You are on page 1of 11

Geosynthetics International, 2009, 16, No.

Numerical modeling of geosynthetic-encased stone


column-reinforced ground
C. Yoo1 and S.-B. Kim2
1
Professor, Department of Civil and Environmental Engineering, Sungkyunkwan University, 300
Chun-Chun Dong, Jan-An Gu, Suwon, Kyong-Gi Do, Korea 440-746, Telephone: +82 31 290 7518,
Telefax: +82 31 290 7549, E-mail: csyoo@skku.edu
2
Graduate Student, Department of Civil and Environmental Engineering, Sungkyunkwan University,
300 Chun-Chun Dong, Jan-An Gu, Suwon, Kyong-Gi Do, Korea 440-746, Telephone: +82 31 290 7518,
Telefax: +82 31 290 7549, E-mail: ksb9677@hanmail.net

Received 14 October 2008, revised 10 January 2009, accepted 11 January 2009

ABSTRACT: This paper presents the results of a comparative study on different finite element
modeling approaches for modeling geosynthetic-encased stone column-reinforced ground for use in
rapid embankment construction. The specific models considered include: (1) an axisymmetric unit
cell; (2) a three-dimensional (3D) column; and (3) a full 3D model. The validity of the unit cell
model was tested by comparison with the results from the 3D models. The applicability of
continuum elements for modeling the geosynthetic encasement was also investigated. The results
show that the 3D column model yielded practically identical results when compared with those of
the full 3D model whereas the two-dimensional axisymmetric unit cell model tended to yield
results that were 10 to 20% larger in terms of the vertical effective stress and lateral deformation
of the stone column. It is also shown that a layer(s) of continuum elements can be used to model
the geosynthetic encasement instead of membrane elements which are not readily available in
commercial software for geotechnical analysis, provided that the axial stiffness of the geosynthetic
encasement is taken into consideration. Based on the results of analysis, the effect of geosynthetic
encasement on the performance of stone columns installed in soft ground under embankment
loading is also discussed.

KEYWORDS: Geosynthetics, Stone column, Finite element method, Membrane effect, Consolidation

REFERENCE: Yoo, C. and Kim, S.-B. (2009). Numerical modeling of geosynthetic-encased stone
column-reinforced ground. Geosynthetics International, 16, No. 3, 116126.
[doi: 10.1680/gein.2009.16.3.116]

Numerical modeling can be used as an effective tool


1. INTRODUCTION
when investigating the effect of geosynthetic encasement in
Stone columns are recognized as an effective means of stone column applications for soft ground improvement, as
improving soft ground for rapid embankment construc- it can reasonably simulate the relevant mechanisms by
tion. It is known that for a single granular pile, such as a adopting the stresspore pressure coupled formulation.
stone column, the most probable failure mechanism is Embankment construction on stone column-reinforced
bulging failure regardless of whether it is floating or ground is a three-dimensional (3D) problem that may
bearing on a firm layer (Madhav 2006). Recently, the require a 3D modeling approach to represent the mechani-
concept of encasing individual granular piles with geosyn- cal behavior of the inner and outer columns. The unit cell
thetics over the full or partial length of the columns in approach (Barksdale and Bachus 1983; Kempton et al.
order to increase their load-carrying capacity has been 1998; Han and Gabr 2002), however, has often been
acknowledged by numerous researchers (Raithel and adopted to represent the interior portions of the embank-
Kempfert 2000; Alexiew et al. 2005; Murugesan and ment, away from the influence of the embankment side
Rajagopal 2006, 2007; Yoo et al. 2007). Another example slopes. Therefore, the unit cell modeling approach may also
of the use of geosynthetics in stone column applications be adopted when modeling the geosynthetic-encased stone
involves the concept of a geosynthetic-reinforced load column application in soft ground improvement. The
transfer platform for improving the performance of appropriateness of the unit cell approach for modeling
rammed aggregate columns in embankment construction geosynthetic-encased stone columns (GESC) in soft ground
(Abdullah and Edil 2007). improvement application needs further examination.
1072-6349 # 2009 Thomas Telford Ltd 116
Numerical modeling of geosynthetic-encased stone column-reinforced ground 117

One of the key features in numerical modeling of the assumed to be placed as a sand mat after construction of
geosynthetic-encased stone columns is perhaps the model- GESCs.
ing of the geosynthetic encasement. The membrane effect
provided by the geosynthetic encasement needs to be 2.2. Geosynthetic-encased stone columns
correctly taken into consideration in a numerical model if Referring to Figure 1, 0.8 m diameter, crushed-stone-filled
realistic results are to be obtained. As indicated by Smith GESCs were assumed to extend to the firm layer. The
and Filz (2007), structural elements, such as beam and GESCs were arranged in a square grid pattern with 2.4 m
cable elements, may be used to represent geosynthetic center-to-center spacing, giving an area replacement ratio
reinforcement in plane strain models but cannot be used in (aE ) of 9%. It was assumed that the stone columns were
an axisymmetric model. Murugesan and Rajagopal (2006), fully encased with geosynthetics having an average axial
in particular, used a layer(s) of continuum elements to stiffness of J 2500 kN/m. The groundwater table was set
represent the geosynthetic reinforcement in their study on at the top surface of the clay layer. Construction of the
the effect of stone column encasement in terms of embankment was assumed to be done in 2 m increments,
improving the short-term load-carrying capacity of stone each lift of which was completed in 15 days with another
columns. Such a modeling approach, however, needs to be 10 days of waiting period for consolidation.
further examined for its appropriateness to establish a
sound numerical modeling strategy for the geosynthetic-
encased stone columns. 3. FINITE ELEMENT MODELING
In this paper, the results of a comparative study of
different modeling approaches for geosynthetic-encased 3.1. Constitutive models
stone column-reinforced ground are presented. The model- With regard to the constitutive modeling, the soft clay was
ing approaches examined include an axisymmetric unit modeled using the Modified Cam Clay material model. A
cell model, a 3D column model, and a full 3D model. The linear-elastic, perfectly plastic model with the Mohr
appropriateness of the axisymmetric unit cell and the 3D Coulomb failure criterion was used to model the crushed
column models was first examined by making compari- stone, the sand, and the fill. The geosynthetic was
sons with the results from the full 3D model. The modeled as a linear elastic material. Note that a grid of
applicability of a thin layer(s) of continuum elements for linear elastic material was used to model geosynthetic
modeling the geosynthetic encasement was also examined reinforcement for out-of-plane behavior by Smith and Filz
by comparing the results from the axisymmetric model in (2007). A summary of the constitutive model parameters
which continuum elements were used for the encasement of the materials is given in Table 1.
with results from the same model but with membrane
elements, which are specifically applicable for modelling 3.2. Description of finite element models
the membrane effect provided by the geosynthetic reinfor- A commercial finite element code, Abaqus (Abaqus
cement. The fundamental working mechanisms of the 2006), was selected for the analysis in order to take
geosynthetic-encased stone column were also investigated. advantage of its robustness in numerical solution strategy
The following sections describe the unit cell and the 3D for soil nonlinearity and stresspore pressure coupled
column representations of GESC-reinforced ground, nu- problems. The first author has successfully used Abaqus
merical models, the validation of the numerical model, for modeling a number of stresspore pressure coupled
and comparisons between axisymmetric and 3D modeling problems (Yoo 2005; Yoo and Jung 2007) as well as
of an example GESC-supported embankment. geosynthetic-reinforced earth walls (Yoo and Song 2006;
Yoo and Kim 2008).
Three modeling approaches were considered in this
2. CASE CONSIDERED
study: an axisymmetric unit cell modeling, a 3D column
A hypothetical embankment construction on GESC-rein- modeling, and an equivalent full 3D modeling. An
forced ground was considered. Details of the case are axisymmetric unit cell and the 3D column representation
given in the following sections. Stone columns without regions are shown in Figure 2. The diameter of the
encasement are referred to herein as ordinary stone axisymmetric unit cell model was selected to produce
columns (OSC). exactly the same area for the 3D column model so that the
two models would have the same area replacement ratio.
2.1. General description In the full 3D modeling, only the center row portion of
The embankment geometry with the geosynthetic-encased the domain was considered as the embankment usually
stone column-reinforced ground is shown in Figure 1, spans a considerable distance with approximately uniform
representing the right half of the domain on account of the cross-sectional geometry in the direction normal to the
line of symmetry about its center. As can be seen in this plane. Figure 3 shows the models considered. In all
figure, the embankment, filled with sandy soil, is 45 m models, the meshes were arranged in order to have the
wide and 6 m high with a 1V : 2H side slope. The clay same element size vertically to eliminate any possible
layer, which is 10 m deep overlying a firm layer, was errors arising from the mesh arrangement.
assumed to be very soft with a standard penetration test Figure 3(a) shows the axisymmetric unit cell model.
blow count N value less than 3, or cone penetration test The model was horizontally fixed on the vertical sides and
tip resistance less 5 MPa. A 1.0 m thick sand layer was full fixity on the base was assumed. For the hydraulic
Geosynthetics International, 2009, 16, No. 3
118 Yoo and Kim

CL
45 m El. (m)

6.0
Hemb 6 m Embankment 1
2 0.0
1.0 m Sand mat
1.0

Hc 10 m
Soft clay
GESC
11.0
800 mm diameter GESCs with area replacement ratio of 9% Firm layer

Figure 1. Cross section of GESC-reinforced ground

Table 1. Summary of constitutive model parameters for stone column, clay, and sand/fill materials

Property Stone column Clay Sand/fill

Model type

MohrCoulomb Modified cam clay MohrCoulomb

Unit weight (kN/m3 ) 19 18 19


Youngs modulus (kPa) 40 000 15 000
Poissons ratio 0.3 0.3 0.3
Cohesion, c9 (kPa) 5 3
Friction angle, 9 (8) 40 28
Dilation angle, (8) 20 10
Critical state stress ratio, M 1.0
Logarithmic hardening constant for plasticity, 0.2
Logarithmic bulk modulus for elastic material behavior, k 0.02
Initial yield surface size, ao , (kPa) 50
Initial void ratio, eo 1.0
Permeability, k (m/s) 1.2 3 102 1.2 3 106 1.2 3 102

Stone column not to scale


B 1.185 m

Soil

r 0.4 m r 0.4 m

B 1.185 m Duc 1.338 m


(a) (b) (c)

Figure 2. Three-dimensional column and unit cell representations: (a) stone column layout; (b) 3D column; (c) axisymmetric
unit cell

boundary condition, the phreatic level was set at the top axisymmetric elements (CAX8P) were used to represent
surface to generate a hydrostatic pore water pressure the clay layer and the stone column.
profile in the domain with the top boundary of the clay In the 3D column representation, only one-quarter of
layer having a zero pore pressure boundary condition to the column and its tributary area, bounded by four vertical
model free drainage. The embankment loading was simu- planes of symmetry for which there were no horizontal
lated by adding layers of elements representing the displacements in the directions perpendicular to the
embankment. Eight node stresspore pressure coupled symmetry planes (Figure 3(b)), was modeled. Similar to
Geosynthetics International, 2009, 16, No. 3
Numerical modeling of geosynthetic-encased stone column-reinforced ground 119

the axisymmetric model, the vertical side boundaries were node stress only elements (C3D20R) were used for the
horizontally fixed with full fixity at the base. The same embankment fill.
hydraulic boundary condition and the loading condition The geosynthetic reinforcement was modeled using the
were imposed on the model. The model was discretized membrane elements, MAX2 and M3D8R, respectively, in
using 20 node stresspore pressure coupled brick elements both the axisymmetric and the 3D modeling, available in
with reduced integration (C3D20RP). Abaqus. The membrane elements shown in Figure 4(a) are
With reference to Figure 3(c), the equivalent full 3D surface elements that offer strength in the plane of the
model had a width of 55 m covering the lateral portion of element but have no bending stiffness, and are known to
the domain but with only one row of stone columns. The be particularly useful in modeling the geosynthetic encase-
same displacement and hydraulic boundary conditions ment as they offer resistance against out-of-plane bulging.
were imposed as those in the 3D column model. Similar Referring to Figure 4(a), the top surface of a membrane
to the 3D column model, 20-node stresspore pressure is the surface in the positive normal direction and is called
elements with reduced integration (C3D20RP) were used the SPOS face for contact definition while the bottom
to represent the clay layer and the stone column, while 20- surface is in the negative direction along the normal and is

Sand Sand
mat mat

Embankment
Stone
Stone Sand mat
column
column

Soft clay

Soft Soft
clay clay

Stone columns &


Geosynthetic geosynthetics encasement
Geosynthetic
encasement encasement

(a) (b) (c)

Figure 3. Finite element models considered: (a) axisymmetic unit cell; (b) 3D column; (c) equivalent full 3D

Stone column

n face SPOS
4 3

n 3
12

Geogrid
Z
Y
face SNEG 1
X 2

(a) (b)

Figure 4. Membrane element and geosynthetic encasement modelling: (a) membrane element; (b) geosynthetic encasement
modeling (3D)
Geosynthetics International, 2009, 16, No. 3
120 Yoo and Kim

called the SNEG face. In the 3D modeling, the membrane The basis of comparison for the numerical model
elements were used to wrap around the stone columns as adopted in this study and for the field measurements of
shown in Figure 4(b). Details of the two-dimensional and the test embankment was the settlement at the center of
3D membrane elements can be found in Abaqus (2006). the embankment. A comparison was also made between
The applicability of continuum elements for modeling the the current model and that of Tan et al. (2008) using the
geosynthetic reinforcement was also examined using the pore water pressure at the mid-depth of the clay layer. In
axisymmetric model. Figure 5(a) in which the time-settlement relation obtained
from the current full 3D model is compared with the field
3.3. Simulation procedure measurements, it can be observed that the current model
After establishing the initial stress and pore pressure appears to capture the general trend of the measured data
conditions with appropriate boundary conditions, the stone well. Furthermore, the time-excess pore water pressure
columns and the geosynthetic reinforcement were acti- relation obtained from the current model agrees fairly well
vated as wished-in-place. The embankment construction with that from the 3D model by Tan et al. (2008) as
was then simulated in three equal stages with 2 m fill shown in Figure 5(b). These results demonstrate that the
placement. The embankment loading was simulated by proposed model is appropriate for modeling the stone
adding layers of elements simulating the fill in all models. column-reinforced ground.
Each embankment fill placement was assumed to be
completed in 15 days, followed by a 10-day consolidation
period. After full placement of the embankment, the 0
analysis was carried out until the excess pore water Field measurement
pressure fell below a specified near zero value (i.e. 1 kPa). Present study
20
Settlement (mm) Tan et al. (2008)
3.4. Model validation
A case history concerning OSC-reinforced ground for 40
rapid embankment construction, reported by Tan et al.
(2008) was used to validate the modeling approach, with
particular reference to the stresspore pressure coupled 60

formulation. Tan et al. (2008) reported the measured


settlement as well as the 3D modeling results for the test 80
embankment. The stone columns were arranged in a 0 20 40 60 80 100 120 140
Time (days)
square grid pattern at a 2.4 m spacing, covering a distance
(a)
from the embankment base to a depth of 6 m. Details of
the test embankment can be found in Tan et al. (2008). 20
In this study, the full 3D modeling approach, described Present study
Excess pore pressure (kPa)

in the previous section, was adopted for analysis of the 16 Tan et al. (2008)
test embankment, following the simulation procedure
adopted by Tan et al. (2008) in their 3D analysis of the 12
test embankment using Plaxis (Brinkgreve et al. 2006).
The materials, including the clay, were assumed to follow 8
the MohrCoulomb (MC) failure criteria as assumed by
Tan et al. (2008). Table 2 summarizes the material 4
parameters used in the analysis. The embankment con-
struction was done in three equal stages, namely a 0.6 m 0
increment in embankment height. Each stage of embank- 0 20 40 60 80 100 120 140
Time (days)
ment construction was simulated with a 3 day consolida- (b)
tion period. The simulation was terminated when the
excess pore water pressure fell below 1 kPa. Details of the Figure 5. Comparison between field measurements and 3D
simulation procedure can be found in Tan et al. (2008). model: (a) settlement; (b) pore pressure

Table 2. Material parameters used for model validation

Material t (kN/m3 ) sat (kN/m3 ) v E (kPa) kh (m/s) kv (m/s) C9 (kPa) 9 (8)

Fill 18 20 0.3 15 000 1.16 3 105 1.16 3 105 3 33


Crust 17 17 0.3 15 000 3.47 3 107 1.16 3 107 3 28
Soft clay 15 15 0.3 1100 3.47 3 109 1.16 3 109 1 20
Stiff clay 18 18 0.3 40 000 3.47 3 109 1.16 3 109 3 30
Stone column 19 19 0.3 30 000 1.16 3 104 1.16 3 104 5 40

E, Youngs modulus; t , total unit weight; sat , saturated unit weight; kh , horizontal permeability coefficient; kv , vertical permeability coefficient.

Geosynthetics International, 2009, 16, No. 3


Numerical modeling of geosynthetic-encased stone column-reinforced ground 121
80
Embankment Ht. 6 m
4. UNIT CELL MODEL VERSUS 3D Replacement ratio 9%

Eff. str. increase, v,cl (kPa)


MODELS Full encasement
60
The results from the aforementioned three models were
compared in terms of settlement, geosynthetic strains, and
40 3D Column
stresses in the ground. Note that the geosynthetic rein-
forcement was modeled using membrane elements in all Full 3D

three models, and therefore any discrepancies between the 20


Axis. unit cell
models solely represent the differences in the different Full 3D
(w/o goegrid)
modeling approaches. The results for the OSC-reinforced
ground, obtained using the full 3D model, are also 0
0 20 40 60 80 100 120
presented to illustrate the effect of the encasement. It Elapsed time (days)
should be noted that the results from the full 3D model (a)
represent those for the interior columns.
5.0
4.1. Settlement and excess pore water pressure
Figure 6 shows the development of the excess pore water 4.0

Str. concen. ratio, v,sc


pressure (u) and the surface settlement (Sc ), the former 3D Column
of which was measured at mid-depth of the clay layer 3.0
Full 3D
between two adjacent stone columns. The variations of
Axis. unit cell
vertical effective stress increase in the clay layer at mid- 2.0
Full 3D
depth 9v,cl and the stress concentration ratio SCR at the Embankment Ht. 6 m (w/o goegrid)
same location with time are shown in Figure 7. Note that 1.0 Replacement ratio 10%
SCR is defined as the ratio of vertical effective stress in Full encasement
the stone column (9v,sc ) to 9v,cl , by which the degree 0.0
of the embankment load transfer to the stone columns can 0 20 40 60 80 100 120
Elapsed time (days)
be quantified. The results for the OSC case using the full
(b)
3D model are also shown.
In Figure 6, it can be seen that the 3D column and the Figure 7. Development of vertical effective stress and stress
concentration ratio: (a) effective stress; (b) SCR
10
Embankment Ht. 6 m 3D Column
Replacement ratio 9% Full 3D PEMAG PEMAG
Excess pore pr., u (kPa)

8 Full encasement (Avg: 75%) (Avg: 75%)


Axis. unit cell 8.0e02 8.0e02
7.3e02 7.3e02
Full 3D 6.7e02 6.7e02
6 6.0e02
(w/o goegrid) 6.0e02
5.3e02 5.3e02
4.7e02 4.7e02
4.0e02 4.0e02
4 3.3e02
3.3e02
2.7e02
2.0e02 2.7e02
1.3e02 2.0e02
6.7e03 1.3e02
2 0.0e00 6.7e03
0.0e00

0
0 20 40 60 80 100 120
Elapsed time (days)
(a)

0
Embankment Ht. 6 m
Replacement ratio 9%
100 Full encasement
Settlement, Sc (mm)

(a) (b)
200 3D Column
Full 3D Figure 8. Plastic strain magnitude (PEMGA) contour plots:
300 (a) GESC; (b) OSC
Axis. unit cell
Full 3D
400 (w/o geogrid)

full 3D models essentially gave the same results in terms


500 of u and Sc . The axisymmetric unit cell model however
0 20 40 60 80 100 120 gave slightly larger values for u and Sc (0.7 kPa and
Elapsed time (days)
(b)
12 mm, respectively) than the 3D column model, although
the area replacement ratios of the two models were kept
Figure 6. Development of excess pore pressure and the same. Such disparities suggest that the geometrical
settlement: (a) pore water pressure; (b) settlement conversion from rectangularity in the 3D column model to
Geosynthetics International, 2009, 16, No. 3
122 Yoo and Kim

the unit cell could have caused some errors in u and Sc , condition, the embankment load level in addition to the
although the area replacement ratios were the same. replacement and the modulus ratios should therefore be
Nevertheless, the unit cell and the 3D column modeling considered.
approaches appear to provide the results in parallel with The effect of encasement can also be observed in Figure
the full 3D model, in terms of the excess pore water 8, in which the magnitudes of plastic strain (PEMAG) in
pressure and the settlement. Figure 7(a) also shows that the clay, which are defined as
the three different models yielded almost identical results r
2 pl
in the embankment load-induced effective stress increase  :  pl
9v,cl . On the other hand, for the stress concentration ratio 3
SCR at the final stage, the unit cell model yielded a
slightly larger SCR value than the 3D models, namely 4.7 where  pl is the plastic strain and the symbol (:) denotes a
for the unit cell model and 4.2 for the 3D models. The scalar product operation, are compared. It can be seen that
difference in SCR was in fact caused by an 11% larger the plastic strains for the GESC case were only one-
effective stress increase in the stone column 9v,sc in the quarter of those for the OSC case, owing to the added
unit cell than in the 3D models with no difference in lateral confinement effect provided by the encasement and
9v,cl . Although the difference in SCR of 0.5 may not be the reduced vertical stresses.
of practical importance, a slightly larger SCR should be
expected when adopting a unit cell model rather than a 3D 4.2. Lateral deformation of stone column and
model. geosynthetic strains
The effect of geosynthetic encasement on the perform- Geosynthetic encasement provides resistance to lateral
ance of stone column-reinforced ground is shown in bulging of a stone column when loaded with an embank-
Figures 6 and 7. For example, in Figure 6(a), the maxi- ment fill. The geosynthetic strains, hoop strains in parti-
mum u for the OSC was approximately 9 kPa whereas cular, need to be accurately evaluated for a given loading
less than 3 kPa of u was developed for the GESC, the condition to check the adequacy of the geosynthetic rein-
difference in which was a direct consequence of less forcement in terms of rupture. Any numerical model
embankment load transfer to the clay layer. Such a trend evaluating the performance of GESC-reinforced ground
is well supported in the progressive development of Sc , as needs to have the ability to correctly predict strains in the
shown in Figure 6(b), in which the final settlement of the geosynthetic encasement.
GESC was reduced by more than half when compared Figures 9 and 10 illustrate the lateral deformation (h )
with the OSC. The reduced settlement for the GESC is of the stone column and the geosynthetic hoop strain h
thought to be a direct consequence of the increased profiles at the selected embankment loading stages. Figure
stiffness of the stone column by the geosynthetic encase- 11 presents the development of h,max with time. Note that
ment. In addition, Figure 7 shows that the added confine- the hoop strain was defined as h h /ro, where ro is the
ment effect by the geosynthetic encasement also decreased initial radius of the encasement. As expected, comparisons
9v,cl , as much as 10 kPa, and the stress concentration between the different models in terms of h and h
ratio from 4.8 to 4.2. This trend is important in terms of profiles reveal that the results from the 3D column model
design in that the stress concentration ratio is not constant were almost identical to those from the full 3D model.
but increases with the embankment load level (fill height). The axisymmetric unit cell model however yielded
When selecting a stress concentration for a given design 10,20% larger h and h than the 3D models, in terms of

12 12 12
Embankment Ht. 2 m Embankment Ht. 4 m Embankment Ht. 6 m

3D Column 3D Column
3D Column
Full 3D Full 3D Full 3D
Depth from top of SC (m)

Depth from top of SC (m)


Depth from top of SC (m)

Axis. unit cell Axis. unit cell Axis. unit cell


8 8 Full 3D 8
Full 3D Full 3D
(w/o goegrid) (w/o goegrid) (w/o goegrid)
Replacement ratio 9% Replacement ratio 9%
Replacement ratio 9% Full encasement Full encasement
Full encasement
4 4 4

0 0 0
0 1 2 3 4 0 2 4 6 0 5 10 15 20 25
Geogrid lateral deflection (mm) Geogrid lateral deflection (mm) Geogrid lateral deflection (mm)
(a) (b) (c)

Figure 9. Lateral deformation of stone column for different embankment fill heights: (a) Hemb 2 m; (b) Hemb 4 m;
(c) Hemb 6 m
Geosynthetics International, 2009, 16, No. 3
Numerical modeling of geosynthetic-encased stone column-reinforced ground 123
12 12 12
Embankment Ht. 2 m Embankment Ht. 4 m Embankment Ht. 6 m
3D Column
Full 3D
Axis. unit cell
Replacement ratio 9%
Depth from top of SC (m)

Depth from top of SC (m)

Depth from top of SC (m)


8 8 Full encasement 8

3D Column
Full 3D
Axis. unit cell 3D Column
4 4 4
Full 3D
Axis. unit cell Replacement ratio 9%
Replacement ratio 9% Full encasement
Full encasement

0 0 0
0 0.1 0.2 0.3 0 0.1 0.2 0.3 0.4 0.0 0.2 0.4 0.6 0.8 1.0 1.2
Geogrid hoop strain (%) Geogrid hoop strain (%) Geogrid hoop strain (%)
(a) (b) (c)

Figure 10. Hoop strains in goesynthetic encasement for different embankment fill heights: (a) Hemb 2 m; (b) Hemb 4 m;
(c) Hemb 6 m

their maximum values. A similar observation can be seen trend shown in the lateral deformation plot. Such a trend
in Figure 11 which shows that the axisymmetric unit cell has an important implication in terms of selection of
model yielded 20% larger h,max than the 3D models, geosynthetic encasement depth (Denc ) for design as it has
although the general trends were identical. been reported in the studies by Murugesan and Rajagopal
The effect of geosynthetic encasement is clearly illu- (2006, 2007) concerning the short-term load-carrying
strated in Figure 9 as the OSC exhibited considerable capacity of geosynthetic-encased stone columns that the
lateral bulging, as great as 22 mm, whereas the maximum optimum Denc beyond which no further benefit in terms of
lateral bulging of the GESC was limited to only 2.5 mm. the encasement effect could be obtained was 2D. This
Another interesting trend is the location of the maximum suggests that different optimum encasement depths should
lateral deformation. For example, for a relatively low be adopted for different loading conditions, i.e., short or
embankment height, namely Hemb 2m, the bulging long term. Further study is needed in this area.
mainly occurred in the upper 3D zone, where D is the
diameter of stone column. The bulging zone then moved
downward, as the embankment height increased, with the
5. MODELING OF GEOSYNTHETIC
maximum bulging occurring 8D below the top of the stone
ENCASEMENT
column, suggesting that the encasement needs to be The applicability of a layer(s) of continuum elements in
provided beyond a depth of 8D to ensure the maximum modeling the geosynthetic encasement was examined
effect. Such a trend can also be observed in Figure 10. For using the axisymmetric model by comparing the results
example, the maximum hoop strain of h,max 0.2% from the same model but with membrane elements. When
occurred at 1,2D below the top when Hemb 2m. The modeling the geosynthetic encasement with continuum
maximum hoop strain h,max then increased to 1.0% at the elements, the axial stiffness J was kept the same as that of
final embankment height of 6 m, occurring at a depth of the membrane elements. Note that J is the secant stiffness
8D below the top of the stone column, which supports the of the geosynthetic, which is the ratio of tensile force per
unit width to the average strain in the geosynthetic and is
1.2 defined as J Et, where E is the elastic modulus and t is
the thickness of the layer of continuum elements repre-
Embankment Ht. 6 m
1.0
Replacement ratio 9%
senting the geosynthetic encasement. Note that the thick-
Full encasement ness of the layer of continuum elements representing the
Hoop strain, h (%)

0.8
geosynthetic encasement was assigned as t 15 mm with
an elastic modulus of 1.7 3 105 kPa, yielding
0.6
3D Column J 2500 kN/m.
0.4 Full 3D In Figure 12, the progressive development of settlement
Axis. unit cell Sc at the embankment base and the stress concentration
0.2 ratio SCR during the embankment construction are illu-
strated. As shown, no significant discrepancies between
0.0 the two models in Sc and SCR are apparent, suggesting
0 20 40 60 80 100 120
Elapsed time (days) that a layer(s) of continuum elements can produce almost
the same results as the membrane elements, provided that
Figure 11. Development of maximum hoop strain with time the axial stiffness of the layer(s) of continuum elements is
Geosynthetics International, 2009, 16, No. 3
124 Yoo and Kim
0
of the stone column as well as the geosynthetic hoop
Membrane element
strain profiles from the two models are practically iden-
40 Solid element
tical.
Settlement, St (mm)

Based on the results presented in this section, it can be


80 Embankment Ht. 6 m
Replacement ratio 9%
concluded that a layer(s) of continuum elements can
Full encasement represent the geosynthetic encasement when modeling
120 J 2500 kN/m GESC-reinforced ground if the axial stiffness of the
geosynthetic encasement is properly taken into considera-
160 tion.

200
0 20 40 60 80 100 120 6. CONCLUSIONS
Elapsed time (days)
(a)
The results of a comparative study on different approaches
for modeling geosynthetic-encased stone column-rein-
5.0
forced ground for use in rapid embankment construction
are presented. In particular, the appropriateness of the unit
Str. concen. ratio, v,sc / v,cl

4.0 cell modeling approach was fully examined by comparing


the results from axisymmetric and 3D column models with
3.0
Membrane element
those from a full 3D model in terms of settlement,
Solid element
stresses, lateral deformation, and geosynthetic strains. The
2.0 applicability of continuum elements for modeling the
Embankment Ht. 6 m geosynthetic encasement was also examined by comparing
Replacement ratio 9%
1.0 Full encasement the results with those from otherwise the same model but
J 2500 kN/m with membrane elements, which is known to be valid
0.0 for modeling geosynthetics. The fundamental working
0 20 40 60 80 100 120
mechanisms of the geosynthetic-encased stone columns
Elapsed time (days)
(b)
were also investigated. The following conclusions can be
drawn based on the findings from this study.
Figure 12. Development of settlement and maximum hoop
strain with time (continuum versus membrane elements): 1. The results of the 3D column model showed good
(a) settlement; (b) stress concentration ratio agreement with those from the full 3D model in all
aspects, i.e., excess pore pressure, settlement, vertical
stress, lateral deflection, and geosynthetic strains.
made the same as that of the membrane elements simulat- The axisymmetric unit cell model, however, tended
ing the geosynthetic encasement. Such a conclusion is to give 10,20% larger results than the 3D models,
fully supported by the lateral deformation profiles of the particularly for the vertical effective stress and lateral
stone column and the geosynthetic hoop strains for differ- deformation of the stone column, and the geosyn-
ent embankment loading stages, shown in Figures 13 and thetic strains, partly because of the error caused by
14, respectively. As shown, the lateral deformation profiles the geometrical conversion from rectangularity in the

12 12 12
Embankment Ht. 2 m Embankment Ht. 4 m Embankment Ht. 6 m
Membrane element
Solid element
Depth from top of SC (m)

Depth from top of SC (m)

Depth from top of SC (m)

8 8 8

Membrane element Membrane element


Solid element Solid element
4 4 4 Embankment Ht. 6 m
Embankment Ht. 6 m Embankment Ht. 6 m Replacement ratio 9%
Replacement ratio 9% Replacement ratio 9% Full encasement
Full encasement Full encasement J 2500 kN/m
J 2500 kN/m J 2500 kN/m

0 0 0
0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5
Lateral deflection (mm) Lateral deflection (mm) Lateral deflection (mm)
(a) (b) (c)

Figure 13. Lateral deformation of stone column for different embankment fill heights (continuum versus membrane elements:
(a) Hemb 2 m; (b) Hemb 4 m; (c) Hemb 6 m
Geosynthetics International, 2009, 16, No. 3
Numerical modeling of geosynthetic-encased stone column-reinforced ground 125
12 12 12
Embankment Ht. 2 m Embankment Ht. 4 m Embankment Ht. 6 m

Membrane element
Solid element
Depth from top of SC (m)

Depth from top of SC (m)

Depth from top of SC (m)


8 8 8
Membrane element
Membrane element
Solid element
Solid element

Embankment Ht. 6 m Embankment Ht. 6 m


4 Replacement ratio 9% 4 Replacement ratio 9% 4
Full encasement Full encasement Embankment Ht. 6 m
J 2500 kN/m J 2500 kN/m Replacement ratio 9%
Full encasement
J 2500 kN/m

0 0 0
0.0 0.5 1.0 1.5 0.0 0.5 1.0 1.5 0.0 0.5 1.0 1.5
Geogrid hoop strain (%) Geogrid hoop strain (%) Geogrid hoop strain (%)
(a) (b) (c)

Figure 14. Hoop strains in geosynthetic encasement for different embankment fill heights (continuum versus membrane
elements): (a) Hemb 2 m; (b) Hemb 4 m; (c) Hemb 6 m

3D column model to the unit cell, although the area encasement depth of two to three times the stone
replacement ratios were the same. The absolute column diameter in a short-term loading condition. It
magnitudes of the discrepancies between the axisym- appears that different optimum encasement depths
metric unit cell and the 3D models were, however, of should be adopted for different loading conditions,
little practical importance, and therefore it can be i.e., short- or long-term, although further study is
stated that the axisymmetric unit cell and the 3D needed in this area.
column modeling approaches are adequate for
modeling GESC-reinforced ground in rapid embank-
ment construction for interior columns.
ACKNOWLEDGEMENTS
2. In modeling the geosynthetic encasement, a layer(s) This work was supported by the Korea Ministry of Con-
of continuum elements gave practically the same struction and Transportation under grant C105A1000017-
results in comparison with the membrane elements, 05A0300-01700 and by the Basic Research Program of the
thus suggesting that continuum elements can be used Korea Science & Engineering Foundation under grant
as alternatives to membrane elements, which are not R01-2004-000-10953-0. This financial support is gratefully
readily available in commercial softwares for geo- acknowledged. Special thanks go to my former student
technical analysis, provided that the axial stiffness of A.-R. Song.
the geosynthetic encasement is taken into considera-
tion.
3. For geosynthetic-encased stone columns installed in
NOTATIONS
soft ground for embankment construction, the use of Basic SI units are given in parentheses.
geosynthetic encasement decreased the embankment
load-induced excess pore water pressures as well as aE replacement ratio (dimensionless)
vertical stresses in the ground, thereby decreasing the ao initial yield surface size (Pa)
associated settlement. Such a trend is attributed to c9 effective cohesion (Pa)
the increased stiffness of the stone column due to the E Youngs modulus (Pa)
added level of confinement provided by the eo initial void ratio (dimensionless)
encasement. Denc encasement depth (m)
4. The stress concentration ratio was not constant but Hc thickness of clay layer (m)
increased with the level of embankment load. When Hemb height of embankment (m)
selecting a stress concentration for a given design J axial stiffness of geogrid (N/m)
condition, the embankment load level in addition to k permeability (m/s)
the replacement and the modulus ratios should be kh horizontal permeability coefficient (m/s)
considered. kv vertical permeability coefficient (m/s)
5. For GESC-reinforced ground in rapid embankment M critical state stress ratio (dimensionless)
construction, the maximum lateral bulging of the ro initial radius of encasement (m)
stone column tended to occur at the bottom end of Sc settlement of embankment (m)
the stone column, suggesting full encasement is SCR stress concentration ratio (dimensionless)
needed to achieve the maximum encasement effect. t thickness of encasement (m)
Such a trend is contrary to the reported critical t total unit weight (N/m3 )
Geosynthetics International, 2009, 16, No. 3
126 Yoo and Kim

sat saturated unit weight (N/m3 ) of piled embankment. Proceedings of the 6th International
9v,sc effective stress increase in stone column (Pa) Conference on Geosynthetics, Atlanta, GA, USA, International
Geosynthetics Society, pp. 767772.
9v,sl effective stress increase in clay (Pa)
Madhav, M. R. (2006). Engineering of ground by stone columns/granular
h lateral deformation of stone column (m) piles. Proceedings, ATC-7 Workshop on Stone Column in Soft
logarithmic hardening constant for plasticity Deposits, Busan, Korea, Korean Geotechnical Society, pp. 117.
(dimensionless) Murugesan, S. & Rajagopal, K. (2006). Geosynthetic-encased stone
h hoop strain in geosynthetic (dimensionless) columns: Numerical evaluation. Geotextiles and Geomembranes,
 pl plastic strain (dimensionless) 24, No. 6, 349358.
Murugesan, S. & Rajagopal, K. (2007). Model tests on geosynthetic-
k logarithmic bulk modulus for elastic material encased stone columns. Geosynthetics International, 14, No. 6,
behavior (dimensionless) 346354.
9 effective internal friction angle (8) Raithel, M. & Kempfert, H. G. (2000). Calculation models for dam
dilation angle (8) foundations with geotextile-coated sand columns. Proceedings of
GeoEngineering 2000, Melbourne, Technomic Publishing, Lan-
caster, PA, USA, p. 347.
REFERENCES Smith, M. & Filz, G. (2007). Axisymmetric numerical modeling of a unit
cell in geosynthetic-reinforced, column-supported embankments.
Abaqus Inc. (2006). ABAQUS Users Manual, version 6.7. Abaqus, Inc., Geosynthetics International, 14, No. 1, 1322.
Pawtucket, Providence, RI, USA. Tan, S. A., Tjahyono, S. & Oo, K. K. (2008). Simplified plane-strain
Abdullah, C. H. & Edil, T. B. (2007). Behaviour of geogrid-reinforced modeling of stone-column reinforced ground. Journal of Geo-
load transfer platforms for embankment on rammed aggregate piers. technical and Geoenvironmental Engineering, 134, No. 2, 185194.
Geosynthetics International, 14, No. 3, 141153. Yoo, C. (2005). Interaction between tunneling and groundwater-numerical
Alexiew, D., Brokemper, D. & Lothspeich, S. (2005). Geotextile encased investigation using three dimensional stress-pore pressure coupled
columns (GEC): load capacity, geotextile selection and pre-design analysis. Journal of Geotechnical and Geoenvironmental Engineer-
graphs. Proceedings of the Geo-Frontiers Conference, Austin, ing, 131, No. 2, 240250.
Texas, USA, ASCE, Reston, VA, USA, Geotechnical Special Yoo, C. & Song, A. R. (2006). Effect of foundation yielding on
Publication, pp. 497510. performance of two-tier geosynthetic-reinforced segmental retaining
Barksdale, R. D. & Bachus, R. C. (1983). Design and Construction of walls: a numerical investigation. Geosynthetics International, 13,
Stone Columns. Report No. FHWA/RD-83/026, Office of Engineer- No. 5, 181194.
ing and Highway Operations Research and Development, Federal Yoo, C. & Jung, H. Y. (2007). Case history of geosynthetics reinforced
Highway Administration, Washington, DC, USA. segmental retaining wall failure. Journal of Geotechnical and
Brinkgreve, R. B. J., Broere, W. & Waterman, D. (2006). Plaxis 2D- Geoenvironmental Engineering, 132, No. 12, 15381548.
Version 8. Plaxis bv, Delft, The Netherlands. Yoo, C. & Kim, S. B. (2008). Performance of a two-tier geosynthetic
Han, J. & Gabr, M. A. (2002). Numerical analysis of geosynthetic- reinforced segmental retaining wall under a surcharge load: Full-
reinforced and pile-supported earth platforms over soft soil. Journal scale load test and 3D finite element analysis. Geotextiles and
of Geotechnical and Geoenvironmental Engineering, 128, No. 1, Geomembranes, 26, No. 6, 460472.
4453. Yoo, C., Song, A. R., Kim, S. B. & Lee, D. Y. (2007). Finite element
Kempton, G., Russell, D., Pierpoint, N. D. & Jones, C. J. F. P. (1998). modeling of geogrid-encased stone column in soft ground.
Two- and three-dimensional numerical analysis of the performance Journal of Korean Geotechnical Society, KGS, 23, No. 10, 133150.

The Editor welcomes discussion on all papers published in Geosynthetics International. Please email your contribution to
discussion@geosynthetics-international.com by 15 December 2009.

Geosynthetics International, 2009, 16, No. 3

You might also like