Вы находитесь на странице: 1из 17

Vi&b,

G. & Atkinson,

J. H. (1995). Gdotechnique

45, No. 2, 249-265

Stiffness of fine-grained
G. VIGGIANI*

and J. H. ATKINSON*

The stiffness of soil at very small strain G, is a


useful parameter for characterizing the non-linear
stressstrain
behaviour of soil for monotonic
loading and it is required for analyses of the
dynamic and small strain cyclic loading of soils.
Tests were carried out on fine-grained soils in a
hydraulic triaxial cell fitted with bender elements
and with local axial gauges. From the results of
these tests simple expressions were obtained which
describe the variation of GO with current state in
terms of the current stress and overconsolidation
ratio. The parameters in these expressions were
found to depend on plasticity index. The simple
expressions for G, were found to apply generally
at larger strains, with the values for the parameters
also depending on the current strain. Values of G,
measured in laboratory tests on reconstituted
London clay agree well with values measured in
tests on undisturbed samples and in field tests
which make allowance for the different states in
the various tests.
KEYWORDS:
clays; dynamics; elasticity;
equipment; laboratory tests; stiffness.

soil at very small strains

laboratory

INTRODUCTION
The shear stiffness of soil measured in dynamic
field and laboratory tests is generally significantly
greater than the shear stiffness measured in conventional
triaxial
tests,
assuming
that
the
stress-strain
behaviour is linear. As a result it was
generally believed that stiffnesses measured
in
dynamic tests did not represent the stiffness of
soil in monotonic
loading and were applicable
only to dynamic loadings such as earthquakes,
shocks or machine
vibrations.
Dynamic
tests
investigating the variation of shear modulus with
shear strain amplitude showed that the stiffness
reduced with increasing
strain as in Fig. l(a)
(Anderson
& Richart, 1976). Conventional
triaxial tests, often with local measurement
of strain,
showed a similar reduction of shear modulus with
strain, as in Fig. l(b) (Jardine, Symes & Burland,
1984). The non-linearity
of the stress-strain

Manuscript
received 22 March 1993; revised manuscript accepted 23 March 1994.
Discussion on this Paper closes 1 September 1995; for
further details see p. ii.
* City University, London.

La rigidite G, dun sol, sous tres faible deformation, est un parametre interessant qui permet de
caractiriser la non-linearite du comportement en
contrainte-deformation
de ce sol lors dun
chargement monotone et danalyser les cycles de
chargement dynamique a faible deformation. Des
essais ont iti! realises en celhde triaxiale hydraulique sur des sols finement genus equip&s de cap
teurs en flexion et de jauges axiales locales. Les
resultats obtenus au tours de ces essais ont permis
de definir des relations simples donnant la variation de GO en fonction de la contrainte courante et
du degre de surconsolidation. Les parametres dependent de Iindice de plasticite. Une expression
simple de G, est applicable a de plus fortes deformations, les parametres &ant alors en plus fonction de la deformation courante. Les valeurs de GO
mesurees en laboratoire sur de largile de Londres
reconstituQ sont en bon accord avec celles
obtenues sur des Cchantillons intacts ou lors
dessais in-situ et rendent compte des differents
&tats rencontres lors des differents essais.

behaviour of soils was also inferred from backanalysis of field observations (Simpson, Calabresi,
Sommer & Wallays, 1979).
Figure l(c) is an idealization
of soil stiffness
over a large range of strains, from very small to
large, and approximately
distinguishes
strain
ranges. At very small strains the shear modulus
reaches a nearly constant limiting value G,. For
reconstituted
soils the strains
at which the
stiffness starts to decrease varies with plasticity
from about 0.001% for low-plasticity
soils to
about 0.01% for plastic clays (Georgiannou,
Rampello & Silvestri, 1991; Lo Presti, 1989). At
strains exceeding about 1% the stiffness is typically an order of magnitude
less than the
maximum, and it continues to decrease as the
state approaches failure. In the intermediate small
strain range the stiffness decreases smoothly with
increasing
strain. Strains in the ground
near
structures in stiff soils are generally in the small
strain and very small strain regions (Burland,
1989).
Non-linear numerical analyses have been used
successfully to predict movements around engineering structures
(Jardine,
Potts, St. John &
Hight, 1991) where the introduction
of non-

250

VIGGIANI

AND ATKINSON

linearity alone considerably improved the quality


of the prediction.
Non-linear
models for soil
behaviour may be developed using the theory of

10-Z
Shear ytryin

%1-

10-l
Shear strain:

10

1
%

(b)

IL.arger
Small

strains

(\

:e;trains

-L -

Shear strain (log scale)


(4

Fig. 1. Typical variation of stiBoess with strain for soil:


(a) dynamic tests on fine-grained soils (after Anderson 81
Richart, 1976); (b) triaxial tests on London clay (after
Jardine et al., 1991); (c) idealization for a wide range of
strains (after Atkinson & Sallfors, 1991)

elasticity with empirical non-linear


stress-strain
curves obtained from laboratory stress path tests
(Jardine & Potts, 1988) or using kinematic hardening elasto-plastic
models with multiple plastic
potential
surfaces
(Stallebrass,
1990). In the
former case the initial value of shear modulus is
useful for defining the starting point of an empirical stress-strain
curve; in the latter case it is
needed to define the stiffness inside the innermost
yield surface.
There is considerable
evidence that soil behaviour within the region of very small strain is
linear and elastic. In both slow and dynamic
cyclic loading tests stress-strain
loops show little
or no hysteresis, which means that the behaviour
is conservative
and little or no energy is dissipated (Papa, Silvestri & Vinale, 1988; Silvestri,
1991). Volumetric
and shear deformations
are
fully recoverable (Lo Presti, 1989) and uncoupled
so that no pore pressures are generated during
undrained
shear (Georgiannou
et al., 1991; Silvestri, 1991).
Values of the shear modulus at very small
strains G, can be measured using dynamic techniques in field and laboratory
tests (Atkinson &
Sallfors, 1991), in which the deformation
properties of the soil are related to elastic shear wave
velocities. In the laboratory, the most common is
the resonant column test in which the response of
a cylindrical sample subjected to forced harmonic
torsional
vibrations
is measured. The resonant
column test can be used to evaluate the stiffness
of soils at shearing
strains
ranging
from
O-00001 % to 1%. However, since analyses of resonant column tests are based on the assumption
that the behaviour of the soil is linear and elastic,
analyses of the test data are strictly valid only in
the region of very small strain (Isenhower, 1979).
An alternative
laboratory
technique
involves
transmitting
and receiving shear waves using
small electro-mechanical
transducers
known as
bender elements (Shirley & Hampton,
1977). In
bender element tests the strains are not constant
throughout
the sample because of both material
and geometric
damping.
The maximum
shear
strain generated in the soil is, however, very small
and was estimated to be less than 10m3% (Dyvik
& Madshus, 1985). In field tests the velocities of
shear waves can be measured from the surface,
using refraction surveys or Rayleigh wave techniques, or at depth, using the cross-hole or downhole techniques
(Yoshimi,
Richart,
Prakash,
Barkan & Ilyichev, 1977). Once again, the strains
involved in field dynamic tests are not constant,
decreasing from a maximum close to the source.
The shear strain amplitude can be obtained from
the ratio of the particle velocity to the shear wave
velocity, E = u&, so that estimates of the strain
amplitude can be made in individual experiments

STIFFNESS

OF FINE-GRAINED

by using calibrated
geophones.
Field dynamic
tests generally develop strains in the field of
10-3-10-4%
and less.
Direct comparison
between the shear modulus
obtained from dynamic tests and the very small
strain shear modulus
relevant
to monotonic
loading in triaxial compression
or extension tests
is difficult, as the rates of strain and the modes of
shearing are very different in the two types of test.
Recent laboratory test data are available comparing the stiffness of sands under dynamic and
static conditions
(Iwasaki, Tatsuoka & Takagi,
1978; Ni, 1987; Bolton & Wilson, 1989). The
results of these tests indicate that the stiffness of
sands at very small strain is independent
of the
rate of loading. Less experimental
evidence is
available for saturated clays, probably because of
the difficulties connected with the correct definition of the drainage conditions and the measurement of pore pressures
generated
during fast
dynamic loading. Nevertheless,
the present evidence is that values of shear modulus obtained
from dynamic and slow loading triaxial tests are
approximately
equal (Rampello & Pane, 1988;
Georgiannou
et al., 1991).
If the mechanical behaviour of soil is taken to
be essentially frictional the mechanical properties,
including both strength and stiffness, vary linearly
with the mean stress. However, if the soil is idealized as an assembly of elastic spheres in contact,
the theories developed by Hertz as reported by
Richart, Hall & Woods (1970) lead to the result
that the shear modulus should depend on the
mean stress raised to the power of l/3.
The observed behaviour of soil lies somewhere
between these two limits. For sands Wroth &
Houlsby (1985) proposed
a general expression
relating shear modulus to mean stress in the form
G

-_=A

Pr

t
0

SOIL AT VERY SMALL

STRAINS

251

with the contribution


from inelastic slipping and
rearrangement
becoming larger with increasing
strain and stress ratio. Thus the overall behaviour
of soil should be expected to correspond
to a
value of n a little greater than l/3 at very small
strains, because of the small contribution
from
slipping and rearrangement,
with n increasing to
a value very close to 1.0 at large strains as the
contribution
from the elastic deformations
of the
particles becomes relatively small.
Based on experimental
observations
on sands,
Hardin & Black (1966) proposed that the shear
modulus at very small strains could be related to
the mean effective stress raised to a power of 0.5
and to the voids ratio and stress history. Hardin
& Black (1968) assumed that the same relationship would hold also for normally consolidated
clays. A more general expression was proposed
by Hardin (1978) based on theoretical
elastic
stress-strain
relationships
by Rowe (1971) and
empirical equations
for initial tangent modulus
by Janbu (1963) and Hardin & Black (1968). This
can be written in the form
G, = Sf(u)OCRkp,-p

(2)

where S is a dimensionless
coefficient
which
depends on the nature of the soil, f(u) is a function
of the specific volume, p is the mean effective
stress, p, is the atmospheric pressure and OCR is
the overconsolidation
ratio defined as the ratio of
the maximum past stress to the current stress.
Results of tests on soils in resonant column tests
(Hardin & Drnevich, 1972) show that n is less
than 1.0 and k increases from 0 to 0.5 as the plasticity index increases from 0 to 100.
In order to consider the effect of anisotropic
stress states on the very small strain stiffness of
soils, Ni (1987) proposed a more general form of
equation (2)

P,

where the dimensionless


parameters
A and n
depend primarily on the nature of the soil and on
the current strain. The reference pressure p, is
included in equation (1) so that the parameters A
and n are dimensionless
but their numerical
values will depend on the choice of reference pressure. Experimental
results from both dynamic
and static tests on sands indicate that the value of
n varies significantly with strain from values near
0.5 at very small strain to 1.0 at large strain
(Wroth, Randolph, Houlsby & Fahey, 1979).
Recent numerical
studies of the mechanical
behaviour
of large random
systems of elastic
spheres of different sizes show that the overall
deformation
always includes both elastic deformations
of the particles
and slippage
and
rearrangement
(Dobry, Ng & Petrakis,
1989),

G, = Sf(o)OCRkp, -c:cr;c$

(3)

where oe is the stress in the direction of wave


propagation,
or is the effective stresses in the
direction of particle motion and cr8is the effective
stress orthogonal
to the plane of shear, and n =
the results
of resonant
n, + n, + r18. From
column tests on hollow cylindrical samples of dry
washed mortar sand, both in biaxial loading conditions and in true triaxial conditions, Ni (1987)
concluded that G, is affected equally by cre and
err (n, w nr) while the influence of ergis practically
negligible, so that nB z 0.
Equations
(2) and (3) contain
the specific
volume u, the current stess cr or p and the overconsolidation
ratio. If the overconsolidation
ratio
is redefined in terms of the stress at the intersection of a swelling and reconsolidation
line with
the intrinsic normal consolidation
line for recon-

252

VIGGIANI AND ATKINSON

Intrinsic normal

(7)

;
5
g

__________-------

0
E
P
lz

I
P
In

PP

Fig. 2. Relationship between specific volume, stress and


overconsolidation ratio

stituted samples as in Fig. 2, the state at A is


defined by any two of the parametrics
v, p and
R, It is usual to choose the current stress as one
of the parameters
to define the current state of
soil, but the choice of the other parameter v or R,
is a matter of convenience.
For coarse-grained
soils, for which it is often difficult to locate the
intrinsic normal consolidation
line, the most convenient parameter is the specific volume; for finegrained
soils the overconsolidation
ratio is
usually preferred.
At large strains the states of normally consolidated and lightly overconsolidated
soils are on
the state boundary surface and the stiffness at a
particular strain increases linearly with the mean
stress p (Wroth, 1971; Wroth et al., 1979) so that
n = 1 in equation
(1). From
the results of
undrained triaxial tests on normally consolidated
and overconsolidated
samples of a glacial till soil,
Atkinson & Little (1988) found that the value of
G/p at a particular strain increased linearly with
the logarithm
of the overconsolidation
ratio,
which is consistent with the behaviour derived by
Wroth (1971) from a reinterpretation
of tests on
undisturbed
London clay by Webb (1967). These
results can be expressed in the form
G
-=

P'

0
--$

The work described in this Paper was experimental and consisted largely of tests on reconstituted samples in a hydraulic triaxial cell in which
values of G, were measured using bender elements. The principal purpose of this work was to
examine the variation of G, with current stress
and overconsolidation
ratio and to evaluate the
parameters
for some typical fine-grained
soils.
Additional work was carried out to examine the
variation of the shear modulus with stress and
overconsolidation
at larger strains and to determine values of G, for London clay in laboratory
tests on undisturbed samples and in situ.

LABORATORY
PROCEDURES

I
..

Bender

or

PI

07
G

R;;
nc

,.

Slgnal generator
Local axial
gauges

jj:jj:ji/: ,I.j::,:.:

(5)

where G,, is the stiffness of a normally consolidated sample at the same strain and the same
mean effective stress and c is a constant. Houlsby
& Wroth
(1991) expressed
the variation
of
stiffness with overconsolidation
ratio using a
power function of the type

AND

jijjjj j.: :jj::i:iji:

elements

G
-=

EQUIPMENT

The laboratory
tests were carried out on 38
in a computer-controlled
mm dia. samples
hydraulic triaxial cell of the type described by
Atkinson, Evans & Scott (1985). The apparatus is
shown in Fig. 3. The very small strain shear
modulus G, was measured using bender elements
of the type developed at the Norwegian Geotechnical Institute by Dyvik & Madshus (1985), and
the stiffness at larger strains was measured using
both an external displacement
transducer
and a
pair of Hall-effect local axial gauges (Clayton &
Khatrush,
1986). For tests on reconstituted
samples which were reconsolidated
in the hydraulic triaxial cell the measurements
of axial strain
were approximately
the same for both sets of
instruments,
provided allowance was made for
the compliance of the apparatus in the readings
obtained from the external displacement
transducer. Using either an external transducer
or

c 108Ro)

G/G, = 1 + c log R,

TEST

~~~

:::::
.....A
.:: :. .:.:.::.:
.,.,.
..:..:
...
::::. ::::: /
::::. ......A..
:.
.:.:.:
: :.::.,.:.
:;;i.:::+-::.
.::.
.. : L..
:

Lziijiiqei
Fig. 3. Laboratory test apparatus

STIFFNESS

OF FINE-GRAINED

local axial gauges the smallest reliable measurement of strain was about 0.002% and the first
reliable determination
of stiffness was at a strain
of about 0.004%.
Reconstituted
samples were prepared by onedimensional
consolidation
of a slurry in a tall
floating ring oedometer
until the samples were
sufficiently
strong to handle. They were then
transferred to the hydraulic triaxial cell and consolidated,
usually isotropically,
to the required
initial state. The samples were then loaded and
unloaded as required by the particular test and
the very small strain shear modulus G, measured
at various states during the test using the bender
elements. At the same time the overall strains in
the sample were measured using the external displacement
transducer
or the local axial gauges
and a conventional
Imperial College type volume
gauge. Additional tests were carried out on a 38
mm dia. undisturbed
sample of London clay prepared by hand-trimming
a 100 mm dia. thin wall
tube sample.
The test results were interpreted in terms of the
deviatoric stress parameter q = (oa - or) and the
mean stress parameter p = l/3(0, + 2~,), where
era and or are the axial and radial effective
stresses respectively.
The corresponding
strain
parameters
were the shear strain E, = 2/3(~, - E,)
and the volumetric strain a, = a, + 2~,, where E,
and E, are the axial and radial strains. The state of
soil is defined by the current values of q, p and
the specific volume u.
Bender elements were used to determine
the
shear modulus at very small strains G, by measuring the velocity of shear waves through the
sample. Bender elements are piezoelectric electromechanical transducers
that bend as an applied
voltage is changed
or, conversely,
mechanical
bending of the elements produces a change of
voltage.
A transmitter
element
and receiver
element are fixed into the top and bottom platens
of the triaxial cell so as to protrude about 3 mm
into the sample. A change of voltage applied to
the transmitter
causes it to bend and generate a
shear wave that propagates
through the sample.
The arrival of the shear wave is recorded as a
change of voltage by the receiver.
The electronics required to operate the bender
elements are shown in Fig. 3. A Farnell FGl
function generator was used to supply the transmitter with the driving voltage. This normally
consisted of a square wave with a frequency of 50
Hz and an amplitude of 10 V (20 V peak to peak).
The frequency of the square wave was always sufficiently low that the subsequent step of the wave
did not interfere with the received wave generated
by the previous step. The amplitude of the wave
was limited by the necessity of avoiding depolarization of the bender elements. The signal used to

SOIL AT VERY SMALL

drive
signal
tronix
The

STRAINS

253

the transmitter
element and the output
from the receiver were displayed on a Tek2211 50 MHz digital storage oscilloscope.
shear modulus G, was calculated from

G, = pV, = pL2/t2

(8)

where p is the density of the soil and V, is the


velocity of the shear wave which is determined
from the effective length L through which the
shear wave travels and the travel time t. The
values of p for a cylindrical sample can be determined with high precision, but there are uncertainties about the value for L (which could vary
between the overall length of the sample and the
distance between the tips of the bender elements)
and the precise instant of arrival of the shear
wave at the receiver.
These uncertainties
were examined in detail by
Viggiani (1992) and Viggiani & Atkinson (1995).
By means of direct calibration they found that the
effective length of travel should be taken as the
distance between the tips of the elements. They
also found that the arrival of the shear wave corresponds
closely to the first major reversal of
polarity of the received signal rather than to the
point of first deflection which probably
corresponds to the arrival of the near field components
of the shear wave (Salinero, Roesset & Stokoe,
1986), travelling at the velocity of compression
waves. Both findings are in good agreement with
previous experimental
observations
and numerical studies (Brignoli & Gotti, 1992; Dyvik &
Madshus,
1985; Mancuso,
Simonelli & Vigale,
1989).
For the tests described in this Paper the results
were always interpreted
by taking the effective
length as the distance between the tips of the elements and the arrival of the shear wave as the
first significant reversal of polarity of the received
signal. From the work described by Viggiani &
Atkinson (1995), the calculated values of G, could
have overestimated
the true values by up to 14%.
However, because the choice of the arrival time
was consistent
throughout
the analysis of the
data, this error does not apply to comparisons
between
shear moduli and affects only their
absolute
values when compared
with other
methods of measurement.
At strains sufhciently
large to be measured
using an axial displacement
transducer
or local
strain gauges the tangent shear modulus G was
calculated from the test results from the gradient
of the deviatoric
stress-shear
strain curve. In
comparing
the values of shear modulus determined from bender element tests and from direct
measurements
of deviatoric stress and strain on
the same sample it has been assumed that the
results are comparable,
despite differences that

254

VIGGIANI

may arise due to different modes


stress path rotations and rate effects.

VARIATION

AND ATKINSON

of shearing,

OF G, WITH ISOTROPIC STATE

Tests were carried out on reconstituted


samples
of speswhite kaolin clay, powdered
slate dust,
London clay and North Field clay. North Field
clay is a glacial till from a test bed site adjacent to
the Building Research Establishment
at Watford
(Abbiss, 1981). These tests examined the variation
of the very small strain shear modulus G, with
isotropic effective stress p and overconsolidation
ratio R, given by

R, = P~'IP

(9)

of a
where pP is the stress at the intersection
swelling line with the isotropic normal consolidation line (see Fig. 2). As soils may creep at constant mean effective stress, the value of pp may

2.5

I,

* Dynamic

p = 50kPa

100 kPa

200 kPa

I,,

reading

of Go

400 kPa

not correspond
to the maximum past stress p,.
Also, this definition of R, is different from the
usual definition
of overconsolidation
ratio in
which the current vertical effective stress is related
to the maximum
past vertical effective stress.
With R, defined as in equation (9), the current
specific volume is defined by the current values of
p and R, and so is not independent.
Each sample was isotropically compressed and
swelled following the path in Fig. 4(a). Typically
the samples were brought to states with p = 50400 kPa and R, = l-8. Due to the hysteresis in
the unloading-reloading
loops the values of R,,
defined by equation (9), at a particular stress on a
particular loop are different and so the values of
R, indicated in Fig. 4(a) are nominal values.
The samples were generally compressed
and
swelled
by continuous
drained
loading
and
unloading,
but in a few stages the load was
applied as a single increment
followed by isotropic consolidation.
The solid line in Fig. 4(a) is
the state path calculated on the assumption
that
any excess pore pressures could be neglected. The
small vertical sections at the end of some stages
represent additional small volume changes due to
dissipation
of small excess pore pressures developed during continuous loading or unloading or
to creep. The line for R, = 1 in Fig. 4(a) through
the equilibrium
states represents
the isotropic
normal compression line.
Figure 4(b) shows the values of Go measured at
the states indicated in Fig. 4(a). The data show
that the value of Go increases with mean effective
stress p, although the variation is non-linear, and
that the value of G, at a particular
stress
increases with overconsolidation
ratio. At a particular nominal value of R, and at a given stress,
values of G, may be slightly different. This is

I
105 -

Nominal

Nominal

values

values
d
0
0

Mean effective

stress p: kPa
(b)

Fig. 4. (a) Isotropic stress states at which bender


element tests were carried out; (b) variation of G, with
mean stress p and nominal overconsolidation ratio

104
.-lo

102

J
103

P!Pr

Fig. 5. Variation of G, with stress and overconsolidation


for reconstituted samples of speswhite kaolin

STIFFNESS

OF FINE-GRAINED

SOIL AT VERY SMALL

because, due to hysteresis


in the unloadingreloading loop in the An p plane, the specific
volumes and hence the actual values of R, are
different.
Figure 5 shows a typical set of values of G,
measured on samples of kaolin clay at the states
indicated
in Fig. 4, together
with additional
values obtained
at intermediate
states on the
normal compression
line. This shows values of
G,/p, against p/p,, both to a logarithmic
scale,
where p, is a reference pressure which, for convenience, has been taken as 1 kPa. The data points
for the normally consolidated
samples fall close
to a single straight line given by

,
1

255

STRAINS

,
3

,
4

,
5

, ,,(
6

769

Ro

Fig. 6. Variation of G,, with stress aad overconsolidation


for reconstituted samples of speswhite kaolin

or

(11)
where A and n are non-dimensional
soil parameters. For the test data for speswhite kaolin
shown
in Fig. 5 the values are A = 2088
and n = 0.653, with coefficient
of correlation
r2 = 0.996 and standard deviation a = 0.009. The
values of A and n obtained by fitting a straight
line through all the available data for normally
consolidated
speswhite kaolin were A = 1964 and
n = 0.653. The general form of equation
(ll),
which relates the shear stiffness of kaolin clay at
very small strain G, to the current stress, is essentially the same as equation (l), which was proposed by Wroth & Houlsby (1985) for the shear
modulus of sands. The values of A and II depend
on the value taken for the reference pressure.
InFig. 5 the data points for overconsolidated
samples fall above the line for normally consolidated samples and, for each value of R, , they fall
close to lines parallel to the line for normally consolidated samples. Considering the data shown in
Fig. 5, together with results of a number of addi-

%!=
P,

A
0

0p &

(12)

P,

where m can be regarded as another soil parameter. For the data shown in Fig. 6, m = 0.196
with coefficient
of correlation
r2 = 0.830 and
standard deviation e = 0.021. The increase of G,
with log R, given by equation (12) is similar to
that reported by Houlsby & Wroth (1991).

Plastiaty

Plasticity index
0 Speswhite kaolin
0 London clay

tional tests on reconstituted


speswhite kaolin, the
values of n in equation
(11) found by fitting
straight lines through all the available data for a
particular overconsolidation
ratio indicate that n
is independent
of R, , at least for R, less than 4.
Figure 6 shows the same data as those in Fig.
5, together with data from additional
tests on
reconstituted
kaolin clay. The data are plotted as
(GdGe,,) where GOnf is the value of G, corresponding to normally consolidated
samples at the
same mean effective stress. The data points fall
reasonably close to a straight line which, making
use of equation (1 l), is given by

Plasticity index

index

North Feld clay


Slate dust

Fig. 7. Variation of stitTness parameters for G, with plasticity index

l
l

Various clays (Weller, 1966)


Fucino clay (Pane & Butghignoli,

1966)

VIGGIANI AND ATKINSON

256
500,,

, , , , , , , , , , , , , , , , , , , , , , I1 = 0.75

043

0.30

0.00

4OO-

only the general trends of the data. Also shown in


Fig. 7 are points representing
tests on various
other fine-grained
soils reported
by Pane &
Burghignoli (1988) and Weiler (1988). They generally fall near the trends for the data obtained as
part of the present work. These data illustrate the
significant influence of plasticity index on small
strain stiffness through the parameter A, and the
relatively small importance
of overconsolidation
ratio through the parameter m.

VARIATION
STRESS
a',:kPa

Fig. 8. Anisotropic stress states at which bender element


tests were carried out on reconstituted samples of speswhite kaolin

Results obtained from tests on the other soils


all conformed to the general relationships
found
for speswhite kaolin clay shown in Fig. 6, but
with different values of the parameters
A, n and
m. Values for these parameters are shown in Fig.
7 plotted against plasticity index. The broken
lines in Fig. 7 are not best-fit lines and indicate

o
q

u'r= lOOkPa
a',= 2OOkPa
dr = 400kPa

OF G, WITH

ANISOTROPIC

STATE

For anisotropic
states for which ua # 6, the
value of G, might vary primarily with the mean
stress p as in equation (1 l), or primarily with the
stress in the direction of travel of the shear waves,
or with one of the stresses orthogonal
to the
direction of travel. Following Ni (1987), equation
(11) can be extended for anisotropic
states of
stress in a similar way to that in which equation
(2) was extended to equation (3). In a bender
element test the direction of wave propagation
is
always the direction of the axial stress and oe =
0,; the direction of particle motion is orthogonal
to the surface of the bender element and the
direction
normal to the plane of vibration
is

o
q

u',=
50kPa
o',=lOOkPa
0'8 = POOkPa

Fig. 9. Variation of G,, with axial and radial stress for normally consolidated
speswhite kaolin: (a) compression, a. > a,; (b) extension, u, < u,

STIFFNESS OF FINE-GRAINED

parallel to the bender element. In a triaxial


sample these are equal and err = CT*= Q,, where
ur is the radial stress.
As the influence of the stress orthogonal to the
plane of vibration has a negligible influence on G,
(because n, x 0 in equation (3)), equation (11) can
be written as

where B, and cr are the axial and radial stresses


in a triaxial sample, A, is equivalent to A in equation (11) and, in general, the parameters n, and n,
are not equal. The data shown in Figs 4-l were
obtained from tests on isotropically
compressed
samples for which p = oq = or and so A, = A
and n = n, + n,. To examme the variation of G,
with axial and radial stress for anisotropic states,
tests were carried out on reconstituted
samples of
kaolin clay consolidated
to the isotropic
and
anisotropic stress states following the stress paths
shown in Fig. 8, where the stress ratio 9 is q/p.
For the paths in Fig. 8 the value of the mean
effective stress was increased and so the samples
can be considered
to be normally consolidated
throughout all stages of each test.
The test results are shown in Figs 9 and 10. In
Fig. 9 the observed value of G, increases with
both axial stress CT=
and radial stress or. In Fig.
9(a), for compression,
the gradients of the lines
and hence the values of n, and n, are approximately equal, the exponent n, being only slightly
larger than the exponent n,. This indicates that,
in triaxial compression,
G, depends about equally
on the stress in the direction
of wave propagation and on the lateral stress, which is consistent with the observations
by Ni (1987). In Fig.
9(b), for extension, the gradients of the lines and
hence the values of n, and n, are significantly different from each other, the exponent n, being significantly larger than the exponent n, although n,
equal to n in both triaxial
+ n, is approximately
extension
and compression.
Ni (1987) did not
report the results of triaxial extension tests. In
Fig. 10 there is a unique relationship between G,
and p for the normally
consolidated
samples
which is independent
of the value of the stress
ratio g and is the same for compression
and
extension. The line in Fig. 10 is the same as that
obtained
for isotropically
compressed
samples
shown in Fig. 5. These data indicate that the
value of G, in reconstituted
normally consolidated kaolin clay depends primarily on the value
of mean stress p rather than on the value of the
stress either in the direction of travel of the shear
waves or orthogonal
to the direction of travel, at
least
for
stress
ratios
in
the
range
-0.60 < q < 0.75 and which are not close to
failure.

257

SOIL AT VERY SMALL STRAINS

106 _

Extension

Compression
0
0
A
*
,drlos

?J =
I =
?f =
rf =

0.00
0.43
0.75
0.30

.
.
.

I
I
I
Tf

=
0.00
= -0.36
= -0.60
- -0.27

102
PlPr

Fig. 10. Variation of G, with mean stress and stress


ratio for normally consolidated samples of speswhite
kaolin

VARIATION OF SHEAR MODULUS


TRIAXIAL COMPRESSION

DURING

A constant p drained triaxial compression


test
was carried out on a reconstituted
sample of
kaolin clay and values of G, were measured using
the bender elements at various points on the
stress path as indicated in Fig. 11(a). The corresponding
stress-strain
curve is shown in Fig.
11(b). The overall stress-strain
behaviour is represented by the smooth non-linear curve drawn
through the many data points; the stars indicate
bender element tests.
The test results are shown in Fig. 12. The triangles represent values of G, determined
from
shear wave velocities using the bender elements
plotted at values of overall strain measured using
the external axial displacement
transducer.
The
circles represent
values of the tangent
shear
modulus calculated
from deviator stresses and
shear strains. Values for the shear strains were
calculated from axial strains measured using an
external linear variable differential
transformer
(LVDT) and from volumetric
strains measured
using an external volume gauge. The samples
were reconstituted
and reconsolidated
in the
hydraulic triaxial cell, and so bedding and seating
errors were very small and the axial strains measured using the external LVDT were not significantly different from those measured using the
local axial gauges. The lower limit of strain for
which the values of tangent stiffness become unreliable was estimated
to be about 0404%. For
strains of up to 0.02% the tangent stiffnesses were
determined from a fifth-order polynomial fitted to
the data points. For strains greater than 0.02%
the tangent stiffnesses were calculated using the
methods described by Atkinson, Richardson
&
Woods (1986).

VIGGIANI AND ATKINSON

258

The data in Fig. 12 show that the value of G,


remains nearly constant, irrespective of the deviator stress or the shear strain.
The slight
reduction in G, by about 5% can be attributed to
small excess pore pressures in the sample which
would have the effect of reducing the current
value of mean effective stress p. The tangent
shear modulus decreases smoothly from a value
that approaches
the value of G,, corresponding
to very small strains, towards a value of G = 0 at
large strains. At very small strain, G, varies with
p as shown in equation (11). At strains greater
than about 0.1% the shear modulus G has been
found to vary linearly with p (Atkinson & Little,
1988). A question then arises as to the nature of
the relationship between stiffness and mean stress
in the intermediate range of small strain.
Figure 13 shows stiffness-strain
curves for a set
of constant p drained triaxial compression
tests
carried out on normally consolidated
samples of

l_l.
0

20

40

60
p: kPa

kaolin clay. The data points start at strains of


about 0X)04%, which was the limit of reliable
measurements
of stiffness. Each stiffness-strain
curve has the characteristic
shape,
with G
* decreasing with increasing strain. Also shown in
Fig. 13 are the values of G, obtained from bender
element measurements
in tests at different values
of p.
Figure 14 shows values of stiffness at a particular strain extracted
from the data in Fig. 13
plotted against mean stress, both normalized with
respect to the reference pressure pr as in Fig. 5.
The uppermost
line represents
results obtained
from the bender elements corresponding
to Go.
For each strain the variation
of stiffness with
mean stress can be represented
by equation (1 l),
with the values of A and n both varying with
increasing strain. Figure 15 shows the values of A
and n obtained from the data in Fig. 14 plotted
against strain on a logarithmic scale. The values
of A decrease by a factor of about three over the
strain range OXrO5%-0.05%, which is a consequence of the highly non-linear behaviour of soil
over this range of strain. In Fig. 15(b) the parameter n increases smoothly in the strain range
0.005-0.05%
from n = 0.12, which is just above
the value for Go obtained using the bender elements, towards
n = 1, corresponding
to large
strain when G is proportional
to p.

G, IN UNDISTURBED

100

80

120

(a)

LONDON

CLAY

A further test was carried out on an undisturbed sample of London clay. The sample was
trimmed from a 100 mm dia. tube sample taken
from a depth of about 6 m at a site at Chattenden
in Kent about 45 km east of central London

50
____~---*----b---d-*___
40
m
PS0
0:
Dynamic

reading

of Go
20

10

4
Shear

6
strain:

10

(b)

Fig. 11. Stress path and stressstrain curve for special


triaxial test with additional bender element tests

0
10-d

10-S

10-z
Shear

lo-
Wan:

10

Fig. 12. Variation of G and G, with strain for the special


triaxial test shown in Fig. 11

STIFFNESS OF FINE-GRAINED

soIL AT VERY

SMALL

259

STRAINS

p = 400 kPa
A D = 200 kPa

p = 200 kPa
Go = 62 MPa

0 p =

50 kPa

60
m
$
ij

p = 100 kPa
Go = 39 MPa

40

p = 50

kPa
Go = 25 MPa

Shear strain: %

Fig. 13. Variation of G with strain for a set of constant p trinxial compression teats oa reconstituted kaolin clay

(Abbiss & Viggiani, 1994). The sample was compressed and swelled in the hydraulic triaxial cell
to a number of different isotropic and anisotropic
states which encompassed
the range of states estimated for the sample in situ, and at each state the
value of G, was measured using bender elements.
Figure 16(a) shows the variation of Go for the
undisturbed
sample of London clay with p, both
normalized
with respect to a reference pressure
p, = 1 kPa as before. The undisturbed
sample
was
heavily
overconsolidated
in situ
and
remained overconsolidated
even at the maximum
stress achieved in the triaxial cell, so that its overconsolidation
ratio varied with reconsolidation
pressure as indicated in Fig. 16(a). The values for
the overconsolidation
ratio R, given in Fig. 16(a)
were calculated from equation (9) taking the iso-

2000

From bender elements


A = 1964

Shear strain: %
(a)

1.1 :
1.0 -

t t
0.6

At larger strain ~
rl=l.)

From bender elements


= 0.653

lo-

10-Z

10

Shear strain: %
PIP.

Fig. 14. Variation of G with mean stress and strain


extracted from the data in Fig. 13

(b)

Fig. 15. Variation of the parameters A aad II extracted


from the data in Fig. 14

260

VIGGIANI AND ATKINSON

tropic normal compression


line for reconstituted
samples. Also shown in Fig. 16(a) is the relationship between G, and p obtained from tests on
normally consolidated reconstituted
samples.
In Fig. 16(a) the data points for the undisturbed samples represent
states with different
overconsolidation
ratios because the maximum
stress applied in the triaxial cell was always considerably smaller than the estimated maximum
past stress in the ground. Consequently
the data
points for the undisturbed
sample fall above
those for the reconstituted
samples, and the bestfit lines through the two sets of data are not
parallel.
The data for the intact sample fall close to a
line given by

Go

-_=c(

PI

OS5
_a-_--*

*9-

Jl,r;

_Q_-9=-

104.

:
-

0 Ro= 62
a&=43
OR,=38
. R. = 21
AR,=18
AI&=13
OR,=12

103
10

_____
-m--

Undisturbed
Reconstituted

(R, = 1)

I
103

102
PIP,
6)

P'
P,

(14)
lo55
L

where a = 3000 and p = 0.50. From equations


and (12)

-I

(9)

@
so a = A(JQ/~,)~ and p = n - m. (Test results for
clay
gave
n = 0.76,
reconstituted
London
m = 0.25 and n - m = O-51, which is very close to
the value obtained from Fig. 16(a).)
Equation (12) can be rewritten as
Go

-=,I

P~Ro"

0
-P

P,

(16)

o Undisturbed
l Reconstituted

lol
10

Fig. 16(b) shows values of Go/p,Rom plotted


scale; the
against p/p,, both to a logarithmic
value of m was taken as 0.25 corresponding
to the
value obtained from the tests on reconstituted
samples of London clay already described. The
data points corresponding
to the results obtained
from the undisturbed
samples now fall very close
to the line obtained from the tests on the reconstituted samples. This result demonstrates
that,
for the London clay examined, the value of Go
depends on the current state (determined by both
the current stress and the overconsolidation
ratio)
and is unaffected by structure and fabric. This
apparently surprising result can be explained if, at
the very small strains involved in dynamic tests,
most of the deformation
is connected to elastic
deformations
and local slipping at the points of
contact
between soil particles
rather than to
major slippage and rearrangement
of particles.

VARIATION OF G, WITH
OVERCONSOLIDATION

The general variation of shear modulus


small strain G, with overconsolidation

910:

at very
can be

102

103

PlP,
(W

Fig. 16. Variation of GO with mean stress and overconsolidation ratio for undisturbed and reconstituted
samples of London clay

represented
by equation
(12); for the shear
modulus
at larger strains
the corresponding
relationship could be either equation (4) or equation (6), which is similar to equation
(12).
However, for the data shown in Fig. 6 from which
equation
(12) was
developed,
the
overconsolidation
ratios were limited to about 7 due
to the maximum
pressures
available
in the
hydraulic triaxial cell; for tests on intact samples
the overconsolidation
ratios were larger since the
maximum stresses in the ground were generally
greater than those available in the triaxial cell.
Figure 17(a) shows values of G,/G,,,
obtained
from bender element tests on undisturbed
and
reconstituted
samples of London
clay plotted
against R,, both to a logarithmic
scale corresponding to equation (6). The data points obtained
from the tests on the undisturbed sample fall very
close to the best-fit straight line through
the
results obtained from the tests on reconstituted

STlFFNESS

OF FINE-GRAINED
-F

A Undisturbed
q Reconstituted

Best-fit line from


reconstiiuted
samples

SOIL AT VERY SMALL

STRAINS

261

and log R, in Fig. 17(a) agrees with the relationship proposed by Houlsby & Wroth (1991) and
test results given by Hardin & Drnevich (1972).
Figure 17(b) shows the same data as those in
Fig. 17(a), but with G,/G,,, plotted to a natural
scale corresponding to equation (4). In this case
the data points for the undisturbed sample at
values of R, greater than about 10 depart from
the best-fit line through the data for reconstituted
samples. On the basis of these results G, can be
related to stress and overconsolidation
by an
expression of the form of equation (6) or equation
(12) over a wide range of overconsolidation
ratios.

100

A Undisturbed
q Reconstituted

est-fit line from


reconstituted
samples

0.5P

100

10

Fig. 17. Variation of G, with overcoasolidatioo ratio for


undiiturbed and reconstituted samples of London clay

samples. These results again indicate that the


stiffness of soil at very small strain is basically
unaffected by whether the sample is undisturbed
or reconstituted, provided that undisturbed and
reconstituted samples are brought to the same
mean effective stress and overconsolidation ratio.
The unique relationship between log (Go/Go,,)

MEASUREMENTS OF G, IN SITU AND


IN LABORATORY TESTS
Measurements of Go in situ can be made by
observing the velocities of either shear waves or
Rayleigh waves in the ground; a number of different techniques are available (Yoshimi et al., 1977;
Atkinson & Sallfors, 1991). Measurements were
made using Rayleigh waves at the site on London
clay at Chattenden from which the undisturbed
sample was obtained. Rayleigh waves, also
known as surface waves, travel largely in a layer
that is about one wavelength deep, and their
velocity varies with the shear stiffness of the
ground. The Rayleigh wave velocity V, is related
to the shear wave velocity V, through Poissons
ratio v, but the relationship is not sensitive to the
value of v. For most practical purposes it is sufficient to take V, as 0.9V, and the error will be
limited to less than 5% for a very wide range of
values of v (Richart et al., 1970).
In the most common form of the Rayleigh
wave method (Richart et al., 1970) continuous
seismic waves generated at the surface by a vibrator are detected by two geophone receivers placed
at the surface at a distance from the vibrator and
at a known distance apart as in Fig. 18. For a
half space in which the stiffness increases with
depth the velocity of the Rayleigh waves between
the two receivers varies with the wavelength, and
hence with the frequency, of the excitation

Vibrator

Oscillator
00

and amphfier
-

Fig. 18 Apparatus for in situ measurements of shear modulus using Rayleigh waves

VIGGIANI

262

AND ATKINSON

because different
wavelengths
sample material
with different stiffnesses.
The travel time between the two receivers is
given by
T(f)

?!a L

(17)

2% f

where f is the excitation frequency and @ is the


phase shift between the two received signals. The
Rayleigh wave velocity and wavelength
corresponding to the frequencyfare
given by

v,(f)
= wxf)

(18)

b(f) = UfM-

(19)

where d is the distance between the receivers. The


Rayleigh
wave velocity
determined
by this
method may be thought of as representative
of
the properties
of the ground in a characteristic
layer. Following
Vettros
(1990) and Gazetas
(1982), this was taken as one third of the wavelength determined from equation (19). By repeating the experiment
with different
excitation
frequencies, a profile of Rayleigh wave velocity
with depth can be obtained.
Hence profiles of
shear wave velocity and shear modulus G, are
obtained.
Details of the equipment
and experimental
methods used for the Rayleigh wave tests are

yGo

Field observations
0 Summer
D Summer

1989
1990

.-- Based on
laboratory results

given by Abbiss & Viggiani (1994). These are


similar to those described by Nazarian & Stokoe
(1986a, 1986b), except that they generated Rayleigh waves as single pulses rather than as continuous vibrations at fixed frequencies. The general
configuration
of the equipment
used in the
present work is shown in Fig. 18. Rayleigh waves
were generated using a Ling Dynamics 400 electromagnetic
vibrator attached to a plate, about
0.2 m in diameter, resting on the ground and tests
were carried out at frequencies in the range 8-200
Hz. The receivers were Sensor SM6 vertically polarized geophones. The signals were recorded on a
Hewlett Packard dual channel spectrum analyser
and the phase shift of the main Fourier components was determined.
Figure 19 shows the variation of shear modulus
with depth obtained from Rayleigh wave measurements at the site on London clay at Chattenden. The values of Go measured
in situ are
generally in the range lo-30 MPa, increasing
slightly with depth and with a few higher values
recorded in the top metre in the crust. (The difference in the values measured one year apart has
not yet been explained satisfactorily.)
Values for G, in the ground can be calculated
from the results of laboratory tests together with
estimates of the current state of stress and overconsolidation
in the ground. The relationship
ratio is
between G, , stress and overconsolidation
given by equation (12). Values for the parameters
A, n and m were determined from bender element
tests on both undisturbed
and reconstituted
samples of London clay from the site at Chattenden; from these tests A = 400, n = 0.76 and
m = 0.25. The in situ stresses were calculated
from the measured unit weights of undisturbed
samples and from the position of the water table,
with values of K, varying with overconsolidation
ratio as proposed by Wroth (1975). The reduction
in vertical effective stress due to erosion of 1500
kPa was estimated using data given by Skempton
(1961) for a site where the geology resembles that
at Chattenden.
The values of shear modulus G,
calculated from the laboratory tests are shown in
Fig. 19; they fall within the scatter of the results
obtained from the in situ measurements.

SUMMARY

Fig. 19. Variation of G, with depth in the ground in


London clay at Chattemien

AND CONCLUSIONS

The work investigated


the variation
of the
shear modulus of fine-grained soils with state (i.e.
current stress and overconsolidation
ratio) and
strain. From the results of laboratory
tests on
reconstituted
samples of a number of different
fine-grained
soils using bender elements, it was
found that the shear modulus
at very small
strains G, could conveniently
be related to the
current state through an expression of the form of

STIFFNESS

OF FINE-GRAINED

equation (12), where p, is a reference pressure


included so that the parameters
A, n and m are
dimensionless,
although
their numerical
values
depend on the choice of the reference pressure.
Values for the parameters
A and n depend on
the plasticity index of the soil. The value of n is in
the region of 0.6-03, depending on the plasticity
index as shown in Fig. 7(b). The value of A
decreases with plasticity index as shown in Fig.
7(a), and this shows the very significant influence
of plasticity
on soil stiffness. The relationship
between G, and current state given by equation
(12) was found to hold for both isotropic and
anisotropic stresses that were not close to failure.
A limited investigation
of undisturbed
London
clay showed that the stiffness at very small strain
G, was the same as for reconstituted
samples at
the same state. Furthermore,
values of G, measured in situ using Rayleigh waves compared well
with the values obtained in the laboratory
tests
normalized
to the in situ stress and overconsolidation
ratio.
At strains larger than those corresponding
to
the region of very small strain in Fig. 1, the
dependence of shear stiffness on state can still be
expressed using equation (12), but with values of
A, n and m that depend addithe parameters
tionally on the magnitude of the strains. In particular, for a given soil, the value of A was found
to decrease with increasing strain as shown in
Fig. 15(a); this is a consequence
of the highly
non-linear
stress-strain
behaviour
of soil. The
value of the exponent n gradually increases with
strain as shown in Fig. 15(b) and, at large strains,
n x 1.0. The change of n can be attributed
partly
to a change from essentially elastic behaviour at
very small strains to essentially plastic behaviour
at large strains.
The value of shear modulus at very small strain
G, is a useful parameter
for characterizing
the
highly non-linear
stress-strain
behaviour of soil
for monotonic
loading. It is also required for
analyses
of dynamic
and small strain cyclic
loading of soils. The present work has shown that
the variation
of G, with stress and overconsolidation
is similar to that for the shear
modulus at larger strains. Values for G, can be
obtained relatively easily from measurements
of
shear wave velocities in field or in laboratory tests
using a number of techniques. In laboratory tests
shear wave velocities can be measured
using
bender elements which can be incorporated
into
conventional soil testing apparatus.

ACKNOWLEDGEMENTS
Dr Viggiani was supported by an SERC Case
award
in collaboration
with
the
Building
Research Establishment.
The field tests at Chat-

SOIL AT VERY SMALL

tenden were carried


C. P. Abbiss.

STRAINS

out in collaboration

263
with Dr

NOTATION
A

nondimensional factor relating G,


to pl
d distance between receivers in Rayleigh wave test
f frequency
G shear modulus
shear modulus for normally conG
solidated soil
GO shear modulus at very small strain
of overconsolidation
k exponent
ratio
L effective length of travel of shear
waves in bender element tests
LR wavelength of Rayleigh wave
m exponent for R,
narne, n,, ns>4 exponents relating G, to p
OCR overconsolidation ratio defined as
l&/a
PI mean effective stress, 1/3((r, + 20,)
maximum past mean effective
P,
stress
PP effective stress at the intersection
of a swelling line with the normal
compression line
reference pressure (taken as 1 kPa)
P,
deviator stress, ur - 6,
overconsolidation ratio defined as
E

n,

PplP

S non-dimensional

v,
v,
a

factor in equation (2)


time shift between received signals
in Rayleigh wave test
travel time of shear waves in
bender element tests
specific volume
Rayleigh wave velocity
shear wave velocity
non-dimensional factor relating G,
to p for intact samples
shear strain, 2/3(s, - E,)
volumetric strain, 6. + 2.5,
exponents for mean pressure for
intact samples
soil density
effective stress in the direction of
wave propagation
effective stress in the direction of
particle motion
effective stress out of the plane of
wave motion
phase shift in Rayleigh wave tests

REFERENCES
Abbiss, C. P. (1981). Shear wave measurements of the
elasticity of the ground. Gkotechnique 31, No. 1, 94104.
Abbiss, C. P. & Viggiani, G. (1994). Surface wave and
damping measurements of the ground with a correl-

264

VIGGIANI

AND ATKINSON

ator. Proc. 13th Int. Conf: Soil Mech., New Delhi,


1329-1332.
Anderson,
D. G. & Richart, F. E. (1976). Effects of
straining on the shear modulus of clays. J. Geotech.
Engng Div. Am. Sot. Civ. Engrs 102, GT9,975-987.
Atkinson,
J. H., Evans, J. S. & Scott, C. R. (1985).
Developments
in stress path testing equipment.
GroundEngng 18, No. 1, 15-22.
_ _
Atkinson. J. H. & Little. J. A. 11988). Undrained triaxial
strength and stress-&rain
characteristics
of a glacial
till soil. Can. Geotech. J. 25, No. 3,428-439.
Atkinson, J. H., Richardson,
D. & Woods, R. I. (1986).
Determination
of tangent stiffness parameters
from
soil test data. Int. J. Comput. Geotech. 2, No. 3, 131140.
Atkinson,
J. H. & Sallfors, G. (1991). Experimental
determination
of soil properties.
Proc. 10th Eur.
Conf: Soil Mech., Florence 3,915-956.
Bolton, M. D. & Wilson, J. M. R. (1989). An experimental and theoretical
comparison
between static and
dynamic torsional soil tests, Gtotechnique 39, No. 4,
585-599.
Brignoli, E. & Gotti, M. (1992). Misure della velocit$ di
onde elastiche di taglio in laboratorio
con limpiego
di trasduttori
piezoelettrici. Riu. Ital. Geotec. 26, No.
1, 5-16.
Burland, J. B. (1989). Small is beautiful: the stiffness of
soils at small strains. Can. Geotech. J. 26,499-516.
Clayton, C. R. I. & Khatrush, S. A. (1986). A new device
for measuring
local axial strains on triaxial specimens. Gtotechnique 36, No. 4,593-597.
Dobry, R., Ng, T.-T. & Petrakis, E. (1989). Deformation
characteristics
of granular soils in the light of particulate mechanics. Atti delle conferenze di geotecniche
di Torino, vol. 3, pp. l-34. Torino: Politecnico
di
Torino.
Dyvik, R. & Madshus, C. (1985). Laboratory
measurements of G,,, using bender elements. Proc. Am. Sot.
Civ. Engrs Cono., Detroit, pp. 186-196.
Gazetas, G. (1982). Vibrational
characteristics
of soil
deposits with variable wave velocity. Int. J. Numer.
Analyt. Meth. Geomech., 6, l-20.
Georgiannou,
V. N., Rampello, S. & Silvestri, F. (1991).
Static and dynamic
measurement
of undrained
stiffness of natural
overconsolidated
clays. Proc.
10th Eur. Conj Soil Mech., Florence 1,91-96.
Hardin. B. 0. (1978). The nature of stress-strain
behaviou; for soils. State-of-the-art
report. Proceedings of
specialty conference on earthquake engineering and
soil dynamics, pp. 3-90. New York: American
Society of Civil Engineers.
Hardin, B. 0. & Black, W. L. (1966). Sand stiffness
under various triaxial stresses. J. Soil Mech. Fdns
Div. Am. Sot. Civ. Engrs 92, SM2, 27-42.
Hardin, B. 0. & Black, W. L. (1968). Vibration modulus
of normally consolidated
clay. J. Soil Mech. Fdns
Div. Am. Sot. Ciu. Engrs 95, SM6, 1531-1537.
Hardin, B. 0. & Drnevich, V. P. (1972). Shear modulus
and damping
in soils. I. Measurement
and parameter effects. J. Geotech. Engng Div. Am. Sot. Civ.
Engrs 102, GT9,975-987.
Houlsby, G. T. & Wroth, C. P. (1991). The variation of
shear modulus of a clay with pressure and overconsolidation
ratio. Soils Fdns 31, No. 3, 138-143.
Isenhower,
W. M. (1979). Torsional simple shear/
resonant column properties of San Francisco Bay

mud. MSc thesis, University of Texas at Austin.


Iwasaki, T., Tatsuoka,
F. & Takagi, Y. (1978). Shear
moduli of sands under cyclic torsional shear loading.
Soils Fdns 18, No. 1, 39-56.
Janbu, N. (1963). Soil compressibility
as determined by
oedometer
and triaxial tests. Proc. 3rd Eur. Conf.
Soil Mech., Wiesbaden 1, pp. 19-24.
Jardine, R. J. % Potts, D. M. (1988). Hutton tension
platform foundation:
an approach to the prediction
of pile behaviour. Gdotechnique 38, No. 2,231-252.
Jardine, R. J., Potts, D. M., St. John, H. D. & Hight,
D. W. (1991). Some practical applications
of a nonlinear ground
model. Proc. 10th Eur. Conf: Soil
Mech., Florence 1,223-228.
Jardine, R. J., Symes, M. J. R. P. & Burland, J. B. (1984).
The measurement
of soil stiffness in the triaxial
apparatus.
Gtotechnique 34, No. 3,323-340.
Lo Presti, D. C. F. (1989). Proprieta
dinamiche
dei
terreni. Atti delle conferenze di geotecnica di Torino,
vol. 2, pp. l-62.
Mancuso,
C., Simonelli, A. L. & Vinale, F. (1989).
Numerical analysis of in situ S-wave measurements.
Proc. Int. 12th Co& Soil Mech.. Rio de Janeiro. 277280.
Nazarian,
S. & Stokoe, K. H. (1986a). In situ determination of elastic moduli of pavement systems by spectral analysis of surface wave method. Practical
aspecfs. Research report 368-IF. Centre for Transportation Research, University of Texas at Austin.
Nazarian, S. & Stokoe, K. H. (1986b). In situ determination of elastic moduli of pavement systems by spectral analysis of surface wave method. Theoretical
aspects. Research report 437-2. Centre for Transportation Research, University of Texas at Austin.
Ni, S-H. (1987) Dynamic properties of sand under true
triaxial stress states from resonant column/torsion
shear tests. PhD thesis, University
of Texas at
Austin.
Pane, V. & Burghignoli,
A. (1988). Determinazione
in
laboratorio
delle
caratteristiche
dinamiche
dellargilla de1 Fucino. In Atti de1 I Conuegno de1
Gruppo Nazionale di Coordinamento per gli Studi di
Ingegneria Geotecnica, Monselice, pp. 115-140.
Papa, V., Silvestri, F. & Vinale, F. (1988). Analisi delle
proprieta di un tipico terreno piroclastico
mediante
prove di taglio semplice. In Atti de1 I Convegno de1
Gruppo Nazionale di Coordinamento per gli Studi di
lngegneria Geotecnica, Monselice, pp. 265-286.
Rampello,
S. & Pane, V. (1988). Deformabilita
non
drenata statica e dinamica di unargilla fortemente
sovraconsolidata.
Atti de1 Convegno de1 Gruppo
Nazionale di Coordinamento per gli Studi di Ingegneria Geotecnica, Monselice, pp. 141-160.
Richart, F. E., Hall, J. R. & Woods, R. D. (1970). Vibration of soils and foundations. Englewood
Cliffs:
Prentice-Hall.
Rowe, P. W. (1971). Theoretical meaning and observed
values of deformation
parameters
for soils. In
Stress-strain behaoiour of soils, pp. 143-194. Cambridge: Foulis.
Salinero, I. S., Roesset, J. M. & Stokoe, K. H. (1986).
Analytical studies of body wave propagation and
attenuation. Report GR86-15. University of Texas at
Austin.
Shirley, D. J. & Hampton,
L. D. (1977). Shear-wave
measurements
in laboratory
sediments. J. Acoust.

STIFFNESS

OF FINE-GRAINED

Sot. Am. 63, No. 2, Feb., 607-613.


Silvestri, F. (1991). Stress-strain
behaviour
of natural
soils by means of cyclic/dynamic
torsional
shear
tests. In Experimental characterization and modelling
of soils and soft rocks, pp. 7-73. University of Napoli
Federico Il.
Simpson, B., Calabresi, G., Sommer, H. & Wallays, M.
(1979). Design parameters
for stiff clays. Proc. 7th
Eur. Co@ Soil Mech., Brighton 1,91-125.
Skempton, A. W. (1961). Horizontal
stresses in an overconsolidated
Eocene clay. Proc. 5th Int. Conf Soil
Mech., Paris 1, 351-357.
Stallebrass, S. E. (1990). Modelling the effect of recent
stress history on the deformation of overconsolidated
soils. PhD thesis, City University, London.
Vettros, C. H. (1990). In-plane vibrations of soil deposits
with variable shear modulus: I surface waves. Int. J.
Numer. Analyt. Meth. Geomech. 14, 209-222.
Viggiani, G. (1992). Small strain stiffness of fine grained
soils. PhD thesis, City University, London.
Viggiani, G. & Atkinson, J. H. (1995). Interpretation
of
bender element tests. Gtotechnique 45, No. 1, 149154.
Webb, D. L. (1967). The mechanical properties of undisturbed samples of London clay and Pierre shale. PhD
thesis, University of London.

SOIL AT VERY SMALL STRAINS

265

Weiler, W. A. (1988). Small strain shear modulus of clay.


Proceedings of Geotechnical Engineering Division
specialty conference on earthquake engineering and
soil dynamics, vol. 2, pp. 331-345. New York: American Society of Civil Engineers.
Wroth, C. P. (1971). Some aspects of the elastic behaviour of overconsolidated
clay. In Stress-strain behaoiour ofsoils, pp. 347-361. Cambridge:
Foulis.
Wroth, C. P. (1975). In situ measurement
of initial
stresses and deformation
characteristics.
Proceedings
of Geotechnical Engineering Division specialty conference on in situ measurement of soil properties, vol.
2, pp. 181-230. New York: American
Society of
Civil Engineers.
Wroth, C. P. & Houlsby, G. T. (1985). Soil mechanicsproperty characterisation,
and analysis procedures.
Proc. 11th Conf: Soil Mech.. San Francisco 1, l-55.
Wroth, C. P., Rgndolph,
MI F., Houlsby,
G. T. &
Fahey, M. (1979). A review of the engineering properties of soils with particular reference to the shear
modulus. OUEL
Report
1523/84, University
of
Oxford.
Yoshimi, Y., Richart, F. E., Jr, Prakash,
S., Barkan,
D. D. & llyichev, V. A. (1977). Soil dynamics and
its application
to foundation
engineering. Proc. 9th
Int. Conf: Soil Mech., Tokyo 2, 605-650.

Вам также может понравиться