Soils and Foundations 2014;54(2):109–115

The Japanese Geotechnical Society

Soils and Foundations

www.sciencedirect.com
journal homepage: www.elsevier.com/locate/sandf

Clay soil settlement: In-situ experimentation and analytical approach

S. Bensallamn, L. Bahi, H. Ejjaaouani, V. Shakhirev, K. Rkha Chaham
Ecole Mohammadia d’Ingénieurs, Université Mohammed 5 – Agdal, Morocco Laboratoire Public d’Essais et d’Etudes, Morocco
Received 19 July 2012; received in revised form 4 September 2013; accepted 2 October 2013
Available online 3 April 2014

Abstract

On the basis of the existing relationship between soil structure and water content, and in order to avoid the multiple factors influencing the
study of soil behavior in the laboratory, a full-scale in-situ testing was performed in Ouarzazate (Morocco) to quantify the soil vertical
displacements according to the environmental conditions.
The study presented in this paper is devoted to analyzing the load-settlement relationship of active clay soil during the drying process. A time-
dependent model is presented to quantify the soil settlement amplitude according to the hydraulic and mechanical states of the soil mass.
& 2014 The Japanese Geotechnical Society. Production and hosting by Elsevier B.V. All rights reserved.

Keywords: Vertical displacements; Shrinkage; Settlement; Clay soil; Predicting model; In-situ testing

1. Introduction Gens and Alonso (1992) presented a conceptual basis for a model
for expansive soils, where two distinct levels are distinguished: a
The total volume change observed in expansive clay soils is microstructural level where the swelling of active minerals takes
basically due to the respective changes associated with place, and a macrostructural level responsible for major structural
each component phase. The interdependence of the physical and rearrangement. At the microstructural scale, volumetric stress–
chemical processes that occurs in the soil mass, however, leads to strain behavior is considered, while volumetric and deviatoric
the complex mechanics of volume change. This volume change effects are considered in the macrostructural scale. On the other
can be defined as a coupling of hydraulic and mechanical hand Annette (1998) presented a model where the strain increment
processes; deformation resulting from the variation of the water in a small time step has two components: an elastic increment
content due to weather conditions represents the hydraulic part, and which is the direct response to changes in an effective stress state
deformation resulting from the variation of vertical stresses due to and a plastic increment which is a time dependent response.
the soil–structure interaction represents the mechanical part. To Mitchell and Soga (2005) presented an analytical model which
simplify the analytical approach of problematic volume changes, describes an attempt to combine both elastic and plastic behaviors;
it is assumed in this model that the strain is a sum of the elastic and
n
Corresponding author. plastic components (cited by Nelson and Miller (1992)).
E-mail address: Saad.bensallam@gmail.com (S. Bensallam). The Barcelona Expansive Model (BExM) proposed by Alonso
Peer review under responsibility of The Japanese Geotechnical Society.
et al. (1999) is considered as a theoretical reference framework to
study the behavior of unsaturated expansive clays. In the literature
several prediction models were developed to assess the volume
changes coupling the hydraulic and stress states (Ejjaaouani et al.,
2008; ARGIC-BRGM/RP-54862-FR et al., 2006; Mitchell and

http://dx.doi.org/10.1016/j.sandf.2014.02.003
0038-0806 & 2014 The Japanese Geotechnical Society. Production and hosting by Elsevier B.V. All rights reserved.
110 S. Bensallam et al. / Soils and Foundations 54 (2014) 109–115

Soga, 2005; Rees and Thomas, 1993; Alonso and Lloret, 1982; structural damage. The expansive clay layer subjected to the
Fredlund and Rahardjo, 1993), and the study of the time-dependent experimental study is located from 0.5–1.7 m to 2.2 m under the
behavior of clay soils remains one of the most discussed approaches ground level, and the thickness of the studied layer was not
used to quantify their long term behavior. Alonso et al. (2005), uniform.
Nowamooz (2007), Oka et al. (2010) and Wang et al. (2013) have The mineralogical description of the clay fraction of the
highlighted the unsaturated expansive soils' densification, as the studied layer is as follows:
increase of the mechanical load decreases the deformation due to the
hydraulic variations. However, it is more or less considered that
10–15% illite;
most of the analytical models are developed for a specific
25–40% kaolinite;
environment or a specific kind of soil, and this is mainly due to
5–15% palygorskite;
the significant mineral associations and environmental conditions.
45–55% smectite.
In arid and semi-arid areas like the studied field (Ouarzazate),
active clays are in general associated with water deficient condi-
tions, which means they are considered unsaturated soils. Accord- The main geotechnical properties of the studied layer are
ing to Popescu (1986), due to the development of a theoretical summarized in Table 1. Intact soil specimens were collected at
framework for unsaturated soils and of several models, expansive a depth of 20 cm from the embedment level of the tested clay
clay soils are treated as if they were unsaturated soils although the layer in order to characterize the soil parameters for each
fundamental mechanisms of their volume change behavior are moisture state. A summary of the internal friction angle
different. However, it is assumed that they exihibit identical behav- according to the soil moisture state and the soil vertical
iors during the drying process. This will be taken into consideration displacements is included in Table 2. The internal friction
in the proposed prediction model presented in this paper. angle values were determined from the direct shear test,
In order to provide a contribution to quantify the settlement of according to the French Standard NF_P94-071.
clay soil, an analytical model is presented in this paper on the basis After removing the non-expansive covering layer, the devel-
of a full-scale in-situ test performed by the Laboratoire Public oped experimental device consisting of four rigid foundations (one
d'Essais et d'Etudes (LPEE) and the Laboratoitre Central des Ponts square meter for each one) was placed over the clay layer as
et Chaussées (LCPC). Since the experimental device used in this presented in Fig. 1. A rotating steel beam fixed to the center of the
experiment was not developed to assess the three dimensional experimental area served as the loading equipment, and the
deformations, only axial deformation measurements were allowed. rotative movement of the loading beam allowed it to be placed
over each foundation independently and a load was applied with
2. Materials and methods a hydraulic jack with a 380 cm2 surface. Each foundation was
used at a certain moment of the experiment, relative to a specific
2.1. Experimental device and experimentation moisture state. For the moistening stage, the water supply was
possible thanks to small pipe lines placed over the entire tested
The in-situ tests were performed in the northwestern part of area (water supplying network, Fig. 1), and the entire test
Ouarzazate city (Morocco), a location renowned for its reported area was submerged until a saturated steady state was reached.

Table 1
Physical and mechanical soil characteristics.

Plasticity index (Ip) Liquid limit Shrinkage limit Plasticity limit Blue methylene Soil's dry Soil's density
(WL) (Ws) (Wp) value (Vbs) density (ρd) (ρh)

30% 57 16 27 5.12 g/100 g 1.61 t/m3 1.81 t/m3

Grading curve characteristics Void ratio (e) Liquidity index (IL) Swelling pressure (sg) Free swelling Particles density (ρs) Natural water
deformation (εg) content (W)
C2 mm ¼10%, C20 mm¼ 80%, 0.662
0.482 380 kPa 0.114 2.66 t/m3 12.5%
C80 mm¼ 93%, 2 mm¼100%

Table 2
Water content and vertical displacement (Ejjaaouani et al., 2008).

Foundation Test date Water content Free swelling Free settlement Active zone Internal friction
(at
20 cm) (mm) (mm) thickness (m) angle (ϕ)

FS-1 Oct 2002 12.54% – – 1.7 35

FS-1 Jul 2003 35% 93 – 1.7 24
FS-2 Sep 2003 27% 76 4–6 1.6 26
FS-3 Mar 2004 20% 87 20–29 1.5 27
FS-4 Sep 2004 16% 80 35–50 1.45 30
 S. Bensallam et al. / Soils and Foundations 54 (2014) 109–115 111

Each Shallow Foundation (SF) used in this experiment was used displacements, and their stems were based on two metal
at a defined decreasing water content state, and the accompanying beams embedded on both sides of the foundation.
vertical displacements and continuous loading–unloading mea-
Elastic band attached to fixed piles.
surements were taken at time intervals. Figs. 1 and 2 below show A detailed description of the experimental device and the
the position of each component of the experimental device. The testing process is presented by Ejjaaouani et al. (2008).
loading-time of each shallow foundation was as follows:

 The first Shallow Foundation FS-1 was loaded twice: before 2.2. In-situ testing results
the soil wetting in order to test its behavior in a natural state
(w=12.5%), and after total saturation (w=35% the average In situ measurements not only provide valuable data needed to
water content under the foundation). assess soil conditions for characterization, but also allow the
 Foundation FS-2 was tested after the wetting period's end, quantitative assessment ofsoil properties in place. They allow
when the average water content was 27%. measurements under field environmental conditions with a mini-
 Foundation FS-3 was tested when the average water content mum amount of disturbance due to sampling and handling.. For
was 20%. this study, the performed in-situ tests were carried out during the
 Foundation FS-4 was tested when the average water content drying process after the water supply was stopped (at saturation
was 16%. state). Vertical displacements were taken at five hydraulic states;
the loading curves are shown in Fig. 3.
During the experimental process, each hydraulic state was
To check the soil vertical displacements, three measurement
accompanied by a specific ground settlement. The measured
devices were used:
settlement (Sr) over the entire tested area in terms of the
moisture states is as follows:

Electronic linear displacement sensors firmly fixed to two
beams installed on both sides of the foundation, with a
 4–6 mm for w ¼ 35–27%
support based outside the area affected by the soil
 20–29 mm for w ¼ 27–20%
deformation. The sensors stems were based on ceramic
 35–50 mm for w ¼ 20–16%
plates placed on the foundation corners.

0.01 mm accuracy classical displacement sensors, fixed on
supports placed outside the area affected by the vertical

Fig. 1. Shallow foundations position in the experimental area (Ejjaaouani

et al., 2008). Fig. 3. Vertical displacements of the foundations.

Fig. 2. Diagram of the beam used to charge the shallow foundations (Ejjaaouani et al., 2008).
112 S. Bensallam et al. / Soils and Foundations 54 (2014) 109–115

Table 2 summarizes the values of the foundation displace- representative of the real soil behavior), the predictive analy-
ments according to the water content and thickness of the tical modeling seems to be the best alternative to quantify the
active zone. amplitude of soil deformations. In this study and on the basis
At the beginning of the drying process, the soil lost its water of the in-situ full-scale test results, an analytical model is
intensively. The soil water content passed from 35% to 27% in 30 developed to predict the settlement amplitude of clay active
days and remained steady for 6 months, with some daily soil during the drying process. The choice to use the in-situ test
fluctuations which could be explained by the diurnal difference results was made in order to take the anisotropic behavior due
in temperature. This intense loss of water can be explained by the to structure anisotropy and/or to stress induced anisotropy into
evaporation of the free water located outside the sphere of action consideration, and to avoid major scale and handling problems.
of the attraction molecular forces (of the clay-particles); and the Because the experimental device was not developed to
higher the temperature became, the more vulnerable the weakly characterize the three-dimensional deformations, the only
linked water was to evaporation. Basically, in the soil mass, three deformation analyzed in this study was the vertical displace-
types of water can be distinguished: strongly linked water around ment. The experimental device used in this experiment was not
the particles, weakly-linked water which constitutes the diffuse capable of analyzing three dimensional soil deformations. The
layer around the particles, and free water between the closest presented modeling is focused on the behavior of clay soil
particles. It is well known that the soil volume change in active during the drying process from a saturation state corresponding
clay soils is closely related to the thickness of the diffuse layer to the water content (wm) and the swelling limit (Δhg).
variation (Lebedev, 1936). Using thermo-analysis, it is possible to As noted before in the text, the volume change process is a
evacuate each kind of water using a specific range of tempera- coupling of hydraulic and mechanical processes, and the main
tures. According to the laboratory tests, the temperature range factors influencing the volume change potential are the
required to evacuate each kind of water is 30–60 1C for free intrinsic soil characteristics that influence the basic equilibrium
water; 105 1C for the weakly linked water; and up to 200 1C for of the soil structure, and the environmental factors (hydraulic
the strongly linked water. and stress states) that influence the changes that may occur in
As noted above, at the beginning of the in-situ tests, water the internal force system. To quantify the volume change
content decreased dramatically from 35% to 27% due to the during the drying process, it is necessary to quantify the water
evaporation of the free water. This reduction continued for 30 days content state relative to the boundaries hydraulic conditions.
accompanied by a settlement from 4 to 6 mm through the tested For the prediction model, the hydraulic state in the considered
area. Then the water content and the settlement remained stable active zone as a function of depth “z” is presented as the ratio
over 6 months. According to the soil states, it can be assumed that wm
wi
the quantity of the linked water in this soil is approximately 27%. ; ð1Þ
wl
In order to quantify the strongly linked water, an intensive drying
of some intact samples was performed, and the measured water where wm–wi represents the range of water content in the
content was 4%. Furthermore, it is possible that some non- saturation state and the water content state used in calculations
connected clay matrix pores remained partially filled with water, as a function of the shrinkage limit wl.
which can lead to erroneous measurements of the quantities of the At the depth “z”, the vertical stress is the sum of the top
weakly and strongly linked water (Bensallam et al., 2012). layer weight and external applied loads. The weight of the top
From the previous results, it can be concluded that layer increases with depth, while the effect of the applied loads
lessens due to stress diffusion in the soil mass. The stress
 the quantity of the strongly linked water is 4%; diffusion in the soil mass as a function of depth can be defined
 the quantity of the weakly linked water is 27-4 ¼ 23%; using the Bousinesq relationship (Eq. 2).
 the quantity of the interstitial free water is 35-27 ¼ 8%. abs
sðext loadÞ ¼ ð2Þ
ða þ zÞðb þ zÞ
Even though the values estimated above are approximate, where a and b are respectively the foundation length and
they provide a quantitative estimation of the different kinds of width, and s is the applied load.
water present in the soil. On the basis of the Eq. (2), Ejjaaouani et al. (2008)
On the other hand, the value of the swelling pressure developed a similar equation which takes the diffusion angle
measured in the field was sg ¼ 150 kPa, while that measured into account (Eq. 3).
in the laboratory on undisturbed samples was sg lab ¼ 380 kPa. abs
The difference between these two values can be explained in sðext loadÞ ¼ ð3Þ
ða þ 2ztgβÞðb þ 2ztgβÞ
terms of the scale effect, sample handling and the manner in
which the laboratory test was carried out. The weight of the top layer can be defined as a function of
its hydraulic state by the following equation:
3. Theory: settlement prediction wm
wi
sðtop layer loadÞ ¼ ρz ð4Þ
wl
Compared to in-situ tests, which are both costly and time-
consuming (but with more accurate results which are also more where ρ is the soil density.
 S. Bensallam et al. / Soils and Foundations 54 (2014) 109–115 113

The total stresses considered at the depth “z” can be 4. Results and discussion
expressed as follows:
The proposed prediction model was developed to esti-
sðloadsÞ ¼ sðext loadÞ þ sðtop layer loadÞ ð5Þ
mate the settlement amplitude according to the intrinsic soil
characteristics and the environmental factors influencing the
abs wm
wi
sðloadsÞ ¼ þ ρz ð6Þ soil. From the field observations, the average thickness of the
ða þ 2ztgβÞðb þ 2ztgβÞ wl clay layer subjected to the in-situ tests was z¼ 1.7 m, and the
where β is the stress diffusion angle in the soil mass and can be main characterictics taken into consideration when predicting
expressed by the following equation (according to Ejjaaouani the settlement amplitude were collected from the field:
et al. (2008)): Δhg ¼ 0.093 m, wm ¼ ¼ 35%, wl ¼ 16%, and ρ¼ 18 kPa. The
 eð1
ðZ i =Z pl ÞÞe values of C and φ were determined from the laboratory shear
Zi test: C ¼ 65 kPa and φ ¼ 241. The deformation modulus “E” in
β¼ φ ð7Þ
Z pl the limit of the plastic zone used for calculation is considered
as the proportionality limit between the loading and the
where “φ” is the internal friction angle;“e” is the void ratio; settlement values.
“Z pi ” is the soil plastic zone below the foundation; Z i is the In the limit of the plastic zone Zpl¼ 0, by using the
concerned depth (0r Z i r Z pl ); and “C” is the soil cohesion. expression of the plastic zone thickness (Eq. 8) it is possible
sa ð1
sin φÞ
2C cos φ to deduce the applied load at the limit of proportionality:
Z pl ¼ exp
πtgφ ð8Þ
ρð1þ sin φÞ 2C cos φ
s¼ ¼ 200 kPa
According to Berezantsev et al. (1961), the soil deformability 1
sin φ
in the concerned active soil mass depth can be expressed by the The received applied load at the limit of proportionality is
following expression: then used to deduce the value of “E” using the Schleicher
ð1
ν2 Þ formula:
z ð9Þ s
E E ¼ C f ð1
ν2 Þb :
S
where ν is Poisson's ratio and E is the deformation modulus.
It is widely recognized that the volume change character is a where b represents the foundation width (b ¼ 1 m) for a
time dependent process. In the present model, the time factor is corresponding shape factor Cf ¼ 0.88. s and S are respectively
expressed by the following ratio: the applied load and the vertical displacement at the limit of
ti proportionality, determined from the loading curves (Fig. 3).
; ð10Þ The received value of “E” according to the Schleicher
tm
formula is E ¼ 1753 kPa.
where “tm” is the needed time to reach a dried state and “ti” is On the other hand, the value of “E” was verified as equal to
the considered time for calculation. 1750 kPa from the loading–unloading curve using the follow-
This prediction model is focused on the behavior of the ing equation.
active clay soil during the drying process from a saturation  
2ν2 si
si
1
state. The prediction of the settlement amplitude with time E ¼ ð1þ e0 Þ 1
ð13Þ
may be obtained using a direct calculation by subtracting a 1
ν ei
ei
1
combination of Eqs. (1), (6), (9) and (10) to the limit swelling In order to simplify calculations, the deformation modulus “E”
amplitude (Δhg). The analytical model to predict the settlement is represented by a single value in the limit of the plastic zone.
amplitude is expressed as follows (Eq. 11). The received The settlement amplitude calculations were performed for
prediction model depicts both the elastic and the plastic different water contents ðw1 ¼ 27%; w2 ¼ 20%; w3 ¼ 16% Þ
behaviors in the limit of the active zone. determined from a depth of 0.2 m from the expansive layer
 embedment level. The results of the ground level predicting
abs
Sr ¼ Δhg
settlement under loads as a function of drying time ti for
ða þ 2ztgβÞðb þ 2ztgβÞ
   ðwm
wi Þ=wm foundations FS-2, FS-3 and FS-4 are shown in Fig. 4. The
wm
wi ð1
ν2 Þ wm
wi ti observation of foundation FS-2 was conducted during a drying
þ ρz z
wl E wl tm time evolution t 1 ¼ 90 days when the average soil water
ð11Þ content was 27%. The observations of FS-3 and FS-4 were
carried out respectively in t 2 ¼ 270 days for a water content of
In the case where the applied load is negligible or equal to
w2 ¼ 20% and t 3 ¼ 450 days for a water content of w3 ¼ 16%.
zero (s¼ 0), the settlement amplitude of the ground level is
The final values of the shallow foundations settlement,
given by the following equation:
calculated on the basis of the applied load, are presented in
    Table 3. At the end of the first drying period t 1 ¼ 90 days, the
wm
wi 2 2 ð1
ν2 Þ t i ðwm
wi Þ=wm settlement was 6 mm at a shrinking water content wR ¼ 27%.
Sr ¼ Δhg
ρz ð12Þ
wl E tm At the end of the second drying period t 2 ¼ 270 days, the
114 S. Bensallam et al. / Soils and Foundations 54 (2014) 109–115

clay soil behavior. We can deduce that the experimental

and the prediction models follow the same logic; the foundation
settlement increases with increasing aeration time and applied
load. The predicting settlement amplitude values correspond
practically to the in-situ soil behavior during the drying process.

5. Conclusion

For soil-structure stability, it is necessary to predict the

imposed stress-strain conditions at a given point in a soil mass
due to the weather and the loading conditions. In this study,
a time-dependent prediction model was developed to quantify
the settlement amplitude of clay soil during the drying process,
capable of representing both the elastic and the plastic
Fig. 4. Calculated soil's shrinking curve in time.
behavior of the tested soil in the limit of the active zone.
The modeling theory was developed by taking the following
Table 3 aspects, all of which govern the soil behavior, into account: the
Foundations settlement according to the applied load calculated using the intrinsic soil characteristics, the hydraulic state, and the stress
proposed model. states. This prediction model is focused on the behavior of
active clay soil during the drying process from a satura-
Foundation Drying time Water content Applied load (kPa)
(days) (%) tion state considered as the swelling limit.
0 210 350 500 The modeling consistency was confirmed using in-situ full-
Foundation settlement (mm) scale test results. The performed in-situ test was developed to
follow and analyze the swelling–shrinkage behavior of active
FS-2 90 27 6 17.5 25.2 33.4
FS-3 270 20 20.9 41.4 55 69.7 clay soil, during a single wetting–drying process. The achieved
FS-4 450 16 33.5 55.6 70 86 analytical results show that the prediction model is in good
agreement with the in-situ tests, which reflects the real
behavior of the studied clay soil during the drying process.
While the prediction and the experimental test results
Table 4 showed good agreement, a limit was observed when the
Ground level settlement during the drying process. applied load was increased, producing considerably different
experimental and analytical results. However, because this
Foundation Water content (at Experimental Calculated

20 cm) settlement (mm) settlement (mm) modeling approach uses only intrinsic soil parameters that can
be easily obtained from the usual geotechnical tests, and
FS-1 12.5% – – because this approach takes all the factors influencing the
FS-1 35% – – soil behavior into consideration, this proposed analytical
FS-2 27–28% 4–6 4.55–5.94
FS-3 18–20% 20–29 20.9–26.84
approach is thought to be sufficient for use in predicting the
FS-4 14–16% 35–50 33.5–40.96 soil settlement amplitude, especially for active clay soil.

settlement was 20.9 mm for wR ¼ 20%: At the end of the third References
drying period t 3 ¼ 450 days, the settlement was 33.5 mm for
wR ¼ 16%, which corresponds to the shrinkage limit obtained Alonso, E.E., Lloret, A., 1982. Behavior of partially saturated soil in undrained
without a load ðs ¼ 0Þ in the laboratory. loading and step by step embankment construction. In: Proceedings of the
IUTAM Conference on Deformation and Failure of Granular Materials.
The results represented in Table 4 indicate that good Alonso, E.E., 1999. Modelling expansive soil behaviour. In: Key Note Lecture,
agreement was achieved between the predicted settlement Proceedings of the II International Conference on Unsaturated Soils,
amplitude and the measured settlement in-situ. The observed UNSAT 98, Beijing, China, vol. 2, pp. 37-70.
dispersion of some of the compared results can be explained by Alonso, E.E., Romero, E., Hoffmann, C., Garcia-Escudero, E., 2005. Expan-
the following: (1) the expansive layer thickness was not uniform sive bentonite–sand mixtures in cyclic controlled suction drying and
wetting. Eng. Geol. 81 (3), 213–226.
under the surface test, (2) the soil was not heterogeneous and (3) Annette, K., 1998. Constitutive model with time-dependent deformations. Eng.
the moisture migration over the soil mass was not even. Even Geol. 49 (285), 292.
though good agreement was reached between the predicted ARGIC-BRGM/RP-54862-FR. Vincent, M., Bouchut, J., Fleureau, J.-M.,
outcome and the experimental tests, a limitation was observed Masrouri, F., Oppenheim, E., Heck, J.-V., Ruaux, N., Le Roy, S., Dubus,
with an increase in the applied load, where a significant I., Surdyk. N., Oct 2006. Étude des mécanismes de déclenchement du
phénomène de retrait-gonflement des sols argileux et de ses interactions
difference was noted between the experimental and the analy- avec le bâti.
tical results. Nevertheless, it appears that the predicted Bensallam, S., Bahi, L., Ejjaaouani, H., Shakhirev, V., 2012. Shrinkage curve:
settlement amplitudes are quite representative of the studied experimental study and modelling. Int. J. Eng. 25, 203–210.
 S. Bensallam et al. / Soils and Foundations 54 (2014) 109–115 115

Berezantsev, V.G., Ksenofontov, A.I., Platonov, E.V., Sidorov, N.N., Nowamooz, H. 2007. Retrait/Gonflement des sols Argileux Compactés et
Yaroshenko, V.A., 1961. Mécanique des Sols et Fondations, Moscou. Naturels (Ph.D. thesis). Institut National Polytechnique de Lorraine, Nancy,
Ejjaaouani, H., Magnan, J.P., Shakhirev, V., 2008. Interactions des Fondations France.
et des Sols Gonflants: Pathologie, Calculs et études Expérimentales (Ph.D. Oka, F., Kodaka, T., Suzuki, H., Kim, Y.S., Nishimatsu, N., Kimoto, S., 2010.
thesis). École Nationale des Ponts et Chaussées, Paris, 248 pp. Experimental study on the behavior of unsaturated compacted silt under
Fredlund, D.G., Rahardjo, H., 1993. Soil Mechanics for Unsaturated Soil. John triaxial compression. Soils Found. 50 (1), 27–44.
Wiley & Sons, New York, 560. Rees, S.W., Thomas, H.R., 1993. Simulating seasonal ground movement in
Gens, A., Alonso, E.E., 1992. A framework for the behaviour of unsaturated unsaturated clay. J. Geotech. Eng., ASCE, 119; 1127–1143.
expansive clays. Can. Geotech. J. 29, 1013–1032. Popescu, E., 1986. A comparison between the behavior of swelling and of
Lebedev, A.F., 1936. L'eau du Terroir et Souterraine, Selkhozgiz. collapsing soils. Eng. Geol. 23 (2), 145–163.
Mitchell, J.K., Soga, K., 2005. Fundamentals of Soil Behavior, third ed. John Wang, Q., Tang, A.M., Cui, Y., Delage, P., Barnichon, J., Ye, W., 2013. The
Wiley & Sons Inc., New York, 558. effects of technological voids on the hydro-mechanical behaviour of
Nelson, J.D., Miller, D.J., 1992. Expansive Soils: Problems and Practice compacted bentonite sand mixture. Soils Found. 53 (2), 232–245.
in Foundation and Pavement Engineering. John Wiley & Sons Ltd.,
New York.