APPENDIX A

Mohr’s circle for two-dimensional stress

Compressive stresses have been taken as positive because we shall almost exclusively be
dealing with them (as opposed to tensile stresses) and because this agrees with the
universal practice in soil mechanics. Once this sign convention has been adopted we are
left with no choice for the associated conventions for the signs of shear stresses and use of
Mohr’s circles.

Fig. A.1 Stresses on Element of Soil

The positive directions of stresses should be considered in relation to the Cartesian

reference axes in Fig. Al, in which it is seen that when acting on the pair of faces of an
element nearer the origin they are in the positive direction of the parallel axis. The plane
on which the stress acts is denoted by the first subscript, while the direction in which it acts
is denoted by the second subscript. Normal stresses are often denoted by a single subscript,
for example, σ ' x instead of σ 'xx .
For Mohr’s circle of stress (Fig. A.2) we must take counterclockwise shear as
positive, and use this convention only for the geometrical interpretation of the circle itself
and revert to the mathematical convention for all equilibrium equations. Hence
X has coordinates (σ 'xx , − τ xy )
and Y has coordinates (σ ' yy , + τ yx )
But from equilibrium we require that τ xy = τ yx .

Fig. A.2 Mohr’s Circle of Stress

Suppose we wish to relate this stress condition to another pair of Cartesian axes (a,
b) in Fig. A.3 which are such that the counterclockwise angle between the a- and x-axes is
+θ.
Then we have to consider the equilibrium of wedge-shaped elements which have
mathematically the stresses in the directions indicated.
207

Fig. A.3 Stresses on Rotated Element of Soil

Resolving forces we get:

σ ' +σ ' yy σ ' xx −σ ' yy ⎫
σ 'aa = xx + cos 2θ + τ xy sin 2θ ⎪
2 2 ⎪
− (σ ' xx −σ ' yy ) ⎪
τ ab = sin 2θ + τ xy cos 2θ ⎪⎪
2 (A.1)
⎬
σ 'xx +σ ' yy σ 'xx −σ ' yy
σ 'bb = − cos 2θ − τ xy sin 2θ ⎪
2 2 ⎪
(σ ' xx −σ ' yy ) ⎪
τ ba = − sin 2θ + τ xy cos 2θ = τ ab . ⎪⎪
2 ⎭

In Mohr’s circle of stress

A has coordinates (σ 'aa , − τ ab )
and B has coordinates (σ 'bb , + τ ba ).
A very powerful geometric tool for interpretation of Mohr’s circle is the
construction of the pole, point P in Fig. A.4. Through

Fig. A.4 Definition of Pole for Mohr’s Circle

any point on the circle a line is drawn parallel to the plane on which the corresponding
stresses act, and the pole is the point where this line cuts the circle. In the diagram XP has
been drawn parallel to the y-axis, i.e., the plane on which σ 'xx and τ xy act.
This construction applies for any point on the circle giving the pole as a unique
point. Having established the pole we can then reverse the process, and if we wish to know
the stresses acting on some plane through the element of soil we merely draw a line
through P parallel to the plane, such as PZ, and the point Z gives us the desired stresses at
once.
208

This result holds because the angle XCA is +2θ (measured in the counterclockwise
direction) as can be seen from eqs. (A.1) and by simple geometry the angle XPA is half
this, i.e., +θ which is the angle between the two planes in question, Ox and Oa.
209
210
211
212
213
214
215
216

APPENDIX C
A yield function and plastic potential for soil under
general principal stresses

The yield function, F(p, q), for Granta-gravel, from eq. (5.27), is
⎛ p ⎞
F = q + Mp⎜⎜ ln − 1⎟⎟ = 0 (C.1)
⎝ pu ⎠
where
σ 'l +2σ 'r ⎛ Γ −v⎞
p= , q = σ 'l −σ 'r , and pu = exp⎜ ⎟.
3 ⎝ λ ⎠
We can treat the function F* as a plastic potential in the manner of §2.10, provided we
know what plastic strain-increments correspond to the stress parameters p and q. In §5.5
we found that υ& υ corresponded to p, and ε& corresponded to q. Therefore, from eq. (2.13),
we can first calculate
∂F
= vε& = 1,
∂q
so that the scalar factor v is
1
v= , (C.2)
ε&
and then we can calculate
∂F υ& 1 υ& p ⎛ q⎞
=v = = M ln = ⎜⎜ M − ⎟⎟.
∂p υ ε& υ pu ⎝ p⎠
This restates eq. (5.21) and thus provides a check of this type of calculation.
We wish to generalize the Granta-gravel model in terms of the three principal
stresses and obtain a yield function F * (σ '1 ,σ '2 ,σ '3 , pu ) where pu remains as specified
above. Let us retain the same function as before, eq. (C.1), but introduce the generalized
parameters of §8.2,
⎛ σ ' +σ '2 +σ '3 ⎞
p* = ⎜ 1 ⎟
⎝ 3 ⎠
and
1
{(σ '2 −σ '3 )2 + (σ '3 −σ '1 )2 + (σ '1 −σ '2 )2 }2 .
1

q* =
2
The function F* then has equation
1
{(σ '2 −σ '3 )2 + (σ '3 −σ '1 )2 + (σ '1 −σ '2 )2 }2
1

F* =
2
(C.3)
⎛ σ '1 +σ '2 +σ '3 ⎞⎧ ⎛⎜ σ '1 +σ '2 +σ '3 ⎞⎟ ⎫
+M ⎜ ⎟⎨ln⎜ ⎟ − 1⎬ = 0
⎝ 3 ⎠⎩ ⎝ 3 pu ⎠ ⎭
This function F* generates a surface of revolution about the diagonal of principal stress
space as shown in Fig. 5.1. Variation of pu generates successive surfaces as indicated in
Fig. 5.2.
217

Let us treat F* as a plastic potential. Clearly, the stress parameters (σ '1 ,σ '2 ,σ '3 )
are associated with plastic strain-increments (ε&1 , ε&2 , ε&3 ) since the loading power is
E&
σ '1 ε&1 + σ '2 ε&2 + σ '3 ε&3 = . (C.4)
υ
Therefore, from eq. (2.13) we calculate
∂F * 3{σ '1 −[(σ '1 +σ '2 +σ '3 ) 3]} M ⎛ σ '1 +σ '2 +σ '3 ⎞
= v * ε&1 = + ln⎜ ⎟⎟
∂σ '1 {[
2 (σ ' −σ ' ) + (σ ' −σ ' ) + (σ ' −σ ' )2 ] 2} 3 ⎜⎝
1
2 2 2 3 pu ⎠
2 3 3 1 1 2

∂F * 3{σ '2 −[(σ '1 +σ '2 +σ '3 ) 3]} M ⎛ σ '1 +σ '2 +σ '3 ⎞
= v * ε&2 = 1 + ln⎜ ⎟⎟ (C.5)
∂σ '2 {[
2 (σ '2 −σ '3 ) + (σ '3 −σ '1 ) + (σ '1 −σ '2 ) 2 2
2 2 2
] }
3 ⎜⎝ 3 pu ⎠
∂F * 3{σ '3 −[(σ '1 +σ '2 +σ '3 ) 3]} M ⎛ σ '1 +σ '2 +σ '3 ⎞
= v * ε&3 = 1 + ln⎜ ⎟⎟
∂σ '3 {[
2 (σ '2 −σ '3 ) + (σ '3 −σ '1 ) + (σ '1 −σ '2 ) 2 2
2 2 2
] }
3 ⎜⎝ 3 pu ⎠

If we introduce ε&*, a scalar measure of distortion increment that generalizes eq. (5.6) and
eq. (5.9), in the form
2
{ }
(ε&2 − ε&3 )2 + (ε&3 − ε&1 )2 + (ε&1 − ε&2 )2 2
1

ε&* = (C.6)
3
then, as in eq. (C.3), we find from eq. (C.5) that
1
v* = (C.7)
ε& *
It is now convenient and simple to separate (C.5) into two parts:
ε&1 + ε&2 + ε&3 q*
=M − , (C.8)
ε& * p*
and
ε&2 − ε&3 3 σ '2 −σ '3
=
ε& * 2 q*
ε&3 − ε&1 3 σ '3 −σ '1
= (C.9)
ε& * 2 q*
ε&1 − ε&2 3 σ '1 −σ '2
=
ε& * 2 q*
The first part, eq. (C.8), is a scalar equation relating the first invariant of the plastic strain-
increment tensor to other scalar invariants. The second part, eqs. (C.9), is a group of
equations relating each component of a plastic strain-increment deviator tensor to a
component of a stress deviator tensor. We will now show that these equations can be
conveniently employed in two calculations.
First, we consider the corner of the yield surface. When q* = 0 and
σ '1 = σ '2 = σ '3 = p*, we find that eqs. (C.9) become indeterminate, but eq. (C.8) gives
⎛ ε& + ε& + ε& ⎞
ε&* = ⎜ 1 2 3 ⎟ (C.10)
⎝ M ⎠
Here, as in §6.6, we find that a plastic compression increment under isotropic stress is
associated with a certain measure of distortion.
218

Next, we consider what will occur if we can make the generalized Granta-gravel
sustain distortion in plane strain at the critical state where (ε&1 + ε&2 + ε&3 ) = 0. In plane strain
ε&2 = 0, so at the critical state, ε&1 + ε&3 = 0. With eqs. (C.9) these give
3 σ '2 −σ '3 ε& ε& 3 σ '1 −σ '2
=− 3 =+ 1 =
2 q* ε& * ε& * 2 q *
from which
σ ' +σ '
σ '2 = 1 3 . (C.11)
2
We satisfy this equation if we introduce the simple shear parameters (s, t) where
σ '1 = s + t ,σ '2 = s,σ '3 = s − t. In terms of these parameters, the values of p* and q* are
{ 1 2
} ( )
1

p* = s and q* = t + 4t 2 + t 2 2 = 3 t
2
The yield function F* at the critical state reduces to
q * −Mp* = 0,
so that
M
t= s. (C.12)
3
This result was also obtained by J. B. Burland1 and compared with data of simple shear
tests. In fact the shear tests terminated at the appropriate Mohr-Rankine limiting stress ratio
before the critical state stress ratio of eqn. (C. 12) was reached.

1
Burland, J. B. Deformation of Soft Clay, Ph.D. Thesis, Cambridge University, 1967.