Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 697e702

Contents lists available at ScienceDirect

Journal of Rock Mechanics and

Geotechnical Engineering
journal homepage: [Link]

Full length article

Interaction analysis of back-to-back mechanically stabilized earth

walls
Sadok Benmebarek*, Samir Attallaoui, Nama Benmebarek
NMISSI Laboratory, Biskra University, Biskra 07000, Algeria

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 22 December 2015
Received in revised form
11 May 2016
Accepted 14 May 2016
Available online 15 July 2016

Back-to-back mechanically stabilized earth walls (BBMSEWs) are encountered in bridge approaches,
ramp ways, rockfall protection systems, earth dams, levees and noise barriers. However, available design
guidelines for BBMSEWs are limited and not applicable to numerical modeling when back-to-back walls
interact with each other. The objective of this paper is to investigate, using PLAXIS code, the effects of the
reduction in the distance between BBMSEW, the reinforcement length, the quality of backll material
and the connection of reinforcements in the middle, when the back-to-back walls are close. The results
indicate that each of the BBMSEWs behaves independently if the width of the embankment between
mechanically stabilized earth walls is greater than that of the active zone. This is in good agreement with
the result of FHWA design guideline. However, the results show that the FHWA design guideline underestimates the lateral earth pressure when back-to-back walls interact with each other. Moreover, for
closer BBMSEWs, FHWA design guideline strongly overestimates the maximum tensile force in the
reinforcement. The investigation of the quality of backll material shows that the minor increase in
embankment cohesion can lead to signicant reductions in both the lateral earth pressure and the
maximum tensile force in geosynthetic. When the distance between the two earth walls is close to zero,
the connection of reinforcement between back-to-back walls signicantly improves the factor of safety.
2016 Institute of Rock and Soil Mechanics, Chinese Academy of Sciences. Production and hosting by
Elsevier B.V. This is an open access article under the CC BY-NC-ND license ([Link]
licenses/by-nc-nd/4.0/).

Keywords:
Back-to-back walls
Numerical analysis
Geosynthetic
Factor of safety
Lateral earth pressure
Maximum tensile force
Reinforcement

1. Introduction
Mechanically stabilized earth (MSE) walls are well-recognized
alternatives to conventional retaining walls due to many advantages such as ease of construction, economy, and aesthetics. For
this, limit equilibrium and numerical methods were basically used
to evaluate the stability of MSE walls (Leshchinsky and Han, 2004;
Han and Leshchinsky, 2006, 2007, 2010). In recent years, back-toback MSE walls (BBMSEWs) have been increasingly used for
bridge approaches, ramp ways, rockfall protection systems, earth
dams, levees and noise barriers. However, there are insufcient
studies and guidelines concerning the behavior of BBMSEWs.
FHWA design guideline (Berg et al., 2009) addressed the design of
back-to-back walls, as illustrated in Fig. 1. Berg et al. (2009) divided
back-to-back walls into two cases:

* Corresponding author. Tel.: +213 670071109.

E-mail address: sadok_benmebarek@[Link] (S. Benmebarek).
Peer review under responsibility of Institute of Rock and Soil Mechanics, Chinese Academy of Sciences.

(1) Case 1: When the distance between the MSE walls, D, is

greater than H1tan (45
 4/2), where H1 is the height of the
higher wall and 4 is the friction angle of the backll, the
width of the ramp or embankment allows for construction of
two separate walls with sufcient spacing between them to
ensure that each wall can act independently. Hence each wall
can be designed individually.
(2) Case 2: When D 0 and the overlap length exceeds 0.3H2,
where H2 is the height of the lower wall, two walls are still
designed independently for internal stability but no active
thrust to the reinforced zone is assumed from the backll. In
other words, no active earth thrust from the backll needs to
be considered for external stability analysis. In this case, the
two walls are assumed to act as a whole, without backll to
exert an external destabilizing thrust.
For intermediate geometries between Cases 1 and 2, when
0 < D < H1tan (45
 4/2), Berg et al. (2009) suggested to interpolate linearly the earth pressure between full active earth pressure
in Case 1 and zero earth pressure in Case 2. However, no justication was provided for this suggestion. Using numerical modeling

[Link]
1674-7755 2016 Institute of Rock and Soil Mechanics, Chinese Academy of Sciences. Production and hosting by Elsevier B.V. This is an open access article under the CC BYNC-ND license ([Link]

S. Benmebarek et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 697e702

45+/2

H2

H1

L2

L1

(a) Case 1.

LR

L1

H2

H1

L2
Wb
(b) Case 2.
Fig. 1. Back-to-back mechanically stabilized earth walls (after Berg et al., 2009).

for the case of limit equilibrium state (i.e. the factor of safety FS 1),
Han and Leshchinsky (2010) indicated that the FHWA design
guideline underestimates the interaction distance, and for W/H1 (W
is the distance between two opposing wall facings) ranging from 2
to 3, the back-to-back walls still interact with each other. Recently,
El-Sherbiny et al. (2013) analyzed different wall width to height
ratios of BBMSEW using the nite element modeling. The numerical model was validated against an instrumented large-scale test
wall (Won and Kim, 2007). It was indicated that when D/H1 < 1, the
two MSE walls interact with each other and the earth pressure
behind the wall decreases because the failure wedge behind the
wall is not fully developed.
In the above-mentioned studies, the interaction distance was
identied when the critical failure surfaces in two opposing walls
did not intercept each other. This seems to be not identical to that
dened by the FHWA design guideline as shown in Fig. 1. In other
words, a single failure surface may occur in one wall.

L=4.2 m

4.2 m

Reinforced
& retained fill

0.75 m

Reinforcement
UX-1400

5m
Stiff foundation

45+/2
(a) Case 1.

W=(0.81.4)H
LR

H=6 m

2. Numerical modeling
In this study, the PLAXIS software was utilized to perform the
two-dimensional (2D) numerical analysis in the condition of plane
strain. The geometry of the baseline model of BBMSEW (Fig. 2)
considered in this study has the same conguration as that reported by Han and Leshchinsky (2010). The height of the walls is
kept constant, equal to 6 m; and the soil foundation depth is equal
to 2 m. The distance between the walls varies from 3H to 0.8H
(large to narrow backll width). Two soils are distinguished:
backll and base soils. The backll material used for reinforced
soil walls is assumed to be granular ll. A stiff soil like rock is
chosen as the base soil to minimize its inuence on the behavior
of reinforced soil. The constitutive relation used for both soil types

W=(1.43)H
D

Precast concrete
panels

H=6 m

45+/2

is the MohreCoulomb model. The properties of the two soils are

shown in Table 1. The Tensar UX-1400 uniaxial geogrid was
adopted to reinforce the BBMSEWs. The soils were simulated using 15-node triangular elements and the geogrid was modeled
using an elastic-perfectly plastic model dened by the stiffness
and tensile strength of geogrid. The vertical spacing of each layer
of geogrid is 0.75 m. The length of reinforcement, L 4.2 m, was
selected to give L/H 0.7. This ratio is the minimum value recommended by the FHWA design guideline for static design (Berg
et al., 2009), except for Case 2 where L/H 0.6 for the geometry with the overlap length, LR, greater than 0.3H. The geogrid
properties used in modeling are summarized in Table 2. The wellknown segmental precast concrete panels were considered in the
current study to simulate the wall. Each wall contains 4 segmental
concrete panels of 1.5 m in width and height and 0.14 m in
thickness. The panels are modeled as a linear elastic material. For
the panels, the Youngs modulus E 25 GPa, the Poissons ratio
n 0.2, and the unit weight g 23.5 kN/m3 Table 3 summarizes
the panel properties as inputs to PLAXIS. The base of the wall is set
to be hinged (i.e. the displacement of the wall is limited in vertical
direction, but it is free to rotate and move in the horizontal
direction).
In the numerical modeling, the geostatic stresses are rstly
generated for the base soil. Secondly, the walls are constructed in
stages, simulating the real construction process of these structures.
The working stresses, strains, deformations, and tensile forces in
the reinforcement are also evaluated in this phase. Then, reductions
in 4 and c (Brinkgreve et al., 2008) are conducted in models to
determine the factor of safety. Finally, the methodology described
above is validated by simulating the well-instrumented Founders/
Meadow segmental bridge abutment reported by Abu-Hejleh et al.
(2002).

2m

698

L=(0.60.7)H

(b) Case 2.
Fig. 2. Dimensions and parameters of the models.

S. Benmebarek et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 697e702

699

Table 1
Material properties of backll and base soil.
Unit weight, g (kN/m3) Friction angle, 4 (
) Dilation angle, j (
) Cohesion, c (kPa) Elastic modulus, E (MPa) Poissons ratio, n

Materials

Elastic perfectly plastic Backll soil 18

MohreCoulomb
Base soil
22

30, 35, 40
30

Table 2
Properties of geosynthetic soil reinforcement.
Model

Ultimate tensile
strength (kN/m)

Elastoplastic 70

Allowable tensile
strength, Ta (kN/m)

Axial stiffness
(kN/m)

25.6

1100

3. Computation results
3.1. Overall factor of safety
The factor of safety against shear failure was obtained using a
shear strength reduction technique (Brinkgreve et al., 2008) for D
values ranging from 0 to 1.6H (W/H 1.4e3). The calculated factors
of safety for Case 1 are presented in Fig. 3. The results are shown as
the normalized distance between the back-to-back walls (D/H).
For soils with different friction angles (4 40
, 35
and 30
, as
shown in Fig. 3), the factor of safety of the BBMSEW decreases
rstly with the increase in the distance between the walls and then
converges to a constant value, indicating the total attenuation of
the interaction between two walls. For 4 40
, 35
and 30
, the
interaction distance D, based on the FHWA design guideline (Berg
et al., 2009), as shown in Fig. 3, is found to be equal to 0.47H,
0.52H and 0.58H, respectively. It is clearly illustrated that the
interaction distance obtained from the numerical analysis is
smaller than that from the FHWA method. This is because that, in
the FHWA method, the Rankine failure plane is assumed in the limit

5
0

0
100

30
200

0.3
0.2

equilibrium state to determine the interaction between two backto-back walls, while this study is based on a working stress and
the failure planes are not formed in the numerical models when
FS > 1.4. This nding is different from Han and Leshchinsky (2010)
based on the limit equilibrium state (i.e. FS 1). Decreasing D leads
to the increase in the factor of safety for different friction angles.
Fig. 4 shows the factor of safety of back-to-back walls when
D  0 and the friction angle of backll material 4 35
. It is shown
that the responses of the BBMSEW with overlapping and continuous reinforcements are different.

3.4

LR=0.4H

Continuous layer

3.2

LR=0.4H

LR=0.3H

Overlapping layer L/H=0.7

Overlapping layer L/H=0.6

LR=0.3H

2.8
Factor of safety

Model

LR=0.2H

2.6

LR=0.2H

2.4
LR=0.1H

2.2
LR=0.1H

LR=0.0H

1.8
LR=0.0H

Table 3
Properties of facing panels (as input to PLAXIS).

1.6
0.8

Model Axial stiffness,

EA (kN/m)

Bending stiffness,
EI (kN m2/m)

Thickness, Weight, w
d (m)
(kN m2)

Poissons
ratio, n

Elastic 3.5  106

5717

0.14

0.2

3.29

1
1.1
1.2
Wall width to height ratio, W/H

1.3

1.4

Fig. 4. Factor of safety of narrow walls when D  0 (W/H  1.4).

2.25

D=0
D=0.3H
D=0.6H
D=1.6H
Rankine
At-rest

FHWA design guideline D=Htan(45/2)

2.15

D=0.47H
4.5
=40

1.95

=35
1.85
D=0.52H

=30

1.75
1.65

c=0 kPa
Ta=25.6 kN

1.55

Wall height (m)

2.05
Factor of safety

0.9

=35
c=0 kPa

1.5

D=0.58H
1.45
1.35

0.4
0.8
1.2
1.6
Normalized distance between back-to-back walls, D/H

Fig. 3. Factor of safety of separate walls when D  0 (W/H 1.4e3).

10
20
30
40
Lateral earth pressure behind reinforced wall (kPa)

50

Fig. 5. Distribution of lateral earth pressure behind the reinforced wall.

700

S. Benmebarek et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 697e702

For the case of D 0, the comparison between connected and

unconnected walls shows that the connection signicantly improves the factor of safety. This can be explained by the fact that the
pullout from the middle becomes impossible and each reinforcement can mobilize all its strength.
The factor of safety continues to increase with the decrease in
the distance between the walls for the case of overlapping reinforcement. This is due to the increasing reinforcement overlapping
length. However, for the case of walls with continuous layers of
geogrid, the factor of safety slightly decreases with the decreasing
W/H ratio. This may be related to the slight reduction in critical
failure surfaces within walls. Furthermore, the decrease in the
reinforcement length from 0.7H to 0.6H, as suggested by FHWA
design guideline (Berg et al., 2009) for Case 2, induces the decrease
in the factor of safety.

Present study

100
90

D = 0.52H (FHWA)

80
FHWA design guideline

70

=35
c=0 kPa

60
50

Case 2
LR=0.3H

40

Case 1
D=0.52H

20

Linear interpolation of active earth thrust

10
0

100%

30
36.6%

Normalized lateral force behind reinforced wall (%)

110

3.2. Lateral earth pressure behind the reinforced wall

0

0.4
0.8
1.2
Normalized distance between back-to-back walls, D/H

1.6

Fig. 6. Percentage of lateral thrust behind the reinforced wall.

Wall height (m)

3
W/H=1.4
2

W/H=2
W/H=3

Ka zSv

0
0

10

15

20

The lateral earth pressure behind the reinforced wall is presented in Fig. 5. The average lateral earth pressure behind the wall
is close to the active Rankine lateral earth pressure when the value
of D/H is large. Nevertheless, in the lower 1/4 of the wall, corresponding to the rst precast segment, the lateral earth pressure
increases and approaches the at-rest earth pressure. This is
attributed to the restraint imposed at the base of the wall by the
precast wall foundation, as clearly observed by wall displacements.
Similar observations have been made from full-scale walls (Won
and Kim, 2007; Huang et al., 2010). However, when the spacing D
decreases from 1.6H to 0, the lateral earth pressure decreases. It is
evident that the lateral earth pressure exists behind the reinforced
wall even for D 0.
The ratio of the active lateral thrust behind the reinforced wall
to the theoretical active Rankine lateral thrust is presented in Fig. 6,
which shows the inuence of D/H on the mobilization of the lateral
thrust. The lateral earth thrust exceeds the active Rankine earth
thrust when D/H is greater than 0.6. This is related to the increase in
lateral earth pressure at the wall base. Nevertheless, the lateral
earth thrust evidently decreases when D is lower than 0.5H. The
lateral earth thrust decreases to 77% of the active Rankine lateral
thrust as D approaches 0.

Maximum tensile force in reinforcement, Tmax (kN/m)

(a) D>0.

5
4.5
W=1.4H
Wall height (m)

Wall height (m)

2
W/H=1.4 connected
1

W=0.8H
3

1.5

W/H=1.4 unconnected

0
5
10
15
20
Maximum tensile force in reinforcement, Tmax (kN/m)
(b) D=0.

Fig. 7. Maximum tensile force in reinforcement at the end of construction.

0.005
Normalized displacement, x/H
Fig. 8. Wall displacement at the end of construction.

0.01

S. Benmebarek et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 697e702

6
c=0 kPa
c=5 kPa

=35
W/H=2

Wall height (m)

701

geometries between Cases 1 and 2, it is underestimated by FHWA

guideline. For instance, when D 0 and no connection exists, the
active earth thrust obtained in the present study is 77% of the active
Rankine lateral thrust, whilst the interpolation as suggested by
FHWA gives 36.6%. A great value of lateral earth thrust (85% at
D 0) was also found by Han and Leshchinsky (2010).
3.3. Tensile force in the reinforcement

0
0

10
20
30
40
50
Lateral earth pressure behind reinforced wall (kPa)

Fig. 9. Effect of cohesion on lateral earth pressure.

The comparison of the present numerical results to the values

given by FHWA guideline (Berg et al., 2009), which suggested that
the lateral earth pressure for external analysis should be ignored if
D 0 and the overlapping length exceeds 0.3H, and that estimated
by linear interpolation for D between 0 (LR 0.3H) and 0.52H
shows that the pressure given by FHWA guideline (Berg et al., 2009)
agrees well with the estimated one for Case 1, but for intermediate

Soil reinforcement in MSE walls causes tensile forces to develop,

which helps to stabilize the sliding mass of the wall. The tensile
force developed in each layer of reinforcement is not uniform. The
maximum tensile force in each layer of reinforcement obtained
from the numerical analyses is presented in Fig. 7. For the case of
D > 0 (Fig. 7a), the tensile forces in the reinforcement at different
W/H ratios are found very close. The theoretical values for the
normalized tensile force in each reinforcement layer using limit
equilibrium methods based on Coulomb theory are also shown in
Fig. 7a for comparison purpose based on the analyses of one wall,
without consideration of wall interaction. These values are given by
KagzSv (Berg et al., 2009), where Ka is the active earth pressure
coefcient, g is the unit weight of the soil, z is the depth of reinforcement layer under consideration, and Sv is the vertical spacing
between reinforcement layers. The computation results of tensile
forces in the reinforcement for different W/H ratios covering Cases
1 and 2 of FHWA design guideline (Berg et al., 2009) match with the
results obtained by the limit equilibrium method, which vary linearly with depth, except the lower quarter of the wall where the
tensile force decreases due to the toe restraint at the base of the
wall, as observed by Huang et al. (2010).
When D 0 (W/H 1.4), the tensile forces in the reinforcement
(Fig. 7b) for both connected and unconnected walls are found very
close. From the obtained results, we can conclude that the tensile
forces in the reinforcement layers are lowly sensitive to the distance
between the BBMSEWs. The limit equilibrium method can be used
for computing the tensile forces with security for back-to-back

c=0 kPa

Plastic
point
Tension
point

c=5 kPa
Plastic
point
Tension
point

Fig. 10. Effect of cohesion within walls at W/H 2.

702

S. Benmebarek et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 697e702

6
c=0 kPa
c=5 kPa

Wall height (m)

4
=35
3

0
0

5
10
15
20
Maximum tensile force in reinforcement, Tmax (kN/m)

Fig. 11. Effect of cohesion on the maximum tensile force in geosynthetic.

stabilized retaining walls. The present computation results contradict those of the FHWA design guideline (Berg et al., 2009), which
stated that the connected reinforcement creates an unyielding
structure, developing an at-rest stress state (K0) from the top to the
bottom of the wall, resulting in much higher tensile forces in reinforcement than the FHWA design guideline (Berg et al., 2009). Berg
et al. (2009) suggested that, to determine the tensile force in the
reinforcement and connection, the increases in lateral stress must be
considered in the design of facing elements. For closer BBMSEWs
with multi-layer of geogrid running across the wall and for the case
of overlapping reinforcement, the present computation results at W/
H 1.4 and 0.8 indicate that the walls lateral displacements at the
end of construction (Fig. 8) are more sufcient to develop the active
earth pressure rather than the at-rest earth pressure.
3.4. Effect of cohesion
The effect of embankment cohesion is examined by increasing it
from 0 to 5 kPa, which may cover practical values. As illustrated in
Fig. 9, the lateral earth pressure behind the reinforced wall decreases with the increase in soil cohesion. It is observed that the
lateral earth pressure at c 5 kPa is equal to about zero at the top
1 m of the wall, as the cohesion greater than zero allows a tensile
crack to develop at the wall top (Fig. 10).
It can be also seen from Fig. 11 that the maximum tensile force in
geosynthetic decreases with the increase in soil cohesion at an
average rate of 40%.
4. Conclusions
The nite element code PLAXIS was used to investigate the effect of the wall width to height ratio on internal and external

stability of BBMSEW where the limit equilibrium method is rarely

used. By comparison of computation results in this study with
those of FHWA design guideline, the following conclusions can be
drawn:
(1) The results of this study are consistent with those of
FHWA design guideline that considers the signicant
interaction between the back-to-back walls when
D < Htan (45
 4/2).
(2) The FHWA design guideline underestimates the lateral earth
pressure when back-to-back walls interact with each other.
The interaction is marked by both the decrease in the lateral
earth pressure behind the reinforced wall and the increase in
the factor of safety against shear failure.
(3) When D is close to zero, connection of reinforcement in backto-back walls signicantly improves the factor of safety.
(4) The maximum tensile force in reinforcement layers is
nearly independent of the distance between the BBMSEWs
even for very close walls. The maximum tensile force
computed is found very close to that obtained by the limit
equilibrium method. Thereby, for closer BBMSEWs, FHWA
design guideline strongly overestimates the maximum
tensile force.
(5) The results of this study indicate that a minor increase in
embankment cohesion induces signicant reductions in
both lateral earth pressure and maximum tensile force in
geogrid.

Conict of interest
The authors wish to conrm that there are no known conicts of
interest associated with this publication and there has been no
signicant nancial support for this work that could have inuenced its outcome.
References
Abu-Hejleh N, Zornberg JG, Wang T, Watcharamonthein J. Monitored displacements
of unique geosynthetic-reinforced soil bridge abutments. Geosynthetics International 2002;9(1):71e95.
Berg RR, Christopher BR, Samtani NC. Design and construction of mechanically
stabilized earth walls and reinforced soil slopes e Volume 1. Publication No.
FHWA-NHI-10e024. Washington, D.C.: U.S. Department of Transportation,
Federal Highway Administration; 2009.
Brinkgreve R, Swolfs W, Engin E. PLAXIS nite element code. Delft, Netherlands:
Delft University of Technology, PLAXIS bv; 2008.
El-Sherbiny R, Ibrahim E, Salem A. Stability of back-to-back mechanically stabilized
earth walls. In: Proceedings of Geo-Congress 2013: Stability and Performance of
Slopes and Embankments III. Reston, USA: American Society of Civil Engineers
(ASCE); 2013. p. 555e65. [Link]
Han J, Leshchinsky D. Analysis of back-to-back mechanically stabilized earth walls.
Geotextiles and Geomembranes 2010;28(3):262e7.
Han J, Leshchinsky D. Stability analyses of geosynthetic-reinforced earth structures
using limit equilibrium and numerical methods. In: Proceedings of the 8th
International Conference on Geosynthetics (8ICG). Yokohama, Japan; 2006.
p. 1347e50.
Han J, Leshchinsky D. Stability analysis of back-to-back MSE walls. In: Proceedings
of the 5th International Symposium on Earth Reinforcement (IS Kyushu07).
Fukuoka: Japan; 2007. p. 487e90.
Huang B, Bathurst RJ, Hatami K, Allen TM. Inuence of toe restraint on reinforced
soil segmental walls. Canadian Geotechnical Journal 2010;47(8):885e904.
Leshchinsky D, Han J. Geosynthetic reinforced multitiered walls. Journal of
Geotechnical and Geoenvironmental Engineering 2004;130(12):1225e35.
Won MS, Kim YS. Internal deformation behavior of geosynthetic-reinforced soil
walls. Geotextiles and Geomembranes 2007;25(1):10e22.