You are on page 1of 11

Geotextiles and Geomembranes 28 (2010) 292–302

Contents lists available at ScienceDirect

Geotextiles and Geomembranes


journal homepage: www.elsevier.com/locate/geotexmem

Geosynthetic-encased stone columns in soft clay: A numerical study


S.R. Lo a, *, R. Zhang b, J. Mak c
a
University of New South Wales, ADFA Campus, Canberra, ACT2600, Australia
b
University of New South Wales, ADFA Campus, Canberra, ACT2600, Australia
c
Roads and Traffic Authority, New South Wales, Australia

a r t i c l e i n f o a b s t r a c t

Article history: This paper presents the findings of a series of numerical studies on the contribution of geosynthetic
Received 27 May 2008 encasement in enhancing the performance of stone columns in very soft clay deposits. In this study, the
Received in revised form imposed loading is from a fill embankment, and the stone columns act like reinforcements. Observed
7 January 2009
settlement of a trial embankment built on very soft clay strengthened with stone columns indicated that
Accepted 15 May 2009
Available online 14 October 2009
the stone columns alone were not adequately effective in reducing settlement because the very softy clay
could not provide adequate confining stress to the stones. An alternative system utilizing geosynthetic
encasement was examined numerically. As the primary issue is the development of settlement with time
Keywords:
Geosynthetic enforcement after the completion of stone column installation, a fully coupled analysis was performed. To reduce the
Soft clay computational effort, a unit cell idealization was adopted. This study showed that the use of geosynthetic
Stone columns encasement has the potential of significantly enhancing the effectiveness of stone columns in very soft
Coupled analysis clay and the simplified analysis presented in earlier work is valid. Furthermore, the predicted perfor-
mance was found to be insensitive to assumed stiffness parameters of the compacted stone. However, it
was found to be dependent on the locked-in stress in the geosynthetic encasement induced during
installation.
Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction Oh et al. (2007) reported the settlement performance of a 4 m


high trial embankment constructed on soft estuarine clay improved
Stone columns have been used as a ground improvement by stone columns. The observed settlement at natural ground level
technique for a wide range of projects. Potential functions of the (NGL), over a period of 457 days, of the stone columns treated
stone columns are increase in bearing capacity, reduction in total section was only slightly less than that of the untreated section. The
settlement, and reduction in post-construction settlement (by clay of this site, which is located in south-east Queensland,
performing the function of vertical drains). Stone columns may be Australia has a high compressibility Index, CC, of w1.5. It is also
used to support a column load in a manner similar to piles, i.e. the noted that there is no stiff crust overlying the soft clay layer, and
external load is applied to the top of a stone column but not to the water table is close to natural ground level. It was hypothesized
surrounding soft soil. Alternatively, a large number of stone that the stone columns bulged and compressed excessively because
columns can be used to strengthen a weak soil stratum for sup- of lack of confinement. It is pertinent to note that such ground
porting a fill embankment. In such an application, the stone condition is not uncommon for estuarine deposits along the coast
columns function in a manner similar to soil reinforcement. This between northern New South Wales and south-east Queensland.
paper is for the latter application. Design procedures for deter- The observed settlement performance of the above trial embank-
mining the bearing capacity of stone columns are well documented ment sections founded on a very soft clay strengthen with stone
in Barksdale and Bachus, 1983. The prediction of settlement is, columns raises the question about the effectiveness of stone
however, less certain. For a road embankment section that leads to columns in reducing settlement of very weak deposits. Concerns on
a piled abutment, the role of stone columns in limiting the settle- the effectiveness of stone columns in soft clay were also raised by
ment is crucial. Applying a generous factor of safety to a calculated Malavizhi and Ilamparuthi (2007).
bearing capacity may not be an effective means to guaranteed The strength and stiffness of a stone column are dependent on
compliance with settlement limit. the effective confining stress provided by the surrounding soil. A
high effective confining stress can normally be induced by the
installation process, with the stones being expanded against the
* Corresponding author. surrounding soil. For very soft clay, this may not be achieved effec-
E-mail address: r.lo@adfa.edu.au (S.R. Lo). tively. The mobilization of additional confining stress on the stones,

0266-1144/$ – see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.geotexmem.2009.09.015
S.R. Lo et al. / Geotextiles and Geomembranes 28 (2010) 292–302 293

and thus the generation of higher bearing capacity, can still be columns to enhance the ability of a soft clay layer in supporting a fill
realized during or after placement of fill because axial deformation embankment involved complicated interaction mechanisms. The
of a stone column is always accompanied by lateral expansion embankment weight is shared between stone columns and soft
against the surrounding soil. However, for very soft soil, the gener- clay in accordance with their relative stiffness values, and both
ation of this additional confining stress requires a high radial being affected by the behaviour of the surrounding soft clay.
expansion of the stones, which in turn means high axial deformation This paper presents a numerical study of a stone column rein-
of the stone column. Alternatively, the generation of confining stress forced with a geosynthetic encasement. In this study, stone
maybe enhanced by encasing/wrapping the stone with geosynthetic columns were use to enhance the performance of a soft clay
reinforcement as illustrated in Fig. 1. A construction system ultilis- stratum in carrying a road embankment in the form of 4 m of fill.
ing high stiffness and creep resistant geotextile for encasing stone The time-dependent interaction between a stone column and the
columns was presented in Alexiew et al. (2005). surrounding soft clay will be captured by using a coupled analysis.
The benefits of using geosynthetic to encase or wrap geo- This study does not assume any particular form of geosynthetic.
materials were well illustrated in studies on soil bag pile by
Lohani et al. (2006). Murugesan and Rajagopal (2006) presented 2. Stone columns as reinforcing elements
an axi-symmetric unit cell analysis to demonstrate the effective-
ness of geosynthetic encasement in improving the performance of For the situation where stone columns were use to improve the
a stone column functioning as a single pile. However the ability of a soft clay stratum in supporting a fill embankment, the
surrounding clay is characterized by a non-linear elastic model in role of stone columns is time dependent as explained below.
terms of total stress. This makes it difficult to relate the computed Immediately after the imposition of fill loading, most of the
results to the actual time-dependent performance, especially imposed total stress is taken by pore water pressure in the clay soil
when the stone columns perform as reinforcements. The rein- and thus the stone columns may only play a small role in resisting
forcing mechanism of stone columns in soft clay is addressed in the fill loading. It is only with dissipation of excess pore water
a later section of this paper. Furthermore, the geosynthetic pressure with time that the clay will settle and the weight of the fill
encasement appears to be modeled as isotropic and this will will ‘‘arch over’’ to the stone columns. The extent this can occur
introduce an extra vertical reinforcing tube of unknown effects. depends on both the strength and stiffness of the stone columns.
Murugesan and Rajagopal (2007) presented 1g-model test results During this process, the stone columns will deform both axially and
to demonstrate the effectiveness of the geosynthetic encasement radially, the latter leading to an increase in confining stress from
in enhancing the axial load bearing capacity of a short stone the surrounding soil due to a cavity expansion mechanism. Some of
column under short duration loading. However, the inherent scale the fill loading will still be transferred to the clay as effective stress
effects and the rather atypical clay parameters reported (for and this also leads to increase in confining stress. Therefore, the
example the clay had a vane shear strength of 2.5 kPa at 47% mechanism involves the interaction of the stone columns with
moisture content) make it difficult to translate the findings to dissipation of excess pore water pressure, i.e. consolidation, of the
a field problem. An alternative approach of reinforcement a stone surrounding soft clay. The need of conducting an analysis that
column by the inclusion of horizontal geosynthetic sheets at close couples the mechanical behaviour (in terms of effective stress) and
intervals was studied by Wu and Hong (2008). Wu and Hong flow of pore water (based on Darcy Law) is evident.
(2009) presented laboratory test results on geosynthetic-encased
granular columns in triaxial testing. Their test results showed that
at low column strain of about 1% in the axial direction, the 3. Geosynthetic encasement
improvement due to geosynthetic encasement was several times
less than that at maximum test load. As shown in Fig. 2, the radial stress acting on the stone column,
The studies discussed above were focused on the performance sr,s, is induced by the radial stress of the surrounding clay, sr,c, and
of a stone column as a single pile, i.e. with external load applied the hoop tension, T, in the geosynthetic encasement. Thus:
only to the column top but not to the surrounding soil. As pointed .
out by Alexiew et al. (2005), the use of a large number of stone sr;s ¼ sr;c þ T R (1)

stones
surrounding soil
σr,c
geosynthetic encasement

σr,s
Stone
weak deposits
Column

T T

2R
Fig. 1. Geosynthetic encased stone column (Excerpted from Lo et al. 2007). Fig. 2. Hoop tension and radial stress in stone column (Excerpted from Lo et al. 2007).
294 S.R. Lo et al. / Geotextiles and Geomembranes 28 (2010) 292–302

where R is the radius of the stone column. The second term can be Top of sand blanket
viewed as the additional radial stress due to the geosynthetic natural ground level
encasement. Unless stated otherwise to the contrary, all stresses are
effective stresses and this is consistent with a coupled analysis.
T(i)
Both T and sr,c can be decomposed into two parts; the initial value zext
(i.e. after stone column installation) and the increase due to Text limit
placement of fill and time-dependent deformation. Therefore, Eq. T(i) profile
(1) can be re-written as:

Depth
. .
sr;s ¼ sr;c ðiÞ þ Dsr;c þ TðiÞ R þ DT R
. (2)
¼ sr;c ðiÞ þ Dsr;c þ srp þ DT R

Where (i) denotes the initial (as-installed) state, ‘‘D’’ denotes


increase due to loading and subsequent consolidation, and Tpre
srp ¼ T(i)/R. Note that the as-installed hoop tension, T(i), is generally
non-zero and its magnitude depends on the installation method. It
is essential to have T(i) included in the modelling. One can also
express Eq. (2) as:
Fig. 3. Initial hoop tension.
sr;s ¼ sr;c ðiÞ þ sr;p þ Dsr;s (3)
.
where Dsr;s ¼ Dsr;c þ DT R earth pressure coefficient of the soft clay. Noting rK0/Kp < < 1, Eq.
To enable the stones to develop adequate strength and stiffness, (5) can be approximated as:
sr,s has to be of adequate magnitude. If an adequately high sr,c(i) can
 
be generated by the installation process, then both T(i) and DT are 1 Tpre
not needed. Indeed the value of Dsr,c will also be low as the axial zext ¼  g t
b b (5a)
g0s RKp
strain and thus the radial expansion of the stone column will be
small. However, one can compensate for a low sr,c(i) value, say due
to the surrounding clay being very soft, by having adequately high 4. Finite element modelling
values of T(i) and DT.
4.1. Unit cell idealization
The value of T(i) is governed by two considerations: i) the
pre-straining action due to installation and ii) triaxial extension
This paper examines the condition where the fill area is large
failure of the stones at the top zones. The compaction of the
relative to the thickness of the soft clay and that a large number of
stones and consequent radial expansion of the stone column
stone columns are installed. The problem can be idealized by a unit
with the prefabricated geosynthetic encasement to its final
cell as illustrated in Fig. 4. This idealization is a reasonable
diameter will pre-strain the geosynthetic encasement. This pre-
approximation in areas away from the edges of the embankment. It
straining will induce a preloading, Tpre, in the geosynthetic
was assumed that stone columns were installed in a regular square
encasement. As the pre-straining action is related to the radial
or triangular pattern with a centre-to-centre spacing denoted as
expansion of the stone column as a result of compacting the
‘‘s’’. Based on equivalence in externally loaded area per stone
stones, Tpre, may be approximated to be constant, i.e. as an input
column, the radius of the unit cell may be approximated as 0.55s for
parameter to the analysis. However, near the top of the stone
a square pattern and 0.525s for a triangular pattern. In our analysis,
column, the magnitude of T(i) is limited by sr,s(ext), the limiting
the top boundary of the clay layer was modeled as free draining,
radial stress at triaxial extension failure of the stones. sr,s(ext) is
given by: whereas the outer and bottom boundaries of the unit cell were
modeled as impermeable. The assumed dimensions for the analysis
sr;s ðextÞ ¼ Kp sz;s ðiÞ (4) are listed in Table 1.
Both the stones-geosynthetic and geosynthetic-clay interfaces
where sz,s(i) is the in-situ vertical stress, Kp ¼ (1 þ sinf)/(1  sinf), were assumed to be full strength. This is because i) the installation
and f is the secant friction angle of the stones. Eq. (4) leads to: of a stone column will automatically leads to an undulating inter-
face, and ii) these interfaces are internal drainage nodes. Therefore
Text ðiÞ ¼ Rsr;s ðextÞ ¼ RKp sz;s ðiÞ (4a)
preferential slippage cannot occur and there is no need to introduce
where Text(i) denotes hoop tension limited by triaxial extension any special internal interface element.
failure of the stones. Therefore, T(i) is the lesser of Tpre(i) and Text(i).
Assuming water table is at natural ground level (NGL), the resultant 4.2. Construction sequence
T(i) profile is given by the bilinear solid line of Fig. 3. At the kink of
the bilinear relationship, Tpre ¼ Text. Substitution this condition into In addition to using a coupled analysis, the following
Eq. (4)a) gives zext, depth of the kink as measured from natural construction sequence was modeled.
ground level as:
1. Initialize in-situ stress of soft clay deposit under green field
Tpre =RKp  gb tb condition.
zext ¼   (5)
g0s 1  rK0 =Kp 2. Place sand blanket in 4 days.
3. Turn appropriate regions of soft clay into stone column
where g’s ¼ effective unit weight of stones, gb ¼ unit weight of sand elements and activate the geosynthetic elements.
blanket, r ¼ ratio of effective unit weight of clay to that of 4. Build the embankment in a layer-by-layer manner at a rate of
stones ¼ 0.65, tb ¼ thickness of sand blanket ¼ 1 m, and K0 ¼ at-rest 0.25 m/day.
S.R. Lo et al. / Geotextiles and Geomembranes 28 (2010) 292–302 295

K0 ¼ 1  sin4 ¼ 0.535 for normally consolidated clay. As the analysis


Embankment models the coupled process of time-dependent dissipation of excess
pore water pressure, permeability parameters are also needed. The
horizontal permeability of the soft clay was assumed as
2.3  1010 m/s, and a horizontal to vertical permeability ratio of two
was assigned. A range of typical undrained shear strength profiles
were also examined. From these undrained strength profiles, it was
inferred, following Potts and Ganendra (1991), that the top 3 m was
over-consolidated even though the soil was soft. The over-consoli-
unit cell dation was characterized by pc, the effective mean stress at the apex
of the Cam-Clay ellipse. The value of pc was assumed to be 70 kPa at
natural ground level and reducing to 40 kPa at 3 m depth.

Axes of
Symmetry 5.2. Fill

The fill was modeled as a Mohr–Coulomb elastic-plastic mate-


rial with a non-associative flow rule. The parameters adopted for
the analysis are given in Table 1.
Fill
natural ground level 5.3. Geosynthetic
sand blanket

geosynthetic The geosynthetic encasement was modeled as a cross-aniso-


encasement (if used) tropic elastic element. The horizontal stiffness was taken to be
2000 kN/m for the reference analysis. The axial stiffness was taken
Stone Column

to be 3% of the horizontal stiffness in line with Lo et al. (2007) so that


soft clay it will not ‘‘numerically’’ act as a vertical cylindrical reinforcing tube.
Poisson ratios were taken to be zero to eliminate cross-coupling
between axial and radial stress. The use of an elastic element
inherently implies that tensile rupture of the geosynthetic will not
occur, and therefore the locked-in tension, T(i) does not influence
the modelling of the geosynthetic. However, T(i) will induce a higher
initial radial stress in the stones as represented by the term srp of Eq.
(2). This will lead to a higher radial stress in the stones in all stages of
the analysis, the effects of which can be modeled by a stone column
element as explained in the next section.
Fig. 4. Unit cell (adapted from Lo et al. 2007).

5.4. Stone column

5. Time-stepping for 10 years so as to track the dissipation of


The stone column was modeled as a free draining material. A
excess pore water pressure and thus development of settle-
stone column element was incorporated into the finite element
ment after the completion of embankment.
code, AFENA (Carter and Balaam, 1995). This element is in fact
a modified Mohr–Coulomb elastic-plastic element with a non-
Note that the sand blanket needs to be placed first to form
linear elastic part similar to that in the Duncan-Chang model
a platform to support the equipment for installing stone columns.
(Duncan and Chang, 1970), but with the unloading and loading
Furthermore, step-3 automatically simulates the effects of T(i) and
stiffness selected based on stress increment direction relative to the
associated srp due to installation.
isotopic stress axes as explained below.
The stress path followed by an element along the centre of the
5. Material models stone column can be schematically illustrated in Fig. 5. This illus-
tration is also considered to be an approximation of other stone
5.1. Soft clay column elements. As a result of T(i) and the corresponding locked-
in confining stress in the stones, sr,s(i) > sz,s(i). Therefore, in the
The soft clay was modeled by the modified Cam-Clay model. The early phase of embankment loading as denoted by ‘‘IO’’ of Fig. 5, any
assumed Cam-Clay soil parameters are given in Table 1. In-situ stress increase in sz,s due to embankment loading will lead to a reduction
was assigned based on an effective unit weight of 6 kN/m3 and in stress ratio and the stress state moving towards the isotropic
stress axes. Thus the element will behave in an ‘‘unloading mode’’,
and the Young’s modulus is given by the initial Young’s modulus of
Table 1 Duncan-Chang given below:
Dimensions of unit cell.
 
sr;s n
Items Dimension(m) E0 ¼ K pa pa
  (6)
Embankment height 4.0 srp þsr;c ðiÞþDsr;s n
Sand blanket thickness 1.0 ¼ K pa pa
Diameter of stone column 0.6
Unit cell radius 2.0 Once the stress state crossed and traced away from the isotropic
Depth of ground water table 0.0 axes, the Young’s modulus was calculated using the Duncan-Chang
Thickness of soft clay 10.0
Chang Eq as expressed below.
296 S.R. Lo et al. / Geotextiles and Geomembranes 28 (2010) 292–302

B In the simplified analysis presented in Lo et al. (2007), an


average Young’s modulus,E, calculated using Eqs. (6) and (8) as well
es
σz,s ax as neglecting Dsr,s were used, thus giving the following equation.
c
pi
A tro  2
is
o 1 þ 1  rf s ðiÞn
r;s
E ¼ pa
2 pa
 n
sr;s ðiÞ
O z0:5 pa (9)
pa
The results of the preliminary analysis showed that significant
number of stone column elements were close to Mohr–Coulomb
failure even at early stage of time-stepping. This may present
σz,s (i) I numerical problem (in the form of very low Bulk modulus) as
discussed in Lo (2001). To suppress such a numerical problem, the
Poisson ratio was taken to be a function of S following Lo (2001) and
σr,s expressed as Eq. (10) below:
σr,s(i) = σr,c(i)+σrp
pffiffiffi
m ¼ m0 þ ð0:496  m0 Þ S (10)
Fig. 5. Effective stress path of a typical stone column element.
Thus the Poisson’s ratio of Table 2 is in fact m0.
 2
6. Results of finite element analysis
E ¼ E0 1  rf S (7)

where E0 (initial Young’s modulus) is given by Eq. (8) and 6.1. Results of reference analysis

ðs1  s3 Þð1  sinfÞ An analysis conducted based on the dimensions listed in Table 1,
S ¼ (8) material models presented in Section-5, material parameters listed
2s3 sinf þ 2c:cosf
in Table 2, and Tpre of 100 kN/m is referred to as the reference
It is recognized that as the stress path cross state ‘‘O’’ but before analysis. Substituting this Tpre value into Eq. (5)) gives zext ¼ 3.95 m.
reaching state ‘‘A’’ defined by a stress ratio equal to that of the as-
installed state ‘‘I’’, the stiffness is actually transitioning from Eq. (6) 6.1.1. Settlement
to that of Eq. (8). However, the criterion of moving away from the Plots showing the development of settlement at natural ground
isotropic stress state is used to trigger the use of the Duncan-Chang level (NGL) with time are presented in Fig. 6a. The settlement-time
equation for loading for the sake of simplicity. relationship computed by the simplified analysis presented in Lo et al.
The parameters assigned in the analysis are given in Table 2, (2007) was also plotted in this figure. It is evident that the settlement
noting that the Duncan-Chang parameters were inferred conser- predicted by the reference analysis is higher than that of the simplified
vatively from triaxial tests results of a well-graded sandy gravel analysis. However, the overall trend in the development of settlement
compacted to maximum dry density as determined by standard is essentially the same for the two analyses. The incorporation of
proctor test. These triaxial test results gave a curved failure surface. geosynthetic encasement considerably reduced settlement both at
The approximation of this curved surface by the linear Mohr– column location and at edge of the unit cell. Without geosynthetic
Coulomb failure function led to a small non-zero cohesion intercept encasement, the settlement at the top of stone column at 10 yr was
of 15 kPa. approximately 0.87 m. The application of geosynthetic encasement
with Tpre ¼ 100 kN/m reduced the settlement at top of stone column
Table 2 to 0.24 m. The settlement at the edge of the unit cell attained a higher,
Soil parameters. but still relatively small, value of 0.32 m at 10 yr. Furthermore the
Material Parameter Value stone column essentially ceased settlement after 2000 days.
Fill f 30

The settlement profiles at natural ground level and at top of the
c 20 kPa fill were plotted in Fig. 6b. Note that settlement at top of fill was

j 5 relative to end of construction. The settlement profile at foundation
Young’s modulus 30  103 kPa
level manifested a ‘‘bump’’ near the perimeter of the stone column.
Unit weight 20 kN/m3
However, the settlement profile at top of fill was ‘‘bump-free’’. This
Soft Clay M 1.1 settlement profile ‘‘intersected’’ that of natural ground level; and

f 27.7
therefore, at edge of the unit cell, the settlement was ‘‘less than’’
l 0.65
k/l 0.1 that at natural ground level. This apparent inconsistency is due to
ecs 4.1 the settlement at top of the fill being referenced to end of
kr 2.3  1010 m/s construction. At centre of the unit cell, the settlement at top of fill
kr/kz 2.0 was significantly higher than that at natural ground level. This is

Stone f 45 because the fill above the stone column is subject to significantly
c 15 kPa higher stress because of the stone column force.
K 2000
n 0.65
rf 0.7 6.1.2. Force in stone column
The distributions of computed column force with depth are
4 ¼ friction angle; c ¼ cohesion; j ¼ dilatancy angle; kr ¼ permeability in radial
(horizontal direction); kz ¼ permeability in vertical direction; (M, l, k, ecs) are
presented in Fig. 7 at two time steps: end of construction and
parameters for modified Cam-Clay model; (K, n, rf) are parameters for Duncan- 10 yr. The corresponding column force distributions given by the
Chang model. simplified analysis (presented in Lo et al., 2007) were also plotted
S.R. Lo et al. / Geotextiles and Geomembranes 28 (2010) 292–302 297

a 0
NGL
0

-0.2
-2
Settlement at NGL (m)

-0.4

Depth (m)
-4

-0.6
-6

2m radius unit cell


-0.8 edge of unit cell
2m radius unit cell
center of unit cell
-8 End of construction
10 yr
Simplified analysis: centre of unit cell
Simplified analysis: End of construction
Simplified analuysis: edge of unit cell
Simplified analysis: 10 yr
No encasement: centre of unit cell
-1
-10
0 1000 2000 3000 4000
0 100 200 300 400 500 600 700 800
Time (day)
Force in stone column (KN)

Centre of unit cell Edge of unit cell Fig. 7. Distribution of column force with depth: Reference analysis.
b -0.2
6.1.3. Coupled behaviour
The coupling between column force and consolidation behav-
iour is highlighted by examining the evolution of [sz,c  sz,c(i)]/q
and uex/q with time in Fig. 9, where [sz,c  sz,c(i)] denotes increase
-0.3
in vertical stress in the clay due to embankment loading, (i) denotes
Settlement (m)

start of embankment construction, uex ¼ excess pore water pres-


sure and q ¼ average fill loading ¼ 80 kPa. Normalization with
respect to q allows easier comparison. The three sets of plots of
Fig. 9 corresponds to three radial locations, refer to as near-column,
-0.4
mid-distance and near-edge, all at mid-depth of the clay. The stress
values for these three sets of plots are average values of three
2.0m radius unit cell corresponding clay elements: next to stone column, mid-distance
Top of fill
NGL between stone column and edge of unit cell, and next to edge of
unit cell. All plots commenced from 16 days when the embankment
-0.5
was at full height.
0 0.5 1 1.5 2
The dissipation of excess pore water pressure proceeded with
Distance from center (m)
increase in vertical effective stress, as expected. The maximum
Fig. 6. Settlement response: Reference analysis. a) Comparison of settlement-time value of uex/q was only 0.82 because of some dissipation occurring
plots; b) Settlement profile at 10 yr. during embankment construction. Both the dissipation of excess
pore water pressure and increase in effective stress at the near-
column location were significantly faster than the other two
to allow a comparison of the two analyses. Although the stone
column force of the simplified analysis is ‘‘noisy’’, the computed
results of the simplified analysis are similar to those given by the
800
current analysis. The column force distribution at end of
construction was different from that at 10 yr. This is because of 700
the coupling between the reinforcing role of a stone column and
consolidation of the soft clay. At an early stage when the extent of 600
consolidation of the soft clay is small, the pore water pressure in
the soft clay provides significant support to the fill loading. It is 500
Force (KN)

only with development of consolidation that a greater portion of


the fill loading is transferred to the stone column. This justified 400 2.0m radius unit cell
1.5-2.0m depth from NGL
the need for a coupled analysis. The column force at 10 yr 4.5-5.0m depth from NGL
increased with depth until a maximum value of w750 kN was 300 8.0-8.5m depth from NGL

achieved at about mid-depth. This is due to negative drag-down


200
from the surrounding clay that tends to settle more than that of
the stone column. The evolution of column force with time was
100
examined in further detail in Fig. 8 at three depths referred to as
near-top, mid-depth and near-toe. At all three depths, the column
0
force increased with time. At mid-depth and near-toe locations,
0 1000 2000 3000 4000
the force in the stone column attained an asymptotic value at Time (day)
about 1500 day. At the near-top location, the force approached an
asymptotic value at w2000 day. Fig. 8. Evolution of column force with time: Reference analysis.
298 S.R. Lo et al. / Geotextiles and Geomembranes 28 (2010) 292–302

0.6 50
1.5-2.0m depth from NGL
efective vertical stress
4.5-5.0m depth from NGL
near column
8.0-8.5m depth from NGL
[σz,c-σz,c(i)] / q

mid-distance
0.4 40
near-edge

30

ΔT (kN)
0.2

20
0

10
0.8
excess pore water pressure
near-column
mid-distance
0
near-edge
0.6 0 1000 2000 3000 4000
Uex / q

Time (days)

0.4 Fig. 10. Increase in geosynthetic tension with time: Reference analysis.

6.2. Influence of lower stone stiffness


0.2

The influence of lower stone stiffness was examined by


0 repeating the analysis with the Duncan-Chang parameter K
reduced to 1000, while keeping all other parameters identical to
0 1000 2000 3000 4000
Time (days) those of reference analysis. This stiffness value is considered to be
relatively low for compacted stone.
Fig. 9. Consolidation response in clay: Reference analysis. The computed evolution of settlement with time was presented
in Fig. 11a, whereas the settlement profiles at natural ground level
and top of fill were plotted in Fig. 11b. The settlement-time rela-
locations. This is due to the drainage provided by the stones. For all tionship and settlement profiles manifested trends similar to those
three locations, the increase in effective stress after 300 days was
slight. However, significant dissipation of uex continued to occur at
the mid-distance and near-edge locations. This is because dissipa- a 0
Settlementat NGL (m)

tion of uex can occur as a result of load being re-distributed to the


-0.1
stone column. This explanation is consistent with the evolution of
column force shown in Fig. 9. When uex/q / 0, say after 3000 days, -0.2
the [sz,c  sz,c(i)]/q value at near-column location was still higher
-0.3
than those of other two locations, and that the [sz,c  z,c(i)]/q values
2.0m radius unit cell: K=1000
at all three locations were significantly and considerably less than -0.4 edge of the unit cell
unity. These two features imply that the consolidation process also center of the unit cell
-0.5
leads to significant embankment loading being re-distributed to
0 1000 2000 3000 4000
the stone column.
Time (day)
The coupling between consolidation and geosynthetic tension
was manifested in the DT versus time relationship, where
DT ¼ T  T(i) ¼ increase in geosynthetic tension after embankment b -0.2
construction. Fig. 10 presents three DT-time plots at three depths:
near-top, at mid-depth, and near-toe. The depths of these three
locations are 1.5–2 m, 4.5–5 m and 8–8.5 m and reinforcement
tension is calculated as the average value of an element. All three
plots showed increase in DT with time (and hence consolidation). -0.3
Settlement (m)

The increase in stone column force with consolidation leads to axial


and radial deformation of the stone column, which in turn induces
DT. This DT provides further confinement to the stones and enables
the stone column to continue attracting more load as consolidation
proceeds. The DT values for mid-depth and near-toe locations were -0.4
close and small relative to T(i). However, the DT value at near-top
2.m radius unit cell: K=1000
location was significantly higher. This is because at the mid-depth NGL
and near-toe locations, T(i) ¼ Tpre ¼ 100 kN/m. Such a high geo- Top of fill

synthetic tension provides very high confinement and thus stiff-


ness to the stones. At the near-top location, z < zext and therefore -0.5

T(i) < Tpre as explained in Eq. (5) and Fig. 3. This smaller T(i) value 0 0.5 1 1.5 2
Distance from center (m)
means the development of higher axial strain in the stones with
loading, which in turn generates significantly higher DT values to Fig. 11. Settlement response: Reduced stone stiffness, K ¼ 1000. a) Settlement-time
enable the stone column to attract further loading. plots; b) Settlement profile.
S.R. Lo et al. / Geotextiles and Geomembranes 28 (2010) 292–302 299

of reference analysis. The settlement was higher than that of the 800
reference analysis and the settlement profile at top of fill was
‘‘bump-free’’. However, the increase in settlement was only w20% 700
in spite of a 50% reduction in stiffness. This somewhat unexpected
small increase in settlement can be explained by examining the 600
computed distributions of column force with depth as presented in
500

Force (KN)
Fig. 12. The distributions for both time step: end of construction and
10-year were only marginally less than those of the reference
400
analysis (Fig. 8). Thus at a high value of Tpre (which is the case for
both analysis), the performance of the unit cell was not sensitive 2.0m radius unit cell: K=1000
300 1.5-2.0m depth from NGL
the stiffness of the compacted stones. The underlying mechanism 4.5-5.0m depth from NGL
for such a ‘‘forgiving’’ and desirable attribute will be explained in 200
8.0-8.5m depth from NGL

a subsequent paragraph that examines the coupling between


development of geosynthetic tension and the consolidation 100
process. The time-dependent nature of the stone column force was
illustrated in Fig. 13 showing the evolution of column force with 0
time at three depths: near-top, mid-depth and near-toe. It is 0 1000 2000 3000 4000
evident that the contribution of stone column to resisting the fill Time (day)
loading is time dependent.
Fig. 13. Evolution of column force with time: Reduced stone stiffnes,s K ¼ 1000.

6.2.3. Coupled behaviour


stiffness parameter of the stones was reduced by 50% in this anal-
The computed [sr,c  sr,c(i)]/q - time and uex/q - time relation-
ysis. This is consistent with the evolution of settlement and stone
ships for three radial locations at mid-depth of the soft clay were
column force with time as discussed in an earlier section.
plotted in Fig. 14 following the same rationale and format as those
The coupling between consolidation and geosynthetic tension
for the reference analysis. These three plots displayed characteris-
was manifested in the DT versus time relationship presented in
tics similar to those for the reference analysis, viz:
Fig. 15 following the same format as that for the reference analysis.
All three plots showed an increase in DT with time (and hence
- Increases in sz,c after 300 days were slight at all three locations,
consolidation). The trends displayed were similar to those of the
but significant dissipation of uex continued to occur at the mid-
reference analysis, thus indicating similar coupling between devel-
distance and near-edge locations. This was because dissipation
opment of geosynthetic tension and the consolidation process.
of uex also occurred as a result of the embankment load being
re-distributed to the stone column.
- The soil element next to the stone column attracted a higher 6.3. Effects of locked-in stress in geosynthetic
effective stress compared with the other two locations as
a result of interaction with the stone column (in addition to The influence of a lower locked-in force in the stone column was
that due to drainage). studied by repeating the analysis with Tpre reduced to 50 kN/m (i.e.
- Even when uex / 0, say after 3000 days, the value of
[sr,c  sr,c(i)]/q for all three locations were significantly and
0.6
considerably less than unity because the load carried by the effective vertical stress
stone column increased with the consolidation process. near-column
mid-distance
[σz,c-σz,c(i)] / q

0.4 near-edge
At 10 yr, the value of [sr,c  sr,c(i)]/q near the stone column was
w20% higher than that of the reference analysis even though the
0.2

NGL
0
0

-2
0.8
excess pore water pressure
near-column
mid-distance
-4 near-edge
Depth (m)

0.6
Uex / q

-6
0.4

2.0m radius unit cell: K=1000


-8 End of construction 0.2
10 yr

-10 0
0 100 200 300 400 500 600 700 800 0 1000 2000 3000 4000
Force on stone column (KN) Time (days)

Fig. 12. Distribution of column force with depth: Reduced stone stiffness, K ¼ 1000. Fig. 14. Consolidation response in clay: Reduced stone stiffness K ¼ 1000.
300 S.R. Lo et al. / Geotextiles and Geomembranes 28 (2010) 292–302

50 location (i.e. at stone column centre), the settlement was 0.40 m


1.5-2.0m depth from NGL
which is w67% higher than that of reference analysis. At edge
4.5-5.0m depth from NGL location, the computed settlement was 0.44 m, which was w55%
8.0-8.5m depth from NGL
40 higher than that of reference analysis. However, these settlement
values are still significantly lower than those without geosynthetic
encasement. Furthermore, the settlement profile along top of fill
30
was still ‘‘bump-free’’. Thus road constructed on such an embank-
ΔT (kN)

ment will still has a high quality riding surface.


20 The distribution of column force with depth at end of construc-
tion and 10-year were plotted in Fig. 17. The overall shapes of the
distributions were similar to those of reference analysis. The
10 distribution at 10 yr was considerably different from that of end of
construction, thus re-confirming that contribution of column force
to resisting fill load was time dependent. For the distribution at 10 yr,
0
the column force in general was distinctly smaller, but only by
0 1000 2000 3000 4000
a small extent. The force at 1 m depth was 555 kN which is 86% of
Time (days)
that given by the reference analysis (with a higher Tpre of 100 kN/m).
Fig. 15. Increase in geosynthetic tension with time: Reduced stone stiffness, K ¼ 1000. The maximum column force was 675 kN which is 90% of that given
by the reference analysis. The underlying mechanism for such an
‘‘apparently unexpected’’ behaviour will be explained in a subse-
halved), while keeping all other parameters identical to those of quent paragraph that examines the coupling between development
reference analysis. Substituting this Tpre value into Eq. (5) gives of geosynthetic tension and the consolidation process. The time-
zext ¼ 0.92 m. dependent nature of the stone column force was illustrated in Fig.18.
The computed evolution of settlement with time was presented It is evident that the contribution of stone column to resisting the fill
in Fig. 16a, whereas the settlement profiles at natural ground level loading is time dependent and following a trend similar to that of the
and foundation level were plotted in Fig. 16b. Although the devel- reference analysis.
opment of settlement with time and the overall shapes of the
settlement profile profiles followed patterns similar to those of the 6.3.3. Coupled behaviour
reference case, this analysis gave settlement values considerably The computed [sr,c  sr,c(i)]/q – time and uex/q – time relation-
higher compared to those of reference analysis. At centre-line ships for three radial locations at mid-depth of the soft clay were
plotted in Fig. 19 following the same rationale and format as those
for the reference analysis. These three plots displayed characteris-
tics similar to those for the reference analysis, viz:
a 0
- Increases in sz,c after 300 days were slight at all three locations,
Settlement at NGL (m)

2m radius unit cell: Tpre=50 kN/m


-0.1 edge of unit cell but significant dissipation of uex continued to occur at the mid-
center of unit cell
-0.2 distance and near-edge locations. This was because dissipation
of uex also occurred as a result of the embankment load being
-0.3 re-distributed to the stone column.
-0.4 - The soil element next to the stone column attracted a higher
effective stress compared with the other two locations as
-0.5
0 1000 2000 3000 4000
Time(days)
NGL

b -0.2 0

-2

-0.3
Settlement (m)

Depth (m)

2m radius unit cell: Tpre=50 kN/m -4


NGL
top of fill

-6
-0.4
2m radius unit cell: Tpre=50kN/m
-8 end of construction
10 yr

-0.5 -10
0 0.5 1 1.5 2 0 100 200 300 400 500 600 700 800
Distance fromcenter (m) Force on stone column (KN)

Fig. 16. Settlement response: Reduced locked-in stress, Tpre ¼ 50 kN/m. a) Settlement- Fig. 17. Distribution of column force with depth: Reduced locked-in stress,
time plots; b) Settlement profile at 10 yr. Tpre ¼ 50 kN/m.
S.R. Lo et al. / Geotextiles and Geomembranes 28 (2010) 292–302 301

800 50
8.0-8.5m depth from NGL
4.5-5.0m depth from NGL
700 1.5-2.0m depth from NGL

40
600

500 30

ΔT (kN)
Force (KN)

400
20
2m radius unit cell: Tpre = 50 kN/m
300 1.5-2.0m depth from NGL
4.5-5.0m depth from NGL
8.0-8.5m depth from NGL 10
200

100
0
0 1000 2000 3000 4000
0
Time (days)
0 1000 2000 3000 4000
Time (days) Fig. 20. Increase in geosynthetic tension with time: Reduced locked-in stress,
Tpre ¼ 50 kN/m.
Fig. 18. Evolution of column force with time: Reduced locked-in stress, Tpre ¼ 50 kN/m.
The coupling between consolidation and geosynthetic tension
was manifested in the DT versus time relationship presented in
a result of interaction with the stone column (in addition to Fig. 20 following the same format as that for the reference
that due to drainage). analysis. Although all three plots showed an increase in DT with
- Even when uex/q / 0, the value of [sr,c  sr,c(i)]/q was signifi- time (and hence consolidation), the magnitude of DT for this case
cantly and considerably less than unity at all three locations was considerably higher for all three locations. Thus the system
because the load carried by the stone column increased with was trying to compensate for a lower locked-in geosynthetic
the consolidation process. tension by generating higher DT during the consolidation
process. The lower T(i) value led to higher axial and radial
At 10 yr, the value of [sr,c  sr,c(i)]/q for this case was signifi- deformation of the column, which in turn induced a higher DT.
cantly higher than that of the reference analysis, by about w40% This additional confinement provided by a higher DT enabled the
next to the stone column location and w23% for the other two stone column to continue attracting embankment load. It is
locations,. This is because of the higher settlement and smaller load pertinent to note that the DT-time plots for near-top and mid-
carried by the stone column as a result of a 50% reduction in Tpre depth locations were close, despite the confining stress from the
adopted in this case. surrounding clay is higher at mid-depth. This is because of
a higher column force at mid-depth and T(i) ¼ Tpre at both
locations. The lower DT value at near-toe is due to the combi-
nation of a smaller column force and a higher confining stress
0.6
from the surrounding clay.

7. Discussion on locked-in tension


[σz,c-σz,c(i)] / q

0.4

The above analysis indicates that the locked-in geosynthetic


tension, T(i), has a significant influence on the performance of
0.2 effective vertical stress a stone column. The T(i) profile is characterized by Tpre which is as
vstress-near col
vstress-mid dist an input constant to the analysis as expressed in Eq. (5a) and
vstress-edge illustrated in Fig. 3. It is pertinent to examine the factors that
0
influence Tpre. It is envisaged that the geosynthetic encasement will
be installed as a ‘‘prefabricated cage’’. The compaction of the stones
0.8
and consequent radial expansion of the stone column with the
prefabricated geosynthetic cage to its final diameter will pre-strain
the geosynthetic encasement, thus leading to the preload, Tpre.
0.6 excess pore water pressure Therefore, Tpre ¼ Jeg(i), where J is the long term horizontal stiffness
Uex / q

pwp-near column
of the geosynthetic and eg(i) is the pre-strain in the geosynthetic
pwp-mid distance
pwp-edge induced by installation. The later can be approximated by dRi/R,
0.4 where R is the radius of the stone column and dRi is the radial
expansion during installation. Thus, Tpre can be assessed by the
following equation
0.2
 
dRi
Tpre ¼ J (11)
R
0
Eq. (11) above relies on the assumption of eg(i) ¼ dRi/R which
0 1000 2000 3000 4000
Time (days)
implies that the stones must be jacketed by the geosynthetic
encasement, i.e. no escaping of the stones through the geo-
Fig. 19. Consolidation response in clay: Reduced locked-in stress, Tpre ¼ 50 kN/m. synthetic during installation. This requires proper design of the
302 S.R. Lo et al. / Geotextiles and Geomembranes 28 (2010) 292–302

prefabricated geosynthetic cages. It is also evident from Eq. (11) Research Scholarship, University of New South Wales at ADFA.
that geosynthetics with an adequately high J-value has to be However, the opinions expressed in this paper are solely those of
used. Determining a ‘‘design value’’ for of dRi is less straight the authors.
forward despite the prefabricate size of the geosynthetic cage
and the final size of the stone column can be specified. This is
References
because the control of the latter, for a construction project, has
to be achieved indirectly via controlling the installation method. Alexiew, D., Brokemper, D., Lothspeich, S., 2005. Geotextile Encased Columns: Load
The importance of controlling the installation method via field Capacity, Geotextile Selection and Pre-Design Graphs. Proceedings of the Geo-
trials for certain geotechnical systems were discussed in Lo and Frontiers Conference, Austin, Texas, January. Geotechnical Special Publication
No. 131. ASCE, pp. 497–510.
Li (2002). Barksdale, R.D., Bachus, R.C., 1983. Design and construction of stone columns. FHWA
Report No. RD-83/026, 194p.
8. Conclusion Carter, J.P., Balaam, N.P., 1995. AFENA – a General Finite Element Algorithm –
Users Manual. School of Civil and Mining Engg. University of Sydney, NSW,
Australia.
A numerical study was undertaken to examine the reinforcing Duncan, J.M., Chang, C., 1970. Nonlinear analysis of stress and strain in soils.
role of stone columns in soft clay. The analysis models the time- Journal of Soil Mechanics and Foundations Division 96 (SM5), 1629–1653.
ASCE.
dependent interaction of the encased stone column and
Lo, S.R., Mak, J., Zhang, R., 2007. Geosynthetic Encased Stone Columns in Soft Clay.
surrounding soft clay by a fully coupled analysis. Locked-in stress in Proc of International Symposium on Earth Reinforcement, Kyushu. Taylor and
the geosynthetic was modeled in the analysis. The following may Francis, pp. 751–756.
Lo, S.R., Li, K.S., 2002. Influence of permanent liner on skin friction of large diameter
be concluded from the findings of this numerical study.
bored piles in Hong Kong granitic saprolites. Canadian Geotechnical Journal 40
(4), 793–805.
1) The overall trends established from the simplified analysis of Lo Lo, S.R., 2001. The Application of Numerical Analysis to Innovative Design of Geo-
et al. (2007) are confirmed. synthetic Reinforced Soil Structures. 10-th Intl Conference on Computer
Methods and Advances in Geomechanics, Jan 2001. Balkema Publishers, Tucson,
2) Geosynthetic encasement of stone columns installed in such Arizona, USA. pp.1567–1572.
a manner that induces significant locked-in geosynthetic Lohani, T.N., Matsushima, K., Aqil, U., Mohri, Y., Tatusoka, F., 2006. Evaluating the
tension is highly effective in enhancing the performance of fill strength and deformation characteristics of a soil bag pile from full-scale
laboratory tests. Geosynthetic International 13 (6), 246–264.
embankment on very soft clay. Malavizhi, Ilamparuthi, 2007. Comparative study on the behavior of encased
3) The role of stone columns in supporting the fill loading evolves stone column and conventional stone column. Soils and Foundations 47 (5),
with time. In order to adequately capture the behaviour of such 873–885.
Murugesan, S., Rajagopal, K., 2006. Geosynthetic-encased stone columns: numerical
a system, coupled analysis or long term monitoring of field evaluation. Geotextiles and Geomembranes 24 (6), 349–358.
performance is essential. Murugesan, S., Rajagopal, K., 2007. Model tests on geosynthetic-encased stone
4) The computed results are relatively insensitive to the stiffness columns. Geosynthetic International 14 (6), 346–354.
Oh E.Y.N., Balasubramaniam, A.S., Bolton, M., Surarak, C., Bolton, M., Chai, G.W.K.,
parameters of the compacted stones.
Huang, M., 2007. Behavior of a highway embankment on stone columns
5) The computed settlement is significantly affected by the improved estuarine clay. Proceedings of 16th Southeast Asian Geotechnical
locked-in stress in the geosynthetic. However, the system has Conference, Malaysia, vol. 1, pp. 567–572.
Potts, D.M., Ganendra, D., 1991. Discussion on Finite element analysis of the collapse
to ability to self-compensate.
of reinforced embankment on soft ground by Hird C.C., Pyrah I.C., Russel D.
Geotechnique 41 (4), 627–630.
Aknowledgement Wu, C.S., Hong, Y.S., 2008. The behaviour of a laminated reinforced granular column.
Geotextiles and Geomembranes 26 (4), 302–316.
Wu, C.S., Hong, Y.S., 2009. Laboratory tests on geosynthetic encapsulated sand
The second author was supported by the Cheung Kong columns. Geotextiles and Geomembranes 27 (2), 107–120. doi:10.1016/
Endeavour, Asia (China) Award and University College Postgraduate j.geotexmem.2008.09.003.

You might also like