Documentos de Académico
Documentos de Profesional
Documentos de Cultura
Mechanical properties and behaviour of a partially saturated lime-treated, high
plasticity clay
Xiwei Zhang, Maria Mavroulidou, M.J. Gunn
PII:
DOI:
Reference:
S0013-7952(15)00161-1
doi: 10.1016/j.enggeo.2015.05.007
ENGEO 4044
To appear in:
Engineering Geology
Received date:
Revised date:
Accepted date:
3 March 2014
27 April 2015
9 May 2015
Please cite this article as: Zhang, Xiwei, Mavroulidou, Maria, Gunn, M.J., Mechanical properties and behaviour of a partially saturated lime-treated, high plasticity clay,
Engineering Geology (2015), doi: 10.1016/j.enggeo.2015.05.007
This is a PDF le of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its nal form. Please note that during the production process
errors may be discovered which could aect the content, and all legal disclaimers that
apply to the journal pertain.
ACCEPTED MANUSCRIPT
IP
SC
R
NU
MA
a. Address during the presented research: London South Bank University, 103 Borough
Road, London, SE1 0AA, UK Present address: Associated research fellow, Key Laboratory of
TE
CE
P
b. Reader of Geotechnical Engineering, London South Bank University, 103 Borough Road,
London, SE1 0AA, UK
AC
figure captions)
Number of Figures: 17 figure captions
Number of Tables: 8
ACCEPTED MANUSCRIPT
*Corresponding author
SC
R
IP
Dr Maria Mavroulidou
NU
MA
AC
CE
P
TE
Email: mavroum@lsbu.ac.uk
ACCEPTED MANUSCRIPT
IP
SC
R
ABSTRACT: This paper investigates the effect of suction and lime treatment on the volumetric and
shear behaviour of a partially saturated high plasticity clay (London Clay). A series of triaxial tests
were performed on statically compacted London Clay and lime-treated London Clay specimens. These
NU
were prepared at the same target void ratio and tested under partially saturated conditions. The tests
MA
concerned isotropic compression and two different types of shearing tests (a) shearing at constant
suction and (b) shearing at constant water content, using the axis translation technique. Based on the
results, the effects of suction and lime-induced bonding are evaluated and quantified within an
CE
P
TE
Keywords: lime-treated clay; partially saturated soil; suction controlled triaxial testing
AC
Notation
ACCEPTED MANUSCRIPT
h specimen height
M critical state stress ratio
IP
M critical state stress ratio with respect to the mean net stress (partially saturated soil)
SC
R
Mb critical state stress ratio with respect to the matric suction (partially saturated soil)
N(s) the intercept of the normal compression line
NU
na and nb components of shear strength associated respectively with (p-ua) and (ua uw)
p mean stress
MA
TE
CE
P
AC
ACCEPTED MANUSCRIPT
v soil specific volume
vw specific volume of water
IP
SC
R
MA
NU
TE
d dry density
CE
P
and
internal angle of friction, peak angle of friction, and critical state angle of friction
AC
, p , c
friction angle terms associated respectively with net stress and matric suction effects
1 Introduction
Chemical soil improvement using additives such as lime is a technique that has been
used extensively in construction, most commonly for pavement applications. With
this application in mind, most international literature on lime treated soils focuses on
simple tests such as California Bearing Ratio (CBR) or unconfined compressive
5
ACCEPTED MANUSCRIPT
strength (UCS) tests. With an increasing use of the technique in a wider range of
applications, the need has emerged for more sophisticated testing and thorough
IP
SC
R
describe the behaviour of lime-treated soils under saturated and partially saturated
conditions. The latter aspect is particularly relevant, as lime-treated soils are typically
compacted after treatment and hence, by definition, partially saturated. Therefore
NU
MA
To the Authors knowledge such data has been lacking in the literature.
The aim of this paper is to present results from an extensive experimental programme
TE
CE
P
AC
2. Background
2.1 Lime treatment of clayey soils
The treatment of clayey soils with lime is a widely used technique of ground
improvement. The improvement consists in a more or less instantaneous reduction in
plasticity (Sherwood, 1993) as well as shrinkage and swelling characteristics
(provided that sulphates are not present) (e.g. Bell, 1996; Al-Rawas et al, 2005).
6
ACCEPTED MANUSCRIPT
These immediate/short term effects on the soil upon lime treatment, commonly
mentioned as lime modification (NLA, 2004), are attributed to rapid ion exchange
IP
reactions between exchangeable clay ions and calcium ions provided by the lime.
SC
R
Cation exchange would be the first step towards more permanent changes; following a
modification of the electrolyte content in the water due to the increased exchangeable
calcium ion concentration, flocculation and agglomeration of the soil particles occurs
NU
and transforms the plastic soil to a granular and less plastic material (Bell, 1996). It
MA
was also argued that this flocculation could also be due to the early formation of small
quantities of calcium silicate or calcium aluminate hydrates; these could create some
bridging between particles and consequently flocculation (Diamond & Kinter, 1965).
TE
This could justify the increase in CBR and UCS shortly after lime addition, frequently
CE
P
If enough lime beyond the Initial Consumption of Lime (ICL) content is present (i.e.
AC
the lime percentage which increased the pH of the soil to 12.35-12.45), UCS and CBR
were shown to continue to increase potentially over long time periods, through slower
pozzolanic reactions between lime, silica and alumina, which produce cementing
agents (Brandtl, 1981; Bell, 1996; Sherwood, 1993). These reactions are usually
reported as stabilisation reactions (NLA, 2004, Sherwood, 1993). They are caused
by the highly alkaline environment upon lime addition, which promotes dissolution of
siliceous and aluminous compounds from the clay mineral lattice. The compounds
dissolved from the clay mineral lattice react with calcium ions in pore water to form
calcium silicate hydrates, calcium aluminate hydrates and hydrated calcium alumino7
ACCEPTED MANUSCRIPT
silicates, which coat the soil particles and subsequently crystallise to bond them (Bell,
1996). However strength does not increase linearly with lime content; it may in fact
IP
SC
R
amount of silica; when all silica in the clay is used up, further lime addition would not
result in the formation of any new cementation products; thus any further lime may
reduce strength as lime has no good frictional properties (Bell 1988; Brandtl 1981;
NU
Hausmann 1990).
MA
An additional reaction between soil and lime is carbonation in the presence of carbon
dioxide. Namely, when the CO2 is dissolved in the soil pore water it reacts with the
TE
hydroxyle ions, forming carbonate ions, which subsequently react with the calcium
ions. This results in the formation of caclium carbonate CaCO3, a weak cement whose
CE
P
formation is normally undesirable, as the reaction consumes lime which would have
otherwise been used in pozzolanic reactions for the formation of stronger cementitious
AC
bonds. Another undesirable effect of carbonation is the fact that it delays penetration
of ions on the surface of the clay and increases the time for these to reach the reaction
sites (Barker, 2002).
ACCEPTED MANUSCRIPT
1990; Toll, 1990; Wheeler & Sivakumar, 1995; Cui & Delage, 1996; Sharma, 1998;
Romero, 1999; Rampino et al., 2000; Cabarkapa, 2001, Jotisankasa, 2005). It is
IP
common knowledge that suction increases the shear strength of the soil. The efforts of
SC
R
NU
soils was proposed by Bishop (1959), using an effective stress approach. Later
MA
TE
independent stress state variables (net stress and matric suction), written as:
(Eqn. 1)
CE
P
where f refers to the shear strength of the soil, c ' is the cohesion (true or apparent),
the net stress normal to the shear plane at failure, is the angle of
AC
ua f is
internal friction related to the applied net normal stress and considered equal to that of
saturated soils, and b a friction angle term associated with matric suction ua uw
effects. Based on data fitting from other researchers work, within suction ranges
between 0 and 200 kPa, early work by Fredlund et al. (1978) suggested that the value
of b would be a constant. Later, new experimental evidence from tests including
higher suctions, demonstrated that the increase of shear strength due to suction
becomes nonlinear when the range of suction is extended to large values (Escario &
Saez, 1986; Fredlund et al., 1987; Cabarkapa, 2001). Modifications of the above
9
ACCEPTED MANUSCRIPT
criterion were therefore suggested to introduce an appropriate expression for the non-
IP
linear variation of the parameter b (Fredlund et al, 1996, Vanapalli et al, 1996).
SC
R
NU
saturated soil depends on the initial and final stress state with respect to the mean net
MA
stress and matric suction, as well as on the particular path followed from the initial to
final state (Alonso et al.1990). The yield stress increases with increasing suction; the
slope of the loading and unloading-reloading curve of a partially saturated soil is not
TE
the same as that of the same soil in a saturated state (i.e. suction tends to stiffen the
CE
P
partially saturated soils against the loading). This behaviour is similar to that
manifested in natural fully saturated cemented soils, as a result of cementation. As for
AC
On the other hand, concerning the effect of the suction under a certain constant mean
net stress (p-u), it was shown that a change in suction may induce irrecoverable
volumetric strain for most clay soils. A soil subjected to a wetting or drying path (i.e.
decreasing or increasing suction respectively), may swell or shrink, under the
combined effect of suction and mean net stress (i.e. it may either swell at low mean
10
ACCEPTED MANUSCRIPT
net stress values or collapse upon wetting at high mean net stress values, and contract
upon the drying path, as described e.g. in Matyas & Radhakrishna, 1968 or Alonso et
IP
al. 1990). Upon drying, shrinkage is generally observed for most soil specimens that
SC
R
are initially saturated, often with an apparent yield point beyond which an irreversible
volumetric contraction can be observed. Experimental evidence shows that this
threshold value (yield suction) depends both on the suction history and the initial void
NU
ratio (Chen et al, 1999). For compacted expansive clays the volume change due to
MA
variations in suction upon drying/wetting was found to be much larger than for nonexpansive clays. Substantial irreversible shrinkage or swelling strains may occur.
TE
compaction pressure and subsequent stress paths (as a function of suction or net
CE
P
stress) and not solely on the soil type (e.g. Cui et al, 2002).
AC
Whereas the above findings have now been established for compacted soils that have
not been subjected to chemical treatment, there is paucity in experimental evidence
concerning the behaviour of partially saturated chemically stabilised soils. For such
soils, both stiffness and shear strength behaviour are likely to be affected by the
relative magnitude and interaction between chemically-induced bonding and suction.
It is therefore of interest to investigate how the trends followed by these soils relate to
the findings concerning untreated partially saturated soils.
ACCEPTED MANUSCRIPT
3.1 Materials
The soil used in this study was London Clay, a high plasticity stiff overconsolidated
IP
marine clay (in its natural state) extensively encountered in construction in the
SC
R
London area and the South Eastern England (a very densely populated area with
intensive industrial activity) including pavement construction, airports (e.g. Heathrow
Terminal 5), underground railway (an example of recent engineering works being the
MA
NU
The London Clay samples used in this study came from an excavation at Westminster
Bridge in London and depths between 30-31 m below ground level. The soil was air-
TE
dried at an average temperature of 220 C and a relative humidity of 60% for a month
CE
P
and was subsequently pulverised. Figure 1 shows the particle size distribution of the
portion of two samples of the pulverised soil passing the BS 425 m sieve (BSI
AC
1990a), (represented by the two different curves in Fig. 1). The composition of the
soil is shown in Table 1. Note the presence of smectite, which causes the London Clay
soil to manifest a swelling/shrinking soil behaviour.
Commercially available hydrated lime was used in this research. Its suitability for soil
12
ACCEPTED MANUSCRIPT
stabilisation was confirmed (BSI 1990b). The relative proportion of calcium
hydroxide to calcium oxide was found to be 4.88:1.00 based on chemical analysis on
IP
the lime sample carried out in duplicate. Plasticity tests performed on London Clay
SC
R
mixed in dry condition with lime at percentages of 0%-8% lime by dry mass of soil
showed no change in the plasticity characteristics of the lime-treated soil beyond
approximately 4% of lime addition. Hence 4% was considered to be the minimum
NU
necessary lime percentage for treating this clay. The percentage was confirmed by
MA
initial consumption of lime (ICL) test results (BSI 1990b) which showed that the ICL
of this soil was between 3 and 4% (see Mavroulidou et al, 2013a and b). Therefore an
amount of 4% lime by dry soil mass was used to treat this soil. The main physical
CE
P
TE
AC
Table 2 Physical characteristics of untreated and 4% lime treated London Clay soils
ACCEPTED MANUSCRIPT
(saturated soil testing at zero suction) two different specimen sizes were used (see
Table 3), it was subsequently found that all strength results plotted as unique strength
IP
lines and hence the difference in the specimen size did not have an effect on the
SC
R
results. Similarly, the results of the isotropic compression tests conducted on larger
specimens (100 mm height and 50mm diameter) were confirmed to be consistent with
the isotropic stage results of the smaller triaxial testing specimens (of 76 mm height
MA
NU
and 38 mm diameter).
The soil was placed in the split moulds in six or eight equal layers (for the 76 mm and
TE
rate of 1 mm/min until the required height was reached. The loading ram was then
CE
P
held in contact with the soil for another five minutes to reduce the rebound upon
unloading (Jotisankasa 2005). This static compaction procedure was selected as the
AC
best way of exerting sufficient control over the compaction process of a clayey soil
(Sivakumar, 1993; Sharma, 1998; Jotisankasa, 2005), so that almost identical
specimens were prepared. It was observed that the two soils (treated vs. untreated) had
different dynamic (standard Proctor compaction) characteristics due to the effect of
lime treatment (see Table 2). Thus for consistent comparisons and due to the length of
the tests limiting the amount of feasible investigations within the research time scales,
a decision had to be made to keep either the same target compaction dry density or the
same compactive effort for the two soils during static compaction, while
acknowledging that the variation of either parameter would have some effect on the
14
ACCEPTED MANUSCRIPT
resulting properties and behaviour of the compacted soils (see e.g. Chen et al. 1999;
Sivakumar and Wheeler, 2000; Alonso and Pinyol, 2008). It was decided to compact
IP
statically all specimens at the same target dry density of 1.43 g/cm3 (corresponding to
SC
R
the maximum standard Proctor dry density of the untreated soil) and not at each soils
(i.e. untreated vs. treated soil) Proctor optimum characteristics (although this resulted
in a higher compactive effort for the lime-treated soil, i.e. 550 kPa for the untreated
NU
vs. 1000 kPa for the treated soil). The water content for both types of specimen was
MA
kept to the dry side of the respective Proctor optimum to ensure that the resulting
structure after compaction was qualitatively similar for the two soils. Thus for the
untreated London Clay the water content was about 25% (which is slightly drier of the
TE
Proctor optimum); for the lime treated specimens an additional 2% of water was used
CE
P
(also dry of the Proctor optimum of the lime treated soil), to ensure that enough water
was available for chemical reactions, considering that the optimum water content of
AC
the treated soil was higher than that of the untreated soil (i.e. 30% for the treated vs
26% for the untreated soil). Using the filter paper method the suctions of the ascompacted specimens were determined and found to be fairly consistent i.e.
approximately 650 kPa for untreated London Clay and 600 kPa for 4% lime treated
London Clay specimens, with average as-compacted degrees of saturation of about 74
% and 78% respectively for the two soils.
After compaction, the lime treated specimens were left to cure for the required time
(one week) in several layers of cling film and stored at controlled environmental
15
ACCEPTED MANUSCRIPT
conditions (constant temperature and humidity). The required curing time of one week
was determined based on prior Unconsolidated Undrained (UU) triaxial testing. This
IP
showed that for this amount of lime (4%) and curing method, and for six different
SC
R
curing periods between 1 day and 166 days (the latter period covering the typical
duration of the tests presented here), curing beyond seven days did not result in any
MA
NU
further improvement of the shear strength of the soil, suggesting no further evolution
Fig. 2 Deviator stress vs axial strain UU testing plots for different curing periods
TE
The characteristics of triaxial testing specimens after compaction as well as the list of
CE
P
tests presented are summarised in Table 3. Note that in places the tests will be
discussed in comparison with results based on a set of triaxial tests conducted at s=
AC
0kPa. These were published elsewhere (Mavroulidou et al, 2011); hence they are not
shown in detail in this paper.
Figure 3 shows indicative Soil Water Retention Curves derived from filter paper
testing of compacted specimens (subject to subsequent saturation, followed by a
drying and then a wetting path). These curves are included here to illustrate the
partially saturated behaviour of the two soils (untreated and treated) and place these in
16
ACCEPTED MANUSCRIPT
the context of the partially saturated soils analysis carried out in the following sections
of the paper. From the figures it can be seen that according to the filter paper tests the
IP
SC
R
saturated in the range of suctions considered. It can also be seen that the treated soil
appears to present a point of maximum curvature / air entry value at lower suctions
compared to the untreated soil (which is expected due to the flocculation of the
NU
particles induced by lime). Conversely the slope of the curves of the treated and
MA
untreated soils beyond these points of maximum curvature (which is linked to the
CE
P
TE
Fig. 3 Indicative Soil Water Retention Curves based on filter paper tests
AC
As the suctions of the specimens at compaction were higher than the target suctions
for triaxial testing, the specimens had to be wetted (following a suction
unloading/decrease path) to achieve the target suction levels for testing, namely 0 kPa
(saturation), 50, 100, 200 and 300 kPa respectively. A small mean net stress of 20 kPa
was applied throughout the suction equalisation stage, to minimise specimen
disturbance, avoid damaging of the membranes and ensure a good contact of the soil
with the ceramic disk interface. Based on Sivakumar (1993) the suction equalisation
stage was considered to have been completed when the rate of the pore water volume
change was less than 0.044 cm3/d and 0.1 cm3/d for specimens of 38 mm diameter and
17
ACCEPTED MANUSCRIPT
76 mm height, and specimens of 50 mm diameter and 100 mm height respectively.
When the above flow criteria were met, the particular suction level was maintained
IP
SC
R
throughout the specimen. This also gave the specimen enough time to deform under
the applied suction. The suction equalisation period typically varied from 10 to 30
NU
MA
consolidation). During loading, the mean net stress (p-ua) was incremented at a
TE
sufficiently slow rate to minimise the development of excess pore water pressure and
CE
P
maintain constant suction during testing (Zhan, 2003). The maximum excess pore
pressure uex induced in the partially saturated soil by the loading, was calculated as
AC
(Sivakumar, 1993):
uex
RL h 2
2cv
(Eqn. 2)
where uex is the excess pore water pressure; RL is the loading rate of the ramped
consolidation; h is the height of the specimen; cv is the coefficient of consolidation.
Taking into account the above equation and based on average cv values derived from
preliminary tests, a rate of 0.6 kPa/h on the mean net stress was used for the isotropic
18
ACCEPTED MANUSCRIPT
loading/unloading tests; its adequacy was confirmed by the monitoring of the pore
water pressures during testing, which showed that this rate allowed for excess pore
SC
R
IP
NU
varying between 100-300 kPa); during these tests the specimens were sheared at a
MA
very slow rate of 1.14 m/min, following a q/(p-ua)=3 path, while maintaining a
constant cell pressure and a constant suction during the shearing (confirmed by the
monitoring of the pore water pressures during shearing). The rate of shearing was
TE
CE
P
addition, three constant water content (CW) tests were performed under undrained
conditions at a constant rate of axial strain (2.42 m/min). The faster rate adopted for
AC
the latter tests was justified as these were undrained tests. During shearing the pore
water pressure was measured, and hence the suction changes were determined.
In addition to the triaxial tests, two suction unloading (wetting path) tests were
performed in the triaxial cell under two different constant mean net stresses of 100
kPa and 200 kPa. They aimed at studying the effects of the mean net stress on the
deformation of the lime-treated soil during wetting. After suction equalisation at the
target suction, the specimens were compressed isotropically up to the target mean net
stress of 100 kPa or 200 kPa (see column 5 of Table 3). During the suction unloading
19
ACCEPTED MANUSCRIPT
/decrease tests (following a wetting path) performed in the triaxial cell, the above
mean net stresses were then kept constant (and so was the air pressure) while the pore
IP
water pressure changed at a specified rate 1 kPa/hour to the respective target value
SC
R
resulting in a decrease in suction. This was followed by a 2-4 day period during which
suction was maintained for excess pore water pressure equilibrium within the
specimen. Water volume changes and overall volume changes were recorded during
NU
the testing. To find the suction decrease line, the results of these tests were
MA
complemented by filter paper results, not discussed here in detail as many of these
were published elsewhere (Zhang et al, 2010; Mavroulidou et al., 2011; Mavroulidou
TE
CE
P
4. Presentation of results
AC
The presented results refer to wetting induced-swelling from different tests and hence
conditions, namely: zero mean net stress (based on filter paper results not detailed
here), 20 kPa net stress (obtained from the suction equalisation stage preceding
triaxial testing) and mean net stress of 100 kPa and 200 kPa (wetting soil water
retention curves obtained using the triaxial cell). For the purposes of comparison the
results of the wetting tests at all different net stresses in terms of specific volume
versus matric suction (v-s), are plotted together in Fig. 4 (a) and (b). The latter plots
the same results of the in a semi-logarithmic scale (v: ln[(s+pat)/pat)], normalised for a
20
ACCEPTED MANUSCRIPT
reference pressure pat (atmospheric pressure). This representation will be used for the
analyses of the results shown in section 5.1. To indicate qualitatively the differences
IP
during swelling between the untreated and the lime-treated soil, specific volume
SC
R
versus suction results from filter paper testing and some available results of wetting
under 20 kPa net stress of untreated London Clay soil are also added. From Figures
4(a)-(b) it can be seen that although the lime-treated soil still shows some swelling
NU
under low confinement stresses, this is considerably reduced across the suction ranges
MA
TE
Fig. 4. Specific volume vs. suction: (a) arithmetic scale; (b) semi-logarithmic scale
CE
P
Based on the curves of the wetting path results from the filter paper specimens (i.e. at
a zero mean net stress), it can be seen that upon wetting, the specific volume initially
AC
appears to increase rather linearly, but after a suction of approximately 200 kPa was
reached, some considerable swelling was observed, indicating yielding due to wetting.
The specific volume after wetting versus suction under the 20 kPa mean net stress
curves (i.e. from the suction equalisation stage in the triaxial test), shows deformation
patterns similar to those of the tests under zero mean net stress tests but the suction
yield stress point can now be identified at about 180 kPa. The maximum specific
volume is about 2.06 (i.e. much smaller than that for tests under zero net stress) as
swelling was partly suppressed due to the higher mean net stress. Specimens LTLCW1 and LTLC-W2 subject to wetting under a mean net stress of 100 and 200 kPa
21
ACCEPTED MANUSCRIPT
respectively, showed swelling volumetric strains of 0.7% and 0.3% for a mean net
stress of 100 kPa and 200 kPa, respectively, i.e., with an increase in the mean net
IP
stress, the swelling deformation decreased. For the specimen LTLC-W1 tested at 100
SC
R
kPa mean net stress, a likely yield point during wetting was identified at about 90-100
kPa suction corresponding to an increase in the rate of change of the specific volume
during wetting (swelling). For the specimen tested at a mean net stress of 200 kPa, the
NU
yield suction was not possible to identify as the swelling deformation curve was
MA
linear; it will therefore be assumed to be zero in the analysis of the results (see section
TE
5.1).
CE
P
AC
the starting points in the plot were controlled by the specific volumes of the
specimens at the end of the suction equalisation stage (i.e., after swelling) shown in
Fig 4(a) and (b)).
c)
Casagrandes method (see Table 4). These yield stress points were subsequently used
to establish the LC yield curve in the s: space, as discussed later (see section 5.2).
22
ACCEPTED MANUSCRIPT
From Table 4, it can be seen that the yield stresses of the lime-treated soil were higher
than that of the untreated soil at the same suction level, which is consistent with the
IP
behaviour of bonded geomaterials. It can also be seen that, as for the untreated
SC
R
partially saturated soil, yield stress increased with suction. However, although suction
increased the yield stress for both soils, for the lime-treated soil the additional effect
of chemically-induced bonding was proven to be considerable, comparing the yield
NU
stress values of the two soils at the same suction level. Table 4 also shows the values
MA
of N(s) and (s) (i.e. respectively the intercept at the reference pressure and the slope
of the normal compression line) for different suctions. For uncemented partially
saturated soils, these values are commonly used to describe the post-yield
CE
P
relationship, i.e.:
TE
compression curves, for which Wheeler and Sivakumar (1995) suggested a linear
v N ( s) ( s) ln
p
pat
(Eqn. 3)
AC
where the atmospheric pressure, pat =100 kPa, was used as a reference pressure to
make the expression dimensionally consistent;
Note that the slope (s) value for the lime-treated soil was not clearly identifiable as
the curves were slightly non-linear. This is realistic because, if extrapolated to higher
mean net stresses that the available equipment for this study could not match, the
gradient (s) during the de-bonding processing would be expected to continually
increase with the mean net stress until the cemented soil curve coincides with that of
23
ACCEPTED MANUSCRIPT
the uncemented soil when the breakage of all cementation bonds is complete. In this
study, the compressibility line of the cemented lime-treated soil did not eventually
IP
converge to that of the untreated soil. This implies that the range of pressures applied,
SC
R
only captured perhaps some initial yield linked to the beginning of a gradual, partial
breakage of the cementation bonds but not that corresponding to full destructuration.
According to Rao & Shivananda (2005), who tested saturated lime-treated black
NU
MA
destructuration was reached at pressures of 3-13 MPa, far beyond the ranges
TE
An interesting observation made based on Figure 5, is that for the lime-treated soil
CE
P
there is no clear indication that suction allows the Normal Compression Line (NCL)
to cross the zero suction NCL of the lime-treated soil (assuming extrapolated NCL
AC
curves due to the limited extent of the actual experimental curve). This is unlike the
behaviour of uncemented partially saturated soils reported in the literature (e.g.
Alonso et al., 1990; Wheeler & Sivakumar, 1995), which was also noted for the
untreated London Clay soil tested here (see Fig. 5). The isotropic compression stage
results of all subsequent shearing triaxial tests (not shown here for the sake of brevity)
based on additional specimens, were very consistent with the independent isotropic
compression testing results shown in Figure 5, confirming the findings. Moreover the
findings are consistent with those obtained from suction-controlled oedometer
apparatus (K0 compression) testing for the same soils (which reached higher
24
ACCEPTED MANUSCRIPT
compression levels, up to 2000 kPa) published previously (Mavroulidou et al 2013a).
These also showed that the partially saturated lime-treated soil NCL did not cross the
IP
zero suction NCL. This would indicate that the behaviour of the lime-treated soil in
SC
R
isotropic compression for the ranges of suction studied was largely controlled by the
lime-induced bonding rather than suction. For this statement to be conclusive the data
MA
NU
needs to be extrapolated to yet higher pressures, which were beyond the capacity of
Overall, based on the results in Table 4 and Figure 5, it can be concluded that lime
CE
P
TE
AC
4.3 Shearing
The results are presented in terms of peak and ultimate (end of test) strengths, also
considering the critical state line of the untreated London Clay soil. Due to the
number of results, these are plotted in several figures for the sake of clarity. A
summary of the results is also given in Table 5.
25
ACCEPTED MANUSCRIPT
Figures 6(a)-(d) show comparative stress-strain plots, namely: 6 (a) shows limetreated specimens sheared under constant suction (CS) at suctions of 100 and 200 kPa
IP
and varying mean net stresses; 6(b) plots constant suction (CS) testing results for
SC
R
specimens sheared at a suction of 300 kPa and varying mean net stresses, compared
with the respective constant water content (CW) results for specimens subjected to an
initial suction of 300 kPa; 6(c) and (d) compare the lime treated soil results with
NU
available results from untreated London Clay specimens to assess the differences in
MA
Fig. 6. Deviator stress- axial strain plots (a) CS tests of lime-treated soil for suctions
TE
of 100 and 200 kPa; (b) comparative CS and CW results for suctions of 300 kPa; (c)
CE
P
comparative results for untreated and lime treated soil for p' or
of 200 kPa
AC
comparative results for untreated and lime treated soil for p' or
From Figures 6(a)-(d), it can be seen that whereas the untreated compacted London
Clay specimens show a strain hardening behaviour without any apparent peak in the
stress, all lime-treated specimens show a very pronounced peak in the strength within
relatively low strain levels; this was followed by strain softening presumably due to
the breakage of cementation bonds, with the stress decreasing dramatically after only
25.6 % of axial strain. The lime treatment thus causes a similar behaviour to that of
overconsolidated clays, which is consistent with the observations by Leroueil and
Vaughan (1990) regarding similarities in the behaviour between naturally cemented
26
ACCEPTED MANUSCRIPT
geomaterials (soft rocks) and overconsolidated clays. The elastic behaviour range is
IP
SC
R
Increasing suction resulted in an increase in the peak strength and stiffness of the
lime-treated London Clay. Whereas this is the expected behaviour of partially
saturated uncemented soils due to the effect of suction, in this instance, the behaviour
NU
is likely to be due to the combined effect of suction and the cementation bonds. Both
MA
types of soil approach constant values of deviator stress as the tests proceed (which is
also consistent with the behaviour noted from the volumetric strain curves and the
pore pressures for the constant suction and constant water content tests respectively,
TE
shown in Fig. 7a and b and discussed later). The lime-treated specimens tested under
CE
P
the same mean net stress reached essentially the same ultimate strength irrespective of
the suction level considering the usual scatter in the experimental results (with some
AC
anomaly in one point of the 200 kPa suction results from specimen LTLC-CS5). This
is also clearly depicted in Fig. 8, plotting the deviator stress versus matric suction
results as well as the q-p results plotted in Fig 14 in section 5.3 below. This is
consistent with the observed behaviour of uncemented partially saturated soils (e.g.
Cui and Delage, 1996 and Hamid, 2008). Conversely the peak deviator stress is
shown to clearly increase with mean net stress.
The stress-strain curves obtained from constant water content (CW) tests are similar
to those obtained from constant suction (CS) shearing; the peak strengths are slightly
27
ACCEPTED MANUSCRIPT
lower compared to the CS tests (see Fig. 6(b)). This could be attributed to the
reduction in the matric suction, consistent with the increase in pore water pressure
IP
when reaching the peak (plotted in Figure 7 (b)). In the post peak region the pore
SC
R
pressures gradually reduced until ultimate stresses were achieved. These ultimate
stresses were essentially the same as those of the respective CS specimens, with a
discrepancy noted between the LTLC-CS8 and LTLC-CW2 only (at 200 kPa suction),
MA
NU
Figure 7(a) presents selected volumetric strain vs. axial strain results. The reason why
the latter plots are shown for some of the specimens only is because not all triaxial
TE
systems used in this study could give reliable volume change measurements for the
CE
P
brittle specimens. Namely two of the triaxial systems were equipped with local LVDT
transducers whereas one system used the newly developed and validated laser sensor
AC
Fig. 7 (a) Volumetric strain and (b) pore water pressure variation with axial strain
28
ACCEPTED MANUSCRIPT
The volumetric strain vs. axial strain results (Fig. 7(a)) show that as opposed to the
untreated London Clay specimens, which were continuously contracting (consistently
IP
SC
R
NU
MA
the reported behaviour of partially saturated uncemented soils (e.g. Hamid, 2008).
With continuous shearing, the majority of the specimens appear to tend towards
constant total volumetric strains, which is consistent with the stressstrain behaviour
TE
of the soils presented in Fig. 6(a)-(c). Note that the lime-treated specimens showed the
CE
P
dilating tendency after, rather than before the peak stress, i.e., the peak strength was
mobilised well before the maximum rate of dilation. This conforms with published
AC
results on lithified materials (Leroueil and Vaughan, 1990; Vaughan, 1993) and
implies that the extra component of strength manifested as peak strength is not due to
dilatancy, as it would be the case for an uncemented particulate material but instead
due to the cementation bonding created by lime treatment. It can thus be argued that
as dilation only happens upon softening, this is consistent with the breakage of the
cementation bonds created by lime treatment.
Figure 8 shows plots of peak and ultimate deviator stress versus matric suction. The
figures are complemented with peak stress results for the saturated lime-treated soil,
29
ACCEPTED MANUSCRIPT
based on results presented elsewhere (Mavroulidou et al. 2011 and 2013b). Figure 8
clearly illustrates the dependence of the peak deviator stress on the suction level and
IP
suggests a non-linear relationship between peak shear strength and matric suction,
SC
R
whereas the ultimate deviator stress is shown to be practically unaffected by the levels
of suction considered in this study (there is some variation and one anomaly for the
200 kPa suction-200 kPa net stress results -noted earlier- but other than this, the
MA
NU
variation in the rest of the results does not appear to be considerable, in view of the
TE
Fig. 8 Peak and ultimate deviator stress envelopes for various suction levels
CE
P
AC
The analysis refers to the results of the suction unloading (suction decrease due to
wetting) tests of the lime-treated soil plotted in Figure 4(b). The results will be
analysed using the concept of a suction decrease (SD) yield locus postulated in the
well-known BBM model (Alonso et al. 1990), to refer to the occurrence of plastic
deformation due to wetting. From Figure 4(b) it can be seen that for low mean net
stresses the resulting curve tends to be bilinear; the turning point between the two
lines can be defined as the yield suction (Zhan, 2003). The bilinear relationship was
identified as:
v vs s ln(
s Pat
) within the elastic zone
pat
(Eqn. 4a)
30
ACCEPTED MANUSCRIPT
s Pat
) within the plastic zone
pat
(Eqn. 4b)
v N s s ln(
IP
where s and s are stiffness parameters with respect to a change in matric suction in
SC
R
the elastic and plastic zone, respectively; vs is the specific volume corresponding to
NU
Combining the yield suction at different mean net stress obtained from the various
MA
tests (see Fig. 4(a) and (b)), the SD yield line s p 200 was plotted in Figure 9.
The inclination of the SD yield line shows the effect of the mean net stress on the
TE
yielding of the lime treated London Clay upon wetting. The assumption of zero yield
at 200 kPa net stress gives a straight SD yield line inclined at 45 in the s: plane as
CE
P
initially suggested by Gens & Alonso (1992), and unlike data for expansive soils,
which can give inclination angles of the SD yield line larger than 45 (e.g. Zhan,
AC
2003). This is consistent with the improvement brought about by lime in terms of
volumetric stability upon wetting noted in section 4.1 above, which showed that the
nature of the initially shrinking/swelling London Clay soil changed to that of a soil of
low swelling potential due to lime treatment.
ACCEPTED MANUSCRIPT
concept of the Loading- Collapse (LC) yield curve, postulated in the BBM model
(Alonso et al. 1990). This curve is defined from the yield points of specimens with
IP
identical stress histories subjected to isotropic loading under constant suction. The LC
SC
R
will expand once the load applied exceeds the yield stress, reflecting the suction
NU
The Loading Collapse (LC) yield curves of the lime treated soil based on the isotropic
MA
loading tests were identified according to the approach proposed by Alonso (1990)
(see Fig 10). The shape of the resulting LC yield curve was consistent with that
proposed by Alonso et al. (1990). Although only three points were available for the
TE
untreated soil these were also used to draw a tentative LC curve for the purposes of
CE
P
comparison. The two curves reflect the considerable increase in the yield stress of the
lime treated London Clay compared to that of the untreated London Clay and for this
AC
reason the lime-treated LC curve moved to the right-hand side of the untreated soil
LC. The increased yield stress of the lime-treated soil indicates that the elastic range is
enlarged, which can be attributed to the effect of the lime on the soil structure
/cementation bonding.
Fig. 10 LC yield curves of lime-treated and untreated London Clay based on isotropic
compression tests
To express the LC curve mathematically, both the BBM model (Alonso et al., 1990)
32
ACCEPTED MANUSCRIPT
and Wheeler and Shivakumars (1995) model involve the parameter (s). However as
discussed above the value of (s) was not easily identifiable due to the non-linearity
IP
of the NCL of the lime-treated soil and due to the fact that the slopes (s) for lime-
SC
R
treated soil found in this study are likely to be related to initial yield rather than yield
associated with the complete breakage of cementation bonds. An expression that
would not involve the post yield parameter (s) was therefore introduced for the
NU
description of the presented experimental data. This reflects the fact that the isotropic
MA
yield behaviour is associated with both the loading and wetting paths. Therefore both
paths were considered in the suggested alternative mathematical expression of the LC
curve. To demonstrate the procedure, data of the wetting path under a mean net stress
TE
of 20 kPa was used; first the relationship of wetting-induced swelling with respect to
CE
P
the decrease in suction in the v:s space from data plotted in Figure 11(a) was
established (in this figure the plotted v values represent the final specific volumes
AC
s pat
s pat
) , ln(
) sd
pat
pat
v A2 B2 ln(
s pat
),
pat
ln(
s pat
) sd
pat
(Eqn. 5a)
(Eqn. 5b)
where A1 and A2 are the intercepts at suction s=0 kPa of the lines defined in Eqn 5a
and 5b respectively; B1 and B2 are the slopes of the lines defined in Eqn 5a and 5b
33
ACCEPTED MANUSCRIPT
respectively; sd is the yield suction factor during the wetting where pat is the
IP
Based on the experimental data, the resulting curve-fitting equations were as follows:
ln(
s pat
) 1.03
pat
s pat
s pat
) , ln(
) 1.03
pat
pat
(Eqn. 6a)
(Eqn. 6b)
s pat
s pat
) a2 exp(b2
)
pat
pat
(Eqn. 7)
TE
v a1 exp(b1
MA
NU
s pat
),
pat
SC
R
CE
P
Using units of kPa for the suction and atmospheric pressure, the curve-fitting gave the
AC
following expression:
(Eqn. 8)
The final specific volume recorded during wetting is the initial specific volume prior
to the isotropic compression. This can now be plotted against mean net stress. The
yield points identified using Casagrandes method presented in Table 4 are also added
on the graph (Fig. 11(b)). From this figure, it can be noted that in the v:
plane, the
yield points can be reasonably assumed to fall on a straight line. This line can be fitted
as:
34
ACCEPTED MANUSCRIPT
p
v N y y ln
pat
(Eqn. 9)
where Ny and y are respectively the intercept on the yield compression line at a
IP
reference pressure of pat and is the slope of the yield compression line in v: plane.
SC
R
For the case of the lime-treated soil the fitted line was found to be:
p
v 2.0625 0.086 ln
pat
(Eqn. 10)
NU
Note that before yield the change in specific volume of the lime-treated soil was very
MA
small and was therefore ignored. Thus the specific volumes represented in Figure
11(a) (end of wetting stage volumes) are also assumed to be the same until the yield.
TE
planes, an
CE
P
expression relating directly suction and yield stress can be obtained and hence, the LC
yield curve (Fig 11 (c)), thus described without the use of the parameter (s). For
AC
example using the double-exponential function (Eqn 7) and Eqn (9), and equating the
right hand side parts of the two equations gives:
a1 * exp(b1
p
s pat
s pat
) a2 * exp(b2
) N y y ln
pat
pat
pat
(Eqn. 11)
Note that following the above procedure a relationship can also be determined
between suction and isotropic compression yield mean net stress shown in Figure 11.
35
ACCEPTED MANUSCRIPT
The predicted specific volume and suction relationship using the two different
descriptions of the curve (bilinear and double exponential respectively) is shown in
IP
Figure 12. The predicted LC curves using the above procedure and the two different
SC
R
descriptions of the v:s curves are shown in Figure 13. From the figures it can be seen
that both expressions fit the available results very well; there are some small observed
differences in the fitted curves based on the two different expressions at suctions
NU
higher than 100 kPa; more experimental data would be needed to confirm which
MA
TE
CE
P
AC
Based on the values of Table 5, the Peak Strength Line (PSL) and Ultimate Strength
Line (USL) were therefore plotted in Figure 14. On the same graph, the results of
36
ACCEPTED MANUSCRIPT
saturated lime treated and untreated London Clay specimens published elsewhere
(Mavroulidou et al, 2011) were also included for the sake of comparison between
IP
SC
R
NU
was found that the peak angle of friction was p =26.5, whereas the results for both
MA
soils (treated and untreated) could be expressed by a unique Critical State line in the
q:p' plane, with a slope of M=0.88 corresponding to an angle of internal friction of c
=22.5 (Mavroulidou et al, 2011). This value is consistent with values reported in the
TE
literature for London Clay (Schofield and Wroth, 1968). This implies that lime did not
CE
P
affect the frictional characteristics of the clay soil and that the observed improvement
in the shear strength manifested in terms of peak strengths was thus due to the
AC
differences in the soil structure before and after lime treatment (including bonding due
to cementing compounds induced by lime).
For the partially saturated lime-treated soil (Fig. 14) the ultimate strength lines
according to the results appear to be slightly different than the CSL of the saturated
soils; from linear regression the slopes of the USLs were found to be 0.82, 0.86 and
0.95 for suctions s, of 100 kPa, 200 kPa and 300 kPa respectively (presumably due to
suction effects) giving however an average slope of 0.88 which is the value of the
37
ACCEPTED MANUSCRIPT
critical state parameter M of the saturated untreated (and treated) soil1. Note that in
the original BBM model, Alonso et al (1990) suggest that a critical state line (CSL)
IP
for non-zero suction will represent the increased strength induced by suction in terms
SC
R
of an increase in the apparent cohesion, while maintaining the slope M of the CSL for
saturated conditions. The intercepts of the best fit lines through the ultimate stress
points for each suction level are found to decrease with suction i.e., about 50 kPa for
NU
s=100 kPa, 46 kPa for s=200 kPa (which is actually quite close to that of 100kPa and
MA
could be considered the same due to the usual scatter of experimental data), and 17
CE
P
TE
The critical state of the partially saturated soil in the q-(p-u) plane can be written as
AC
(Toll, 1990):
q= M (p - u)+Mb(u- uw)
(Eqn. 12)
where M is the critical state stress ratio with respect to the mean net stress (p - u)
and Mb is the critical state stress ratio with respect to the matric suction (u- uw).
Based on Alonso et al (1990) assumptions that M=M=0.88 (where M is the critical
state parameter of the saturated soil) the values of Mb were calculated as shown in
1 The anomalous point of the LTLC-CS5 was excluded from the analysis of the results (curve fitting
and determination of the parameter Mb)
38
ACCEPTED MANUSCRIPT
Table 6. To fit the presented data, it was found that Mb was not a constant and
generally decreased with suction (with one irregularity in the trends for the LTLC-02
IP
specimen, probably because the test terminated earlier than the others). This implies
SC
R
that there is a drop in Mb with the gradual desaturation of the material as reported
elsewhere for untreated partially saturated soils (Vanapalli et al, 1996; Toll and Ong,
2003). Unlike the findings in Toll et al (2008) for an artificially bonded sand, showing
NU
MA
(hence Toll et al suggested that a value of a > should have been used to overcome
this issue), here the value of M set equal to M gave sensible values of Mb, clearly
TE
lower than M.
AC
variables
CE
P
Table 6. Partially saturated lime-treated specimens: critical state values of the state
From Figure 14, linear regression gave the slopes of the PSLs as 0.73, 0.69 and 0.64
for suctions of 100 kPa, 200 kPa and 300 kPa respectively, i.e. the PSL slope
decreased with suction. The PSL was much higher than the USL in each individual
suction level, suggesting that the additional peak strength is mainly dependent on the
lime-induced bonding.
39
ACCEPTED MANUSCRIPT
For the mathematical description of the peak strength of partially saturated lime
treated specimens we adapted Tolls (1990) critical state line expression for partially
IP
saturated soils. It was thus assumed that the peak strength of artificially cemented
SC
R
specimens could be associated with three components: (a) a mean net stress
component, (b) a suction component and (c) a cementation bonding component.
NU
q n ( p u ) n (u u ) f (c)
a
(Eqn. 13)
MA
where the stress ratio parameters na and nb describe the components of shear strength
associated respectively with the uncoupled stress state variables (p-ua) and (ua uw) in
TE
a similar fashion as in Toll (1990) and Toll and Ong (2003); f(c) is the additional
cementation bonding effect, which is a function of suction (because the difference in
CE
P
peak values between the treated and untreated partially saturated soil was not constant
for different suctions); f(c) can thus be expressed as:
f (c) n (u u ) q
AC
(Eqn. 14)
q n ( p u ) (n n )(u u ) q
a
(Eqn. 15)
40
ACCEPTED MANUSCRIPT
where (b+c) is the stress ratio incorporating chemically-induced bonding and
suction, whose effect is likely to be coupled 2. When suction is zero, the above
IP
equation reduces to the PSL expression for saturated soils. It should be noted that the
SC
R
For the constant suction tests a was found from regression analysis of the
NU
experimental data for each specimen. The value of (b+c) was then determined from
MA
Eqn 15. The average value of (b+c) at each suction level was then calculated and
used for the prediction of the peak strength according to Eqn 15 (note that a was not
averaged). The calculated parameters a and mean (b+c) and resulting predictions of
CE
P
TE
deviator stresses for the constant suction tests are given in Table 7.
For the constant water content tests the parameters a and average (b+c) were
AC
n 0.0005 * (u u ) 0.7873
a
(Eqn. 16)
2 To determine each one of these two parameters nb and nc individually, a relationship between
cementation bonding component and suction is needed, which would necessitate a large number of data
on the untreated partially saturated soil, in addition to those of the lime-treated partially saturated soil.
41
ACCEPTED MANUSCRIPT
n n 55.42 * (u u )
b
0.606
(Eqn. 17)
SC
R
IP
NU
Using the above expressions, the peak deviator stresses qp of the three CW specimens
MA
were successfully predicted (with a maximum error of -1.37%), (see Table 8 and
TE
Figure 17 depicting the predicted and the measured peak deviator stresses).
CE
P
Table 7. Values of , (b+c) and predicted peak deviator stresses qp for lime treated
AC
Table 8. Predicted peak deviator stresses qp for lime treated specimens (CW tests)
6. Conclusions
The mechanical behaviour of lime treated London Clay during wetting, compression
and shearing was investigated via suction-controlled triaxial testing. The paper
42
ACCEPTED MANUSCRIPT
provided valuable data of suction controlled testing of lime-treated soils and
addressed the corresponding behaviour of the lime-treated soil, within an unsaturated
IP
soil mechanics framework. Although the results presented are specific to London
SC
R
Clay, experience indicates that the behaviour of other highly plastic clays treated with
lime will follow similar trends.
NU
Overall the experimental results showed the beneficial effect of the lime on the
MA
volumetric properties (reduced swelling upon wetting, increased yield stress and
reduced compressibility upon compression) and shear strength within ranges of strain
relevant to engineering design (despite the observed strain softening behaviour). They
TE
also indicated that for the range of suctions considered in this study chemically
CE
P
induced bonding had a major effect on the properties and behaviour of the soil.
AC
It was also shown that although the properties of the soil have obviously changed
upon lime treatment, the behaviour of the partially saturated lime treated soil
presented behaviour trends consistent with the reported behaviour of uncemented
partially saturated soils regarding the effect of suction or mean net stress, with the
exception of the compressibility behaviour of the lime-treated soil. It was thus shown
that the partially saturated lime treated soil could be described within the framework
of constitutive models commonly used for untreated partially saturated soils with only
occasional modifications. Thus, some simple procedures for the analysis of the results
of this type of soil were suggested and successfully applied to the mathematical
43
ACCEPTED MANUSCRIPT
description of the results. These consisted in (a) an alternative, simple method of
determining the LC yield equation combining the wetting induced swelling curve and
IP
the LC yield compression lines and (b) an expression used to interpret the peak
SC
R
strengths of the partially saturated lime treated soil to account for the combined effect
of mean net stress, suction and cementation bonding.
NU
Acknowledgements
MA
The work presented in this paper was carried out at London South Bank University
during the doctoral studies of the first author, funded by the UK Engineering and
TE
CE
P
References
Alonso, E. E., Gens, A., Josa, A., 1990. A constitutive model for partially saturated
AC
ACCEPTED MANUSCRIPT
Barker J. E., 2002. Ion migration associated with lime piles, PhD Thesis, University
of Birmingham, UK.
IP
Bell F.G.,1996. Lime stabilization of clay minerals and soils. Engineering Geology,
SC
R
42, 223-237.
Bishop, A. W., 1959. The principle of effective stress. Tecknisk Ukeblad 106 (39):
859-863.
NU
Brandl, H., 1981. Alteration of soil parameters by stabilization with lime, In:
MA
British Standards Institution BSI, 1990a. Methods of test for Soils for Civil
TE
CE
P
British Standards Institution BSI, 1990b. Stabilized Materials for Civil Engineering
purposes. BS 1924: Part 2. BSI, Milton Keynes.
AC
Cabarkapa, Z., 2001. Mechanical behaviour and modelling of unsaturated soils. PhD
thesis. South Bank University, London, UK.
Chen, Z.-H., Fredlund, D., Gan, J.K., 1999. Overall volume change, water volume
change, and yield associated with an unsaturated compacted loess. Canadian
geotechnical journal 36, 321-329.
Cui, Y. J. and Delage, P., 1996. Yielding and plastic behaviour of an unsaturated
compacted silt. Gotechnique, 46(2), 291-311.
Cui, Y., Yahia-Aissa, M., Delage, P., 2002. A model for the volume change behavior
of heavily compacted swelling clays. Engineering Geology, 64, 233-250.
45
ACCEPTED MANUSCRIPT
Delage, P., 2004. Experimental unsaturated soil mechanics. In: Unsaturated soils
Proc. 3rd Int. Conf. on Unsaturated Soils, UNSAT2002, Juca JFT, De Campos
IP
TMP, Marino FAM (eds), Recife, Brazil, Balkema, vol 3, pp. 973-996.
SC
R
NU
MA
TE
CE
P
AC
ACCEPTED MANUSCRIPT
Durham, UK, July 2-4 2008, CRC Press/Balkema, London, pp. 429-434.
Harris P, Scullion T., 2009. Stabilizer Diffusion in Swelling Soils. In: GeoFlorida
IP
SC
R
GSP 187, Iskander M, Laefer D. F., and Hussein, M.H. (Eds) ASCE ISBN (print):
978-0-7844-1023-3
NU
New York
MA
Jotisankasa, A., 2005. Collapse behaviour of a compacted silty clay. PhD thesis,
TE
CE
P
AC
ACCEPTED MANUSCRIPT
2011: Theory and Practice, Proceedings of the 5th Asian-Pacific Conference on
Unsaturated Soils, Papaya, Thailand , Jotisankasa et al., (eds), Kasetsart University
IP
SC
R
Mavroulidou M., Zhang X., Gunn M J., Cabarkapa Z., 2013a. Water Retention and
Compressibility of a Lime-Treated, High Plasticity Clay, Geotechnical and Geological
Engineering, 31(4), 1171-1185.
NU
Mavroulidou M., Zhang X., Kichou Z., Gunn M.J., 2013b. Hydro-mechanical
MA
properties of lime-treated London Clay, In: 18th International Conference for Soil
Mechanics and Geotechnical Engineering, Paris 2-5 September 2013, pp. 1151-1154.
CE
P
Available from
TE
Lime stabilization & lime modification. Bulletin 326, NLA, Arlington (Online).
http://www.lime.org/documents/publications/free_downloads/construct-
AC
ACCEPTED MANUSCRIPT
Schofield, A. and Wroth, P., 1968. Critical State Soil Mechanics. McGraw Hill,
London.
IP
SC
R
Sherwood, P.T., 1993. Soil Stabilization with Cement and Lime.HMSO, London,
UK.
NU
Sivakumar, V., 1993. A critical state framework for unsaturated soils. PhD Thesis,
MA
TE
31-44.
CE
P
Toll, D. G., 1990. A framework for unsaturated soil behaviour. Gotechnique, 40(1),
AC
Toll, D. G. and Ong, B. H., 2003. Critical state parameters for an unsaturated residual
sandy clay. Gotechnique, 53(1), 93103.
Toll, D. G., Ali Rahman, Z., Gallipoli, D., 2008. Critical state conditions for an
unsaturated artificially bonded soil. In: Toll, D. G., Augarde, C. E., Gallipoli, D. &
Wheeler, S. J. (eds.) Proceedings 1st European Conference on Unsaturated Soils,
Durham, CRC Press. pp. 435-440.
Vanapalli, S. K., Fredlund, D. G., Pufahl, D. E., Clifton, A. W., 1996. Model for the
prediction of shear strength with respect to soil suction. Canadian Geotechnical
Journal. 33, 379-392.
49
ACCEPTED MANUSCRIPT
Vaughan P.R., 1993. Engineering behaviour of soft rocks. Some answers and
some
questions.
IP
SC
R
1741-1766.
Wheeler, S. J. and Sivakumar, V., 1995. An elasto-plastic critical state framework for
unsaturated soil. Gotechnique, 45(1), 35-53.
NU
MA
Zhan, L. T., 2003. Field and Laboratory study of an unsaturated expansive soil
TE
associated with rain-induced slope instability. PhD thesis. Hong Kong University of
CE
P
Zhang, X. Mavroulidou, M. Gunn MJ, Cabarkapa, Z., 2010. A study of soil water
AC
ACCEPTED MANUSCRIPT
Table 1 Composition of London Clay soil used in this study
Clay content %
51
SC
R
50
26
15
9
4%
45%
NU
Illite (%)
Smectite (%)
Kaolinite (%)
Chlorite (%)
Sand
Silt
IP
of which:
(%)
(%)
64
CE
P
Lime-treated
wP
London Clay
wL
26
Standard
Proctor
wopt
(%)
Standard
Proctor
dmax
(g/cm3)
2.75
26
1.43
2.73
30
1.26
IP
GS
38
35
TE
Soil Type
MA
Table 2 Physical characteristics of untreated and 4% lime treated London Clay soils
89
54
London Clay
AC
(4% lime)
51
ACCEPTED MANUSCRIPT
Specimen Dia.
mm
Suction level,
kPa
p' or (p-ua),
kPa
Stress Path
Initial conditions
Type
w
(%)
Sr
(%)
0%
50
100
3=0
73.2
LC-SAT2
0%
38
200
3=0
74.9
LC-IS1
0%
38
5-570-5
N/A
N/A
24.95 0.928
73.9
LC-IS2
0%
38
300
200
N/A
N/A
24.80 0.923
73.9
LC-CS1
0%
38
200
100
3=0
Drained
LC-CS2
0%
38
300
200
3=0
Drained 24.8
LTLC-IS1
4%
38
300
N/A
LTLC-IS2
4%
38
50
20-400-20
LTLC-IS3
4%
38
100
LTLC-IS4
4%
50
LTLC-IS5
4%
50
LTLC-CS1
4%
38
LTLC-CS2
4%
38
LTLC-CS3
4%
LTLC-CS4
4%
LTLC-CS5
4%
LTLC-CS6
4%
LTLC-CS7
4%
LTLC-CS8
4%
LTLC-CS9
4%
LTLC-CW1
4%
LTLC-CW2
4%
38
LTLC-CW3
4%
38
LTLC-W1
4%
50
LTLC-W2
4%
50
NU
LC-SAT1
AC
SC
R
Lime %
Specimen ID
Shearing
IP
Consolidation/
Compression
73.8
0.923
73.9
N/A
27.00 0.918
80.2
N/A
N/A
26.87 0.931
78.6
20-530-23-100
N/A
N/A
26.91 0.941
77.9
200
20-1000-20
N/A
N/A
26.97 0.933
78.8
300
20-1200
N/A
N/A
26.92 0.925
79.3
100
3=0
CS
26.93 0.930
78.9
200
3=0
CS
26.82 0.928
78.8
38
300
3=0
CS
26.97 0.932
78.9
38
100
3=0
CS
26.82 0.931
78.5
200
3=0
CS
26.69 0.938
77.5
300
3=0
CS
26.82 0.931
78.5
100
3=0
CS
26.78 0.928
78.6
200
3=0
CS
26.89 0.931
78.7
38
300
3=0
CS
26.78 0.931
78.4
38
100
3=0
CW
27.00 0.940
78.3
200
3=0
CW
26.80 0.931
78.4
300
3=0
CW
26.60 0.931
77.9
100
N/A*
N/A
26.60 0.945
76.7
200
N/A*
N/A
26.90 0.938
77.0
D
TE
CE
P
38
MA
0.931
100
200
38
38
38
300
300
500-0 (wetting
path)
25
*These were Soil Water Retention Curve tests demonstrating the effect of stress state (see Fig 4a and b)
52
ACCEPTED MANUSCRIPT
Table 4. Soil parameters from isotropic compression data
p c
(s)
N (s)
0
200
300
0
50
100
200
300
35
80
180
130
180
260
340
400
0.139
N/A
0.047
0.064
0.027
0.031
0.058
0.065
2.27
N/A
2.16
2.041
2.028
2.014
2.048
2.049
IP
s (kPa)
SC
R
Specimen
LC-IS1
LC-CS1*
LC-IS2
LTLC-IS1
LTLC-IS2
LTLC-IS3
LTLC-IS4
LTLC-IS5
NU
*Specimen LC-CS1 was used to complement the data for the yield stress but the (s)
MA
and N(s) could not be determined due to the limited extent of the data
LTLC-CS0*
LTLC-CS4
LTLC-CS5
TE
LTLC-CS1
LTLC-CS2
LTLC-CS3
Peak
Strain range (%) for
Mean net
Ratio of peak
Suction deviator
peak strength
stress
/ultimate
(kPa)
stress
(kPa)
deviator stress
(kPa)
3%
100
689
3.59
2%
100
200
791
2.95
1.62%,
300
880
2.11
4.2%
50
668
1.67
CE
P
Specimen ID
784
849
LTLC-CS6
300
966
3%
2.66%
2.53%
LTLC-CS7
LTLC-CS8
100
200
836
957
5%
5.6
300
1032
3.96%
300
870
945
1015
1.6%,
3.1%
4.2%
AC
100
200
200
300
LTLC-CS9
LTLC-CW1
100
LTLC-CW2
200
LTLC-CW3
300
4.22
2.53
2.25
4.80
3.38
2.29
4.22
2.53
2.25
*Not plotted in Figures 6(a)-(b) for the sake of clarity in the presentation of the plots
Table 6. Partially saturated lime-treated specimens: critical state values of the state
53
ACCEPTED MANUSCRIPT
Mb
0.88
0.88
0.88
0.88
0.88
0.88
0.88
0.88
0.49
0.14
0.31
0.22
0.20
0.12
0.08
0.18
IP
(u -uw)
(kPa)
100
100
100
200
200
300
300
300
SC
R
(p-u)
(kPa)
162
289
439
162
443
158
295
450
NU
q
(kPa)
192
268
417
186
429
174
283
450
variables
MA
Table 7. Values of , (b+c) and predicted peak deviator stress qp for lime treated
LTLC-CS0
(ua-ua)p
(kPa)
50
LTLC-CS1
LTLC-CS2
100
LTLC-CS4
LTLC-CS5
LTLC-CS6
200
AC
LTLC-CS7
LTLC-CS8
na
regressed
(nb+nc)
calculated
(nb+nc)
mean
300
LTLC-CS9
qp(kPa)
predicted
668
423
0.78
5.166
5.166
668
689
330
0.73
3.683
3.564
677
784
461
0.73
3.672
3.564
773
836
579
0.73
3.336
3.564
859
791
364
0.69
2.300
2.243
779
849
483
0.69
2.179
2.243
862
957
619
0.69
2.249
2.243
956
880
393
0.64
1.828
1.822
878
966
522
0.64
1.840
1.822
961
1032
644
0.64
1.799
1.822
1039
CE
P
LTLC-CS3
qp(kPa) (p-ua)p
Measured (kPa)
TE
Specimen
ID
Table 8. Predicted peak deviator stresses qp for lime treated specimens (CW tests)
Specimen
ID
(ua-ua)p
(kPa)
qp(kPa)
Measured
LTLC-CW1
276
868
389
0.65
1.891
856.125
-1.37
LTLC-CW2
274
945
515
0.65
1.899
936.935
-0.85
LTLC-CW3
247
1010
637
0.65
2.021
1000.070
-0.98
(p-ua)p
(kPa)
na
(nb+nc)
regressed mean
qp(kPa)
predicted
Error
(%)
54
ACCEPTED MANUSCRIPT
AC
CE
P
TE
MA
NU
SC
R
IP
Fig. 1
55
MA
NU
SC
R
IP
ACCEPTED MANUSCRIPT
AC
CE
P
TE
Fig. 2
56
ACCEPTED MANUSCRIPT
AC
CE
P
TE
MA
NU
SC
R
IP
Fig. 3
57
ACCEPTED MANUSCRIPT
AC
CE
P
TE
MA
NU
SC
R
IP
Fig. 4a
58
ACCEPTED MANUSCRIPT
AC
CE
P
TE
MA
NU
SC
R
IP
Fig. 4b
59
ACCEPTED MANUSCRIPT
AC
CE
P
TE
MA
NU
SC
R
IP
Fig. 5
60
ACCEPTED MANUSCRIPT
AC
CE
P
TE
MA
NU
SC
R
IP
Fig. 6a
61
ACCEPTED MANUSCRIPT
AC
CE
P
TE
MA
NU
SC
R
IP
Fig. 6b
62
ACCEPTED MANUSCRIPT
AC
CE
P
TE
MA
NU
SC
R
IP
Fig. 6c
63
ACCEPTED MANUSCRIPT
AC
CE
P
TE
MA
NU
SC
R
IP
Fig. 6d
64
ACCEPTED MANUSCRIPT
AC
CE
P
TE
MA
NU
SC
R
IP
Fig. 7a
65
ACCEPTED MANUSCRIPT
AC
CE
P
TE
MA
NU
SC
R
IP
Fig. 7b
66
ACCEPTED MANUSCRIPT
AC
CE
P
TE
MA
NU
SC
R
IP
Fig. 8
67
ACCEPTED MANUSCRIPT
AC
CE
P
TE
MA
NU
SC
R
IP
Fig. 9
68
ACCEPTED MANUSCRIPT
AC
CE
P
TE
MA
NU
SC
R
IP
Fig. 10
69
ACCEPTED MANUSCRIPT
AC
CE
P
TE
MA
NU
SC
R
IP
Fig. 11
70
IP
ACCEPTED MANUSCRIPT
AC
CE
P
TE
MA
NU
SC
R
Fig. 12
71
IP
ACCEPTED MANUSCRIPT
AC
CE
P
TE
MA
NU
SC
R
Fig. 13
72
ACCEPTED MANUSCRIPT
AC
CE
P
TE
MA
NU
SC
R
IP
Fig. 14
73
ACCEPTED MANUSCRIPT
AC
CE
P
TE
MA
NU
SC
R
IP
Fig. 15
74
ACCEPTED MANUSCRIPT
AC
CE
P
TE
MA
NU
SC
R
IP
Fig. 16
75
ACCEPTED MANUSCRIPT
AC
CE
P
TE
MA
NU
SC
R
IP
Fig. 17
76
ACCEPTED MANUSCRIPT
Highlights
We investigated the mechanical behaviour of lime treated London Clay
IP
testing)
We described the behaviour of the lime-treated soil
We proposed simple expressions for the analysis of results for this soil type
AC
CE
P
TE
MA
NU
SC
R
77