Vous êtes sur la page 1sur 8

MECHANICS

RESEARCH COMMUNICATIONS

Mechanics Research Communications 29 (2002) 375382


www.elsevier.com/locate/mechrescom

Study of the swelling behavior of a compacted soil


using exible odometer
T. Windal, I. Shahrour

Department of Civil Engineering, Laboratoire de M


ecanique de Lille, EUDIL University of Sciences and Technologies of Lille,
59 655 Villeneuve dAscq cedex, France
Received 4 December 2001

Abstract
This paper includes an experimental study of the swelling behavior of a compacted soil. The study is performed using
a exible odometer, which allows for lateral deformation during soil expansion and the measurement of the lateral swell
pressure. The paper is composed of two parts. The rst one describes the exible device used in this study while the
second presents experimental results and discusses the inuence of the rigidity of the odometer ring on the swelling
behavior of a compacted soil.
2002 Elsevier Science Ltd. All rights reserved.

1. Introduction
The design of civil engineering structures such as tunnels and retaining walls in expansive soil requires
a reliable determination of the lateral swell pressure. Generally, the characterization of expansive soil is
carried out using the classic odometer, which permits the measurement of the axial swelling deformation
and the swell pressure under a condition of zero lateral deformation. Yet, this condition is not representative of in situ conditions particularly in the case of exible structures. Several studies have showed that the
swell pressure highly decreases if a small deformation of the soil sample is allowed during the swelling
process (Ofer, 1981; Kabbaj, 1989). Steiner (1993) and Chen and Huang (1987) also showed that swell
pressure measured in laboratory using classic odometer is generally higher than that observed in situ. This
phenomenon can be partially attributed to the high lateral stiness of the conventional odometer ring.
In this paper, we suggest to study the swelling behavior of a compacted soil using a exible odometer,
similar to that used by Franklin (1984). This device allows for lateral deformation during soil expansion
and for the measurement of the corresponding lateral swell pressure. The lateral expansion of the sample is
controlled through the stiness of the odometer ring. The variation of the ring stiness allows for the realization of a wide range of swelling tests, which can vary from free-swelling tests (very exible rings) to

Corresponding author.
E-mail address: isam.shahrour@eudil.fr (I. Shahrour).

0093-6413/02/$ - see front matter 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 9 3 - 6 4 1 3 ( 0 2 ) 0 0 3 0 8 - 7

376

T. Windal, I. Shahrour / Mechanics Research Communications 29 (2002) 375382

traditional swelling tests (very sti rings). These tests provide experimental data, which can be used for the
validation of three-dimensional swelling constitutive relations such as those proposed by Einstein and
Bischo (1976), Wittke and Pierau (1979) and Kiehl (1990). Moreover, it makes possible to investigate the
swelling behavior of anisotropic soils.

2. Experimental device
2.1. Description
The device used in this study is composed of an odometer loading frame and a thin odometer ring. The
ring is instrumented with strain gages, which allow for the measurement of the lateral strain. The lateral
swell pressure is deduced from the lateral strain measurement using a calibration curve established for each
ring. The magnitude of the lateral swelling strain is controlled by the relative stiness between the ring and
the soil sample.
The stiness of the ring Kr denotes the relation between the internal pressure pl and the radial strain er :
pl Kr er

According to the theory of elasticity, the expression of Kr is given by


Kr

eEr
R

where Er , e and R refer to the Youngs modulus and the thickness and radius of the ring, respectively. In this
study, aluminum is used for sti rings, while an alloy of copper and beryllium is used for exible rings. The
aluminum ring is 1 mm thick; the thickness of the copper rings is equal to 0.15 mm.
2.2. Calibration of the exible ring
Flexible rings were calibrated by means of compression odometric tests performed on a rubber specimen.
According to the theory of elasticity, the application of an axial stress ra on a rubber sample placed inside a
exible ring induces the following lateral pressure:
mc
pl
ra
3
1  mc KEcr
Ec and mc denote the rubber Youngs modulus and Poissons ratio, respectively. As the stiness of the rubber
Ec (3 MPa) is very small in comparison with that of the ring Kr (between 800 and 3000 MPa) and as
Poissons ratio is equal to 0.5, Eq. (3) yields
pl ra

Note that the use of Eq. (4) permits the calibration of the each exible ring from odometer tests on the
rubber specimen. The stiness of the ring is determined from Eq. (1).
2.3. Selection of the rings stiness
Selection of the ring stiness for particular application mainly depends on the soil-structure relative
stiness. The conventional odometer can be used for sti structures concerned by soft soils. For exible
structures and cases lacking engineering experience, the ring stiness can be estimated from preliminary
analyses of the soil-structure interaction using swelling properties determined from conventional odometer

T. Windal, I. Shahrour / Mechanics Research Communications 29 (2002) 375382

377

tests. Indeed, such analyses permit an estimation of both stresses and strains induced by soil swelling, and
consequently allow for the determination of the ring stiness and characteristics using Eqs. (1) and (2). This
procedure has to be veried by further analyses using results obtained with the exible odometer.

3. Study of the expansion of a compacted soil


3.1. Experimental procedure
The exible odometer was used to study the expansion of a compacted soil. Soil samples were cut in
blocks prepared by a dynamic compaction of Bavent clay at a water content w 24%. The characteristics of this clay are plastic limit wP 28%, liquid limit wL 44%, plasticity index IP 16% and shrinkage
limit ws 12:5% (Al Shihabi et al., 2002). Tests were performed on air-dried samples at an initial water
content ranging between 11.5% and 14%.
Two rings were used in this study a exible ring with Kr 850 MPa and a sti ring with Kr 3045 MPa.
With each ring, two series of tests were performed. The rst series comprises free-swelling tests performed at
four values of the axial stress (ra 63, 173, 340, 732 kPa). The second series concerns swelling tests performed according to the HuderAmberg method (Huder and Amberg, 1970). The rst series allows for the
determination of the swelling characteristics of the soil under a constant overburden pressure while the
second series permits the determination of the soil response to unloading stress paths such as those induced
by soil excavation.
3.2. Free-swelling tests
Results of free-swelling tests are illustrated in Figs. 1 and 2. It can be observed that the evolution of the
axial strain and the lateral pressure induced by the soil expansion depends on both the axial stress and
the stiness of the odometer ring. At low axial stresses, the lateral swell pressure reaches a peak-value. The
magnitude of this peak decreases with the increase in the axial stress and disappears for high stress values.
The presence of a peak in the response of expansive soils was observed on laboratory tests (Erol and

Fig. 1. Swelling tests performed with the exible ring (Kr 850 MPa): (a) evolution of the axial strain and (b) evolution of the lateral
pressure.

378

T. Windal, I. Shahrour / Mechanics Research Communications 29 (2002) 375382

Fig. 2. Swelling tests performed with the sti ring (Kr 3045 MPa): (a) evolution of the axial strain and (b) evolution of the lateral
pressure.

Ergun, 1994) and on large-scale experimentation (Chen and Huang, 1987; Joshi and Katti, 1980). According to Chen and Huang (1987), the reduction of the lateral pressure is due to a gradual change in the
soil structure and clay particle orientation associated with the saturation process. This phenomenon can
also be attributed to stress relaxation phenomena in the soil. Analysis of the evolution of the axial strain
shows that the rate of swelling increases with the augmentation of the axial stress. For example, the time
required to develop 50% of the nal axial swell increases from 0.03 to 0.22 day when the axial stress ra
decreases from 732 to 63 kPa.
The inuence of the axial stress on the axial swelling and lateral swell pressure is illustrated in Tables 1
and 2. It can be observed that the increase in the axial stress causes a decrease in the axial swelling strain
and an increase in the lateral swell pressure. Tests performed with the exible ring (Kr 850 MPa) show
that the increase in the axial stress from 63 to 732 kPa leads to an increase in the lateral swell pressure from
403 to 566 kPa and a decrease in the axial swelling strain esa from 7.8% to 0.2%. The same increase in the
axial stress in the sti ring (Kr 3045 MPa) induces an increase in the lateral swell pressure from 675 to
1187 kPa and a decrease in the axial swelling strain esa from 9.4% to 1.8%. The ratio of the maximum lateral
Table 1
Inuence of the axial stress on the expansion of a compacted Bavent clay/results obtained with the exible ring (Kr 850 MPa)
ra (kPa)
Axial swell strain esa (%)
Lateral swell pressure rs1 (kPa)
rs1 =ra

63
7.8
403
6.4

173
3.2
461
2.7

340
2.5
500
1.5

732
0.2
566
0.8

Table 2
Inuence of the axial stress on the expansion of the compacted Bavent clay/results obtained with the sti ring (Kr 3045 MPa)
ra (kPa)
Axial swell strain esa (%)
Lateral swell pressure rs1 (kPa)
rs1 =ra

63
9.4
675
10.7

173
5.5
907
5.2

340
3.9
1167
3.4

732
1.8
1187
1.7

T. Windal, I. Shahrour / Mechanics Research Communications 29 (2002) 375382

379

swell pressure to the axial stress (rsl =ra ) is also reported in Tables 1 and 2. This ratio is highly dependent on
both the stiness of the odometer ring and the applied axial stress. For the exible ring, it decreases from
6.4 to 0.8 when the axial stress increases from 63 to 732 kPa. For the sti ring, this ratio decreases from 10.7
to 1.7 when ra increases from 63 to 732 kPa. This result indicates that attention should be paid for soil
expansion at shallow depths, because such situation may induce a high lateral swell pressure compared to
the axial stress. Consequently, this may lead to extension-type soil failure.
Concerning the inuence of the ring stiness on the soil expansion, it can be noted that the increase in the
ring stiness causes an increase in both the axial swelling strain esa and lateral swell pressure rsl . At a low
axial stress (ra 63 kPa), the increase in the ring stiness from 850 to 3045 MPa induces an increase of
about 20% in the axial swelling strain and an increase of about 67% in the lateral swell pressure. At an axial
stress ra 340 kPa, the same increase in the ring stiness induces an increase of about 50% in the axial
swelling strain and an increase of about 133% in the lateral swell pressure. This result well agrees with
experimental results reported by Ofer (1981), which showed that the authorization of a lateral deformation
during soil expansion leads to a signicant decrease in the swell pressure.
Fig. 3 illustrated the variation of the axial swelling strain (esa ) in terms of the logarithm of the axial stress
(ra ). It can also be observed that experimental results are in good agreement with the relation generally
assumed for conventional odometer (Grob, 1972):
 
ra
s
ea Ks log
5
rsa
where Ks denotes the swelling index, while rsa designates the axial swell pressure that corresponds to zero
axial strain in the (esa , logra ) plane. Furthermore, it can be observed that the ring stiness slightly aects
the swelling index Ks (0.068 for the exible ring and 0.07 for the sti ring), but signicantly aects the axial
swell pressure. Indeed, when the ring stiness Kr increases from 850 to 3045 MPa, the swelling index Ks
increases from 0.068 to 0.07, while the axial swell pressure increases from 730 to 1225 kPa. In comparison
with the lateral swell pressure, it can be observed that the axial swell pressure in the sti ring (rsa 1225
kPa) is close to the lateral swell pressure measured with the same ring at high axial stress (rs1 1187 kPa),
while it is about 30% higher than rs1 measured with the exible ring.

Fig. 3. Free-swelling tests: inuence of the axial stress on the axial swelling strain.

380

T. Windal, I. Shahrour / Mechanics Research Communications 29 (2002) 375382

3.3. HuderAmberg tests


Fig. 4 shows results of swelling tests carried out according to the HuderAmberg method (Huder and
Amberg, 1970) using exible and the sti rings. The path (ABCD) corresponds to the stress path experienced by the soil sample before the swelling test, (DE) corresponds to free-swelling under a constant
axial stress, while (EF) denotes swelling induced by the reduction of the axial stress.
The evolution of both axial strain and lateral pressure during swelling is depicted in Fig. 5. It can be
observed that each reduction in the axial stress causes an immediate increase in the axial strain and an
immediate decrease in the lateral pressure. This behavior is then followed by a small variation before full
stabilization stage. The swelling process and magnitude depend on both the axial stress and the ring
stiness. Table 3 summarizes the ratio between the induced lateral pressure (Drs1 ) and the axial stress

Fig. 4. Variation of the axial strain during HuderAmberg tests (1970): (a) exible ring (Kr 850 MPa) and (b) sti ring (Kr 3045
MPa).

Fig. 5. Inuence of the ring stiness on the evolution of the axial strain and lateral pressure during HuderAmberg tests.

T. Windal, I. Shahrour / Mechanics Research Communications 29 (2002) 375382

381

Table 3
Inuence of the axial stress and the stiness of the ring on the ratio Drs1 =Dra (HuderAmberg tests)
Unloading step (kPa)
Flexible ring (Kr 850 MPa)
Sti ring (Kr 3045 MPa)

732340
0.42
1.04

340173
0.65
1.37

17363
0.94
1.94

6326
1.23
2.33

Fig. 6. HuderAmberg tests: inuence of the ring stiness and axial stress on the axial swelling strain.

decrement (Dra ). This ratio increases with both the increase in the ring stiness and the reduction in the
axial stress. Test performed with the exible ring (Kr 850 MPa) shows that the ratio Drs1 =Dra increases
from 0.42 at the rst unloading step to 1.23 at the last step, while for the sti ring (Kr 3045 MPa) this
ratio increases from 1.04 at the rst step to 2.33 at the last step.
Fig. 6 depicts the variation of the axial swelling strain esa in terms of the logarithm of the axial stress. It
shows that the swelling index Ks increases from 0.0316 to 0.0512 when the ring stiness Kr increases from
850 to 3045 MPa. For the same variation in the ring stiness, the axial swell pressure increases from 695 to
1470 kPa. It can be noted that values of Ks measured using the HuderAmberg method are lower than those
measured under free-swelling condition, in particular for the exible ring where the dierence between the
two values is about 53%. On the other hand, the axial swell pressure is less inuenced by the test procedure.
In this case, swell pressure measured by the HuderAmberg method is about 16% higher than that measured by the free-swell procedure.

4. Conclusion
This paper includes a study of the swelling behavior of a compacted soil using a exible odometer, that
allows for a lateral deformation of the soil during expansion and the measurement of the swell lateral
pressure. This device makes it possible to investigate soil expansion under dierent stress paths. Also, it
provides valuable experimental data which can be used to verify three-dimensional constitutive relations
related to expansive soils.

382

T. Windal, I. Shahrour / Mechanics Research Communications 29 (2002) 375382

Experimental results obtained in this study showed that an increase in the stiness of the odometer ring
will lead to signicant increase in both lateral swell pressure and axial swell strain. Furthermore, reducing
the soil swelling in one direction leads to a notable increase in the soil swelling in the other directions.
At low axial stresses, lateral swell pressure attains peak values that are generally higher than those of the
applied axial stress. This signies that swelling of soils at shallow depths may lead tt extension-type soil
mass failure.

References
Al Shihabi, O., Shahrour, I., Mieussens, C., 2002. Experimental study of the inuence of cyclic dryingwetting on the compressibility
of a compacted soil. In: Proc. 3rd Int. Conf. Unsaturated Soils, UNSAT 2002, March 2002, Recife, Brazil, Vol. 2, AA. Balkema
Publishers, pp. 541545.
Chen, F.H., Huang, D., 1987. Lateral expansion pressure on basement walls. In: Proc. 6th Int. Conf. Expansive Soils, New Delhi, 14
December 1987 (1). pp. 5559.
Einstein, H.H., Bischo, N., 1976. Dimensionnement des tunnels en roche gonante. Tunnels et ouvrages souterrains 15, 109119.
Erol, O., Ergun, U., 1994. Lateral swell pressure in expansive soils. In: Proc. 13th ICSMFE, New Delhi. pp. 15111514.
Franklin, J.A., 1984. A ring swell test for measuring swelling and shrinkage characteristics of rock. International Journal of Rock
Mechanics and Mining Sciences 21 (3), 113121.
Grob, H., 1972. Schwelldruck im Belchentunnel. In: Proc. Int. Symp. f
ur Untertagebau, Lucerne.
Huder, J., Amberg, G., 1970. Quellung in Mergel, Opalinuston und Anhydrit. Schweizerische Bauzeitung 43, 975980.
Joshi, R.P., Katti, R.K., 1980. Lateral pressure development under surcharge. In: Proc. 4th Int. Conf. Expansive Soils, Denver (1).
pp. 227241.
Kabbaj, M., 1989. Sols gonants: mesure des proprietes en laboratoire. Revue Marocaine du Genie Civil 26, 1727.
Kiehl, J.R., 1990. Ein dreidimensionales Quellgesetz und seine Anwendung auf den Felshohlraumbau. In: Proc. 9th Natn.
Felsmechanik Symp., Aachen, Germany. pp. 185207.
Ofer, Z., 1981. Laboratory instrument for measuring lateral soil pressure and swelling pressure. Geotechnical Testing Journal 4 (4),
177182.
Steiner, W., 1993. Swelling rock in tunnels: rock characterization, eect of horizontal stresses and construction procedures.
International Journal of Rock Mechanics and Mining Sciences 30 (4), 361380.
Wittke, W., Pierau, B., 1979. Fundamentals for the design and construction of tunnels in swelling rock. In: 4th ICRM Montreux (2).
pp. 719729.