Stress Strain Behaviour of Soft Clays
Section A: Critical state stress-strain theories and
behavior of soft clays
2.1 Introduction
One of the main objectives of this research is to develop appropriate stress-strain
models to be used in ground improvement works of soft clays with varying organic
contents. As such a critical review of the existing theories based on elasto-plastic
component forms an important part of this report. key publications and contributions,
which led to the development of the critical state theories, form the essential part of
this chapter.
The other object of this report is to present a systematic procedure for establishing a
hierarchy of models, starting from the familiar Cam clay models, which include
internal friction, non-associated flow rules and a variety of biased volumetric/shear
hardening laws. These models have the advantage of automatically satisfying the
second law of thermodynamics. This is not as trivial an exercise as may at first be
thought, since, as will be seen, the well-known original Cam clay model actually
violates the second law, even though it satisfies Drucker’s postulate. It will also be
shown that some models with significant amounts of internal friction actually have
yield surfaces parts of which are concave, a property that violates Drucker’s postulate
but not the second law. The classical experimental contributions related to the
behaviour of normally and overconsolidated states of soft clays are also examined in
the latter half of the chapter.
With the use of several types of computer softwares in Geotechnical Engineering
Practice, it is important to have a good command on the basic stress-strain behaviour
of soils and their strength characteristics. Critical state soil mechanics involves the
study and modelling of the mechanical behaviour of soils. It covers nearly all aspects
of the behaviour of most engineering soils. In this section the original contributions
made in relation to the development of the Cambridge stress strain theories will first
be made and then the theories will be briefly derived. Sections will then follow this on
the actual behaviour of soft clays as studied by other notable researchers.
The stress strain characteristics of soils were studied in detail since 1950 and the
behaviour was modelled in a rational manner using concepts, which are well founded
from experimental observations. This classical work led to the development of a state
boundary surface in the (p, q, e) space (Roscoe et al., (1958)). The state boundary
surface incorporates all states of a particular soil from isotropic consolidation to the
failure state both under compression and extension conditions under a wide variety of
applied stress paths in undrained and drained conditions.
Using such a state boundary surface it was possible to calculate the volumetric
strains in normally consolidated clays. The over consolidated states lie within the state
boundary surface and are found to experience both plastic volumetric and shear
strains of a magnitude smaller than the normally consolidated states. In addition,
using such a state boundary surface it was possible to calculate the volumetric strain
in drained shear and the pore pressure development during undrained shear. An
incremental stress strain theory was proposed originally by Poorooshasb (1961) to
model the strains in drained conditions using the data from undrained conditions and
isotropic and anisotropic consolidation. Poorooshasb (1961) illustrated how the
dilatancy ratio in normally consolidated clays can be expressed as a function of the
stress ratio, q / p . This dilatancy ratio was the maximum under isotropic
consolidation and gradually reduced to zero, when the critical state was reached.
Using an approach similar to the boundary energy correction of Taylor (1948) and
Rowe (1957), Roscoe et al. (1963) proposed an energy balance equation that
described the plastic dilatancy ratio in soils with an associated type of flow rule. The
plasticity concepts were able to capture most of the dilatants behaviour of various
types of soils. Later, Drucker et al. (1957) proposed a normality condition that offered
an alternate method by which the plastic dilatancy ratio can be determined from the
yield locus with an associated flow rule for volumetric yielding in soils. Drucker et al.
(1957) have provided a successful framework to investigate the stress-strain and
strength of soil. Caladine (1963) suggested the method from which a volumetric yield
locus can be established using the state boundary surface and the elastic swelling
under isotropic conditions. These studies formed the corner stones for the
establishment of the Critical state theories (Roscoe et al. (1963); Roscoe and Burland
(1968)).
The critical state models were successfully used first to model the behaviour of
Bangkok clays under a variety of applied stress paths (Balasubramaniam and Hwang,
(1980); Balasubramaniam et al., (1993); Balasubramaniam et al., (1999)). The CRISP
program was then used to make successful predictions of a large number of test
embankments and excavations with and without ground improvement in Thailand and
Malaysia (Bergado et al., (1996); Bergado et al., (1998); Bergado et al., (1999); Chai
et al. (1995); Indraratna et al. (1992)).
There are a variety of ways of constructing mathematical models for examining the
mechanical behaviour of clay. One is to curve-fit experimental data and formulates
the basic equations needed to describe the model in terms of these empirically derived
functions. However, even in this empirical approach, it is necessary to have some
underlying theoretical framework that enables the model to have the necessary
predictive qualities. The well-developed theory of rate-independent elastic or plastic
materials is by far the most popular such background framework currently in use, and
the theory developed by Lade and co-workers is one of the best known of such
empirically based models (Lade, 1977). The second approach in modelling is to start
with some specific theoretical assumptions, such as a dilatancy law, an expression for
‘plastic work’ or ‘energy dissipation’, as in the original Cam clay model (Schofield
and Wroth, 1968; Gens and Potts, 1988; Wood, 1990), or to make some other micro-
mechanical assumptions, as in the recently developed models involving particle
crushing by McDowell and Bolton (1998) and McDowell (2000). These models (Cam
clay and critical-state models) and the vast hierarchy of extensions are excellent.
Although a large number of models are developed based on critical state soil
mechanics, a very limited success has been achieved to simulate the real soil
behaviours. It was soon realised from the extensive field experience of soft clays in
Southeast Asia that successful theories based on the critical state concept must be able
to model accurately.
1 The strains in natural clays which are some-what over consolidated but undergo
plastic strains within and without the state boundary surface
2 The small strains experienced in excavation problems under undrained
conditions
3 Undrained creep at higher stress ratios approaching the critical state
4 The continuous creep and secondary settlements in peaty soft clays with high
organic content, where surcharging techniques are used to erase the long-term
settlements under secondary consolidation.
This research will be concentrating on these aspects of the modelling of soft clay
behaviour. With these objectives in mind, a critical review of the available literature in
these areas will first be made and from these insights, a proposal of work both
theoretical and experimental needs to be carefully formulated in the latter chapter. The
elasto-plastic fundamentals will be described and reviewed in the next section on Soil
Plasticity.
2.2 Soil Plasticity
2.2.1 Yield Surface
A yield surface is defined as a surface in the stress space within which elastic strains
take place and for stress states on the surface or outside of it plastic deformations
occur. In general, the function for yield surface is expressed in terms of Stress State,
the plastic strain and the hardening parameter, i.e.
f = F (σ ij , ε ij , k ) (2.1)
The derivative of this function by stress components is used to judge the loading
conditions.
∂f ∂f ∂f (2.2)
f& = σ& ij + ε& ij ( p ) + k&
∂σ ij ∂σ ij
( p)
∂k
2.2.2 Flow Rule
Flow rule specifies how the plastic incremental strain develops by a next incremental
stress when the present stress condition is on the yield surface, and the next
incremental stress is loading. The relationship can be established from the concept of
plastic potential function g as
dε ij = dλ∂g / ∂σdf (2.3)
( p)
Where dλ is a positive scalar of proportionality dependent on the state of stress and
stress history. If f=g, the flow rule is called associated flow rule otherwise it is called
non- associated flow rule.
2.2.3 Hardening Rule
Hardening rule describes the expansion of yield surface. The strain-hardening rule is a
relationship between the position of the yield surface in the stress space and the
plastic strains experienced by the material in arriving the specific stress point under
consideration. A material is said to be work hardening or strain hardening if the
following two conditions hold true.
dσ ij dε ij > 0 Upon loading (2.4)
dσ ij (dε ij − dε ij( e ) ) = dσ ij dε ijp > 0 Upon load removal (2.5)
2.2.4 Development of Equation
In the development of an incremental elasto-plastic stress strain relationship, the total
strain increment is assumed to be the sum of the elastic strain component dεije and the
plastic strain component dεijp.
dεij = dεije +dεijp (2.6)
The next section is fully devoted to the derivation of the Cambridge stress –strain
theories
2.3 Review of Critical State Concepts
2.3.1 State Boundary Surface
The State Boundary Surface is a unique surface in the three dimensional stress space,
which correlate the void ratio (e), and the stress parameters p and q. The unique
characteristic of the SBS is that it separates the states of an element of soil from those
that are not admissible. This surface is formed by two surfaces namely the Roscoe
surface- the surface on which volumetric yielding takes place and is defined by the
undrained stress path of Normally consolidated clay; and the Hvorslev surface- the
locus of the failure points of highly overconsolidated soils. The existence of such a
unique State Boundary Surface have been verified by many investigators (Parry 1960;
Wroth and Loudon, 1967; Balasubramaniam, 1969)
2.3.2 Critical State Line
The end points of all specimens, when they are sheared to failure lie on a unique line
defined as the Critical State Line. This line separates the Roscoe surface and Hvorslev
(Lade, 1979) failure surface. When the state of sample reaches the critical state,
unlimited distortion occurs with no further change in effective stress or volume. The
projection of the Critical State Line on (e, ln p) plot is a straight line, which is parallel
to the virgin consolidation line. The area between the Critical State Line and the
normally consolidated line in (e, ln p) plot is called wet of critical, while the area to
the left of the critical state line is called dry of critical. The soil specimens when
sheared in the wet zone develop a positive pore water pressure under undrained
condition and reduce in volume if sheared under drained condition. On the other hand,
when the soil specimen is sheared in the dry zone, its volume would increase under
drained condition and negative pore water pressure would develop under undrained
condition.
2.4 Review of the Stress Strain Theories Based on
Critical State Concepts
2.4.1 Incremental Stress-Strain Theory
Assuming a unique (p, q, e) surface, an incremental stress strain theory was proposed
by Roscoe and Poorooshasb (1963), in the form:
⎛ dε ⎞ ⎛ dε ⎞ (2.7)
dε 1 = ⎜⎜ 1 ⎟⎟ dη + ⎜ 1 ⎟ dε v
⎝ dη ⎠ v ⎝ dv ⎠ η
dε 1
Where, ( ) v corresponds to the variation of ε1 with η in an undrained test and
dη
dε 1
( ) η represents the variation of ε1 with v in a constant η stress path. The total strain
dv
experienced at any instant for any state path on the State Boundary Surface is
calculated by integrating the equation.
2.4.2 Cam Clay Theory
The Cam Clay Theory (Roscoe et al. 1963) was developed for normally consolidated
and slightly overconsolidated soils assume that only plastic shear strain contributes to
the energy dissipation.
p dεv p+q dεσ p =Mp dεs p (2.8)
The Flow rule is
dε sp 1 (2.9)
=( )
dε vp M −η
The yield locus is
q= M p ln(po /p) (2.10)
The Strain Components are
λ −k 1 dq q dp (2.11)
dε vp = ( )[1 + ( − )]
1+ e M dp p p
λ −k 1 1 dq q dp (2.12)
dε sp = dε s = ( )( )[1 + ( − )]
1+ e M −η M dp p p
State Boundary Surface is
λM p (2.13)
η=( ln( o )
λ − k) p
The volumetric yield locus considered in this theory is of bullet shaped about p-axis.
2.4.3 Modified Cam Clay Theory
The Modified Cam Clay Theory (Roscoe and Burland, 1968) differs from the original
Cam Clay Theory in the formulation of energy balance equation, which consider both
the plastic volumetric strain and plastic distortional strain for the determination of
energy dissipation as:
p
(δW) Dissipated = p (δε v ) 2 + ( Mδε s ) 2
p
(2.14)
The Flow rule is
dε vp M 2 −η 2
= (2.15)
dε sp 2η
The yield locus is
p M2
=( 2 ) (2.16)
po M +η 2
The Strain Components are
λ −κ 2ηdη dp
dε vp = ( ){( )+ } (2.17)
1+ e M +η
2 2
p
λ −κ 2η dp 2η
dε sp = dε s = ( )( )[ + ( 2 ) dη ] (2.18)
1+ e M −η
2 2
p M +η 2
The equation of State Boundary Surface is
κ
p M2 (1− )
( )=( 2 ) λ (2.19)
po M +η 2
The Volumetric yield locus is elliptical in shape in this model whereas in Cam Clay
the volumetric yield locus is log spiral in type shape.
2.4.4 Revised Theory
Roscoe and Burland (1968) proposed revised flow rule and the corresponding theory
as revised theory. In this theory the contribution of the undrained shear strain can be
expressed as:
dε sp
dε q = ( ) p dη (2.20)
dη v
Where, v p implies plastic volumetric strain is constant. Hence the incremental plastic
shear strain in the revised theory is:
dε sp dε p
dε sp = ( ) v p dη + ( sp )η dε vp (2.21)
dη dε v
Pender (1978) developed a stress-strain theory, which can contribute both plastic
volumetric and distortional strains below the State Boundary Surface. His
contribution is mainly within the critical state concepts but employed a non-associated
flow rule. By carefully analysis the experimental observations available on
overconsolidated clays, Pender made a series of hypothesis in his theory.
Overconsolidated soil experience plastic strains when there is a change in stress
ratio. Constant stress ratio yield locus is given by the function “f”.
f = q −ηi p = 0 (2.22)
∴ df = dq − η i dp = pdη (2.23)
The undrained stress paths are parabolic in (p, q) plane
PO
(1 −
)
η − η o 2 PCS P
( ) = [ ] (2.24)
AM − η O P PO
(1 − )
PCS
Where, A= -1 for extension and +1 for compression
The ratio of the plastic shear strain increment to plastic volumetric strain increment
is given by:
dε sp P P
( p ) = ( AM − η O ) 2 /( AM ) 2 {( O ) − 1}{( AM − η O ) − (η − η O ) } (2.25)
dε v PCS PCS
Hence
2κ ( p / p cs )(η − η o )dη
dε sp = (2.26)
( AM ) (1 + e)(2 p o / p − 1)[( AM − η O ) − (η − η O )( p / p cs )]
2
2κ ( PO / PCS − 1)( p / PCS )(η − η O )dη
dε VP = (2.27)
( AM − η ) 2 (1 + e)(2 p o / p − 1)
Appendix 1 contains the derivation and results of other models, rather complicated
but making use of the ideas in critical state soil mechanics.
2.5 Critical comments and concluding remarks on
stress-strain models
A basic requirement of the successful models is that they satisfy the basic laws of
physics. For example, any equations describing the properties of isotropic materials
must be expressed in terms of the invariants of stress, strain, strain increment, etc. The
second law of thermodynamics is one of these basic laws that govern the dissipative
behaviour of materials, but it is seldom invoked in geomechanical theories. Until
recently thermodynamics and ‘heat flow’ were seldom seen as relevant to
geotechnical problems. A noticeable exception is the overview paper by Mitchell
(1991). However, this situation is now changing, particularly as a result of the
development of geo-environmental engineering (Smith, 2000). During the early days
of the development of the theory of elastic or plastic materials, quasi-thermodynamic
postulates were introduced in an effort to ensure that the dissipation of energy was
always positive in a closed cycle of stress (Drucker’s postulate) or strain (Iliushin’s
postulate). However, it was soon realised that these postulates were actually
classifications of types of material behaviour and not, in any sense, equivalent to the
second law of thermodynamics (Drucker, 1988; Lubliner, 1990).
In the last 20 years, however, there have been major developments in the theory of
thermo-mechanics of continua, which have thrown new light on many long-standing
issues in a number of application areas. In the case of isothermal deformations of rate-
independent elasto-plastic solids, the new theories have shown that, under certain
minimal assumptions, knowledge of the material’s free energy and dissipation rate
potentials is sufficient to uniquely determine the elasticity law, the yield function, the
flow rule and the hardening rules (both isotropic and kinematic) of the material. The
simultaneous determination of both the yield function and the plastic potential is
particularly significant, since, in geomechanics, these two functions are normally
chosen independently. (A notable exception to this is the modeling process used by
Chandler (1985, 1988) and Chandler and Song (1990), whose approach using the
theory of envelopes has several similarities to the more general thermomechanical
procedure.)
The early developments of these general ideas were due to Ziegler (1983) and
Ziegler and Wehrli (1987), who noted some applications of their general theory to the
classical Coulomb model. A rigorous general theory was then developed by a number
of researchers in France, accounts of which can be found in the books by Maugin
(1992, 1999), Maugin et al (2000), Besseling and van der Giessen (1994) and
Lemaitre and Chaboche (1990). Applications of these ideas to soil mechanics were
pioneered by Houlsby (1981, 1982, 1993). A comprehensive analysis of the
isothermal thermomechanics of geomaterials was given by Collins and Houlsby
(1997), who demonstrated that a non-associated flow rule is a necessary property of a
‘frictional material’, in which the plastic deformations are governed by stress ratios
rather than by the magnitudes of certain yield stresses. Houlsby and Puzrin (2000)
generalised some aspects of this work to non-isothermal conditions, and in a series of
papers have developed a family of sophisticated models, based on the use of internal
functions (Puzrin and Houlsby, 2001). These authors have proposed the term
‘hyperplasticity’ to describe theories in which the irreversible plastic behaviour is
determined by a dissipation potential function, by analogy with the well-established
term ‘hyperelasticity’ used in the analogous situation in elasticity theory.
In spite of these extensive developments, the critical state theories when understood
and used in a proper manner with the full knowledge of their limitations will indeed
make excellent predictions of strains in drained behaviour together with pore pressure
predictions in undrained conditions. The most outstanding issue to be resolved is how
to extent such models for the behaviour in the over consolidated state. Pender’s
approach though empirical is the closest to the realistic behaviour both in the
laboratory tests as well as in the field states. Because these theories are based on
conceptual models of macro or micro behaviour, such models are capable of giving
deep physical insights into the engineering behaviour of soils, although quite
frequently at the expense of detailed numerical accuracy. The pedagogical merits of
this approach are elegantly argued in Wood (2000). The choice of complexity of the
model depends on the practical purpose underlying the model construction. There are
now some very complex models available, involving large numbers of material
parameters, but as pointed out by Kolymbas (2000), such models are seldom
‘transportable’ from one research group to another.
Although the basic philosophy of Cam Clay and Modified Cam Clay and Revised
Theory, presented above is same, but the Revised theory of Roscoe and Burland
(1968) is found to be more successful in predicting the undrained shear strength, the
excess pore pressure and shear strain for isotropic normally consolidated samples. As
these theories are developed for normally consolidated clay, but almost all the natural
deposits of soft clay are lightly overconsolidated due to weathering and aging effects.
Also most of the clays are under anisotropic condition to some extent. These
conditions are looked as the limitations to use these theories. Hence, there is a need to
develop a theory, which can consider these aspects equally. It is found in literature
that the contribution of Sekiguchi and Ohta (1977) and Dafalias (1987) seems
reasonable to model the anisotropic behavior, but its use is limited to simulate the
overconsolidated behavior. On the other hand, Pender’s model seems valuable to
model most of the natural soil behaviour. But it is found in literature that the use of
hypothesis, especially regarding undrained stress path and plastic dilatancy ratio,
which does not represent the real soil behaviour, limits its use to model the real soil
behaviour.
The next section would be fully devoted to the empirical and experimental
contributions by notable authors on the stress strain behaviour of soft clays and other
related soils.