JARDINE,R. J., Porrs, D. M., FOURIE,A. B.

& BURLAND,
J. B. (1986). GCorechnique
36, No. 3, 377-396

Studies of the influence of non-linear stress-strain

characteristics in soil-structure
interaction
R.

J. JARDINE,*

D. M. POTTS,*

A. B. FOURIEt

Recent field and laboratory studies have shown that,

even at very small strains, many soils exhibit non-linear
stress-strain
behaviour.
Nevertheless,
because of its
convenience,
linear elasticity will continue to play an
important
role in the analysis of such problems
as
settlement, deformation
and soil-structure
interaction.
In this Paper the measured
non-liqear
stress-strain
properties of a low plasticity clay are used in the finite
element analysis
of footings,
piles, excavations
and
pressuremeter
tests to assess the influence of small
strain non-linearity
in comparison
with linear elastic
behaviour.
In all cases non-linear
behaviour
results in
the concentration
of strain and deformation
towards
the loading boundaries.
This is shown to have important consequences
for soil-structure
interaction
problems such as settlement profiles, pile group interaction
and contact
stress distributions.
Small strain nonlinearity also has a significant influence on the interpretation in terms of equivalent elastic moduli of in situ
deformation
tests (e.g. plates and pressuremeters)
and
of field measurements.
It is concluded
that, although
linear elasticity remains a convenient tool for expressing
measurements
of soil stiffness, unless the non-linear
nature of soils is taken into account,
soil-structure
interaction computations
and the interpretation
of field
measurements
can be misleading.

and J. B. BURLAND*

tions de contraintes
au contact.
La non-1inCaritC B
faibles contraintes
influence aussi de faqon importante
IinterprCtation
des essais de d&formation en place, par
exemple, plaques et pressiomitres,
et les mesures in situ
en fonction de modules klastiques kquivalents.
Tandis
que 16lasticitt lintaire reste encore une methode correcte pour exprimer des mesures de la rigidit du sol, on
tire la conclusion que les calculs de Iinteraction entre le
sol et la construction
et Iinterpr&tation
des mesures in
situ peuvent induire en erreur, & moins quon ne tienne
compte de la nature non-IinCaire des ~01s.
KEYWORDS:
elasticity; excavation;
settlement; soil-structure
interaction.

field tests; piles;

INTRODUCTION

Analyses of soil-structure
interaction
frequently
involve
the prediction
of deformations
and
stresses, both in the surrounding
soil mass and
over areas of contact with the loading boundaries. In recent years it has become possible to
compute
solutions
with increasingly
complex
descriptions
of the soil properties.
However, the
use of non-linear calculations in engineering practice is restricted by time and cost. Moreover high
quality stress-strain
data are difficult to obtain.
Des essais in situ et des Ctudes en laboratoire
de date
rtcente ont dtmontrb que beaucoup de sols ont un comThere is therefore a need for sensitivity studies
portement
contrainte_dCformation
non-IinCaire,
m&me
using advanced soil models to investigate the sigg des dkformations
extrtmement
faibles. Cependant
$
nificance of various features of soil behaviour
cause de sa commoditB l&lasticit& lintaire continuera
B
such as non-linearity
at small strains and local
jouer un rble important
dans Ianalyse des problbmes
failure.
tels que le tassement,
la dkformation
et Iinteraction
The most common types of analysis continue
entre le sol et la construction.
Dans cet article les propto be based on the theories of linear elasticity.
riCtts contrainte42formation
non-liniaires
dune argile
The underlying
assumption
is either that at
de faible plasticit
ont utilis&s dans Ianalyse B &men&
working loads the soil mass is behaving in a linfinis des semelles, des pieux, des excavations
et des
essais pressiomttriques
afin dkvaluer Iinfluence de la
early elastic manner or that the stress changes in
non-IinCaritk $ faibles contraintes
en comparaison
avec
the soil are close to those given by linear elasticity
le comportement
Clastique IinCaire. Dans tous les cas le
even though the soil itself may be non-linear. As
comportement
non-lintaire
produit une concentration
pointed out by Eisenstein & Medeiros (1983), the
de contraintes
et de d&formations
vers les limites de
work of Wroth (1971) and Burland (1975) has
chargement.
On dkmontre que ceci a des cons&quences
encouraged
the former view for stiff clays and
importantes
pour les problimes
dinteraction
entre le
weak
rocks.
The finding that the vertical stresses
sol et la construction,
tels que les profils de tassement,
beneath flexible loaded areas are relatively insenIinteraction
entre des groupes de pieux et les distribusitive to the stress-strain
law has greatly promoted the second assumption
(Morgenstern
&
Discussion on this Paper closes on 1 January 1987. For
Phukan, 1968). The accuracy of predictions
has
further details see inside back cover.
thus been seen to hinge on the determination
of
* Imperial College of Science and Technology.
appropriate
in situ elastic moduli (E,, E, G, K
t University of Queensland;
formerly Imperial College
etc.) and their variations
with depth. However, it
of Science and Technology.
Delivered by ICEVirtualLibrary.com
to:
377
IP: 129.31.210.87
On: Fri, 26 Aug 2011 17:33:26

378

JARDINE,

POTTS,

FOURIE

AND BURLAND

has proved to be difficult to measure these elastic

struction of a strutted excavation.
Comparisons
parameters.
In particular the results of convenwith linear elastic behaviour are then made to
tional laboratory
tests frequently give stiffnesses
draw out the significance of the non-linear
soil
which are far lower than those back analysed
properties.
from field measurements.
This discrepancy
has
To facilitate the interpretation
of the results,
been explored for London Clay by St John (1975)
only the simplest form of failure criterion is used.
and has resulted in a strong move towards in situ
This allows the problems to be treated in terms of
testing (Marsland, 1971; Windle & Wroth, 1977).
total stresses but in each study the initial ground
The combination
of careful field measurements
stresses were specified in relation to appropriate
and linear elastic theory for back analysis and
K,, effective stress conditions.
prediction
has been relatively
successful,
but
The objectives of the present study are
several limitations
have become apparent.
For
(a) to identify
several important
features
of
example, it is well known that in many problems
behaviour which stem from non-linear stresslocal enclaves of fully plastic behaviour
can
strain characteristics
develop at working loads. Their existence leads to
(b) to discuss the problems of selecting appropristresses
and patterns
of deformations
which
ate values of apparent
elastic moduli for
depart significantly from elastic behaviour. If the
simple elastic calculations
material is brittle such local failure zones may
(c) to draw attention to the difficulties of interpropagate
rapidly leading to instability at relapreting in situ tests and field measurements
tively small displacements.
using linear elasticity, particularly
when the
Field observations
have identified other inconderived soil moduli are to be used in a differsistencies.
Burland
& Hancock
(1977)
for
ent type of boundary value problem.
example, drew attention
to the fact that the
profile of ground movements outside excavations
Before presenting
the results of the various
cannot
be accounted
for by linear elasticity.
boundary value analyses, a brief resume is given
of the characteristics
observed
in the recent
Simpson, ORiordan & Croft (1979) showed that
the use of a bilinear stress-strain
relationship
laboratory experiments.
Following from this, the
before failure with an initial very stiff portion
details of the simple soil model developed for
accounted considerably
for the observed behavthese studies are set out.
iour.
Almost
concurrently
new laboratory
techniques began to be developed for the accurate
UNDRAINED STRESS-STRAIN CHARACmeasurement
of strains locally on soil samples
TERISTICS OBSERVED IN SPECIALLY
INSTRUMENTED LABORATORY TESTS
(Daramola,
1978; Costa-Filho,
1980; Burland &
Recent research by Daramola
(1978), CostaSymes, 1982). The results obtained
with these
Filho (1980) and Jardine et al. (1984) has shown
techniques throw into doubt the validity of the
that conventional
methods of determining
axial
assumption
of linear elastic behaviour
under
strains in triaxial experiments can lead to imporworking
conditions.
The tests show that the
initial stress-strain
behaviour
of many soils is
tant errors. Even with the most careful sample
preparation
and calibration for the compliance of
much stiffer than indicated
by conventional
equipment,
bedding effects at the specimen ends
strain measurements,
and that the undrained
and rotation of the sample under load can cause
stress-strain
characteristics
of a wide range of
the externally measured strains to exceed those
soil types are markedly
non-linear
(Jardine,
measured
locally on the sample. In the early
Symes & Burland, 1984). Careful analysis of spestages of a test the external strains can be ten
cially instrumented
field tests confirms that these
characteristics
are also representative
of in situ
times larger than those measured on the sample.
Maswoswe
&
By means of displacement
transducers
mounted
Fourie,
behaviour
(Jardine,
on the soil specimens, Costa-Filho
& Vaughan
Burland, 1985).
The purpose of this Paper is to examine the
(1980) found that the true secant stiffness of
significance
of the laboratory-observed
stresssamples of London Clay, at around 0.1% strain,
agreed well with average stiffness values obtained
strain characteristics
in a range of practical probfrom the back analysis of field measurements
(St
lems. A typical low plasticity clay is considered
John, 1975). It therefore appears that the prewhich exhibits both realistic non-linear behaviour
before failure and plastic flow when its failure
viously reported differences between laboratory
and field stiffnesses of London Clay result more
criterion is satisfied. Finite element analyses are
from inadequacies
in conventional
laboratory
presented
which allow a detailed study of the
strain measuring techniques than from sampling
undrained response of this one soil when loaded
by footings, piles, cavity expansion
and the
condisturbance to:
or time-dependent
threshold effects.
Delivered
by ICEVirtualLibrary.com

IP: 129.31.210.87
On: Fri, 26 Aug 2011 17:33:26

NON-LINEAR

STRESS-STRAIN

CHARACTERISTICS

Awal

3
siran

379

E %

(a)

Ax1.3

stram

E: %

03
Fig. 1. Measured undrained stress-strain behaviour of a reconstituted K, consolidated low plasticity clay at various overconsolidation ratios as indicated: (a) strain on an arithmetic scale; (b) strain
on a logarithmic scale

ticity clay which were consolidated

from a slurry
Using new local strain measuring techniques,
to various
overconsolidation
ratios.
Fig. I(b)
Jardine (1985) carried out a comprehensive
range
shows the same data with strain plotted on a
of triaxial tests on a wide spectrum
of soils.
logarithmic scale to show the detail of the initial
Experiments
on intact and reconstituted
low and
stages of loading. In Fig. 2 the results of tests RI
medium plasticity clays, sand and intact chalk all
and R2 have been plotted
as E,jC, against
exhibited
high initial stiffness and non-linear
log(strain)
where E, is the secant undrained
stress-strain
behaviour,
both
these
factors
Youngs modulus. The results of tests on two
depending on soil type, method of formation and
lightly overconsolidated
intact samples of the
stress
history.
To illustrate
these
features
same soil (I1 and 12) are also shown and give
Fig. l(a)
shows
the
undrained
stress-strain
similar results. Sample I1 was tested unconsolibehaviour of reconstituted
samples of a low plasDelivered by ICEVirtualLibrary.com to:

IP: 129.31.210.87
On: Fri, 26 Aug 2011 17:33:26

JARDINE, POTTS, FOURIE AND BURLAND

380

Axial strain E. %

Fig. 2. Stiffness-&rain curves for a low plasticity clay: Rl and R2 resonstituted, sheared from K,
conditions; I1 and 12 lightly overconsolidated intact clay from 672 m below the sea bed; I1 sheared
from unconsolidated conditions, 12 reconsolidated to K. = @52 before undrained shearing

dated and sample 12 was anisotropically

reconsolidated
to its estimated
in situ stress state.
Further details of these tests are given by Jardine
et al. (1984).
EMPIRICAL
UNDRAINED
RELATIONSHIP

STRESS-STRAIN

To use the measured stress-strain

relationships
in the analysis of boundary value problems it is
necessary to find a simple mathematical
expression that fits the data reasonably
well. In this
section such an empirical expression is described.
No doubt other expressions
could be found
which fit the data equally well.
The data presented in Fig. 2 suggest that the
general form of the relationship
between the
secant Youngs modulus E, and the logarithm of
axial strain before failure can be conveniently
represented by a periodic logarithmic function

as shown in Fig. 3. When considering

more
general effective stress models, equations of the
same form may be used to describe the variations
in shear and bulk modulus with their respective
strain invariants. The empirical constants A, B,
C, c( and y can be quickly determined
from test
data as described in Appendix 1. Equation
(1)
only holds for a specified range of strain values.
For strains below a lower limit E,,,~and above an
upper limit E,,,,
fixed tangent
stiffnesses
are
assumed.
Over this elastic range a Poissons
ratio of 0.49 is specified and, if yield is to be
modelled, a suitable criterion and flow rule must
be included. Care is required to ensure compatibility between E,,, and the onset of plastic yield.
In Fig. 4 equation (1) is fitted to a selection of
the stiffness test data given by the low plasticity
clay. The values of the associated empirical constants are given in Table 1. It can be seen that the
formulation
is appropriate
for soil behaviour
before yield.

E
-=4+Bcos{ol[log,,(~)~}
(1)
CU
Table I. Non-linear soil parameters from three tests

Test

c: %

e,(min): %

EJmax): %

RI
R2
12

850
3100
1420

1000
3200
1380

OGO8
0.0007
0.009

2.023
1,349
2.098

0.5943
0.6385
0.5050

0@05
0,003
0.0045

0.20
0.20

Delivered by ICEVirtualLibrary.com to:

IP: 129.31.210.87
On: Fri, 26 Aug 2011 17:33:26

1.5

NON-LINEAR

STRESS-STRAIN

381

CHARACTERISTICS

,Observed, or protected. maximum

/

/E,/C,

= A + E cos

[a (log,, dC)yI

u=

E =

c-

r10~3

Projected minimum

Fig. 3.

Curve fitting to stiffnessstrain

data

The majority of numerical

procedures
make
use of the tangent modulus E,, rather than the
secant modulus. Differentiating
and rearranging
equation (1) gives

eralized
invariant
E =

by

substituting

the

deviatoric

strain

12[(~1 - Q) + (et - E#

0
Buyl-
E,,
c = A + B cos (d) - x
sin (czP)

where I = log,,

(E,/C). Equation

(2)

(2) can be gen-

+ (E2- &3)2]12
for 312&, and this allows equation
porated
into non-linear
elastic

Fig. 4.

Curve fitting for tests Rl, R2 and I2

(3)

Test data
Equatfon (1) (Rl,
Equation (1) (12)

Delivered by ICEVirtualLibrary.com to:

IP: 129.31.210.87
On: Fri, 26 Aug 2011 17:33:26

R2)

(2) to be incorfinite element

JARDINE,

382

POTTS,

FOURIE

AND BURLAND

-0C

0.
103

Fig. 5.

10~2

o -

= 110 kN/m

Experiment R2
Finite element slmulatfon

10-i
eA %

100

Comparison between experimental and fitted curves for test R2

layer. The footing is loaded undrained to a mean

bearing
pressure
of 3C, where
C, is the
undrained
strength of the clay. Hence the load
factor on undrained
bearing failure is approximately 0.5. Throughout
the clay layer it is
assumed that the soil has the same initial stress
history and stress-strain
properties
as given by
the low plasticity clay in test R2 (see Fig. 5).
For this illustrative problem it is assumed that
the changes in the total stresses crv and CT~
beneath
the centre of the footing can be obtained
by
means of linear elasticity. At any depth Z/D the
value of (a, - a,)/2 can be calculated and the corresponding value of vertical strain obtained from
Fig. 5. Fig. 6(b) shows the distribution
of vertical
strain beneath the centre of the footing obtained
using this procedure. In Fig. 6(c) the variation in
normalized
settlement
6/S, is plotted
against
depth and compared with the distribution
for a
homogeneous
linear elastic material.
Fig. 6(d)
shows the variation in secant modulus E,/C, with
depth corresponding
to the strain distribution
given in Fig. 6(b).
SIMPLE
PRACTICAL ILLUSTRATIVE
The following important practical conclusions
PROBLEM
can be drawn from this simple illustration.
In this section a simple problem is analysed to
illustrate some of the practical conclusions
that
(a) From Fig. 6(b) it can be seen that the axial
arise from the effects of non-linear stress-strain
strains beneath the centre are always less than
behaviour. An approximate
method is employed
O-l%. Therefore, even though the load factor
for this preliminary study but in the remainder of
is as high as 0.5, the ground response is domithe Paper a finite element analysis is used. The
nated by its small strain properties.
Thus
problem considered is that of a rigid circular load
laboratory
testing to evaluate the stiffness
which might, for example, represent a foundation
properties
of the ground will require precise
for which the underlying settlements are required
measurement
of strains to at least an accuracy
or an instrumented
loading test from which
of 0.01%.
deformation properties are to be deduced.
(b) From Fig. 6(c) it is evident that the settleFigure 6(a) shows a rigid smooth
circular
ment for LPC2 reduces much more rapidly
footing of diameter
D resting on a layer of
with depth than for a homogeneous
linear
uniform clay of thickness 5D underlain by a rigid
elastic material.
Delivered by ICEVirtualLibrary.com to:
computer programs. For the present analyses the
empirical formulation
has been combined with a
perfectly plastic (non-hardening)
Tresca failure
criterion
and plastic potential.
The undrained
stress-strain
relationship
chosen for this study is
referred
to as LPC2 and corresponds
to the
relationship
obtained for test R2 given in Fig. 1
(see Table 1 for the appropriate
empirical
constants).
Figure 5 shows a comparison
of the source
data from test R2 and the finite element simulation using material LPC2. The agreement is excellent over almost four logarithmic cycles of strain.
Test R2 was chosen for the studies described in
this Paper, as its yielding behaviour most closely
approximated
to the simple Tresca criterion.
Model LPC2 represents
a stiff low plasticity
clay with an overconsolidation
ratio of 2. For the
range of data presented by Jardine et ul. (1984)
the model shows a higher normalized
stiffness
than average but is not particularly non-linear.

IP: 129.31.210.87
On: Fri, 26 Aug 2011 17:33:26

NON-LINEAR

STRESS-STRAIN

CHARACTERISTICS

Apparent

E: %

CD_*

383

E,IC,

5000

0.1

T-r
:2-0

E,/C,

= 1000 + 1250UD

(apparent
linear
Of Stiffness
with

(4

varlallon
depth)

id)

Fig. 6. Approximate calculations of strain, settlement and apparent stiffness beneath the centre of
a rigid footing: (a) circular footing on a uniform layer of clay; (b) vertical strains deduced under the
centre line for an elastic stress distribution; (c) variations in settlement with depth for LPC2 and a
homogeneous linear elastic soil; (d) variation in apparent linear modulus with depth for LPC2 using
elastic stresses and strains from (b)

(c) The results given in Fig. 6(d) show that, had

of settlement
at various
field measurements
depths been made down to Z/D = 3.5, a
linear
elastic
interpretation
would
have
pointed to a linearly increasing stiffness with
depth. This conclusion has been reached from
many back-analysed
field measurements.
An
erroneous conclusion that a given soil profile
has increasing
stiffness with depth, rather
than non-linear small strain stiffness properties, can have significant
and unfortunate
practical consequences.
Of course in many
practical
situations
both
effects
will be
present.

The remainder
of this Paper is devoted to
exploring some of the practical implications
of
small strain non-linearity
(and failure) and the
limitations of linear elastic predictions when such
non-linearity
is present.
ANALYSIS

vertical

displacements

yields

100

02

0.4
Normalired

Fig. 7.

Pressuresettlement

BOUNDARY

0.6
settlement

VALUE

As mentioned
previously, the purpose of this
study is to analyse a range of undrained boundary value problems using non-linear elasto-plastic
stress-strain
characteristics
in such a way as to
identify any major differences between the results
obtained and those predicted from linear elastic

,Applled

1 st element

OF SOME

PROBLEMS

of footing

m-

0.8
8/D

curve for a rigid footing resting on soil type LPC2

Delivered by ICEVirtualLibrary.com to:

IP: 129.31.210.87
On: Fri, 26 Aug 2011 17:33:26

10

JARDINE,

384

POTTS,

FOURIE

theory. The boundary value problems considered

fall in four groups which will be discussed separately
(a)
(6)
(c)
(d)

AND BURLAND

6/6,

0.2

0.4

0.6

0.8

1 .o
A

footings
cavity expansion problems
axially loaded piles
strutted excavations.

The analyses were carried out by means of the

finite element
method
using eight-noded
isoparametric elements and reduced integration.
Circular rigidfooting
The problem considered is that of a vertically
loaded rigid smooth circular footing of diameter
D resting on the surface of a uniform layer of
clay. The depth assumed for the clay layer was
5D, as this case approaches
the asymptote
for
infinite
depth
with a linear elastic material
(Poulos & Davis, 1974).
Two soil types were considered in the study
(a) the undrained
non-linear
soil model LPC2
described
previously,
with K,, = 0.72 and
C, = 220 kPa (Fig. 5)
(6) undrained
linear elastic soil with E, = 1056
MN/m2 (i.e. the initial value for LPC2) and
Poissons ratio 0.49.

SOL
Fig. 8. Profiles of normalized settlement with depth for
a rigid footing (see Fig. 7)

Figure 9 shows the predicted ground surface

profiles at load factors of 0.3 and 0.52 (i.e. before
and after first yield) for the LPC2 model compared with linear elasticity. It can be seen that the
influence of the stiffness variation before yield,
compared with linear elasticity, is to concentrate,
The finite element mesh used for the analysis is
shown by the inset in Fig. 7 and loading was
the settlements strongly around the loaded area.
carried out by applying increments
of uniform
The onset of local yield further accentuates
this
behaviour. It is important
to note that a linear
vertical displacements
to the footing surface. For
elastic material
with stiffness increasing
with
the elastic case very close agreement was obtained
depth (a Gibson soil) also exhibits a concentrawith the exact solution given by Poulos & Davis
tion of settlement
towards
the loaded
area
(1974).
(Gibson, 1967; Burland, Sills & Gibson, 1973).
The load-settlement
relationship
predicted
These steep local gradients of displacement
have
with LPC2 is plotted in Fig. 7. The analysis was
important
implications
for soil-structure
intercontinued until a settlement ratio 6/D of 0.02 had
action.
been reached, by which point the mean bearing
Figure
10 presents
contours
of deviatoric
pressure was within 2% of the solution qua z (3
strain at a load factor of 0.52. The small region of
+ n)C, given by Vesic (1975) and others.
local failure is shown shaded. It can be seen that
The analysis gives first yield under the centre
at a normal working load most of the soil is expeline at a load factor L, = 0.58 and plastic flow
riencing smaller shear strains than those develfirst develops at the edge of the footing at a lower
oped in a triaxial test at 0.1% axial strain.
factor of 0.38. The latter ratio is dependent on the
In Fig. 11 the elastic stress changes beneath the
finite element mesh geometry, and if an infinitely
centre of the loaded area are compared with the
fine mesh were to be used for the analysis yield
LPC2 model for load factors of 0.3 and 0.52. It
would occur under the perimeter
at even the
can be seen that the influence of the non-linearity
lightest loads. The approximation
allows the
of the stress-strain
characteristics
is to increase
effects of non-linearity
and local failure to be
the vertical and horizontal stresses. The effects are
separately assessed.
most pronounced
for the radial stresses at a
Figure 8 gives plots of normalized
settlement
depth beneath the footing and are less marked for
against depth and these show settlement reducing
the vertical stresses. It is of considerable
interest
far more rapidly with depth than elastic theory
to note that the changes in deviator stress are
predicts. These features develop long before first
much less sensitive to non-linearity.
Thus the
yield occurs and the profile at L, = 0.52 shows
approximation
employed
earlier in the simple
that the trend is accelerated by the onset of local
illustrative problem would not have given rise to
failure.
Delivered by ICEVirtualLibrary.com to:

IP: 129.31.210.87
On: Fri, 26 Aug 2011 17:33:26

NON-LINEAR

STRESS-STRAIN

385

CHARACTERISTICS

0
1

2-O

,o

__

L, = O-52
>__--

Fig. 9.

50

4.0

3.0

______-_--------_______-----

_-----

6.0
--

Profiles of surface settlement adjacent to a rigid footing (see Fig_ 7)

large errors in the calculation of undrained centre

line settlement.
Figure 12 shows the vertical base contact stress
distributions.
It can be seen that the influence of
non-linearity
and local yield is to distribute the

stresses
more
uniformly
by shedding
load
towards the centre and to decrease the stress concentrations
considerably
at the edges (where the
stresses are infinite in the linear elastic cases).
The assumption
that, for a known surface
stress distribution,
soil stresses can be calculated
from linear elasticity is central to routine foundation engineering. To investigate this aspect calculations were carried out for a circular flexible load
with Z/D = 5.0. The plots of centre line stresses
for load factors of 0.3 and 0.5 are compared with
elastic profiles in Fig. 13. In this case the vertical
stresses are insensitive to the constitutive law, but
elasticity underestimates
the radial stresses and
overpredicts the deviator stress profile. This result
provides
further
evidence
of the validity
of
employing
Boussinesq
theory for the vertical
stresses in settlement calculations.
In summary,
with footings,
the non-linear
stress-strain
characteristics
have a dominant
influence on the form and scale of the displace-

(4

(0

e
C

contour1

E: %

+%EiT
; I ;:y
D

0.05

0.1

0.35
1 0.5

Fig. 10. Contours of deviatoric strain beneath a rigid

footing when L, = 05 (see Fig. 7)

Fig. 11.

Distributions

(a) radial stress changes;

(b)

of stress increment beneath the centre of a rigid footing (see Fig. 7):
(b) vertical stress changes;

(c) deviator

stress changes

Delivered by ICEVirtualLibrary.com to:

IP: 129.31.210.87
On: Fri, 26 Aug 2011 17:33:26

JARDINE, POTTS, FOURIE

386

features with the horizontal loading of piles. The

problem geometry was idealized by considering a
disc-like assembly of finite elements, with the
mesh extending to 100 times the internal diameter
of the cavity. Plane strain in the vertical direction
was imposed, and increments of radial displacement were applied at the inner radius of the
cavity. (Solutions to this problem can be derived
without resort to finite element methods although
it is frequently just as convenient to make use of
them.) The analyses were again performed using
LPC2 and a linear elastic soil with E, = 1056
MN/m and v, = 0.49. The analysis was continued until the cavity had been expanded to 1.5
times its initial diameter, when LPC2 gave a total
pressure of 10,9C,.
Figure 14(a) shows the predicted relationships
between P/P,,,
and A V/( V, + A V) where P is the
increase in cavity pressure from the initial radial
stress, V, is the initial volume of the cavity and V
is the change in the volume of the cavity. The
ratio P/P,,,
is a load factor L, and it can be seen
that yield occurs at a relatively early stage.
The curve predicted
using LPC2 may be
treated as being equivalent to field data obtained
from an ideal undrained pressuremeter
test, and it
is of interest to plot the curve of normalized
Youngs modulus EJC, against P/P_,
and P/C,,
as shown in Fig. 14(b). In most pressuremeter
testing a single shear stiffness G = EJ3 is evaluated from an unload-reload
loop, the magnitude
of which can vary between 0.X
and 3.5C,,
depending on the judgement
of the operator. A
linear shear modulus is then determined
from a
mean line drawn through the loop. It is clear
from Fig. 14(b) that the deduced modulus is likely
to be highly sensitive to the size of the imposed
pressure cycle for soils such as LPC2.

---'Elastic'
----_L,
-L,
D

*-

03

0.52
average footing
=

stress

2
4
PA
:
2z
iO8iz
v)

-__-_---

m
_e

-_----

5 04

I
I

0.2

0.4

Fiad~al distance

O-8

0.6

from

centre

of footing

10

r/r0

Fig. 12. Vertical contact stresses beneath a rigid footing

(see Fig. 7)

ment distributions
and a less marked, but none
the less significant, influence on the stress distribution. It is of interest to note that a Gibson
soil (linearly elastic with stiffness increasing hnearly with depth) gives rise to very similar effects,
i.e. a concentration
of vertical displacements
beneath and around the loaded area, a reduction
in variations
in base contact stresses beneath
rigid footings and relatively small deviations of
vertical
stresses
from homogeneous
elasticity
beneath uniformly loaded areas.
Expansion

of long cylindrical

cavity

Cylindrical cavity expansion analysis is of considerable practical interest, as it is used to interpret pressuremeter
data, and has some common

02

04

(a)

06

02

AND BURLAND

04

06

08

ib)

10

(Cl

Fig. 13. Distributions of stress increment beneath the centre of a flexible circular footing with the
geometry shown in Fig. 7: (a) radial stress changes; (b) vertical stress changes; (c) deviator stress
changes

Delivered by ICEVirtualLibrary.com to:

IP: 129.31.210.87
On: Fri, 26 Aug 2011 17:33:26

NON-LINEAR

STRESS-STRAIN

CHARACTERISTICS

12
AVl(V,

387

20

16

AV). %

(a)
P/C
2

10

I
I
(
I
I
I
I

1 Range for magnitude

o-2

0.4

0.6
Load factor P/P,,,

0.8

1.o

ib)

Fig. 14. (a) Pressure-volume change curves for cylindrical cavity expansion and (h) the
variation in apparent secant modulus with load factor for cylindrical cavity expansion in
soil type LPC2

Delivered by ICEVirtualLibrary.com to:

IP: 129.31.210.87
On: Fri, 26 Aug 2011 17:33:26

388

JARDINE,

erkal

POTTS, FOURIE

displacements

applied to top 01 pile

A+7

de7

50 m

+
Fig. 15.

AND BURLAND

Finite element mesh for a pile 30 m long

The analysis of a full pressuremeter

test was
not attempted because of the assumptions
which
would have to be made concerning
the stressstrain
characteristics
associated
with loading
reversals. However, the curves obtained from high
quality tests in clays often show a non-linear
similar to that given in Fig. 14(a)
response
(Windle & Wroth, 1977; Powell & Uglow, 1985).

the maximum
value given by LPC2 (as
described later, additional
elastic runs were
carried out with a range of E, values to aid in
the interpretation
of equivalent soil stiffness
from load-settlement
data)
(c) P3: as for Pl with the pile modulus increased
by a factor of 103.

The relationships
between load and settlement
of the pile head are given in Fig. 16. In both of
the elasto-plastic
cases local plastic
failure
Axial loading of pile 30 m long
(indicated by arrows) was reached at settlements
The third class of problem to be investigated
of less than 2 mm and the stiffness of the pile can
was that of a compressible
pile. A solid pile,
be seen to have an important
influence on the
0.75 m in diameter and 30 m long, was selected
behaviour. (The finite element mesh designed for
with a modulus
of 30 x 10 MN/m.
Such a
this study was not chosen to give particularly thin
stiffness is appropriate
to either a steel pipe pile
elements close to the pile. With the shaft shear
or a reinforced concrete pile, and was 28 times the
stresses being projected from points around r,J5
maximum soil stiffness. The finite element mesh
from the interface, there was a tendency for the
for the study is shown in Fig. 15, and no account
ultimate capacity to be overpredicted
by as much
was taken of any effects of installation
on soil
as 7%. In analyses where the calculation of ultiproperties
or initial conditions.
Loading
was
mate loads is of greater importance, the accuracy
simulated by applying increments of vertical discan be improved by using a finer mesh.)
placement to the top of the pile. Three cases were
Figure 17 shows the radial profiles of relative
considered
surface settlement (6,/d,), where 6, is the settlement at radius r from the pile centre and 6, is the
(a) PI: the soil was everywhere represented
by
settlement of the pile. The profiles for Pl and P3
the non-linear model LPC2 including the soil
correspond
to a load factor of 0.5. It is apparent
immediately
adjacent to the pile shaft (i.e.
that linear elastic theory gives a poor estimate of
Lx= 1)
the surface settlement profile around a typical pile
(b) P2: the soil was linear elastic Delivered
with E, taken
as
by ICEVirtualLibrary.com
to:

IP: 129.31.210.87
On: Fri, 26 Aug 2011 17:33:26

NON-LINEAR

STRESS-STRAIN

389

CHARACTERISTICS

LPC2(cr=1)(P1)Ep=30X103MN/m2
- - ~
-----xx

(Arrows

lndlcate

points

Linear

e&AC

LPC2(a

of first

D
0.8

b
z
1 .o

Fig. 17.

30 X 1 O3 MN/m*

30X106

MN/m

20

16

mm

curves for a pile 30 m long

installed in a soil with the stress-strain

characteristics of LPC2. However, it is of interest to note
that the elastic profile can be almost recovered if
the pile-soil stiffness ratio is increased by 103, i.e.
the pile becomes very stiff.
The location
and extent of local failure at
working load is shown in Fig. 18 in which contours of a deviatoric strain E are plotted for case
Pl with a load factor of 0.5. It can be seen that
yielding has occurred near the ground surface
and that, except for a very narrow zone immediately adjacent to the pile, the bulk of the soil

12

Load-settlement

(I??) E,

l)(P3)Ep

yield)

Settlement:

Fig. 16.

experiences
only
very
small
strains,
with
E<O.05%.
Fig. 19 shows the mobilization
of
shaft resistance r/C, with depth for Pl and P3 at
L, = 0.5. It can be seen that progressive failure is
taking place and that the extent of this is dependent on both the relative compressibility
of the
pile and the pre-failure stress-strain
characteristics of the soil.
The same tendencies
of load shedding
and
plastic flow near the pile head give rise to the
pronounced
concentrations
of ground movements
near to the pile. It is therefore clear that the

/i*
-

o-x-x

--O

LPC2,

L, =

Linear

elastic,LPCZ

O-5, E,

As Pl

but with

E,

=
E
=

30X
=

lo3

9 0 X lo6

Profiles of surface settlement adjacent to a pile 30 m long (see Fig. 16)

Delivered by ICEVirtualLibrary.com to:

IP: 129.31.210.87
On: Fri, 26 Aug 2011 17:33:26

MN/m(Pl)

30 X 1 O3 MN/m2
MN/m2

(P3)

(P2)

JARDINE,

390

POTTS, FOURIE

AND BURLAND

Zone of plastic
behaviour
to pile

close
I

Fig. 18. Contours of deviatoric strain around a pile 30 m long when

L,O= 05 (see Fig. 16)

overall behaviour is sensitive to the combination

of large strain properties close to the pile and the
small strain characteristics
of the surrounding
soil. These features are significant in design and
are likely to affect the capacities and working settlements of both single piles and pile groups. It is
apparent
that linear elastic theory will tend to
Normalized

shear

stress:

TJC,

overpredict group settlement ratios and to exaggerate the non-uniformity

of loads within rigidly
capped pile groups. In assessing the interaction of
pile groups it is thus necessary to consider the
initial response of the soil to shearing with the full
accuracy afforded by the new laboratory
techniques, and this is particularly so when considering the response of large offshore piled structures
(Jardine, 1985).

Strutted excavation
The prediction of structural forces and ground
movements around deep excavations has important implications
for construction
in built-up
areas. These problems
have prompted
much
research, and some of the difficulties of assuming
linear elastic soil behaviour
have first become
apparent from the monitoring
of such structures.
It was therefore considered important to include
an analysis of a hypothetical
strutted excavation
in the present series of studies.
The excavation considered was infinitely long,
40 m wide, 15.26 m deep and was supported by a
diaphragm
wall 20 m deep, propped
at the
surface by means of rigid struts before excavation.
The finite element mesh is shown in the inset to
Fig. 20 and excavation was simulated by sequentially removing layers of elements from within the
excavation.
The final depth of 15.26 m was
LPC2,Ei,=30X103MN/m(P1)
chosen to give a factor of safety F, of 1.5 in terms
--x
~
LPC2
Ep= 30 X 1 O6 MN/m2 (P3)
of undrained shear strength. The wall adhesion
factor was taken as 0.X,. The material composFig. 19. Variation in shear stress with depth down a pile
ing the diaphragm
wall was specified as being
30 m long for two pile stiffnesses when L, = 05 (see
linear elastic with E, = 28 x lo3 MN/m
and
Fig. 16)
Delivered by ICEVirtualLibrary.com to:

IP: 129.31.210.87
On: Fri, 26 Aug 2011 17:33:26

NON-LINEAR

STRESS-STRAIN

391

CHARACTERISTICS

employlng-

analws

Dlsplacemenl

towards

excavatfon:

Excavation

1 ,o

2.0

3.0

1 ,o

cm

Normalzed

Fig. 20. Variation in maximum horizontal wall displacement with depth of excavation for a propped diaphragm retaining wall

depth

15-26 m
6

0.6
horizontal

0.2
displacement

w/wloe

Fig. 21. Horizontal deflected profiles of a diaphragm

wall for a depth of excavation of 1526 m (see Fig. 20)

is very sensitive to the soil model. The incorporation of non-linearity

and plasticity
into the
model gives rise to a pronounced
settlement
trough close to the excavation which cannot be
(a) LPC2
matched using linear elasticity. As mentioned in
(b) linearly
elastic with E, = 1056 MN/m
and
the introduction,
this phenomenon
was noted in
v, = 0.49.
the field by Burland & Hancock (1977) and was
For both cases the initial stresses were assumed
later reproduced
analytically
by Simpson et al.
to increase linearly with depth Z such that crV=
(1979) using a bilinear elastic perfectly plastic soil
202 kN/m and un = 0,7a,. The undrained shear
model.
strength was taken everywhere as 110 kN/m.
The non-linear
and elastic solutions also give
Figure
20 shows the relationship
between
rise to different
stress distributions
over the
maximum inward displacement
of the wall, wmaxr
retaining
wall, and hence different
bending
and the depth of excavation. Local failure is initimoments and prop forces. Thus, for a depth of
ated before the load factor L, = l/F, reaches 0.2.
excavation of 15.26 m the LPC2 model predicts a
The load-displacement
curve computed with the
prop force of 387 kN/m while linear elasticity
LPC2 model is clearly non-linear
with wall
gives 316 kN/m.
deflexions accelerating as excavation takes place.
Contours
of deviatoric
strain are given in
Fig. 21 shows the normalized horizontal deflected
Fig. 23 with the excavation at 15.26 m. A zone of
profiles of the wall, w/wtoer at the final depth of
contained
plastic behaviour
is found with the
excavation found with the two soil models. It can
non-linear
soil, and the contours show that the
be seen that the shapes of the wall profiles do not
deviatoric strains in the surrounding
ground fall
differ appreciably.
between 0.3% and 0.05%.
Figure 22 shows the profiles of surface settleThese findings have been further reinforced by
ment with distance from the wall, expressed as a
recent field and analytical studies of the strutted
proportion
of w,,,.
Unlike the displacement
excavations
for the Bell Common cut and cover
profile for the wall, the surface settlement profile
tunnel (Fourie, 1984; Tedd, Chard, Charles &
Delivered by ICEVirtualLibrary.com to:
v, = 0.15. Two cases
soil modelled
as

were

considered

with

the

IP: 129.31.210.87
On: Fri, 26 Aug 2011 17:33:26

392

JARDINE,

POTTS,

FOURIE

AND BURLAND

Linear elastic
analysis

Fig. 22.
Fig. 20)

Vertical

displacement

profiles

adjacent

to a strutted

Symons,
1984; Hubbard,
Potts,
Miller
&
Burland, 1984). Fourie carried out finite element
analyses of the excavations using an elasto-plastic
effective stress model for London Clay, in which
the pre-yield behaviour was described in a similar
way to the LPC2 model. The stiffness parameters
were derived from the instrumented
laboratory
tests described by Jardine et al. (1985) and it is
encouraging
that excellent agreement was found
between the measured and predicted behaviour.

excavation

(see

calculated
by relating a characteristic
displacement to a known loading condition,
such as
centre line settlement to mean bearing pressure.
The same method has been applied to the loaddisplacement
data calculated using model LPCZ.
Thus the computed loaddisplacement
curves are
treated as if they were experimental data gathered
in the field. The variations in apparent modulus
with load factor L,, produced
by the different
boundary value problems, can then be compared.
Figure 24 shows the variation in E,*/C, and L,
for the rigid footing on a deep clay layer. It can
INTERPRETATION
OF LOAD-DISPLACEMENT
be seen that even for load factors as low as oneBEHAVIOUR USING LINEAR ELASTICITY
third the value of E,* reduces from its initial
The finite element
studies
presented
here
value by about 40%. The broken line in Fig. 24
provide insight into the effects of non-linear soil
represents the variation in secant E,* with L, for
properties
in soillstructure
interaction.
In this
a triaxial test with the soil model LPC2 (i.e. test
section the results of the various studies are used
R2 in Figs l(b) and 5). In this case L, =
to investigate
the choice of equivalent
elastic
(q - qO)/(qr- qo) where q is the deviator stress
design parameters
and to draw attention to the
and q. and qf are the initial and failure values
difficulties of linear elastic interpretations
of in
respectively. It can be seen that the two curves
situ tests and full-scale field monitoring.
are almost identical up to a load factor of about
It is common
to interpret
field
load0.5, i.e. for most practical ranges of working load.
displacement
behaviour in terms of linear elasHowever, as L, increases above 0.5 and the zones
ticity. An apparent
Youngs modulus
EUA, is
of local failure spread,
the apparent
moduli
derived from the displacements
of the footing fall
RIgId
below the values from the triaxial test.
Similar relationships
between E,* and L, have
been derived for the other boundary value problems and are plotted in Fig. 25, together with the
result from the triaxial compression
test. This
latter curve may be conveniently
used as a basis
for comparison.
The relationship
for the 30 m
long pile needs special mention since it is complicated by the effect of pile compressibility.
The
relationship
between
E,* and 1oad:settlement
ratio was obtained by carrying out eight linear
Fig. 23. Contours of deviatoric strain around a strutted
elastic analyses in which E,* was varied but the
excavation for a factor of safety on strength F. = 1.5
(see Fig. 20)
pile stiffness was fixed. Hence, for any given point
Delivered by ICEVirtualLibrary.com to:

IP: 129.31.210.87
On: Fri, 26 Aug 2011 17:33:26

NON-LINEAR

Rigld

--

02

Tnaxml

04

fooilng
test

O-6

Load

factor

STRESS-STRAIN

Z/D

on curve Pl in Fig. 16, it was possible to obtain

the apparent modulus E,* and thereby to derive
the relationship
between E,*/C, and L, given in
Fig. 25.
For all the cases referred to in Fig. 25, with the
exception of the strutted excavation, the apparent
modulus E,* relates to the deflexion of the point
of application of the load (or stress), and the load
factor is clearly defined. For the strutted excavation the value of E,* relates to the maximum
horizontal
deflexion of the wall and the load
factor is the inverse of the factor of safety on
undrained strength.
Bearing in mind that Fig. 25 relates to a specific non-linear
soil model applied to particular
boundary value problems the following observations can be made.

5.0

R2

08

393

CHARACTERISTICS

(a) In each case the apparent

modulus reduces
continuously
as the load factor increases, and
in no case can the continuum
behaviour be
properly described as linear elastic at working
loads.
(b) For any given load factor there is a considerable range of values of E,*/C,. For example,
at L, = 0.5 the rigid pile gives the stiffest
response
(E,*/C, = 3600) and the cavity

10

L,

Fig. 24. Variations in apparent secant modulus with

load factor for a rigid footing and an undrained triaxial
test for soil type LPC2
5OOOr
RIgId

0.4

0.2
LOad

30

factor

L/L,,,,

m long,

0.75

dia.

and

1 IF s

0.6
PIP,,,

ptle

0.6

1 .o

Fig. 25. Summary diagram showing variations in apparent modulus

with load factor (l/F, for the excavation)

Delivered by ICEVirtualLibrary.com to:

IP: 129.31.210.87
On: Fri, 26 Aug 2011 17:33:26

394

JARDINE,

POTTS, FOURIE

AND BURLAND

expansion
curve gives the softest response
(E,*/C, = 300) and these two curves fall to
either
side of the triaxial
characteristic.
(Intuitively, similar results might be expected
for the horizontal
loading of the pile.) Thus
field moduli deduced from experiments in the
same soil, but with different types of boundary conditions, can be radically different even
for an isotropic material such as LPC2.
(c) First plastic yield occurred over a wide range
of load factors. Hence the values of E,*/C, at
the onset of yield vary from 3700 for the strutted excavation (L, = 0.14) to 800 for the triaxial test (L, = 1.0). If an infinitely fine mesh
had been employed for the footing analysis,
first yield would have occurred with L, = 0.0
when E,*/C, = 4800.

the limitations of the study, the following preliminary conclusions can be made.
There can be considerable
dihiculties in applying linear elastic theory to the prediction
of
ground movements and soil stresses induced by
different types of structure. In all the practical
cases studied, the modelling
of realistic small
strain non-linearity
and the consideration
of local
failure have important implications in considering
soil-structure
interactions at working loads.
In footings and excavations
the small strain
characteristics
appear to have the greatest influence on the deflexion profiles around the loaded
boundary.
With piled foundations
the onset of
local failure appears to be at least as significant,
and the combination
of these two kinds of nonlinearity appears to control pile group interaction
and progressive failure.
The two extreme cases in Fig. 25 are of interest
For all the cases studied, the large mass of the
(i.e. the expanding cavity and the rigid pile) since
soil influenced
by the boundary
loading was
both form the basis of in situ tests. It is clear that
strained to less than 0.1% deviator strain and
great care is needed in evaluating the stiffness of
frequently to less than 0.05%. If representative
the ground from such data. The value of these
soil parameters
are to be determined experimentests would be increased if the full characteristic
tally, highly accurate measurements
are required.
of apparent
modulus
with load factor were
Thus
the
precision
offered
by
the
new
laboratory
reported, rather than a single arbitrary stiffness
techniques is of considerable practical value.
value. It may be feasible to use such a characterIn problems such as footings it may be reasonistic to estimate the non-linear stress-strain
charable to combine stresses predicted
from linear
acteristics of individual soil elements.
elastic theory and measured
non-linear
stressIt is of considerable
practical interest to note
strain characteristics
to carry out approximate
the reasonably
good agreement
between
the
evaluations
of centre line settlements.
In other
results derived from the settlement
of footings
cases, such as the estimation
of group displaceand the triaxial element test. Good agreement is
ments for large piles, or the calculation
of wall
also found between the 30 m pile and the triaxial
bending
moments
in deep excavations,
linear
test when a realistic pile stiffness is assumed.
elastic theory is less satisfactory.
However, a rigid pile gives a far stiffer characterThe studies show that the back analysis of fullistic than the triaxial curve and explains why the
scale performance
or in situ tests is likely to lead
modulus back calculated from pile tests is often
to a wide range of possible values for deformation
so much higher than that obtained
by other
moduli, even for a uniform isotropic material.
methods.
Recent research has confirmed
the expected
DISCUSSION AND CONCLUSIONS
sensitivity of stiffness to boundary conditions and
The series of analyses are intended to give a
the loading level in field and laboratory
tests on
preliminary
appraisal of the effects of the nonLondon Clay (Jardine et al., 198.5). The initial
linear soil behaviour
observed in recent laborstiffnesses observed in triaxial tests can exceed
atory tests. To restrict the number of variables,
overall values deduced from either high quality in
only the undrained behaviour of a homogeneous
situ tests or the back analysis of full-scale perlayer of an isotropic material under monotonic
formance.
loading is considered.
Although linear elasticity remains a convenient
The simple empirical stress-strain
expression
tool for expressing measurements
of soil stiffness,
used for the calculations
provides a good fit to
its limitations must be recognized. In particular
the undrained
behaviour
of a lightly overthe importance
of load factor and the tendency
consolidated
low plasticity clay in triaxial comtowards concentrations
of strain close to loading
Non-homogeneity
can be considered
pression.
boundaries
must be taken into account. If the
without undue difftculty, but if drained conditions
non-linear
nature of soils is not acknowledged,
(or cycles of loading) were to be considered
a
comparisons
of field and laboratory
measuremore complex model is required. The material
ments can be confusing.
considered,
LPC2, is probably stiffer than most
The use of instrumentation
systems in field
soils but is not unusually non-linear. Recognizing
monitoring
which allow the direct determination
Delivered by ICEVirtualLibrary.com to:

IP: 129.31.210.87
On: Fri, 26 Aug 2011 17:33:26

NON-LINEAR

STRESS-STRAIN

of strains (and their comparison

with stresses) is
likely to be of great value in understanding
soil
behaviour and assessing the applications
of the
laboratory techniques.
Although
the apparent
agreement
between
laboratory
non-linear
characteristics
and measured field behaviour is most encouraging, further
studies are required to investigate the effects of
different stress paths, drainage
conditions
and
strain rates.
Finally the pre-yield
non-linear
formulation
described in this Paper can be used more generally and has been included
in analyses
where
more sophisticated,
models have been
Jardine
(1985).

effective
stress, elasto-plastic
required;
see Fourie (1984) and

CHARACTERISTICS

395

measurement
of soil stiffness in the triaxial apparatus. Gtotechnique 34, No. 3, 323-340.
Marsland,
A. (1971). Laboratory
and insitu measurements of the deformation
moduli of London clay.
Proc. Symp. Interaction of Structure and Foundation,
July. Midland
Soil Mechanics
and Foundation
Engineering
Society.
(Also
Building
Research
Station Current Paper CP 24/73.)
Morgenstern,
N. R. & Phukan, A. L. T. (1968). Stresses
and displacements
in a homogeneous
non-linear
foundation.
Proc. Int. Symp. Rock Mech., Madrid,
pp. 3 13-320.
Poulos, H. G. & Davis, E. H. (1974). Elastic solutions fin
soil and rock mechanics. New York: Wiley.
Powell, J. J. M. & Uglow, I. M. (1985). A comparison
of
Menard, self-boring and push-in pressuremeter
tests
in a stiff clay till. Adv. Underwat. Technol. Offshore
Engng 3,201-219.
Simpson, B., ORiordan,
N. J. & Croft, D. D. (1979). A
computer
model for the analysis of ground movements in London Clay. Geotechnique 29, No. 2, 149175.
St John, H. D. (1975). Field and theoretical studies of the
behaviour of ground around deep excavations in
London clay. PhD thesis, University of Cambridge.
Tedd, P., Chard, B. M., Charles, J. A. & Symons, I. F.
(1984). Behaviour of a propped embedded retaining
wall in stiff clay at Bell Common Tunnel. Gtotechnique 34, No. 4, 513-532.
Vesic, A. S. (1975). Bearing capacity of shallow foundations (eds H. F. Winterkorn
and H. Y. Fang),
Chap. 3, Foundation
Engineering
Handbook.
New
York: Van Nostrand Reinhold.
Windle, D. 8~ Wroth, C. P. (1977). lnsitu measurements
of the properties
of stiff clays. Proc. 9th Inc. Conf
Soil Mech. Fdn Engng, Tokyo 1, 347-352.
Wroth, C. P. (1971). Some aspects of the elastic behaviour of overconsolidated
clay. Proc. Roscoe Memorial Symp., pp. 347-361. London: Foulis.

REFERENCES
Burland, J. B. (1975). Some examples of the influence of
field measurements
on foundation
design and construction. Proc. 6th Regional Conf for Africa Soil
Mech. Fdn Engng, Durban 2.
Burland, J. B. & Hancock, R. J. R. (1977). Underground
car park at the House of Commons:
geotechnical
aspects. Struct. Engr 55, 877100.
Burland, J. B., Sills, G. C. & Gibson, R. E. (1973). A
field and theoretical
study of the influence of nonhomogeneity
on settlement. Proc. 8th Inc. Con/ Soil
Mech. Fdn Engng, Moscow 1.3, 31-46.
Burland, J. B. & Symes, M. (1982). A simple axial displacement
gauge for use in the triaxial apparatus.
Geotechnique 32, No. 1,62X5.
Costa-Filho,
L. M. (1980). A laboratory investigation of
the small strain behaviour of London clay. PhD
thesis, University of London.
Costa-Filho,
L. M. & Vaughan P. R. (1980). Discussion
on A computer
model for the analysis of ground
movements in London clay. Geotechnique 30, No. 3,
336339.
APPENDIX
1
Daramola,
0. (1978). The influence of stress history on
Calculation
of non-linear
parameters from
test
the deformation of sand. PhD thesis, University
of
data. Referring to Fig. 3, first locate the observed, or
London,
projected, maximum
stiffness point. Now maxima for
Eisenstein,
Z. & Medeiros,
L. V. (1983). A deep
equation (1) occur when
retaining
structure
in till and sand: part II, performance and analysis. Can. Geotech. J. 20, No. 1,
131-141.
cos
0
Fourie, A. B. (1984). The behaviour of retaining walls in
i.e.
when
sttrclays. PhD thesis, University of London.
Gibson, R. E. (1967). Some results concerning displaceflog,,
2nn
ments and stresses in a non-homogeneous
elastic
half-space. Geotechnique 17, No. 1, 58-67.
assuming
n = 0 for the observed
maximum
gives
Hubbard,
H. W., Potts, D. M., Miller, D. & Burland, J.
log,, (EJC) = 0 and C = A.
B. (1984). Design of the retaining walls for the M25
Next the crossing point where the angular part of
cut and cover tunnel at Bell Common. Geotechnique
equation (1) must equal n/2 is located, so that
34, No. 4, 495-512.
Jardine, R. J. (1985). Investigations ofpile-soil behaviour
with special reference to the foundations of offshore
(5)
structures. PhD thesis, University of London.
Jardine, R. J., Fourie, A., Maswoswe, J. & Burland, J. B.
then the minimum point where the angular part must
(1985). Field and laboratory
measurements
of soil
equal rr is located, so that
stiffness. Proc. 11th Inc. Conf Soil Mech. Fdn Engng,
San Francisco 2,51 I-514.
Jardine, R. J., Symes, M. J. & Burland, J. Delivered
B. (1984). The
by ICEVirtualLibrary.com to:

{fIog,,(;)J} ?=
@)y=

IP: 129.31.210.87
On: Fri, 26 Aug 2011 17:33:26

396

JARDINE, POTTS,

AND BURLAND
The two
parameters
A
B can
taken
directlv
the stiffness
corresnondine.
to
strainsc,
D
E.
Equation (1) should then be evaluated for a number
of points to find the degree of departure
from the test
data near the upper and lower limits of strain. The
limits should be selected to prevent negative tangent
stiffnesses from being predicted,
and the lower limit
should not usually be less than O+JOl %I.
The maximum, minimum and crossing points can be
reselected if the degree of fit is unsatisfactory.
If it is required to evaluate the expressions
for A
-c C, problems will arise when raising the logarithmic
terms to a fractional power. In this case pre- and postmultiplication
by - 1 will be required.
I

(7
From the result of equation
from equation (5) and

(7) a can be obtained

42
a = Clog,,

wc)Iy

Delivered by ICEVirtualLibrary.com to:

IP: 129.31.210.87
On: Fri, 26 Aug 2011 17:33:26