Environmental Engineering Reference
In-Depth Information
where
p
and
r
are the peak and residual strengths defined under the same effective nor-
mal stress.
However, as indicated by Vaughan & Hamza (1977) and by Chandler (1984) the brit-
tleness index alone is not sufficient to characterise the susceptibility of a soil to progres-
sive failure and the rate at which the strength decreases from peak strength to ultimate
strength is also important. D'Elia et al. (1998) propose a generalised Brittleness index, I
BG
,
defined as follows:
p
ob
(6.3)
I
100%
BG
p
mob
is the mobilised shear stress at the considered strain or displacement.
I
BG
thus varies with strain or displacement from 0 at the peak to a value equal to I
B
at
large displacements. I
BG
, is associated to the stress paths that are representative of those
followed
in situ
, and must thus not be seen as a fundamental characteristic of a soil. With
this extended definition, not only overconsolidated clays, clay shales, sensitive clays,
residual soils and loess, but also cohesionless soils such as loose sands, may appear brittle
in undrained conditions.
Contractant conditions may occur in mine tailings, end-tipped mine waste dumps, uncom-
pacted or poorly compacted road or railway fills, normally and lightly over-consolidated
clays, poorly compacted clay fills (e.g. old dams which were compacted by horses hooves or
rolled with light rollers in thick layers), puddle cores, and in the core of large dams where the
confining stresses exceed the pre-consolidation effects of compaction, and potentially in loose
dumped dirty rockfill in dams (although the authors are not aware of this happening).
Cooper et al. (1997) discuss an example where the use of conventional effective stress
analysis over-estimated by about 50% the factor of safety of an earth dam constructed
with poorly compacted clay and with concrete core wall.
As has been recognized for a long time, e.g. Jamiolkowski et al. (1985), Ladd (1991)
and Kulhawy (1992), the undrained shear strength used in the analysis should take
account of the failure mechanism. As shown in
Figure 6.10
the loading conditions under
an embankment are best simulated by triaxial compression, while beyond the toe, they are
best simulated by triaxial extension and direct simple shear. Centrally within an embank-
ment they are probably represented by triaxial compression, and at the upstream and
downstream toe by triaxial extension.
Figure 6.11
shows the mean normalized undrained strength ratios for the major labo-
ratory tests.
For this figure the reference strength ratio is given by the modified Cam Clay model,
which Kulhawy (1992) indicates is reasonable for relatively unstructured soils. For sensi-
tive, cemented and other structured fine grained soils, the reference strength is a lower
bound. It can be seen that to use Ko triaxial compression tests is unconservative for those
parts of the failure surface best modelled by triaxial extension and direct simple shear tests.
The behaviour of contractant granular soils is a complex, developing science, and will
not be further discussed here. Reference should be made to Fell et al. (2000) and the ref-
erences given therein and to the recent literature.
in which
6.1.4
Laboratory testing for drained strength parameters, and common errors
The following discussion on shear strength of soils briefly outlines the test procedures and
some of the problems which arise in testing and interpretation of results. It is the authors'
experience that many organisations, at least in the past, failed to follow correct test pro-
cedures despite the fact that they have been long established.
