Professional Documents
Culture Documents
4, 137–147 137
Quite often, foundation engineers encounter a thick layer Très souvent, les ingénieurs s’occupant de fondations
of collapsible soil. This type of soil possesses considerable rencontrent une couche épaisse de sol sujette à l’effondre-
strength, which is largely lost when the soil becomes wet. ment. Ce type de sol possède une résistance considérable
Excessive settlement and lateral deformation accompany qui disparaı̂t en grande partie lorsque le sol s’humidifie.
this significant loss in strength. As an alternative to Un affaissement excessif et une déformation latérale
conventional deep foundations, stone columns encapsu- accompagnent cette perte de résistance significative. Pour
lated in geofabric reinforcement may be used to transmit remplacer les fondations profondes conventionnelles, on
foundation loads to suitable bearing strata below the peut utiliser des colonnes de pierres encapsulées dans un
collapsible soil layer. This paper presents an experimental renfort géotextile afin de transmettre les charges de fonda-
investigation on the performance of stone columns encap- tion à une strate porteuse située en dessous de la couche
sulated in geofabric installed in a collapsible soil layer sujette à l’effondrement. Cet exposé présente une investi-
and subjected to inundation. The carrying capacity of the gation expérimentale de la performance de colonnes de
columns and their settlement characteristics were investi- pierres encapsulées dans des géotextiles et installées dans
gated. Columns made of sand were tested with different une couche de sol sujette à l’effondrement et soumise à
lengths, degrees of inundation and different strengths of l’inondation. Nous avons étudié la capacité porteuse des
geofabric. Based on the results of the present experimental colonnes ainsi que leurs caractéristiques d’affaissement.
investigation, it can be reported that unreinforced sand Nous avons testé des colonnes faites de sable avec
columns in collapsible soil did not contribute significantly plusieurs longueurs, degrés d’inondation et résistances
to the soil’s performance. Furthermore, premature failure géotextiles. En nous basant sur les résultats de la présente
of the column was noted. The carrying capacity of investigation expérimentale, nous pouvons rapporter que
encapsulated sand columns increases owing to an increase les colonnes de sable non renforcées dans le sol sujet à
in strength of the geofabric material and/or an increase in l’effondrement ne contribuent pas de manière significative
column length. The settlement of the column’s head due to à la performance du sol. De plus, nous avons noté la
external loading and inundation decreases owing to an défaillance prématurée de la colonne. La capacité porteuse
increase in column rigidity and/or an increase in column des colonnes de sable encapsulées augmente en raison
length (up to a maximum value equal to the thickness of d’une augmentation de la résistance des matériaux géotex-
the collapsible soil layer). Theoretical models were devel- tiles et/ou d’une augmentation de la longueur de colonne.
oped to predict the carrying capacity and settlement of Le tassement de la tête de colonne à cause de la charge
these columns. Comparison of the results predicted by the extérieure et de l’inondation diminue en raison d’une
proposed theory and experimental results of the present augmentation de la rigidité de la colonne et/ou d’une
investigation and those reported in the literature showed augmentation de la longueur de la colonne (jusqu’à une
good agreement. valeur maximum égale à l’épaisseur de la couche de sol
sujette à l’effondrement). Nous avons développé des
modèles théoriques pour prédire la capacité porteuse et
l’affaissement de ces colonnes. Les résultats prédits par la
théorie proposée, les résultats expérimentaux de l’investi-
Keywords : collapsible soil; foundation engineering; gation et ceux rapportés dans la documentation montrent
geofabrics; inundation; stone columns une bonne corrélation.
Pu ultimate carrying capacity confirms that design theories are lagging behind the con-
QR correction factor for cases in which failure does not occur struction of these columns.
QT correction factor for effects of delayed installation In the literature, reports can be found describing the
q surcharge on ground surface of soil failure mechanism of these columns by bulging at about four
S0 settlement of unreinforced column diameters from the top of the column (Poulos and Mattes,
S1 settlement of top of collapsing layer
1969; Williams, 1969; Hughes and Withers, 1974). Others
r0 initial radius of column
t thickness of geofabric material have reported that the carrying capacity of a stone column
˜ total settlement of stone column, caused by inundation under in loose soil could be reduced by 50% in the case of raising
external load P the groundwater table (Simpson et al., 1989). Grigoryan
˜H foundation settlement on collapsible soil with stone column (1967), Zur and Wiseman (1973), Maswoswe (1985) and
ª0 soil unit weight Lawton et al. (1991) have reported that, during the inunda-
1 elastic settlement of column due to axial load P tion of a soil element in a collapsible soil layer, the lateral
2 settlement caused by downdrag force due to consolidation of pressure around that element decreases drastically, which
surrounding soil will be further manifested in additional axial deformation or
3 settlement due to lateral deformation of column excessive settlement.
4 settlement component of layer below column tip, caused by
inundation
FS axial movement of column at failure
n settlement of nth slice Experimental programme
r0 lateral deformation of column
v vertical settlement Tests were performed on prototypes of the sand column
r radial strain model in a stress-controlled chamber that contained loose,
Ł cirumferential strain collapsible fill. The set-up used in this investigation was
9h0 effective lateral stress of soil prior to column installation
similar to that used by Hamitouche and Ayadat (1996). The
9hmax maximum effective lateral pressure that soil and geotextile
chamber was made of cylindrical steel, with 390 mm inside
can support
9r effective lateral pressure around column diameter, 520 mm depth and 17.5 mm wall thickness (Fig. 1).
9v vertical stress acting on column The chamber was resting on a circular steel plate 25 mm
9vlim vertical stress acting on column thick, which was bolted to its bottom flange. Another 25 mm
Ł tangential stress thick circular plate served as a lid on the top end of the
a tensile strength of geofabric material cylinder with a hole in the centre of 40 mm diameter. In
9 angle of shearing resistance of stone column material order to minimise the friction between the collapsible soil
and the walls of the chamber, a thin layer of silicone grease
was applied to the walls before filling up the chamber with
soil.
The collapsible soil used in this investigation was a
Introduction mixture gap-graded soil. The mixture consisted of 78%
concrete sand, 10% Leighton buzzard sand (less than 90 m),
Collapsible soils can be defined as unsaturated soils that go and 12% speswhite kaolin clay. The grain-size distribution of
through a radical rearrangement of particles and loss of this mixture is presented in Fig. 2. The values of the
volume upon wetting with or without additional loading. coefficient of earth pressure at rest (k0 ), the angle of shearing
Collapsible soils are found in formations such as loess, resistance (9), and the cohesion (c9) for this material were
alluvial, residual, aeolian subaerial, colluvial and gypsifer- determined from the conventional triaxial consolidated
ous silts.
In the literature, reports can be found dealing with
methods of identifying collapsible soil and determining the Loading lever
amount of settlement that may occur when it is wetted
Pressure plate
(Hassani and Goel, 1982; Lutenegger and Saber, 1988). LVDT
tube
Some researchers have suggested soil stabilisation techni- Klinger valve
ques to improve its performance (Bara, 1976), and recom- Pressure plate O-ring
mendations have been made to install reinforced stone
columns with geofabric material in weak and compressible
soils as a soil improvement technique (Gray et al., 1982;
Gorle and Thijs, 1989; Al-Joulani, 1995). Nevertheless, Bleed valve
practising engineers are still unable to design conventional
foundations on this soil with a high degree of confidence.
Rising construction costs along with a high level of
uncertainty will undoubtedly make stone columns a more
Column
attractive alternative to conventional pile foundations when
dealing with collapsible soil.
Collapsible soil Terram
Stone columns are considered as a viable technique as a
ground treatment method. Their use in foundation engineer-
ing goes back for more than 50 years for reinforcement of
soft soil. In spite of the frequent recommendations to use
stone columns as a treatment for collapsible soils (Bara,
1976; Fargher et al., 1979; Ronan, 1980), there is little that can
be found in the literature dealing with the subject matter. Frame
From a tap
Stroud and Mitchell (1990) reported a case history where
stone columns failed to strengthen chalk fill layer. This Fig. 1. Cross-section of testing apparatus
138
Encapsulated stone columns for collapsible soil
100
ing the sand columns. Four non-woven geofabrics were
90 tested in this investigation (Terram 700, 1000, 1500 and
2000). Table 2 summarises the physical and mechanical
80
properties of this material. This information was compiled
Percentage passing: %
139
Ayadat and Hanna
load–settlement curves for collapsible soil with and without increases with settlement linearly up to approximately 20–
unreinforced sand column. It can be seen that the presence 30% of the ultimate load, beyond which the column exhibits
of the unreinforced sand column has slightly improved both relatively larger amounts of settlement owing to a small
the carrying load and the settlement measurements of this increase in the load. The performance of the collapsible soil
soil. The performance improved further when the sand was improved still further by the presence of reinforced
columns were driven to the entire depth of the collapsible sand columns with geofabric material. Furthermore, the
layer. performance was significantly enhanced by the increase of
Figure 4 presents typical load–settlement curves for the strength (stiffness) of the geofabric material. It was
collapsible soil with sand columns reinforced with geofabric observed that, before inundation, sand columns failed by
material. It can be seen that the load on these columns bulging at relatively shallow depths (typically less than four
140
Encapsulated stone columns for collapsible soil
160 110
100
140
90
120
80
100 70
Load: N
Load: N
80 60
50
60
40
Full inundation
40 Partial inundation
30
SC (L 5 410 mm) 20
20 SC (L 5 250 mm)
Soil without SC SC (L 5 410 mm)
SC (L 5 250 mm)
10 Soil without SC
0
0 0·5 1·0 1·5 2·0 2·5 3·0 0
Settlement: mm 0 5 10 15 20 25 30 35 40 45 50 55 60 65
Settlement: mm
Fig. 3. Load–settlement curves for collapsible soil with/without sand
columns (SC) Fig. 5. Load–settlement curves for collapsible soil with/without sand
columns after partial or full inundation (load ¼ 30%)
400
Soil alone
360 Sand column
Column in T700
Column in T1000 column during inundation (Carr, 1970; Selig and Grangaard,
320 Column in T1500
Column in T2000
Rigid pile 1970). This proved not to be effective because of the damage
280 and disturbance caused to the instruments during failure of
240
the sand column. Figs 6(a) and 6(b) show respectively the
sand column in place before and after failure, which was
Load: N
141
Ayadat and Hanna
Fig. 6. Stone column in collapsible soil subjected to inundation caused by rise of water table: (a) before inundation; (b) during inundation
550 2·6
Soil alone
500 Sand column L 5 250 mm
Column in T700 2·4 L 5 300 mm
Column in T1000
450 Column in T1500 L 5 410 mm
Column in T2000
Carrying capacity ratio, CCR
350 2·0
Load: N
300
1·8
250
1·6
200
150 1·4
100
Full inundation 1·2
50
0 1·0
10 000 100 000 1 000 000 10000000
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75
Modulus of elasticity of piles, Ep: kPa
Settlement: mm
Fig. 7. Load–settlement curves for various foundation supports, of length Fig. 8. Variation of carrying capacity ratio (CCR) with stiffness of columns
L ¼ 410 mm, after full inundation under applied load equal to 80% Pu
Si
¼ (2)
S0
where Si is the settlement of the reinforced column, and S0 is 1·0
the settlement of the unreinforced column. P 5 20% Pu
Figure 9 shows the reduction factor plotted against the 0·9 P 5 50% Pu
P 5 80% Pu
modulus of elasticity of the sand column, Ep . It can be seen 0·8
that decreases with increase of the stiffness of the sand
Settlement reduction factor, â
142
Encapsulated stone columns for collapsible soil
lateral deformation is the main contributor to the total 9
v9 lim ¼ tan 2 þ h9 max (4)
settlement of that column. 4 2
Based on the results of the present experimental investi-
gation, it can be reported that partially penetrated columns where v9 lim is the maximum effective vertical stress acting
will settle excessively owing to the existence of a collapsible on the column; 9 is the angle of shearing resistance of the
layer below the column’s tip, which will be reduced owing material of the stone column; and h9 max is the maximum
to an increase of the column length. However, the increase effective lateral pressure that the soil and the geofabric can
in the column length will increase its contact area with the support, given by
surrounding soil, and hence will increase the column h9 max ¼ h9 s þ ˜ (5)
settlement caused by the downdrag force. Nevertheless, the
settlement due to the drag force is significantly less than where h9 s is the maximum effective lateral pressure pro-
the settlement produced by the collapse of the layer vided by the soil around the column, and ˜ is the
situated below the column’s tip. additional pressure that contributes to the reinforcement of
In this investigation, the collapse potential was introduced the column.
in terms of the collapse potential (CP), defined as the ratio of Based on the work of Hughes and Withers (1974) and
column settlement to the thickness of the collapsing layer: Hughes et al. (1975) it can be assumed that
˜H h9 s ¼ h9 0 þ kc9 (6)
CP ¼ (3)
H0
and
where ˜H is the settlement of the foundation on collapsible
ªL
soil with/without stone column, and H0 is the thickness of h9 0 ¼ k0 qþ (7)
2
the collapsing layer.
Figure 10 presents the variation of the factor CP with the where h9 0 is the effective lateral stress of the soil before
stiffness of the sand column, represented by the modulus of installing the column; c9 is the drained cohesion of the
elasticity, Ep . It can be seen that CP decreases significantly, collapsible soil; k0 is the coefficient of earth pressure at rest;
owing to the decrease of foundation settlement caused by q is the surcharge applied on the ground surface; L is the
the increase of column stiffness and the column’s depth of length of the stone column; and k is a coefficient (k ¼ 4;
embedment. Specifically, CP dropped from 13% for soil Hughes and Withers, 1974)
alone to about 4% for the sand column with L ¼ 410 mm, The geofabric material enveloping the stone column
encapsulated in T2000 fabric. This significant reduction in provides superior resistance to the lateral pressure than the
CP demonstrates the contribution of the reinforcement soil alone. This fabric, when stretched, exerts additional
material. pressure on the column that contributes to its reinforcement.
This pressure (˜) can be estimated based on the tensile
strength (a ) of the geofabric material. Referring to Fig. 11, at
equilibrium the following equation can be written:
Theory ˜ 2r0 ¼ 2a t (8)
Ultimate carrying capacity a t
˜ ¼ (9)
The ultimate carrying capacity of a stone column installed r0
in a soil mass is governed by the lateral confining earth
pressure mobilised in the surrounding soil. A stone column where r0 is the initial radius of the column, and t is the
encapsulated in a geofabric material can be simulated by a thickness of the geofabric material.
cylindrical element made of a cohesionless material and Substituting h9 s and ˜ in equation (5), the carrying
subjected to triaxial loading. The confining pressure is the ultimate load of a stone column encapsulated with geofabric
total of the lateral resistance of the soil and the geofabric can be estimated by the following equation:
enveloping the column. According to Briaud (1991), the
2 9 a t
vertical stress acting on an unreinforced stone column is v9 lim ¼ tan þ h9 0 þ kc9 þ (10)
given by 4 2 r0
10
maximum strength of the geofabric in equation (10) (i.e.
˜ ¼ a t=r0 ) will lead to overestimation of the confining
8
pressure around the column, and accordingly the carrying
SC in T700 capacity of the column. Thus in this investigation a coeffi-
6
SC in T1000 cient Æ was introduced in equation (10) as a reduction factor
for the value of a t/r0 , as follows:
4 SC in T1500
SC in T2000
L 5 250 mm
9 a t
v9 max ¼ tan 2 þ h9 0 þ kc9 þ Æ (11)
2 L 5 300 mm 4 2 r0
L 5 410 mm
0
0 10000 20000 30 000 40 000 50 000 60 000 70 000 80 000 90 000 where: 0 < Æ < 1.
Modulus of elasticity: kPa The experimental values of the coefficient Æ depend
Fig. 10. Variation of collapse potential (CP) with stiffness of columns, mainly on the stiffness of the column. The carrying capacity
subjected to full inundation under load P ¼ 20% Pu . SC, sand column values of the column obtained experimentally were used
143
Ayadat and Hanna
ÄP 5 Äó r0
ô ô
t 2 r0 t
(a) (b)
together with the proposed theoretical model (equation tion. Nevertheless, the comparison is regarded as acceptable
(11)), to provide proper evaluation for Æ. The different in the geotechnical engineering field.
values obtained for this coefficient are plotted against
column stiffness in Fig. 12. The following empirical formula
has been proposed for the coefficient Æ with a satisfactory Settlements
coefficient of correlation (R2 ¼ 0.972): The total settlement of a fully penetrating stone column in
:37 a collapsible soil subjected to an external load P, caused by
Æ ¼ 3:2 3 10 5
E1
p (12)
inundation, is made up of three components:
˜ ¼ 1 þ 2 þ 3 (13)
where Ep is the modulus of elasticity of the stone column.
Furthermore, the constant term in equation (12) depends where ˜ is the total settlement of a stone column, caused by
mainly on the type of reinforcing material (i.e. its tensile inundation under an external load P; 1 is the elastic
strength). settlement of the column due to the axial load P, 2 is the
The experimental results for the ultimate carrying capa- settlement caused by the downdrag force due to consolida-
cities of confined stone columns reported by Al-Joulani tion of the surrounding soil; and 3 is the vertical settlement
(1995) were compared with the predicted values produced due to the lateral deformation of the column.
by the present theory (see Table 6). It can be seen from Table The settlement components 1 and 2 were obtained from
6 that the theory overestimates the results by about 8–20%. the analytical models of Poulos and Davis (1975), developed
This may be explained by the fact that the coefficient Æ was for compressible piles in compressible soils, whereas 3 was
developed for the geofabric material used in this investiga- predicted based on the theoretical model of Hughes and
Withers (1974).
According to Poulos and Davis (1975), 1 and 2 are given
0·45 by
250 mm 300 mm
0·40 410 mm Pa L
1 ¼ (14)
Ep Ap
0·35
0·30
and
Coefficient á
0·25
2 ¼ FS QR QT (15)
Trend with
0·20
0·15 2qL2 ªL
FS ¼ c9 þ k0 tan 9 þ1 (16)
Ep d 3q
0·10
0·05
where Pa is the applied axial load; Ep is the modulus of
0 elasticity of the column; Ap is the cross-sectional area of the
0 10000 20000 30 000 40 000 50 000 60 000 70 000 80 000 90 000
column; d is the diameter of the column (d ¼ 2r0 ); FS is the
Modulus of elasticity, Ep of column: kPa
axial movement of the column at failure; QR is a correction
Fig. 12. Variation of coefficient Æ with column stiffness factor for cases when full pile–soil slip does not occur
Table 6. Comparison between computed capacities and laboratory test results (Al-Joulani, 1995)
Confining Geofabric strength: Ultimate carrying capacity: kPa
pressure: kPa kN/m
Laboratory test results Present theoretical results
(Al-Joulani, 1995)
103.5 0 1000 980.3
86.2 2400 2600
119.1 2750 3100
207 0 1500 1532
86.2 2750 3200
119.1 3000 3600
144
Encapsulated stone columns for collapsible soil
(Poulos and Davis, 1980); and QT is a correction factor for Pa L 2qL2 ªL r0 L
˜¼ þ c9 þ k0 tan 9 þ 1 QR QT þ 2˜P
the effects of delayed installation (Poulos and Davis, 1980). Ep Ap Ep d 3q Et
Assuming uniform and axisymmetrical deformation of the (30)
column, the summation of settlement contributions from a
discrete horizontal slice of the column can be written as For a floating column, the displacement of the column due
v ¼ n n (17) to the settlement component (4 ) of the layer situated below
the column’s tip, and caused by inundation, has to be added
r0
n ¼ 2 H n (18) to the total settlement (˜) of the column. By assuming that
r0
the soil settlement varies linearly from S1 at the surface to
zero at the base, the settlement 4 was expressed as
Substituting H0 ¼ nHn , the vertical settlement can be defined
as H0 L
4 ¼ S1 (31)
H0
r0
v ¼ 2 H0 (19)
r0 where H0 is the thickness of the collapsing layer, L is the
where n is the number of slices; n is the settlement of the length of the pile, and S1 is the settlement of the top of the
nth slice; Hn is the length of the nth slice; H0 is the total collapsing layer.
height of the column (H0 ¼ L); and r0 /r0 is the radial strain Comparisons between the measured and predicted values
of the column at the given slice. of the settlement reduction factor (see equation (2)) at the
The radial strain r0 /r0 was estimated by assuming top of a column, caused by inundation of the column under
compatibility of lateral deformations between the column an external axial load, are shown in Figs 13 and 14
material and the cylinder fabric around it, and further by respectively. It can be seen that good agreement between
considering axisymmetric strain conditions. Thus the cylin- these values can be found. Comparison of the measured and
der fabric of a thickness t and a radius r0 will be subjected to predicted values of triaxial test results on stone columns
internal pressure of ˜P. From equilibrium: encapsulated with geogrid (Al-Joulani, 1995) is given in Fig.
15. It can be seen from Fig. 15 that there is reasonable accord
2r0 ˜P ¼ 2 Ł t (20) between the model behaviour and the experimental data.
Therefore
r0 1·0
Ł ¼ ˜P (21) Predicted
t 0·9 Experimental
Given 0·7
Ł ¼ EŁ (22) 0·6
r0 r0 1·0
r ¼ ¼ ˜P (25)
r0 Et Measured
0·9 Sand column
By substitution in equation (19): Column in T1000 Sand column
0·8 Column in T2000
r0 Rigid pile
Settlement reduction factor, â
v ¼ 2 H0 ˜P (26) 0·7
Et
0·6
Therefore
r0 L 0·5
3 ¼ 2˜P (27)
Et 0·4 Column in T1000
where 0·3
Column in T2000
˜P ¼ v9 Kas r9 (28) 0·2
2 9 0·1
______ Predicted
Kas ¼ tan (29)
4 2 Rigid pile
0
0 2·5 5·0 7·5 10·0 12·5 15·0 17·5 20·0 22·5 25·0
v9 is the vertical stress acting on the column, and r9 is the Ratio L/d
effective lateral pressure around the column. Fig. 14. Variation of settlement reduction factor (equation (2)) with ratio
Thus the total settlement can be expressed as L/d (P ¼ 20%)
145
Ayadat and Hanna
146
Encapsulated stone columns for collapsible soil
147