You are on page 1of 15

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/332344762

Soil spatial variability impact on the behavior of a reinforced earth wall

Article  in  Frontiers of Structural and Civil Engineering · June 2020


DOI: 10.1007/s11709-020-0611-x

CITATIONS READS

3 474

3 authors:

Adam Hamrouni Daniel Dias

12 PUBLICATIONS   85 CITATIONS   
Antea Group
347 PUBLICATIONS   5,142 CITATIONS   
SEE PROFILE
SEE PROFILE

Badreddine Sbartai
Badji Mokhtar - Annaba University
76 PUBLICATIONS   236 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Soil improvement View project

Assessment of the seismic risk of the urban areas in Algeria View project

All content following this page was uploaded by Adam Hamrouni on 07 November 2020.

The user has requested enhancement of the downloaded file.


Front. Struct. Civ. Eng.
https://doi.org/10.1007/s11709-020-0611-x

RESEARCH ARTICLE

Soil spatial variability impact on the behavior of a


reinforced earth wall
Adam HAMROUNIa* , Daniel DIASb,c, Badreddine SBARTAId
a
Laboratory InfraRES, Mohammed Cherif Messaadia University, SoukAhras 41000, Algeria
b
School of Automotive and Transportation Engineering, Hefei University of Technology, Hefei 230009, China
c
Antea Group, Antony 92160, France
d
Department of Civil Engineering, University of Annaba, Annaba 23000 & LMGHU Laboratory, University of Skikda, Skikda 21000,
Algeria
*
Corresponding author. E-mail: hamrouni.adam@yahoo.fr

© Higher Education Press and Springer-Verlag GmbH Germany, part of Springer Nature 2020
ABSTRACT This article presents the soil spatial variability effect on the performance of a reinforced earth wall. The
serviceability limit state is considered in the analysis. Both cases of isotropic and anisotropic non-normal random fields
are implemented for the soil properties. The Karhunen-Loève expansion method is used for the discretization of the
random field. Numerical finite difference models are considered as deterministic models. The Monte Carlo simulation
technique is used to obtain the deformation response variability of the reinforced soil retaining wall. The influences of the
spatial variability response of the geotechnical system in terms of horizontal facing displacement is presented and
discussed. The results obtained show that the spatial variability has an important influence on the facing horizontal
displacement as well as on the failure probability.

KEYWORDS reinforced earth wall, geosynthetic, random field, spatial variability, Monte Carlo simulation

1 Introduction convert the random field discretization into a finite number


of random variables so that the probability of failure can be
In recent years, the probabilistic analysis of geotechnical determined. Several methods to generate random fields are
structures has devoted a great deal of effort. Various already available in literature like: Expansion Optimal
uncertain parameters were modeled using random vari- Linear Estimation (EOLE), Stochastic Spectral (SS),
ables based on some simplified methods in which the soil Karhunen-Loève (K-L)), Random Finite Element Method
is considered as a uniform material. However, in nature, as (RFEM), stochastic response surface method (SRSM),
a result of deposition and post-deposition processes, soil polynomial chaos expansion (PCE). The use of these
parameters (elastic properties, shear strength parameters, methods has allowed several authors to study the effect of
etc.) vary spatially (horizontal and vertical direction). The soil spatial variability on the behavior of geotechnical
spatial variability of the soil properties will then affect the structures [1–18].
behavior of geotechnical structures (bearing capacity, A mechanically stabilized earth (MSE) wall is a
foundation settlement, slope stability, etc.). Therefore, the composite material formed by the combination of soil
need to represent the soil parameters considering random and metallic or synthetic strips able to sustain significant
fields is essential. Advanced probabilistic approaches were tensile loads. The reinforcing strips give to the soil mass
proposed in the literature. The deterministic models used to anisotropic properties in the direction perpendicular to the
represent geotechnical structures are frequently based on reinforcement Schlosser and Elias [19]. To improve the
the finite difference method (FDM) or the finite element optimization of the design method, it is essential to
method (FEM). In these approaches, it is necessary to understand the behavior of such structures. Several studies,
experimental, theoretical or numerical, were carried out
Article history: Received Feb 6, 2019; Accepted Aug 11, 2019 with this objective. Using numerical studies, two and
2 Front. Struct. Civ. Eng.

three-dimensional models based on finite elements or finite MCS methodology is used to determine the probability
differences (Abdelouhab et al. [20,21], Riccio et al. [22]) density function (PDF) of the system response variability
allowed to analyze the deformation and the influence of of the reinforced soil retaining wall. The aim of this study
soil parameters of reinforced soil walls. is to determine the effect of the different governing
The reliability analysis of reinforced retaining walls statistical characteristics (autocorrelation distances and
permits to consider the uncertainties due to the soil coefficient of variation) on the facing maximum horizontal
properties variability [23–28]. These authors have included wall displacement.
in their studies the uncertainties related to the reinforce-
ment properties. Chalermyanont and Benson [29] evalu-
ated the Hasofer-Lind reliability index through the 2 Presentation of the numerical model
application of Monte Carlo simulations (MCSs). Relia-
bility diagrams were also developed to select the Abdelouhab et al. [20,21] used the Lagrangian explicit
geosynthetic material which is able to meet the safety finite-difference code FLAC2D to study the behavior of a
requirements [30]. However, the representation of such a mechanically reinforced earth wall reinforced by geosyn-
realistic structure for reliability analysis by analytical thetic strips. The wall of reference is 6 m high and is made
mechanisms is not satisfactory. So the use of the of 4 superimposed panels (beam elements). This wall is
continuum mechanics (FEM or FDM) is essential. Thurner reinforced by 8 levels of 4 m long reinforcement layers
and Schweiger [31] found some difficulties to apply (strip elements) (Fig. 1(a)). The real cruciform geometry of
reliability calculation methods such as the first order the panels (Fig. 1(b)), leads to a complex geometry of the
reliability method (FORM) or the second order reliability wall facing. A real panel have 4 points of connection for
method (SORM) when the performance is not explicitly the strips. This three-dimensional geometry and the
available in an analytical form (such as the displacements staggered layout are simplified into a two-dimensional
of the reinforced retaining wall). Sayed et al. [25] assessed model using some simplifications. The panels are modeled
the maximum failure probability of MSE using the RSM; as rectangular plates of 1.5 m by 1.5 m. The simplification
this model is based on numerical simulations using FEM of the geometry permits to use a two-dimensional model
model. Lin et al. [32] analyzed the probabilistic maximum with continuous reinforcements (plane strain configura-
deformations of three reinforced soil walls using a tion). The parameters of these reinforcements are calcu-
combination of a FDM and the RSM technique. This lated by homogenizing their geometrical characteristics
work was extended by Yu and Bathurst [33] using different (Fig. 1(c)). The reinforcement surface is the ratio between
geosynthetic reinforcement configurations. The response the individual reinforcement area multiplied by the number
surface methodology optimized by genetic algorithm was of reinforcement and the model width.
used by Hamrouni et al. [34] to study the reliability of For the boundary conditions, the horizontal displace-
reinforced earth walls. ments are blocked at the vertical limits of the model.
In the literature, very few studies have been done to Vertical and horizontal displacements are blocked at the
investigate the effect of spatial variability on the response model bottom.
of a reinforced earth wall. Previous studies only considered Concerning the interface friction at the soil/reinforce-
random variables. However, Dodagoudar et al. [35] ment, the apparent coefficient of friction is used in order to
attempted to consider the influence of soil spatial take into account the effect of the dilatancy and the
variability on reinforced earth walls. A linear elastic reinforcements shape.
behavior was assumed for the soil and the soil Young The model consists of two soils (Fig. 1(a)). The
modulus was considered as a 2D isotropic random field embankment soil is made of a uniform fine sand, known
and supposed to follow a normal distribution. Ning and as Hostun RF sand [38,39]. The constitutive model used
Bathurst [36] investigated random fields of soil properties for this sand is linear elasticity with perfectly plasticity
on reinforced soil slopes. The RFEM developed by (shear failure criterion of Mohr-Coulomb type). This
Griffiths and Fenton [37] was used. Their work was only constitutive model is characterized by five parameters: 2
focused on the reinforced slopes stability at the ultimate elastic parameters (E: Young’s modulus, υ: Poisson’s ratio)
limit state. and 3 plastic parameters (φ: friction angle, C: cohesion, ψ:
This article presents a probabilistic analysis at the dilatancy angle). The parameters of this constitutive model
serviceability limit state (SLS) of a reinforced earth wall. were defined for the embankment soil by calibration on
The spatial variability of the soil is modeled using non- triaxial tests carried out under confinement of 30, 60, and
Gaussian anisotropic random fields. Random fields are 90 kPa [20,21]. For the soil foundation, a linear elastic
discretized into a finite number of random variables using model is used. This model is characterized by two elastic
the K-L expansion. The facing maximum horizontal parameters (E: Young’s modulus and υ: Poisson’s ratio)
displacement is used to represent the system response. (Table 1).
The deterministic model considered is based on a finite Beam elements are used to model the concrete panels.
difference numerical approach (FLAC2D [20,21]). The These structural elements are isotropic and linear elastic.
Adam HAMROUNI et al. Soil spatial variability impact on the behavior of a reinforced earth wall 3

Fig. 1 Presentation of the studied numerical model. (a) Geometry of the earth reinforced wall; (b) cruciform panel; (c) two-dimensional
model; (d) numerical model adopted.

Table 1 Mechanical and geometrical properties of the model


properties (unity) fill soil foundation concrete panels reinforcement GS 50 Abdelouhab et al. [41]
E (Young’s modulus) (MPa) 50 50 15000 2500
υ (Poisson’s ratio) 0.3 0.3 0.2 –
φ (friction angle) (°) 36 – – –
Ψ (dilation angle) (°) 6 – – –
C (cohesion) (kPa) 0 – – –
3
density (t/m ) 1.56 2 2.5 –
width (m) – – – 0.1
thickness (mm) – – – 3
strip tensile yield-force limit (kN) – – – 100
maximum compressive strength (kPa) – – – 0
tensile failure strain limit of strip (%) – – – 12

They can resist to bending moments, compressive, and considered by reducing artificially the beam section
tensile efforts (Table 1). keeping the real inertia moment.
The beams were represented by pins to reproduce the Geosynthetic strip type of reinforcement containing
flexibility of the real wall facing. Wall panel spacers made high-tenacity polyester yarns protected by polyethylene
of ribbed elastomeric pads are inserted to prevent the sheath are studied and modeled by the use of structural
panels from having contact points and peeling of the elements of “Strip” type [40]. Its properties are presented
concrete. In numerical modeling, the elastomeric pads are in Table 1. These elements are specially designed to
4 Front. Struct. Civ. Eng.

simulate the behavior of reinforcing strips used in of the second soil layer and installation of the first strip
reinforced earth embankments. Strip elements consider between the two layers of the reinforced backfill
tensile strength and compression but cannot withstand (equilibrium under self weight).
bending moments [41]. Stage 2. Placement of the third and the fourth layer and
The soil and the beam elements are linked using installation of the second strip between the two layers of
interface elements in order to simulate the frictional the reinforced backfill (equilibrium under self-weight).
interaction between the backfilling soil and the concrete Stage 3. Set up of the second beam, of the fifth and sixth
facing panels (Table 2). The FLAC recommendations are layer and installation of the third strip between the two
used to calculate the normal and the shear stiffness. The layers of the reinforced backfill.
value of the friction angle is estimated to be equal to 2/3 of These phases are repeated up till 6 m high.
the soil friction angle [40].
A nonlinear shear failure envelope that varies as a 2.1 Results (displacements)
function of the confining pressure defines the shear
behavior of the strip/soil interface. The interface para- Using the reinforcement reference parameters, the max-
meters are the shear stiffness kb at the soil/strip interface imum displacement calculated on the MSE wall is equal to
and the apparent friction coefficient f*. The values of the 7.29 cm. This value is located just behind the concrete
shear stiffness (kb) and parameter the friction coefficient facing, at the level of the 3rd reinforcing bed (Fig. 2). The
(f*) used were defined by calibration on laboratory pullout extensibility of the synthetic strips (low elastic modulus)
tests (Abdelouhab et al. [41]). According to this calibra- leads to this high wall deformation.
tion, the parameters used in the numerical modeling are
presented in Table 2.
A non-uniform mesh composed of around 2500 zones is
used (Fig. 1(d)). These zones increase gradually away from
the ends of the reinforced area until they become equal to
0.80 m in areas far from the retaining wall but below the
maximum limit (1 m) proposed by Der Kiureghian and Ke
[42]. In the vertical direction, the backfill is divided into 24
areas refined uniformly. This mesh have a constant vertical
width equal to 0.25 m and does not exceed a maximum of
0.50 m vertically (recommendation proposed by Der
Kiureghian and Ke [42]). These values are used in Fig. 2 Horizontal facing displacement of the MSE Wall (6 m
accordance with the projection condition of a random height).
field on the finite difference mesh.
The compaction of the soil layers is not considered in the 2.2 Determination of the significant soil parameters on the
reference study. A calculation considering compaction by wall behavior
an equivalent loading pressure is performed in the
parametric study. To reduce the number of variables to be considered in the
Calculations were made under drained conditions as reliability approach, a parametric study was used to define
there is no water presence. the input parameters influence on the wall behavior. The
To reproduce the real building steps, the setting up of the output parameter considered in this study is the facing
embankment is modeled by layers of 0.375 m thickness in maximum horizontal displacement noted as Uhmax =
3 stages: 7.29 cm which is calculated with the reference values. A
Stage 1. Set up of the first concrete panel of the first and parametric study is performed here to check the sensitivity

Table 2 Concrete panel/soil interface and Soil/reinforcements interface characteristics


parameter concrete panel/soil interface soil/reinforcements interface
constitutive model Coulomb sliding Coulomb sliding
normal stiffness (MPa) 1000 –
shear stiffness (MPa) 1000 –
friction angle at panel/soil interface (°) 24 –
initial apparent friction coefficient at the soil/strip interface “f 0*” – 1.2
minimum apparent friction coefficient at the soil/strip interface “f 1*” – 0.6
shear stiffness at the soil strip interface kb (MN/m /m)2 – 0.22
Adam HAMROUNI et al. Soil spatial variability impact on the behavior of a reinforced earth wall 5

of the facing maximum horizontal displacement to the five fields generated in this study are two-dimensional and
parameters of the linear elastic perfectly plastic constitutive follow a lognormal distribution law. The mean and
model (E, υ, c, φ, and ψ) and of the panel/soil interface standard deviation of the field in (Z) normally distributed
friction angle (°). When studying the influence of one are named μZ, σZ, respectively, while the average (μln) and
parameter, the other parameters are kept at their reference standard deviation (σln) of the lognormal Z field are given
value. A total of 18 simulations corresponding to three by
different values of each parameter see Table 3 are studied, qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
as shown in Fig. 3. The Young’s modulus, the Poisson’s ln ¼ lnð1 þ COV 2Z Þ, (1)
ratio, the dilation angle and the friction angle at the panel/
soil interface have a negligible influence on the Uhmax
ln ¼ lnðZ Þ – 0:5ln : (2)
value (difference versus the reference case inferior to 2%).
However, two parameters have an influence on the output Among the available methods used to generate corre-
parameters: the friction angle φ and the soil unit weight γ. lated parameters randomly, the Cholesky decomposition
The soil friction angle influence is higher than the one of [43,44] and K-L expansion [45,46] are often used. Since
the soil unit weight respectively of 123% and 20%. In the the considered problem deals with spatially correlated
following study, only the friction angle φ will be fields, the K-L decomposition is recommended and was
considered as a random variable. employed in this study.
The soil Young’s modulus has a negligible effect on the The random field defined by the method of expansion in
horizontal displacement Uhmax. This is due to the rigidity series of K-L presented by Spanos and Ghanem [45] is
of the reinforcements which is stiffer than the soil one. used to express it through a truncated series using M
number of terms [47,48]. The field is then divided into a
determined portion represented by the average, and a
3 Random field method used for random element depending of fi eigenvalues and of li
discretization functions of the corresponding covariance function:
 XM pffiffiffiffi 
As already mentioned, random fields are defined by a ZðX ,Þ  exp ln þ i¼1
l i fi ðX Þ i ðÞ , (3)
distribution law and an autocorrelation function. Random

Table 3 Influence of soil parameters on the wall behavior


parameters value reference variation DU/Uref (%)
min max
Young’s modulus (MPa) 50 37.5 62.5 0.49 – 0.23
Poisson’s ratio 0.3 0.225 0.375 0.329 – 1.83
friction angle (°) 36 27 45 122.9 – 40.38
dilation angle (°) 6 4.5 7.5 0.57 – 1.17
unit weight (kN/m3) 15.6 11.7 19.5 – 18.97 20.05
friction angle at panel/soil interface (°) 24 18 30 1.64 – 0.82

Fig. 3 Influence of the input parameters on the maximum displacement of the wall face.
6 Front. Struct. Civ. Eng.

with X the vector of coordinates of the field points where and autocorrelation function. However, the accuracy of the
the values of the Z property are to be generated; θ is the random fields generated depends on the number of terms M
variable of the random draw and i(θ) is a normal random of the K-L series. To select the number of terms M to be
variable centered and reduced. used in the random field discretization of the friction angle,
The main advantage of this method lies in the the average distribution functions and autocorrelation are
decoupling of spatial and stochastic variables (x and θ, drawn for many realizations of this random field as a
respectively). The eigenvalues and eigenfunctions (fi and function of the number of terms M, and compared to the
li, respectively) of the covariance function C (X1, X2) are target functions.
solutions of the following equation: A random field generation of the friction angle is
proposed using a lognormal distribution with a mean value
!DCðX 1,X 2ÞfiðX 1ÞdX 1 ¼ lifiðX 2Þ: (4) μj = 36° and a covariance COVj = 15%. An exponential
autocorrelation function with autocorrelation distances
This integral can be analytically solved for some types of equal to 1 m in both directions is chosen to illustrate the
autocorrelation function. The analytical solution for the influence of the number of terms M. A large number of
exponential autocorrelation function here adopted is given outputs (350) are generated for each variation of the
by Ghanem and Spanos [49]. The stochastic dependency number of terms M. For each case, the cumulative
appears only through the random variables i. The spatial probability function and the autocorrelation function of
correlation is defined by the eigen modes (fi and li) of the the generated values are plotted. The average of these
covariance kernel (Eq. (4)). functions is then compared to the target theoretical
The expression of the covariance function appears in functions.
Eq. (5b) and introduces both the variance 2 of the Figure 4 shows a comparison between the average
Gaussian random field and the correlation length L. Each cumulative probability functions (CDF) using 350 random
term C12 of the covariance matrix C is the value of the field realizations for different numbers of terms M and the
covariance function computed between the two nodes 1 target cumulative probability function. Figure 5 shows a
and 2 of the spatial mesh, whose position is given by X1 comparison between the autocorrelation function of the
and X2. target (Eq. (7)) and of the average autocorrelation functions
  generated in the horizontal direction after 350 random field
kX 1 – X 2 k realizations (Eq. (8), considering different numbers of
C 12 ¼ CðX 1 ,X 2 Þ ¼ V exp – : (5a)
L terms M). The good concordance between these two curves
has permitted to validate the generated stochastic field.
In our problem, two spatial directions (horizontal and
vertical) are distinguished. Therefore, two correlation
lengths are also introduced. When a 2D field is considered,
the covariance function is given as:
 
jx1 – x2 j jy1 – y2 j
C 12 ¼ ln exp –
2
– : (5b)
Llnx Llny
Equation (5b) is the covariance between nodes 1 and 2,
for which (x1, y1) are the coordinates of node 1 (id. for node
2). Llnx and Llny are the autocorrelation distances in the
horizontal direction and in the vertical one, respectively.
Note that the number of terms M is selected depending
on the targeted accuracy. The estimated error εrr(X) values
generated by the K-L expansion series for a number of
terms is given by M [50].
 X
1
εrr ðX Þ ¼ 1 –
M
l f2 ðX Þ:
i¼1 i i
(6)
ln
Fig. 4 Comparison between the distribution function and the
average functions generated using 350 realizations of the friction
angle random field, for several numbers of terms M.
4 Choice of the number of terms of the K-L
series
Figures 4 and 5 show that more the number of K-L terms
When a random field is generated, the values of this field increases, more the functions generated get closer to the
should reflect the initial theoretical statistical distribution target theoretical functions, especially for 200 and 250
Adam HAMROUNI et al. Soil spatial variability impact on the behavior of a reinforced earth wall 7

by μln and σln, respectively. The dependency of the friction


angle values for two distinct points of the soil mass is
represented by an exponential autocorrelation function.
To model a realization of the friction angle random field
and pass it to the proposed numerical model, a 2D field
representing the soil mass is discretized into Nx  Ny
surface elements, where Nx and Ny are the number of
elements in the x and y directions. With the implementation
of a function using MATLAB, a friction angle value is
given to each soil element. In fact, each element is assumed
to be homogeneous and is affected by a single friction
angle value at its center using the K-L method of expansion
in series, developed by Ghanem and Spanos [49]. In other
Fig. 5 Comparison between the theoretical autocorrelation words, for each realization of the random field, a given
function and the average autocorrelation functions generated configuration of the soil is generated: a friction angle value
using 350 realizations of the friction angle random field, for several is assigned to the center of each of its elements depending
numbers of terms M. of its position in space, following the distribution law and
the selected autocorrelation function. It should be noted
terms M. A number of terms M equal to 250 is adopted in that for each realization, the values of the random field (φ)
the following calculations. were determined at the centroid of each element of the
The domain and the model mesh must be selected to deterministic mesh (FLAC2D) using the middle method
meet certain conditions (fine mesh in areas near the facing). [55].
A transfer of the stochastic data to the mesh is necessary. Finally, it should be mentioned that a link between
For this purpose, two transfer methods exist: FLAC2D and MATLAB was performed to automatically
1) The middle method used by Refs. [42,51,52]: in this exchange the data in both directions and thus to decrease
method, the random field is represented by its value at the the computation time.
centroid of the finite element. One seeks the position of the The spatial variability of the soil friction angle can be
centroid in the stochastic mesh and assigned to the finite defined by a lognormal distribution with μj = 36 °, COVφ =
element mesh the value of the field corresponding to the 15% and a second-order exponential function with an
position of its centroid in the stochastic mesh. autocorrelation distance equal to 1.0 m in both directions.
2) The local averaging method (or spatial averaging) The embankment is represented by a random field (Fig. 6).
proposed by Vanmarcke [53,54]: it consists in allocating to The field is generated in two dimensions using the
each cell of the domain a value determined by averaging following dimensions: 20 m in the longitudinal direction X
the stochastic mesh field values that fall within the finite and 6 m in the vertical direction Y.
element mesh. This domain is discretized into elements of 0.25 m 
Der Kiureghian and Ke [42] noted that the method of the 0.25 m.
middle over-represents the variability of the field while the
local averaging method tends to under-represent the true
variability of the field. They also demonstrated that the 6 Probabilistic analysis with Monte Carlo
spatial averaging method is more suitable for Gaussian
fields. Popescu [55] conducted a comparative study 6.1 Selection of the optimal number of MCS
between these two methods and concluded that the middle
method is best suited for non-Gaussian random fields since The effect of the soil spatial variability on the wall facing
it keeps the probability distribution of the initial field. In horizontal displacement is done using the Monte Carlo
this study, the middle method was used since the random (MC) method for several combinations of autocorrelation
field representative soil properties are considered as non- distances in the two directions. This probabilistic analysis
Gaussian fields. requires, for each combination studied, the generation of a
large number of samples, each represented by a random
field realization. The maximum facing horizontal displace-
5 Random field of soil properties ment (Umax ) is then calculated for each of these
achievements and the average (μU) and standard deviation
The numerical model proposed in this study focuses on the (σU) of all values of the maximum facing horizontal
spatial variability of the soil friction angle. The corre- displacement are then compared. The number Ns of
sponding field is expected to follow a lognormal distribu- simulations required is the one for which the values of
tion, the soil friction angle being a strictly positive the two statistical moments converge.
quantity. The mean and standard deviation are denoted Table 4 shows the values of the first critical pressure
8 Front. Struct. Civ. Eng.

Fig. 6 Transformation of the random field data to the numerical model.

Table 4 Mean μ(u) and coefficient of variation COV (U) of the maximum facing horizontal displacement for different numbers of MC simulations
and number of terms M of the K-L series.
Ns M
50 100 150 200 250
μ (U) (cm) COV (U) (%) μ (U) (cm) COV (U) (%) μ (U) (cm) COV (U) (%) μ (U) (cm) COV (U) (%) μ (U) (cm) COV (U) (%)
50 7.4572 12.29 6.9660 15.47 7.7780 12.84 8.7580 14.44 9.6880 16.15
100 7.5887 12.56 6.2480 13.65 8.5020 13.10 7.1200 14.84 8.4020 14.20
200 7.6827 12.30 8.5770 14.71 6.8680 14.21 7.2020 14.27 8.7300 14.18
300 7.7630 12.68 8.7410 15.18 5.9760 14.68 8.2630 14.48 10.7700 14.22
400 7.7478 12.88 8.7480 14.92 8.5550 14.96 10.8400 14.80 8.4070 13.88
500 7.7528 13.03 9.4610 14.84 7.2360 14.98 8.7240 14.59 6.6880 14.04

statistical moments obtained for different simulation distribution law with an autocorrelation distance of 1 m.
numbers Ns, while varying the number of terms of K-L, The mean and coefficient of variation of the soil internal
M. The corresponding curves are shown in Figs. 7 and 8. friction angle φ adopted are equal to 36 degrees and 15%,
The optimal number of MCSs is equal to 300. This respectively. Dark regions correspond to small values of φ
number will be used for the following calculations. and the lighter ones to larger values where the values vary
between 28° and 44°. Figure 10 shows a realization of a
6.2 Effect of the soil spatial variability on the facing random field of the anisotropic soil internal friction angle
horizontal displacement (i.e., Lx = 10 m; Ly = 1 m).
The considered combinations are as follows:
To study the soil spatial variability effect on the facing 1) Case 1D: Lx ≠ Ly
horizontal displacement, several combinations of the These cases represent stratified soils knowing for each
autocorrelation distances (L) in the two directions were layer its friction angle value. The layers are parallel to the
used. In fact, when a small distance autocorrelation is plane formed by the directions for which a large
considered in a given direction, the soil friction angle autocorrelation distance is adopted.
values are highly correlated and quickly change from one 2) Case 2D: Lx = Ly
point to another. However, when the distance is large, the These cases represent heterogeneous soils with friction
values change slowly with the distance and can become angle varying from one point to another.
homogeneous. Several cases of autocorrelation distances (L) are
Figure 9 illustrates a realization of an isotropic random considered using L = 1, 2, 5, and 10 m.
field of the soil friction angle according to a lognormal The two first statistical moments of the maximum facing
Adam HAMROUNI et al. Soil spatial variability impact on the behavior of a reinforced earth wall 9

Fig. 7 Mean maximum horizontal displacement of the facing as a function of the simulation number MC, for several numbers of terms
M of the series K-L.

Fig. 8 Coefficients of variation of the facing maximum horizontal displacement in terms of the simulation number MC, for several
numbers of terms M of the series K-L.

Fig. 9 Sample of internal friction angle random field (degrees).


10 Front. Struct. Civ. Eng.

Fig. 10 Sample of isotropic internal friction angle random field (degrees) (Lx = 10 m, Ly = 1 m).

displacement are calculated for 300 realizations of the 2) The movement of the ground is greater in areas where
random field for each of the 3 considered cases. The results there is a low value of friction angle and especially in the
are then compared and evaluated. reinforced area. For intermediate autocorrelation distances
(L ≈ length of strips), agglomerates of weak and strong
6.3 Effect of the soil spatial variability on the average areas are created. Their distribution within the reinforced
maximum facing displacement zone varies from one realization to another.
3) For larger autocorrelation distances: for the 1D case,
Figure 11 shows the variation of the average on the the layers and soil columns form alternating small and
maximum facing displacement as a function of the distance large values of the friction angle. The soil zones with a
autocorrelation for different random fields types (1D and high friction angle value will constitute a stabilizing barrier
2D). for the soil movements. It induces a small value of the
All curves have the same shape. With the increase of the maximum displacement mean of the facing and a more
autocorrelation distance, the average of the maximum stable structure. As for the 2D case, the soil becomes more
displacement increases to reach a maximum at a facing homogeneous especially for the case where Lx = Ly =
height of L ≈ H/2, and then decreases. This maximum 1000 m. In this case, the mean of the maximum facing
indicates that the soil behind the facing is at least stable for displacement has a difference of 5.7% compared to the
this autocorrelation distance. Similar results are observed deterministic displacement.
by Ref. [6] for shallow foundations. This can be explained 4) The cases considering a spatial variability in a single
as follows: direction (vertical one), lead to values very close to the
1) For small autocorrelation distances, the soil varia- deterministic displacement of the facing but lower than
bility is very high and it results in a mix of small and large those obtained for the case 1D according to the horizontal
friction angle values. This gives a displacement close to the direction. This is due to the fact that the variation of the
deterministic value. friction angle along the vertical axis leads to “layered”

Fig. 11 Variation of the average facing maximum displacement in terms of autocorrelation distance for 1D and 2D random fields.
Adam HAMROUNI et al. Soil spatial variability impact on the behavior of a reinforced earth wall 11

vertical soil, perpendicular to the reinforcement plane of of soil, vary considerably from one realization to other,
symmetry, which facilitates their collapse and makes this leading to more scattered values of the maximum facing
case as safe as the case according to the horizontal displacements.
direction. Figure 12 shows that the greatest variation of the
5) Finally, for the case 2D (Lx = Ly), the mean maximum maximum facing displacement is obtained for the case 2D
displacement of the facing is greater than the one of the 1D and is followed by the 1D cases according to the Y axis.
case according to the horizontal and vertical direction. This The smaller variation is observed for the case 1D
is due to the introduction of the vertical variability which according to the X axis. This can be explained in the
reduces the deformations of the facing, as explained above. following way:
In contrast, the 2D case reveals lower movements of the For very large autocorrelation distances (Lx = Ly =
facing compared to the 1D case. 1000 m), the standard deviation of the maximum facing
6) Observing the effect of the spatial variability displacement tends to a maximum constant value which
direction on the mean maximum displacement of the corresponds to the case of a random variable. In fact, the
facing reveals that taking into account the same variation in friction angle values for one realization case are highly or
the angle of friction in the two directions leads to a less even perfectly correlated, thus generating a homogeneous
stable facing. However, considering spatial variability in ground mass and affected everywhere by the same friction
the longitudinal and/or vertical directions improves the angle value. However, the friction angle values depend on
deformations of the facing and therefore a smaller value of the theoretical statistical distribution adopted. The soil
the displacement is required to avoid the failure of the mass thus has a wide variation of the friction angle values
structure. which has a direct effect on the system response. (i.e.,
maximum facing displacement).
6.4 Effect of the soil spatial variability on the standard The cases considering a spatial variability in one
deviation of the maximum facing displacement direction (case 1D), exhibit a smaller variability of the
maximum facing displacement than the case of “random
Figure 12 shows that the variation of the standard deviation variables” discussed above, i.e., Lx = Ly = 1000 m. In fact,
of the facing maximum displacement depends of the by introducing a variation of the friction angle in a given
distance autocorrelation. All curves have the same direction, all the maximum facing displacement values will
tendency regardless the considered variability. The stan- tend to increase, thus excluding the small displacement
dard deviation increases with the autocorrelation distance values. Consequently, the maximum facing displacement
increase. variability will decrease compared to the case of homo-
geneous soil mass.

6.5 Influence of the random field coefficient of variation

Figure 13 presents the PDF of the maximum facing


displacement for three different configurations of the

Fig. 12 Variation of the standard deviation facing maximum


displacement versus the autocorrelation distance (1D and 2D
random fields).

Indeed, for small autocorrelation distances, the gener-


ated friction angle values are very heterogeneous, i.e., a
mixture of large and small values, which leads to an
“averaging” of the backfill properties values. The resulting
facing maximum displacements are more homogeneous. Fig. 13 Effect of the random field’s coefficient of variation
For larger autocorrelation distances, the friction angle (friction angle) on the PDF of the maximum facing displacement
values, although more homogeneous within the same mass (Lx = Ly = 1 m).
12 Front. Struct. Civ. Eng.

random field’s coefficients of variation. Table 5 presents References


the statistical moments of the maximum facing displace-
ments. Figure 13 and Table 5 show that the variability of 1. Griffiths D V, Fenton G A. Bearing capacity of spatially random
the maximum facing displacements increases with the soil: The undrained clay Prandtl problem revisited. Geotechnique,
random field coefficient of variation. 2001, 51(4): 351–359
2. Griffiths D V, Fenton G A, Manoharan N. Bearing capacity of rough
Table 5 Effect of the random fields coefficients of variation “φ” on the
rigid strip footing on cohesive soil: Probabilistic study. Journal of
statistical moments of the maximum facing displacement. Geotechnical and Geoenvironmental Engineering, 2002, 128(9):
item Uh  max ðcmÞ Uh  max ðcmÞ COV (%) 743–755
COVφ = 10% 7.5022 0.7784 10.3756 3. Fenton G A, Griffiths D V. Bearing capacity prediction of spatially
COVj = 15% 7.6901 1.1297 14.6903 random c-φ soils. Canadian Geotechnical Journal, 2003, 40(1): 54–
65
COVj = 20% 7.9875 1.4617 18.2998
4. Popescu R, Deodatis G, Nobahar A. Effect of random heterogeneity
of soil properties on bearing capacity. Probabilistic Engineering
Mechanics, 2005, 20(4): 324–341
7 Conclusions 5. Ahmed A, Soubra A H. Extension of subset simulation approach for
uncertainty propagation and global sensitivity analysis. Georisk:
The influence of the soil spatial variability on the behavior Assessment and Management of Risk for Engineered Systems and
of a reinforced earth wall at the SLS is considered in this Geohazards, 2012, 6(3): 162–176
work. Both cases of isotropic and anisotropic non-normal 6. Al-Bittar T, Soubra A H. Bearing capacity of strip footing on
random fields are considered for the soil properties. The spatially random soils using sparse polynomial chaos expansion.
K-L method is used for the discretization of the random International Journal for Numerical and Analytical Methods in
field. A deterministic model based on numerical simula- Geomechanics, 2013, 37(13): 2039–2060
tions using the explicit code Lagrangian finite-difference 7. Gheris A, Hamrouni A. Treatment of an expansive soil using
code FLAC2D is used to calculate the maximum vegetable (DISS) fibre. Innovative Infrastructure Solutions, 2020, 5:
horizontal facing displacement. The MCS technique was 34
used to determine the deformation response variability of 8. Luo N, Bathurst R J, Javankhoshdel S. Probabilistic stability
the reinforced soil retaining wall. The main results of this analysis of simple reinforced slopes by finite element method.
study are the following ones: Computers and Geotechnics, 2016, 77: 45–55
1) The main input parameter which influence the 9. Pan Q, Dias D. An efficient reliability method combining adaptive
movement of the wall is the internal friction angle for the Support Vector Machine and Monte Carlo Simulation. Structural
studied case. Safety, 2017, 67: 85–95
2) The numerical results have shown that the variability 10. Pan Q, Dias D. Sliced inverse regression-based sparse polynomial
of the maximum facing displacement increases when the chaos expansion for reliability analysis in high dimensions.
random fields coefficient of variation increase. Reliability Engineering & System Safety, 2017, 167: 484–493
3) A maximum probabilistic mean is reached for an 11. Hamrouni A, Dias D, Sbartai B. Probabilistic analysis of a piled
intermediate value of the autocorrelation distance (roughly earth platform under a concrete floor slab. Soils Found, 2017, 57(5):
equal to the reinforcement length), whatever the studied 828–839
scenario. This maximum indicates that the reinforced area 12. Hamrouni A, Dias D, Sbartai B. Reliability analysis of shallow
is the least stable for this autocorrelation distance. tunnels using the response surface methodology. Underground
4) Considering a friction angle spatial variability along Space, 2017, 2(4): 246–258
the transverse direction, it induces a decrease of the 13. Pan Q, Dias D. Probabilistic analysis of a rock tunnel face using
horizontal facing displacement. As a result, it leads to polynomial chaos expansion method. International Journal of
higher displacement values than those obtained in cases of Geomechanics, 2018, 18(4): 04018013
spatial variability in the longitudinal and/or 2D directions. 14. Kroetz H, Do N A, Dias D, Beck A T. Reliability of tunnel lining
The last two scenarios would lead to a minimum facing design using the hyperstatic reaction method. Tunnelling and
displacement. Underground Space Technology, 2018, 77: 59–67
5) The standard deviation of the maximum facing 15. Guo X, Dias D, Carvajal C, Peyras L, Breul P. Reliability analysis of
displacement increases with the autocorrelation distance embankment dam sliding stability using the sparse polynomial
increase for all the considered scenarios, chaos expansion. Engineering Structures, 2018, 174: 295–307
Finally, the obtained results show the importance of 16. Hamrouni A, Sbartai B, Dias D. Probabilistic analysis of
considering the spatial variability of soil properties due to ultimate seismic bearing capacity of strip foundations. Journal of
the fact that some observed phenomena (such as the non- Rock Mechanics and Geotechnical Engineering, 2018, 10(4): 717–
symmetric soil failure) cannot be seen when homogenous 724
soils are considered. 17. Guo X, Dias D, Carvajal C, Peyras L, Breul P. A comparative study
Adam HAMROUNI et al. Soil spatial variability impact on the behavior of a reinforced earth wall 13

of different reliability methods for high dimensional stochastic 34. Hamrouni A, Dias D, Sbartai B. Reliability analysis of a
problems related to earth dam stability analyses. Engineering mechanically stabilized earth wall using the surface response
Structures, 2019, 188: 591–602 methodology optimized by a genetic algorithm. Geomechanics and
18. Hamrouni A, Dias D, Sbartai B. Probability analysis of shallow Engineering, 2018, 15(4): 937–945
circular tunnels in homogeneous soil using the surface response 35. Dodagoudar G, Sayed S, Rajagopal K. Random field modeling of
methodology optimized by a genetic algorithm. Tunnelling and reinforced retaining walls. International Journal of Geotechnical
Underground Space Technology, 2019, 86: 22–33 Engineering, 2015, 9(3): 229–238
19. Schlosser F, Elias V. Friction in Reinforced Earth. Pittsburgh: A.S. 36. Ning L, Bathurst R J. Probabilistic analysis of reinforced slopes
C.E. Convention, 1978, 24–28 using RFEM and considering spatial variability of frictional soil
20. Abdelouhab A, Dias D, Freitag N. Numerical analysis of the properties due to compaction. Georisk: Assessment and Manage-
behavior of mechanically stabilized earth walls reinforced with ment of Risk for Engineered Systems and Geohazards 2017, 12(2):
different types of strips. Geotextiles and Geomembranes, 2011, 87–108
29(2): 116–129 37. Griffiths D V, Fenton G A. Probabilistic slope stability analysis by
21. Abdelouhab A, Dias D, Freitag N. Two-dimensional numerical finite elements. Journal of Geotechnical and Geoenvironmental
modeling of Reinforced Earth walls. European Journal of Environ- Engineering, 2004, 130(5): 507–518
mental and Civil Engineering, 2012, 16(10): 1143–1167 38. Gay O. Physical and numerical modeling of the effect of a slow slide
22. Riccio M, Ehrlich M, Dias D. Field monitoring and analyses of the on foundations of structures. Dissertation for the Doctoral Degree.
response of a block-faced geogrid wall using fine-grained tropical Grenoble: Joseph Fourier University, 2000
soils. Geotextiles and Geomembranes, 2014, 42(2): 127–138 39. Flavigny E, Desrues J, Palayer B. Technical note: Hostun RF sand.
23. Chun B S, Kim K M, Min D K. A study on reliability analysis for Revue Française de Géotechnique, 1990, 53: 67–70
reinforced earth retaining walls. In: The Third Asian Geotechnical 40. Itasca Consulting Group, Inc. FLAC2D–Fast Lagrangian Analysis
Conference on Geosynthetics. Seoul: Millpress Science Publishers, of Continua in 2 Dimensions. Ver. 7, User’s Manual. Minneapolis:
2004, 248–254 Itasca, 2011
24. Sayed S, Dodagoudar G R, Rajagopal K. Reliability analysis of 41. Abdelouhab A, Dias D, Freitag N. Physical and analytical modelling
reinforced soil walls under static and seismic forces. Geosynthetics of geosynthetic strip pull-out behaviour. Geotextiles and Geomem-
International, 2008, 15(4): 246–257 branes, 2010, 28(1): 44–53
25. Sayed S, Dodagoudar G R, Rajagopal K. Finite element reliability 42. Der Kiureghian A, Ke J B. The stochastic finite element method in
analysis of reinforced retaining walls. Geomechanics and Geoengi- structural reliability. Probabilistic Engineering Mechanics, 1988,
neering, 2010, 5(3): 187–197 3(2): 83–91
26. Miyata Y, Bathurst R J. Reliability analysis of soil-geogrid pullout 43. Vu-Bac N, Rafiee R, Zhuang X, Lahmer T, Rabczuk T. Uncertainty
models in Japan. Soil and Foundation, 2012, 52(4): 620–633 quantification for multiscale modeling of polymer nanocomposites
27. Miyata Y, Bathurst R J, Allen T M. Reliability analysis of geogrid with correlated parameters. Composites. Part B, Engineering, 2015,
creep data in Japan. Soil and Foundation, 2014, 54(4): 608–620 68: 446–464
28. Miyata Y, Bathurst R J. Reliability-based analysis of combined 44. Vu-Bac N, Lahmer T, Zhuang X, Nguyen-Thoi T, Rabczuk T. A
installation damage and creep for the tensile rupture limit state of software framework for probabilistic sensitivity analysis for
geogrid reinforcement in Japan. Soils and Foundations, 2015, 55(2): computationally expensive models. Advances in Engineering Soft-
437–446 ware, 2016, 100: 19–31
29. Chalermyanont T, Benson C H. Reliability-based design for internal 45. Spanos P D, Ghanem R. Stochastic finite element expansion for
stability of mechanically stabilized earth walls. Journal of random media. Journal of Engineering Mechanics, 1989, 115(5):
Geotechnical and Geoenvironmental Engineering, 2004, 130(2): 1035–1053
163–173 46. Sahraoui Y, Chateauneuf A. the effects of spatial variability of the
30. Sia A H I, Dixon N. Deterministic and reliability-based design: aggressiveness of soil on system reliability of corroding under-
Veneer cover soil stability. Geosynthetics International, 2008, 15(1): ground pipelines. International Journal of Pressure Vessels and
1–13 Piping, 2016, 146: 188–197
31. Thurner R, Schweiger H F. Reliability analysis for geotechnical 47. Cho S E. Effect of spatial variability of soil properties on slope
problems via finite elements—A practical application. In: ISRM stability. Engineering Geology, 2007, 92(3–4): 97–109
International Symposium. Melbourne: International Society for 48. Cho S E, Park H C. Effect of spatial variability of cross-correlated
Rock Mechanics and Rock Engineering, 2000, 19–24 soil properties on bearing capacity of strip footing. International
32. Lin B H, Yu Y, Bathurst R J, Liu C N. Deterministic and Journal for Numerical and Analytical Methods in Geomechanics,
probabilistic prediction of facing deformations of geosynthetic- 2010, 34: 1–26
reinforced MSE walls using a response surface approach. Geotex- 49. Ghanem R G, Spanos P D. Stochastic Finite Elements: A Spectral
tiles and Geomembranes, 2016, 44(6): 813–823 Approach. New York: Springer, 1991
33. Yu Y, Bathurst R J. Probabilistic assessment of reinforced soil wall 50. Sudret B, Berveiller M. Stochastic finite element methods in
performance using response surface method. Geosynthetics Inter- geotechnical engineering. In: Reliability-based Design in Geotech-
national, 2017, 24(5): 524–542 nical Engineering: Computations and Applications. Oxford: Taylor
14 Front. Struct. Civ. Eng.

& Francis, 2008 53. Vanmarcke E. Probabilistic modeling of soil profiles. Journal of
51. Shinozuka M, Dasgupta G. Stochastic fields and their digital Geotechnical Engineering, 1977, 103: 1227–1246
simulation. In: Stochastic Methods in Structural Dynamics. Boston: 54. Vanmarcke E. Random Fields: Analysis and Synthesis. Cambridge:
Martinus Nijhoff Publishers, 1987, 93–133 MIT press, 1983
52. Deodatis G. Stochastic FEM sensitivity analysis of nonlinear 55. Popescu R. Stochastic variability of soil properties: Data analysis,
dynamic problems. Probabilistic Engineering Mechanics, 1989, digital simulation, effects on system behaviour. Dissertation for the
4(3): 135–141 Doctoral Degree. Princeton, NJ: Princeton University, 1995

View publication stats

You might also like