You are on page 1of 14

*Corresponding author. Fax: 91-44 235 0509; e-mail: gopal@civil.iitm.ernet.

in
Geotextiles and Geomembranes 17 (1999) 171184
Behaviour of sand conned with single
and multiple geocells
K. Rajagopal*, N.R. Krishnaswamy, G. Madhavi Latha
Geotechnical Engineering Division, Department of Civil Engineering, Indian Institute of Technology Madras,
Chennai 600 036, India
Received 14 August 1998; received in revised form 9 December 1998; accepted 10 December 1998
Abstract
This paper studies the inuence of geocell connement on the strength and stiness behav-
iour of granular soils. A large number of triaxial compression tests were performed on granular
soil encased in single and multiple geocells. The geocells were fabricated by hand using dierent
woven and nonwoven geotextiles and soft mesh to investigate the eect of the stiness of the
geocell on the overall performance of geocellsoil composite. In general, it was observed that
the granular soil develops a large amount of apparent cohesive strength due to the connement
by the geocell. The magnitude of this cohesive strength was observed to be dependent on the
properties of the geosynthetic used to fabricate the geocell. The stiness of the composite was
also found to increase with the provision of geocell reinforcement. The results have shown that
using three interconnected cells in the testing programme is adequate to simulate the perfor-
mance of geocell reinforcement layer consisting of many interconnected cells. A simple meth-
odology has been presented in the paper to estimate the magnitude of the apparent cohesive
strength developed by the granular soil as a function of the geometric and material properties of
the geocell. 1999 Elsevier Science Ltd.
Keywords: Geocell reinforcement; Triaxial compression tests; Apparent cohesion; Geosyn-
thetics; Junction peel strength
1. Introduction
The use of geosynthetics for constructions over soft foundation soils has gained
considerable popularity in the recent past because of the savings in cost and time that
02661144/99/$ see front matter 1999 Elsevier Science Ltd.
PII: S 0 2 6 6 1 1 4 4 ( 9 8 ) 0 0 0 3 4 X
Table 1
Properties of geosynthetics used in the test programme
Wide width Secant modulus
tensile Seam Mass per seam at
Type of geosynthetic strength strength Thickness unit area 5% strain
material (kN/m) (kN/m) (mm) (g/m) (kN/m)
Woven geotextile (white) 65 8 0.6 308 70
Woven geotextile (black) 54.5 7.5 0.7 286 50
Nonwoven geotextile 9 9* 2.5 378 5
Soft mesh 1 1* 1.1 105 0.5
*Failure in parent material (seam was intact).
can be gained by using these soil reinforcement products. For example, geocells, which
are three-dimensional, polymeric, honey-comb like structure of cells interconnected
by joints, have been found to be useful for base reinforcement of embankments, on
steep slopes and in applications where the soil should withstand high tensile stresses.
Very few studies were reported in the literature on the mechanical behaviour of
geocell conned soils. Mitchell et al. (1979) conducted a series of laboratory model
tests on sand lled square shaped paper grid cells and identied the dierent possible
modes of failure in these cells. They found the optimum dimensions for these cells
relative to the size of the loading plate. Bathurst and Karpurapu (1993) reported the
results from large-scale triaxial compression tests on cohesionless soils conned by
a single thin walled exible geocell. Field applications of geocells for construction over
soft clays have been described by Bush et al. (1990), Cowland and Wong (1993) and
Hendricker et al. (1998).
This paper reports the results from a series of triaxial compression tests carried out
on sand conned with single and multiple geocells. The main objective of the testing
program is to determine the improvement in the strength and stiness properties of
soil reinforced with geocell. The tests were carried out with multiple geocells in order
to simulate the real conditions in which a number of cells are interconnected.
2. Materials used for testing
The soil used in the tests was uniformly graded river sand with an eective size (D

)
of 0.3 mm and coecients of uniformity (C
"
) and curvature (C

) of 2.17 and 1.04,


respectively. This soil can be classied as SP according to the Unied Soil Classica-
tion System. The maximum and minimum dry unit weights of the soil are 18.4 and
15.7 kN/m, respectively. All the tests were performed by using pre-weighted quantity
of soil placed in the sample mould in 20 layers with light tamping. This procedure has
produced uniform soil samples with a relative density of 55%.
The testing programme considered four dierent types of geosynthetics for forming
the geocells. These are polypropylene woven geotextiles (white and black in colour),
needle punched nonwoven geotextile (polypropylene) and a soft plaster reinforcement
172 K. Rajagopal et al. /Geotextiles and Geomembranes 17 (1999) 171184
Fig 1. Dierent congurations of cells used in triaxial tests.
mesh made of unoriented polymer. This is referred to as soft mesh in this paper. The
geocells were fabricated by stitching the geotextiles into circular shapes with a height
equal to that of the sand sample, i.e. 200 mm. Several cells were stitched together in the
case of tests with multiple geocells.
The number of stitches and the thread used was the same throughout the investiga-
tion so as to have uniform conditions for all the tests. The tests with a single geocell
were performed using all the above referred geosynthetic materials while the tests with
multiple geocells were performed using only the woven white geotextile.
The properties of the geosynthetics used in this test programme are given in Table 1.
The seam strength referred to in the table corresponds to that of the stitch used to
form the geocells. In the case of nonwoven geotextile and the soft mesh the seam was
found to be much stronger than the parent material it self. In all the cases, the seams
were stitched using a cotton thread to promote failure in the samples at reasonable
compression load levels that can be applied in the test apparatus.
3. Test procedure
The model geocells were fabricated from dierent types of geotextiles and mesh
elements. The geocellsand composites are of 100 mm in diameter and 200 mm in
height. The dierent congurations used in the test programme are shown in Fig. 1.
The photograph in Fig. 2 shows the conguration with four interconnected cells. For
K. Rajagopal et al. /Geotextiles and Geomembranes 17 (1999) 171184 173
Fig 2. Photograph of a triaxial test sample with four interconnected cells.
Table 2
Dierent types of triaxial compression tests performed
Type of reinforcement Congurations of cells Aspect ratio of cells (h/d

)
Unreinforced
Woven geotextile (white) Single 2
Double 4
Three 4.3
Four 4.83
Woven geotextile (black) Single 2
Nonwoven geotextile Single 2
Plaster reinforcement mesh Single 2
the sake of clarity, the photograph was taken with partially lled cells. Table 2 shows
the dierent test congurations in the test programme including the aspect ratios
(h/d

) of individual cells in dierent tests in which h is the height of the sample and
d

is the initial diameter of the cell pocket. Many of these tests were repeated twice to
check the consistency of results.
A triaxial test apparatus that can accommodate 100 mm diameter samples was used
for all the tests. The samples were encapsulated in a 0.2 mm thick rubber membrane.
All the specimens were prepared in dry condition within a split cylinder mould. The
174 K. Rajagopal et al. /Geotextiles and Geomembranes 17 (1999) 171184
Fig 3. Loadstrain behaviour of geotextile samples with seam ( junction peel test).
tests were performed at three dierent conning pressures of 100, 150 and 200 kPa.
The axial load was applied on the sample through a pre-calibrated proving ring at an
axial strain rate of 0.625% per minute.
The tensile loadstrain behaviour of all the geosynthetic materials used in the tests
was determined from wide width tension tests performed according to the guidelines
given in ASTM D4595. The junction peel strength of the geocells was determined as
per the procedure explained by Cancelli et al. (1993) with geotextile specimens having
a horizontal seam at mid length. In all the cases, the specimens were of 200 mm wide
and 100 mm long. The tensile load was applied at a strain rate of 5% per minute. The
load in the geosynthetic was measured continuously during the test using a 10 kN
capacity load cell.
4. Results and discussion
4.1. Failure patterns
The loaddeformation behaviour observed from the junction peel tests on dierent
geosynthetic materials is shown in Figs. 3 and 4. The seamed woven white geotextile
sample had the highest strength and stiness while the soft mesh is extremely soft as
K. Rajagopal et al. /Geotextiles and Geomembranes 17 (1999) 171184 175
Fig 4. Loadstrain behaviour of soft mesh with seam ( junction peel test).
shown in Fig. 4. The nonwoven geotextile was found to be extremely soft with the
peak load occurring at or above 80% strain.
The failure in both the single and multiple geocell cases was observed to be by
bursting of the seams at the mid-height of the samples. The seam strength was much
lower than that of the parent geosynthetic material leading to the failure at the
junctions. In the case of samples with multiple geocells, the bursting has started from
the seams of the outer cells and has slowly propagated towards the inner cells.
Whereas the seams of the outer cells have showed clear rupture, the seams of the inner
cells were damaged to a lesser extent.
The failure pattern of the samples encased in the geocells made of nonwoven
geotextile and mesh was very much similar to that of unreinforced soil samples
because of the excessive stretching of these geocells. Because of the low stiness of
these meshes, the geocells underwent such large lateral expansions that the soil may
not have been conned eectively by these geocells at all.
4.2. Inuence on strength
Some typical stressstrain curves from the tests and the pq diagrams from dierent
triaxial compression tests are shown in Figs. 5 and 6 and Figs. 7 and 8, respectively.
The p and q in these diagrams are the mean normal stress and shear stress which are
176 K. Rajagopal et al. /Geotextiles and Geomembranes 17 (1999) 171184
Fig 5. Stressstrain curves for sand with dierent types of geocells (conning pressure"100 kPa).
Table 3
Shear strength parameters from triaxial tests
Type of reinforcement c (kPa)
Unreinforced 40.5 0
Woven geotextile (white)
Single cell 39.2 98.9
Two cells 39.2 134.9
Three cells 39.0 159.2
Four cells 39.0 169.1
Woven geotextile (black) 40.5 77.1
Nonwoven geotextile 40.9 17.3
Mesh reinforcement 40.5 11.0
dened as (

)/2 and (

)/2, respectively. The

and

are the maximum


axial stress and conning pressure, respectively. The interpreted c properties of the
soil with dierent types of reinforcement, obtained from regression analysis, are
shown in Table 3.
The tests on the unreinforced sand gave a friction angle () of 40.5 and a zero
cohesion intercept (c). Interestingly, the reinforced soil samples have exhibited a
K. Rajagopal et al. /Geotextiles and Geomembranes 17 (1999) 171184 177
Fig 6. Stressstrain curves for sand with dierent congurations of geocells (conning pressure"100 kPa).
similar friction angle while displaying a signicant cohesion intercept. As the dry
sand has no internal cohesion, the cohesion intercept can be interpreted as the
apparent cohesive strength imparted by the connement oered by the geocell
reinforcement.
Comparing the cohesive strengths of the samples encased in single geocell, it is
clear that samples with stier geocells developed higher cohesive strength. The
two samples with nonwoven geotextile and mesh reinforcement did not develop
appreciable apparent cohesive strength due to the low stiness of these geosynthetic
materials.
There is a signicant increase in the value of the apparent cohesion when the
number of cells was increased beyond one. However, when the number of cells was
increased from three to four, there was only a marginal improvement in the perfor-
mance of the samples. The increase in cohesive strength is only marginal in this case.
As the number of cells was increased, more area of the soil is conned by the cell
pockets (Table 3), i.e. connement oered by cells per unit volume of soil increases as
the number of cells increases. However, this increase in unit connement has not
resulted in progressive increase in the apparent cohesion of soil as illustrated in
Fig. 8 and Table 3. This result cannot be attributed to the changes in aspect ratio, as
the change is only marginal (Table 2). This result can only be explained from the
interaction that takes place between the dierent cells. The test results indicate that
178 K. Rajagopal et al. /Geotextiles and Geomembranes 17 (1999) 171184
Fig 7. pq diagrams for sand samples with geocells of dierent materials.
the improvement in the performance due to this interaction is not signicant beyond
three cells. Hence, we may conclude that the strength behaviour of three inter-
connected cells may represent the mechanism of geocells having a large number of
interconnected pockets.
4.3. Inuence on stiness
In addition to the increase in the strength of soil, there was a corresponding increase
in the stiness of the soil, which is indicated by steeper stressstrain curves in
Fig. 5 and 6. Because of the additional conning pressure on the soil due to the
membrane stresses, the peak stresses occurred at larger strains. This is similar to the
unreinforced soils developing peak stress at higher strains at higher conning pres-
sures. As the number of cells were increased the stiness of the soil sample has also
increased. The stressstrain response of samples with three and four geocells was
found to be almost identical, Fig. 6. Hence, we may once again conclude that the use of
three interconnected cells in model tests is adequate to represent the stiness behav-
iour of geocells with many interconnected cells.
K. Rajagopal et al. /Geotextiles and Geomembranes 17 (1999) 171184 179
Fig 8. pq diagram with dierent congurations of geocells.
5. Analysis of cohesive strength due to geocell reinforcement
The apparent cohesive strength induced due to the three-dimensional connement
from geocells was obtained from the vertical intercept of the pq diagrams as
discussed in the previous section. It is interesting to analyse this data to derive
a relation between the apparent cohesive strength and the geocell properties. The
rubber membrane theory proposed by Henkel and Gilbert (1952) was used by
Bathurst and Karpurapu (1993) to analyse the cohesive strength of soil encased in
a single geocell.
The mechanism of the development of cohesion may be understood to be due to the
increased conning stresses generated in the soil due to the membrane stresses in
the walls of geocells. The relation between the induced apparent cohesive strength and
the additional connement stress due to the geocell may be derived with reference to
Fig. 9.
In Fig. 9, the small circle shows the Mohr circle corresponding to that of the
unreinforced soil. Because of the provision of geocell reinforcement, the conning
stress increases from

to

due to which the ultimate normal stress increases


to

from
"
. The Mohr circle corresponding to this state is given by the intermedi-
ate circle in the gure. The same ultimate stress can also be represented with the larger
Mohr circle which has a conning pressure of

and an apparent cohesion of c


'
. The
180 K. Rajagopal et al. /Geotextiles and Geomembranes 17 (1999) 171184
Fig 9. Mohr circles for calculations of apparent cohesion for geocellsoil composite.
relation between the c
'
and

can be obtained by applying the MohrCoulomb


failure theory to the two Mohr circles.
The ultimate stress that can be applied on a soil subjected to a conning pressure of

and having apparent cohesive strength of c


'
is can be written as

"
1#sin
1!sin

#2c
P

1#sin
1!sin
"K

#2c
'
(K

. (1)
If the same soil is considered as an unreinforced granular soil subjected to a conning
pressure of

, the ultimate stress can be written as

"K

). (2)
However, both the ultimate stresses should be equal as illustrated in the gure from
which the following relation can be easily derived,
c
'
"

2
k

. (3)
The additional conning stress on the soil due to the membrane stresses developed in
the walls of the geocell (

) can be estimated from the rubber membrane theory of


Henkel and Gilbert (1952) which was originally developed to correct for the eects of
sti rubber membrane. The derivation is as follows,
If the volume of the soil sample remains constant during the test, the relation
between the diameter of the sample (d), original diameter (d

) and the axial strain (

)
can be derived as follows by comparing the initial and volume after application of
strain,

4
d

"

4
d,
K. Rajagopal et al. /Geotextiles and Geomembranes 17 (1999) 171184 181
therefore
d"
d

(/

"
d

(1!

, (4)
in which

is the initial length and is the length of the sample at an axial strain of

Then the circumferential strain (

) can be calculated as

A
"
d!d

"
d!d

"
1!(1 !

(1 !

. (5)
Then the additional conning pressure due to the membrane stresses can be written as
(Henkel and Gilbert 1952),

"
2M

d
1
(1!

)
"
2M
d


1!(1!

1!


(6)
in which

is the axial strain at failure, d

is the initial diameter of individual cell


pocket, and M is the secant modulus of the membrane of the cell at axial strain of

.
In the present case, M is the secant modulus from the junction peel tests.
The above equations are applicable for soil only when it is completely encased
within the geocell. However, as the number of cells was increased, some part of the soil
remains outside the cells as illustrated in Fig. 1, i.e. some part of the soil will behave as
a reinforced soil whereas some other part will behave like an unreinforced soil. Hence,
correction has to be applied on the cohesive strength computed from Eq. (3) to
account for the reduced area of soil in the geocell. This correction can be applied in
terms of the total cross-sectional area of the soil sample and the area of the soil
encased within the geocell as follows,
c
' '''
"c
' '"''
;(area of soil conned by geocells/total area). (7)
Eqs. (3)(7) were used to estimate the apparent cohesive strength developed by
dierent samples and a comparison of the same is shown in Table 4. The axial strain
corresponding to the peak deviator stress was used in these calculations. The correc-
ted apparent cohesive strengths from Eq. (7) compare reasonably well with those
determined from experiments. The validity of the above equation for aspect ratios in
the range of 2 to about 5 has been established in this paper while its validity for aspect
ratio of unity was established from the results published in Bathurst and Karpurapu
(1993). Hence the above equations may be used for extrapolating the strength
properties of geocell conned soil for a wide range of aspect ratios. The above theory
is illustrated with an example in the following.
Let us consider a geocell with pocket size of 1 m diameter made of a geogrid with
a modulus of 200 kN/m. Let the friction angle of the granular soil lled in this geocell
be 35 and the allowable vertical strain be 5%. The additional conning pressure can
be calculated from Eq. (6) as 10.7 kPa by substituting the relevant parameters. Then
from Eq. (3), the apparent cohesive strength of this soil composite can be obtained as
20 kPa.
182 K. Rajagopal et al. /Geotextiles and Geomembranes 17 (1999) 171184
Table 4
Comparison of experimental results with theoretical calculations
Area of soil
No. of cells within in geocells

(kPa) c
'
(kPa) c
' '''
the membrane (mm) from Eq. (6) from Eq. (3) (kPa) Eq. (7) c
' `
(kPa)
Single
Woven-white 7854 87.7 92.0 92.0 98.9
Woven-black 7854 73.9 80.0 80.0 77.0
Nonwoven 7854 13.3 14.6 14.6 17.4
Woven-white
Two 3927 245.7 257.5 128.8 134.8
Three 5160 264.8 277.5 182.2 159.2
Four 5750 247.2 259.1 189.8 169.1
On the other hand, if the required additional cohesive strength is specied, a suit-
able diameter of the geocell can be arrived at from the above equations. For example,
if the required cohesive strength is 40 kPa for the above problem, the required
diameter of the geocell pockets can be obtained as 0.5 m.
6. Summary and conclusions
This paper has investigated the inuence of geocell connement on the strength and
stiness behaviour of granular soils. The inuence of the stiness of geocell material
was also investigated by conducting tests with geocells made of dierent materials.
The shear strength parameters were determined from the pq diagrams by regression
analyses. The geocell reinforced samples have exhibited signicant apparent cohesive
strength. Both the reinforced and the unreinforced soils have shown almost the same
frictional strength. The following conclusions can be drawn from the results of this
investigation.
1. The geocell reinforcement imparts apparent cohesive strength even to cohesionless
soils.
2. The frictional strength of granular soils is not aected by the geocell reinforcement
i.e. the failure envelopes of both the reinforced and the unreinforced samples are
parallel to each other.
3. The induced apparent cohesive strength depends on the tensile modulus of the
geosynthetic used to form the geocell.
4. The results have clearly shown that using a single geocell in triaxial tests is
not adequate to represent the real behaviour of geocell reinforced soils. For
experimental purposes, a test conguration with atleast three interconnected cells
should be used to simulate the performance of soil encased by many interconnected
cells.
K. Rajagopal et al. /Geotextiles and Geomembranes 17 (1999) 171184 183
5. The membrane correction procedure as proposed by Henkel and Gilbert (1952) can
be applied for the interpretation of the apparent cohesive strength developed by
the geocell reinforced soils.
Acknowledgements
The Department of Science and Technology, New Delhi is gratefully acknowledged
for funding a research project titled Geosynthetics in Reinforced Earth Embankments
(vide sanction No. III 4(36)/90-ET, dt. 24.2.95) at the Department of Civil Engineering,
IIT Madras, under which this work was performed.
Notation
c, shear strength properties of the soil
c
'
apparent cohesion induced to the soil
d

initial diameter of the cell pocket


d diameter of the cell pocket at an axial strain of

circumferential strain
h height of the geocell

initial length of the sample


length of the sample at an axial strain of

M modulus of the geosynthetic from junction peel tests (kN/m)

increase in conning pressure due to membrane forces in cell

major principal stress

minor principal stress (conning pressure)


p mean normal stress, (

)/2
q shear stress, (

)/2
References
Bathurst, R.J., Karpurapu, R. 1993. Large scale triaxial tests on geocell reinforced granular soils. Geotech-
nical Testing Journal 16 (3), 296303.
Bush, D.I., Jenner, C.G., Bassett, R.H., 1990. The design and construction of geocell foundation mattress
supporting embankments over soft ground. Geotextiles and Geomembranes 9, 8398.
Cancelli, A., Rimoldi, P., Montanelli, F. 1993. Index and performance tests for geocells in dierent applica-
tions. In: Geosynthetic Soil Testing Procedures, ASTM STP 1190, S.C. Jonathan Cheng, (Ed.), pp. 6475.
Cowland, J.W., Wong, S.C.K., 1993. Performance of a road embankment on soft clay supported on
a geocell mattress foundation. Geotextiles and Geomembranes 12, 687705.
Henkel, D.J., Gilbert, G.D., The eect of the rubber membrane on the measured triaxial compression
strength of clay samples. Geotechnique 3 (1), 2029.
Hendricker, A.T., Fredianelli, K.H., Kavazanjiam, Jr., E., McKelvey III, J.A., 1998. Reinforcement require-
ments at a hazardous waste site. In: Proc. of 6th Internat. Conf. On Geosynthetics, Atlanta, USA, Vol. 1,
pp. 465468.
Mitchell, J.K., Kao, T.C., Kavazanjiam, Jr. E., 1979. Analysis of grid cell reinforced pavement bases.
Technical Report No. GL-79-8, U.S. Army Engineer Waterways Experiment Station, July 1979.
184 K. Rajagopal et al. /Geotextiles and Geomembranes 17 (1999) 171184

You might also like