Effect of Geocell Type on Load-Carrying Mechanisms ofGeocell-Reinforced Sand Foundations

Sujit Kumar Dash1

Abstract: In this study, a series of model tests has been carried out to develop an understanding of the influence of the geocell material on theload-carrying mechanism of the geocell-reinforced sand foundations under strip loading. Geocells of different types were prepared using geo-grids of different types. The parameters studied are as follows: the footing load-settlement response, deformation on the fill surface, strain in thegeocell, pressure transmitted to the subgrade soil underlying the geocell mattress, and load dispersion in the geocell mattress. The test resultsindicate that the strength, stiffness, aperture opening size, and orientation of the rib of the geocell material influence the performance of thereinforced-sand foundation bed. Geocells made of geogrids of higher strength, relatively smaller size aperture opening, and ribs of orthogonalorientation give better performance improvement. DOI: 10.1061/(ASCE)GM.1943-5622.0000162. © 2012 American Society of CivilEngineers.

CE Database subject headings: Sand (material); Foundations; Load factors; Geomaterials.

Author keywords: Reinforced-sand foundation; Type of geocell reinforcement; Load-carrying mechanism.

Introduction

The techniques of reinforcing the earth have been used by man forcenturies, in the form of bamboo- or reed-supported clay wall, roadsover soft soil, among others. However, it is the pioneering work ofVidal (1969) that has brought amajor growth of interest in this subject.The basic attributes of reinforced earth are the overall economy andease of construction, coupled with simplicity, providing an addedattraction for the engineers. In earlier days, the concept of reinforcingthe earth was mainly pertaining to metallic reinforcements. Over thepast four decades, planar reinforcement in the form of metallic stripsand meshes, polymeric fabrics, and grids (i.e., geotextiles and geo-grids) havebeen extensively used, all over theworld, for improvementof foundations, roads, and the construction of walls (Binquet and Lee1975; Guido et al. 1986; Khing et al. 1993; Hatami et al. 2001;Léonard et al. 2002). Subsequent advancement in this field is toprovide three-dimensional confinement to the soil using geocells. Thegeocell foundationmattress is a series of interlocking cells constructedfrom polymer geogrid reinforcement, which contains and confines thesoil effectively. It intercepts the potential failure planes, and its rigidityforces them deeper into the foundation soil, thereby increasing thebearing capacity.

The beneficial effect of geocell reinforcement in soil fills hasbeen described by several researchers. Rea and Mitchell (1978) andMitchell et al. (1979) identified various modes of failure of geocellsby conducting model tests on sand-filled paper grid cells. Bathurstand Jarret (1989) have reported the use of geocell reinforcement forthe construction of pavements over extremely soft peat subgrades.Krishnaswamy et al. (2000) and Madhavi Latha et al. (2006) have

shown that geocell reinforcement is advantageous in increasingthe load-bearing capacity and reducing the deformations of earthembankments over weak foundation. Through model load tests,Dash et al. (2001, 2003), Sitharam et al. (2005), Yoon et al. (2008),Zhou and Wen (2008) have observed that soil reinforcement in theform of interconnected geocells gives rise to a large increase in load-carrying capacity and visible reduction in settlement of the footing.Shimizu and Inui (1990) and El Sawwaf and Nazer (2005) haveshown that confining foundation soil even with a single three-dimensional cell can bring a substantial increase in bearing ca-pacity and reduction in settlement of the footing. Field applicationsof geocells have been reported by Bush et al. (1990), Cowland andWong (1993), and Hendricker et al. (1998). Dash et al. (2004) haveobserved that the geocell mattress, made out of geogrid, is a superiorform of reinforcement over the planar geogrid system.

Review of the literature shows that geocell reinforcement is aneffective means of improving the performance of foundation soils.Besides, it provides a cost-effective solution (Robertson and Gilchrist1987; Paul 1988; Dean and Lothian 1990). As a result, the geocellsare being used in many civil engineering projects. They are broadlyof two types. When manufactured from high-density polyethylenesheets (solid or perforated), they are ultrasonically welded togetherinto a honeycomb pattern (called geowebs) and are typically of 100–200 mm in height (Bathurst and Rajagopal 1993). When they are ofa higher height, they are fabricated directly at the site, using geogrids(Bush et al. 1990). Such grid cells are popularly called geocells.Several case studies reported (Bush et al. 1990; Dean and Lothian1990; Cowland and Wong 1993; Forsman et al. 1998) demonstratethe efficacy of these grid cells in improving the performance offoundation soils. The effect of geometry of the geocell reinforcement(i.e., height, width, pocket size, pattern of formation, position ofplacement) on the performance of the foundation beds has beenextensively studied (Dash et al. 2007; Madhavi Latha et al. 2009).However, the influence of the properties of the materials, making upthe geocells, hasn’t yet been well understood. It is envisaged todevelop an understanding of the effect of geocellmaterial on the load-carrying mechanisms of the geocell-reinforced foundation beds. To

1Associate Professor, Dept. of Civil Engineering, Indian Institute ofTechnology Guwahati, Guwahati 781 039, India. E-mail: sujit@iitg.ernet.in

Note. This manuscript was submitted on April 22, 2010; approved onAugust 2, 2011; published online onAugust 4, 2011.Discussion period openuntil March 1, 2013; separate discussions must be submitted for individualpapers. This paper is part of the International Journal of Geomechanics,Vol. 12, No. 5, October 1, 2012. ©ASCE, ISSN 1532-3641/2012/5-537–548/$25.00.

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / SEPTEMBER/OCTOBER 2012 / 537

Int. J. Geomech. 2012.12:537-548.

Dow

nloa

ded

from

asc

elib

rary

.org

by

DA

PS L

IBR

AR

Y o

n 08

/31/

13. C

opyr

ight

ASC

E. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved.

achieve this objective, model load tests have been carried out oninstrumented geocell-reinforced sand foundation beds, with geocellsmade of geogrids of varying strength, stiffness, aperture opening size,and orientation of ribs.

Experimental Program

Themodel testswere conducted in a steel tankmeasuring 1,200mm inlength3 332 mm inwidth3 700 mm inheight.The lengthof the sidesof the tank was made of a thick perspex sheet and was braced withangle iron to avoid yielding under loading. The perspex wall, apartfromminimizing the friction between the soil and the tank, permittedobservationof thepattern of deformations of the sandduring tests. Theside-wall friction effects, on the model test results, were further re-duced by coating the inside of the perspex walls with colorless pe-troleum jelly. The model foundation was made of steel and measured330mmin length3 100 mm inwidth3 25 mm in thickness.A rough-base condition was achieved by cementing a thin layer of sand onto thebase of the model foundation with epoxy glue. The footing wascentered in the tank, with the length of the footing parallel to thewidthof the tank. Because the length of the footing was almost equal to thewidth of the test tank, a plane strain condition was generally main-tained during the tests.On each side of the tank, a 1-mmgapwas givento prevent contact between the footing and the sidewalls.

The soil used is a dry uniformly graded river sand (Unified SoilClassification System SP) with properties Cu 5 2:318, Cc 5 1:03,D5050:46mm, gmax517:410 kN=m3, and gmin514:30 kN=m3.Tests were carried out at a relative density (ID) of 70%. The peakfriction angle of the sand at a relative density of 70%, as determinedfrom triaxial compression tests, is 42:2� and one obtained underplane-strain conditions (i.e., direct shear tests) is 46�. Fig. 1 shows thethree different geogrids used for making the geocells of differenttypes. One is a biaxial grid (BX)made of oriented polymer, while theother two are made of nonoriented polymers, referred to as the NP 1and NP 2 grid. These geogrids were chosen to have varying tensilestrength and stiffness, aperture opening size, and shape (i.e., orienta-tion of ribs). The load-strain response of the geogrids, obtained throughthe standard multirib tension tests carried out as per ASTM D-6637(ASTM 2001), is depicted in Fig. 2. The properties of the geogrids arepresented in Table 1.

The geocellmattresseswere preparedby placing the geogrid stripsin transverse and diagonal directions with joints at the connections

(Bush et al. 1990). The joints of the geocellswere formedusing 6-mmwide and 3-mm-thick plastic strips cut from commercially availablebodkins made of low-density polypropylene. However, in the case oftheNP 2 geogrid, thinmild steel rods of 2-mmdiameterwere used forforming the joints because of the small aperture opening of thegeogrid. All the tests were performed with a single layer of geocellreinforcement, and the geocells were formed in a chevron pattern[Fig. 3(b)].

To achieve uniform density of soil in the foundation beds, a sand-raining technique was used. The accuracy of sand placement and theconsistency of the placement density were checked during rainingby placing small aluminum cans with known volumes at differentlocations in the test tank. The difference in the densities measuredat various locations in the test tank was found to be less than 1%.The geocell walls being of a geogrid, have a substantially highpercentage of openings. As a result, the free flow of sand duringraining hasn’t been affected much, leading to marginal reductionin the placement density. At the desired depth, the raining of sandwastemporarily ceased, and the geocellmattresswas placed on the surfaceof the prepared sand bed. After this, the sand raining continued to fillthe cells.

The footing was loaded by a hydraulic jack supported againsta reaction frame. The load was applied incrementally. With geocellreinforcement, the number of load increments was more than 20.Each load increment was maintained constant until the footing set-tlement stabilized and there was no significant change in settlement(i.e.,,0:02 mm=min). The load applied to the footing wasmeasuredthrough a precalibrated proving ring suspended from the spindle ofthe jack through an adapter and resting on the footing through a ballbearing. Settlements of the footing were measured by two dialgauges (Dg2, Dg3) (Fig. 3) placed in diagonal directions. Thedeformations on the fill surface (heave/settlement) were measuredby dial gauges (Dg1,Dg4) (Fig. 3) placed at a distance of 2.5 times thewidth of footing (B) to the left and right of the footing center line,respectively. The footing settlement (s) and surface deformation (d)are the average value of the two dial gauge readings. The geometryof the problem is shown in Fig. 3. The pocket size (d) of the geocellsis taken as the diameter of an equivalent circular area of the geocellpocket opening. A typical geocell pocket opening is shown throughthe hatch mark in Fig. 3(b). In all the tests, the pocket size of thegeocells (d), height of the geocell layer (h), width of the geocell layer(b), and the depth of the placement of the geocell layer (u) were keptconstant, that is, d=B 5 1:6, h=B 5 1:2, b=B 5 8, and u=B 5 0:1,

Fig. 1. Geogrids used for making geocell mattresses

538 / INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / SEPTEMBER/OCTOBER 2012

Int. J. Geomech. 2012.12:537-548.

Dow

nloa

ded

from

asc

elib

rary

.org

by

DA

PS L

IBR

AR

Y o

n 08

/31/

13. C

opyr

ight

ASC

E. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved.

because it is found to be the optimumconfiguration givingmaximumperformance improvement (Dash et al. 2008).

The strain developed in the geocell reinforcement was measuredthrough electrical resistance-type strain gauges of 10-mm-gaugelength. For geocells made of BX and NP 1 grid where the ribs areorthogonal, the strain gaugeswere placed in the horizontal direction.However, for NP 2 grid where the ribs are inclined at 45�, the straingauges were placed inclined at 45�. The strain gauges were fixed tothe geocell wall using a commercially available adhesive. All thestrain gauges were mounted on the transverse members of thegeocell reinforcements, along the width of the geocell mattress, atmidheight. A typical layout of the strain gauges on the geocell wall isshown in Fig. 3.

At each gauge location, the geogrid surface was mildly rubbedwith a fine sand paper and then wiped clean. Next, the strain gaugeswere pasted with quick setting adhesive. Unlike the planar re-inforcement system, the geogrid in the geocell reinforcement is heldby bodkin joints, which arrest the cross-plane bending. In the ab-sence of cross-plane bending, most of the strain induced in the re-inforcement is because of in-plane axial deformation. To verify thisfact, pilot tests were carried out with pairs of strain gauges pastedback to back on both sides of the geocell wall. A typical observationis presented in Fig. 4. Both the strain gauges have recorded tensilestrains (1) in the geocell wall with almost equal magnitude, in-dicating that there was no cross-plane bending of the geocell wall.

Fig. 2. Load-strain behavior of the geogrids

Table 1. Properties of Geogrids Used in Tests

Type of geogridAperture opening

shapeAperture opening

size (mm)Ultimate tensilestrength (kN/m)

Axial strain atfailure (%)

Secant modulus at 5%strain (kN/m)

Biaxial grid Square 353 35 20.0 25 160Nonoriented polymer 1 Square 503 50 4.5 10 70Nonoriented polymer 2 Diamond 83 7 7.5 55 70

Fig. 3.Geometry and details of instrumentation of the geocell-reinforcedfoundation bed

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / SEPTEMBER/OCTOBER 2012 / 539

Int. J. Geomech. 2012.12:537-548.

Dow

nloa

ded

from

asc

elib

rary

.org

by

DA

PS L

IBR

AR

Y o

n 08

/31/

13. C

opyr

ight

ASC

E. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved.

Therefore, in the present test program, a single strain gaugewas usedat each location for measurement of in-plane axial strain in thegeocell walls. Dummygaugeswerefixed to a piece of geogrid placedin another sand bed, just outside the tank, in order to simulate thesame environmental conditions for both the active and dummygauges. The length of the lead wires connected to both the active anddummy gaugeswere kept nearly equal in order to avoid drift in straingauge readings. The strain measurements are reported at variousnormalized footing load levels (bearing pressure ratio [BPR]. TheBPR is defined as the ratio between the footing pressure with thegeocell (q) and the ultimate footing pressure (qult) of unreinforcedsoil. The compressive strains are reported with a negative (2) sign,and the tensile strains are reported with a positive (1) sign.

The vertical earth pressures (s) were measured by placing straingauge-type earth pressure cells below the geocell layer. The overalldiameter and thickness of the pressure cells were 60 and 10 mm,respectively. In total, three pressure cells were used in each test: onebelow the center of the footing and the other two at a distance of 1.1Boneither sideof the footing centerline. Each cellwas calibrated a prioriby embedding the cell in the sand bed placed inside a calibrationchamber and applying known pressures uniformly on the fill surface(Dunnicliff 1988). In all the model tests, the earth pressure cells wereplaced below the pocket openings of the geocells rather than below itswalls, as to avoid any abnormal pressure readings because of stress-concentration effects. In the case of unreinforced earth beds, the earthpressures were measured at a depth corresponding to the base level ofthe geocell mattress. These data are used for comparison with thosemeasured below the geocells. At the required depth, the raining ofsand was temporarily stopped in order to place the earth pressurecells, and then the sand raining was continued. Hadala (1967) hasreported that this method of placement of the earth pressure cells(i.e., cells simply set on the surface, followed by normal constructionprocedure to complete the fill) is best. The pressures were recordedthrough a digital display unit. The measured pressures were nor-malized with respect to the applied footing pressure (q). The nor-malized pressure (s=q), which represents the percentage of the

footing pressure transmitted to the base of the geocell mattress, wasplotted at different footing loads, expressed in terms of the BPR.

The deformation pattern of the subgrade soil underlying thegeocell layer was observed by placing thin horizontal layers of whitecolored sand at 50mmvertical intervals. On completion of each test,the deformed shape of the colored lines was recorded by tracing ontransparent paper. The angle of load dispersion in the geocell mat-tress was deduced from the observed rupture surface, delineatedthrough the discontinuity in the colored lines.

Results and Discussion

The bearing pressure versus settlement responses of the footing fordifferent types of geocell reinforcement in the foundation bed areshown inFig. 5. The pattern of variation of the subgrade modulus (kr)of the foundation bed, obtained as the secantmodulus of the pressure-settlement responses (i.e., slope of the line joining the point on thecurve at a given settlement to the origin) is depicted in Fig. 6. Thefoundation shows almost equal bearing pressure with geocells madeof the BX and NP 2 grids up to a settlement equal to about 20% of thefooting width. Fig. 6 shows that at a relatively lower settlement, thestiffness of the foundation bed with an NP 2 grid geocell is evenhigher than that with a BX grid. This is in spite of the fact that thestiffness of the BX grid is much higher than that of the NP 2 grid(i.e., 5% strain secant modulus is more than 2 times higher) (Table 1).Because the aperture opening size of the NP 2 grid is almost 5 timessmaller than that of theBXgrid (Table 1), it offers higher confinementto the encapsulated soil. Consequently, a better composite materialis formed that redistributes the footing load over a wider area, givingrise to increased performance improvement. This could further besubstantiated from the observation that, though both NP 1 and NP 2grids have the same stiffness, in a relatively lower settlement range(i.e., s=B# 20%), performance with the NP 1 grid is comparativelyinferior. This is because of the higher aperture opening size of theNP 1 grid.

In Fig. 7, the settlement of the fill surface, recorded at a distanceof 2.5B from the center of footing, is the maximum in the case of theNP 2 geogrid. This indicates that the geocell earth bed made of theNP 2 grid acts as a better composite body, which enables it to deflectas a coherent mass under footing penetration. This establishes thehigher confining efficiency of the geocells made of geogrid havingsmaller aperture openings.

Besides, the geocell mattress being an interconnected cagederives anchorage from both sides of the loaded area through mo-bilization of interfacial friction between soil and reinforcement,interlocking of the soil through apertures of the geogrid, and passiveresistance through bearing at the soil to the grid cross-bar interface.With a decrease in aperture opening size of the geocell walls, theeffective reinforcement area available for mobilization of anchoragefrom soil increases, thereby giving rise to increased performanceimprovement.

The improvement in bearing capacity because of the geocellreinforcement is quantified using a nondimensional improvementfactor (If ), which is defined as the ratio of footing pressure (q) withgeocell, at a given settlement, to the pressure on unreinforced soil(qo) at the same settlement. If the footing has reached its ultimatecapacity at a certain settlement, the bearing pressure qo is assumedto remain constant at its ultimate value for higher settlements. Thebearing capacity improvement factor (If ) is different from the factorBPR.TheBPR is obtainedwith respect to the ultimate pressure of theunreinforced soil, which is independent of settlement, and hencea constant for the whole of this test program. Therefore, the BPRrepresents the normalized load applied unto the footing, while If

Fig. 4. Readings of a typical strain gauge pair mounted back to back onboth sides of the geocell wall

540 / INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / SEPTEMBER/OCTOBER 2012

Int. J. Geomech. 2012.12:537-548.

Dow

nloa

ded

from

asc

elib

rary

.org

by

DA

PS L

IBR

AR

Y o

n 08

/31/

13. C

opyr

ight

ASC

E. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved.

quantifies the bearing capacity improvement at different settlementlevels of the footing.

Variation of the bearing capacity improvement factor (If ) with thenormalized aperture opening size of the geogrid (da=D50) at differentsettlement levels of the footing (s=B) are presented in Fig. 8. In theinitial stages of loading, when the footing settlement is low, theimprovement factor (If ) reduces with an increase in the apertureopening size of the geogrid, making the geocells. Subsequently,under a relatively larger settlement of footing, the dense sand tends toflow. Because of the large extent of soil in the downward direction,the majority of the soil flow tends to be in the sideways upwarddirection, as observed in classical bearing capacity experiments(Chummar 1972). In the present tests, the relatively stiff geocellwalls, standing vertical, resist this flow through bearing on thegeogrid ribs. As a result, the equilibrium of the soil mass is restored;therefore, it continues to sustain increased loading (Jewell andWroth 1987). With geocell walls having larger openings (becauseof an increased aperture opening of the geogrid making up thegeocells), their ability to restrain the deforming soil mass reduces.This in turn results in a reduced mobilization of strength of geocellreinforcement, leading to reduced performance improvement (If ).

However, at a higher settlement of footing, the bearing capacityimprovement (If ) is at a maximum for the BX grid, whose strengthand stiffness is highest. At higher settlements, the sand under andaround the footing shears and starts moving away, as can be seenfrom Fig. 7 in terms of surface heaving. Because of this, a largeportion of the footing load gets directly transferred to the geocells;

hence, at this stage the stiffness of the geocell material (i.e., geogrid)influences the overall behavior of the foundation bed significantly.

Further, from Fig. 5, in the case of the BX grid geocell, at around30% settlement (s=B), the slope of the pressure-settlement responsereduces substantially. However, beyond that, the slope once againpicked up, with the bearing capacity increasing, with an increase inthe footing settlement. In the case of the NP 1 grid, the slope of thepressure-settlement response gradually reduces and tends to becomevertical beyond a settlement of around 32% of the footing width,indicating that the foundation bed has undergone failure. At a highersettlement, the soil within the geocell pockets, in the region belowthe footing, overcomes interfacial friction and interlocking over thegeocell wall and gets pushed down. With the shearing of encap-sulated soil, the geocell-soil composite structure breaks and hencea substantial part of the footing load is directly transferred to thegeocell matrix. At this stage, the strength of the geocell materialplays a dominant role in supporting the footing against the appliedloads, through mobilization of anchorage from both sides of theloaded area because of frictional, interlocking, and soil passiveresistance. The NP 1 grid having very low strength is found to haveundergone severe yielding at the joints, as observed in the posttestexhumed geocell walls. With the shearing of soil and yielding ofgeocell walls, the geocell-soil matrix undergoes complete failure, asobserved in the pressure-settlement response. Because the BX gridis of much higher strength, the geocell matrix continues to sustaina much higher intensity of footing pressure. The visible reductionin the slope of the pressure-settlement response at around 30%

Fig. 5. Bearing pressure-footing settlement: responses for different geocell materials

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / SEPTEMBER/OCTOBER 2012 / 541

Int. J. Geomech. 2012.12:537-548.

Dow

nloa

ded

from

asc

elib

rary

.org

by

DA

PS L

IBR

AR

Y o

n 08

/31/

13. C

opyr

ight

ASC

E. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved.

Fig. 6. Subgrade modulus-footing settlement: responses for different geocell materials

Fig. 7. Settlement and heave on the fill surface-footing settlement: responses for different geocell materials

542 / INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / SEPTEMBER/OCTOBER 2012

Int. J. Geomech. 2012.12:537-548.

Dow

nloa

ded

from

asc

elib

rary

.org

by

DA

PS L

IBR

AR

Y o

n 08

/31/

13. C

opyr

ight

ASC

E. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved.

settlement is because of the shearing of the encapsulated soil. Oncethe encapsulated soil shears away, the footing rests on the geocellmatrix, which enables it to further carry load, as observed in thepressure-settlement response.

In the case of geocells made of theNP 2 geogrid, a sudden drop inthe slope in the pressure-settlement response is observed at a settle-ment around 20% of the footing width. Beyond this, the pressure-settlement curve is almost vertical, indicating that the foundationbed has undergone sudden failure. This is because in the case of theBX and NP 1 grid, the aperture opening shape is square, and the ribsare in horizontal and vertical directions in the geocell wall; thereby,they effectively resist against footing penetration through mobili-zation of vertical compression and horizontal anchorage. The ribs ofthe NP 2 geogrid, being in an inclined direction (owing to diamondshape of its aperture opening), are unable to effectively resist thepenetration of the footing in a postsoil shearing stage. Indeed inthe posttest observation, upon removal of the top cushion of sand, thegeocell walls (of the NP 2 geogrid) were found to have folded inthe region under the footing. This folding of the geocell walls isbelieved to have caused the early failure of the footing in the caseof theNP2geogrid. In theother twocases (i.e., theBXandNP1grids),the geocell walls were heavily deformed and buckled, indicating thatthe geocell strength was substantially mobilized. Hence, for betterperformance improvement, the geocell mattress should bemade up ofa geogrid having a square/rectangular aperture opening, where theribs in the geocell walls are perpendicular and parallel to the footing.

The relatively early failure of the foundation bed with NP 2 gridgeocells compared with the ones having NP 1 and BX grid geocells(i.e., failure settlement is 20% in the case of NP 2, while it is 30% inthe case ofNP1 andBX) (Fig. 5), indicates that the encapsulated soilshears off relatively easily in the case of the NP 2 grid cells. Theresistance of the geocell encapsulated soil against footing pene-tration is attributed to two factors: (1) through mobilization of thefrictional resistance, at reinforcement-soil interface; and (2) throughsoil-to-soil friction in the soil mass passing through the apertureopenings in the geocell walls, which can be considered as imaginarysoil beams interlocking through the aperture openings in the geocellwalls. Because of the soil-reinforcement interface, friction is lowerthan the soil-soil interface friction (Juran et al. 1988); with a reducedsize of the openings in the geocell walls, the overall resistanceagainst punching down the encapsulated soil mass gets reduced,leading to failure of the foundation at a much lower settlement. Thegeocells with solid walls would underperform compared with theones having perforations on their walls. Therefore, geocells madeof geogrids would provide better performance compared withthe solid-wall geocells made up of polyethylene sheets, which aretypically called geowebs. Indeed, Adams andCollin (1997), throughlarge-scale model load tests, have observed that the bearing capacityof a geoweb-reinforced sand foundation was even lower than thatof a planar geogrid reinforced system.

The typical variation of strain in the horizontal direction along thewidth of the geocell mattress at its midheight, for geocells made of

Fig. 8. Bearing pressure improvement factor-aperture opening size of the geogrid used to make a geocell mattress: responses at different settlementlevels of footing

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / SEPTEMBER/OCTOBER 2012 / 543

Int. J. Geomech. 2012.12:537-548.

Dow

nloa

ded

from

asc

elib

rary

.org

by

DA

PS L

IBR

AR

Y o

n 08

/31/

13. C

opyr

ight

ASC

E. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved.

different types of geogrids, is shown in Fig. 9. The strain in re-inforcement is largest at the center of footing and decreases rapidlyaway from the footing, indicating that the reinforcing effect of thegeocell mattress is largest beneath the footing. In this zone, thetensile and compressive strength of the geocell reinforcement ishighly mobilized. The extended portions of the geocell mattressbeyond the footingwidth contribute in a secondarymanner by virtueof arresting the potential failure planes and deriving anchorage. Thenumerical values of the strains at the center of the geocell mattress(0, h=2) are given in Table 2.

In Fig. 9 and Table 2, at the region under the footing where themobilized strength of reinforcement is at maximum, the geocell wallmade up of the NP 2 grid has undergone compression, contrary toother two cases where it is in tension. This is because of the inclined

Fig. 9. Pattern of strain variation in geocell reinforcement: responses for different types of geogrids used to make geocells

Table 2. Strain, «h ð%Þ Measured in Transverse Member at Center ofGeocell Mattress (0, h=2) for Different Types of Geogrid Used to MakeGeocell Mattress

Bearing pressure ratio

Type of geogrid used to make geocell

NP1 BX NP2

0.378 0.081 0.028 20.0830.757 0.228 0.090 20.1271.136 0.380 0.172 20.1401.515 0.592 0.286 20.1751.894 0.850 0.410 20.1852.273 1.172 0.587 20.2142.652 1.661 0.863 20.221

544 / INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / SEPTEMBER/OCTOBER 2012

Int. J. Geomech. 2012.12:537-548.

Dow

nloa

ded

from

asc

elib

rary

.org

by

DA

PS L

IBR

AR

Y o

n 08

/31/

13. C

opyr

ight

ASC

E. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved.

Fig. 10. Pattern of variation of normal pressure in the subgrade soil underlying the geocell mattress: responses for different types of geogrids used tomake geocells

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / SEPTEMBER/OCTOBER 2012 / 545

Int. J. Geomech. 2012.12:537-548.

Dow

nloa

ded

from

asc

elib

rary

.org

by

DA

PS L

IBR

AR

Y o

n 08

/31/

13. C

opyr

ight

ASC

E. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved.

orientation of the NP 2 geogrid ribs. Because the geosyntheticsmaterial ismuch stronger in tension than in compression, the strengthof the geocell mattress made up of the NP 2 grid is underutilized.

Fig. 10 shows the normalized vertical pressure (s=q) distributionat the base of the geocell mattress for different types of geogrid usedto form the geocellmattresses. Table 3 presents the normalized valueof the pressures recorded at the midsection below the geocellmattress (i.e., x=B5 0). The geocell made of theNP 2 geogrid showsa better elastic response, as all the graphs (in prefailure stage) arefound to be falling within a narrow range (i.e., almost the samepercentage of load transferred to the base). In Table 3, in the pre-failure stage, the pressure transmitted to the base of the geocellmattress increases with a decrease in the aperture opening of thegeogrid (i.e., NP 1 . BX . NP 2). This may be directly related tothe higher confinement offered to the soil because of the smalleraperture openings of the geogrid used to fabricate the geocell. Sucha confinement induces higher compressive strength to the encap-sulated soil, thereby transmitting the footing pressure more effec-tively to the underlying soil layer. In the case of the NP 2 gridgeocells, the percentage of pressure transmitted to the side pressurecells, placed at a distance of 1:1B from the center of footing, isrelatively higher (Fig. 10). This indicates a better load redistributionbecause of the increased coherence of the geocell mattress.

The reduction in the values of the normalized pressure in the laterstages of loading is attributed to the shearing of soil within thepockets of the geocell mattress. At a higher loading stage there islocal shearing of sand immediately below the footing because thegeocell reinforcement starts sharing a higher proportion of load,thereby bringing forth a reduction in the pressure transmittedthrough the encapsulated soil to the subgrade soil.

The colored sand layers with the uncolored sand mass providea relative medium, where the deformation pattern of the coloredlayers, formed because of loading, indicates the course of rupture inthe foundation soil. A typical rupture pattern is shown in Fig. 11.There is deformation, in the foundation soil, even at a depth morethan about 3B (third line from top). In contrast, the rupture surface inthe unreinforced sand bed is limited to a maximum depth of about0:9–1:1B from the base of the footing (Jumikis 1961; Selig andMcKee 1961; Chummar 1972). These observations establish thatthe geocell reinforcement intercepts the rupture surfaces and forcesthem deeper into the foundation soil. Besides, the top line is de-formed over a length much larger than the footing width, indicatingthat the loading of width B is now spread through the geocellmattress to give an increased width of loading on the underlyingsubgrade soil. Hence, the geocell mattress behaves as a wide slab

that transmits the footing pressure to the underlying soil layer andredistributes it over a wider area, of width (B1DB), where DB[i.e., 5 2ðh1 uÞ3 tana, a is the load spreading angle within thegeocell mattress] is the increase in footing width at the base of geocellmattress (i.e., at depth of u1 h) because of the wide slab effect. DBwas measured from the observed rupture surface in the sand subgradedelineated through the discontinuity in the white-colored sand layers.Using the previous formulation and the measured values of DB, theload-spreading angle in the geocell mattress is calculated as

tana ¼�

DB2ðh þ uÞ

�ð1Þ

The variation of tana with the normalized aperture opening sizeof the geogrid (i.e., da=D50, diameter of the equivalent circular area ofthe aperture opening of the geogrid to the average particle size ofsand) used to fabricate the geocell mattress is illustrated in Fig. 12.The angle of load dispersion is found to decrease with an increase inthe size of the aperture opening of the geogrid. The geocell wall withsmaller openings offers higher confinement to the encapsulated sandthat induces a better coherency and hence higher stiffness in thegeocell-soil composite, thereby redistributing the footing pressuremore effectively onto the subgrade below. This is in agreement withthe observation that pressure transmitted to the side pressure cells(placed at 1:1B distance from the center of the footing) is relativelyhigher for the NP 2 grid geocell. These mechanisms pertain to theprefailure stage when the geocell-soil composite structure is intact.In the postfailure stage, in the region where soil shearing takes place,the geocell and soil almost function as individual entities, hence thecomposite mechanics are no longer valid. At this stage, the geocellreinforcement sustains the footing loading mostly through itsflexural and compressive resistance in the region under the footingand anchorage from the stable soil mass in the region around.

The pressure-settlement responses, with geocell reinforcement,are almost linear until a settlement of about 20% of the footing width(Fig. 6). Besides, the settlement recording dial gauges (Dg2 andDg3)placed at both sides of the footing have shown almost equal readingover this range of settlement (i.e., s=B5 20%), which indicates thatthe differential settlements of the footing are practically negligible.With most of the settlement being elastic and the differential set-tlement being marginal, the design permissible settlement of thefooting on the geocell-reinforced foundation beds, if increased toa relatively large value, is not likely to cause much inconvenience.The findings can also be of used to structure where the permissiblesettlement is relatively high, for example liquid storage tanks,

Table 3. Normal Pressure (s=q) Measured below Geocell Mattress at(x=B5 0) for Different Types of Geogrid Used to Make Geocells

Type of geogrid used to make geocell

Bearing pressureratio

Nonorientedpolymer 1

Biaxialgrid

Nonorientedpolymer 2

0.378 0.362 0.390 0.4420.757 0.383 0.403 0.4381.136 0.399 0.424 0.4481.515 0.420 0.438 0.4551.894 0.430 0.452 0.4702.273 0.418 0.468 0.4322.652 0.277 0.504 0.3293.031 — 0.540 —

3.409 — 0.410 —

3.788 — 0.399 —

4.167 — 0.387 —

Fig. 11. Posttest deformation pattern of subgrade soil underlyinga geocell mattress made of a BX grid

546 / INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / SEPTEMBER/OCTOBER 2012

Int. J. Geomech. 2012.12:537-548.

Dow

nloa

ded

from

asc

elib

rary

.org

by

DA

PS L

IBR

AR

Y o

n 08

/31/

13. C

opyr

ight

ASC

E. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved.

low-cost unpaved roads, large stabilized parking areas, and plat-forms for oil exploration.

Conclusions

This paper, through a series ofmodel load tests, studied the influenceof geocell material on the load-carrying mechanisms of geocell-reinforced foundation beds. Based on the findings, the perfor-mance of geocell-reinforced foundation beds is dependent on thestrength, stiffness, aperture opening size, and orientation of the ribsof the geogrid used to make the geocell mattress. With a decrease inaperture opening size of the geogrid, higher confinement and hencehigher compressive strength is induced to the encapsulated soil,thereby giving rise to better performance improvement. At a rela-tively higher settlement range when the encapsulated soil tendsshearing, the strength of geocells plays a major role. The bearingcapacity improvement because of the geocell reinforcement in-creases with an increase in strength of the geocell material.When thegeogrid ribs, in the geocell wall, are in the horizontal and verticaldirections, they effectively resist against footing penetration throughthe mobilization of vertical compression and horizontal anchorage.Hence, the geocell mattress should be made of the geogrid with asquare/rectangular aperture opening and with its ribs oriented per-pendicular and parallel to the footing.

Large-scale tests carried out byMilligan et al. (1986) and AdamsandCollin (1997) indicate that the generalmechanisms and behaviorobserved in themodel tests are reproduced at a large scale. Therefore,this study provides insight into the basic mechanisms dealing withthe influence of geocell material characteristics, on the bearingpressure versus settlement responses of the geocell-reinforced sandfoundation beds. These results will be of use in providing guide-lines for design and construction of geocell-reinforced sand founda-tions, conducting large-scale model tests, and developing numericalmodels.

Acknowledgments

I would like to thank Professor K. Rajagopal for his valuable com-ments and suggestions.

References

Adams, M. T., and Collin, J. G. (1997). “Large model spread footing loadtests on geosynthetic reinforced soil foundations.” J. Geotech. Geo-environ. Eng., 123(1), 66–72.

ASTM. (2001). “Standard test method for determining tensile propertiesof geogrids by single or multi-rib tensile method.” D6637, WestConshohocken, PA.

Bathurst, R. J., and Jarrett, P. M. (1989). “Large-scale model tests of geo-composite mattresses over peat subgrades.” Transp. Res. Rec. No. 1188,Transportation Research Board, Washington, DC.

Bathurst, R. J., and Rajagopal, K. (1993). “Large-scale triaxial compressiontesting of geocell- reinforced granular soils.” Geotech. Test. J., 16(3),296–303.

Binquet, J., andLee,K. L. (1975). “Bearing capacity tests on reinforced earthslabs.” J. Geotech. Eng. Div., 101(12), 1241–1255.

Bush, D. I., Jenner, C. G., and Bassett, R. H. (1990). “The design andconstruction of geocell foundation mattress supporting embankmentsover soft ground.” Geotextiles Geomembranes, 9(1), 83–98.

Chummar, A. V. (1972). “Bearing capacity theory from experimentalresults.” J. Soil Mech. Found. Div., 98(12), 1311–1324.

Cowland, J. W., and Wong, S. C. K. (1993). “Performance of a road em-bankment on soft clay supported on a geocell mattress foundation.”Geotextiles Geomembranes, 12(8), 687–705.

Dash, S. K., Krishnaswamy, N. R., and Rajagopal, K. (2001). “Bearingcapacity of strip footings supported on geocell-reinforced sand.” Geo-textiles Geomembranes, 19(4), 235–256.

Dash, S. K., Rajagopal, K., and Krishnaswamy, N. R. (2004). “Performanceof different geosynthetic materials in sand foundations.” Geosynth. Int.,11(1), 35–42.

Dash, S. K., Rajagopal, K., and Krishnaswamy, N. R. (2007). “Behaviour ofgeocell reinforced sand beds under strip loading.” Can. Geotech. J.,44(7), 905–916.

Dash, S. K., Reddy, P. D., and Raghukanth, S. T. G. (2008). “Subgrademodulus of geocell-reinforced sand foundation.” Proc. Institution CivilEng.-Ground Improve., 161(2), 79–87.

Dash, S. K., Sireesh, S., and Sitharam, T. G. (2003). “Model studies oncircular footing supported on geocell reinforced sand underlain by softclay.” Geotextiles Geomembranes, 21(4), 197–219.

Dean, R., and Lothian, E. (1990). “Embankment construction problemsover deep variable soft deposits using a geocell mattress.” Proc. Conf.Perform. Reinforced Soil Struct., British Geotechnical Society, London,443–447.

Fig. 12. Variation of the load-spreading angle with an aperture opening size of the geogrid used to make geocells

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / SEPTEMBER/OCTOBER 2012 / 547

Int. J. Geomech. 2012.12:537-548.

Dow

nloa

ded

from

asc

elib

rary

.org

by

DA

PS L

IBR

AR

Y o

n 08

/31/

13. C

opyr

ight

ASC

E. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved.

Dunnicliff, J. C. (1988). Geotechnical instrumentation for monitoring fieldperformance, Wiley, New York.

El Sawwaf, M., and And Nazer, A. (2005). “Behaviour of circular footingsresting on confined granular soil.” J. Geotech. Geoenviron. Eng., 131(3),359–366.

Forsman, J., Slunga, E., and Lathinen, P. (1998). “Ground and geocellreinforced secondary road over deep peat deposit.” Proc. 6th Int.Conf. Geosynth., International Geosynthetics Society, Jupiter, FL,777–778.

Guido, V. A., Chang, D. K., and Sweeny, M. A. (1986). “Comparison ofgeogrid and geotextile reinforced earth slabs.” Can. Geotech. J., 23(4),435–440.

Hadala, P. F. (1967). “The effect of placement method on the responseof soil stress gauges.” Proc. Int. Symp. Wave Prop. Dyn. Prop.Earth Mater., University of New Mexico Press, Albuquerque, NM,255–263.

Hatami, K., Bathurst, R. J., and Di Pietro, P. (2001). “Static response ofreinforced soil retaining walls with nonuniform reinforcement.” Int. J.Geomech., 1(4), 477–506.

Hendricker, A. T., Fredranelli, K. H., Kavazanjian, E., Jr., and Mckelvy,J. A., III. (1998). “Reinforcement requirements at hazardous wastesite.” Proc. 6th Int. Conf. Geosynth., Vol. 1, International Geo-synthetics Society, Jupiter, FL, 465–468.

Jewell, R. A., and Wroth, C. P. (1987). “Direct shear tests on reinforcedsand.” Geotechnique, 37(1), 53–68.

Jumikis, A. R. (1961). “The shape of rupture surface in sand.” Proc. 5th Int.Conf. Soil Mech. Found. Eng., Vol. 1, Dunod Press, Paris, 693-698.

Juran, I., Guermazi, A., Chen, C. L., and Ider, H. H. (1988). “Modelling andsimulation of load transfer in reinforced soils: Part 1.” Int. J. Numer.Anal. Methods Geomech., 12(2), 141–155.

Khing, K. H., Das, B. M., Puri, V. K., Cook, E. E., and Yen, S. C. (1993).“The bearing capacity of a strip foundation on geogrid reinforced sand.”Geotextiles Geomembranes, 12(4), 351–361.

Krishnaswamy, N. R., Rajagopal, K., andMadhavilatha, G. (2000). “Modelstudies on geocell supported embankments constructed over soft clayfoundation.” Geotech. Test. J., 23(1), 45–54.

Léonard, D., Vanelstraete, A., and Parewyck, S. (2002). “Structural designof flexible pavements using steel netting as base reinforcement.” Int. J.Geomech., 2(3), 291–303.

Madhavi Latha, G., Dash, S. K., and Rajagopal, K. (2009). “Numericalsimulation of the behaviour of geocell reinforced sand in foundations.”Int. J. Geomech., 9(4), 143–152.

Madhavi Latha, G., Rajagopal, K., and Krishnaswamy, N. R. (2006).“Experimental and theoretical investigations on geocell supportedembankments.” Int. J. Geomech., 6(1), 30–35.

Milligan, G. W. E., Fannin, R. J., and Farrar, D. M. (1986) “Model and full-scale tests of granular layers reinforced with a geogrid.” Proc. 3rd Int.Conf. Geotextiles, Vol. 1, International Geosynthetics Society, Jupiter,FL, 61–66.

Mitchell, J. K., Kao, T. C., andKavazanjiam, E., Jr. (1979) “Analysis of gridcell reinforced pavement bases.” Technical Rep. No. GL-79-8, U.S.Army Waterways Experiment Station, Vicksburg, MI.

Paul, J. (1988). “Reinforced soil systems in embankments—Constructionpractices.” Proc. Int. Geotech. Symp. Theory Pract. Earth Reinforce.,Balkema, Rotterdam, Netherlands, 461–466.

Rea, C., and Mitchell, J. K. (1978). “Sand reinforcement using paper gridcells.” ASCE Spring Convention and Exhibit, ASCE, Pittsburgh, 24–28.

Robertson, J., and Gilchrist, A. J. T. (1987). “Design and construction ofa reinforced embankment across soft lakebed deposits.” Proc. Int. Conf.Found. Tunnels, M.C. Forde Engineering Technics Press, Edinburgh,U.K., Vol. 2, 84–92.

Selig, E. T., and McKee, K. E. (1961). “Static and dynamic behaviour ofsmall footings.” J. Soil Mech. Found. Div., 87(6), 29–47.

Shimizu, M., and Inui, T. (1990) “Increase in the bearing capacity of groundwith geotextile wall frame.” Proc. 4th Int. Conf. Geotextiles Geo-membranes Relat. Products, Vol. 1, International Geosynthetics Society,Jupiter, FL, 254.

Sitharam, T. G., Sireesh, S., and Dash, S. K. (2005). “Model studies ofa circular footing supported on geocell-reinforced clay.” Can. Geotech.J., 42(2), 693–703.

Vidal, H. (1969) “The principle of reinforced earth.” Highway ResearchRecord No. N.282, Highway Research Board, Washington, DC.

Yoon, Y.W., Heo, S. B., andKim, K. S. (2008). “Geotechnical performanceof waste tires for soil reinforcement from chamber tests.” GeotextilesGeomembranes, 26(1), 100–107.

Zhou, H., and Wen, X. (2008). “Model studies on geogrid- or geocell-reinforced sand cushion on soft soil.”Geotextiles Geomembranes, 26(3),231–238.

548 / INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / SEPTEMBER/OCTOBER 2012

Int. J. Geomech. 2012.12:537-548.

Dow

nloa

ded

from

asc

elib

rary

.org

by

DA

PS L

IBR

AR

Y o

n 08

/31/

13. C

opyr

ight

ASC

E. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved.