Behavior of Geocell-Reinforced Sand

Under a Vertical Load

Jie Han, Xiaoming Yang, Dov Leshchinsky, and Robert L. Parsons

Geocells have a three-dimensional cellular structure, which can be used As Yuu et al. pointed out, despite the effectiveness of the geocell
to stabilize foundations by increasing bearing capacity and reducing system, the acceptance of geocells for base reinforcement of roads
settlements. However, a considerable gap exists between the applications is limited because of the lack of design methods (1). However, the
and the theories for the mechanisms of geocell-reinforced foundations. An lack of design methods results from limited understanding of the load
experimental and numerical study on the behavior of geocell-reinforced transfer mechanisms, limited methods for quantifying the benefits,
sand under a vertical load is presented. A single geocell was filled with and limited full-scale performance data. Most studies have been
sand and subjected to a vertical load to failure. This test process was experimental and have not investigated the load transfer mechanisms
modeled by using the FLAC3D numerical software to investigate the within the geocell. A three-dimensional numerical method pro-
mechanisms of geocell and sand interactions. Experimental and numer- vides an opportunity for in-depth investigation of the load transfer
ical results both demonstrated that the geocell increased the ultimate mechanisms between the three-dimensional geocell and soil.
bearing capacity and the modulus of the sand. The numerical results In this study, continuum mechanics-based three-dimensional soft-
include the distributions of displacements in the sand and geocell walls ware, FLAC3D, was used to investigate the impact of geocell as
and the distributions of tensile stresses and shear stresses acting on the base reinforcement. Laboratory tests were conducted to calibrate the
geocell walls. The numerical results for geocell-reinforced sand are models used in this numerical study.
compared to those for sand without geocell.

EXPERIMENTAL STUDY
Geocell is a three-dimensional form of geosynthetic (polymer)
material with interconnected cells infilled with compacted soil. Geo- Material Properties
cells have been successfully used worldwide to construct retaining
walls and slopes and to reinforce road bases. The key benefit of geo- Sand
cells is to confine fill inside the three-dimensional polymeric honey-
comb cells and thus reduce the lateral movement of soil particles The infill material used in this study was a poorly graded sand known
and form a stiffened mattress to distribute applied loads to a wider locally as Kansas River sand. The grain size distribution of this sand
area. Yuu et al. provide a detailed review of geocells used as base is shown in Figure 1. The mean grain size of the sand was 2.6 mm,
reinforcement over weak subgrade (1). and the particles were subrounded. The minimum and maximum
As summarized by Yuu et al. (1), experimental studies on geo- void ratios of Kansas River sand were 0.384 and 0.560, respectively.
cells have been conducted in three ways: (a) by using triaxial cells Triaxial tests showed the sand at 70% relative density, having the
(2) or resilient modulus cells (3) to investigate the confinement peak and residual friction angles of 42.0° and 39.5°, respectively.
effect as increasing apparent cohesion or reducing long-term perma- Direct shear tests were done at lower normal stresses to represent
nent plastic deformations; (b) by using laboratory model tests to the test conditions. The direct shear tests produced peak and resid-
investigate the reinforcement effect as increasing bearing capacity ual friction angles of 45.0° and 35.0°, respectively, for this sand at
and reducing settlement under static or dynamic loading (4, 5); and 70% relative density.
(c) by using full-scale trafficking tests to investigate the overall
effect as reducing rut depth and prolonging road life (6). Geocell
Yuu et al. identified the following key influence factors on the
performance of geocell-reinforced bases over weak subgrade: geo- The geocells were provided by PRS Mediterranean Ltd. in Israel.
metric variables (height and width) of geocells, quality of infill soil, The height of the geocells used in this study was 50 mm. The perfo-
subgrade strength, and loading type and location (1). rated geocells consist of high-density polyethylene with a density
0.95 g/cm3 (±1.5%) having a cell length of 210 mm and a cell width
of 250 mm. Single cells with a cell area of 26,200 mm2 were welded
J. Han, X. Yang, and R. L. Parsons, Civil, Environmental, and Architectural Engineer-
together to form a uniform geocell mattress. The cell walls are per-
ing Department, University of Kansas, 2150 Learned Hall, 1530 West 15th Street,
Lawrence, KS 66045. D. Leshchinsky, Department of Civil and Environmental forated with holes 10 mm in diameter. The holes are arranged in hor-
Engineering, University of Delaware, Newark, DE 19716. Corresponding author: izontal rows that are staggered and separated 12 mm relative to the
J. Han, jiehan@ku.edu. hole centers. The total open area is 16% of the cell wall area. The
surfaces of cell walls are textured with a multitude of rhomboidal
Transportation Research Record: Journal of the Transportation Research Board,
No. 2045, Transportation Research Board of the National Academies, Washington,
indentations over the entire strip area at a surface density of 2,200 to
D.C., 2008, pp. 95–101. 3,100 mm2 and a depth of 0.35 to 0.85 mm. The tensile stress–strain
DOI: 10.3141/2045-11 relationships of the cell walls were determined by five tensile tests

95
96 Transportation Research Record 2045

100
90
80
70
60

% Finer
50
40
30
20
10
0
0.1 1 10
Particle size (mm)

FIGURE 1 Grain-size distribution of Kansas River sand.

conducted at a strain rate of 10% per minute and are shown in ness (the box was marked at different depths). The thickness of
Figure 2. The average width of the cell wall samples was 55 mm and each of the two layers after compaction was 50 mm (same as the
the gauge length of the tensile tests was 150 mm. The mean tensile cell height) and 20 mm, respectively; therefore, the total fill thick-
stresses at strains of 2% and 5% were 5.0 and 8.2 kN/m, respec- ness was 70 mm. In the geocell-reinforced cases, the geocell was
tively, and the mean ultimate tensile strength was 9.3 kN/m. The preplaced first and then backfilled with sand. The sand inside the
thickness of the geocell sheet was 1.778 mm. The tensile stiffness geocell was placed simultaneously with that outside the geocell.
of geocell sheet at 2% strain was determined to be 250 kN/m based To avoid direct contact between the geocell and the load plate, the
on the tensile stress–strain relationship. second sand layer serves as a cushion between the geocell and the
plate. As the main purpose of this study was to identify and inves-
tigate the load transfer mechanisms between infill and geocell, no
Test Setup and Procedures subgrade was used in this study.
After the placement of two layers of sand, a plate was placed on
Four plate load tests were conducted in a test box by using a load top of the sand ready for testing under a vertical load. The distance
plate (100 × 90 mm), as shown in Figure 3. The size of this plate from the bottom of the plate to the top of the geocell was 20 mm. The
was selected not to touch the sides of the geocell. Before the test, plate was placed on top of the sand but within the cell area, as shown
the Kansas River sand was compacted to a target density (70% in in Figure 3. The center of the plate coincided with the center of the
this study). All the samples were prepared in a dry condition. The geocell. Each vertical load increment was applied and maintained
soil was placed and compacted in two layers. To control the den- for 10 min, at which time the next load increment was applied until
sity of the sample, the mass of each layer was measured as calcu- the failure of the unreinforced sand or the reinforced sand. To ensure
lated. After each layer of sand was filled into the box, compaction the repeatability of the test results, both unreinforced and reinforced
was applied by tamping until the sand reached the desired thick- sands had two plate load tests each.

10

7
Tension (kN/m)

6
Test 1
5
Test 2
4
Test 3
3 Test 4
Test 5
2

0
0 5 10 15 20 25
Strain (%)

FIGURE 2 Tension–strain relationship of geocell specimens.

Han, Yang, Leshchinsky, and Parsons 97

and two reinforced cases similar to those evaluated experimentally

were modeled and analyzed. The mesh for the reinforced case is pre-
13 mm
Load sented in Figure 5 as an example. For the unreinforced cases, no
Plate geocell was included in the model.
Geocell
100 mm The Kansas River sand was modeled as a linearly elastic perfectly
plastic material with a Mohr–Coulomb failure criterion. The geocell
Sand mm
70 was modeled by using the structural (geogrid) elements provided by
38

the FLAC3D software (7 ). The geogrid elements were used to form

0m

mm
480 the geocell as shown in Figure 5. The 19-mm-thick steel load plate
m

was modeled as an elastic solid material with a large elastic modulus.

The properties of these materials are provided in Table 1.
A very low cohesion (practically insignificant) was used for the
FIGURE 3 Test box and test setup.
Kansas River sand to avoid numerical instability and to simulate the
apparent cohesion under a small normal stress. The friction angle of
the sand was chosen on the basis of the direct shear test under a low
Test Results
normal stress. The moduli of the Kansas River sand in Table 1 were
The test results of the four plate load tests on unreinforced or rein- backcalculated by fitting the numerical results to the experimental
forced sand are presented in Figure 4. It is clearly shown that both the data for the unreinforced and reinforced cases. These modulus val-
unreinforced and reinforced cases had excellent repeatability. Figure 4 ues are relatively small because of the low overburden stresses in the
also shows that the inclusion of geocell increased the bearing capaci- soil box. The tensile stiffness of geocell walls were determined on
ties and the stiffness as evidenced by the initial slopes of the curves. the basis of the tension-strain relationship from the tensile tests. A typ-
The load deformation curve is linear up to a deformation of 1.25 mm. ical interaction coefficient and an interface shear stiffness between
For this deformation the load increased from approximately 70 to geocell and sand were selected (8). The boundary conditions were
115 kPa (an approximately 65% increase) with the inclusion of the fixed in the x, y, and z directions, and the other directions were free.
geocell. The modulus of the base increased similarly. It should be The authors recognized the possible effect of boundary conditions
pointed out that this test was based on a single cell. More benefit on the test results; however, the numerical models used the same
would be expected if the sand had been reinforced with multiple cells. dimensions as those in the experimental tests for consistency.
Because FLAC3D software uses x, y, and z for both global and
local coordinate systems, actual directions are marked in the related
NUMERICAL MODELING figures to avoid confusion.

Models and Material Properties

Numerical Results and Analyses
FLAC3D software, a suitable tool for analyzing three-dimensional
polymeric honeycomb geocells, was used in this study. To investigate The numerical results presented herein are based on a single cell in
the interactions between geocells and infilled sand, two unreinforced the base course, which is different from the geocells used in actual

Applied pressure (kPa)

0 50 100 150 200
0

2
Displacement (mm)

4 Unreinforced #1 (test)
Unreinforced #2 (test)
Reinforced #1 (test)
5 Reinforced #2 (test)
Unreinforced (numerical)
Reinforced (numerical)
6

FIGURE 4 Results of plate load tests.

98 Transportation Research Record 2045

Geocell TABLE 1 Properties of Materials Used in Numerical Analysis

Material E (MPa) ν γ (kN/m3) c′ (kPa) φ′ (°)

a b
Kansas River 3.2 /6.0 0.33 17.3 1 45°
sand
Steel plate 200,000 0.33 20 — —
Geocell (geogrid J = 250kN/m, ci = 0.8, k = 71,000kN/m/m
element)

NOTE: E = elastic modulus, ν = Poisson’s ratio, γ = unit weight, c′ = effective

cohesion, φ′ = effective friction angle, J = tensile stiffness of geocell sheet,
ci = interaction coefficient between geocell sheet and sand, k = interface shear
stiffness between geocell sheet and sand.
a
Elastic modulus of sand without geocell.
b
Elastic modulus of sand with the geocell.
FIGURE 5 Three-dimensional mesh of model.

applications. A group of geocells under vertical loads would have more As shown in Figure 6, the patterns of vertical displacements for
complicated interactions among geocells; however, the understanding the unreinforced and reinforced cases are very similar, that is, com-
of the behavior of a single geocell under a vertical load is helpful for pression under the load plate but heave away from the load plate.
investigating the behavior of group geocells. The study on the behav- However, the magnitude of the vertical displacement for the unre-
ior of group geocells in the base course is underway; therefore, it is inforced sand was much larger than that for the reinforced sand
beyond the scope of this study. under the same vertical load.

Vertical Displacement Horizontal Displacement

As shown in Figure 4, the vertical displacements were significantly Figure 7 presents the contours of horizontal displacements for the
reduced by the inclusion of the geocell. The main benefit of the geo- unreinforced and reinforced cases. The largest horizontal movement
cell is to provide its confinement to the sand. In the current tests, the developed under the edge of the load plate and at the bottom of the
modulus of the sand was increased from 3.2 MPa to 6.0 MPa, appro- soil mass. The comparisons in Figure 7 show that the maximum hor-
priately doubling the modulus. izontal displacement for the unreinforced sand was approximately

Contour of Z-Displacement
Magfac = 0.000e+000 (Unit: m)
Heave
Live mech zones shown
-2.3990e-003 to -2.0000e-003
-2.0000e-003 to -1.5000e-003
-1.5000e-003 to -1.0000e-003
-1.0000e-003 to -5.0000e-004
-5.0000e-004 to 0.0000e+000
0.0000e+000 to 5.0000e-004
Compression 5.0000e-004 to 1.0000e-003
1.0000e-003 to 1.1779e-003
Interval = 5.0e–004
(a)

Contour of Z-Displacement
Magfac = 0.000e+000 (Unit: m)
Live mech zones shown
-1.0346e-003 to -1.0000e-003
-1.0000e-003 to -8.0000e-004
-8.0000e-004 to -6.0000e-004
-6.0000e-004 to -4.0000e-004
-4.0000e-004 to -2.0000e-004
-2.0000e-004 to 0.0000e+000
0.0000e+000 to 2.0000e-004
2.0000e-004 to 4.0000e-004
4.0000e-004 to 5.1409e-004
Interval = 2.0e–004
(b)

FIGURE 6 Vertical displacements at 100 kPa: (a) unreinforced and (b) reinforced.
Han, Yang, Leshchinsky, and Parsons 99

Contour of X-Displacement
Magfac = 0.000e+000 (Unit: m)
Live mech zones shown
-8.3963e-004 to -8.0000e-004
-8.0000e-004 to -6.0000e-004
-6.0000e-004 to -4.0000e-004
-4.0000e-004 to -2.0000e-004
-2.0000e-004 to 0.0000e+000
0.0000e+000 to 2.0000e-004
Movement 2.0000e-004 to 2.3815e-004
Interval = 2.0e–004
(a)

Contour of X-Displacement
Magfac = 0.000e+000 (Unit: m)
Live mech zones shown
-3.3982e-004 to -3.0000e-004
-3.0000e-004 to -2.5000e-004
-2.5000e-004 to -2.0000e-004
-2.0000e-004 to -1.5000e-004
-1.5000e-004 to -1.0000e-004
-1.0000e-004 to 5.0000e-005
-5.0000e-005 to 0.0000e+000
Movement 0.0000e+000 to 5.0000e-005
5.0000e-005 to 1.0000e-004
1.0000e-004 to 1.1409e-004
Interval = 5.0e–005
(b)

FIGURE 7 Horizontal displacements at 100 kPa: (a) unreinforced and (b) reinforced.

2.5 times that for the reinforced sand under the same load. The limited
horizontal displacement resulted from the confinement of the geocell.

Deformation of Geocell

The deformations of the geocell in the sand are presented in Figure 8.

It is clearly shown that the walls of the geocell were pushed out by
the applied pressure. An increase of the applied pressure from 100 kPa
to 150 kPa increased the maximum displacement of the geocell wall
from 0.29 mm to 2.52 mm. In addition, under the high pressure of
150 kPa (close to the ultimate bearing capacity), two corners of the
geocell moved in. It is also shown that the movement close to the
bottom of the geocell was larger than that close to the top. This result Out
may be true for the small height of geocell used in this study. (a)
Figure 9 presents the variation of the maximum displacement of
the geocell walls with the applied pressure. The maximum displace-
ment was nearly proportional to the applied pressure in the initial
portion. After an applied threshold pressure of 120 kPa, the dis- In
placement increased exponentially. This phenomenon is similar to
the change of the vertical displacement of the load plate with the
applied pressure.

Forces in Geocell
In
Figure 10 presents the forces in the geocell in the vertical and hori-
Out
zontal directions. Figure 10a shows that there was compression at
the four locations corresponding to the corners of the load plate, which (b)
were close to the sides of the geocell walls. In addition, there was FIGURE 8 Displacement vectors of geocell walls:
vertical tension at the four welding joints, which corresponded to the (a) 100 kPa, maximum displacement 0.29 mm, and
locations where the heave developed, as shown in Figure 6. (b) 150 kPa, maximum displacement 2.52 mm.
100 Transportation Research Record 2045

Applied pressure (kPa)

0 20 40 60 80 100 120 140 160
0

Maximum displacement of geocell

1

(mm)
2

FIGURE 9 Maximum displacement of geocell walls versus applied pressure.

More important, tension developed in the horizontal direction cell. However, there was a significant increase in the maximum
in the geocell walls, as shown in Figure 10b. High tension devel- tension in the geocell when the applied pressure was higher than
oped at two welding joints, whereas low tension developed at 140 kPa, which is close to the ultimate bearing capacity of the
another two welding joints. It is clearly shown that the high ten- geocell-reinforced sand.
sion was close to the bottom of the cell. This result is consistent
with the distribution of the displacements in Figure 8 and true for
the small height of the geocell used in this study. Figure 10b also Interface Shear Stresses Between Geocell
implies the importance of the welding strength of the joints between and Sand
geocell walls.
The variation of the maximum tension in the geocell with the The interface shear stresses between the geocell and the sand is pre-
applied pressure is presented in Figure 11. In general, an increase sented in Figure 12. It is shown that the high interface shear stresses
of the applied pressure increased the maximum tension in the geo- developed at the locations close to the corners of the load plate. In

x
SEL sres-Nx (Unit: N/m)
Magfac = 0.000e+000
-7.2390e+001 to -6.0000e+001
-6.0000e+001 to -4.0000e+001
Tension Compression -4.0000e+001 to -2.0000e+001
-2.0000e+001 to 0.0000e+000
0.0000e+000 to 2.0000e+001
2.0000e+001 to 4.0000e+001
2.0000e+001 to 6.0000e+001
6.0000e+001 to 7.6475e+001
Interval = 2.0e+001
SurfX = (1.00, 0.00, 0.00)
(a)

SEL sres-Ny (Unit: N/m)

y Magfac = 0.000e+000
2.7706e+001 to 5.0000e+001
5.0000e+001 to 7.5000e+001
7.5000e+001 to 1.0000e+002
Large tension Small tension 1.0000e+002 to 1.2500e+002
1.2500e+002 to 1.5000e+002
1.5000e+002 to 1.7500e+002
1.7500e+002 to 2.0000e+002
2.0000e+002 to 2.2500e+002
2.2500e+002 to 2.5000e+002
2.5000e+002 to 2.6164e+002
Interval = 2.5e+001
(b)

FIGURE 10 Force distributions in geocell walls at 150 kPa: (a) vertical force
and (b) horizontal force.
Han, Yang, Leshchinsky, and Parsons 101

700

Maximum tension in geocell walls (N/m)

600

500

400

300

200

100

0
0 20 40 60 80 100 120 140 160
Applied pressure (kPa)

FIGURE 11 Maximum tension in geocell walls versus applied pressure.

Geogrid Coupling Stress

Magfac = 0.000e+000 (Unit: Pa)
4.4730e+002 to 5.0000e+002
5.0000e+002 to 1.0000e+003
1.0000e+003 to 1.5000e+003
1.5000e+003 to 2.0000e+003
2.0000e+003 to 2.5000e+003
2.5000e+003 to 3.0000e+003
3.0000e+003 to 3.2769e+003
Interval = 5.0e+002

FIGURE 12 Interface shear stress between geocell walls and sand at 100 kPa.

addition, the high interface shear stresses were close to the bottom REFERENCES
of the geocell walls.
1. Yuu, J., J. Han, A. Rosen, R. L. Parsons, and D. Leshchinsky. Technical
Review of Geocell-Reinforced Base Courses over Weak Subgrade. Proc.,
First Pan American Geosynthetics Conference & Exhibition, 2–5 March,
CONCLUSIONS Cancún, Mexico, 2008, pp. 1022–1030.
2. Gourves, R., P. Reffsteck, and J. F. Vignon. Study of Confinement Effect
Experimental and numerical studies both showed that geocells could in Geocells. In Geosynthetics: Applications, Design and Construction,
increase the bearing capacity and elastic modulus of the reinforced 1996, pp. 455–458.
3. Mengelt, M., T. B. Edil, and H. H. Benson. Resilient Modulus and Plastic
sand by providing confinement to the infill material. The experi- Deformation of Soil Confined in a Geocell. Geosynthetics International,
mental results showed the load deformation curve is linear up to a Vol. 13, No. 5, 2006, pp. 195–205.
deformation of 1.25 mm. For this deformation, the load increased 4. Rea, C., and K. Mitchell. Sand Reinforcement Using Paper Grid Cells.
approximately 65% with the inclusion of the geocell. The modulus Proc., Symposium on Earth Reinforcement, ASCE, Pittsburgh, 1978,
pp. 644–663.
of the base increased similarly. Numerical analyses showed that the 5. Chang, T.-T., C.-H. Chang, and S.-W. Pai. Investigation of the Bearing
maximum displacement and the maximum tension within the geocell Capacity and Dynamic-Elastic Behavior of Mechanical Stabilization of
existed close to the bottom of the cell, probably because of the small Sandy Subgrade Using Geocells. Presented at 87th Annual Meeting of the
Transportation Research Board, Washington, D.C., 2008.
height of the geocell used in this study. The position of the load plate 6. Webster, S. L. Investigation of Beach Sand Trafficability Enhancement
affected the distributions of the maximum compression force and the Using Sand-Grid Confinement and Membrane Reinforcement Concepts.
maximum interface shear stresses on the geocell walls. It is important U.S. Army Engineer Waterways Experiment Station, Vicksburg, Miss.,
to have sufficient strength at the welding joints of the geocell. 1979.
7. FLAC3D Version 3.1 User’s Guide. Itasca Consulting Group, Minneapolis,
Minn., 2006.
8. Huang, J., J. Han, and J. G. Collin. Geogrid-Reinforced Pile-Supported Rail-
ACKNOWLEDGMENTS way Embankments: The Three-Dimensional Numerical Analysis. In Trans-
portation Research Record: Journal of the Transportation Research Board,
No. 1936. Transportation Research Board of the National Academies,
This research was funded jointly by the University of Kansas, Trans- Washington, D.C., 2005, pp. 221–229.
portation Research Institute, a U.S. Department of Transportation
grant, and PRS Mediterranean, Inc. The authors appreciate the support. The Geosynthetics Committee sponsored publication of this paper.