MaksimoviC, M. (1989). G~ofechnique 39, No.

2,347-351

TECHNICAL NOTE

On the residual shearing strength of clays

M. MAKSIMOVIc*

KEYWORDS: clays; friction; shear strength; land- of the failure envelope is of hyperbolic type and
slides; slopes; Atterberg limits can be defined as a variation of the angle of the
residual shearing resistance in the form
INTRODUCTION
The angle of residual shearing resistance &, often Y, = #BJ + 4@r/(l + O'"lPN,J c-4
depends on the magnitude of normal stress acting where +‘r,r is the basic angle of residual friction,
on the failure plane, as shown by Skempton & 4#, is the maximum angle difference and pN,r is
Petley (1967), Bishop (1971), Lupini, Skinner & the median angle normal stress. The parameters
Vaughan (198 l), Chandler (1984), Skempton and variables related to the effective stress on the
(1985), Lambe (1985) and others. Residual failure plane are shown in Fig. 1.
strength envelopes of pure clay minerals are Equation (2) describes in simple terms the angle
described as curved lines by Kenney (1967). The of shearing resistance as the sum of two parts and
curvature of the envelope of residual shearing has the following properties: when u’” = 0 then
strength is not well understood, but it may be @, = #B,r + 4@, = @o,r; when c’, + 00 then
attributed to the different degrees of orientation @, = &‘B,r. The asymptotic value I$‘~,~is equal to
of platy particles with increasing parallelism to the angle of the shearing resistance for high con-
the failure plane as the normal stress on the fining stress levels at which the perfect orientation
failure plane increases. of platy particles at large shearing displacements
The cohesion intercept in terms of effective governs the failure in soils with high contents of
stresses is conventionally obtained by extrapo- clayey particles.
lation of the straight line envelope towards the The second part of equation (2) contains the
zero normal stress level, and its apparent value is value of 4+‘, , the maximum angle difference, and
mainly a consequence of the non-linearity of the the hyperbolic function of the stress level, which
failure envelope. describes the rate of decrease of the angle as the
A number of different expressions describing stress level increases. The median angle normal
the non-linear failure envelopes of both cohesive stress is defined as the value of the normal effec-
and cohesionless materials have been proposed in tive stress for which the angle of the residual
the literature. A detailed discussion of the various shearing resistance amounts to the mean value
expressions is beyond the scope of this Technical between 4’rrr and @o.r.
Note.

DETERMINATION OF PARAMETERS FROM

PROPOSED EXPRESSION FOR ENVELOPE
As most of the tests for the residual shearing EXPERIMENTAL DATA
strength are carried out either by use of direct In this section the subscript ‘r’(for residual) will
shear box or ring apparatus, the shearing strength be omitted, as equation (2) can also be used for
is related to the normal stress on the failure plane. the description of the peak shearing strength in
Here the conventional interpretation will be terms of effective stresses of most non-cemented
assumed, which (for c’ = 0) defines the angle of soils. If results for only three stress levels are con-
the residual shearing resistance as sidered, (this being the smallest number of results
required for the derivation of parameters), the
#, = tan - ’ (~,,,/a’,) (1) system of three equations can be solved for
unknown parameters, if stresses on the failure
Several possible expressions for the description
plane are in a sequence such that cri,i < 01,~ <
of the non-linearity of the failure envelope were
06,~ and the following condition is satisfied
described by Maksimovic (1979, 1988). It is the
Author’s opinion that the most convenient simple
expression for the description of the non-linearity

Discussion on this Technical Note closes on 6 October An interpretation of this kind is necessary if
1989. For further details see p. ii. results are reported for three stress levels only, or
* University of Belgrade. if the envelope is given in the form of a smooth

347
348 MAKSIMOVIC

and the parameter b is

b = 1/(&o - #a) = l/A@ (6)
or the maximum angle difference is
A# = lJb (7)
Also, it can be shown that

a = POW’ (8)
or the median angle stress is
Effective normal stress f~‘.
pN = aA&. (9)
Fig. 1. Variation of secant angle of shearing resistance
$’ with change of effective normal stress on failure plane The equivalent equation (4) can be rewritten in
and definition of parameters for description of non-linear the form
failure envelope.
a + ba’, = a’J(c$‘,, - 4’) (10)
The left hand side is a straight line. The right
and continuous graph and digitalization for more hand side can be assigned for each data point if
than three points does not seem justified. This the value of @O is assumed. The problem now
procedure will be called ‘the perfect fit’ technique reduces to the linear regression analysis (least
implying that perfect fit is achieved in three square fit) to obtain a and b, with variation of
points only. &,, . The solution is reached when the mean error
It is preferable that tests are made for more has the smallest value for all possible choices of
than three stress levels, in the widest range pos- @o. The final solution might be influenced by the
sible, and that results are presented in digital choice of the adopted mean error. In this Note
form. If more than three data points are available, the mean deviation is defined as the mean of
the parameters can be obtained by linearized absolute values of differences between the com-
curve fitting that minimizes the certain norm of puted angles and data values for the same normal
error, as equation (2), (Maksimovik, 1979), can be stress, or
written in the equivalent form
4’ = #,, - a’,/@ + bo’,) (4) average error = A $1 tfl - 4’1 (11)
The asymptotic value of 4’ when stress tends to
where m is the number of data points, #1 is the
infinity is
angle of shearing resistance obtained by testing
&a = #(co) = &O - l/b (5) for certain stress and 4’ is the angle of shearing

Table 1. Residual shear streneth narameters for non-linear failure envelopes for some clays and clay minerals. Field
values obtained by back-analysii are marked with *.
-I-

Number Material Ad,: PN.r: Data

% deg kPa
1 Brown London clay-ring shear test 6.9 5.6 57.0 Bishop (1971)
2 Brown London clay-ring shear-average 8.0 9.0 20.0 Chandler (1984)
3 London clay landslides-average * 8.0 9.0 120.0 Chandler (1984)
4 London clay landslides-average * 10.3 6.0 50.0 Skempton (1985)
5 Lias clay-ring shear-average 6.5 13.5 20.0 Chandler (1984)
6 Lias clay landslides-average 6.5 13.5 50.0 Chandler (1984)
I Lias clay landslides-average * 4.8 10.9 133.3 Skempton (1985)
8 Upper Carboniferous-Walton’s Wood * 10.8 4.0 150.0 Skempton (1985)
9 Etruria marl-Bury Hill * 11.1 2.6 30.0 Skempton (1985)
10 Atherfield clay-Sevenoaks bypass * 7.0 8.9 130.0 Skempton (1985)
11 Weald clay-ring shear test 8.2 11.8 55.7 Bishop et al. (1971)
12 Amuay clay PI = 15% (chart) 17.9 15.6 31.3 Lambe (1985)
13 Amuay clay PI = 30% (chart) 10.9 15.6 31.3 Lambe (1985)
14 Amuay clay PI = 50% (chart) 5.4 15.6 31.3 Lambe (1985)
15 Hydrous mica-laboratory test 19.6 13.5 loo.0 Kenney (1967)
16 Kaolinite-laboratory test 10.6 13.4 55.0 Kenney (1967)
17 Na montmorillonite-laboratory test 2.1 4.3 110.0 Kenney (1967)
 CLAY RESIDUAL SHEARING STRENGTH 349

resistance computed from calculated parameters

and respective stress. Parameters can be obtained
by minimizing the error practically to zero when
conditions for ‘perfect fit’ for three data points are
satisfied.
Parameters for examples presented in this
Technical Note are obtained by application of the
rigorous method described. A program for inter-
active computing and graphical presentation on a
personal computer has been written to perform PN = 42.6 kPa
6 I
this task. However, in practical work, proposed 0 0.4 0.8
parameters can be estimated or computed results Effective normal stress a’“: MPa

checked with reasonable accuracy from graphs, Fig. 3. Curve fitting for results of ring shear test for the
and from simple, clear definitions of parameters. residual strength of Weald clay. Results by Bishop et al.
Some other, rather obvious and simple approx- (1971)
imate methods for derivation of parameters,
which will not be discussed in this Note, are also
atus known to the Author are these on Weald
possible.
clay (case 11). Six data points, the curve of
analytical approximation, and parameters are
EXAMPLES FOR JUSTIFICATION shown in Fig. 3.
Examples selected from the literature justify the (4 Skempton (1985) examined the residual
proposed expression (2) for the non-linear failure strength of some English clays and compared
envelope. Care is taken that test results con- field and laboratory values. One of the report-
sidered in this analysis are of high quality, ed conclusions was that for most clays the
published by recognized authors, so that the relation between residual strength and normal
results are well known to the professional and effective pressure is non-linear. He gave values
easily accessible in the given references. Prefer- for three stress levels in a digital form, and
ence is given to results which are presented described the field residual strength with
numerically, but graphs of plotted results are also values of the angles of shearing resistance for
used. The results shown in Table 1 contain three stress levels: 50, 100 and 150 kPa. His
derived parameters for selected results of results, which satisfy the condition required
published tests performed on natural clays and for ‘the perfect fit’ method are shown as cases
some pure minerals. Four examples from Table 1 4 and 7-10 in Table 1. The graph for the
(cases 1, 11, 4 and 12-14), are presented in some average value of the angle of the shearing
detail. resistance obtained from the back-analysis of
various landslides in London clay is shown in
(a) Bishop (1971) gave results of ring shear tests Fig. 4.
for weathered brown London clay, (case I), (4 The non-linear dependence of the residual
which: were plotted by Chandler (1971) in a angle as the function of the stress level and
manner shown in Fig. 2. By using the pro- plasticity index was presented by Lambe
posed method, the same set of results was (1985) for Amuay clays in graphic form. By
analysed and a good fit was obtained as
shown by the parameters obtained from four
data points in Fig. 2. I_ @‘o,,
= 16.3”
(b) The published results which cover the widest 16

normal stress range in the ring shear appar-

A@‘, = 6.0
F

‘612
u
P
4: q+‘s,, = 10.3”

:‘
PN = 50.0 kPa
8 I
0 100 200
Effectwe normal stresso’,: kPa

Fig. 4. Curve fitting for average residual shearing

Effective normalstresso’,: kPa
strength obtained by back-analysis of various landslides
Fig. 2. Ring shear test for residual shearing strength of in London clay. Perfect fit method for data by Skemptoa
London clay. Data by Bishop (1971) (1985)
 350 MAKSIMOVl6

~,q = 31.3 kPa

01
0 20 60
Plastratyindex(PI): 2’
01 I

0 20 40
Plastlaty index (PI): % Fig. 7. Approximate correlation of proposed para-
meters and plasticity index. Typical values for pN,r are in
Fig. 5. Parameters for residual strength of Amuay
range 10-30 kPa in shear test, but in field values are 3-5
clays. Proposed parameters based on graph by Lambe
times larger
(1985)

applying the regression analysis the graph in the degree of freedom of continuity of the
shown in Fig. 5 was constructed, which shows failure surface in the field compared with the
that the initial angle @0 is a function of plasti- forced shearing in the apparatus.
city index. In terms of equation (2) it can be The estimated range of parameters that
concluded that the basic angle 41B depends on describe the residual non-linear failure envelope is
the plasticity index, that the maximum angle shown in Fig. 7. The maximum angle difference
difference is constant A$’ = 15.6” (for the seems to decrease with increasing plasticity. The
graph analysed), and the median angle normal median angle normal stress is in the range l&
stress is also a constant at 31.3 kPa or 150 kPa.
approximately 30 kPa. Failure envelopes for
three values of the plasticity index are shown
in Fig. 6.
SUMMARY AND CONCLUSIONS
On limited evidence, it can be estimated that The non-linear residual failure envelope in
the initial angle of the residual failure envelope terms of effective stresses can be described for
might be in the range between the values of the
most soil types by a simple three parameter
peak shear resistance of a remoulded normally expression
consolidated sample of the same soil and its fully-
softened or critical state value. L = cfn
tanC~J’~,~
+ WA1 + ~‘,h,,)l (12)
The accumulated experience suggests
(Chandler, 1984; Skempton 1985) that the ring- which contains the basic residual angle of friction,
shear apparatus underestimates the field strength. the maximum angle difference and the median
This Technical Note indicates, at least tentatively, angle normal stress.
that the ring-shear test underestimates the The proposed expression reduces to the con-
median angle stress parameter pN only, by some ventional Coulomb’s straight line envelope for
significant factor in the approximate range from 3 three combinations of parameters, i.e., when
to 5 times, but the other two parameters, $IB and A@, = 0, or when pN,r= 0 or when pN,ris infinite.
A@, would still be directly applicable to field con- In this sense the conventional linear envelope is
simply one special case of the proposed non-
ditions. An explanation could probably be found
linear law of failure in soils.
The basic angle of residual friction r#~‘~,~ is con-
stant for a particular soil. It represents the angle
of the shearing resistance of perfectly orientated
particles at some high stress level.
~~ The maximum angle difference A@ has the fol-
lowing significance.
(a) It is approximately equal to or slightly smaller
0 150 200 than the difference between the angle of the
Effective normal stress 0’“: kPa shearing resistance of the remoulded normally
Fig. 6. Failure envelopes for three values of plasticity consolidated clay and the basic angle defined
index for Amuay clays based on parameters shown in previously. It is the consequence of the change
Fig. 5 of particle orientation from random to perfect.
 CLAY RESIDUAL SHEARING STRENGTH 351

(b) The change of the strength from peak for the Proc. Oslo Geotech. ConJ on the Shear Strength
normally consolidated clay to its residual Prop. of Natural Soils and Rocks, 1, 124-129.
value can be described approximately by the Lambe, T. W. (1985). Amuay landslides. Proc. Ilth Int.
Conf on SMFE, Golden Jubilee Volume, pp. l37-
change of the mean angle stress pN from infin-
158, San Francisco, Brookfield VT: Balkema.
ity to some rather small value.
Lupini, J. F., Skinner, A. E. & Vaughan, P. R. (1981).
The drained residual strength of cohesive coils, Gfo-
technique 31, No. 2, 181-213.
REFERENCES Maksimovii-, M. (1979). Limit equilibrium for nonlinear
Bishop, A. W., Green, G. E., Garga, V. K., Andresen, A. failure envelope and arbitrary slip surface. Third Int.
& Brown, J. D. (1971). A new ring shear apparatus Conf on Numerical Methods in Geomechanics,
and its application to the measurement of residual Aachen, pp. 769-777.
strength. GCotechnique 21, No. 4, 273-328. Maksimovii, M. (1988). General slope stability software
Bishop, A. W. (1971). Shear strength parameters for package for micro-computers. 6th Int. ConJ on
undisturbed and remolded soil specimens. Stress- Numerical methods in Geomechanics, Innsbruck 3,
strain behaviour of soils. Roscoe memorial sympo- 2145-2150.
sium, pp. 3-58. Henley: Foulis. Skempton, A. W. & Petley, D. J. (1967). The strength
Chandler, R. J. (1971). Discussion. Roscoe memorial along structural discontinuities in stiff clays. Proc.
symposium, 733-735. Oslo Geot. Conf on Shear Strength Prop. of Natural
Chandler, R. J. (1984). Recent European experience of Soils and Rocks 2, 29-46.
landslides in overconsolidated clays and soft rocks. Skempton, A. W. (1985). Residual strength of clays in
4th Int. Symp. on Landslides, Toronto, 1, 61-81. landslides, folded strata and the laboratory. Gio-
Kenney, T. C. (1967). The influence of mineralogical technique 35, No. 1, 3-18.
composition on the residual strength of natural soils.