You are on page 1of 11

ARTICLE IN PRESS

Geotextiles and Geomembranes 25 (2007) 50–60


www.elsevier.com/locate/geotexmem

Behavior of strip footing on geogrid-reinforced sand


over a soft clay slope
Mostafa A. El Sawwaf
Structural Engineering Department, Faculty of Engineering, Tanta University, Tanta, Egypt
Received 1 March 2006; received in revised form 15 June 2006; accepted 24 June 2006
Available online 1 September 2006

Abstract

The potential benefits of reinforcing a replaced layer of sand constructed on near a slope crest was studied. Model tests were carried
out using model footing of 75 mm width and geogrids. Several parameters including the depth of replaced sand layer and the location of
footing relative to the slope crest were studied. Particular emphasis is paid on the reinforcement configurations including number of
layers, spacing, layer length and depth to ground surface. A series of finite element analyses were performed on a prototype slope using
two-dimensional plane strain model using the computer code Plaxis. The soil was represented by non-linear hardening soil model, which
is an elasto-plastic hyperbolic stress–strain model while reinforcement was represented by elastic elements. A close agreement between the
experimental and numerical results is observed. Test results indicate that the inclusion of geogrid layers in the replaced sand not only
significantly improves the footing performance but also leads to great reduction in the depth of reinforced sand layer required to achieve
the allowable settlement. However, the efficiency of the sand–geogrid system increases with increasing number of geogrid layers and layer
length. Based on the theoretical and experimental results, critical values of the geogrid parameters for maximum reinforcing effects are
established.
r 2006 Elsevier Ltd. All rights reserved.

Keywords: Bearing capacity; Strip footing; Reinforced sand slope; Geogrid reinforcement; Soft clay; Finite element analysis

1. Introduction chemical grouting, using soil reinforcement, or installing


continuous or discrete retaining structures such as walls or
The subject of reinforcing soil beneath footings has piles.
gained considerable attention in the past few years (e.g. Several case studies described the successful use of
Dash et al., 2003; Boushehrian and Hataf, 2003; Ghosh geogrids to reinforce a weak subgrade such as variable soft
et al., 2005; Bera et al., 2005; Patra et al., 2005, 2006). This clay (Samtani and Sonpal, 1989; Tsukada et al., 1993;
paper is interested in the many situations where footings Tensar International, 1995; Alawaji, 2001; British Rail
are constructed on/or adjacent to soft clay sloping surfaces Research, 1998; Otani et al., 1998; Maharaj, 2003; Dash
such as footings for bridge abutments on sloping embank- et al., 2003; Yetimoglu et al., 2005). Tsukada et al. (1993)
ments. In these case, two major problems arise; the low investigated the use of geogrids for roadway foundation
bearing capacity of soft clay and the potential failure of the and reported that settlement response and pressure
slope itself. Therefore, over some years, the subject of distributions were directly related to the thickness and
stabilizing earth slope has become one of the most configuration of the geogrid-reinforced foundation. Ala-
interesting areas for scientific research and several techni- waji (2001) discussed the effects of reinforcing sand pad
ques have been suggested to improve the stability of earth over collapsible soil and reported that successive reduction
slope and hence improve the bearing capacity. Typical in collapse settlement up to 75% was obtained. British Rail
examples include modifying the slope surface geometry, Research (1998) has demonstrated that geogrid inserted in
the ballast where tracks lie over soft ground can help
Tel.: +20 2 106814464. extend maintenance intervals. Otani et al. (1998) and
E-mail address: Mos_sawaf@hotmail.com. Maharaj (2003) studied the behavior of strip footing

0266-1144/$ - see front matter r 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.geotexmem.2006.06.001
ARTICLE IN PRESS
M.A. El Sawwaf / Geotextiles and Geomembranes 25 (2007) 50–60 51

constructed on reinforced clay. Settlement was found to be 1-Manual winch


reduced with the increase in reinforcement size, stiffness 2-Raining box
3-Hydraulic jack
and number of layers. The load carrying capacity of a 2
4-Proving ring
footing has been found to increase more on soil in which 5-Model footing
reinforcements are provided at closer spacing. Dash et al. 6-Dial gauges
7-Geogrid reinforcement
(2003) performed model tests in the laboratory to study the 8-Sand
response of reinforcing granular fill overlying soft clay beds 9-Soft clay
and showed that substantial improvements in the load 10-Test tank 3
11-Loading frame
carrying capacity and reduction in surface heaving of the

225.0
4
1
foundation bed were obtained. Yetimoglu et al. (2005)
6 5
performed Laboratory California Bearing Ratio on sand

800
50.0cm
fills reinforced with randomly distributed discrete fibers 7
8
overlying soft clay and reported that adding fiber inclu-
sions in sand fill resulted in an appreciable increase in the
peak piston load. Although, these investigations have 60 9 60
demonstrated that both the ultimate bearing capacity and
the settlement characteristics of the foundation can be
significantly improved using soil reinforcement, it should 10
11
be mentioned that the aim of this reinforcement is not to
eliminate the settlement, as this is not possible, but rather Lab ground
limiting the settlement to values that could be tolerated by 10.0
Dimensions are in centimeters
10.0
the structure. 125.0
A conventional solution to improve the bearing capacity
of soft clay slope is to remove part of the existing weak soil Fig. 1. Schematic view of the experimental apparatus.
and to replace it by granular soil (partial replacement). The
depth of replaced fill depends on the required bearing These columns are firmly fixed in two horizontal steel
capacity and the allowable settlement. Sometimes this beams, which are firmly clamped in the laboratory ground
technique leads to great heights of soil replacement and using four pins. The glass side allows the sample to be seen
hence excessive cost. As a cost-effective alternative to this during preparation and sand particle deformations to be
solution, geosynthetic reinforcement can be provided in the observed during testing. The tank box was built sufficiently
sand fill layer. Although, several studies reported the rigid to maintain plane strain conditions by minimizing the
behavior of footing constructed on stabilized sandy slope out of plane displacement. To ensure the rigidity of the
(Huang et al., 1994; Yoo, 2001; El Sawwaf, 2005), the tank, the back wall of the tank was braced on the outer
behavior of such footing on a reinforced soil slopes surface with two steel beams fitted horizontally at equal
constructed on soft or yielding foundations has not been spacing. The inside walls of the tank are polished smooth
investigated. Many questions still remain as to the overall to reduce friction with the sand as much as possible by
behavior of the footing and slope and vertical and attaching fiber glass onto the inside walls.
horizontal deformations of soils. Therefore, the object of The loading system consists of a hand-operated hydrau-
this paper is to address the aspect of both the bearing lic jack and pre-calibrated load ring. Since the sand raining
capacity improvement (BCI) and settlement reduction technique was used to deposit the sand inside the tank, the
through using both soil replacement and optimum soil beam was designed to swing about one end. Therefore, the
reinforcement. The aim is to study the relationships beam can be swung out during sand deposition and
between the footing response and the variable parameters returned back to the original loading position, when sand
including replaced sand depth, relative density of sand, set up is completed. The sand-raining box was made from
location of the footing relative to slope crest and wood and is 0.85  0.48 m in plan and 0.10 m depth. The
geosynthetic configurations. sand particles were rained from the box through a square
grid of holes (4 mm diameter and 20 mm spacing) in the
2. Laboratory model tests base plate. The height of sand raining, measured from the
bottom of the box to sand surface in the tank, is adjusted
2.1. Model box up or down by using a manual winch.

The main elements of the laboratory apparatus are a 2.2. Model footing
tank, a horizontal steel beam over the tank, and a sand-
raining box. The test box, having inside dimensions of A model strip footing made of steel with a hole at its top
1.00  0.50 m in plan and 0.5 m in depth is made from steel center to accommodate a ball bearing was used. The
with the front wall made of 20 mm thickness glass and is footing was 498 mm in length, 75 mm in width and 20 mm
supported directly on two steel columns as shown in Fig. 1. in thickness. The footing was positioned on the sand bed
ARTICLE IN PRESS
52 M.A. El Sawwaf / Geotextiles and Geomembranes 25 (2007) 50–60

with the length of the footing running the full width of the size distribution was determined using the dry sieving
tank. The length of the footing was made almost equal to method and the results are shown in Fig. 2. The effective
the width of the tank in order to maintain plane strain size (D10), the mean particle size (D50), uniformity
conditions. The two ends of the footing plate were polished coefficient (Cu), and coefficient of curvature (Cc) for the
smooth to minimize the end friction effects. A rough base sand were 0.15 mm, 0.50 mm, 4.07 and 0.77, respectively.
condition was achieved by fixing a thin layer of sand onto Sand beds were placed in 25 mm thick layers by a raining
the base of the model footing with epoxy glue. The load is technique in which sand is allowed to rain through air at a
transferred to the footing through a bearing ball as shown controlled discharge rate and height of fall of 55 cm to give
in Fig. 1. Such an arrangement produced a hinge, which uniform densities. The relative density achieved during the
allowed the footing to rotate freely as it approached failure tests was monitored by collecting samples in small cans of
and eliminated any potential moment transfer from the known volume placed at three different locations in the test
loading fixture. tank. The raining technique adopted in this study provided
a uniform relative density of approximately 80% with a
2.3. Test material unit weight of 19.10 kN/m3. No particle segregation was
noticed during raining and uniformity tests showed that the
2.3.1. Soft clay obtained relative densities from the three samples did not
The cohesive soil used in the model tests was locally depend on the location of the mold. A series of direct shear
available. The soil was identified as CL according to unified tests were performed on specimens prepared by dry
classification system. A series of undrained triaxial tamping at the same relative density of the sand and the
compression tests on cylindrical sample of clay with estimated internal friction angle was 431.
different confining pressures were performed along with
the consistency tests. For the model tests, soil particles 2.3.3. Geogrid reinforcement
passing through B.S. sieve No. 36 was kept in an oven for Tenax TT Samp with peak tensile strength of 45 kN/m
24 h. A specific quantity of water was added to the soil to was used as reinforcing material for the model tests. These
achieve a 32% water content. The soil was thoroughly geogrids, were manufactured by extruding and mono-
mixed and placed by hand into the model box to the directional drawing of high-density polyethylene (HDPE)
required geometry of the slope. Then the soil was grids. Typical physical and technical properties of the grids
consolidated by adding vertical pressure on the top surface were obtained from manufacturer’s data sheet and are
of clay. Finally, soil was unloaded for 24 h before given in Table 2.
depositing sand and geogrid reinforcement and carrying
out the test. A series of direct shear tests on clay specimens 2.4. The experimental setup and test program
with different normal stresses gave an internal friction
angle of 51. The properties of this soil are given in Table 1. An experimental program was carried out to investigate
the partial replacement of soft clay slope and to evaluate
2.3.2. Sand the effects of reinforcing the replaced pad of sand on the
The sand used in this research is medium to coarse, bearing capacity of a strip footing adjacent to the slope
washed, dried and sorted by particle size. It is composed of crest. A 425 mm in height soil model slopes samples were
rounded to sub-rounded particles. The specific gravity of constructed in layers with the bed level and slope observed
the soil particles was determined by the gas jar method. through the front glass wall. The soil was set up to form a
Three tests were carried out producing an average value of slope of 3 (H): 2 (V). Initially beds of soft clay layer were
2.654. The maximum and the minimum dry unit weights of
the sand were found to be 19.95 and 16.34 kN/m3 and the 100
corresponding values of the minimum and the maximum 90
void ratios are 0.305 and 0.593, respectively. The particle 80
70
percent finer, %

Table 1 60
Properties of cohesive soil
50
Liquid limit (%) 39 40
Plastic limit (%) 21
30
Shrinkage limit (%) 13
Plasticity index (%) 18 20
Optimum moisture content (%) 17 10
Average moisture content during testing (%) 31
0
Consistency index 0.44
0.01 0.1 1 10 100
Dry unit weight of soil during testing (kN/m3) 15.84
Undrained cohesion (kN/m2) 25 Grain size, mm
Angle of internal friction (deg.) 5
Fig. 2. Grain size distribution of the sand.
ARTICLE IN PRESS
M.A. El Sawwaf / Geotextiles and Geomembranes 25 (2007) 50–60 53

placed followed by depositing of sand by raining technique. parameter while the other variables were kept constant.
In the reinforced tests, layers of geogrid were placed in the The varied conditions include the depth of replaced sand
sand at predetermined depths during preparing the ground layer (d), horizontal distance between footing and the slope
slope. The inner faces of the tank were marked at 25 mm crest (b), number of geogrid layers (N), length of geogrid
intervals to facilitate accurate preparation of the sand bed layers (L), and vertical spacing between layers (x). Table 3
in layers. On reaching the reinforcement level, a geogrid and Fig. 3 summarize all the tests programs with varied
layer was placed and a layer of sand is rained and so on. parameters used. Several tests were repeated at least twice
One to four geogrid layers with width equal to the full to verify the repeatability and the consistency of the test
width of the tank were placed with different lengths and data. The same pattern of load-settlement relationship with
spacing. The preparation of the sand bed above the geogrid difference of the ultimate load values less than 1.5% were
cell was continued in layers up to the level required for a obtained. The difference was considered to be small and
particular depth of embedment. Great care was given to neglected.
level the slope face using special rulers so that the relative
density of the top surface was not affected. The footing was 3. Prototype study
placed at desired position and the load was applied
incrementally by the hydraulic jack until reaching failure. 3.1. Finite element analysis
Each load increment was maintained constant until the
footing settlement had stabilized. This settlement was A series of two-dimensional finite element analyses
measured using two 0.001 mm accuracy dial gauges, placed (FEA) on a prototype footing-slope system was performed
on opposite sides across the center of the footing. in order to verify the laboratory model tests results and
Many model tests in two different test programs were understand the deformations trends within the soil mass.
carried out. Initially, the response of the model footing The analysis was performed using the finite element
supported on the unreinforced replaced sand layer of program Plaxis software package (professional version 8,
different heights overlying soft clay slope was determined. Bringkgreve and Vermeer, 1998). Plaxis is capable of
Then, six series of tests were performed to study the handling a wide range of geotechnical problems such as
inclusion effect of the geogrid layers on the footing deep excavations, tunnels, and earth structures such as
behavior. Tests were conducted to find out the best retaining walls and slopes. Prototype slopes were assumed
location, length and configuration of the geogrid layers to rest on a yielding foundation and to extend laterally to a
that give the maximum improvement in footing response. distance of 1.5 times the slope height (H) from the toe of
Each series was carried out to study the effect of one
b B

Table 2
Engineering properties of geogrid u
x
x d
Structure Mono-oriented geogrids L
Aperture shape Oval apertures
Aperture size (mm  mm) (13/20)  220
Weight (g/m2) 300.00 Hs
Polymer type HDPE 0.5B Xs Soft clay
Tensile strength at 2% strain (kN/m) 11
Tensile strength at 5% strain (kN/m) 25
0.5B
Peak tensile strength (kN/m) 45
Yield point elongation (%) 11.5
Long term design strength (N/m) 21.2 Fig. 3. Geometric parameters of reinforced sand slope overlying soft clay
model.

Table 3
Model tests program

Series Constant parameters Variable parameters

1 Tests on non-reinforced slope, b/B ¼ 0 d/B ¼ 0.5, 1, 1.5, 2, 3


2 b/B ¼ 0, d/B ¼ 1.5, N ¼ 1, L/B ¼ 5 u/B ¼ 0.25, 0.5, 0.75, 1.0, 1.25
3 b/B ¼ 0, d/B ¼ 1.5, N ¼ 2, u/B ¼ 0.25, L/B ¼ 5 x/B ¼ 0.25, 0.5, 0.75, 1.0
4 b/B ¼ 0, d/B ¼ 1.5, N ¼ 3, u/B ¼ 0.25, x/B ¼ 0.4 L/B ¼ 2, 3, 4, 5, 6
5 b/B ¼ 0, d/B ¼ 1.5, u/B ¼ 0.25, x/B ¼ 0.4, L/B ¼ 6 N ¼ 1, 2, 3, 4
6 Tests on non-reinforced slope, d/B ¼ 1.5 b/B ¼ 0, 1, 2, 3,4
7 d/B ¼ 1.5, N ¼ 3, u/B ¼ 0.25, x/B ¼ 0.4, L/B ¼ 5 b/B ¼ 0, 1, 2, 3,4

Note: See Fig. 3 for definition of the variable.


Footing width (B) is always constant ¼ 75 mm.
ARTICLE IN PRESS
54 M.A. El Sawwaf / Geotextiles and Geomembranes 25 (2007) 50–60

the slope. The geometry of the prototype footing-slope place. A prescribed footing load was then applied in
system was assumed to be 10 times the laboratory model increments (load control method) accompanied by iterative
(the footing width B ¼ 0.75 m and thickness, t ¼ 0.2 m, soil analysis up to failure. The modeled boundary conditions
height (sand layer and soft clay layer) ¼ 4.25 m. The same were assumed such that the vertical boundaries are free
inclination of model test slopes, 3 (H): 2 (V), and the vertically and constrained horizontally while the bottom
material of geogrid, sand and soft clay were used in the horizontal boundary is fully fixed. The software allows the
prototype study. The software allows the automatic automatic generation of six node triangle plane strain
generation of six or fifteen node triangle plane strain elements for the soil, and three node beam elements for the
elements for the soil, and three or five node beam elements footing and the geogrid. The number of element used in
for the footing while three or five node elastic elements reinforced tests are 213 element while in unreinforced tests
were used for the geotextile elements. the number is 160. The analyzed prototype slope geometry,
generated mesh, and the boundary conditions are shown in
3.2. Finite element modeling Fig. 4.
An internal angle of friction and secant Young’s
The non-linear behavior of sand was modeled using modulus E ref50 representing dense sand conditions derived
hardening soil model, which is an elastoplastic hyperbolic from a series of drained triaxial compression tests were
stress–strain model, formulated in the framework of used for the sand. A value of 25 kN/m2 to the undrained
friction hardening plasticity. The foundation was treated cohesion (c) for the soft clay derived from undrained
as elastic beam elements based on Mindlin’s beam theory triaxial compression tests was used. The hyperbolic
with significant flexural rigidity (EI) and normal stiffness parameters for the sand and clay were taken from database
(EA). A basic feature of the hyperbolic model is the stress provided by the software manual as shown in Table 4.
dependency of soil stiffness. The interaction between the
geogrid and soil is modeled at both sides by means of 4. Results and discussion
interface elements, which allow for the specification of a
reduced wall friction compared to the friction of the soil. A total of 32 model tests were carried out on model
The limiting state of stress are described by means of the plane strain footing supported on sand pads overlying soft
secant Young’s modulus E ref 50 , tangent stiffness modulus clay ground slope. The effect of geogrid parameters on the

for primary compression E ref oed , Poisson’s ratio (u), ultimate load and displacement were obtained and
effective cohesion (c), angle of internal friction (f), angle discussed. An additional numerical study on the effect of
of dilatancy (c), failure ratio (Rf) and interface reduction reinforcing the sand pad on the behavior of a prototype
factor (Rint). footing was carried out using the finite element model.
A refined mesh was adopted to minimize the effect of
mesh dependency on the finite element modeling of cases 4.1. Bearing capacity behavior
involving changes in the number, length, and the location
of geogrid layers. In the finite element modeling, as the The BCI of the footing on the reinforced replaced sand is
slope surface is not horizontal, the initial stress condition represented using a non-dimensional factor, called BCI
of the slope was established first by applying the gravity factor. This factor is defined as the ratio of the footing
force due to soil in steps with the geogrid reinforcements in ultimate pressure with the slope reinforced (qu reinforced) to

A A
The Footing

The geogrid
The interface

Fig. 4. Prototype slope geometry, generated mesh, and boundary conditions.


ARTICLE IN PRESS
M.A. El Sawwaf / Geotextiles and Geomembranes 25 (2007) 50–60 55

Table 4
Hardening soil-footing model parameters used in the finite element analysis

Parameter Dense sand Soft clay Footing Geogrid



Primary loading stiffness E 50 (kN/m2)
ref 40 000 1000 — —
Cohesion (c) (kN/m2) 0.00 25 — —
Friction angle (f) 40.00 5 — —
Dilatancy angle (c) 10.00 0.0 — —
Soil unit weight (g) (kN/m3) 18.90 15.84 — —
Poisson’s ratio (u) 0.3 0.33 — —
Failure ratio (Rf) 0.90 0.9 — —
Interface reduction factor (Rint) 0.80 0.5 — —
EA of the footing (kN/m) — — 5 000 000 —
EI of the footing (kN m2/m) — — 8500 —
EA of the geogrid (kN/m) — — — 2000

the footing ultimate pressure in tests without slope q, kPa


reinforcement (qu). The footing settlement (S) is also 0 5 10 15 20 25 30
0
expressed in non-dimensional form in terms of the footing b/B = 0
width (B) as the ratio (S/B, %). The ultimate bearing
2 d/B = 0.5
capacities for the both the model and prototype footing are d/B = 1.0
determined from the load–displacement curves as the d/B = 1.5
pronounced peaks, after which the footing collapses and 4
d/B = 2.0
the load decreases. In curves which did not exhibit a d/B = 3.0
definite failure point, the ultimate load is taken as the point 6
S/B, %

at which the slope of the load-settlement curve first reach


zero or steady minimum value (Vesic, 1973). The measured 8
bearing load of the model footing when located on non-
reinforced and reinforced replaced sand layer of depth 10
1.5B with horizontal ground surface and overlying soft
clay are 24 and 38N, respectively. The corresponding 12
theoretical bearing loads obtained from the FEA are 210
and 320N. 14
Typical variations of bearing capacity pressure (q) with
settlement ratios (S/B) for small-scale model and prototype Fig. 5. Variations of q with S/B for model slope with different depths of
replaced sand.
footing for different depths (d/B) of unreinforced replaced
sand are shown in Figs. 5 and 6. In this series, all the tests
were performed on the footing placed at slope crest while
the depth of the sand layer was varied. The figures clearly q, kPa
show that partial replacement of soft clay much improves 0 20 40 60 80 100 120 140
both the initial stiffness (initial slope of the load-settlement 0
b/B = 0.0
curves) and the bearing load at the same settlement level. d/B = 0.5
Also, the figure demonstrates that as the thickness of sand 1 d/B = 1.0
layer increases, the mode of failure changes from punching d/B = 1.5
d/B = 2.0
shear failure (at d/B ¼ 0.5) to the general shear failure in 2 d/B = 3.0
which a pronounced peak can be seen, after which the load
comes down. Both figures confirm the significant increase
S/B, %

in the bearing capacity of the footing with the increase of 3


replaced sand thickness. For example, partial replacement
of soft clay with a layer of sand of thickness equal to 3B 4
cause an improvement in measured bearing capacity as
high as four times more than that when the sand is of 5
thickness ¼ 0.5B. The measured and calculated ultimate
loads for footing supported on both reinforced and non-
reinforced slopes for the different studied parameters are 6
given in Tables 5–7. These results are discussed in the Fig. 6. Variations of q with S/B for prototype slope with different depths
following sections. of replaced sand.
ARTICLE IN PRESS
56 M.A. El Sawwaf / Geotextiles and Geomembranes 25 (2007) 50–60

Table 5
Results of footings located near to reinforced slopes (series 2 and 3)

Test results u/B x/B

0.25 0.5 0.75 1.0 1.25 0.25 0.5 0.75 1.0

q (kPa) model 16.75 17.5 17.25 16.5 16.45 20.57 21.9 19.5 18.5
q (kPa) FEA 128 130 129 126 125 154 158 149 141

Table 6
Results of footings located near to reinforced slopes (series 4)

Test results L/B

0 2 3 4 5 6

q (kPa) model 12 16 19 21.5 23 23.40


q (kPa) FEA 93 120 144 162 171 173

Table 7
Results of footings located at different locations (series 6 and 7)

Test results Non-reinforced Reinforced

b/B b/B

0 1 2 3 4 0 1 2 3 4

q (kPa) model 12.0 14.5 18 21 22 23.0 26 30.5 35 36.5


q (kPa) FEA 93 130 168 197 204 171 221 263 301 308

4.1.1. Effect of number of geogrid layers BCI


Two series of studies were carried out in order to study 0 0.5 1 1.5 2 2.5
the effect of varying the number of geogrid layers on the 0
footing-slope performance. In these series, a depth of
2
replaced sand of 1.5B along with geogrid length, location,
and spacing, was kept constant but the number of geogrid 4
layers was varied. Typical variations of BCI measured from
model tests and q obtained from numerical analysis against 6
settlement ratios (S/B) for a footing located at the slope
S/B, %

crest are shown in Figs. 7 and 8, respectively. Also, 8


b/B = 0,
included in the figure for comparison is the footing
d/B=1.5, L/B=5.0
behavior when placed on the replaced sand without 10 u/B=0.25, x/B=0.4
reinforcement. The figures clearly show that the general No reinforcement
12
trends of FEA shown in Fig. 8 agree fairly well with those N=1
of the model slope in Fig. 7. For the same displacement N=2
14
ratio, the figure demonstrates that the inclusion of geogrid N=3
layers resulted in an increase in the load capacity of N=4
16
the model footing. Also, for the same footing load, the
settlement ratio decrease significantly with increasing the Fig. 7. Variations of BCI with S/B for model slope for different N.
number of geogrid layers.
This increase in footing ultimate load can be attributed
to reinforcement mechanism which derived from the the spreading of slope and lateral deformations of sand
passive earth resistance, interlocking in front of the particles. The mobilized tension in the reinforcement
transverse members, and adhesion between the long- enables the geogrid to resist the imposed horizontal shear
itudinal/transverse geogrid members and the sand. The stresses built up in the soil mass beneath the loaded area
mobilized passive earth resistance of soil column confined and transfer them to adjacent stable layers of soils leading
in the geogrid apertures along with the interlocking limit to a wider and deeper failure zone. Therefore, sand
ARTICLE IN PRESS
M.A. El Sawwaf / Geotextiles and Geomembranes 25 (2007) 50–60 57

q, kPa there are an optimum number of reinforcement layers after


0 50 100 150 200 250 which the gain in bearing capacity is not significant. This is
0
consistent with previous studies of strip or square plate
over entirely dry sand (Omar et al., 1993; Das et al., 1994)
2 which demonstrated that there are an optimum number of
geogrid layers after which the BCI remain constant. Also,
test results shown in Fig. 5 and Table 5 demonstrate that
4 reinforcing a replaced sand layer of thickness 1.5B using
S/B, %

three layers of geogrid could bring out an improvement in


bearing capacity greater than that obtained using replaced
6
b/B = 0.0 sand of thickness 3B without reinforcement.
No reinforcement
N=1 4.1.2. Effect of geogrid layer length
8
N=2
N=3 In order to determine how far to extend the geogrid
N=4 layers into the soil mass to provide an adequate anchorage
10 length for each geogrid layer, five tests were carried out to
Fig. 8. Variations of q with S/B for prototype slope for different N. study the effect of varying the layer length on the footing
behavior. Fig. 10 shows the variations of BCI with the
geogrid length for both prototype and model slope-footing
2.25
system. The BCI increases with increasing geogrid length.
This increase in the ultimate bearing capacity with
2 increasing layer length is significant until a value of
(L/B ¼ 5) beyond which further increase in layer length
1.75
of geogrid does not show significant contribution in
increasing the ultimate load of the footing. This is
BCI

consistent with the recommended lengths of geogrid


1.5 reported by Yoo (2001) who showed that BCI of reinforced
d /B = 1.5, u/B=0.25
x /B=0.4, L /B=6.0 sandy slop loaded with strip footing 1.5B away of the slope
1.25 crest remains constant for length ratios 45.5.
Model
FEA
This behavior illustrates that sufficient anchorage
lengths must be provided to maximize the reinforcing
1 effect through full mobilization of pullout capacity of the
0 1 2 3 4 5
N
reinforcements. With short layers of geogrid, the anchorage
length of geogrid in sand is insufficient and the mobilized
Fig. 9. Variations of BCI with number of geogrid layer, N. lateral resistance by passive resistance, interlocking and
friction in the stable mass of soil is less than the transferred
pad–geogrid interaction not only result in increasing the horizontal shear stresses and the geogrid layers will move
bearing capacity due to developed longer failure surface down with the soil movement underneath the footing.
but also results in widening the contact area between sand
and soft clay. As a result, the developed acting net stress 2.25
due to footing load decreased leading to decreasing the
consolidation settlement of soft clay.
Fig. 9 presents comparisons of the variations of the 2
calculated and measured BCI for a footing located at the
slope crest for varying values of N. Although, the BCI
1.75
obtained from the FEA appears to be smaller than that for
BCI

the model slope, the general trends of the manner in which


BCI varies with the number of geogrid are in good 1.5
agreement with those from the model tests. The BCI d/B = 1.5, u/B = 0.25
increases with the increase of N for the two series. x/B = 0.4, N = 3
However, the rate of increase in BCI decreases with 1.25 FEA
increasing number of geogrid layers until N ¼ 3 after model
which the rate of load improvement become much less. For
practical reasons, no tests were carried out using more than 1
0 1 2 3 4 5 6 7
four geogrid layers due to the limited depth of the replaced
L /B
sand fill. However, the figure demonstrates that for the
thickness of replaced sand and reinforcement conditions, Fig. 10. Variations of BCI with geogrid layer length L/B.
ARTICLE IN PRESS
58 M.A. El Sawwaf / Geotextiles and Geomembranes 25 (2007) 50–60

For longer layers, sufficient anchorage length mobilizes 4.1.4. Effect of depth to top layer
larger lateral resistance than that built up underneath the The effect of depth of the geogrid layer to the ground
footing and therefore with footing settlement the geogrid surface u/B is studied using only one layer of geogrid
will not move down with supporting soil but mobilize placed in dense sand at different depths of ground surface.
greater resistance up to maximum pullout capacity of Five tests were carried out on model footing along with five
geogrid layer after which the system fails. studies on prototype footing using FEA. Fig. 12 shows the
variation of the BCI of the footing against the normalized
4.1.3. Effect of footing location relative to slope crest depth u/B for both model and prototype footing. Both
In order to study the effect of the proximity of a footing graphs clearly show that the BCI initially increases with
to the slope crest (b/B), two series of tests were carried out increasing the depth until it attains a maximum value after
on strip footing resting on replaced sand fill overlying soft which the BCI comes down with increasing the depth of
clay slopes. While the first was carried out on non- geogrid layer. The maximum improvement was obtained at
reinforced replaced sand fill, the second was carried out on depth ratio of u/B ¼ 0.6 for both model and FEA,
3-layer of geogrid-reinforced sand. Fig. 11 shows the respectively. A similar trend was observed for level sand
variation of the BCI against the b/B ratios, for both model ground by Das et al. (1994). Also, the variation of BCI with
and prototype results. The curves clearly show good u/B reported by Selvadurai and Gnanendran (1989) and
agreement in the general trend of behavior of both the Yoo (2001) for reinforced sand slope are similar to that
experimental and numerical analysis. It can be seen that, obtained from the present investigation. It was confirmed
while the bearing capacity load significantly decrease as the that there exist a critical value for u/B at which maximum
footing location moves closer to the slope crest, the effect gain in bearing capacity was obtained. The reported values
of soil reinforcement on the bearing capacity significantly varied between 1.0 and 0.5 according to slope geometry,
increase. Also, the figure clearly shows that maximum soil condition and number of geogrid layers. This can be
benefit of slope geogrid reinforcement is obtained when explained as follows; at shallow depths under the footing,
footing is placed at slope crest. The same trend can be both the vertical and horizontal soil displacements are
observed for both model and prototype slopes. As the greater. Maximum benefits could be obtained when soil
footing location moves away from crest, the rate of reinforcement are placed at these depths where mobilized
decrease in BCI of the footing become less until a value lateral resistances for soil lateral displacements are max-
of about b/B ¼ 2.5 after which the change can be imum. When the depth of geogrid layer increases, both
considered insignificant. This change in bearing capacity lateral and vertical soil displacements in the zone between
of the footing with its location relative to slope crest can be the footing and the geogrid layer increase and hence the
attributed to soil passive resistance from the slope side and bearing capacity decreases.
reinforcement effect. When, the footing is placed far away
of the slope, the passive resistance from the slope side to 4.1.5. Effect of vertical spacing of the geogrid
the failure wedge under the footing increases. Also, using Two series of studies were performed on both model and
geogrid reinforcement decreases lateral displacements and prototype footing located at the slope crest of replaced
results in wider and deeper failure zone as discusses in sand layer overlying soft clay ground slope with all
previous sections, leading to increasing the bearing parameters kept constant but vertical spacing between
capacity load. layers was varied. Two geogrid layers were used with

2
1.5

1.8 1.4

1.6 1.3
BCI

BCI

1.4 1.2 b/B = 0,


N = 3, u/B=0.25
N = 1, d /B = 1.5
x/B=0.4, L/B=5.0 x /B=0.4, L/B=5.0
1.2 model 1.1
Model
FEA FEA
1 1
0 1 2 3 4 5 0 0.25 0.5 0.75 1 1.25 1.5
b/B u/B

Fig. 11. Variations of BCI with footing location b/B. Fig. 12. Variations of BCI with depth of geogrid layer u/B.
ARTICLE IN PRESS
M.A. El Sawwaf / Geotextiles and Geomembranes 25 (2007) 50–60 59

vertical spacing values of x/B ¼ 0.25, 0.5, 0.75, and 1.0 displacement vectors at failure for non-reinforced slope are
were studied. The variation of BCI with normalized layer concentrated underneath the footing toward the slope face
spacing x/B is shown in Fig. 13. The curves clearly show while for the reinforced slope, the displacement vectors are
that there is an optimum value of x/B for which maximum widely distributed underneath the footing for greater width
benefit of the geogrid reinforcements is derived. The BCI and depth than that in the non-reinforced case. It is clear
increases with x/B up to approximately a maximum value that the geogrid layers prevent the soil particles from
of 0.5B after which it decreases. A similar trend was lateral movement toward the slope face and pushes them
reported for entirely sand slope by Yoo (2001), showing a downward for greater depth and hence spreads the footing
critical value of x/B ¼ 0.7. load wider and deeper into the soil, which in turn meant a
longer failure surface and greater bearing capacity.
4.2. Displacement vectors
4.3. Scale effects
Fig. 14 presents the failure pattern and deformed mesh
for a footing placed at the crest of both the non-reinforced The model footing adopted in laboratory study is
and three-geogrid layers reinforced slope, respectively. The reduced to a certain scale while the used sand, clay and
figure clearly shows the tendency of the footing rotation geogrids were the same in the model and the prototype
toward the slope face on reinforced test while in tests on analysis. Therefore, model footing or the soil, may not play
non-reinforced slope, the footing tend to fail by punching the same role as in the prototype and it might cause some
shear failure. Typical plots of the displacement vectors influence on the experimental results which is called scale
obtained from the FEA are also presented. Comparing the effects (Vesic, 1973; Ovesen, 1979). These differences occur
plastic flow between these two cases, it can be observed that primarily because of the differences in stress level between
the model tests and the field tests. Also, it was preferred to
2 use a prototype geogrid of lowest allowable stress
(standard one) rather than to use some sort of hand-made
geotextile (unknown one) for similitude. Therefore, due to
1.75 scale effects and the use of prototype geogrid in the model
tests, the results of model footing may be influenced.
It was illustrated in a previous section that, the main
object of this research is to see the reflection of soil
BCI

1.5
reinforcement in the footing response, and not to have
b/B = 0, precise comparison between the model and prototype
d/B = 1.5, u /B = 0.25
results as this can only be achieved by conducting a
1.25 N=2, L /B=5.0
centrifuge study. Therefore, it was the target to have insight
Model into the significant trends of soil reinforcement on the
FEA
footing response both experimentally and numerically. The
1 general trends of both model and prototype study indicated
0 0.25 0.5 0.75 1 1.25
the benefits can be obtained when using geogrid to
x/B
reinforce replaced sand layer over soft clay slopes on the
Fig. 13. Variations of BCI with normalized geogrid layers spacing x/B. behavior of a footing, and provide a useful basis for further

(A) (B)

Fig. 14. Failure pattern and displacement vectors plot for prototype slope: (A) non-reinforced slope and (B) reinforced slope.
ARTICLE IN PRESS
60 M.A. El Sawwaf / Geotextiles and Geomembranes 25 (2007) 50–60

research using full-scale tests or centrifugal model tests Bera, A.K., Ghosh, A., Ghosh, A., 2005. Regression model for bearing
leading to an increased understanding of the real behavior capacity of a square footing on reinforced pond ash. Geotextiles and
Geomembranes 23 (2), 261–286.
and accurate design in application of soil reinforcement.
Boushehrian, J.H., Hataf, N., 2003. 241 Experimental and numerical
investigation of the bearing capacity of model circular and ring
5. Conclusions footings on reinforced sand. Geotextiles and Geomembranes 21 (4),
241–256.
The bearing capacity behavior of a strip footing resting Bringkgreve, R., Vermeer, P., 1998. PLAXIS-Finite Element Code for Soil
and Rock Analysis. Version 7 Plaxis B.V., The Netherlands.
on replaced sand layer (partially replaced) constructed on a
British Rail Research, 1998. Supporting roll—geogrids provide a solution
soft clay slope was investigated. Also, the effect of to railway track ballast problems on soft and variable subgrades.
inclusion of geogrid reinforcement on the footing response Ground Engineering 31 (3), 24–27.
was studied both experimentally and theoretically. Wide Das, B., Khing, K., Shin, E., Puri, V., Yen, S., 1994. Comparison of
ranges of boundary conditions including the depth of sand bearing capacity of strip foundation on geogrid-reinforced sand and
clay. In: Proceedings of the Eighth International Conference on
layer, footing location, and the geogrid parameters were
Computer Methods and Advances in Geomechanics, Morgantown,
considered. Based on the results from this investigation, the WA, USA, pp. 1331–1336.
following conclusions can be drawn: Dash, S., Sireesh, S., Sitharam, T., 2003. Model studies on circular footing
supported on geocell reinforced sand underlain by soft clay.
(1) Soil improvement of soft clay ground slope by partial Geotextiles and Geomembranes 21 (4), 197–219.
El Sawwaf, M., 2005. Strip footing behavior on pile and sheet pile-
replacement with sand layer significantly increases the
stabilized sand slope. Journal of Geotechnology and Geoenviron-
load bearing capacity of a footing placed on/near to the mental Engineering 131 (6), 705–715.
crest of sloping ground. Ghosh, A., Ghosh, A., Bera, A.K., 2005. Bearing capacity of square
(2) The inclusion of soil reinforcement not only improves footing on pond ash reinforced with jute-geotextile. Geotextiles and
the footing behavior but also leads to significant Geomembranes 23 (2), 144–173.
Huang, C., Tatsuoka, F., Sato, Y., 1994. Failure mechanisms of reinforced
reduction of the depth of replaced sand layer over the
sand slopes loaded with a footing. Soils and Foundations 24 (2), 27–40.
soft clay for the same footing settlement, at the same Maharaj, D., 2003. Nonlinear finite element analysis of strip footing on
load levels. reinforced clay. Electronic Journal of Geotechnical Engineering, vol. 8,
(3) The effect of geogrid reinforcements on the footing Bundle (C).
performance is dependent on the footing location Omar, M., Das, B., Puri, V., Yen, S., 1993. Ultimate bearing capacity of
shallow foundations on sand with geogrid reinforcement. Canadian
relative to slope crest. In terms of BCI, geogrid is
Geotechnical Journal 30, 545–549.
most effective when the footing is placed on the slope Otani, J., Ochiai, H., Yamamoto, K., 1998. Bearing capacity analysis of
crest rather than any distance away from the slope reinforced foundations on cohesive soil. Geotextile and Geomem-
crest. branes 16, 195–206.
(4) For a footing located at slope crest resting on 1.5B Ovesen, N.K., 1979. The use of physical models in design: the scaling law
relationship. In: Proceedings of the Seventh European Conference on
thicknesses of replaced sand, an adequate anchorage
Soil Mechanics and Foundation Engineering, vol. 4, pp. 318–323.
length for each geogrid layer should be provided along Patra, C.R., Das, B.M., Atalar, C., 2005. Bearing capacity of embedded
with an optimum number of geogrid layers should be strip foundation on geogrid-reinforced sand. Geotextiles and Geo-
used. The length should be greater than or equal to membranes 23 (5), 454–462.
(L/B ¼ 5) five times the footing width and the Patra, C.R., Das, B.M., Bhoi, M., Shin, E.C., 2006. Eccentrically loaded
strip foundation on geogrid-reinforced sand. Geotextiles and Geo-
recommended number of geogrid layers is 3.
membranes 24 (4), 254–259.
(5) For the studied slope geometry and conditions, the Samtani, N., Sonpal, R., 1989. Laboratory tests of strip footing on
maximum benefit of geogrid reinforcements is depen- reinforced cohesive soil. Journal of Geotechnical Engineering—ASCE
dent on reinforcement configuration. For optimum 15 (9), 1326–1330.
response, the recommended depth of reinforcing Selvadurai, A., Gnanendran, C., 1989. An experimental study of a footing
located on a sloped fill: influence of a soil reinforcement layer.
geogrid (u/B) and geogrid spacing (x/B) are 0.6 and
Canadian Geotechnical Journal 26 (3), 467–473.
0.5 of the footing width. Tensar International, 1995. Tensar geogrid reinforced sub-bases—case
(6) A close agreement between the experimental and studies. New Wellington Street, Blackburn BR2 4PJ, England.
numerical results on general trend of behavior and Tsukada, Y., Isoda, T., Yamanouchi, T., 1993. Geogrid subgrade
the critical values of the geogrid parameters is reinforcement and deep foundation. In: Raymond, Giroud (Eds.),
Proceedings of the Geosynthetics Case Histories. ISSMFE, Committee
observed. In all cases, the BCI calculated from the
TC9, pp. 158–159.
FEA for prototype scale appears to be smaller than Vesic, A., 1973. Analysis of ultimate loads of shallow foundations. Journal
that obtained from the model slope results. of Soil Mechanics and Foundations Division—ASCE 94 (SM3),
661–688.
Yetimoglu, T., Inanir, M., Inanir, O., 2005. A study on bearing capacity of
randomly distributed fiber-reinforced sand fills overlying soft clay.
References Geotextiles and Geomembranes 23 (2), 174–183.
Yoo, C., 2001. Laboratory investigation of bearing capacity behavior of
Alawaji, H., 2001. Settlement and bearing capacity of geogrid-reinforced strip footing on geogrid-reinforced sand slope. Geotextiles and
sand over collapsible soil. Geotextiles and Geomembranes 19, 75–88. Geomembranes 19, 279–298.

You might also like