Vous êtes sur la page 1sur 27

Undrained strength properties of clays

derived from pressuremeter tests


Ramdane Bahar* Fawzia Baidi* Ouarda Belhassani*
Eric Vincens**
* Laboratoire de Gomatriaux, Environnement et Amnagement, Universit
Mouloud Mammeri de Tizi-Ouzou, BP RP 17, 15000 Tizi-Ouzou, Algrie
ramdane_bahar@mail.ummto.dz
** Universit de Lyon, Laboratoire de Tribologie et Dynamique des Systmes, Ecole
Centrale de Lyon, 36 avenue Guy de Collongue, 69134 Ecully Cedex, France
eric.vincens@ec-lyon.fr
ABSTRACT.

The behaviour of saturated clays under undrained conditions around


pressuremeter path is studied by means of a theoretical and a numerical analysis based on
the generalised elastoplastic Pragers model with the Von Mises Criterion which is suitable
for these cases of soils and gives a realistic response when unloading is considered. The soil
constitutive model is described depending only on three parameters. The model response on
pressuremeter path and its identification from experimental data lead to the determination of
the undrained cohesion and the initial Young modulus of clays. Comparisons between the
undrained cohesions determined with the proposed method and those obtained by other
means illustrate the applicability of the method. An application to predict the bearing
capacity and the settlement of bored piles using parameters derived from the proposed
approach is also presented with a comparison to the measured data.
RSUM.

Le comportement des argiles satures en conditions non draines sur un chemin


pressiomtrique est tudi au moyen d'une analyse thorique et numrique base sur le
modle lastoplastique du Prager gnralis avec prise compte du critre Von Mises. Ce
critre est bien adapt aux sols considrs et donne une rponse raliste en dchargement.
Le modle constitutif de sol propos est dfini seulement par trois paramtres. Cette
approche permet la dtermination de la cohsion non draine et le module de Young initial
des argiles satures partir dessais pressiomtriques. La comparaison entre les cohsions
non draines dtermines avec la mthode propose et celles obtenues par d'autres moyens
illustrent son applicabilit. Une application de lapproche propose pour prvoir la capacit
portante et le tassement dun pieu for soumis une charge axiale centre est galement
prsente avec une comparaison avec les mesures exprimentales.
KEYWORDS:

undrained cohesion, clay, pressuremeter, behaviour, bored pile.

MOTS-CLS :

cohsion non draine, argile, pressiomtre, comportement, pieu for.

Revue. Volume X n x/anne, pages 1 X

Undrained strength properties of clays

1. Introduction
In the recent year, the use of the pressuremeter as an in situ testing device by
practising geotechnical engineers has increasingly become popular in Algeria and
others countries. Indeed, it provides the measurement of in situ stress-strain response
of soils, and it is an useful and economical way for obtaining reliable in situ
properties of soils. From this test, the design of foundations can be performed by
using the limit pressure and the pressuremeter modulus. They are derived from the
pressuremeter curve or deduced from existing correlations with undrained cohesion
and internal angle friction (Mnard, 1957; Amar et al., 1972). In particular, these
parameters are used to evaluate the bearing capacity of soil foundations and the
expected settlements. But they can also help to identify usual soil parameters
required by simple constitutive models (eg. Mohr-Coulomb model) for soils in
numerical calculations.
For saturated clays with low permeability, several empirical, analytical or
numerical methods based on pressuremeter tests have been proposed to evaluate the
undrained shear strength and the stress-strain behaviour. The empirical methods give
relationships between soil properties and pressuremeter parameters (Mnard, 1957;
Amar et al., 1972). The analytical ones are based on the cavity expansion theory
assuming idealised behaviour of soils, axial symmetry and plane strain, soil
homogeneity, isotropy and undrained conditions (Baguelin et al., 1972; Ladanyi,
1972; Palmer, 1972; Gibson et al., 1961; Prvost et al., 1975; Denby, 1978; Ferreira
et al., 1992; Monnet et al., 1994). Numerical solutions are used if a more precise
solution of the pressuremeter test involving complex constitutive model for soils is
required (Boubanga, 1990; Bahar, 1992; Zentar et al., 2001, Monnet, 2007).
Boubanga (1990) and Bahar (1992) have developed a methodology called
Pressident, Pressuremeter Identification to analyse pressuremeter tests using a
very simple axisymmetric plane finite element method independent of the used
constitutive model for soil. This program allows identifying the model parameters,
taken into account the whole pressuremeter curve. It has successfully been used to
define the soil parameters using non-viscous and viscous models for soil (Cambou et
al., 1993; Bahar et al., 1995; Bahar, 1998; Bahar et al., 2005).
In this paper, the behaviour of saturated clays under undrained conditions
(constant volumetric deformations) is studied by means of both a theoretical and a
numerical analysis based on the generalised elastoplastic Pragers model with the
Von Mises criterion which is suitable for these kind of soils and gives a realistic
response when unloading is considered along with Bauschinger effect (Iwan, 1967).
In the first part, the soil constitutive model is described. Based on the analytical
representation of the stress-strain curves obtained from triaxial tests proposed by
Olivari (Olivari et al., 1995) and modified by Bahar (Bahar et al., 1999), it is shown
how the parameters of the generalised Prager model, composed of a large number of
elastoplastic slip elements associated in series, can be identified. Also, by
introducing the assumptions of incompressibility for the material and plane strain

Undrained strength properties of clays

condition for the calculation, it is shown that the proposed model depends only on
three parameters, and the response of the pressuremeter test with a cycle of
unloading-reloading is found to be realistic.
In the second part of the paper, the numerical program developed to simulate the
pressuremeter test and taking into account the proposed model is described. This
program allows the soil constitutive model parameters to be defined using
pressuremeter tests. A parametric study is carried on to define the model constants
that could reasonably be identified from pressuremeter tests. Then a strategy for the
identification of the three parameters is presented. The model response on the
pressuremeter path and its comparison with experimental data, lead to the
determination of the undrained cohesion and the elastic Young modulus of clays.
The validity of this method, which helps to obtain in situ mechanical properties, is
compared to usual methods able to provide these properties. The comparison
between the undrained cohesion determined with the proposed method and with
other means illustrates the applicability of the previously described procedure.
In the third part, an application to predict the bearing capacity and settlement of
a bored pile using the mechanical properties derived from this approach is presented
and compared with the results obtained with the Pressident approach and with the
experimental data.
2. Elasto-plastic model with multiple yield surfaces
Many plasticity models have been able to predict the nonlinear stressstrain
behaviour of soils. Based on the use of a kinematic hardening function, a bilinear
stress-strain behaviour can be simply described by means of a single internal
variable (Figure 1(a)). Combining the idea of kinematic hardening and multiple
internal variables together, a piecewise linear stressstrain response can be described
(Figure 1(b)), the stress-strain curve is approximated by linear segments along
which the tangent shear modulus is assumed to be constant. Each linear segment is
associated with a yield surface in the stress space. Iwan (1967) proposed a model
based on the assumption that a general hysteretic system can be constructed from a
large number of ideal elasto-plastic Prager elements having different yield levels. A
purely kinematic hardening rule is adopted for the movement of the yield surfaces in
the stress space when the plasticity is activated, say, the sizes of the yield surfaces
are assumed to remain constant while they translate in the stress space. Initially, all
the yield surfaces are symmetrically arranged with respect to the origin. In this
study, a collection of (n) elasto-plastic Prager elements connected in series is
considered (Figure 2). The model is defined by the compliance of (n) elastic
elements and their associated (n) yield surfaces. Eventually, a single linear spring
elastic element of compliance (Jo) and a single yield surface of threshold stress (S)
can be introduced and connected in series to the collection of elements to represent
the initial elastic strain and failure respectively. Thus the model is defined by (2n+2)
parameters that can be represented by a discrete spectrum of compliance (Figure 3).

Undrained strength properties of clays

(a)

(b)

Figure 1. Stress-strain relationships: (a) bilinear; (b) piecewise linear.

J1

Jn

J2

Jo

S1

Sn

S2

p
1

=e +

np

p
2

+ p

i =1 i

Figure 2. Multiple surface model.

~
1
Jk =
Ck

Jo

Sk
Threshold stresses

Figure 3. Compliance spectrum of the model.

Undrained strength properties of clays

During the initial loading, before the stress reaches the value of the first slip
stress k1, the behaviour is linear elastic and is governed by the elongation of the Espring. After the stress reaches the value of the slip stress k1, the first sliding element
slip and the J1-spring becomes active. The corresponding behaviour is elasto-plastic
with a linear hardening characterized by the tangent modulus E1. After the stress
reaches the value of the slip stress kn*, the n*th sliding slips and the Jn*-spring
become active. The corresponding behaviour is elasto-plastic with a linear hardening
characterized by the tangent modulus En*.
The constitutive equations are:
The elastic component of strain (ij) is calculated according to Hookes law:

ije =

1 +
v
ij kk ij
E
E

[1]

where E and denote the Young modulus and the Poisson ratio respectively.
The equation of the kth yield surface fk in the stress space is given by:

f k s ij X ijk = S k2

[2]

where (k) is a superscript denoting the kth element of the collection, sij denotes
the components of the deviatoric stress tensor, Sk is the threshold stress of the kth
yield surface, and X ijk is the tensor specifying the position of the center of the kth
yield surface in the stress space.
The total plastic strain is the sum of the plastic strain developed by each element.
The number of activated elements is (n*) and, of course, 0 n* n. Moreover, the
components of the tensors Xij that specify the state of the model must be initially
given. In the virgin state, all the residual stresses are equal to zero, hence for each
cell, the hardening variables are also equal to zero, Xij(k) = 0. Since each of the (n)
Prager elements will individually obey a linear work-hardening law, their combined
action leads to a piecewise linear behaviour with kinematics hardening for the
material as a whole.
The advantage of the multiple surface models is clearly that they are able to fit
more accurately the non linear behaviour of certain materials across a wide range of
strain amplitudes. This is important, for instance, for the modelling of geotechnical
materials. The drawback is of course that a large number of material parameters,
associated with each yield surface must be identified.

Undrained strength properties of clays

3. Modelling of the undrained behaviour of clays


3.1. Assumptions

The framework of the short-term behaviour of an incompressible medium is


adopted. The considered stresses are thus total stresses. Viscous effects are also
supposed to be negligible which justifies the use of an elastoplastic model as a first
approximation of the true behaviour of the material. The material is supposed to be
isotropic and normally consolidated or with a low value of the overconsolidation
ratio.
3.2. Yield criterion

The used criterion obeys the condition of plastic incompressibility. Referring to


the results of Habib (1953), obtained on Provinss clay, it is admitted that the failure
happens for the same value of the deviatoric stress in compression and in tension.
Therefore the undrained cohesion is sufficient to characterise the plastic flow
criterion. Consequently, the Von-Mises criterion is used:

)(

3
s ij X ijk s ij X ijk = k k
2

[3]

where k = 2cu, cu is the undrained cohesion.


3.3. Spectrum compliance

The response of this proposed model on a triaxial path is a polygonal line that
can be considered as a discretisation of the experimental curve. Then, if an
analytical representation of the experimental test results for a triaxial stress path is
chosen, the parameters of the Prager model can be easily identified. The following
relationship is used (Olivari et al., 1995; Bahar et al., 1999):

dp = A Ln(1 R ) + (1 2 R )

R=

1 3
( 1 3 ) f

1 R

[4]

[5]

1 and 3 are the principal total stresses, (1 - 3)f is the asymptotic value for the
difference between the major and the minor principal stresses that is related closely
to the strength of the soil, and A is a positive parameter defining the curvature of the
curve as shown in Figure 4.

Undrained strength properties of clays

R =(1 - 3) /(1 - 3)f

0.9
0.8
0.7
0.6

A = 2.00
A = 1.00

0.5

A = 0.50

0.4

A = 0.10

0.3

A = 0.05

0.2
A = 0.01

0.1

A = 0.001

0
0

Deviatoric plastic strain d (%)


p

10

11

12

Figure 4. Theoretical curve for different values of parameter A.

As a consequence to this expression, the non-dimensional yields of the different


~
links are fractions of the units of the chosen discretisation. Then, the compliance J k
can easily be obtained from the successive secant modulus of the experimental
curve. The compliances intervening in the constitutive law are written as:

J0 =

Jk =

1
E

[6]
~
2 Jk
3 2 k cu

with 0 <k <1

[7]

Consequently, the (2n+2) parameters of the proposed model are totally defined
with only three parameters: the initial Young modulus E, the curvature parameter A,
and the undrained cohesion cu. Poisson ratio is equal to 0.5.
4. Model response on a pressuremeter path

The expansion of a pressuremeter probe in a homogeneous soil is considered.


The hypothesis of soil incompressibility and plane strain in the axial direction is
assumed. The axisymmetry imposes that the stress increments in the directions r,
and z are principal (Figure 5). The boundary conditions can be specified either in
displacements or in stresses.
Along the wall cavity r = ro:

ur = uo or r = po

[8]

Undrained strength properties of clays

At an infinite distance r = r :

ur = 0 and r = 0

[9]

where ro is the initial borehole radius and uo is the radial displacement at the
cavity wall.
There are two concentric annular zones around the probe (Figure 5). The first
one is bounded by a circle of radius re in which the material is subject to elastoplastic straining (r0 < r < re). The second one is located beyond re in which the
material behaves elastically (r > re).
The pressuremeter loading is then governed by the following equilibrium
equation:
r r
+
=0
r
r

[10]

r and are the radial stress and the circumferential stress respectively.

z
Elastic zone

rO
z

ro

r
r

dz

Elasto-plastic
zone

Figure 5. Equilibrium of a soil element.

r+dr

re

Undrained strength properties of clays

4.1. Solution of the problem in the elastic zone

Taking into account the radius re of the elastic outside zone with a Poisson ratio
= 0.5, we obtain:

d r =

d r =

re
r2

d =

du e

2 re
E
du e
3 r2

re
r2

du e

d z = 0

2 r
d = E e2 du e
3 r

[11]

d z = 0

[12]

The straining takes place without volume change and at constant mean stress.

4.2. Solution of the problem in the elasto-plastic zone

The plastic incompressibility implies that the yield surfaces are represented by
cylinders in the principal stress space (Von-Mises criterion). Thus, the incremental
strain vector d ijp is normal to the sections of the yield surfaces characterised by the
circles of radius Sk and center X ijp . Furthermore, the condition of plane strain
imposes that the strain path is normal to the z axis. The two conditions lead to the
following relationships between the deviatoric increments:

d r = C (r )ds r

d = d r

(
n*

C (r ) =

)(

Jk
3
+2
s r X rk s r X rk
2
2E
k =1 S k

d z = 0

[13]

[14]

Sk being a constant proportional to the threshold stress of the kth yield surface,
which is a function of the undrained cohesion cu. The coefficient C(r) is a function
depending on the distance r from the studied point of the soil mass to the axis of the
borehole, because the number, n*, of activated links of the model depends upon this
distance.
Further, to solve the problem completely, an unknown isotropic mean stress p(r)
may be introduced:
d r = dp(r ) +

d r
C (r )

[15]

d = dp(r )

d r
C (r )

[16]

d z = dp(r )

[17]

Undrained strength properties of clays

10

This pressure dp(r) is determined from the equilibrium equation (equation [10])
and the boundary conditions (equations [8] and [9]). Therefore, the following
expressions are deduced:
re
re

1
dr

dp(r ) = ro u o
+
2
r 2 C (r )
r 3C (r )
r
r

p o

u o =
ro
re 2 C (r )

[19]

re

+ 2ro

[18]

dr

r C (r )
3

po and uo are the pressures applied to the cavity wall and displacement at
cavity wall respectively.
5. Numerical program

From the analysis of the pressuremeter test results, a computer code called
Clayident is developed in order to automatically identify the three parameters of
the proposed model, particularly, the undrained cohesion cu. This code can be used
in two different ways:
Given the parameters of the model, the corresponding pressuremeter curve can
be deduced;
Given the experimental pressuremeter curve, parameters of the model can be
identified by means of an optimisation method by comparing simulated curve to the
experimental curve.
5.1. Evaluation of the proposed model from pressuremeter test

The simplex algorithm is used to optimise the parameters A, E and cu to produce


the best fitting curve to the data (Nelder et al., 1965). At the beginning of the
optimisation procedure, initial values for E, cu and A must be defined. For this
purpose, analytical formulations taking into account analytical simplifying
assumptions are considered:
A study within the elastic zone of the soil around the pressuremeter allows the
initial value of E to be determined from the measured Menard pressuremeter
modulus or secant pressuremeter modulus at 2% for self boring pressuremeter.
A study within the elastoplastic zone of the soil around pressuremeter allows
the initial value of cu to be determined from the measured conventional limit
pressure and elastic modulus using the simplified formulation of Salenon (1966).

Undrained strength properties of clays

11

Using the Duncan et al. (1970) representation of the stress-strain curve for the
undrained response of the soil, and after some algebraic manipulations, the
following relationship can be obtained for the initial value of parameter A (Bahar et
al., 1999):
A=

6 R f cu Ri 2

R
Ei 1 R f Ri Ln(1 Ri ) + (1 2 R ) i
1 Ri

[20]

where Ei and cu are the initial values of Youngs modulus and undrained
cohesion previously defined, Rf is a reduction factor usually less than one (Rf = 0.7).
A numerical study suggests that there is a relationship between Ri and parameter
used to classify the soils according to Baguelin et al. (1978). Ri is defined from the
pressuremeter test which is a factor describing the first curvature of experimental
pressuremeter curve:
Ri =

p5% p0
p lim p o

[21]

where po, p5% and plim are the initial total horizontal pressure in the ground,
applied pressure at 5% volume change and conventional limit pressure respectively.
Figure 6 compares, for a typical optimisation case, the responses associated with
the initial and optimal values of the three parameters with the experimental data.
The optimal theoretical response is very close to the experimental data, indicating
that the constitutive model is able to describe the undrained stress-strain behaviour
of clays around the pressuremeter adequately. Figure 6 also compares the results
obtained using the proposed approach and those obtained from Pressident finite
element program taking into account the non linear elastic model (Duncan et al.,
1970). It can be noted that the numerical and theoretical curves are very close.
5.2. Influence of the model parameters on the simulated curve

In order to understand the influence of the model parameters on the numerical


response, we carried out a sensitivity study of the three parameters changing the
value of one parameter by 50% of its initial value. The reference parameters are
identified from results of self boring pressuremeter test carried out on the clay of
Cran site (France), at 3.0 m depth (Figure 6). The results are presented in Figures
7(a), 7(b) and 7(c). It may be observed that all the three parameters influence the
response. The elastic modulus E has a preponderant influence not only on the
beginning of the curve but also when large strains are developing. The parameter A
that describes the curvature has a similar influence on the pressuremeter curve.
Finally, the undrained cohesion cu has a negligible influence on the initial tangent of
the curve, but an important influence on the failure behaviour. These observations

Undrained strength properties of clays

12

indicate that none of the three parameters of the model can be fixed to an average
value. The three parameters must be determined by the proposed optimisation
procedure.

140
120

Pressure (kPa)

100
80
60

Optimal parameters (E=4700 kPa, Cu=25 kPa, A=0.067)

Test

40

Bahar & Olivari approach


with optimal parameters
Bahar & Olivari approach
with initial parameters
Pressident approach

20
0
0

250

500

750
1000
1250
Volume change (cm3)

1500

1750

Figure 6. Experimental, theoretical and finite element computation pressure


expansion curves.

5.3. Distribution of total stresses at various levels of cavity expansion

Experimental data simulating the pressuremeter test using a large hollow


cylindrical specimen indicate that the total vertical stress does change during
expansion cavity (Thevanayagam et al., 1994). The results of other analytical
studies simulating a plane strain cylindrical cavity expansion in normally
consolidated clays also indicate changes in total vertical stresses in all elements
surrounding the cavity. An example of results for the total stresses r, and z,
using the proposed model is shown in Figures 8(a), 8(b) and 8(c). The figures show
that the evolution of these stresses is important until a radius of approximately 50
cm around the probe (SBP, Self Boring Pressuremeter, ro = 8 cm and L/D=4). Figure
8(c) indicates that the total vertical stress z initially decreases at very low strain
levels and subsequently increases.

Undrained strength properties of clays

140

140

120

120

Pressure (kPa)

160

Pressure (kPa)

160

100

13

100

80
60
40

80
60
40

0.50 E (- 50%)
E = 4700 kPa (reference)
1.5 E (+ 50%)

20

0.50A (- 50%)
A=0.067 (reference)
1.5A (+ 50%)

20

400

800

1200

1600
3

Volume change (cm )

2000

400

800

1200

1600

Volume change (cm3)

(a)

2000

(b)

160
140

Pressure (kPa)

120
100
80
60
0.50cu (- 50%)
cu=25 kPa (reference)

40
20

1.5cu (+ 50%)

0
0

400

800

1200

1600

Volume change (cm3)

2000

(c)
Figure 7. Influence of the parameters E, A and cu.

5.4. Response of the model on an unloading path

The unloading response of the proposed model on the pressuremeter path is very
easy to obtain. Indeed, the Prager model complies with the Masing rule, so that the
unloading curve has the same shape as the first loading curve, except that the scale
is increased by a factor of two. This result seems to be acceptable for isotropic clays,
as shown in Figure 9 (clay of Cran site). The unloading part of the curve can be used
in a very profitable way to confirm the elastic value for the modulus of the soil.

Undrained strength properties of clays

Radial total stress (kPa)

120
110
100
90
80
70
60
50
40

20% limit pressure (1)


100 % limit pressure (5)

1
0.00

0.25

0.50

Total vertical stress (kPa)

C ircum ferential stress (kPa)

(a)

(b)

100
90
80
70
60
50
40

(c)

14

70
60

0.75
1.00
Radius r (m)

1.25

1.50

20 % limit pressure (1)

100 % limit pressure (5)

50
1

40
30
20
0.00

0.25

0.50

0.75
1.00
Radius r (m)

1.25

1.50

20% limit pressure (1)


100% limit pressure (5)

1
0.00

0.25

0.50

0.75

1.00

1.25

1.50

Radius r (m)

Figure 8. Distribution of total stresses at various levels of cavity.

6. Application of the proposed method to identify undrained cohesion of clays

The proposed method has been used to determine the undrained cohesion of
some clay in Algeria. The results obtained are compared to those derived using
Pressident method, the empirical methods proposed by Menard (1957) and Amar
et al. (1971) and others in situ tests. The two empirical methods are established by
correlation between the limit pressure obtained from pressuremeter tests and the
undrained shear strength obtained from field vane and triaxial tests for soft cohesive
soils.
Pressident (Pressuremeter Identification) is a numerical program developed at
the Ecole Centrale de Lyon, France, which uses the non linear elastic model of
Duncan (Bahar et al., 1993).

Undrained strength properties of clays

15

500

Simulation

450

Test

Pressure (kPa)

400
350
300
250
200
150
100
50
0
0

250

500 750 1000 1250 1500 1750


Volume change (cm3)

Figure 9. Pressuremeter test with unloading (Cran clay).


The method of Menard (1957) is an empirical relationship often used in the
analysis of Menard pressuremeter data. It is given by:

cu =

pl _ po
5.5

[22]

The method of Amar et al. (1972) is an empirical relationship given by:


cu =

pl _ po
+ 25 (kPa)
10

[23]

pl, po, and cu are the limit pressure, the in situ total horizontal stress and the
undrained shear strength respectively.
The undrained shear strength was also determined from cone penetration tests
using an empirical relationship cu=(qc-qo)/Nk. qc, qo and Nk are the cone resistance,
the in situ vertical stress and an empirical factor respectively. The value of Nk ranges
between 10 and 15 for cohesive soils (Cassan, 1988).
6.1. Very soft to soft clay of Annaba

The site is located in the east of Algeria. The soil stratigraphy encountered on
site consists on muddy soft to very soft brownish clays. The thickness of the clay
layer is about 25 to 30 m. The ground water table was about 5 m depth from the
ground surface. The clays are saturated. The natural water content wn varies between

Undrained strength properties of clays

16

18% and 60%. The plasticity index varies between 26% and 35%. The shear
strength parameters derived from consolidated undrained triaxial tests with pore
pressure measurement range from 10 to 21 for the friction angle and from 11 kPa
to 36 kPa for cohesion. A conventional limit pressure ranging from 200 kPa to 800
kPa characterizes the clays. Consolidation testing indicates that the soils are
unconsolidated with a high compressibility index, Cc ranging from 11% to 41%.
Figure 10 shows an identification example concerned with the test realised at
4 m depth, in borehole SP6. The undrained cohesion obtained using the proposed
method and the one obtained using Pressident method and empirical methods
mentioned above are presented in Figures 11(a) and 11(b). It can be noted that the
proposed method gives a values relatively similar to those deduced by Pressident
method. In these figures, it can also be noted that, for limit pressure less than 300
kPa, the undrained cohesion values deduced from the proposed approach are close to
those obtained from the empirical methods. For limit pressure ranging between 300
kPa and 700 kPa the undrained cohesion values deduced from the proposed
approach were on the average 170% higher than those deduced from the empirical
methods. There are number factors that can explain the observed difference in the
values of cu obtained by different methods. The two empirical methods are
established by correlation between limit pressure obtained from pressuremeter tests
and undrained shear strength obtained from field vane and triaxial tests for soft
cohesive soils. In general, pressuremeter undrained shear strength obtained using
cavity expansion methods are significantly higher than the values obtained using
other in situ or laboratory tests.
600
Model parameters
A = 0.0087
E = 8676 kPa
cu = 76 kPa

500

Volume change (cm3)

po = 32 kPa
vo = 135 cm3

400

300

200
Site of Annaba
Borehole SP6

100

Test (Depth : 4 m)
Simulation

0
0

50

100

150

200

Pressure (kPa)

250

300

350

Figure 10. Example of identification of the proposed model parameters.

Undrained strength properties of clays

0
2

Undrained cohesion cu (kPa)


20

40

60

80

100

Annaba site
Borehole SP3

10

10

20

40

60

80

100

120

Annaba site
Borehole SP6

Depth (m)

Depth (m)

Undrained cohesion cu (kPa)

17

12

12

14

14

16

16

18

18

20

Empirical relationship of Amar et al.

20
Empirical relationship of Amar et al.

Empirical relationship of Menard

22

Pressident method

22

Bahar & Olivari method

24

Empirical relationship of Menard


Pressident method
Bahar & Olivari method

24
a)

b)

Figure 11. Undrained cohesion of very soft to soft clays, site of Annaba (Algeria).
6.2. Stiff to very stiff clay of Bab Ezzouar

The site is located in Algiers. The soil stratigraphy encountered on the site
consists on stiff to very stiff clays, which overlies a layer of sandstone material. The
thickness of the clay is about 15 to 18 m. The clays are saturated. The natural water
content wn varied between 7% and 21%. The plasticity index varied between 22%
and 27%. The shear strength parameters derived from consolidated undrained
triaxial tests with pore pressure measurement range from 7 to 21 for the friction
angle and from 14 kPa to 126 kPa for the cohesion. A conventional limit pressure
ranging from 500 kPa to 2600 kPa and pressuremeter moduli ranging from 4700 kPa
to 44000 kPa characterizes the clays. Consolidation testing indicates that the soils
are normally consolidated to slightly overconsolidated with medium compressibility,
Cc ranging from 10% to 17%.
Figure 12 shows an identification example concerned with the test realised at
10 m depth, in borehole SP1. The undrained cohesions, obtained using the proposed

Undrained strength properties of clays

18

method, are presented in Figures 13(a), 13(b). These figures also show the
comparison of the obtained results to those obtained by Pressident method, by
empirical relationships and from cone penetration tests. It can be noted that the
proposed method gives values relatively similar to those deduced by Pressident
method and by the cone penetration test. It is also observed that, in common with the
findings of others investigators, for stiff to very stiff clay the cu values obtained from
the proposed method are consistently higher than the corresponding cu values
obtained by empirical relationships and triaxial tests (factor between 1.35 and 2.30).
Furthermore, the interpreted soil parameters had reasonable values when compared
with the cone penetration test results.
High undrained shear strengths from pressuremeter tests have been frequently
observed. The measured cu will be affected by the in situ or laboratory method used
and the stress path followed during the test (Wroth, 1984). Wroth (1984) showed
that the undrained shear strength derived from pressuremeter tests should be larger
than the strengths derived from field vane tests due to the nature of the different
stress paths. As explained by other researchers (Baguelin et al, 1978), this difference
is due to disturbance during boring prior testing and also possibly due to the
difference in the mode of failure during the test. It has also been recognized by
many researchers that some drainage and creep takes place during pressuremeter
tests in clay (Wroth, 1984). The effect of drainage and creep can result in
overestimation of the undrained shear strength.
600
Model parameters
A = 0.009
E = 44332 kPa
cu = 496 kPa

Volume change (cm3)

500

po = 100 kPa

400

vo = 110 cm3

300
200
Site of Bab Ezzouar
Borehole SP1

100

Test (Depth : 10 m)
Simulation

0
0

250

500

750

1000

1250

Pressure (kPa)

1500

1750

2000

Figure 12. Example of identification of the proposed model parameters.

Undrained strength properties of clays

Undrained cohesion cu (kPa)


100

200

300

400

500

600

100

200

300

Bab Ezzouar Site


Borehole SP1

400

500

600

Bab Ezzouar site


Borehole SP2

Depth (m)

Depth (m)

Undrained cohesion cu (kPa)

19

10

10

12

12

14

14

16

16
Empirical relationship of Amar et al.

18

Empirical relationship of Amar et al.


Empirical relationship of Menard

18

20

CPT (Nk=15)

Triaxial tests (UU)


CPT (Nk=10)

Triaxial tests (UU)


CPT (Nk=10)

Empirical relationship of Menard

20

CPT (Nk=15)
Pressident method

Pressident method

Bahar & Olivari method

Bahar & Olivari method

22

22
a)

b)

Figure 13. Undrained cohesion of stiff to very stiff clays, site of Bab Ezzouar (Algeria).

6.3. Clay of Cran site (France)

The proposed approach was applied to identify the undrained cohesion of clay of
Cran site (Boubanga, 1990; Cambou et al., 1993). Figure 14(a) shows optimised
simulations of different pressuremeter tests performed at different depths leading in
each case to the best set of parameters. The results obtained by the proposed method
agree well with the experimental data. Figure 14(b) shows the profiles of cu
obtained for Cran clay as a function of depth, and lead to a comparison with
different methods used to obtain cu. It appears that the proposed method gives values
of cu slightly greater than those obtained by undrained triaxial tests. Furthermore, it
can be seen that the proposed method gives values in rather good agreement with
field vane test results and larger than the strength derived from triaxial test. For soft
clay of Cran site, the proposed method had reasonable values when compared with
the cone penetration test results. The self boring pressuremeter and the field vane
test shear strengths are similar and appear to increase at the same rate.

Undrained strength properties of clays

Undrained Cohesion (kPa)

2000

0 10 20 30 40 50 60 70 80 90 100

Simulation
Test

0
1

11m 13m

1600

Volume Change (cm3)

20

2
3
3m

1200

6m

Vane test
Triaxial test
Pressident approach
Bahar & Olivari approach

9m

Depth (m)

800

6
7
8
9

10

400

11

(a)
0
0

50 100 150 200 250 300 350 400

Pressure (kPa)

12
13

(b)

14

Figure 14. Soft clay of Cran site. (a) best set of parameters for simulations of
pressuremeter tests performed at different depths, (b) undrained cohesion profiles.

6.4. Identification of the model parameters using the results of a pressuremeter


test with unloading

The proposed approach is used to interpret the Fucino soft clay (Ferreira et al.,
1992) and the obtained results are compared to those obtained by using Ferreira et
al. (1992) and Jefferies (1988) methods. The clay deposit is described as soft,
homogeneous, highly structured and cemented. Ferreira et al. (1992) method is an
extension of Jefferies (1988) method incorporating the unloading portion of the
pressuremeter test to derive the initial shear modulus and undrained shear strength.
The soil behaviour is represented by a hyperbolic (non linear elastic) relationship
between the shear stress and the circumferential strain. Jefferies (1988) method is an
extension of the Gibson-Anderson (1961) theory for pressuremeter test performed in
clays incorporating the complete loading and unloading portion of the test. The
method was based on an ideal elastic perfectly plastic soil model and assumed that
the installation was carried out with minimum disturbance. The ratio of the
unloading strength of the clay was assumed to be known. Commonly, the
pressuremeter results are plotted in terms of radial pressure r versus loge(V/V),
where V/V is a measurement of the cavity strain related to the deformed state.
Jefferies (1988) used computer-aided modelling techniques to visually compare the
measured response with the numerically derived curves.

Undrained strength properties of clays

21

The best simulated curve leading to the best set of model parameters is shown in
Figure 15(a) and Table 1. Table 1 and Figure 15(b) compare the derived parameters
using Jefferies (1988), Ferreira et al. (1992) and proposed methods. The set of
derived parameters leads to a curve which is very close to results provided by other
techniques used to identify the undrained shear strength of clays.
140

Parameters
E=30282 kPa
A=0.0076
cu=114 kPa

Shear strength (kPa)

120
100

Pressure (kPa)

850
800
750
700
650
600
550
500
450
400
350
300
250

80
60
40
Ferreira et al. (1992)
Jefferies (1988)
Bahar & Olivari

20

Simulation
Test (V2P14 - depth : 26m)

0
0 1 2 3 4 5 6 7 8 9 10 11 12

Cavity strain (%)


(a)

Cavity strain (%)


(b)

10

12

Figure 15. Fucino clay test (V2P14, 26m depth), (a) Identification with the
proposed method, (b) Comparison of stress-strain curves using different methods.

Table 1. Results obtained using different methods.

Methods
Jefferies (1988)
Ferreira et al., (1992)
Proposed method

Test V2P10 (18 m) Test V2P14 (26 m)


G (kPa) cu (kPa) G (kPa) cu (kPa)
4300
5952
6068

55
93
60

6800
11188
10094

100
112
114

7. Load-settlement prediction of a bored pile

The proposed method has been used to predict the load-settlement curve of a
bored pile of case test which was organised for the International Symposium ISP52005, taking place at the occasion of the 50 years of pressuremeters (Reiffsteck,
2006). The pile diameter is D=0.5 m and its length is 12m. The pile is embedded in
a 9.6 m thick clay layer, below a 2.4 m thick silt layer. The water table is located 1.8
m below ground level (Figure 16). The laboratory tests carried out on soil samples
extracted close to the pile location showed that the site physical properties were

Undrained strength properties of clays

22

relatively homogeneous. The shear strength parameters derived from consolidated


undrained triaxial tests with pore pressure measurements are c=57 kPa and =23.
The Young modulus at 0.2% strain derived from unconsolidated undrained triaxial
tests is E=50 MPa. The results of three Menard pressuremeter boreholes, given in
the exercise, are used to define the soil parameters by means of the proposed method
and by the Pressident approach. The corresponding best curve and the best set of
parameters are given in Figures 17(a) and 17(b). Figures 18(a) and 18(b) summarize
the evolution of undrained cohesion evaluated by means of the proposed method, by
the Pressident approach, by the empirical relationships proposed by Mnard and
Amar et al., and by triaxial tests. The proposed method and the Pressident
approach give very close results.
For this site, the calculation of the bearing capacity and settlements of a bored
pile using the parameters derived from the proposed method was achieved using
FLAC3D software (Itasca, 2005) considering an elastoplastic model with Mohr
Coulomb criterion. Figure 19 gives the parameters used for achieving the
calculations. Figure 20 compares the results obtained by the proposed method to
those obtained using the parameters derived from Pressident approach an
experimental data (Reiffsteck, 2006). For this case again, one can note that the
proposed method is in very good agreement with the measured load-settlement
curve on site.

Figure 16. Pile and soil sketch (Reiffsteck, 2006).

Undrained strength properties of clays

Depth : 1.0 m
po = 10.8 kPa
Vo = 117 cm3

600

700

Borehole SP1
Test
Simulation

Volume change (cm3)

A = 0.0055
E = 6526 kPa
cu = 63 kPa

500

parameters
500 Model
k = 251, n = 0.5

Depth : 14.0 m
Po = 151 kPa

200

A = 0.0037
E = 65 205 kPa
cu = 248 kPa

100

Rf =0.7

300

Vo = 28 cm3

200

kb = 4190, m = 0.5

400 cu = 111 kPa, =0

400
300

Depth : 1.0 m
Po = 10.8 kPa
600 V = 135 cm3
o

Volume change (cm3)

700

23

Borehole SP2

100

Test
Simulation

0
0

400

800

1200

1600

100 200 300 400 500 600 700

Pressure (kPa)
Pressure (kPa)
(a)
(b)
Figure 17. Identification of model parameters from pressuremeter test results, best
fit curve: (a) proposed method, (b) Pressident method.

Undrained cohesion (kPa)


50

100

150

200

250

Undrained cohesion (kPa)


0

300

50

100

150

10

10

12

12

14

14
Pressident method

Pressident method
Bahar & Olivari method

16

Amar et al. empirical relationship

Bahar & Olivari method


Menard empirical relationship

Menard empirical relationship

18

250

Borehole SP3
2

Depth (m)

Depth (m)

Borehole SP1
2

16

200

18

Amar et al. empirical relationship


Triaxial UU

Triaxial UU

20

20

(a) borehole SP1

(b) borehole SP3

Figure 18. Profile for undrained cohesion derived from different methods.

300

Undrained strength properties of clays

= 18 kN/m3
E = 7972 kPa, = 0.49
Cu = 78 kPa , u = 0

= 18 kN/m3
E = 39312 kPa
= 0.49
Cu = 177 kPa
u = 0

Soil parameters identified from


Bahar & Olivari approach

Layer 1
Silt

Layer 1
Silt

Layer 2
Clay

Layer 2
Clay

= 18 kN/m3
E = 79023 kPa
= 0.49
Cu = 241 kPa
u = 0

12 m

6.60 m

2.40 m

Soil parameters identified from


Pressident approach

Layer 3
Clay

Layer 3
Clay

= 18 kN/m3
E = 12860 kPa, = 0.5
Cu = 80 kPa, A = 0.04

= 18 kN/m3
E = 32257 kPa
= 0.5
Cu = 174 kPa
A= 0.006
= 18 kN/m3
E = 100325 kPa
= 0.5
Cu = 240 kPa
A = 0.003

Figure 19. Pile and model parameters of soils.


Load (kN)
0

200

400

600

800

1000 1200 1400

0
5
10

Settlement (mm)

15
20
25
30
35
Measured

40

24

Bahar & Olivari approach


Pressident approach

45
50

Figure 20. Comparison of measured and calculated load-settlement curves.

Undrained strength properties of clays

25

8. Conclusions

A method to interpret pressuremeter test results carried out in saturated clays is


proposed. The proposed model is specifically developed to describe the undrained
behaviour of clays. Introducing an analytical formulation of the total stress-strain
curve obtained during an undrained triaxial test, the model depends only on three
parameters. The response of the model on the pressuremeter path is in good
agreement with experimental data. An example of application of the proposed
method has shown realistic results compared to those obtained by other methods.
However, further researches are needed to verify these conclusions for various clay
types with both field and laboratory test results.

Acknowledgements
The authors thank the Construction and Testing Engineering Laboratory (CTELAB)
and Laboratoire de lHabitat et de la construction du Centre (LHCC) especially for
some data available for scientific use. They thank also Pr. B. Cambou and Dr. G.
Olivari for their comments and discussions.
9. Bibliographie
Amar S., Jzquel J.F., Essais en place et en laboratoire sur sols cohrents: comparaison des
rsultats, Bulletin de Liaison des Ponts et Chausses, vol. 58, 1972, p. 97-108.
Baguelin F., Jezequel J.F., Leme E., LeMhaut A., Expansion of cylindrical probes in cohesive
soils, J. of the Soil Mechanics and Foundations Division, vol. 98, n 11, 1972, p. 1129-1142.
Baguelin F., Jezequel J.F., Shields D.H., The pressuremeter and foundation engineering,
Switzerland, Trans Tech Publications, 1978.
Bahar R., Analyse numrique de lessai pressiomtrique : application lidentification de paramtres
de comportement des sols, Thse de doctorat. Ecole Centrale de Lyon, 1992.
Bahar R., Cambou B., Fry J.J., Forecast of creep settlements of heavy structures using
pressuremeter tests, Computers and Geotechnics, vol. 17, 1995, p. 507-521.
Bahar, R., Interpretation of pressuremeter tests carried out in stiff clays, Proceeding of 1st
Intern. Conf. on Site Characterization ISC98, Atlanta, USA, 26-28 mai 1998, p. 735-740.
Bahar R., Abed Y., Olivari G., Theoretical analysis of the behaviour of clays around a
pressuremeter, Proceeding of 12th African Regional Conference, Durban, South Africa,
October 25-27 1999, p. 135-142.
Bahar R., Aissaoui T. & Kelanemer S., Comparison of some methods to evaluate the
undrained cohesion of clays, Proceeding of the 16th Inter. Conference on Soil Mechanics
and Geotechnical Engineering, Osaka, Japan. vol. 2, September 12-16 2005, p. 667-670.
Boubanga A., Identification de paramtres de comportement des sols partir de l'essai
pressiomtrique. Thse de doctorat. Ecole Centrale de Lyon, 1990.

Undrained strength properties of clays

26

Cambou B., Bahar R., Utilisation de l'essai pressiomtrique pour l'identification de


paramtres intrinsques du comportement d'un sol, Revue Franaise de Gotechnique, 63,
1993, p. 39-50.
Cassan M., Les essais in situ en mcanique des sols, vol. 1, Paris, Eyrolles (Eds), 1978.
Denby, G.M., Self-boring pressuremeter study of the San Francisco Bay mud, Ph.D thesis, Stanford
University, California. 1978.
Duncan J.M., Chang C.Y., Non linear analysis of stress and strain in soils, Journal of
Geotechnical Engineering Division., SM5, 1970, p. 1629-1653.
Ferreira R., Robertson P.K., Interpretation of undrained self-boring pressuremeter test results
incorporating unloading. Canadian Geotechnical Journal, vol 29, 1992, p. 918-928.
Gibson R.E., Anderson W.F., In-situ measurement of soils properties with the
pressuremeter Civ. Engng. Publ. Wks. Rev., Vol. 56, 1961, p. 615-618.
Habib P., La rsistance au cisaillement des sols, Thse de doctorat, 1953, Paris.
Iwan W.D., On a class of models for the yielding behaviour of continuous and composite systems
Journal of Applied Mechanics, 1967, p. 612-617.
Itasca, FLAC3D, Fast Lagrangian Analysis of Continua. Itasca Consulting Group, Users
manual, Minneapolis, 2005.
Jefferies M.G., Determination of horizontal geostatic stress in clay with self bored
pressuremeter, Canadian Geotechnical Journal, vol. 25, 1988, p. 559 - 573.
Ladanyi B., In-situ determination of undrained stress-strain behaviour of sensitive clays with the
pressuremeter Canadian Geotechnical Journal, Vol. 9, N 3, 1972, p. 313-319.
Mnard L., Mesures in situ des proprits physiques des sols, Annales des Ponts et
Chausses, l.3, 1957, p. 357-376.
Monnet J., Chemaa T., Etude thorique et exprimentale de lquilibre lasto-plastique dun sol
cohrent autour du pressiomtre, Revue franaise de Gotechnique, n 73, 1995, p. 15-26.
Monnet J., Numerical validation of an elastoplastic formulation of the conventional limit
pressure measured with the pressuremeter test in cohesive soil, Journal of Geotechnical
and Geoenvironmental Engineering, Vol. 133, Issue 9, September 2007, p. 1119-1127.
Nelder J..A., Mead R., A simplex method for function minimisation , The Computer Journal, vol.
7, 1965, p. 308-313.
Olivari G., Bahar R., Response of generalized Prager's model on pressuremeter path, Proc.
4th Int. Sym. on Pressuremeter. Sherbrooke, Canada, 17-19 mai 1995, p. 207-213.
Palmer A.C., Undrained plain strain expansion of a cylindrical cavity in clays: a simple
interpretation of the pressuremeter test, Geotechnique, vol. 22, n 3, 1972, p. 451-457.
Paraharam S., Chameau J.L., Alischaefel A.G., Holtz R.D. Effect of disturbance on
pressuremeter results in clays Journal of Geotechnical Engineering, vol. 116, n 1, 1990,
p. 35-53.
Prevost J.H., Hoeg K., Analysis of pressuremeter in strain-softening soil, J. Geotech. Engng. Div.
vol. 101, GT8 , 1975, p. 717-731.

Undrained strength properties of clays

27

Reiffsteck P., Portance et tassements dune fondation profonde - Prsentation des rsultats
du concours de prvision. Comptes rendus Symposium international 50 ans de
pressiomtres (ISP 5), Marne-la-Valle, 22-24 aot 2005, vol. 2, 2006, p. 521-535.
Salenon J., Expansion quasi-statique dune cavit symtrie sphrique ou cylindrique dans
un milieu lastoplastique, Annales des Ponts et Chausses, vol. 3, 1966, p. 175-187.
Thevanayagam S., Chameau J.L., Altschaeffl A.G., Some aspects of pressuremeter test
interpretation in clays, Geotechnique, vol. 44, n 2, 1994, p. 319-334.
Wroth C.P., The interpretation of in situ soil tests, 24th Rankine lecture, Geotechnique, vol.
34, 1984, p. 449-489.
Zentar R, Hicher PY, Moulin G., Identification of soil parameters by inverse analysis,
Computers and Geotechnics, vol. 28, n 2, 2001, p. 129144.

Vous aimerez peut-être aussi