See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/277338052

Failure mechanism of geosynthetic-encased stone columns in soft soils under

embankment

Article  in  Geotextiles and Geomembranes · May 2015

DOI: 10.1016/j.geotexmem.2015.04.016

CITATIONS

99READS

431

4 authors, including:

Some of the authors of this publication are also working on these related projects:

Mechatronics View project

Multi-modal probabilistic analysis of slope stability View project

Jianfeng Chen

Tongji University

78 PUBLICATIONS   508 CITATIONS   

SEE PROFILE

Jianfeng Xue

University of New South Wales, Canberra, Australia

65 PUBLICATIONS   892 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Jianfeng Chen on 08 September 2019.

The user has requested enhancement of the downloaded file.

lable at ScienceDirect

Geotextiles and Geomembranes 43 (2015) 424e431

Contents lists avai

Geotextiles and Geomembranes

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

Failure mechanism of geosynthetic-encased stone columns in softsoils under embankment

Jian-Feng Chen a, Liang-Yong Li a, Jian-Feng Xue a, b, *, Shou-Zhong Feng a, c

a Department of Geotechnical Engineering, School of Civil Engineering, Tongji University, 1239 Siping Road, Shanghai 200092, Chinab Geotechnical and Hydrogeological Engineering Research Group (GHERG), Federation University Australia, Victoria 3842, Australiac Wuhan Guangyi Transportation Science and Technology Co., Ltd, Wuhan 430074, China

a r t i c l e i n f o

Article history:Received 6 December 2014Received in revised form20 April 2015Accepted 24 April 2015Available online 27 May 2015

Keywords:Geosynthetic-encased stone columnSoft soilBending failureUnbalanced lateral loading

* Corresponding author. Tel.: þ61 3 51226448.E-mail address: jianfen.xue@federation.edu.au (J.-

http://dx.doi.org/10.1016/j.geotexmem.2015.04.0160266-1144/© 2015 Elsevier Ltd. All rights reserved.

a b s t r a c t

The reaction of geosynthetic-encased stone columns (GECs) in soft soils under embankment loading wasmodeled with an indoor physical model test and numerical models using three dimensional and twodimensional finite element methods. The experimental and three dimensional numerical modeling re-sults showed that the failure of the GECs is caused by the bending of the columns rather than shear.Three dimensional finite element analysis showed that the distribution of unbalanced lateral loadingacting on the columns is symmetric about a ‘hinge point’ above the plastic hinge, rather than triangle oruniform distribution. An equivalent shear resistance model of the GECs is proposed based on the dis-tribution of the unbalanced lateral loadings on the wall. The stability of the embankment was analyzed intwo dimensional finite element method by transforming the columns into equivalent soil walls usingequivalent bending resistance and equivalent shear resistance methods. It was found that results fromequivalent bending resistance method is closer to the estimations from the three dimensional analysis,which agrees with the bending failure mechanism of the GECs. It is suggested that one more row of suchcolumns may be required to provide higher lateral resistance in the soils in front of the toe to improvethe stability of the embankment.

© 2015 Elsevier Ltd. All rights reserved.

1. Introduction

Stone columns have beenmorewidely used as a cost and energyefficient, and environmental friendly method for soft soil treat-ment. For situations when the undrained shear strength of soil istoo weak, stone columns may lose their effectiveness as the sur-rounding weak soils may not provide enough confinement to thecolumns, which may result in bulging or crushing failure of thecolumns at the upper section of the columns (Hughes et al., 1975).In that case, geosynthetic (i.e. geotextile or geogrid) encased stonecolumns (GECs) overcome the shortcomings and provide lateralconfinement to the stonematerials to improve the bearing capacityof the soils (Raithel et al., 2005; Yoo, 2010; Zhang et al., 2012; Dashand Bora, 2013; Elsawy, 2013; Wu and Hong, 2014).

The columns, such as sand compaction columns, stone columns,and deep mixed columns, can fail due to bending, sliding, rotation,shearing, tension, or a combination of the failure modes under

F. Xue).

embankment load (Kivelo and Broms, 1999; Han et al., 2005;Kitazume and Maruyama, 2006, 2007; Han, 2012; Zhang et al.,2014). Shear failure is the most common failure mode for sandcompaction and stone columns (Abusharar and Han, 2011). Hanet al. (2005) found that rotational failure of deep mixed columnsis dominant under road embankment based on the findings fromnumerical analysis. Through centrifuge tests, Kitazume andMaruyama (2006, 2007) found that the deep mixed columnscould fail under bending. They indicated that the area replacementratio of deep mixed columns influences the bending failuresignificantly and sliding failure might happen to shorter columns.Based on numerical modeling results, Zheng et al. (2010) suggestedthat rigid columns (e.g. concrete piles) are more prone to bendingfailure rather than shear failure under embankment loading.

However, there is very limited literature on the failure mode ofGECs under embankment load. This paper evaluates the stability ofa road embankment built on GECs reinforced soils using laboratorytesting, and three dimensional and two dimensional finite elementanalyses. The performance and failure mode of the encased stonecolumns under embankment loading are extensively studied.

J.-F. Chen et al. / Geotextiles and Geomembranes 43 (2015) 424e431 425

2. Laboratory testing

The authors carried out a test on geotextile-encased stone col-umn reinforced soft soils. The model was built in a1200 mm � 400 mm � 800 mm (length � width � height) tank asshown in Fig. 1 to simulate a 3.5 m high embankment at a scale of1:25 (model size to full size). The full size embankment will bemodeled with finite element method in the later sections. Kaolinwas mixed to the water content of 100%, which is well above theliquid limit (54.2%), to make the 400 mm thick of foundation soil. Astandard medium sand from Pingtan island in the eastern ChinaSea, Fujian Province, China was used in the test for sand cushion.The sand has beenwidely used in China for research purpose (Zhouet al., 2012). The sand used in the test is well graded with mean sizeof 0.34 mm and coefficient of uniformity of 1.542. The sand wascompacted to the density of 15.3 kN/m3 to construct the 50 mmthick sand cushion between the embankment and the foundationsoil. The friction angle of the sand was about 27.3� based on directshear tests at the dry density of 15.6 kN/m3. Silica sand with di-ameters ranging from 2 to 4 mm was compacted to the density of17.2 kN/m3 to construct the stone columns. The mean size of thesilica sand was 2.64 mm, and the coefficient of uniformity was1.861. The friction angle of the silica sand was 36.7�. Non-wovengeotextile was used to encase the silica sand. The tensile strengthof the geotextile was 0.42 kN/m based on tensile tests on six20mm� 20mm samples. The stiffness of the geotextilewas 4.0 kN/m at 5% tensile strain. Steel weights were used to construct theembankment. The weight was 380 g each, with the dimension of5 cm � 5 cm � 2 cm (length � width � thickness).

The Kaolin slurry was poured into the tank and consolidated fortwo weeks under the self-weight with double side drainage, andthe drains were remaining closed during the rest of the test. Afterconsolidation, the undrained shear strength of the soil was ob-tained at 5.2 kPa using a miniature cone penetrometer developed

Fig. 1. Dimensions of the laboratory model embankment on GECs reinforced soft soils(units are in mm): (a) section view; (b) plan view.

by Chen et al. (2012). The layout of the GECs is shown in Fig. 1. Toconstruct the sand columns, a 32 mm outer diameter steel pipewith 0.4 mm wall thick was driven into the soil and an augerextruder was used to remove the soil in the pipe. The non-wovengeotextile was sewn to form a 32 mm diameter and 400 mm longtube, and placed into the steel pipe to form the geotextile casing.Silica sand was poured into the casing and compacted in layers of50 mm to the designed density (17.2 kN/m3). After constructing thesand column, the steel pipe was carefully pulled out. Two pie-zometers (K1 and K2) were installed below themiddle section of theembankment to the depth of 200 mm as shown in Fig. 1(a) and (b)to monitor excess pore water pressure dissipation duringconstruction.

The embankment was constructed using steel weights in threestages (stages 1e3) as shown in Fig. 2. The embankment was con-structed in three stages using steel weights. Each loading stage was10 min (time to place the weights) followed by a resting period forthe dissipation of excess pore water pressure, which is about of30e40 min based on the monitored data from piezometers shownin Fig. 3. After the construction of the embankment, sand bags withequivalent pressure of 21 kPa were applied on the embankmentsurface to fail the structure (stage 4 in Fig. 2). The deformation ofthe ground before the last loading stage is shown in Fig. 4. Theshape of the deformed ground shows that large settlement hasoccurred below the embankment, with heave in the soils in front ofthe toe. The largest curvature of the deformed ground contour islocated below the toe of the embankment, where is also the lastcolumn located.

After removing the weights, the soils were left for 2 weeks withthe drains open to solidify the Kaolin for excavation. After theexcavation, it was found that the columns were bended, with thelargest deflection been observed in the column at the toe of theembankment as shown in Fig. 5, and the closer to the centerline ofthe embankment, the less the deflection in the columns. This agreeswith the ground contour shown in Fig. 4. The deformation of thecolumns shows that bending failure occurred in the columns nearthe toe.

3. Three dimensional finite element model

The physical model tested in the laboratory is relatively small insize, therefore the stress level encountered in situ and the stressvariations in the stone columns and geotextiles during the testcannot be fully studied. To further investigate the reaction of theGECs under road embankment, a 3.5 m high road embankmentwith 10 m wide on top was simulated with a three dimensionalfinite element code Z_Soil developed by the Swiss Federal Instituteof Technology. The software has been used with great success inforensic analysis and designs (Truty et al., 2008). The side slope ofthe embankment is 35�. The thickness of the underneath soft soil is10 m with the groundwater table located at ground surface. Theencased stone columns are 10 m long and 0.8 m in diameter, whichare installed at square pattern with center to center spacing of2.5 m. A 1.25 m road section was simulated in the three dimen-sional model to consider the symmetrical structure as shown inFig. 6. The material properties are shown in Table 1. The soils aremodeled with Mohr Coulomb model and the geotextiles wassimulated with isotropic linearly elastic perfectly plastic membranematerials embedded in the Z_Soil software. The geotextile has atensile stiffness of 2000 kN/m, Poisson's ratio of 0.3 and tensilestrength of 70 kN/m. The soft soil and geotextile interface wasmodeled with linear elastic-perfectly plastic interface model. Theinterface coefficient was 0.7 in the analysis.

Fig. 6 shows the finite element mesh used in the analysis. Themesh consists of 6132 elements. Impermeable boundary conditions

Fig. 2. Loading stages (each cell represents one steel weight).

Fig. 3. Dissipation of excess pore water pressure during construction.

Fig. 4. The deformation in the soils before the last loading stage.

Fig. 6. The numerical model of the GECs-supported embankment (units are in me-ters): (a) three dimensional view; (b) top view.

J.-F. Chen et al. / Geotextiles and Geomembranes 43 (2015) 424e431426

were applied at the bottom and sides of the model. Horizontal andvertical fixities were applied to the bottom boundary, with hori-zontal fixity on lateral boundaries. The embankment was builtusing stage construction method (continuous linear), with a

Fig. 5. The deflection of the GECs after the tests.

loading period of 30 days and resting period of 50 days after theloading.

4. Results from three dimensional finite element analysis

4.1. Lateral deflection of and stresses in the columns and geotextiles

The lateral displacements of the centerline of the columns at theend of the construction period, e.g. the 30th day, were plotted inFig. 7. The displacement curves clearly show the bending of thecolumns. The location of the largest curvature of the bended col-umns is deeper in the columns near the toe. It can be seen that thelargest displacement was observed at the head of the outmostcolumn, e.g. the column below the toe of the embankment, with

Table 1Material properties used in the three dimensional model.

Parameter Embankment fill Soft soil Stone column

Unit weight g (kN/m3) 20 17 22Young's Modulus/E (kPa) 20,000 3000 40,000Cohesion/c0 (kPa) 0.5 1 0.5Friction angle/f0 (�) 30 15 38Dilation angle/j (�) 10 0 10Poisson's ratio 0.3 0.3 0.3Permeability/k (10�6 m/s) 50 2.3e�3 120

Fig. 7. Deflections of the GECs at the end of construction period.

Fig. 9. Axial stresses in the GECs at the end of construction period.

J.-F. Chen et al. / Geotextiles and Geomembranes 43 (2015) 424e431 427

deflections of the columns reduce as the columns getting closer tothe centerline of the embankment. The columns remain almoststraight at greater depth, e.g. deeper than 5 m, for all the columnswith slightly tilting.

The tensile stresses in the geotextiles are shown in Fig. 8. It canbe seen that large tensile stresses have been generated in thegeotextiles in columns 3 and 4, with the largest value very close toits tensile strength, which may result in the tensile failure of thegeotextiles. The axial stresses in the stone columns at the end ofconstruction period are presented in Fig. 9. It can be seen that onthe tension side the axial stresses in column 3 and 4 are nearly zeronear the plastic hinge, where the largest bending moment located.The axial stresses in column 3 are greater than those in column 4, asthe column 3 is located below the shoulder of the embankmentwith higher load levels.

4.2. Lateral loading, shear force and bending moment diagrams ofthe columns

The net lateral loading diagrams of the columns were plotted inFig. 10 by integrating the normal stresses acting on the shafts. Thefigure shows that the columns acted as lateral loaded flexible piles,especially the ones near the toes where lateral displacement of the

Fig. 8. Tensile stress in the geotextiles at the end of construction period.

embankment and foundation soil is large. Fig. 10 shows that the netlateral loading acting on the columns is not zero at the head of thecolumns. The loading decreases with the depth to the ‘hinge point’at the depth of 1.2 m below the head of the columns, where thelateral loading is zero, then the direction of loading changes and thevalues increase to their maximum values at the depth of around2.8 m below the head of the columns. The loading distributionbetween the head and the maximum values is almost symmetricabout the ‘hinge point’.

By integrating the net lateral loading diagrams, the shear forceand bending moment diagrams can be obtained as shown in Fig. 11and Fig. 12. It can be seen that large shear forces and bendingmoments have been generated in columns 3 and 4. There is achange of shear force direction in the columns at the depth of 3 mwhere the maximum bending moment occurs. It seems that thelocations of the maximum shear forces in the columns differ in thecolumns. For example, the maximum shear force in column 3 islocated 1.6 m below the column head, while in column 2 is located2 m below the column head. This may be due to the variation ofstress levels in the soils below the embankment and the shape andlocation of the slip surface as shown later.

Fig. 10. Net lateral loading acting on the GEC shafts at the end of construction period.

Fig. 11. Shear force diagram of the GECs. Fig. 13. Lateral loading acting on the columns: (a) Kitazume et al. (2000) method; (b)Broms (2001) method; (c) assumption used in this paper.

J.-F. Chen et al. / Geotextiles and Geomembranes 43 (2015) 424e431428

Fig. 12 shows that the location of the maximum bendingmoment in each column differs, with the deepest location observedin column 4. This is due to the fact that during sliding of theembankment, slip surface may passes through the columns atdifferent locations, which results in the different degree of bendingof the columns. The location of themaximum curvature in a columnis affected by the depth of slip surface as shown in later sections.The maximum bending moment and shear force in column 3 isnearly 30% greater than the values in column 4 even the lateraldeflection in column 4 is larger. This is due to the fact that thevertical stresses acting on column 3 is much larger than that thoseon column 4 as shown in Fig. 9, which results in less bendingmoment in columns 3.

4.3. Equivalent shear capacity of the columns

At bending failure, Kitazume et al. (2000) assumed that theunbalanced lateral loading acting on the stone columns increaseslinearly with depth above the plastic hinge (Fig. 13(a)). The equiv-alent shear capacity (Pu) of the stone columns can be calculatedusing:

Pu ¼ 3Mu=hH (1)

in which Mu is the bending capacity of the column; hH is the dis-tance from the top of the column to the location of the plastic hinge,where the maximum bending moment is located.

Fig. 12. Bending moment diagram of the GECs.

Broms (2001) suggested uniformly distributed unbalancedlateral loading on piles or stone columns (Fig. 13(b)), then theequivalent shear capacity of the stone columns can be calculatedusing:

Pu ¼ 2Mu=hH (2)

From Fig. 10, we can see that the lateral loading acting on theGECs does not follow the assumptions by Broms (2001) andKitazume et al. (2000). As shown in Fig. 10, in the GECs, above theplastic hinge, the shear loading is almost symmetric about the‘hinge point’. The loading distribution can be assumed as the formshown in Fig. 13(c). Based on the assumption shown in the figure,the equivalent shear capacity of the GECs can be calculated using:

Pu ¼ 1:5Mu=hH (3)

This value is much lower than the values suggested by Broms(2001) and Kitazume et al. (2000) for the same Mu. Further testsneed to be done to validate this assumption: such as measuring theinclination of stone columns in centrifuge or in-situ tests underembankment loading.

4.4. Factor of safety and critical slip surface

The stability of the embankment was analyzed using strengthreduction method. In the method, the factor of safety (FS) of a soilstructure is obtained by reducing the strength parameters of thesoils:

FS ¼ ccr

¼ tan f

tan fr(4)

where c and f are the input cohesion and friction angle, respec-tively; cr and fr are the reduced cohesion and friction angle,respectively.

The factor of safety of the embankment was analyzed using thestrength reduction method built in Z_Soil. The calculated factor ofsafety is just 1.0, which suggests that the embankment is at limitstate. The simulated slip surface is shown in Fig. 14. It can be seenthat the moved soil body in the sliding zone caused the bending ofthe stone columns. The variation of the slip surface depth results indifferent degrees of bending of the stone columns, which may havecaused the variation of the locations of the maximum bendingmoment in the stone columns as discussed earlier.

J.-F. Chen et al. / Geotextiles and Geomembranes 43 (2015) 424e431 429

5. Two dimensional finite element analysis

Two dimensional finite element analysis is commonly used inpractice for its simplicity comparing to three dimensional finiteelement analysis. Three dimensional column-supportedembankment can be converted into an equivalent plane strainproblem using the column-wall method or the equivalent areamethod (Christoulas et al., 1997; Cooper and Rose, 1999; Tanet al., 2008; Abusharar and Han, 2011). Zhang et al. (2014)compared the stability of an embankment reinforced withstone columns by using above two conversion techniques andfound that, using the column-wall method normally gives lowerfactor of safety for short term stability, but for long term stabilitythe results from the two methods are quite comparable. Themain shortcoming of equivalent area method is that stressconcentration near the columns cannot be simulated; thereforefor short term stability, a reduction factor should be applied tothe factor of safety obtained from equivalent area method asrecommended by Zhang et al. (2014). In the following analysisthe reinforced soils were converted into column-walls usingequivalent shear resistance method and equivalent bendingresistance method.

5.1. Equivalent parameters of the column walls

The soil parameters in the equivalent walls are estimated usingthe weighted average strength parameters of the columns and thesoft soil (Terashi et al., 1991; Cooper and Rose, 1999):

cwAw ¼ cceAc þ csAs (5)

tan fwAw ¼ tan fceAc þ tan fsAs (6)

EwAw ¼ EceAc þ EsAs (7)

gwAw ¼ gceAc þ gsAs (8)

in which cw, fw, Ew and gw are the equivalent cohesion, frictionangle, Young's modulus and unit weight of the equivalent wall,respectively; cce, fce, Ece and gce are the apparent cohesion, frictionangle, Young's modulus and unit weight of the encased stone col-umn, respectively; cs, fs, Es and gs are the cohesion, friction angle,Young's modulus and unit weight of the soft soil, respectively; Aw,Ac and As are the areas of the equivalent wall, the encased stonecolumn and the soil, respectively.

Fig. 14. Critical slip surface predicted using the

5.2. Apparent cohesion of geotextile encased stone column

The inclusion of geotextile around the stone columns contrib-utes to the apparent cohesion (cce) of the encased stone columns(Malarvizhi and Ilamparuthi., 2008):

cce ¼s1g

2tan ð45� � fc=2Þ þ

s3g

2tan ð45� þ fc=2Þ (9)

where:

s1g ¼ pDEgεð1� εÞA0

(10)

s3g ¼ 2EgεcDεð1� εÞ (11)

In the equations, s1g is the increase in strength due to thecompression shell effect; s3g is the increase in strength due to hooptension; Eg is the Young's modulus of the geosysthetics; fc is thefriction angle of the stones; ε and εc are the vertical and circum-ferential strain of the stone column, respectively; D and Dε are theinitial diameter and diameter of the column at the axial strain of ε,respectively.

To obtain the apparent cohesion of the encased stone columnsusing Equation (9), the circumference strain induced by axial strainof the stone columns should be determined. Frikhan et al. (2015)carried out tri-axial tests on soft soil reinforced with sand col-umns and found that deviator stresses in the samples stabilize atthe axial strain level of 6e10%. Malarvizhi and Ilamparuthi (2008)compared the results from numerical and experimental tests ongeogrid encased stone columns, and found that using 10% of axialstrain gives the best comparison. In this study, a three dimensionalnumerical model of a geotextile-encased stone column with 0.8 min diameter and 1.6 m in length was built to simulate tri-axial testsof the encased stone columns. The circumferential strain obtainedis 4.8% at the 10% axial strain. By adopting these values, theapparent cohesion of the encased stone column was obtained as472 kPa using Equations (9)e(11). The friction angle of the encasedcolumn is assumed to be equal to the friction angle of the stones fc.

5.3. Equivalent modulus and Poisson's ratio of encased stonecolumns

The equivalent compressive modulus (Ece) and Poisson's ratio(mce) of the encased stone columns can be calculated using thefollowing equations (Zhou et al. 1998):

three dimensional finite element model.

Table 2Material properties of the equivalent soil wall used in the two dimensional model.

Equivalent parameter Equivalent bending resistance method Equivalent shear resistance method

Unit weight (kN/m3) 18.5 18.3Young's Modulus (kPa) 14,300 12,300Poisson's ratio 0.28 0.28Cohesion (kPa) 144.8 118.8Friction angle (�) 23 21.6

J.-F. Chen et al. / Geotextiles and Geomembranes 43 (2015) 424e431430

Ece ¼ ½1þ bð1� mcÞ�.h

1þ b�1� mc � 2m2c

�i(12)

mce ¼ mc

.h1þ b

�1� mc � 2m2c

�i(13)

in which:

b ¼ EgrEc

(14)

where r is the radius of the encased stone columns, Eg is the tensilemodulus of geosynthetics, Ec is the Young's modulus of the stones,mc is the Poisson's ratio of the stones. By using the values listed inTable 1, we obtained Ece ¼ 1.02Ec, and mce ¼ 0.94mc.

Fig. 15. Displacement of the embankment with the soil of: (a) ‘eq

5.4. Properties of the equivalent walls

The GECs reinforced soil zone was converted to equivalent wallsusing above mentioned methods by considering equivalent shearresistance and bending resistance. In the equivalent shear resistantmethod, the soil and stone columns are converted into soil wallswith equivalent shear resistance using Equations (5)e(8). In theequivalent bending moment method, the soil and stone columnsare converted into soil walls with thickness of b with equivalentbending moment:

112

ðDþ tÞb3 ¼ 164

pD4 (15)

uivalent bending resistance’; (b) ‘equivalent shear resistance’.

J.-F. Chen et al. / Geotextiles and Geomembranes 43 (2015) 424e431 431

in which D is the diameter of the encased column and t is thespacing between the columns. Then the area of the equivalent wallis:

Aw ¼ ðDþ tÞb (16)

By using the equations, the encased stone columns weretranslated into equivalent walls with the width of the 0.8 m usingequivalent shear resistance method and 0.459 m using equivalentbending resistance method. The parameters of the equivalent soilwalls for the two methods are shown in Table 2.

5.5. Failure mechanism in two dimensional analysis

The stability of the embankment was analyzed using strengthreduction method in the two dimensional finite element methodusing Z_Soil. For the embankment with the soil of ‘equivalentbending resistance’, the factor of safety obtained was 1.01, and forthe embankment with the soil of ‘equivalent shear resistance’, thefactor of safety obtained was 1.11. The critical slip surfaces with twodifferent equivalent materials are shown in Fig. 15(a) and (b). Thereis not much difference in the critical slip surfaces predicted fromthe two methods. Based on the factor of safety obtained, it seemsthat the results from the equivalent bending resistance method aremore comparable with the ones from three dimensional analysisfor this case.

6. Conclusions

A physical model test was carried out on a model embankmentbuilt on geotextile-encased silica sand reinforced soft soil toinvestigate the failure mechanism of GECs under road embank-ment. A numerical model embankment built on geotextile encasedstone column improved soft soil ground was analyzed using threedimensional and two dimensional finite element software Z_Soil.The reaction of the GECs was extensively studied using the threedimensional finite element code. In the two dimensional analysis,the reinforced soil zone was modeled with equivalent soil wallsusing equivalent bending resistance method and equivalent shearresistance method. Based on the findings from the physical andnumerical models, the following conclusions and recommenda-tions have been made.

1. Bending failure is the main failure mode in the geosynthetic-encased columns under embankment loading. The bending ofthe columns was caused by sliding of the embankment and thefoundation soil, and the unbalanced lateral loading acting on thecolumns.

2. Three dimensional finite element analysis showed that the netlateral loading acting on the columns above the plastic hinge isneither triangle nor uniformed distributed as assumed byKitazume et al. (2000) and Broms (2001). The lateral stressesdecreases from the head of the columns to zero at the ‘hingepoint’ where the lateral stress is zero, and maximize at theplastic hinge where the bending moment is the largest. Thelateral stresses above the plastic hinge are most likely sym-metric about the ‘hinge point’.

3. The bending moment and shear force diagrams in the columnsshowed that failure initiates in the columns at the edge of theslope due to the least lateral resistance in the columns. From thispoint of view, one more row of columns outside of the slope isnecessary to increase the lateral resistance of the soils in front ofthe embankment toe.

4. In two dimensional finite element analysis using equivalent soilwall method, the factor of safety varies with the methodsadopted for processing the columns. The result from equivalent

View publication statsView publication stats

bending resistance method is closer to the predictions fromthree dimensional analysis.

Acknowledgment

The support from the Shanghai Pujiang Program under grant No.14PJD032 is gratefully acknowledged.

References

Abusharar, S.W., Han, J., 2011. Two-dimensional deep-seated slope stability analysisof embankments over stone column-improved soft clay. Eng. Geol. 120 (1e4),103e110.

Broms, B.B., 2001. Discussion on ‘centrifuge model tests on failure envelope of col-umn type deep mixing method improved ground’. Soils Found. 41 (4), 103e107.

Chen, J.F., Liu, J.X., Ma, J., 2012. Calibration of a miniature cone penetrometer forgeotechnical model test. J. Tongji Univ. Nat. Sci. 40 (4), 549e552, 588. (inChinese).

Christoulas, S., Giannaros, C., Tsiambaos, G., 1997. Stabilization of embankmentfoundations by using stone columns. Geotech. Geol. Eng. 15 (3), 247e258.

Cooper, M.R., Rose, A.N., 1999. Stone column support for an embankment on deepalluvial soils. Proc. ICE-Geotech. Eng. 137 (1), 15e25.

Dash, S.K., Bora, M.C., 2013. Influence of geosynthetic encasement on the perfor-mance of stone columns floating in soft clay. Can. Geotech. J. 50 (7), 754e765.

Elsawy, M.B.D., 2013. Behaviour of soft ground improved by conventional andgeogrid-encased stone columns, based on FEM study. Geosynth. Int. 20 (4),276e285.

Frikhan, W., Tounekti, F., Kaffel, W., Bouassida, M., 2015. Experimental study for themechanical characterization of Tunis soft soil reinforced by a group of sandcolumns. Soils Found. http://dx.doi.org/10.1016/j.sandf.2014.12.014.

Han, J., 2012. Recent advances in column technologies to improve soft soils. In:Indraratna, B., Rujikiatkamjorn, C., Vinod, J. (Eds.), Invited Keynote Lecture,Proceedings of International Conference on Ground Improvement and GroundControl, Wollongong, Australia, 30 October to 2 November, vol. 1. ResearchPublishing, pp. 99e113.

Han, J., Huang, J., Porbaha, A., 2005. 2D numerical modeling of a constructedgeosynthetic-reinforced embankment over deep mixed columns. Contemp. Is-sues Found. Eng. 1e11. ASCE.

Hughes, J.M.O., Withers, N.J., Greenwood, D.A., 1975. A field trial of the reinforcingeffect of a stone column in soil. Geotechnique 25 (1), 31e44.

Kitazume,M.,Maruyama,K., 2006.External stabilityofgroupcolumntypedeepmixingimproved ground under embankment loading. Soils Found. 46 (3), 323e340.

Kitazume, M., Maruyama, K., 2007. Internal stability of group column type deepmixing improved ground under embankment loading. Soils Found. 47 (3),437e455.

Kitazume, M., Okano, K., Miyajima, S., 2000. Centrifuge model tests on failure en-velope of column type deep mixing method improved ground. Soils Found. 40(4), 43e55.

Kivelo, M., Broms, B.B., 1999. Mechanical behaviour and shear resistance of lime/cement columns. In: International Conference on Dry Mix Methods: Dry MixMethods for Deep Soil Stabilization, pp. 193e200.

Malarvizhi, S.N., Ilamparuthi, K., 2008. Numerical analysis of encapsulated stone col-umns. In: The 12th International Conference of International Association forComputer Methods and Advances in Geomechanics (IACMAG), Goa, India,pp. 3719e3726.

Raithel, M., Kirchner, A., Schade, C., 2005. Foundation of constructions on very softsoils with geotextile encased columns -state of the art. In: Proceedings ofGeoFrontiers 2005, Austin, Texas, USA, pp. 1e11.

Tan, S.A., Tjahyono, S., Oo, K.K., 2008. Simplified plane-strain modeling of stone-column reinforced ground. J. Geotech. Geoenviron. 134 (2), 185e194.

Terashi, M., Kitazume, M., Minagawa, S., 1991. Bearing capacity of improved groundby compaction piles. In: ASTM Special Technical Publication 1089, Deep Foun-dation Improvement: Construction and Testing, pp. 47e61.

Truty, A., Zimmermarm, T.H., Podles, K., 2008. Z_Soil. PC 2010 Manual. ZACE ServiceLtd, Lausaane.

Wu, C.S., Hong, Y.S., 2014. A simplified approach for evaluating the bearing per-formance of encased granular columns. Geotext. Geomemb. 42 (4), 339e347.

Yoo, C., 2010. Performance of geosynthetic-encased stone columns in embankmentconstruction: numerical investigation. J. Geotech. Geoenviron. Eng. 136 (8),1148e1160.

Zhang, Y.P., Chan,D.,Wang, Y., 2012. Consolidation of composite foundation improvedby geosynthetic-encased stone columns. Geotext. Geomemb. 32, 10e17.

Zhang, Z., Han, J., Ye, G.B., 2014. Numerical investigation on factors for deep-seatedslope stability of stone column-supported embankments over soft clay. Eng.Geol. 168, 104e113.

Zheng, G., Liu, L., Han, J., 2010. Stability of embankment on soft subgrade reinforcedby rigid inclusions (II)-group piles analysis. Chin. J. Geotech. Eng. 32 (12),1811e1820 (in Chinese).

Zhou, Z.G., Zhang, Q.S., Zheng, J.L., 1998. Analysis of mechanism of improved groundwith stone columns reinforced by geogrids. Chin. J. Civ. Eng. 31 (1), 20e25 (inChinese).

Zhou, J., Chen, J.F., Xue, J.F., Wang, J.Q., 2012. Micro-mechanism of interaction be-tween sand and geogrid Transverse Rib. Geosynth. Int. 19 (6), 426e437.