You are on page 1of 14

Some Pitfalls when using Modified Cam Clay

David Potts1 & Lidija Zdravkovic1


1

Imperial College, London, UK, d.potts @ic.ac.uk

Keywords: Constitutive models, numerical analysis, Finite elements

Abstract
This paper considers some of the pitfalls that can arise when performing numerical analysis
of geotechnical problems using the Modified Cam clay constitutive model. In particular the
implications of the shape of the plastic potential in the deviatoric plane on failure strengths
is considered. In addition some of the problems associated with using such a critical state
model for undrained analyses is discussed.

1. Introduction
The use of numerical analyses, such as the finite element method, when applied to
geotechnical problems, can be extremely complex. While in principle the method can be
used to provide a solution to most of the problems that we may wish to analyse, there are
approximations which can lead to errors. These approximations can be classified into two
groups. Firstly, there are approximations in the numerical method (i.e. Finite element
method) itself and secondly, there are approximations arising from the idealisations made
by the user when reducing the real problem to a form which can be analysed, [1].
Examples of the second group are the many potential errors which can be associated with
a users lack of in depth understanding of the constitutive model employed to represent
soil behaviour. This is a common source of error, due to the complexities of many of the
constitutive models currently available. As an example this paper considers the effect of the
shape of the yield and plastic potential surfaces on soil strength. This is demonstrated by
using the modified Cam clay model to show that what seems like sensible input parameters
can result in unrealistic predictions. Experience indicates that this is a very common pitfall
that many users unknowingly fall into, with the results that their analyses predict erroneous

collapse loads which are usually unconservative.


As the undrained strength is not one of the input parameters to the conventional critical
state constitutive models, use of such models to analyse undrained problems can be
problematic. This is particularly so if the site investigation has been designed to obtain
undrained strength values. It is shown that realistic undrained strength profiles can be
obtained by a suitable selection of the model input parameters and initial stress conditions.
However, such conditions can only be achieved if the numerical computer program (i.e.
finite element program) has flexible input facilities.

2. Development of modified Cam clay


Since the work of Coulomb [2] and Rankine [3], there has been a long history of
applications of the theory of plasticity to geomechanics. Slip line theory and the theorems
of limit analysis are in frequent use. However, the development of realistic constitutive
models for soils has lagged significantly behind the corresponding formulations for metals,
in spite of the fact that the features of soil behaviour are clearly consistent with the elastoplastic framework. Early attempts of taking into account the frictional character of geomaterials by extending the von Mises failure criterion [4], or by generalising the MohrCoulomb failure envelope failed to model adequately many basic features of soil behaviour.
The first critical state models were the series of Cam clay formulations, developed at the
University of Cambridge, by Roscoe and his co-workers. The formulation of the original
Cam clay model as an elasto-plastic constitutive law is presented by Roscoe and Schofield
[5] and Schofield and Wroth [6]. Afterwards, Roscoe and Burland [7] proposed the
modified Cam clay model.

2.1 Basic formulation in triaxial stress space


Modified Cam clay, like the original Cam clay model, was originally developed for triaxial
loading conditions. The model is essentially based on the following assumptions:
- A piece of clay, which is subjected to slow, perfectly drained isotropic
(F1N=F2N=FN3) compression, moves along a trajectory in the v-lnpN plane (v =
specific volume = 1+e, p =(F1N+F2N+FN3)/3 ), which consists of a virgin
consolidation line and a set of swelling lines, see Figure 1. Initially, on first loading,
the soil moves down the virgin consolidation line.

If subsequently unloaded from point


b, it moves up the swelling line bc.
If re-loaded, it moves back down this
same swelling line until point b is
reached, at which point it begins to
move down the virgin consolidation
line again. If unloaded from point d, it
moves up the swelling line de. The
virgin consolidation line and the
swelling lines are assumed to be
straight in v-lnpN space and are given by
the following equations:
)LJXUH%HKDYLRXUXQGHULVRWURSLF
FRPSUHVVLRQ

v + (ln p ) = v1
v + (ln p ) = vs

(virgin consolidation line)


(swelling line)

(1)

The values of 6, 8 and v1 are characteristics of the particular type of clay, whereas
the value of vs is different for each swelling line. Volume change along the virgin
consolidation line is mainly irreversible or plastic, while volume change along a
swelling line is reversible or elastic.
-

The behaviour under increasing triaxial


shear stress, q = FvN!FhN= /3J, (where
J=[(F1N!F2N)2+(F2N!F3N)2+(F1N!F3N)2]0.5*
(1//6)) is assumed to be elastic until a
yield value of q is reached, which can
be obtained from the yield function
F({FN},{k}) = 0. As noted above,
behaviour is elastic along swelling lines
and therefore the yield function plots
above each swelling line as shown in
Figure 2. For modified Cam clay the
yield surface is assumed to take the
form:
)LJXUH<LHOGVXUIDFH

J
p

o 1 = 0
F ({ },{k}) =
p

pM J

(2)

where pN is the mean effective stress, J


is the deviatoric stress, MJ is another
clay parameter, and poN is the value of
pN at the intersection of the current
swelling line with the virgin
consolidation line, see Figure 2. The
projection of this curves onto the J-pN
plane is shown in Figure 3 where it can
be seen that the modified Cam clay
yield surface plots as an ellipse. The
parameter poN essentially controls the
size of the yield surface and has a )LJXUH3URMHFWLRQRI\LHOGVXUIDFHRQWR
-SNSODQH
particular value for each swelling line.
As there is a yield surface for each
swelling line, the yield function, given
by Equation (2) defines a surface in v-JpN space, called the Stable State
Boundary Surface, see Figure 4. If the
v-J-pN state of the clay plots inside this
surface, its behaviour is elastic, whereas
if its state lies on the surface it is elastoplastic. It is not possible for the clay to
have a v-J-pN state that lies outside this
surface.
-

Hardening/softening is isotropic and is


controlled by the parameter poN which is
related to the plastic volumetric strain,
,vp, by:
)LJXUH6WDWHERXQGDU\VXUIDFH

dpo
v
= d vp

po

(3)

Equation (3) therefore provides the


hardening rule.

When the soil is plastic (i.e. on the Stable State Boundary Surface), the plastic
strain increment vector is taken normal to the yield curve. Consequently, the model
is associated, with the plastic potential P({FN},{m}) being given by Equation (2).

As noted above, behaviour along a swelling line is elastic. This means that the
elastic volumetric strain, ,ve, can be determined from Equation (1):

d ve =

dv dp
=
v
v p

(4)

This gives the elastic bulk modulus, K, as:


K=

dp
d ve

vp

(5)

In the original formulation, no elastic shear strains are considered. To avoid


numerical problems and to achieve a better modelling inside the state boundary
surface, elastic shear strains are usually computed from an elastic shear modulus,
G, which is an additional model parameter.
In the above form, both the Cam clay and modified Cam clay models require five material
parameters: v1 , 6, 8, MJ and G. Sometimes an elastic Poissons ratio, , is specified instead
of G.

2.2 Extension to general stress space


The original critical state formulation is based, almost exclusively, on laboratory results
from conventional triaxial tests. The portions of stress space in which these tests operate
are severely restricted as the intermediate principal stress must be equal to either the major
or the minor principal stress. Because of this, the basic formulation is developed in terms
of q (=F1N!F3N) and pN. For numerical analysis, the models have to be generalised to full
stress space by making some assumption on the shape of the yield surface and plastic
potential in the deviatoric plane. The first generalisation [7] is achieved by effectively
replacing q by J. This substitution is made in Equation (2). In general stress space this is
equivalent to assuming that the yield and plastic potential surfaces (and hence the failure
surface) are circles in the deviatoric plane, see Figure 5. However, it is well known that a
circle does not represent well the failure conditions for soils, where a Mohr-Coulomb type
failure criterion is more appropriate. Roscoe and Burland [7] suggest that circular (in the

deviatoric plane) yield surfaces should be used


combined with a Mohr-Coulomb failure
criterion. This implies, however, that critical
state conditions can only be reached under
triaxial compression conditions (F2N= F3N).
In order to obtain a Mohr-Coulomb hexagon for
the yield surface in the deviatoric plane, MJ in
Equation (2) must be replaced by g(2):
g ( ) =

sin cs

sin sin cs

cos +
3

(6)
)LJXUH)DLOXUHVXUIDFHVLQWKHGHYLDWRULF
SODQH

where 2 is the Lodes angle (=tan-1[(2 (F2N!F3N)/(F1N!F3N)!1)//3]), NcsN is the critical state
angle of shearing resistance which replaces MJ as an input parameter. This expression gives
the hexagon shown in Figure 5. Equation (2) then becomes:
2

J
p

o 1 = 0
F ({ },{k}) =
p g ( )
p

(7)

Critical state conditions then occur with a constant NcsN. The discontinuity of the MohrCoulomb expression at 2 = !30o and 2 = +30o requires, usually, some ad hoc rounding of
the corners. Although sufficient as a first approximation, and certainly superior to a circle,
the Mohr-Coulomb criterion does not achieve a perfect agreement with observed soil failure
conditions.
Other failure surfaces have been suggested which are continuous and agree better with
experimental results in the deviatoric plane. Matsuoka and Nakais [8] and Lades [9] are
the best known, see Figure 5. In terms of g(2) Matsuoka and Nakais surface can be
expressed as:

g ( ) =

J 2f

(8)

where J20f can be obtained for a specific value of Lodes angle, 2, by solving the following
cubic equation:

(CMN 3) J 2f +

27

CMN sin3 ( J 2f ) 3/ 2 (CMN 9) = 0

(9)

in which:

CMN =

9 3 M J2
2 3 3
M J M J2 + 1
9

where MJ is the gradient of the critical state line in J-pN space, corresponding to triaxial
compression, 2 =!30o. In terms of the critical state angle of shearing resistance in triaxial
compression, (NcsN)2=-30, MJ in Equation (9) can be expressed as:

MJ =

2 3 sin cs
= 30
3 sin cs
= 30

(10)

The Lades surface can also be expressed by Equation (8), with J20f obtained for a specific
value of Lodes angle, 2, and mean effective stress, pN, from the following equation:

J 2f +

2
27

sin3 ( J 2f ) 3/ 2 CL = 0

(11)

in which:

CL =

1 pa

27 3 p

p
1+ 1 a
27 3 p

where 01 and m are material properties, and pa is atmospheric pressure.


As an alternative, Van Eekelen [10] proposes a family of continuous deviatoric plane yield
surfaces (or plastic potentials). They are expressed as:

g ( ) = X (1 + Y sin3 ) Z

[12]

where X, Y and Z are constants. There are restrictions on Y and Z if convex surfaces are
required. The substitution of g(2) into Equation (7) provides a flexible way to incorporate
a desired shape for yield surfaces or plastic potentials in the deviatoric plane. Circular, Lade

and Matsuoka and Nakai surface shapes can also be well approximated by Equation (12).
The importance of the model formulation in the deviatoric plane is highlighted by Potts and
Gens [11]. They demonstrate that the adoption of a plastic potential shape, gpp(2), in the
deviatoric plane and a dilation angle, <, determines the value of the Lodes angle at failure,
2f , in problems involving plane strain deformation. They show that some of the plastic
potential expressions proposed in the literature do not guarantee realistic values of 2f . They
also indicate that it is often necessary to have different shapes of the yield and plastic
potential surfaces in the deviatoric plane. For example, if the yield surface uses Equation
(6), which gives a Mohr-Coulomb hexagon in the deviatoric plane, then a different shape
must be adopted for the plastic potential, otherwise plane strain failure occurs with either
2f =!30o (i.e. triaxial compression) or 2f = 30o (i.e. triaxial extension). The use of different
shapes of the yield and plastic potential surfaces in the deviatoric plane results in a nonassociated constitutive model.

2.3 Undrained strength


As noted previously, the material parameters used to define the modified Cam clay model
include the consolidation parameters (v1, 6 and 8), the drained strength parameter (NcsN or
MJ) and its variation in the deviatoric plane, and the elastic parameter ( or G). They do not
involve the undrained shear strength, Su . As this model is often used to represent the
undrained behaviour of soft clays, whose strength is conventionally expressed in terms of
Su , this can be inconvenient. The undrained shear strength, Su , can be derived from the
input parameters and the initial state of stress as shown by Potts and Zdravkovic [1]. The
resulting equation is:

2(1 + 2 KoOC )
Su
OCR
2
NC
= g ( ) cos
(1 + 2 Ko )[1 + B ]
2
NC
vi
6
(1 + 2 Ko ) OCR [1 + B ]

(13)

By using this equation it is possible to select input parameters (6 , 8 , and NcsN or MJ) and
initial stress conditions (overconsolidated ratio, OCR and coefficient of earth pressure at
rest, Ko), so that the desired undrained strength distribution can be obtained. In this respect
care must be exercised because the undrained strength is always zero when the initial
vertical effective stress is zero.

3. Influence of the shape of the yield and plastic potential surfaces


As noted above, the shape of the plastic potential in the deviatoric plane can affect the
Lodes angle 2 at failure in plane strain analyses. This implies that it will affect the value
of the soil strength that can be mobilised. In many commercial software packages, the user
has little control over the shape of the plastic potential and it is therefore important that its
implications are understood.
Many software packages assume that both the yield and plastic potential surfaces plot as
circles in the deviatoric plane. This is defined by specifying a constant value of the
parameter MJ. Such an assumption implies that the angle of shearing resistance, NN, varies
with the Lodes angle, 2. By equating MJ to the expression for g(2) given by Equation (6)
and re-arranging, gives the following expression for N in terms of MJ and 2:

1 M J cos
= sin
M J sin

(14)

From this equation it is possible to express MJ in terms of the angle of shearing resistance,
NTC, in triaxial compression (2 = !30o), see Equation 15:

sin TC

(J TC ) = 23 3sin

(15)

TC

In Figure 6 the variation of NN with 2,


given by Equation 14, for three values
of MJ are plotted. The values of MJ have
been determined from Equation 14
using NNTC=20o, 25o and 30o. If the
plastic potential is circular in the
deviatoric plane, it can be shown, [1],
that plane strain failure occurs when the
Lodes angle 2=0o. Inspection of Figure
6 indicates that for all values of MJ
there is a large change in NN with 2. For
example if MJ is set to give NNTC=25o,
then under plane strain conditions the
mobilised NN value is NNPS=34.6o. This

Figure 6: Variation of N with 2 for constant


MJ

difference is considerable and much


larger than indicated by careful
laboratory testing. The difference
between NNTC and NNPS becomes greater
the larger the value of MJ.
The effect of 2 on undrained strength,
Su, for the constant MJ formulation is
shown in Figure 7. The variation has
been calculated using Equation 13 with
OCR=1, g(2)=MJ, Ko=1-sinNNTC and
6/8=0.1. The equivalent variation based
on the formulation which assumes a
constant NN, instead of a constant MJ, is
given in Figure 8. This is also based on
Equation 13 and the above parameters,
except that g(2) is now given by
Equation 6. The variation shown in this
figure is in much better agreement with
the available experimental data than the
trends shown in Figure 7.

Figure 7: Effect of 2 on Su, for the constant MJ


formulation

To investigate the effect of the plastic


potential in a boundary value problem
two analyses of a rough rigid strip
footing have been performed. The finite
element mesh is shown in Figure 9. The
modified Cam clay model was used to
represent the soil which had the Figure 8: Effect of 2 on Su, for the constant N
formulation
following material parameters, OCR=6,
v1=2.848, 8=0.161, 6=0.0322 and
=0.2. In one analysis the yield and
plastic potential surfaces were assumed to be circular in the deviatoric plane. A value of MJ
= 0.5187 was used for this analysis, which is equivalent to NNTC = 23o. In the second
analysis a constant value of NN=23o was used giving a Mohr-Coulomb hexagon for the yield
surface in the deviatoric plane. However, the plastic potential still gave a circle in the
deviatoric plane and therefore plane strain failure occurred at 2=0o, as for the first analysis.
In both analysis the initial stress conditions in the soil were based on a saturated bulk unit
weight of 18kN/m3, a ground water table at a depth of 2.5m and a Ko = 1.227. Above the

ground water table the soil was assumed


to be saturated and able to sustain pore
water pressure suctions. Coupled
consolidation analyses were performed
but the permeability and time steps
were chosen such that undrained
conditions occurred. Loading of the
footing was simulated by imposing
increments of vertical displacement.
In summary, the input to both analysis
Figure 9: FE mesh for footing analysis
is identical, accept that in the first, the
strength parameter MJ is specified,
whereas in the second, NN is input. In
both analyses NNTC=23o and therefore
any analyses in triaxial compression
would give identical results. However,
the strip footing problem is plane strain
and therefore differences are expected.
The resulting load displacement curves
are given in Figure 10. The analysis
with a constant MJ gave a collapse load
some 58% larger than the analysis with
a constant NN. The implications for
practice are clear, if a user is not aware
of the plastic potential problem and/or
is not fully conversant with the Figure 10: Load-displacement curves for two
different approaches
constitutive model implemented in the
software being used, he/she could easily
base the input on NNTC=23o. If the model
uses a constant MJ formulation, this would then imply a NNPS=31.2o, which in turn leads to
a large error in the prediction of any collapse load.

4 Using modified Cam clay in undrained analysis


As noted above, the input parameters to the modified Cam clay model are based on drained
soil behaviour and do not involve the undrained shear strength, Su. Consequently, undrained
analyses can be problematic. For example, if constructing an embankment or foundation

on soft clay, short term undrained conditions are likely to be critical from a stability point
of view. It is therefore important for any analysis to accurately reproduce the undrained
strength that is available. It is also likely that establishing the undrained strength profile
would be a priority of any site investigation.
Although the undrained strength is not
one of the input parameters to the
constitutive model, it can be calculated
from the input parameters and the initial
state of stress in the ground, as shown
in Equation 13. Consequently, if the
undrained strength profile is known, it
is possible to use Equation 13 to back
calculate one of either OCR, KoOC or
6/8. For example, if the undrained
strength profile at a site resembles that
shown in Figure 11b, which is typical
of a soft clay deposit, it is possible to Figure 11: Variation of a) overconsolidation
set all parameters, except the OCR, and
ratio and b) undrained strength for soft clay
then to use Equation 13 to calculate the
distribution of OCR which is consistent
with the required Su profile. Such a distribution of OCR is given in Figure 11a. Clearly it
will be necessary for the finite element software to be flexible enough to allow the user to
input such a variation of OCR.
It should be noted that for modified Cam-clay the undrained strength, Su, is linearly related
to the vertical effective stress, FNvi. Consequently, if FNvi=0, then so will the undrained
strength. This explains why it is necessary for the OCR to increase rapidly near to the
ground surface in Figure 11a. However, even if FNvi=0 at the ground surface (i.e. no pore
water suctions present) it is still possible to perform finite element analysis which simulate
a finite undrained strength at the surface. This is possible because the constitutive model
is only evaluated at the integration points which lie a finite distance below the ground
surface.

5. Conclusions
Use of modified Cam clay model in advanced numerical analysis can be problematic. The
model was originally developed for triaxial stress and strain conditions and therefore must

be extended into generalised stress and strain space for use in numerical analysis. At
present there is no universally accepted way of performing this extension and consequently
there are many different forms of the model implemented in the various available computer
codes. In many of these cases the finer details of the model are not documented.
Consequently many potential errors are associated with a users lack of in depth
understanding of the constitutive model being employed.
Two examples of such errors have been discussed in this paper. The first example
considered the influence of the shape of the yield and plastic potential surfaces in the
deviatoric plane. Many different options have been described in the literature and this is one
of the most uncertain areas of the model. However, it has been shown that this shape can
have a dominant effect on both the predicted drained and undrained strengths of the soil.
In the strip footing example described in this paper an increase in the failure load of some
58% can be attributed to just changing the shape of the yield surface from a Mohr-Coulomb
hexagon to a circle. Such results clearly show that it is imperative that the user has an in
depth understanding of the constititutive model being used.
The second example considered the potential problems associated with performing
undrained analysis. Unfortunately the undrained strength of the soil is not an input
parameter to the modified Cam clay model. It can, however, be calculated from the initial
stress conditions in the soil and the model parameters. Knowing the distribution of
undrained strength in the soil profile it is possible to calculate the distribution of either
OCR or Ko which will give rise to this strength distribution. Care must be taken near to the
ground surface because if the effective stress approaches zero then so does the undrained
strength. Again it is critical that a user understands the implications of the model.

6. References
[1]
[2]

[3]
[4]
[5]

Potts D.M. and Zdravkovic L., 1999, Finite element analysis in geotechnical
engineering: theory, Thomas Telford, London
Coulomb C.A., 1776, Essai sur une application des regles de maxims et minims a
quelques problemes de statique, relatifs a larchitecture, Mem. Acad. Royal Soc., 7,
343-382
Rankine W.J.M., 1857, On the stability of loose earth, Phil. Trans. Royal Soc., 147,
9-27
Drucker D.C. and Prager W., 1952, Soil mechanics and plastic analysis of limit
design, Q. Appl. Math., 10, 157-167
Roscoe K.H. and Schofield A.N., 1963, Mechanical behaviour of an idealised wet

[6]
[7]
[8]
[9]
[10]
[11]

clay, 2nd ECSMFE, Wiesbaden, 1, 47-54


Schofield A.N. and Wroth C.P., 1968, Critical state soil mechanics, McGraw Hill,
London
Roscoe K.H. and Burland J.B., 1968, On the generalised stress-strain behaviour of
wet clay, Eng. plasticity, Cambridge Univ. Press, 535-609
Matsuoka H. and Nakai T., 1974, Stress-deformation and strength characteristics
of soil under three different principal stresses, Proc. Jap. Soc. Civ. Eng., 232, 59-70
Lade P.V. and Duncan J.M., 1975, Elasto-plastic stress-strain theory for
cohesionless soil, ASCE, GT Div., 101, 1037-1053
Eekelen H.A.M. van, 1980, Isotropic yield surfaces in three dimensions for use in
soil mechanics, Int. Jnl. Num. Anal. Meth. Geomech., 4, 89-101
Potts D.M. and Gens A., 1984, The effect of the plastic potential in boundary value
problems involving plane strain deformation, Int. Jnl. Num. Anal. Meth. Geomech.,
8, 259-286

You might also like