I

162 INTRODUCTION TO SOIL MECHANICS AND FOUNDATIONS GENERAL SOIL MODELS

7.15 Pore pressure changes in undrained tests: The pore pressure

changes during undrained tests may be estimated by
comparing the total and effective stress paths, as described in
Section 7.4 above. Consider a specimen, tested at a cell
pressure p,, and loaded to failure in undrained triaxial
compression. Any increment of axial total stress Aoa results
in an increase of Aca in q and f Ac, inp. The total stress path
is therefore inclined at 3 to 1 to the p axis.
(M-

Fig. 7.16 Unified plot of the state boundary surface.

For normally consolidated and ligl~tlyoverconsolidated

specimens, the pore pressure increases during the test, but in
heavily overconsolidated specimens uf - u, is negative.

7.16 A unijied plot of the state boundary surface: Figure 7.12 shows
sections on planes of constant v through the state boundary
- surface. All such sections are of the same shape, although of
different sizes. The whole state boundary surface may
therefore be displayed by a single unified plot of p'/pl, and
o v e r consolidated+- normally consolidated q/pt, where p', = exp [(N - v)/A]. This unified plot is shown
in Fig. 7.16. The normal consolidation line is represented by
Fig. 7.15 Pore pressure changes in undrained tests. -Total stress paths. --- Effective
stress paths. (After Atkinson and Bransby [7.8].) the single point B where pt/p', = 1-0 and q/pl, = 0. The
critical state is represented by the single point C, and the
equation of the state boundary surface in the dry region is
Figure 7.15 shows the stress paths for two specimens, one
normally consolidated and the other heavily over-
consolidated. In each case, failure occurs at the critical state,
-4 --
p'e
- "MO + (M
~e
-~ , ) e x rp-( v~ ) / ,eP
so that
But
Pf =Po +f4,
+
= p o fMptf p', = exp(?) and p,/pt = R,
and
v =r -~ i ~ g , ~ ~ , Hence
so that

In the 'wet' region, a number of shapes for the state boundary

Then surface have been suggested. Of these, the most generally
useful seems to be the elliptical surface used in the 'modified
cam clay' model (Roscoe and Burland, [7.14]). The state
boundary surface is defined by the expression

= po - (1 - f M) exp
(' n .)
---
 INTRODUCTION TO SOIL MECHANICS AND FOUNDATIONS GENERAL SOIL MODELS

so that

The unified Rendulic diagram (Fig. 7.8) and the unified

critical state diagram (Fig. 7.1 6) describe essentially the same
behaviour. In each case the diagram shows

(a) a stable yield locus BC in the wet region,

(b) an unstable yield (failure) locus AB in the dry
region, and
(c) a no-tension condition OA.
Additional features of the critical state model are
(a) the concept of a critical state which lightly
consolidated soils reach at failure, and to which
heavily overconsolidated soils revert after failure,
and
(b) mathematical expressions defining the yield con-
ditions and the critical state.

7.17 Yield, failure and the critical state in principal stress space: We
showed in Section 7.9 that we are only entitled to ignore the
effects of the third stress invariant 8 for so long as we limit our
investigation to triaxial compression tests, in which
a', 2 a',. We must now see if the model can be extended to
cover cases in which the two lesser principal stress
components are not equal. We will start by transferring the
unified state boundary surface from Fig. 7.16 to the unified
Rendulic diagram (Fig. 7.17(a)). For the conditions of the
triaxial test, 8 = 30 ",and the plot lies in the upper part of the
plane OCBD, above the space diagonal. For other values of
8, the yield condition, which forms the state boundary
surface in the 'dry' region, is represented by a cone, not
necessarily circular in section, around the space diagonal.
The figure is closed by a cap, again not necessarily circular in
section, representing the yield condition in the 'wet' region.
The general forms of the two yield conditions, and the critical
state may be studied by examining their intersections with
octahedral planes such as ElE,E3 (Fig. 7.2). Notice that,
since we have not stipulated which principal stress
componen t is greatest, the sections must be symmetrical
about El F,, E2F2, and E, F3. They are therefore fully
 INTRODUaION TO SOIL MECHANICS AND FOUNDATIONS GENERAL SOIL MODELS 167

~l't This implies that the yield condition is not affected

by the intermediate principal stress component. It
condi t ion may be shown that the surface defined by such a
condition intersects the octahedral plane in an
irregular hexagon, as shown in Fig. 7.18.
y i e l d cond
7.19 General yield, failure and critical state conditions for real
soils: There are considerable practical difficulties in the
way of conducting tests in which all three principal stress
components are to be controlled independently, and the
evidence about the behaviour of real soils subjected to
general stress changes is therefore far from conclusive.
However, the following points are fairly well established (see
for example references [7.11] to i7.151).
(a) The shear strength of real soils is generally
predicted from the results of triaxial compression
tests. Figure 7.18 shows the failure conditions
which would be predicted from the result of such a
Fig. 7.18 Yield conditions in the octahedral plane. test (represented by point A ) , assuming
(i) a von Mises, and
described by values of 8 between 30" and - 30" (i.e. by the (ii) a Mohr-Coulomb failure condition.
sector El BF3). The evidence seems to show that the failure
condition for most real soils lies somewhere
7.18 The extended von Mises and generalised Mohr-Coulomb models: between the two.
Of the many general soil models which have been suggested, (b) A triaxial compression test interpreted in terms of
only two will be examined here. the Mohr-Coulomb failure condition will slightly
under-estimate the real strength in plane strain, but
(a) Drucker et al. [7.10] suggested an extended form of may predict the strength in triaxial extension
von Mises criterion, in which yield is assumed to almost exactly.
occur when (c) A triaxial compression test interpreted in terms of
roct =f (a'oct) the von Mises failure condition may seriously over-
This implies that roc,is a function of atoctonly, and estimate the strength in plane strain.
that the third invariant 8 does not affect the yield (d) When the intermediate principal stress component
condition. Since a',,, is a constant on any is small (i.e. when 8 is nearly 30°), the real failure
condition may be nearer to the von Mises than to
octahedral plane, and BP = fir,,,, the intersec- the Mohr-Coulomb model.
tion of this yield surface with the octahedral plane (e) Generally similar behaviour is observed at the
is a circle (Fig. 7.18). In Sections 7.9 to 7.16, the critical state.
critical state model was developed without refer- (f) Near the tip of the cap (where the principal stress
ence to 8. If we now assume that the model is components are nearly equal), the yield condition
generally applicable in the form developed above, may be nearer to the von Mises model.
we imply that the yield, failure, and critical state
conditions are of the von Mises type. Although there is nothing unacceptable in the critical state
(b) Alternatively, we might assume that the conditions theory as such, the model is clearly not generally applicable
were of the Mohr-Coulomb type, such that to all possible stress states in the form in which it was derived
in Sections 7.9 to 7.15 above. Of course, there is no reason
why we should not construct a critical state model in which
 INTRODUCTION TO SOIL MECHANICS AND FOUNDATIONS GEP4ERAL SOIL MODELS 169
the yield conditions are defined in terms of all three stress (a) Orientation of the soil particles. Clay particles are
invariants, and such models have been proposed 17.151. flat, and their orientation and mode of assembly
However, the complexity of such models and the difficulty of affects their behaviour under load. Anisotropic
evaluating the parameters, often make this approach consolidation tends to produce a preferred orien-
unproductive. It is usually simpler and more efficient to use tation of the particles normal to the direction of the
either Mohr-Coulomb conditions (which always err on the greatest component of principal stress. Large
safe side) or von Mises conditions with the parameters strains, however, cause the particles to be re-
suitably reduced. The actual choice will depend on the arranged parallel to the direction of sliding. This
purpose for which the model is required. The choice of results in a significant drop in strength.
suitable models is discussed by Zienkiewicz and Pande [7.16] (b) Physico-chemical bonds. These develop gradually
and Naylor 17.171. in the adsorbed layers surrounding the particles in
contact. If the bonds are broken by deformation of
7.20 Applications of the critical state model: Critical state theory the soil, they do not reform immediately, and the
provides an excellent qualitative model of soil behaviour, by shear strength is correspondingly reduced.
bringing together a number of apparently unconnected (c) Intergranular cement. In some soils (see Section
concepts, such as those of Casagrande, Rendulic, Coulomb, 1.6), the particles are cemented with iron oxide,
and Hvorslev, and the critical state concept itself. In calcite, etc. deposited at and around the particle
addition, some simple quantitative problems may be solved contacts. If this cement is broken up by relative
directly. For example, the prediction of pore pressure movement of the particles, the shear strength is
changes is often better done by the stress path techniques permanently reduced.
described in Section 7.15 than by the use of Skempton's pore
pressure parameters, which are very sensitive to stress levels Undrained strength of saturated cohesive soils: The undrained
E7.181. strength of a saturated soil (Section 7.14) is unaffected by
The most promising quantitative application of the model changes in the total normal stress. Then
is as a basis for elasto-plastic stress-strain laws for use with
the finite element method. At the moment, the com- zJ = cu and cp, = 0
putational effort required for this technique restricts its use where c, and cpU are Coulomb's shear strength parameters for
to the research field, but the method appears to have the undrained condition in terms of total stress.
considerable potential 17.171. In overconsolidated soils, however, the pore pressure
For ease and simplicity, most routine problems are still drops sharply as the specimen approaches the critical state.
solved by using simple elastic theory and consolidation In this condition, the strength is slightly less than at failure,
theory to predict displacements, and the classical as a result of sample disturbance. Local soil failure then
Mohr-Coulomb model to predict collapse loads. occurs on quite narrow bands of softened soil, in which the
Nevertheless, it is important to realise the limitations of these pore pressure is less than in the adjacent material. This causes
models when applying them to predict the behaviour of real fairly rapid local migration of water towards these bands,
soils. leading to a rise in pore pressure and a loss of strength. Such
soils, therefore, may not remain truly undrained for long
[7.19].
SHEAR STRENGTH O F REAL SOIL
Variation of the undrained shear strength with depth: The
7.21 The maximum strength of natural cohesive soil: In developing the undrained shear strength (c,) of a normally consolidated
soil models, we considered only remoulded soils. The clay is almost directly proportional to the consolidation
maximum strength of a cohesive soil in its natural pressure (o',).Skempton 17.201 has suggested the following
undisturbed state is commonly significantly greater than that empirical relationship between the index of plasticity (PI)
of the same soil after remoulding. This is because certain and the ratio cU/o1.(See Fig. 7.19):
features which contribute to the shear strength of the soil are
destroyed during remoulding. These features are as follows: <
C

o
= 0.1 1 + 0.003 7 (PI)
 INTRODUaION TO SOIL MECHANICS AND FOUNDATIONS
GENERAL SOIL MODELS 17

09

08

0-7

,a 0.6
0
*-
0.5
a
-
3 0.4
p = 0 il+ 0.0037 (PI.)
0.3 = sheor strength at a
depth where effective
overburden pressure
0.2

0.1

0 0
Plasticity index

Fig. 7.19 Skempton's relationship between undrained shear strength, consolidation pressure
and the index ofplasticityl7.20].

If the ground water level is at the ground surface, o' (and

therefore c,) is often directly proportional to depth, as shown
by the solid line in Fig. 7.20(a).
Figure 7.20(b) shows a typical relationship between shear
strength for a heavily overconsolidated clay, for which the
preconsolidation pressure o', is much greater than the Fig. 7.20 Typical strengthldepth profiles for ( a ) normally consolidated clay [7.21], and
present effective overburden pressure. The effects of ( b ) over consolidated clay [7.22].
weathering and seasonal water content changes near the
surface have curved both lines. Close to the surface of an
otherwise normally consolidated clay, desiccation in dry
weather may cause a large suction (that is, a large negative identical. In an undisturbed natural soil, however, there is
pressure). This results in an increase in effective stress, and usually some loss of strength after failure due to disruption
produces the effect of overconsolidation, even though the of physico-chemical bonds and interparticular cement. The
present overburden pressure has never been exceeded. This is strength of a natural normally consolidated soil at the critical
reflected in the increased strength of the drying crust near the state is therefore generally somewhat below the maximum.
surface in Fig. 7.20(a). Below this crust, there is often a In overconsolidated soils, the lower void ratio is reflected
considerable layer in which the strength has been increased in a higher strength at failure, and
by physico-chemical effects. rs = C' + o', tan q'
7.24 Drained strength of clay soils: The maximum drained strength of a where q' is less than for the same soil when normally
normally consolidated soil may generally be expressed in consolidated, but c' is greater than zero and is a function of
terms of effective stress in the form the preconsolidation pressure. A typical relationship be-
tween the maximum strengths of a soil in the normally and
T, = o', tan q' so that c' =0 overconsolidated conditions is shown in Fig. 7.21. A
For a remoulded and normally consolidated soil, the comparison with Fig. 7.9(a) shows that this is in good
agreement with the behaviour of Hvorslev's model.
strength at failure and at the critical state are almost
 INTRODUaION TO SOIL MECHANICS AND FOUNDATIONS GENERAL SOIL MODELS 173

In heavily overconsolidated soils, there is a substantial 7.25 Shear strength of clay soils after large deformations: In some
drop in strength between failure and the critical state (see overconsolidated clays (particul,arly those without fissures),
Section 7.13 above). In assessing the collapse load of a soil the strength computed from observations of the collapse of
structure, we need to know the average strength within the natural and excavated slopes has been near the measured
zone of plastic yield. Where the strains before collapse are peak strength of the undisturbed material. However, in some
nearly uniform, most of the soil will reach the maximum overconsolidated fissured clays (notably the London clay)
strength at the same moment. The average strength will be there has evidently been a gradual loss of strength before
collapse, extending over many years, and tending towards
the critical state value. This is presumed to be the effect of the
fissures causing progressive failure. In such soils, the critical
state strength is clearly the appropriate value to use in design.
As explained above, this critical state strength is somewhat
less than the maximum strength of the normally con-
solidated soil. Skempton [7.24] suggests that, as a con-
servative estimate, we may take the peak value of q',
assuming c' to be zero.
In previous sections, it has been implied that the critical
state and residual strengths are the same, but in real soils this
is not entirely true. In overconsolidated soil, collapse usually
results in movements of substantial bodies of soil, by sliding
along relatively thin deforming layers within the soil mass.
Displacements of the order of 100 mm along such layers are
generally sufficient to reduce most clays to the critical state,
after which deformation continues at nearly constant
volume. However, further displacements-of the order of
1 m or more-may reduce the strength still further.
Skempton [7.23] originally made tests on London clay using
a reversing shear box (see Section 6.8) and obtained values of
q', of about 16 ". Similar values have been obtained using
cut-plane samples in the triaxial apparatus. These values
may be compared with the value of 223" for the critical state
strength predicted by Schofield and Wroth 17.71. More
recently, Bishop and others, using the ring shear apparatus,
which facilitates much larger displacements, have obtained
much lower values [7.25]. Evidence from field investigations
indicates that this further reduction in strength reflects the
formation of large and continuous slip surfaces within the
soil mass. The displacements necessary to develop these slip
surfaces are generally much greater than are usually
Fig. Shear strength of clay soils in terms of eflectiue stress. observed in cuttings or natural slopes which have not
previously slipped. However, where there is a risk of
nearly equal to the maximum. In some structures, however, reactivating an old slip surface, the residual shear strength
quite large strains occur locally in the soil mass before total which must be used in design may be much less than the
collapse. In these parts, the soil strength will have been critical state strength.
reduced to the critical state value. The average strength then
lies between the maximum and the critical state value, and 7.26 Slip lines and rupture surfaces: There is a fundamental difference
may be much nearer to the latter. between the drained behaviour of a soil initially on the 'wet'
 GENERAL SOIL MODELS 175
INTRODUeTION TO SOIL MECHANICS AND FOUNDATIONS
ROSCOE,K. H. and BURLAND, J. B. 1968. On the
or loose side of the critical state, and one initially on the 'dry' generalised stress-strain behaviour of wet clay. In
or dense side. In a normally consolida.ted or lightly Engineering Plasticity. (Ed. HEYMAN, J. and LECKIE, F.)
overconsolidated soil, having an initial void ratio on the Cambridge Univ. Press (London).
loose side of the critical state line, the approach to failure is VERMEER, P. A. 1978.A double hardening model for sand.
accompanied by continuous progressive strain hardening Geotechnique, 28.
and a general collapse of the soil structure. Dense soil, on the ZIENKIEWICZ, 0 . C. and PANDE, G. N. 1976. Some useful
other hand, dilates during shear and its strength decreases. forms of isotropic yield surfaces for soil and rock
As a result, the strains are concentrated in a few narrow mechanics. Numerical methods in soil and rock mechanics.
bands of weakened material which appear as slip lines. It is (Ed. BORN,G. and MEISSNER, H.), 2. Institut fiir Boden-
important to realise that when, in later chapters, reference is ,mechanik und Felsmechanik, Karlsruhe.
NAYLOR,D. J. 1978. Stress-strain laws for soil. In
made to slip lines and rupture surfaces, it is in the context of
Developments in soil mechanics. (Ed. Scon, C. R.)
the behaviour of a perfectly plastic model soil. In real loosely Applied Science Publ. (London).
packed soil, no rupture surfaces will occur under shear BURLAND, J. B. 1971. A method of estimating the pore
loading, at least until after failure. Even in heavily pressures and displacements beneath embankments on
overconsolidated soils, the rupture surfaces which do soft natural clay deposits. Proc. Roscoe Memorial Symp.,
develop may not be very close to those implied by the Cambridge.
Mohr-Coulomb model. SKEMPTON, A. W. and LAROCHELLE, P. 1965. The Bradwell
Slip: a short term failure in the London clay.
Geotechnique, 15.
TAYLOR, D. W. 1948. Fundamentals of soil mechanics. SKEMPTON, A. W. 1957. Discussion on 'The planning and
Wiley (New York). design of the new Hong Kong airport'. (GRACE,H. and
RENDULIC, L. 1936. Porenziffer und Porenwasserdruck in HENRY, J. K. M.) Proc. Inst. Civil Eng., 7.
Tonen. Der Bauingenieur, 17. BJERRUM, L. 1967. Engineering geology of Norwegian
normally consolidated clays as related to settlements of
HENKEL, D. J. 1960. The shear strength of saturated
remoulded clays. Research ConJ on Shear Strength of buildings. Geotechnique, 16.
SKEMPTON, A. W. and HENKEL, D. J. 1960. Tests on
Cohesive Soils, Boulder, Colorado (ASCE).
London clay from deep borings at Paddington, Victoria
HVORSLEV, M. J. 1937. Uber die Festigkeitseigenshaften and the South Bank. Proc. 4th Int. ConJ Soil Mech. and
gestorter bindiger Boden. Danmarks Naturvidenskabelige Found. Eng.
Samfund. Copen hagen.
SKEMPTON, A. W. 1964. Long term stability of clay slopes.
HVORSLEV, M. J. 1960. Physical components of the shear
Geotechnique, 14.
strength of saturated clays. Research ConJ on Shear
SKEMPTON, A. W. 1970. First-time slides in over-
Strength of Cohesive Soils, Boulder, Colorado (ASCE).
consolidated clays. Geotechnique, 20.
ROSCOE, K. H., SCHOFIELD, A. N. and WROTH, C. P. 1958.
On the yielding of soils. Geotechnique, 8.
BISHOP, A. W., GREEN, G . E., GARGA, V. K., ANDRESEN, A.
and BROWN, J. D. 1971.A new ring shear apparatus and its
SCHOFIELD, A. N. and WROTH,C. P. 1968. Critical state
soil mechanics. McGraw-Hill (London). application to the measurement of residual strength.
Geotechnique, 21.
ATKINSON, J. H. and BRANSBY, P. L. 1978. The mechanics of
soils. McGraw-Hill (London).
PARRY, R. G. H. 1958. On the yielding of soils (Discussion
of ref. 7.6). Geotechnique, 8.
DRUCKER, D. C., GIBSON, R. E. and HENKEL, D. J. 1957.
Soil mechanics and work hardening theories of plasticity.
Trans. ASCE, 122.
BISHOP, A. W. and ELDIN,A. K. G . 1953. The effect of
stress history on the relation between qp and porosity in
sand. Proc. 3rd Int. ConJ Soil Mech. and Found. Eng.
KIRKPATRICK, W. M. 1957. The condition of failure for
sands. Proc. 4th Int. Conf. Soil Mech. and Found. Eng.
BISHOP,A. W. 1966. The shear strength of soils as
engineering materials. Geotechnique, 16.
 ysis, and

PLASTICITY AND COLLAPSE

8.1 Deformation, yield, and collapse: A system of surface and body

forces applied to a body will cause changes in the stresses and
strains within it. The changes in strain will usually be partly
elastic and immediately recoverable, but may also result in
part from irrecoverable plastic yield. As the forces increase, a
growing part of the body may reach the limit of its shear
strength. Once this part is sufficiently extensive to form an
unstable mechanism, unlimited yield will occur. This will
continue as long as the forces can be maintained in the face of
the changing geometry of the body. This unlimited yield is
called collapse, and the system of forces which will produce it
is called the collapse load.
T In principle, the conditions of equilibrium and of strain
/
compatibility, together with a knowledge of the stress/strain
! behaviour of the material, are sufficient to determine the
distribution of stress and displacement up to the moment of
, collapse. In practice, however, the stresslstrain behaviour of
soil is so complex, and so dependent on the loading history,
that analysis of soil structures must be drastically simplified.
It is not that we are evading the mathematical complexities,
great though these are; but a full solution would require a
knowledge of the behaviour of the soil, and of the past and
future loads on it, which in practice we cannot hope to
obtain.
Fortunately, most practical problems may be solved by
answering the following two questions:
(a) What are the structural displacements under the
working loads?
(b) Is the working load substantially less than the
collapse load?
Provided that the applied load is substantially less than the
collapse load, the first question can often be answered with
sufficient accuracy by assuming that the stresses (and
sometimes the displacements also) are those which would
occur in a linearly elastic body. This was discussed in Chapter
177