creep is ignored.

STRESS-STRAIN-TIME

BEHAVIOUR

size p. of the equivalent yield surface continually shrinks because


of stress relaxation
(see
equation (46)).
When shearing is resumed the expanded yield
surface is again engaged, showing initial stiffening of the stress-strain
response expected due to
quasi-preconsolidation.
Further shearing causes
the element to fail at d.

OF WET CLAYS

.
200 -

150.
E
t
y iooti
-

/
APPLICATION
BEHAVIOUR

OF THE MODEL
OF UNDISTURBED

Whop
8. Henkel
(1962),
and SPIN 2D (no creep)

PEPCO.

SPIN

creep

2D with

volumetric

TO THE
BAY MUD

In this section
the constitutive
model
is
evaluated
on the basis of its ability to predict
accurately
the results of simple triaxial and
plane strain laboratory tests on undisturbed
San
Francisco Bay Mud (UBM). Both drained (freeflow) and undrained
(no-flow)
problems
are
numerically analysed using single finite elements
whose convergence
characteristics
have been
previously established.*
The soil parameters
used to model UBM are
summarized
in Table 1. These soil properties
were taken from a comprehensive
summary of
Bay Mud properties
presented
by Bonaparte
&
Mitchell (1979). Except for e, (the void ratio at
p = 1 on the isotropic consolidation
line) which
locates the immediate consolidation
line on the
e-ln p plane, all the soil properties were directly
obtained
from
the results
of conventional
laboratory
tests.
The value of e, can be obtained indirectly by
extrapolating
the primary consolidation
line to
p = 1 and moving up along the void ratio axis, to
the immediate
line, according to the secondary
compression
coefficient 4 (see also Kavazanjian
& Mitchell,
1980). However,
Borja
(1984)
showed that the stress-strain
and pore pressurestrain curves do not show appreciable
sensitivity
to the value of e,. Hence, the primary consolidation curve may also be taken as the immediate
line without introducing serious numerical inaccuracies.
Drained ttiaxial tests
The response
of the soil is said to be fully
drained when the rate of loading is much smaller
than the pore fluid diffusion rate. In this case,
the pore pressure
degree of freedom
can be
suppressed,
allowing the problem to be solved
using a single-phase
continuum formulation.
To demonstrate
the predictive
capability of
the constitutive
model in drained triaxial situations, four drained triaxial compression
tests on
*See

293

Kavazanjian, Borja & Jong (1984) for an


analysis of a test embankment
problem involving the
combined
effects of consolidation
and creep using a
mesh of multiple finite elements.

8
w

9 =

MP
Yield

Surtace

p kNlm

Fig. 6. Drained

test on Weald Clay

UBM performed
by Lacerda (1976) were simulated. In his tests, the specimens were initially
isotropically
consolidated
to confining pressures
(T, of 53.9 kN/m,
102.9 kN/m*, 156.8 kN/m
and 313.6 kN/m and sheared at a strain rate of
E, = 3.2 x lo-% per minute.
Lacerda showed that the deviator stress versus
axial strain diagram can be normalized by dividing the deviator stress by the initial consolidation pressure.
A plot of this normalized
curve
and the volumetric
strain-axial
strain curve are
shown in Fig. 7.
Numerical
tests were run with and without
creep effects using a single axisymmetric
finite
element.
This element
represents
the upper
quadrant
of a triaxial specimen
and approximates the displacement
field using a bilinear
interpolation.
The results of numerical
analyses with and
without creep effects are shown by the full and
open circles in Fig. 7, respectively.
The creepinclusive
analysis,
performed
using the deviatoric scaling option, significantly improves the
prediction for volumetric strain E, particularly at
larger values of E,.
Undrained
Lacerda
undrained

triaxial compression tests


(1976) also performed
a series of
triaxial compression
tests at different

294

BORJA

Table 1. Model parameters

for undisturbed

AND

Bay Mud
Symbol

Parameter
Virgin

compression

KAVAZANJIAN

index*

Value
A

0.37
0.85

C,=log,,h
Recompression

index*
c, = log:,

Secondary

compression

coefficient*

ICI
c, = log,, IL

Hyperbolic

stressstrain

parameters?

a
b

Singh-Mitchell

creep

parametersl

Permeability*
Angle

of internal

Void ratio*

0.054
0.124
0.0065
0.0150

0.0062
1.36 (=1.23
0.95

A
cu
m

3.5 X 10-s per min


4.45
0.75

friction?

34.5

(5)

1.40

ea

2.52

at p,= 1 kPa
time

O,)i

1.00 min

&Ii

1.00 min

Instant

volumetric

Instant

deviatoric

* From
t From
$ From

triaxial isotropically
consolidated
or conventional
consolidation
test.
isotrooicallv
consolidated
undrained
test with pore pressure measurement.
isotropically
consolidated
undrained
creep test.

time

rates,
ranging
from
1.1%
per minute to
7.3~ 10m4% per minute. He observed that both
the initial tangent modulus Ei and the ultimate
strength quit of the deviator stress-axial strain
curves varied linearly with the logarithm of the
axial strain rate E,. Similar observations have
been reported by Kondner (1963). Hence, it
may be inferred that the hyperbolic curves given
by equation (lo), obtained from conventional
undrained triaxial tests, do not exactly represent
the time-independent behaviour because they
also contain creep components.
Interpolating E, and quit from Lacerdas plots
for & values of 1.0% per minute, 0.1% per
minute, 0.01% per minute and 0.001%
per
minute, the corresponding transformed hyperbolic curves are plotted in Fig. 8. Also plotted in
Fig. 8 is the hyperbolic curve obtained by
Bonaparte
(1981) from conventional stresscontrolled undrained triaxial compression tests.
Bonapartes
tests were completed in about
250-300 min to allow pore pressure equalization, compressing the samples to about 10%
axial strain during this period. This is roughly
equivalent to compressing the samples at a strain

from equation

(63))

Variable

h, k

M from equation

strain

2.57

2.0-

1.5bU
B
l.O-

SPIN 20
No creep
Creep Included

r:;~
.

Fig. 7. Drained

test on UBM

STRESS-STRAIN-TIME

BEHAVIOUR

rate of about 0.01% per minute. This is confirmed by the observation that Bonapartes results
plot very closely to the hyperbola corresponding
to this strain rate.
Numerical
analyses using the properties
in
Table 1 were performed
with a single finite
element
that uses a biquadratic
displacement
interpolation
and a linear pore pressure interpolation.* Excellent
agreement
was achieved between Lacerdas results and the numerical experiments,
as shown in Fig. 8. It can be verified
from Fig. 8 that the trace of the Cam Clay
yield surface on the deviator stress-axial
strain
plane under undrained
triaxial compression
is,
approximately,
a hyperbola whose strength and
stiffness are a function of the imposed strain
rate.
When creep is ignored, the data points define
a unique hyperbola
regardless
of whether
the
numerical
tests are stress controlled
or strain
controlled.
It can also be observed
that the
hyperbola
obtained
from conventional
stresscontrolled
triaxial compression
tests does not
generally
represent
the immediate
soil be-

Lacerda
(1976)
(strain controlled)

Bonaparte

(1981)

controlled)
duratfon
f

240-300

ml

OF WET CLAYS

since it may contain


haviour,
amount of creep deformation.

295
a

significant

plane strain. test


Plane strain test results are not as commonly
reported
in the literature as triaxial test results
although the former can be more useful in the
analysis of actual geotechnical
structures such as
dams, embankments
and long excavations.
An undrained
plane strain test on UBM was
performed
by Sinram (1985) to simulate a deep
pressuremeter
test stress condition
(Figs 9 and
10). In this case, the normal off-plane
stress
changes as a result of Poisson effects. Sinram
measured
these stress changes as well as the
in-plane strains and pore pressures in his tests.
To mimic the imposed loading history shown
by the points on the total stress path in Fig. 9, a
numerical
test was performed
using a single
bilinear
displacement,
constant
pore pressure
element
and the deviatoric
creep option for
creep strains. Fig. 10 shows that the induced
pore pressures
due to volumetric
compression
and shearing are initially overpredicted.
Consequently, the deviator stresses during the initial
shearing
stage (oa) in Fig. 9 are also overpredicted.
Overall,
however,
the pore pressure-strain
and stress-strain
curves manifest the stress history to which the soil specimen was subjected.
Further,
the rupture strength
((r,-~~),,,~._ the
pore pressure at the near-failure
condition and
the additional
excess pore pressure during the
Ufldrained

a-

6-

Stress

controlled,

no creep

Straw

controlled,

no creep

Strain

rate

0 1% per mn

A Stress controlled,
duratfon
= 300

Q6

mn

Fig. 8. Undrained triaxial tests on UBM performed


various strain rates

lo

at

*It has been generallv observed (Christian, 1977;


Johnston, 19817 that to obtain compatible coupled
fields the displacement interpolation should be one
order higher than the pore pressure interpolation.

Fig. 9. Stress-strain
URM

diagram for a plane strain test on

296

BORTA

AND

KAVAZANJIAN

9
200-

itjo-

Points

z
5 120

-2

--~
-

on TSP

sinram(1985)

-SPIN

2D

I
k

6
SKiIn

10

12

e22 %

Fig. 10. Pore pressurestrain


strain undrained test on UBM

diagram

for a plane

isotropic loading stage BC were reasonably well


predicted.
The same overprediction
of the deviator stresses in the triaxial stress condition was also reported
by Roscoe
& Burland
(1968). They
suggested that the Cam Clay model be modified
to account for plastic shear deformations
which
may also develop beneath
the state boundary
surface.
Undrained creep tests
Several undrained
triaxial
been performed
at different
10

creep tests have


deviatoric
stress

levels over periods of several logarithmic cycles


of time. Results of three tests performed
by
Lacerda (1976) are shown in Fig. 11 for values
of fi of 0.8, 0.7 and 0.5, obtained
either by
increasing
the axial load or by decreasing
the
lateral (confining) pressures from the initial isotropic condition.
Superimposed
in Fig. 11 are the numerical
tests obtained using a single triaxial element that
interpolates
the displacement
field biquadratitally and the pore pressure field linearly, and the
deviatoric scaling option on creep strain is employed. The excellent agreement between the experimental
and the numerical
test
results
affirmed the validity of the Singh-Mitchell
creep
function
for undrained
triaxial conditions
for
values of fi within the range of engineering
interest. It should be noted that for the condition fi = 0.8, the experimental
(full) curve starts
to deviate from the Singh-Mitchell
creep function for t > 1000 min, an observation
which can
be even more apparent in the failure and nearfailure conditions
(D -+ 1.0).
Combined creep and stress relaxation tests
To illustrate the ability of the model to account for the time and stress history dependence
of the material
response
to applied loads, a
numerical
test was performed
to simulate undrained triaxial creep and stress relaxation tests
on UBM performed
by Lacerda (1976). Biquadratic displacement,
linear pore pressure interpolations were used on a single axisymmetric
quadrilateral
element and the deviatoric scaling
option on creep strain was employed.

r
CR-l-2

SR-1-2

SR-l-3

Fig. 11. Undrained

Lacerda

D = 0.8
0, Icreaslng,o,

Consranl

D = 0 7
m, constanlp,

decreasing

u = 05
v, constant,o,

decreasmg

(19761

creep tests on UBM

STRESS-STRAIN-TIME

BEHAVIOUR

The test consisted of initially consolidating


the
element
to an isotropic
stress of 78.4 kN/m
and shearing the element undrained
at a constant axial strain rate of F = 0.38% per minute.
At
this
point,
the
deviatoric
stress
q =
42.6 kN/m (point A in Fig. 12).
After about 3000 min of stress relaxation
(AB) during which ca= constant,
shearing was
resumed at the same strain rate until the stress
level was close to failure (C). The strain was
then held at about 2.3% for 1320 minutes after
which the specimen
was sheared
again at a
reduced strain rate of e, = 1.6 x lo-% per minute.
Very good agreement
can be observed during
stages OA, AB, BC and CD on the stress-strain
curve of Fig. 12. During reloading at a reduced
strain rate (DE), however, the creep strain rate
compute_d_ from the Singh-Mitchell
function
& zz AeaD exceeded
the actual strain rate of
0.016%
per minute for values of DaO.342.
Thus the numerical
solution had to be terminated at that point.
The predicted pore pressure curve of Fig. 12
shows an offset of about AB = 10 kN/m beyond
point B. The overestimation
of pore pressures
during this stress relaxation
stage is due to the
arresting of secondary
consolidation
in a nearisotropic
condition
(B), a phenomenon
which
can be expected
to occur but was never observed in this laboratory test. The offset AB may
be attributed to the fact that the deviator& scaling option overpredicts
F,t for values of D close
to zero, resulting in the overprediction
of pore
pressures as well.
CONCLUDING

REMARKS

A constitutive model for wet clays capable of


accounting for time-dependent
(creep) effects in
general three-dimensional
stress conditions
has
been developed.
This model is characterized
mathematically
by the following effective stressbased rate constitutive
equation
u;j = C$&&k,- ai,t
and has been incorporated
into Biots (1941)
three-dimensional
theory of consolidation
to account for the influence of hydrodynamic
lag.
The constitutive
model defines the tensor of
rate ail,
moduli ciikl and the stress relaxation
before
and during yielding,
based on Bjerrums (1967) concept that the total deformation
can be decomposed
into time-independent
and
time-dependent
parts.
The time-independent
part of the total deformation
is evaluated using
the classical theory of plasticity and the ellipsoidal yield surface of the modified
Cam Clay
model presented
by Roscoe & Burland (1968).

297

OF WET CLAYS

60
E
t
Y
6-40

20

---

- --SPIN

OL

6
,

Lacerda
(1976)
20

3
Am

Fig. 12. Combined


UBM

Lacerda
(1976)
SPIN 2D

strain

72 %

creep and stress relaxation test on

The time-dependent
part is evaluated
on the
basis of the phenomenological
creep rate expressions
for isotropic
or undrained
triaxial
stress conditions (volumetric and deviatoric scaling respectively).
If the secondary
compression
coefficient
4
is constant,
the volumetric
and the deviatoric
scaling procedures
cannot give identical results
because
rC, cannot be made to vary with the
deviator stress level. An approach presented
by
Borja (1984) for the plane strain case forces the
tensor Eer to satisfy both the volumetric and the
deviatoric creep requirements
without applying
the normality rule.
The above constitutive
equation
has been
numerically
implemented
into a consolidationcreep finite element program capable of solving
two-dimensional
boundary
value problems
in
plane strain and axisymmetric
(torsionless) stress
conditions. The tensor of moduli c,~~,contributes
to the global stiffness matrix, while the stress
relaxation
rate term &<iifis treated as or along
with temperature
stresses which can be superimposed in the vector of applied nodal forces using
the virtual work or variational formulation.
The consolidation
creep program
has been
used to predict
results of triaxial and plane
strain laboratory
tests, with and without creep

298

BOFIJA AND

effects, demonstrating
that creep deformations
can become a major contributor
to the total
deformations
for young clays or in conditions
under which the time of sustained
loading is
comparable
with the geologic age of the soil. By
imposing the condition
of incompressibility,
it
has likewise been shown that undrained
creep
can be of major importance
in the prediction of
excess pore pressures
as a result of either the
arresting of secondary compression
or shearing.
While the examples
discussed
in this Paper
showed simple boundary
conditions
simulating
triaxial or plane strain laboratory
stress conditions, the validity of the above model has also
been investigated
and has been shown to work
in more complicated
axisymmetric
applications
(Borja, 1984) and in an actual plane strain field
situation (Kavazanjian
et al., 1984).
Further
research
is currently
under way as
part of a Stanford
University-University
of
California,
Berkeley,
collaborative
research
project on the stress-strain-time
behaviour
of
cohesive soils, funded by the National Science
Foundation.

REFERENCES
Biot, M. A.

(1941).
General
theory
of threedimensional
consolidation.
J. Appl. Phys. 12, 155164.
Bishop, A. W. & Henkel, D. J. (1962). The menyurement of soil properties in the triaxial test, 2nd edn.
London:
Arnold.
Bjerrum, L. (1967). Engineering
geology of Norwegian
normally
consolidated
marine clays as related to
settlements
of buildings.
7th Rankine
Lecture.
Geotechnique
17, No. 2. 83-117.
Bonaparte,
R. (1981). A time-dependent
constitutive
model for cohesive soils. PhD thesis, University
of
California,
Berkeley.
Bonaparte,
R. & Mitchell, J. K. (1979). The properties
of San Francisco Bay Mud at Hamilton Air Force
Base, California.
Geotechnical
Engineering
Report, University
of California,
Berkeley.
Borja,
R. I. (1984). Finite element analysis of the
time-dependent
behavior of soft clays. PhD thesis,
Stanford University.

KAVAZANJIAN

Christian,
J. T. (1977). Two- and three-dimensional
consolidation.
In Numerical methods in geotechnical engineering,
ch. 12, pp. 399-426
(eds C. S.
Desai and J. T. Christian). London: McGraw-Hill.
Duncan,
J. M. & Chang, C. Y. (1970). Nonlinear
analysis of stress and strain in soils. J. Soil Mech.
Fdns Div. Am. Sot. Civ. Engrs 96, SM5, 16291653.
Johnston,
P. R. (1981). Finite element consolidation
analyses
of tunnel behavior on clay. PhD thesis,
Stanford University.
Kavazanjian,
Jr, E., Borja, R. I. & Jong, H.-L. (1985).
Numerical
analysis of time-dependent
deformations in clay soils. Proc. 11th Int. Conf. Soil Mech.
Fdn Engng, San Francisco,
II-16
Aug., Session
lA, Paper lA30. Rotterdam:
Balkemar
Kavazaniian.
Jr. E. & Mitchell. J. K. (1980). Timedependent
heformation
behavior
of clays.
_I.
Geotech. Engng Div. Am. Sot. Civ. Engrs 106,
GT6, 611-630.
Kondner,
R. L. (1963). Hyperbolic
stress-strain
response: cohesive soils. J. Soil Mech. Fdns Div. Am.
Sot. Civ. Engrs 89, SMI, 115-144.
Lacerda,
W. A: (1976). Stress-relaxation and creep
effects on soil deformation. PhD thesis, University
of California,
Berkeley.
Ladd, C. C. & Foott, R. (1974). New design procedure
for stability of soft clays. J. Geotech. Engng Div.
Am. Sot. Civ. Engrs 100, GT7, 763-786.
Pender, M. J. (1977). A unified model for soil stressstrain behavior. Proc. 9th In?. Conf. Soil Mech. Fdn
Engng, Tokyo, Specialty session 9, pp. 213-222.
Roscoe,
K. H. & Burland,
J. 8. (1968). On the
generalized
stress-strain
behavior of wet clay. In
Engineering plasticity, pp. 535409
(eds J. Heyman
and F. A. Leckie). Cambridge:
Cambridge
University Press.
Roscoe, K. H., Schofield, A. N. & Thurairajah,
A.
(1963). Yielding
of clays in states wetter than
critical. GCotechnique 13, No. 3, 211-240.
Singh, A. & Mitchell, J. K. (1968). General stressstrain-time
functions for soils. J. Soil Mech. Fdns
Div. Am. Sot. Civ. Engrs 94, SMl, 21-46.
Sinram, A. (1985). Cubical shear box tests designed to
simulate pressuremeter loading. Engineers
thesis,
Stanford University,
to be published.
Tavenas, F. & Leroueil, S. (1977). Effects of stresses
and time on yielding of clays. Proc. 9th Int. Conf.
Soil Mech. Fdn Engng, Tokyo, pp. 319-326.
Taylor, D. W. (1948). Fundamentals of soil mechanics.
New York: Wiley.

Das könnte Ihnen auch gefallen