Vous êtes sur la page 1sur 12

Technical Note

Stress Distribution below a Continuous Footing on


Geotextile-Reinforced Soil
Elif Cicek1; Erol Guler, M.ASCE2; and Temel Yetimoglu3
Downloaded from ascelibrary.org by Indian Institute of Technology, Delhi on 01/10/18. Copyright ASCE. For personal use only; all rights reserved.

Abstract: This study investigated the additional vertical stress (Ds ) change occurring under the centerline of a continuous footing on
geotextile-reinforced soil. For different reinforcement parameters and conditions, load tests and finite-element model (FEM) analyses were
performed. It was determined that the presence of reinforcements reduces the additional vertical stress values. There is a good correlation in
some cases between experimental measurements and FEM results, but the values obtained are different from the values proposed based on the-
oretical considerations developed for uniform materials. Each additional reinforcement layer affects Ds , and different parameters can affect
the values of Ds . It was determined that the reinforcements create a discontinuity in the triangular failure wedge under the footing, thereby
affecting the additional stress developed. DOI: 10.1061/(ASCE)GM.1943-5622.0001095. © 2018 American Society of Civil Engineers.
Author keywords: Additional stress; Sand; Vertical pressure; Finite element; Plate load.

Introduction different factors (Cicek et al. 2014). In recent years, a limited number
of studies have been performed to determine the stress behavior of
Reinforced soils are used in many areas of civil engineering, such reinforced soil (Chen 2007; Lovisa et al. 2010; Abu-Farsakha et al.
as foundations, highways, airport pavements, railways, embank- 2016). It was noted that the additional stress was decreased by using
ments, slopes, and retaining walls. It is well known that reinforce- reinforcement materials in soil. However, the studies mainly investi-
ment works together with soil to provide an improved composite gated the stress distribution in dense sands, and consequently, the
performance during compressive loading and changes the differen- stress distribution of the reinforced medium-dense sand is not
tial settlement. For many years, a number of researchers have con- known. As was known from the study by Cicek et al. (2014), the
ducted model tests in laboratories and numerical analyses to deter- stress behavior changes depending on the relative stiffness of the
mine the effect of the reinforcements (Binquet and Lee 1975a, b; soil. In some studies, it was also stated that compaction of sand can
Guido et al. 1986; Huang and Tatsuoka 1990; Al-Qadi et al. 1994, affect the stress distribution. Both laboratory and finite-element
1997, 2008; Adams and Collin 1997; Schlosser et al. 1983; Huang model (FEM) results show this difference (Lovisa et al. 2010;
and Menq 1997; Wayne et al. 1998; DeMerchant et al. 2002; Yetimoglu 1998). However, soil reinforcement is mostly needed in
Kumar and Saran 2003; Michalowski 2004; Hufenus et al. 2006; the case of medium-dense sands in real projects. Many studies can
Ghazavi and Lavasan 2008; Cicek 2011; Abu-Farsakha et al. 2016). be found in the literature on reinforced soil behavior and bearing
It is a known fact that stress distribution in soils was studied by capacity, but there is not much information on additional stress distri-
some well-known researchers for unreinforced soil. The Boussinesq bution behavior of reinforced medium-dense sand. This study pro-
(1885) and Westergaard (1938) theories are among the most fre- vides important information on this aspect.
quently used theories. According to Westergaard’s theory, soils con- In reinforced foundation studies, the parameters related to geom-
taining stronger layers between soft layers tend to prevent lateral etry are commonly normalized to minimize the scale effect. The
spreading, and therefore, the stress at deeper layers is less than that scale effect is one of the problems for laboratory model tests.
formulated for isotropic soil after the Boussinesq approach. The However, in the literature, it was stated that the general mechanisms
Westergaard calculation indicates that stratified soil behaves like a and behavior observed in the model tests are reproduced on a large
soil reinforced by thin and rigid sheets (Naval Facilities Engineering scale, and model-scaled tests can be helpful to determine the mecha-
Command 1982). The assumptions of the additional vertical stresses nism of foundations (Milligan et al. 1986; Adams and Collin 1997;
in the theories are generally adequately correct. Nevertheless, neither Sireesh et al. 2009; Moghaddas Tafreshi and Norouzi 2012; Cicek
theory considers the material properties. However, some literature et al. 2015).
studies show that stress distribution in soils can be affected by Furthermore, for different configurations of reinforcement, the
effects can be changed, so different configurations were selected to
determine the effect on interaction without changing the type of
1
Assistant Professor, Dept. of Civil Engineering, Hacettepe Univ., reinforcement for medium-dense sand. In this study, laboratory
Ankara 06800, Turkey (corresponding author). E-mail: elif.cicek@ tests and numerical analyses were conducted, and the results were
hacettepe.edu.tr
2 compared with literature studies and classical evaluations.
Professor, Dept. of Civil Engineering, Bogazici Univ., Istanbul
34470, Turkey. E-mail: eguler@boun.edu.tr
3
Emeritus Professor, Dept. of Civil Engineering, Ataturk Univ.,
Material Properties and Test Setup
Erzurum 25240, Turkey. E-mail: yetimt@atauni.edu.tr
Note. This manuscript was submitted on February 22, 2017; approved
on September 25, 2017; published online on January 10, 2018. Discussion In this study, uniform dry sand with a specific gravity of 2.65 was
period open until June 10, 2018; separate discussions must be submitted used to investigate the pressure behavior under static load. The min-
for individual papers. This technical note is part of the International imum and maximum dry unit weights of the sand were 13.9 and
Journal of Geomechanics, © ASCE, ISSN 1532-3641. 16.5 kN/m3. The sand had a coefficient of uniformity of 2.5 and an

© ASCE 06018005-1 Int. J. Geomech.

Int. J. Geomech., 2018, 18(3): 06018005


effective size of a particle of 0.22 mm. In all model tests, the aver- Table 1. Properties of Geosynthetic Reinforcement
age unit weight and relative density of the sand were kept constant
Material property Polypropylene
at 15 kN/m3 and 46%, respectively. The friction angle of the sand at
a relative density of 46% was determined to be 38° with triaxial lab- Type Woven
oratory tests. A constant relative density was achieved in the test Ultimate tensile strength (kN/m) 60
tank using a sand raining technique. A woven geotextile was used Failure strain (%) 16
as reinforcement, and its properties are given in Table 1. The test Mass per unit area (g/m2) Thickness (mm) 310 1.01
tank was a steel tank measuring 100 cm (width) by 50 cm (length)
by 100 cm (height). The inside walls and the edges of the tank were
polished to reduce friction as much as possible. The static load was
applied on a steel plate with a width of 10 cm (B), a thickness of
25 mm, and a length of almost the same length as the tank. The plate
Downloaded from ascelibrary.org by Indian Institute of Technology, Delhi on 01/10/18. Copyright ASCE. For personal use only; all rights reserved.

was centered, with the length of the footing parallel to the width of
the tank. The plane-strain condition existed in all tests. The plate
was placed on the surface of the sand bed. The applied load was
measured by a load cell with the help of a data logger. Additional
vertical soil stress values (Ds ) were measured by using miniature
pressure sensors. These gauges have a diameter of 2.667 mm and a
thickness of 0.508 mm. Such small pressure gauges were chosen so
that the stress distribution would not be influenced by creating addi-
tional rigidity. For the same reason, the cabling of the strain gauges
was chosen to be as thin as the electronic circuit allows. Pressure
sensors measured the stresses directly with the help of quarter
bridges and a connector block (Cicek 2011; Cicek et al. 2014). All
pressure gauges came with a calibration, but the calibrations were Fig. 1. Test setup
also checked at the authors’ laboratory to be sure. Many times dur-
ing disassembly, the strain gauges or the cables were harmed.
Therefore, a new set of pressure gauges were used in each test to exponent of the stress-stiffness function = 0.5. The steel plate was
avoid mismeasurements. Miniature pressure sensors were placed modeled using 5-node beam elements, and plate properties were
below the centerline of the plate at various depths (z) from the base represented with flexural rigidity (260 kNm2/m) and axial stiffness
of the footing (z/B = 0.35, 0.55, 0.75, 0.95, 1.65, and 2.2). These (5  106 kN/m) (Cicek et al. 2014). Elastic rigidity for geotextile
locations were selected to enable measurement of the pressure in reinforcement was taken as J = 550 kN/m. The interactions of sand
the reinforced and unreinforced zones. Sensors were located care- and reinforcement happen mostly at the interface, and this interac-
fully during sand placement, and some tests were repeated to be tion depends on the sand particle morphology and the surface
sure of the measurements. However, in some tests, mainly due to roughness characteristics of the reinforcement. In this study, to
the cable breakage of some sensors, measurement for some points account for this, the interface coefficient between soil and geotex-
could not be taken. Additionally, the geotextile reinforcements tiles was taken as 0.7. Because stresses can be measured in the
(with the 1.01-mm thickness) were placed in the soil in different experiments only at discrete points, the FEM results are very impor-
configurations to achieve the maximum benefit from the reinforce- tant for interpretation of the results.
ment, in accordance with the literature survey (Cicek et al. 2015).
After the placement of sand and reinforcements was completed, the
Results
plate footing was placed on the sand surface, and uniform static ver-
tical loads were applied to the center of the model plate. The differ-
To investigate the additional vertical soil stress distribution in rein-
ent configurations of the reinforcement geometry were controlled
forced and nonreinforced soil, experiments were conducted in a lab-
by the vertical spacing between consecutive layers of reinforcement
oratory, and measured stress values were compared with finite-
(h), the total number of reinforcement layers (N), the width of the
element (FE) analysis results. These values were also compared to
geosynthetic reinforcement (L), and total reinforced zone (d). The
the Westergaard and Boussinesq values. First, the results of unrein-
test setup can be seen in Fig. 1. In modeling the problems, the first
forced models were discussed. Then, test results were compared for
reinforcement depth was taken constantly as u = 0.35B.
different reinforcement parameters, namely, number of reinforce-
ment layers, vertical spacing of reinforcement layers, and various
FEM Properties loading levels.
The stress distribution of unreinforced soil was examined and
FEM analyses were performed using the PLAXIS 8 program. reported by Cicek et al. (2014). As can be seen in Fig. 2, the values
Boundary-loading conditions were modeled to simulate the labora- of additional stress in soil created by the surface foundation (Ds )
tory test conditions. The plane-strain method was chosen for the obtained from FEM were similar to the Boussinesq values at greater
two-dimensional (2D) model. The mesh size was very fine. Fifteen- depths, but the vertical stresses measured in the laboratory tests
node triangular elements were used. The problem is symmetrical were lower than both the FEM and Boussinesq predictions for the
regarding the center of the footing, so only half of the system was stress level of q = 50 kPa (Cicek 2011).
modeled. Laboratory test results were used as model properties. For As is known from the literature, the presence of reinforcement
modeling, the hardening-soil model was chosen to model the sand layers changes the soil’s mechanical behavior. Failure does not
soil. The soil properties were as follows: unit weight = 15 kN/m3, occur at the same stress as in unreinforced conditions, and load–
elasticity modulus = 25,000 kN/m2, Poisson’s ratio = 0.25, friction settlement curve behavior can change the bearing capacity of the
angle = 38°, dilatancy angle = 8°, reference pressure = 100 kPa, soil for different reinforcement configurations. So, initially, the

© ASCE 06018005-2 Int. J. Geomech.

Int. J. Geomech., 2018, 18(3): 06018005


load–settlement (q-s) curves obtained from laboratory model tests, that this behavior is due to the soil parameters, such as modulus
and subsequently, FEM analysis results, were compared for unrein- of elasticity, used in the analysis and the values of displacement
forced and geotextile-reinforced soil, as seen in Fig. 3. The number measured in the laboratory. For this reason, it can be said that the
of reinforcements, reinforcement length, and vertical spacing medium-dense soil properties, such as average unit weight and
between reinforcement layers for the reinforced model were relative density, can be changed by the loading because the exces-
selected based on literature studies (Cicek 2011; Cicek et al. 2015) sive stress causes a reduction in the void ratio, and as a conse-
as N = 3, L = 3B, and h = 0.4B, respectively. It is evident that, for quence, the relative density of the soil increases. When an exter-
measurements of small loading and FEM analysis, the results nal load is applied to a moderately dense sand, the stress level
show the same values for both tests. However, this behavior applied until that failure is not great enough to create a further
changes in the case of larger loads, and in this case, the unrein- densification of the sand. However, because the presence of rein-
forced model carries smaller loads than the reinforced soil for the forcement prevents failure, larger loads can be applied. These
same strain. Boushehrian and Hataf (2003) reported that the FE
Downloaded from ascelibrary.org by Indian Institute of Technology, Delhi on 01/10/18. Copyright ASCE. For personal use only; all rights reserved.

high loads and the lateral confinement created by the reinforce-


and physical test results do not match completely. They explained ment may cause a densification in the sand. Therefore, in this pa-
per, results are discussed for the small load (q = 50 kPa) and larger
loads, such as q = 100 kPa.

Effect of Length of Reinforcement on Dr Distribution


It is known that the reinforcement length can change the deforma-
tion behavior in load–settlement curves and bearing capacities of
soils (Cicek et al. 2015). It can be said that the change in the addi-
tional vertical stress due to the reinforcement is a very important
factor influencing this behavior. As was shown by Cicek et al.
(2015), L = 5B is the geotextile reinforcement length, which gives
the maximum bearing capacity, and L = B gives the minimum bear-
ing capacity for reinforced soil models. It can be stated that the rein-
forcement length affects the additional pressure behavior of soil and
the bearing capacity. To emphasize the difference between the dif-
Fig. 2. Additional vertical stresses for unreinforced soil (reprinted
ferent lengths, the results of only L = B and L = 5B are compared in
from Cicek et al. 2014, © ASCE)
Fig. 4. The number of reinforcement layers and the vertical spacing

Fig. 3. Load-settlement curves (q-s): (a) for unreinforced sand (reprinted from Cicek et al. 2014, © ASCE); (b) for reinforced sand

Fig. 4. Additional vertical stress distributions at different points for different reinforcement lengths: (a) L = 5B; (b) L = B

© ASCE 06018005-3 Int. J. Geomech.

Int. J. Geomech., 2018, 18(3): 06018005


of the reinforcement layers were kept constant as N = 3 and h = ratios than the model with L = B for 5-cm settlement. As a result, it
0.4B. It can be seen from Fig. 4 that FEM analysis and model test can be said that the optimum reinforcement-layered soil has a
results show good agreement in some cases for the reinforced zone greater bearing ratio and can be loaded more by decreasing addi-
(z < 0.75 for N = 2, and z < 1.15 for N = 3); however, the results tional vertical stress in the reinforced zone. Each reinforcement
have smaller values than FEM and theoretical values below the re- behaves as a block for loads, and transfer of the loads under soil
inforced zone. It should be mentioned that the vertical stresses can be attempted. Friction and blocking effects are critical, after
decreasing and increasing with depth are related to the presence of the reinforcement layers, soil behaves weaker. Hence, soil takes
reinforcement layers at these depths. At greater depths, the results more loads. Reinforcement effects absorb stresses in the rein-
of the laboratory test and FEM show smaller additional stresses forced zones, and behave as a different rigid plate above the
resembling more closely the Westergaard distribution. Furthermore, weak soil. This can be seen clearly in Fig. 6. Although unrein-
it can be clearly seen that each reinforcement layer affects the addi- forced model maximum pressure values occur under the footing
tional vertical stress (Ds ) of soil and decreases the value of Ds in center, for reinforced models, the pressures can be changed on
Downloaded from ascelibrary.org by Indian Institute of Technology, Delhi on 01/10/18. Copyright ASCE. For personal use only; all rights reserved.

the reinforced zone, especially as the number of reinforcement the reinforcement lines and absorb the energy. They transfer the
layers increases for all reinforcement lengths. Furthermore, the loading effects under the reinforced zone. That is to say, rein-
reinforcements send the vertical stress to a deeper area. The other forcements can move the loads to deeper areas, as can be seen in
important point is that, as can be seen from Fig. 4, the larger geo- Fig. 6.
textile reinforcement length reduces the additional vertical stress
more than the shorter one, which is equal to the width of the plate. Effect of Vertical Spacing of Reinforcement Layers on
Fig. 4(a) shows results for a longer length of the reinforcements Dr Distribution
(L = 5B), and the number of reinforcements is N = 3. This means
the maximum bearing ratios can occur in this condition. When the To investigate the effect of additional vertical stress behavior for
reinforcement layers’ lengths increase, the effect of the reinforce- different vertical spacings of reinforcement layers, model tests were
ment increases. It was thought that reinforcement behaves as a conducted with h/B = 0.6, 0.4, and 0.2. In these series of tests, the
plate and block in front of the loading vectors. And vertical stress number of reinforcements and reinforcement length were taken as
decreases by the effect of this longer block and friction between N = 4 and L = 3B, respectively. The bearing capacity–settlement
the reinforcement and soil particles. Hence, stresses moved to hori- relation was reported by Cicek et al. (2015). Fig. 7 shows the addi-
zontal lines, and for vertical parts, this value can be decreased. This tional vertical stress distribution for different vertical spacings. As
situation can be seen more clearly in Fig. 5, in which FEM analyses can be seen from the graphs, experimental measurements and FEM
are compared to each other for longer and shorter geotextile rein- analysis results have approximately good agreement with each
forcement lengths (L = 5B and B) for N = 3. To investigate the rea- other in some cases for the shape of the curve by reinforcement
son for the reduction in vertical stress, FE results were compared effect, but in some cases, experimental results have smaller vertical
with unreinforced sand for reinforcement lengths of L = 5B and L = stress values than the FE results. In the reinforced zone, additional
B in (Fig. 6). As can be seen from Fig. 6, the reinforcement causes vertical stress values are similar to Westergaard theory results, and
the stresses to be transmitted to deeper layers, probably because the Ds values change with the effects of the reinforcement layers. The
load on the footing plate may enable densification of the soil more effect of the vertical displacement of reinforcement layers is shown
easily with the help of the reinforcement, preventing lateral spread- more clearly in Fig. 8, which is a comparison of effects for bigger
ing. Therefore, the stress could be transferred to the deeper layers (h = 0.6B) and smaller (h = 0.2B) vertical spacings of reinforce-
via this effect. Additionally, reinforcements divide the active zone ments on Ds distribution in sand soil. As can be seen from the
under the plate into parts by tensile load and membrane effect or curves of Fig. 8, Ds values for the smaller vertical spacing of rein-
friction between reinforcement. Shorter reinforcements (L = B) forcement layers (h = 0.2B) until almost z = 1B (z = 10 cm) decrease
could affect only the active zone, but the behavior of the longer more than the greater vertical spacing of reinforcement layers (h =
ones (L = 5B) might impact larger areas, so additional vertical soil 0.6B), because this depth is the reinforced zone for h/B = 0.2. After
stress values become smaller, and the results of applied load can be the depth z = 1B, Ds values start to increase for h = 0.2B. However,
improved. Cicek et al. (2015) reported that the model with a rein- Ds values for h = 0.6B start to become smaller than h = 0.2B
forcement length of L = 5B has almost 50% higher bearing capacity because, for h = 0.6B, the reinforced zone continues until z = 2.15B.
That is, for h = 0.6B, the reinforced zone (d1) is greater than the rein-
forced zone (d2) of h = 0.2B, and vertical spacing of reinforcement
layers can increase the stress levels more. However, the behavior of
both additional vertical stress curves becomes similar at a depth of
z = 3B. Therefore, under the reinforced zone for both d1 and d2,
the degrees of stress start to increase and become the same value.
Indeed, it can be seen that different conditions can affect the addi-
tional vertical stress distribution in the soil and, consequently, the
bearing capacity. In other words, reinforcements decrease the Ds
values, but these values start to increase under the reinforcement,
and then for the second reinforcement layer, Ds values decrease.
So, each reinforcement layer reduces the vertical load and prob-
ably spreads it laterally, and the stress increases again but below
the reinforcement. However, under the reinforced zone, the Ds
values abruptly become far greater. In the reinforced zone, addi-
tional stress behavior is smaller than the literature suggestions,
Fig. 5. Comparison of additional stress distributions for L = 5B and but under this zone, the stress behavior resembles the calculation
via FEM analysis formula of Boussinesq more than that of Westergaard.
Additionally, the results of the FEM show the effect of

© ASCE 06018005-4 Int. J. Geomech.

Int. J. Geomech., 2018, 18(3): 06018005


Downloaded from ascelibrary.org by Indian Institute of Technology, Delhi on 01/10/18. Copyright ASCE. For personal use only; all rights reserved.

Fig. 6. Vertical stress distribution below footings in unreinforced and reinforced soils from FEM analysis: (a) unreinforced; (b) reinforced soil for
\L = B; (c) reinforced soil for L = 5B

reinforcements in different vertical spacings for stress levels follows that the decrease of the additional stress levels can be
(Fig. 9). It can be noticed that, as the vertical spacing of the rein- determined in this way.
forcement layer decreases, the boundary of the active stress zone
changes. Each reinforcement layer affects the active zone and
Effect of Number of Reinforcement Layers on
causes a small amount of swelling on the border of the active
Dr Distribution
zone, dividing it into sections and transferring the stress from the
center of the footing into horizontal lines. The greater vertical A comparison was made for the different numbers of reinforce-
spacing of reinforcement layers affects the deeper areas. Thus, it ments, such as N = 2, 3, and 4 at h = 0.6B for L = 3B in Fig. 10. It

© ASCE 06018005-5 Int. J. Geomech.

Int. J. Geomech., 2018, 18(3): 06018005


Downloaded from ascelibrary.org by Indian Institute of Technology, Delhi on 01/10/18. Copyright ASCE. For personal use only; all rights reserved.

Fig. 7. Effect of vertical spacing of reinforcement layers: (a) h = 0.6B; (b) h = 0.4B; (c) h = 0.2B

the effects of the reinforcement layers more clearly, and different


reinforcement configurations were compared in the same reinforced
zone depths (d). The configuration of the tests is given in Table 2
and the results are shown in Fig. 11, and it can be seen that three dif-
ferent depths of the reinforced zone were discussed (d/B = 0.75,
0.95, and 1.55). In Fig. 11, the reinforced zone depth is equal for
each graph, but the values of the number of reinforcement layers
and, consequently, the vertical spacing of reinforcement layers are
different. From the interpretation of Fig. 11, it can be said that for
the same reinforced zone depth, the number of reinforcement layers
affects the result in small ratios and causes only small changes in
the vertical stress distribution for a small loading value (50 kPa).
Therefore, it can be concluded from these results that the reinforced
zone depth is more influential in decreasing the stress levels in
Fig. 8. Comparison of FEM results for vertical spacing medium-dense soil for small loadings.

Dr Distribution for Different External Load Levels


can be noticed from Fig. 10 that when the number of reinforcements
increases, each geotextile reinforcement decreases the additional To understand the effect of the external stress level, three levels
vertical soil pressure values. The model laboratory test measure- were chosen. One level is nearly below the bearing capacity of the
ments and FEM results show similar behaviors, but the theoretical unreinforced sand (namely, q = 50 kPa). The second level is above
results were slightly larger in some cases. the unreinforced soil’s bearing capacity (namely, q = 100 kPa), and
As is generally agreed in the literature (Chen 2007; Cicek 2011), the last one is for much bigger loadings. First, the number of rein-
the total maximum reinforcement depth that is efficient is 3B. forcement was chosen as N = 4, the length of the reinforcement
However, the effect of the number of reinforcement layers and layers was chosen as L = 3B, and the vertical spacings of the rein-
depth of reinforced zone on the pressure distribution in soil is not forcement layers were h = 0.6B, 0.4B, and 0.2B. As can be seen in
known completely yet. The vertical spacing and the number of rein- Fig. 12, there is a good correlation in some cases between the exper-
forcement layers were studied in this study, but it is important to see imental measurements and FE analysis results. However, the test

© ASCE 06018005-6 Int. J. Geomech.

Int. J. Geomech., 2018, 18(3): 06018005


Downloaded from ascelibrary.org by Indian Institute of Technology, Delhi on 01/10/18. Copyright ASCE. For personal use only; all rights reserved.

Fig. 9. Effect of vertical spacing of reinforcement on vertical stress distribution below the footing for q = 100 kPa found from FEM analysis: (a) h =
0.6B; (b) h = 0.4B; (c) h = 0.2B

results of stresses are slightly smaller than the FE model results, but reinforcements and settlement ratio on bearing ratios. The bearing
for both of them, reinforcements affect the stress distribution. On ratio (BR) is defined as BR = qr/qu. Here, qu is the average loading
increasing the vertical spacing of layers, Ds values become smaller of footing on unreinforced soil, and qr is the average loading of the
in the deeper depths of soil, and the reinforced zone can be more same footing on reinforced soil at the same settlement ratio (s/B). It
effective. In this series, the changing ratio in Ds values is bigger can be noticed that as the settlement rates increase, the BR values
than the q = 50 kPa models. Also, in the reinforced zone, Ds values also increase. This means that if larger settlements are allowed, the
become almost smaller than the Westergaard results, and under the bearing capacity increase becomes much more pronounced, and if
reinforced zone, Ds values are greater than the Boussinesq calcula- only a limited settlement is allowed, the effect of the reinforcement
tion results. on bearing capacity remains more moderate. Additionally, there is a
It is important to see the behavior of a structure under working good correlation in some cases between Figs. 13 and 14. This means
loads and under extreme loading values. Therefore, finally, different that as the loading values on the structure increase, the additional
load levels were compared for the same reinforced zone, as can be vertical stress decreases more than the small loadings. Each rein-
seen in Fig. 13. Here, the depth of the reinforced area is same as in forcement divides the loadings and distributes them. In the rein-
Fig. 13(a) for h = 0.2B and N = 4, and in Fig. 13(b) for h = 0.6B and forced zone, smaller vertical stresses occurred, and settlements can
N = 2. These configurations were already compared for the small be decreased by this effect. In other words, as loading increases,
loading value (50 kPa) in Fig. 11(a). However, in Fig. 13, the com- vertical stress in the reinforced zone decreases by the reinforcement
parisons can be seen for different loading values. Fig. 13 shows that effect, and so the total settlement values of the structure decrease
within the reinforced zone, the magnitude of the additional pressure and bearing ratio values increase. Hence, it can be said that rein-
decreases, and below the reinforced zone, the additional stress val- forced soils can show much more effective improvement in big
ues increase for both N = 2 and N = 4. Furthermore, as the magni- loading values by changing the behavior of vertical stress distribu-
tude of the load increases, the difference in the additional stress in tion, and so settlement behavior can also be affected.
the reinforced and unreinforced zone becomes more pronounced. In the literature, some stress distribution measurements are
As the loading on the footing increases, the settlement ratios reported, but not on the medium-dense sand, and when the figures
increase as well. Fig. 14 shows the effect of the number of given in this paper are examined carefully, it can be said that the

© ASCE 06018005-7 Int. J. Geomech.

Int. J. Geomech., 2018, 18(3): 06018005


Downloaded from ascelibrary.org by Indian Institute of Technology, Delhi on 01/10/18. Copyright ASCE. For personal use only; all rights reserved.

Fig. 10. Distribution of additional vertical stress with number of reinforcement layers: (a) N = 2; (b) N = 3; (c) N = 4

Table 2. Parameters for Comparison of Number of Reinforcements reinforcement configurations. Via different reinforcement parame-
ters (h, L, and N), the behavior of the reinforced zone can be
Vertical spacing of
changed, and the differences between the test and FEM or theoreti-
Number of reinforcement reinforcement layers Depth of the reinforced
layers (N) (h/B) zone (d/B) cal results can be changed. Additionally, different conditions
change not only bearing ratios but also stress behavior in the tests.
2 0.4 0.75 Hence, friction between the soil particles and geotextile can change
3 0.2 the behavior of the results.
2 0.6 0.95 Additionally, there are some limitations, as can be seen for FE
4 0.2 results. A more detailed simulation can be made when the FE pro-
3 0.6 1.55 grams are developed and more details can be introduced. Therefore,
4 0.4 the similar studies can be continued via different programs and
methods in the future. Additionally, the soil was almost loose in this
study, so different conditions can be studied and the results can be
stiffness of the sand can be important in terms of the additional ver- compared with this study. Furthermore, theoretical methods also
tical stress distribution. For the reinforced case, when the unrein- have some limitations. For example, for the Boussinesq method, the
forced bearing capacity is exceeded, the excessive load can densify soil stiffness and other descriptions cannot be determined, so new
the soil. Therefore, with increasing pressure, the density of the sand theoretical methods should be improved for different unreinforced
soil changes, and as a consequence, the additional vertical stress dis- and reinforced models with different reinforcement parameters.
tribution also changes. Additionally, the mean grain size of the sand was 0.22 mm.
In general, there is a good correlation in some cases between the Hence, it can be said that each gauge could connect with 24 sand
experimental measurements and the FE analysis results. However, particles. Therefore, some possible particle sizes may affect some
test measurements for additional stress behavior are smaller than small measurements. However, in this study, the point results in
the FEM results. The unreinforced soil shows more difference, but general have good agreement when the reinforced soil results are
the characteristics of the curves are similar. The reason for this can compared with each other.
be the relative stiffness of the soil. In general, as is known from the Ling and Liu (2003) reported that there can be big differences
literature (Boushehrian and Hataf 2003), this may be due to the soil between the test and computer results. They conducted plate load
parameters, such as modulus of elasticity, used in the analysis and tests on reinforced sand and computed the model by using the same
the values of displacement measured in the laboratory. The sand FE method as this paper (PLAXIS), but they highlighted that there
used in this study was medium-dense sand, and so the FEM models can be differences between the FE and laboratory test results for
cannot model the values completely, but can model behavior and soils. They reported that the stiffness of the geogrid affected the
changing slopes similar to experiments. load–settlement relationships. However, different values of stiff-
FEM analyses were based on the theoretical backgrounds, but in ness gave rise to different strain distributions in the geogrid.
the real test, this condition can be changed via different soil or Additionally, the stiffness of reinforcement can affect the load–

© ASCE 06018005-8 Int. J. Geomech.

Int. J. Geomech., 2018, 18(3): 06018005


Downloaded from ascelibrary.org by Indian Institute of Technology, Delhi on 01/10/18. Copyright ASCE. For personal use only; all rights reserved.

Fig. 11. Comparisons of reinforced zone depths: (a) d = 0.75B; (b) d = 0.95B; (c) d = 1.55B

settlement behavior. The geosynthetic material modeled as a linear- in this study is loose sand, so in some cases, the stress point can
elastic model poorly predicted the strain distributions in the geosyn- occur suddenly by the effect of the loading and being much stiffer
thetic layer. Additionally, test and FE results showed big differen- than the starting point of the experiment. Some differences can
ces in the study by Yetimoglu et al. (1994). Additionally, occur via the loading effect when the soil compacts in the experi-
Yetimoglu et al. (1994) reported that Lambrechts and Leonards ment. Hence, a new approach should be determined by conducting
(1978) indicated that the irreversible strain due to prestressing could experiments on reinforced soils.
increase the reload modulus by more than an order of magnitude Many reasons were mentioned earlier for the differences
over that of the normally consolidated sand, even though no residual between FEM and test methods. For dense sand soil, in some
stresses due to stressing are extant. The presence of residual lateral studies, the comparisons can be the same; however, this study
stress can further increase the soil stiffness by an additional factor shows one important point for not only researchers but also pro-
of two. Lade and Duncan (1976) indicated that prior stress history, ject engineers in the field: that this behavior starts to change for
which may influence stress-strain behavior of a sand very signifi- loose sand, and for some behaviors, the differences can be seen.
cantly, had no discernible effect on the shear strength. There are a By loading the soil, the grains become denser, and the stiffness of
lot of limitations for FE studies, and the theoretical studies the soil changes as mentioned earlier. Another point is that loose
(Boussinesq and Westergaard) have gaps and uncertainties in deter- soil has a different grain distribution from denser ones, and in
mining the behavior of the reinforced soils. This study shows the FEMs, only some parameters can be determined by general soil
importance of experiments and different soil types. The procedure values. Although load–settlement behavior is consistent, pressure
of the study of Lade and Duncan (1976) was based on only limited or stress behaviors of reinforced soils in loose sand can be inco-
parameters but density of soil changed by loading of the plate, and herent when test and FEM results are compared. Load–settlement
the membrane effect (Zornberg 2012) of the reinforcement can occurs about the general form of the structure, but stress behavior
affect the behavior of pressure in the soil. Each reinforcement place- in the soil can be affected by the grain behaviors. Therefore, in
ment and condition parameters, such as vertical displacement of future studies, discrete-element method can be used for similar
reinforcements, numbers, or lengths, can affect the behavior of the researches. Additionally, project engineers should be careful
reinforced soil (Cicek et al. 2015). Via the effect of the reinforce- about using FE because some soil conditions can be give different
ment, the vertical stress behavior decreases. However, the soil used results according to real soil conditions. Loose and medium-

© ASCE 06018005-9 Int. J. Geomech.

Int. J. Geomech., 2018, 18(3): 06018005


Downloaded from ascelibrary.org by Indian Institute of Technology, Delhi on 01/10/18. Copyright ASCE. For personal use only; all rights reserved.

Fig. 12. Additional vertical stress distributions for q = 100 kPa: (a) h = 0.6B; (b) h = 0.4B; (c) h = 0.2B

Fig. 13. Additional stress distributions for different loading values: (a) h = 0.2B, N = 4; (b) h = 0.6B, N = 2

dense conditions have different failure systems from dense condi- Conclusion
tions. Dense sand and rigid materials, such as concrete, can give
the same results as FEM, but discontinuous structures can have In this paper, the additional vertical stress under the center of a strip
different behaviors. This case should be considered for solving plate resting on medium-dense sand soil was investigated through
problems. Theoretical calculations, such as Boussinesq and FEM experiments and FE analysis. Furthermore, measurements and nu-
computations, need revisions for loose and medium-dense soils. merical results were compared with classical theoretical calcula-
However, there are limited studies to compare the results. In the tions. Different load levels were used. The research was repeated
future, new studies can be conducted to find the new theoretical for different reinforcement configuration parameters (L, h, and N),
calculations by helping this parametric study. and their effects were noted. The results show the behavior of

© ASCE 06018005-10 Int. J. Geomech.

Int. J. Geomech., 2018, 18(3): 06018005


Adams, M. T., and Collin, J. G. (1997). “Large model spread footing load
tests on geosynthetic reinforced soil foundations.” J. Geotech.
Geoenviron. Eng., 10.1061/(ASCE)1090-0241(1997)123:1(66), 66–72.
Al-Qadi, I. L., Brandon, T. L., and Bhutta, S. A. (1997). “Geosynthetic sta-
bilized flexible pavements.” Proc., Conf. Geosynthetics ’97, Industrial
Fabrics Association International, St. Paul, MN, 647–661.
Al-Qadi, I. L., Brandon, T. L., Valentine, R. J., Lacina, B. A., and Smith,
T. E. (1994). “Laboratory evaluation of geosynthetic reinforced pave-
ment sections.” Transp. Res. Rec., 25–31.
Al-Qadi, I. L., Dessouky, S. H., Kwon, J., and Tutumluer, E. (2008).
“Geogrid in flexible pavements: Validated mechanism.” J. Transp. Res.
Board, 1, 102–109.
Binquet, J., and Lee, K. L. (1975a). “Bearing capacity analysis of reinforced
Downloaded from ascelibrary.org by Indian Institute of Technology, Delhi on 01/10/18. Copyright ASCE. For personal use only; all rights reserved.

earth slabs.” J. Geotech. Eng. Div., 101(12), 1257–1276.


Fig. 14. Bearing ratio–settlement ratio behavior for different Binquet, J., and Lee, K. L. (1975b). “Bearing capacity tests on reinforced
conditions earth slabs.” J. Geotech. Eng. Div., 101(12), 1241–1255.
Boushehrian, J. H., and Hataf, N. (2003). “Experimental and numerical
investigation of the bearing capacity of model circular and ring footings
medium-dense sand, of which the effects were not known com- on reinforced sand.” Geotext. Geomembr., 21(4), 241–256.
pletely. Based on the results, the following conclusions were Boussinesq, J. (1885). Application des potentiels a l’etude de l’equilbre
et du movement des solids elastiques, Gauthier-Villars, Paris (in
drawn: French).
• There is a good correlation in some cases between the experi-
Chen, Q. (2007). “An experimental study on characteristics and behavior of
mental measurements and the FE analysis results for small val- reinforced soil foundation.” Ph.D. dissertation, Louisiana State Univ.,
ues for load–settlement behavior. However, laboratory test Baton Rouge, LA.
stress values are a little bit smaller than FEM in some condi- Cicek, E. (2011). “Analysis of strip footings on geosynthetic reinforced
tions, but for both of them, reinforcements affect the stress dis- sand.” Ph.D. dissertation, Institute of Science in Ataturk Univ.,
tribution similarly. Erzurum, Turkey.
• Long reinforcement length (L = 5B) reduces the additional ver- Cicek, E., Guler, E., and Yetimoglu, T. (2014). “Comparison of measured
and theoretical pressure distribution below strip footings on sand
tical stress more than the reinforcement that is only as wide as
soil.” Int. J. Geomech., 10.1061/(ASCE)GM.1943-5622.0000347,
the base plate. 06014009.
• Each reinforcement divides the active triangular zone under
Cicek, E., Guler, E., and Yetimoglu, T. (2015). “Effect of reinforcement
the loaded plate into sections. length for different geosynthetic reinforcements on strip footing on sand
• As a general conclusion, it can be stated that for the calculation soil.” Soils Found., 55(4), 661–677.
of the vertical additional stress for small loadings in reinforced DeMerchant, M. R., Valsangkar, A. J., and Schriver, A. B. (2002). “Plate
soil, the Westergaard approach can generally give the maxi- load tests on geogrid reinforced expanded shale lightweight aggregate.”
mum degree of stress expected. But for the levels below the re- Geotext. Geomembr., 20, 173–190.
inforced zone, stresses predicted by the Boussinesq approach Ghazavi, M., and Lavasan, A. A. (2008). “Interference effect of shallow
foundations constructed on sand reinforced with geosynthetics.”
can be exceeded. Furthermore, the presence of reinforcements
Geotext. Geomembr., 26(5), 404–415.
transfers the vertical stress to deeper levels, so if the soil at a Guido, V. A., Chang, D. K., and Sweeney, M. A. (1986). “Comparison of
deeper level is stiffer, the stress increase will cause less geogrid and geotextile reinforced slabs.” Can. Geotech. J., 23(4),
settlement. 435–440.
• For smaller loads on the footings, a difference in the behavior Huang, C. C., and Menq, F. Y. (1997). “Deep-footing and wide-slab effects
exists depending on the reinforcement and the reinforced zone in reinforced sandy ground.” J. Geotech. Geoenviron. Eng., 10.1061
depth. /(ASCE)1090-0241(1997)123:1(30), 30–36.
• The FE and theoretical assumptions have limitations to show Huang, C. C., and Tatsuoka, F. (1990). “Bearing capacity reinforced hori-
zontal sandy ground.” Geotext. Geomembr., 9(1), 51–82.
the real stress-strain behaviors for reinforced soils. Therefore,
Hufenus, R., Rueegger, R., Banjac, R., Mayor, O., Springman, S. M., and
new approaches should be determined via experimental tests Bronnimann, R. (2006). “Full-scale field tests on geosynthetic rein-
in future studies. forced unpaved roads on soft subgrade.” Geotext. Geomembr., 24(1),
21–37.
Kumar, A., and Saran, S. (2003). “Bearing capacity of rectangular footing
on reinforced soil.” Geotech. Geol. Eng., 21(3), 201–224.
Acknowledgments Lade, P. V., and Duncan, J. M. (1976). “Stress-path dependent behavior of
cohesionless soil.” J. Geotech. Engrg., 102(1), 51–68.
The authors thank the Bogazici University Research fund and Lambrechts, J. R., and Leonards, G. A. (1978). “Effects of stress history on
deformation of sand.” J. Geotech. Engrg., 104(11), 1371–1387.
the Scientific and Technical Research Council of Turkey
Ling, H. I., and Liu, H. (2003). “Finite element studies of asphalt con-
(TUBITAK). Additionally, the first author thanks Hacettepe crete pavement reinforced with geogrid.” J. Eng. Mech., 10.1061
University Research Projects Management System (Project /(ASCE)0733-9399(2003)129:7(801), 801–811.
FDS-2015-8791). Lovisa, J., Shukla, S. K., and Sivakugan, N. (2010). “Behaviour of pre-
stressed geotextile-reinforced sand bed supporting a loaded circular
footing.” Geotext. Geomembr., 28(1), 23–32.
References Michalowski, R. L. (2004). “Limit loads on reinforced foundation soils.” J.
Geotech. Geoenviron. Eng., 10.1061/(ASCE)1090-0241(2004)130:
Abu-Farsakha, M., Akond, I., and Chen, Q. (2016). “Evaluating the per- 4(381), 381–390.
formance of geosynthetic-reinforced unpaved roads using plate load Milligan, G. W. E., Fannin, R. J., and Farrar, D. M. (1986). “Model and full-
tests.” Int. J. Pavement Eng., 17(10), 901–912. scale tests of granular layers reinforced with a geogrid.” Proc., Third

© ASCE 06018005-11 Int. J. Geomech.

Int. J. Geomech., 2018, 18(3): 06018005


Int. Conf. on Geotextiles, International Geosynthetics Society, Jupiter, Wayne, M. H., Han, J., and Akins, K. (1998). “The design of geosynthetic
FL, 1, 61–6. reinforced foundations.” Proc., Annual Convention and Exposition,
Moghaddas Tafreshi, S. N., and Norouzi, A. H. (2012). “Bearing ASCE, Reston, VA, pp. 1–18.
capacity of a square model footing sand reinforced with shredded Westergaard, H. M. (1938). “A problem of elasticity suggested by a prob-
tire—An experimental investigation.” Constr. Build. Mater., 35, lem in soil mechanics: Soft material reinforced by numerous strong hori-
547–556. zontal sheets.” The mechanics of solids, Macmillan, New York.
Naval Facilities Engineering Command. (1982). Soil mechanics: Design Yetimoglu, T. (1998). “Bearing capacity of footings on vibratory com-
manual 7.1, Dept. of the Navy, Alexandria, VA. pacted sand.” Teknik Dergi, 9(1), 479–482.
PLAXIS 8 [Computer software]. PLAXIS, Delft, Netherlands. Yetimoglu, T., Wu, J. T. H., and Saglamer, A. (1994). “Bearing capacity of
Schlosser, F., Jacobsen, H. M., Juran, I. (1983). “Soil reinforcement.” rectangular footings on geogrid-reinforced sand.” J. Geotech. Engrg, 10
Genereal Rep., Eighth European Conf. on Soil Mechanics and .1061/(ASCE)0733-9410(1994)120:12(2083), 2083–2099.
Foundation Engineering, Balkema, Helsinki, pp. 83–103. Zornberg, J. G. (2012). “Geosynthetic-reinforced pavement systems.”
Sireesh, S., Sitharam, T. G., and Dash, S. K. (2009). “Bearing capacity of Proc., Fifth European Geosynthetics Conf., EuroGeo5, 16-19 September,
Downloaded from ascelibrary.org by Indian Institute of Technology, Delhi on 01/10/18. Copyright ASCE. For personal use only; all rights reserved.

circular footing on geocell-and mattress overlying clay bed with void.” Spanish Chapter of the International Geosynthetics Society, Tilesa Kenes,
Geotext. Geomembr., 27(2), 89–98. Spain, Keynote Lectures, 49–61.

© ASCE 06018005-12 Int. J. Geomech.

Int. J. Geomech., 2018, 18(3): 06018005

Vous aimerez peut-être aussi