You are on page 1of 10

TRIAXIAL TESTS

ON

WEAKLY BONDED SOIL


IN STRESS PATH

WITH

CHANGES

By V. Malandraki1 and D. G. Toll2


ABSTRACT: The results are presented of drained triaxial tests on a weakly bonded artificial soil in which the
stress path direction has been changed partway through the shearing process. The effects of the previous shearing
path history on the yield and failure surfaces are examined. Yield of the bonds occurs under each stress path
direction followed, even when yield has previously occurred along another path. This demonstrates that bond
breakdown is an anisotropic process. The position of yield was found to be independent of the previous shearing
path history of the soil and occurred at points that corresponded to a yield surface defined for the current
shearing path direction. However, the previous shearing path history of the soil did significantly affect the failure
envelope. It is suggested that the bond yield surface is kinematic, in the sense that it is an expandable/shrinkable
surface, but that it is not a moveable surface. It is postulated that the yield surface expands when volumetric
strains are compressive and shrinks when the volumetric strains are dilatent.

INTRODUCTION
Natural materials are often structured because of weak
bonds between their particles. Although the bonded structure
may arise from different reasons, they present similar characteristics. An important consideration for bonded materials is
that their yield behavior is controlled by the bonded structure,
which can be independent of the previous stress history.
In the last 15 years, attempts have been made to extend the
theoretical models of soil behavior (which have largely been
developed from studying the behavior of remolded or reconstituted soils) and include the effects of bonded structure on
soil behavior. Malandraki and Toll (1994, 1996, 2000) have
put forward a framework that explains the behavior of an artificially weakly bonded soil under a range of different stress
path directions.
Conventially, soils are tested in the triaxial cell under conditions where the cell pressure 3 remains constant; for drained
tests this imposes a constant 3 stress path. However, in the
field, soils will be subjected to many different stress paths. For
example, an element of soil adjacent to an excavation will be
subject to a stress path where the horizontal stress decreases
(as adjacent soil is removed by excavation) but the vertical
stress remains constant. In a free-draining soil this will produce a reduction in effective horizontal stress but with a constant effective vertical stress (if the ground-water level is not
modified by the excavation process). This can be modeled in
a stress path triaxial cell as a constant 1 path. Such excavation
processes, combined with changes in external loading or
changes in ground-water level or both may produce complex
stress paths in the field, in some cases with distinct changes
in stress path direction. Therefore, this study examines the behavior of soil when subjected to such changes in stress path
direction partway through the shearing process.
Malandraki and Toll (1996) discussed the yield behavior of
weakly bonded soils, basing their ideas on the work of
Vaughan (1988) and Jardine (1992). They suggested incorporating a bond yield (originally referred to as second yield
by Vaughan and Malandraki and Toll). This point is in addition
1
PhD, Former Res. Asst., School of Engrg., Univ. of Durham, Durham,
DH1 3LE UK.
2
Sr. Lect., School of Engrg., Univ. of Durham, Durham, DH1 3LE UK.
E-mail: d.g.toll@durham.ac.uk
Note. Discussion open until August 1, 2001. To extend the closing
date one month, a written request must be filed with the ASCE Manager
of Journals. The manuscript for this paper was submitted for review and
possible publication on October 14, 1998; revised October 17, 2000. This
paper is part of the Journal of Geotechnical and Geoenvironmental
Engineering, Vol. 127, No. 3, March, 2001. ASCE, ISSN 1090-0241/
01/0003-02820291/$8.00 $.50 per page. Paper No. 19447.

to the yield points identified by Jardine: Y1 (first yield)


representing the limit of linear elastic behavior; Y2 representing the limit of recoverable behavior; and Y3 (final yield)
representing the complete destruction of any structure within
the soil. Malandraki and Toll suggested that bond yield could
be identified from the tangential stiffness versus axial strain
graph at a point where a major loss of stiffness is initiated.
The patterns of behavior expected of three different soil and
rock types are shown in Fig. 1. Unbonded soils (Curve 1)
show no bond yield. Weakly bonded soils (Curve 2) would be
expected to show a reduction in stiffness beyond Y1 but with
a second clear change in stiffness at bond yield. Soft rocks
(Curve 3) might be expected to show a more significant drop
in stiffness beyond Y1 but with a very clear, more distinct
change in the stiffness decay curve at bond yield.
Malandraki and Toll (2000) found that the position of bond
yield under probing constant 1 and constant p tests was influenced by the rotation of the shearing path direction. Bond
yield, defined from the Etan versus axial strain graph, occurred
at lower axial strains than those observed in constant 3 (conventional drained) tests. The greater the degree of rotation of
the stress path direction, the lower the axial strain at yield.
Different yield loci were observed for each stress path direction. Fig. 2 shows this variation in failure surface and in yield
surface when following constant 3 , constant p, and constant
1 stress paths.
Malandraki and Toll (1996) identified four main zones of
behavior for bonded soils:
In the first zone the bonds entirely control the soils behavior at failure. The failure surface coincides with the
bond yield surface.
In the second zone the bonds only partially control the
soils behavior at failure. The bonded soil exhibits higher
maximum q/p ratios than those of the destructured soil
because of the postyield influence of the bonded structure.
The slope of the failure surface decreases gradually with
an increase of p. In this zone, bond yield is reached before failure; thus, the bond yield surface diverts from the
failure surface.
In the third zone the soils behavior at failure is independent of bonding. Bond yield occurs under shear, and the
failure surface coincides with that of the destructured soil.
In the fourth zone, yield will occur in isotropic compression, and again, the initial bonded structure makes no contribution to the soils strength.
The limits for the different zones were defined by Malandraki and Toll (1996) to be constant 3 paths positioned at
certain stress levels p where the soils behavior changes. Ma-

282 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / MARCH 2001

FIG. 1.

Conceptual Picture of Yield for Bonded Materials [after Toll and Malandraki (2000)]

FIG. 2. Failure Surfaces, Yield Surfaces, and Zones of Bonded Soil Behavior Defined from Drained Probing Tests [after Malandraki
and Toll (2000)]

landraki and Toll (2000) showed that, for constant p and constant 1 tests, the limits between the zones rotated to follow
the stress path direction but the values of p on the isotropic
axis that defined the zones were unchanged (Fig. 2).
Atkinson et al. (1990), from triaxial tests on reconstituted
clays, showed that shear and bulk moduli vary systematically
according to the relative angle developed between the recent
stress path trajectory and the direction of the perturbing path.
Jardine et al. (1991), using data from tests on overconsolidated

intact London clay, showed that the stiffness response depended on the recent stress path directions. Jardine (1994) concluded that the recent stress history has a strong effect on
stiffness within the nonlinear range.
In this paper the work of Malandraki and Toll (2000) is
extended by considering the effects of changes in stress path
direction during shear. The results of drained triaxial tests on
the artificially bonded soil with changes in the stress path direction during shear are analyzed to examine the effects of

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / MARCH 2001 / 283

FIG. 3.

Stress Paths for Triaxial Drained Tests with Changes in Stress Path Directions during Shear

previous shearing path history on the soils behavior with respect to strength, stiffness, and yield conditions.

sheared initially under a constant 1 path, and changed onto a


constant p path.

TEST PROCEDURES

STRESS PATHS AND STRESS-STRAIN BEHAVIOR


FOR BONDED SOIL

The artificial soil developed for this study was prepared using a method similar to that used by Maccarini (1987) and
Bressani (1990); the method of preparation has been described
by Toll and Malandraki (1993). Sand mixed with a small
amount of kaolin (13%) was fired at 500C for 5 h. Under
these conditions the kaolin changes in nature and forms a weak
bond between the sand particles. The samples described in this
paper were all prepared at the same void ratio (e0 = 0.6), the
same as that used for other series of conventional drained and
undrained tests and drained probing triaxial tests (Malandraki
1994; Malandraki and Toll 2000).
The study of the effects of the recent shearing path history
on the artificially weakly bonded soil was performed on 38mm-diameter, 76-mm-high triaxial samples. A stress path triaxial apparatus (Bishop and Wesley 1975) was used with a
computer control system (TRIAX) for the experimental work
(Toll 1993). The computer system controls the stress path cell,
so different shearing paths were followed with accuracy and
without great difficulty. A limitation was that axial strains were
measured using conventional external measurements rather
than sample mounted devices. This meant that small strain
behavior could not be captured well and the first yield points
were not always well defined.
A total of eight drained triaxial tests were carried out on
bonded samples. The samples were first isotropically consolidated to different confining pressures and then sheared initially along a constant 1 path. Shearing for seven of the tests
changed to a constant 3 path after the development of low
percentages of axial strain (0.20.45%). One test that was initially carried out under a constant 1 path changed to a constant p drained path at a > 0.25%. Control of shearing under
the different stress paths took place under a constant rate of
change of deviator stress of 50 kPa/h.
The name of each test indicates the stress paths followed
during shearing and the initial consolidation pressure. Thus,
indicates a sample consolifor example, test c1 (150)
3
dated at p0 = 150 kPa, sheared initially under a constant 1
path, and changed onto a constant 3 path; and test c1
p(300) indicates a sample consolidated at p0 = 300 kPa,

The stress paths for all the tests are plotted in Fig. 3. The
samples were sheared initially along a constant 1 path, followed by a change in direction at higher stresses. The change
in the path direction was initiated at different stress levels for
each test.
The deviatoric stressaxial-strain curves are plotted for all
the tests in Fig. 4. The points of change in the path direction
and the points at which the samples reach the maximum q/p
ratios are also included. Smooth stress-strain curves were followed during shearing along the different stress paths. The
samples showed strain softening behavior and developed shear
surfaces. Tests c1 (300)
and c1 p(300) followed
3
stress paths that finally came together in the effective stress
space, and the samples reached the same maximum value
of q.
EFFECT OF PREVIOUS SHEARING PATH HISTORY
ON TANGENTIAL STIFFNESS
In the following, the stiffness behavior will be shown in
terms of the tangent values of Youngs modulus Etan and shear
modulus Gtan. The stiffness values are normalized with respect
to the current value of mean effective stress p. For unbonded
soils, stiffness is related to stress level and normalization can
remove the effect of stress level. The same normalization is
used for the bonded material to aid comparison between different stress levels.
The development of Etan /p and Gtan /p for test c1
(70)
are shown plotted versus axial or shear strain in a log3
log graph in Figs. 5(a and b). The development of E/p and
G/p for a constant 1 test (which followed the same shearing
path for the first part of shearing) is also included in this graph
to allow comparisons to be made between the different tests.
A small key graph is also presented as an insert in Fig. 5 and
shows the stress paths followed for the two tests under consideration.
Test c1 (70)
presented initially high stiffness values,
3
similar to those presented by test c(70)
(Fig. 5). A first drop
1

284 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / MARCH 2001

FIG. 4.

Deviator Stress versus Axial Strain Curves for Constant Constant


3 and Constant Constant
p Drained Triaxial Tests
1
1

FIG. 5. Normalized Tangential Stiffness versus Strain for


Tests c1 (70)
and c(70):
(a) E/p; (b) G/p
3
1

FIG. 6. Normalized Tangential Stiffness versus Strain for


Tests c1 (70)
, c(11),
and c(35):
(a) E/p; (b) G/p
3
3
3

in stiffness occurred at an axial strain a = 0.13% or a shear


strain s = 0.19%. This loss in stiffness is very close to the
position at which bond yield occurred for test c(70)
. As yield
1
was occurring, the stress path direction was changed to a constant 3 path. For test c1 (70)
, the values of Etan leveled
3
off after the change in path, whereas the values of Gtan actually
increased. In contrast, the constant 1 test showed a further

decrease in E/p and G/p with increasing strain. However, at


a = 0.94% (s = 0.98%), a second drop in stiffness took place
for c1 (70)
.
3
The development of stiffness during the second part of
shearing changed from that of a typical constant 1 test. Because the path changes to a constant 3 path, one can compare
the results with other constant 3 test results. The results for

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / MARCH 2001 / 285

= 0.33% (the point of change in the stress path direction).


During the second part of shearing, E/p continued to decrease
with the increase of axial strain but with a lower slope up to
a = 0.79%. Values of G/p increased sharply with the change
in the stress path direction [Fig. 7(b)] but then started to fall
away with increasing strain. Fig. 7(a) then showed a second
drop in E/p, similar to that which occurred for test c(35)
,
3
but at a lower percentage of axial strain [Fig. 7(a)]. However,
in terms of shear strain [Fig. 7(b)], yield occurred at very
similar values (s = 0.9%) for both tests.
The development of E/p for test c1 (400)
sheared at
3
higher confining stresses is plotted versus axial strain in Fig.
8, and the results for test c(150)
, which followed a similar
3
stress path in the latter stages, are also included. Unfortunately,
transducer readings were not taken sufficiently frequently for
test c1 (400)
; thus, only a limited number of data points
3
are presented. However, a clear pattern of behavior is presented that is similar to that shown by the previous constant
constant
3 tests.
1
Fig. 8(a) shows that, at the beginning of shearing, test c1
(400)
presented higher values of E/p than test c(150)
,
3
3
although values of G/p are more similar [Fig. 8(b)]. A first
drop in E/p took place at a = 0.19%, and stiffness decreased
further, up to the point of the change in the stress path direction. During the second part of shearing, the sample presented
almost constant values of E/p with increasing axial strain up
to a = 0.99%, where the second drop in E/p took place. The
stiffness values were substantially lower than those presented
by test c(150)
, but the second drop in E/p took place at a
3
similar axial strain to that at which bond yield occurred for
the latter test.
It can be seen in Fig. 8(b) that values of G/p were rather

FIG. 7. Normalized Tangential Stiffness versus Strain for


Tests c1 (150)
and c(35):
(a) E/p; (b) G/p
3
3

test c1 (70)
are replotted in Fig. 6, and the development
3
of stiffness for two constant 3 tests, c(11)
and c(35)
, are
3
3
included in the same graph. It can be seen in the key graph
in Fig. 6 that the second part of shearing for test c1
(70)
took place in the stress space between the two conven3
tional drained tests.
During the first part of shearing, test c1 (70)
presented
3
higher E/p values than those developed from tests c(11)
3
and c(35)
[Fig. 6(a)] [similar to those for the equivalent test
3
c(70)]
. However, the G/p values for test c1 (70)
were
1
3
similar to those for c(35)
[Fig. 6(b)]. After the initial loss
3
in stiffness and the change in stress path direction, test c1
(70)
presented similar E/p values to the two conventional
3
drained tests [Fig. 6(a)]. For a > 0.2% the sample maintained
constant E/p values and presented a second drop at higher
strains, which coincides with the point at which bond yield
occurred for test c(11)
and is very close to that for
3
c(35)
. In terms of G/p [Fig. 6(b)], after the initial drop (and
3
then an increase when the path is changed), the values of
G/p became very similar to those for c(35)
and the position
3
of yield is very similar in both tests.
The development of stiffness for test c1 (150)
is plot3
ted versus strain in Fig. 7, and the results for test c(35)
are
3
also included. During the second part of shearing, test c1
(150)
followed a stress path very close to that of test
3
c(35)
. Initially the two samples presented similar E/p values
3
[Fig. 7(a)], although G/p values were higher for c(35)
[Fig.
3
7(b)]. A first drop in E/p was initiated for test c1
(150)
at a = 0.23% or s = 0.3% (similar to that for the
3
constant 1 tests), and stiffness continued to decrease up to a

FIG. 8. Normalized Tangential Stiffness versus Axial Strain for


Tests c1 (400)
and c(150):
(a) E/p; (b) G/p
3
3

286 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / MARCH 2001

erratic after the first drop in stiffness. There seems to be some


increase in G/p on changing the stress path direction, but the
values then drop significantly before recovering. However,
there is some evidence for a change in stiffness behavior at a
shear strain of 1.08% [which coincides with the drop in E/p
at a = 0.99% seen in Fig. 8(a)], but it is difficult to identify
clearly.
Similar behavior was observed from the rest of the tests
(Malandraki 1994). Two main drops in stiffness were observed, each corresponding to the shearing path being followed
at the time.
VOLUMETRIC BEHAVIOR
Yield behavior as demonstrated by the E/p graphs will implicitly reflect changes in volumetric behavior as well as
changes in shear strain. It can be seen from Figs. 58 that the
yield points defined from E/p graphs coincide well with yield
points that can be observed in the G/p plots. This might suggest that the changes in Youngs modulus are a direct reflection
of changes in shear strain, without any contribution from volume change. However, the volumetric behavior will also be
studied to see whether there are changes in volumetric behavior at yield.
Fig. 9(a) shows volume strain plotted against axial strain for
the constant constant
3 tests. A positive volume strain
1
(plotted downward) indicates contraction, and negative values
indicate dilation. It can be seen that all the tests showed dilation during the initial part of the tests when they were fol-

lowing a constant 1 path. This is in agreement with the behavior of constant 1 tests, as can be seen by test c(70)
1
plotted in Fig. 9(b).
The yield points identified earlier from E/p and G/p plots
are shown by symbols in Fig. 9(a). The first drops in stiffness
(while following the constant 1 path) do not seem to show
very significant changes in volume strain behavior. Again this
is similar to the behavior shown by c(70)
[Fig. 9(b)]. How1
ever, slight changes in slope in the volume-strainaxial-strain
plots can be seen [except in c1 (600)]
.
3
As the stress path changes to constant 3 , a significant
change in volumetric behavior can be seen. The volume strain
changes direction in many tests, and the strain increments become contractive, except for c1 (300)
, which neverthe3
less still shows a significant reduction in dilative behavior.
This change in behavior is compatible with the contractive
behavior seen in the initial stages of the constant 3 tests
shown in Fig. 9(b) [c(11)
, c(35)
, and c(150)]
.
3
3
3
The yield points representing the second drop in stiffness
during the constant 3 part of the stress path also show a
distinct change in volumetric behavior. They coincide with
minimum points on the volume-strainaxial-strain plots [or
an upward change in slope for c1 (300)]
. This means
3
that the behavior up to these yield points is generally contractive but dilation is initiated at the point of yield. This is also
in agreement with the behavior observed in constant 3 tests
[Fig. 9(b)] where bond yield is close to the minimum point
representing a change from contraction to dilation.
These changes in volumetric behavior can also be seen in
Fig. 10, where the volume strains are plotted against p. The
constant constant
3 tests show dilative volume strains as
1
a result of reducing mean stress p. Test c1 (70)
shows
3
an identical stress-strain path to that followed by c(70)
up
1
to the point where the stress path changes for c1 (70)
.
3
At the point of change in the stress path direction, the volume
strain increments generally become contractive. When the
bond yield point is reached (when following the constant 3
path), dilative volume strains are induced. The shape of the
volumetric strain-p paths for the latter part of the test follows
the same pattern shown by the constant 3 tests. These initially
show contractive behavior and then dilate after bond yield occurs.
STRESS PATHS AND FAILURE SURFACE PLOTTED
IN STRESS SPACE
The stress paths for the constant constant
3 and con1
stant constant
p tests and the failure envelope for the
1
bonded soil defined from conventional drained and undrained
tests (Malandraki and Toll 2000) are plotted in Fig. 11. The
failure surface for the destructured soil (formed at the same
void ratio as the bonded soil) defined from triaxial drained and
undrained tests is also included.
The samples sheared at low mean stresses (reaching failure
at p < 120 kPa) reach the failure surface for constant 3 tests.
However, samples sheared at higher stresses (reaching failure
at p values between 120 and 460 kPa) sustained lower maximum q/p ratios than the constant 3 failure surface but still
above that defined for the destructured soil. For p > 460 kPa,
the bonded soil sheared under different stress path directions
sustained the same limiting stress ratios as those of the destructured soil.
FIRST AND SECOND LOSS IN STIFFNESS
PRESENTED IN STRESS SPACE

FIG. 9. Volume Strain versus Axial Strain: (a) Tests with


Changes in Stress Path Direction from Constant 1 to Constant
;
3 (b) Constant
1 and Constant
3 Tests

The points of the two major drops in E/p (and G/p) during
the constant constant
3 and constant constant
p tests
1
1
are plotted in stress space in Fig. 12. The failure and the yield

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / MARCH 2001 / 287

FIG. 10. Volume Strain versus Mean Effective Stress for Tests with Changes in Stress Path Direction from Constant 1 to Constant
,
3 Constant ,
1 and Constant
3 Tests

FIG. 11. Stress Paths and Failure Surface for Bonded Soil from Constant Constant
3 Tests and Constant Constant
p Tests
1
1
Compared to Constant 3 Tests and Destructured Soil

surfaces defined from constant 3 , constant p, and constant


1 triaxial tests [from Malandraki and Toll (2000)] are also
included in the plot.
The first drop in stiffness for the tests in which the stress
path directions have been changed during shear is governed
by the yield surface defined from the constant 1 tests. However the second drop in stiffness, for all the samples sheared
under the constant constant
3 tests, closely coincides with
1
the yield surface defined for the constant 3 tests (Fig. 12).
Similarly the second drop in stiffness for the constant
constant
p test coincides with that for the constant p tests.
1
Therefore this suggests that a bonded sample sheared along
different stress paths will yield under each individual shearing
path that is followed in the stress space (Malandraki 1994).

DISCUSSION
The pattern of behavior observed is similar for all the
bonded samples with a previous shearing path history. Initially
the development of stiffness in a constant constant
3 test
1
was similar to that for the equivalent constant 1 test. All the
samples passed through yield before the stress path direction
was changed. During the second part of shearing, the samples
developed E/p values below those presented during a constant
3 test carried out at the same stress level. A second drop in
E/p then occurred, generally, at a lower axial strain than that
at which bond yield takes place in the constant 3 test. Similar
behavior was also shown by the sample sheared along the
constant constant
p path.
1

288 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / MARCH 2001

FIG. 12.

FIG. 13.

First and Second Drop-in Stiffness for Constant Constant


3 Tests Plotted in Stress Space
1

Failure Surface Defined from Constant Constant


3 Tests and Four Zones of Bonded Soil Behavior
1

The yield points defined from E/p graphs coincide with


changes in shear stiffness shown in G/p graphs. In terms of
volume strain, the first drop in stiffness (while following the
constant 1 path) shows only a very small change in slope in
the volume-strainaxial-strain plot. During this portion of the
stress path, the volumetric behavior is dilative, reflecting the
reducing mean stress. However, the second drop in stiffness
(while following the constant 3 path) shows a marked change
in volumetric behavior. After the change in stress path direction, the volumetric behavior is generally initially contractive,
but after bond yield occurs, the behavior changes to dilative.
Yield of the bonded structure of the soil occurs under each
stress path direction followed in the stress space, and the
points of the two major drops in stiffness during the constant
constant
3 and constant p tests coincide with the three
1
yield surfaces previously defined from tests following the three
shearing path directions (Fig. 12). Therefore, it can be seen
that these yield surfaces still hold for samples that have been

previously subjected to yield by shearing along another stress


path direction.
The fact that a sample can show two yield points demonstrates that the process of bond breakdown is anisotropic and
not all bonds break when bond yield is initiated. Only those
bonds subject to shear stresses associated with the current
stress path direction will break. Therefore, on rotating the
stress path direction, unbroken bonds are present that can still
resist shear stresses in the direction imposed by the new path
diection.
The maximum q/p ratios that the bonded soil can sustain
under the different shearing paths were also influenced. The
failure surfaces for the bonded soil from the constant
constant
3 tests, from constant 3 tests, from constant
1
1 tests, and for the destructured soil are plotted in Fig. 13.
The upper boundaries for Zones 1 and 2 [as defined by Malandraki and Toll (1996)] from constant 3 and constant 1
tests are also included.

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / MARCH 2001 / 289

The constant constant


3 tests that reach failure at p <
1
120 kPa sustained the same maximum q/p ratios as the
drained constant 3 and constant 1 tests (note that at this
stress level the failure surfaces for the three stress paths are
coincident, as shown in Fig. 2). This stress level ( p = 120
kPa) coincides with the position of the upper limit of the first
zone (where the failure surface coincides with the bond yield
surface) defined from the constant 3 tests. The rest of the
samples sheared at stress levels that reach failure at p > 120
kPa could not sustain such high limiting stress ratios, and the
failure surface falls closer to that of the destructured soil. For
p > 460 kPa, the soils behavior is governed by that of the
destructured soil. This sets the upper limit for Zone 2 for this
particular shearing path history.
It can be seen that the new failure surface is positioned
between the failure surface defined from the constant 3 triaxial drained tests on bonded soil and the failure surface defined for the destructured soil. It can also be seen that the
failure surface for the constant constant
3 tests converges
1
with the failure surface for the destructured soil (thus defining
the upper limit for Zone 2) at a point between the Zone 2
limits defined for the constant 1 and constant 3 tests. The
new failure surface converges at p = 460 kPa, a much lower
level than for constant 3 tests, which converges at p = 950
kPa. However, it is greater than for the constant 1 tests, where
convergence takes place at p = 330 kPa. Therefore, shearing
a soil along two stress path directions so that yield is induced
twice (on two different paths) produces a loss in strength
within the second transitional zone. The available strength is
smaller than samples sheared to failure entirely along the constant 3 path. However, samples are stronger than if they had
been sheared to failure entirely along a constant 1 path.
Leroueil and Vaughan (1990) and Gens and Nova (1993)
suggested that destructuration of bonded materials can be related to both positive (compressive) and negative (dilatent/
swelling) volumetric strains. Certainly the data presented here
show that yield occurs under both conditions, although it may
well be that it is shear strains that contribute most to bond
breakdown. When following a constant 1 path, yield occurs
while dilatent volume strains are present. In this case the volumetric behavior is only slightly modified by the yield process.
However, when following a constant 3 path, yield occurs
where the volume strain increments are generally contractant.
In this case the volumetric behavior changes to dilation after
yield. It could be that the change in volumetric behavior from
contraction to dilation was responsible for inducing yield.
However, it does seem that the yield observed from E/p and
G/p plots generally slightly precedes the change in volumetric
behavior (Figs. 9 and 10). This suggests that yield induces the
change in volumetric behavior rather than vice versa.
It is interesting to consider the nature of the bond yield
surface, whether it is static (i.e., uniquely located in q-p
space) or kinematic (i.e., moving and evolving with the current
stress state) (Jardine 1992). One might expect that the bond
yield surface would be static, determined only by the strength
of the bonds. The fact that Malandraki and Toll (2000) have
shown that different yield conditions exist for different stress
paths clearly shows that the surface is anisotropic. However,
they also point out that the structure of the bonded soil cannot
account for this because the initial structure of the artificial
weakly bonded soil before shearing (as indicated by microscopic examination) appears isotropic. This suggests that the
initial bond yield surface would be isotropic but that an anisotropic yield surface is developed during shear. This suggests
a kinematic surface.
However, the data presented here show that a soil that has
been subject to a different stress path history (even passing
through yield along a previous stress path) reaches yield at the

same yield point as one that has been sheared entirely along
the final stress path. This would not be expected of a fully
kinematic surface that was being moved and changed by previous stress history. Therefore it is suggested that the bond
yield surface is an expandable/shrinkable surface rather than a
moveable surface. The initial bond yield is modified depending
on the direction of the strain increments. It can perhaps be
postulated that the yield surface expands when volumetric
strains are compressive and shrinks when the volumetric
strains are dilatent. This would be compatible with the results,
as the bond yield surface is larger for constant 3 tests (when
volume strains are contractive) and smaller for constant 1
tests (when volume strains are dilatent).
In practical terms, it is demonstrated that the position of the
bond yield surfaces are not affected by changing the stress
path direction during shear (Fig. 12). However, the failure envelope is very distinctly affected by the changes in the stress
path direction (Fig. 13). The strength available is reduced if
the soil is subject to a complex stress path. This suggests that,
if a structure was being designed (in or on bonded soil) where
the major requirement was limiting ground movements, data
from conventional triaxial tests might not be significantly in
error. However, if the design criteria was a limiting strength,
data from conventional triaxial tests could significantly overestimate the available strength, leading to an unconservative
design.
CONCLUSIONS
The behavior of the bonded soil is strongly influenced by
clockwise changes in the stress path direction during shearing.
Yield of the bonded structure of the soil occurs along each
stress path direction followed. Even after yield has occurred
when following one stress path direction, a further yield is
observed when the stress path direction is changed. This demonstrates that the process of bond breakdown is anisotropic
and not all bonds break when bond yield is initiated. Only
those bonds subject to shear stresses associated with the current stress path direction will break. Therefore, on rotating the
stress path direction, unbroken bonds are present that can still
resist shear stresses in the direction imposed by the new path
direction.
The position of yield was found to be independent of the
previous shearing path history of the soil and occurred at
points that corresponded to a yield surface defined for the current shearing path direction. However, the previous shearing
path history of the soil did significantly affect the failure envelope. The strength available was reduced if the soil was
subjected to a complex stress path. Therefore, for complex
stress paths that may be present in the field, bonding degradation should be considered if the design is based on a limiting
strength. In this case, data from conventional triaxial tests
could significantly overestimate the available strength, leading
to an unconservative design. However, if a structure is being
designed (in or on bonded soil) where the major requirement
is limiting ground movements, then data from conventional
triaxial tests might not be significantly in error.
It is suggested that the bond yield surface is kinematic, in
the sense that it is an expandable/shrinkable surface, but that
it is not a movable surface. It is postulated that the yield surface expands when volumetric strains are compressive and
shrinks when the volumetric strains are dilatent.
ACKNOWLEDGMENTS
The research described in this paper was partially financed by an
E. C. Human Capital and Mobility fellowship.

290 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / MARCH 2001

APPENDIX.

REFERENCES

Atkinson, J. H., Richardson, D., and Stallebrasse, S. E. (1990). Effect


of recent stress history on the stiffness of overconsolidated soil. Geotechnique, London, 40(4), 531540.
Bishop, A. W., and Wesley, L. D. (1975). A hydraulic triaxial apparatus
for controlled stress path testing. Geotechnique, London, 35(4), 657
670.
Bressani, L. A. (1990). Experimental properties of bonded soils. PhD
thesis, University of London, London.
Jardine, R. J. (1992). Some observations on the kinematic nature of soil
stiffness. Soils and Found., Tokyo, 32(2), 111124.
Jardine, R. J. (1994). One perspective of the pre-failure deformation
characteristics of some geomaterials. Proc., Int. Symp. on Pre-Failure
Deformation Characteristics of Geomaterials, Keynote Lecture 5, S.
Shibuya, T. Mitachi, and S. Miura, eds., Preprint Volume, 151182.
Jardine, R. J., St. John, H. D., Hight, D. W., and Potts, D. M. (1991).
Some practical applications of a non-linear ground model. Proc.,
10th Eur. Conf. on Soil Mech., Vol. 1, 223228.
Leroueil, S., and Vaughan, P. R. (1990). The general and congruent
effects of structure in natural soils and weak rocks. Geotechnique,
London, 40(3), 467488.
Maccarini, M. (1987). Laboratory studies of a weakly bonded artificial
soil. PhD thesis, University of London, London.
Malandraki, V. (1994). The engineering behaviour of a weakly bonded
artificial soil. PhD thesis, University of Durham, Durham, U.K.
Malandraki, V., and Toll, D. G. (1994). Yielding of a weakly bonded

artificial soil. Proc., Int. Symp. on Pre-Failure Deformation Characteristics of Geomaterials, S. Shibuya, T. Mitachi, and S. Miura, eds.,
Vol. 1, Balkema, Rotterdam, The Netherlands, 315320.
Malandraki, V., and Toll, D. G. (1996). The defininition of yield for
bonded materials. Geotech. Geological Engrg., Dordrecht, The Netherlands, 14(1), 6782.
Malandraki, V., and Toll, D. G. (2000). Drained probing triaxial tests
on a weakly bonded artificial soil. Geotechnique, London, 50(2), 141
151.
Toll, D. G. (1993). A computer control system for stress path triaxial
testing. Developments in civil and construction engineering computing, B. H. V. Topping, ed., Civil-Comp Press, Edinburgh, 107113.
Toll, D. G., and Malandraki, V. (1993). Triaxial testing of a weakly
bonded soil. Proc., Int. Symp. on Geotech. Engrg. of Hard Soils-Soft
Rocks, A. Anagnostopoulos, F. Schlosser, N. Kalteziotis, and R. Frank,
ed., Vol. 1, Balkema, Rotterdam, The Netherlands, 817824.
Toll, D. G., and Malandraki, V. (2000). The engineering behaviour of
structured soils. Proc., 1st Central Asian Geotech. Symp., Vol. 2, 629
634.
Vaughan, P. R. (1985). Mechanical and hydraulic properties of in situ
residual soilsGeneral Report. Proc., 1st Int. Conf. on Geomechanics
in Tropical Lateritic and Saprolitic Soils, Brasilia, Vol. 3, Brasilian
Society for Soil Mechanics, 231263.
Vaughan, P. R. (1988). Characterising the mechanical properties of the
in-situ residual soil. Proc., 2nd Int. Conf. on Geomechanics in Tropical Soils, Vol. 2, Balkema, Rotterdam, The Netherlands, 469487.

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / MARCH 2001 / 291

You might also like