Archive of SID

Transaction A: Civil Engineering

Vol. 17, No. 4, pp. 273{284
c Sharif University of Technology, August 2010

A New Approach for Estimating the

Seismic Soil Pressure on Retaining Walls
S. Maleki1; and S. Mahjoubi1
Abstract. In this paper, a simple nite element model for seismic analysis of retaining walls is
introduced. The model incorporates nonlinearity in the behavior of near wall soil, wall exibility and
elastic free eld soil response. This model can be employed in nonlinear modeling of retaining walls and
bridge abutments. The advantages of this model are simplicity and exibility in addition to acceptable
precision. Using this nite element model, an analytical study is conducted on several soil-wall systems
using nonlinear time-history analysis by applying real earthquake records. Based on the results of these
analyses, new seismic soil pressure distributions are proposed for di erent soil and boundary conditions.
These distributions are shown to be more accurate than the popular Mononobe-Okabe equations.
Keywords: Seismic analysis; Retaining walls; Soil pressure; Soil-structure interaction; Mononobe-
Okabe; Finite element analysis.

INTRODUCTION Experimental studies since the 1970's have proved

that the M-O method and the modi ed version (Seed
Seismic behavior of retaining walls has been widely and Whitman method) are satisfactory in calculating
investigated by researchers since the 1920's. Three the total seismic soil thrust [4,5,7,8].
pioneer researchers in this eld have been Mononobe, Another prevalent method used by engineers is
Matsuo and Okabe [1,2]. They proposed a pseudo the subgrade modulus method. In this method, the
static method to calculate the seismically induced soil behind the wall is modeled as a series of parallel
active and passive earth pressures using the Coulomb massless springs. Using this concept, Scott [9] studied
theory in a wedge of soil behind the wall. The method, the soil-wall interaction. He modeled the free eld soil
herein called the M-O method, gives two relationships as a vertical shear beam with mass and the interfacing
for seismic active and passive earth pressures with soil as massless linear springs with constant sti ness.
linear distribution along the wall height. Owing to Veletsos and Younan [10] improved Scott's method by
its simplicity, this method has been widely used by applying semi-in nite horizontal bars with constant
practicing engineers. Seed and Whitman [3] modi ed mass per length connected to the wall by massless
the M-O method and suggested new relationships for linear springs. Richards et al. [11], retrospecting
the seismic soil thrust in terms of unit weight of soil and Scott and Veletsos and Younan studies, represented a
peak ground acceleration. They also suggested 0:6 H as method of dynamic analysis for retaining walls using a
the resultant force point of action where H is the height 2D model containing a semi-in nite nonlinear layer of
of the retaining wall. Anderson et al. [4], by a series of cohesionless soil, free at the top and rigidly restrained
centrifuge tests, found that the location of equivalent at the bottom. The retaining wall is connected to
dynamic soil pressure forces varies, but recommended the free eld soil by means of elastic constant sti ness
a value of 0:5 H for its approximate location. Sherif et springs. They studied the model in four modes of
al. [5] and Wood [6] proposed a value of 0:45 H above displacement: rotation about base, rotation about
the base for the seismic earth thrust point of action. top, rigid translation and xed base. Veletsos and
1. Department of Civil Engineering, Sharif University of Tech- Younan [12], using Lagrange's equation, analyzed a
nology, Tehran, P.O. Box 11155-9313, Tehran, Iran. linear model and investigated the e ects of many
*. Corresponding author. E-mail: [email protected] variables, such as base xity and nonuniformity of the
Received 14 September 2009; received in revised form 3 March soil shear modulus, on the soil thrust. They concluded
2010; accepted 17 July 2010 that for realistic wall exibilities, the total wall force is

www.SID.ir
Archive of SID
274 S. Maleki and S. Mahjoubi

one half or less of that obtained for a xed base, rigid propped bridge abutments. Nonlinear nite element
wall. They also found the elastic method to be of good dynamic time history analyses of soil-wall systems are
precision. employed for veri cation. It is shown that the sug-
Recently, with great progress in computer soft- gested equations more accurately predict the seismic
ware and hardware, nonlinear models have been used soil pressure behind retaining walls than the well-
by several researchers to investigate many aspects of known M-O relationships, and are simple to use.
soil-wall interaction and wall behavior during earth-
quakes [13-15]. Green and Ebling [16] employed an SOIL-WALL MODEL
elasto-plastic constitutive model for the soil in con-
junction with the Mohr-Coulomb failure criterion. The In this section, the theoretical background for con-
wall is modeled with elastic beam elements using a struction of an accurate nite element soil-wall analysis
cracked second moment of inertia. Interface elements model is described. Past research has been selectively
were used to model the wall-soil interface. Cheng [17] put together to obtain a simple but yet accurate
proposed relationships for calculating seismic lateral model that ignores nonlinearity in the far eld soil but
earth pressure coecients for cohesive soils using the captures it in the near eld soil adjacent to the wall
slip line method. He found iterative analysis to where it most a ects the wall pressure.
be useful for obtaining passive earth pressure but In an earthquake, the soil-wall system can ex-
unimportant for active earth pressure. He also found perience considerable displacement. If the wall and
that active earth pressure is much more in uenced the free eld displacement of the soil are equal, then
by the PGA of an earthquake than passive pressure. the wall has no e ect on the free eld soil stresses.
Psarropoulos et al. [18] used a nite-element model to However, this is usually not the case, and the di erence
study the dynamic earth pressures developed on rigid in displacements of the soil and wall creates stresses in
or exible nonsliding retaining walls modeling the soil the soil, especially in the vicinity of the wall. Therefore,
as a viscoelastic continuum. Results showed a crude the horizontal stresses in the soil behind the wall can
convergence between the M-O and elasticity-based be written in terms of the di erence between the free
solutions for structurally or rotationally exible walls. eld soil and wall displacements. This phenomenon can
More recently, Choudhury and Chatterjee [19], as be modeled with nonlinear springs attached to the wall
an extension of the Veletsos and Younan study, used a representing the interfacing soil. In addition, the free
mass-spring-dashpot dynamic model with two degrees eld soil can be modeled with elastic plane strain shell
of freedom to arrive at the total active earth pressure elements. These are further explained below.
under earthquake time history loading. They suggested In general, the wall displacement can be obtained
the use of an in uence zone of 10 times the wall height as:
for the soil media behind the wall for dynamic models. uw = uf
u; (1)
They also presented non dimensional design charts for
rapid calculation of active earth pressures. Choudhury where uf is the free eld soil displacement, uw is the
and Subba Rao, in two di erent studies, obtained an wall displacement and
u is the di erence between
estimate for the seismic passive earth pressure against the wall and free eld soil displacements (Figure 1).
retaining walls by using logarithmic spiral and com- The total soil stress behind the wall for increment i (at
posite curve failure surface assumptions and a pseudo location of spring i) can be written as:
static method [20,21]. E ects of cohesion, surcharge, i =  i +
 i ;
wn (2)
fn n
own weight, wall batter angle, ground surface slope, soil
friction angle and seismic accelerations were considered
ni = ksi (uifn uiwn ): (3)
in the analysis.
Although, nowadays, many commercial computer
programs with the ability to perform nonlinear anal-
ysis of continua are available, these programs are
expensive and time-consuming in nonlinear stepwise
analyses. Uncertainty regarding the input for soil pa-
rameters is also a drawback. Programs like FLAC [22],
SHAKE [23] and SASSI [24] are simpler than the
general purpose nite element software, but each has
its own limitations.
The objective of this paper is to obtain a simpli-
ed seismic soil pressure distribution against retaining
walls with di erent boundary and sti ness conditions,
such as rigid walls, exible walls, bridge abutments and Figure 1. Typical soil and wall displacements.

www.SID.ir
Archive of SID
Seismic Soil Pressure on Retaining Walls 275

i
wn = total soil normal stress against the wall at to be 1.35 based on the suggestion of Huang [11,25].
increment i (at location of spring i). Soil pressure is considered bounded by the active and
i
fn = normal soil stress in free eld at passive soil pressures as follows:
increment i.

ni = variation of soil stress because of the Ka : :z  z  Kp : :z; (9)
di erence between the free eld soil and where:
wall displacements at increment i.
ksi = soil subgrade modulus at increment i. 1 sin  1 + sin 
Ka = ; Kp = : (10)
1 + sin  1 sin 
The total horizontal force behind the wall is:
Z H Therefore, the springs between the free eld soil and
Pnt = wn dz; (4) retaining wall are modeled as a bilinear elastic perfectly
0 plastic type. The initial elastic linear part sti ness
and the overturning moment is: is calculated from Equation 8 by substituting Gz for
each spring from Equation 7. The plastic portion has a
Z H maximum/minimum force of Pp =Pa . These are spring
Mnt = wn zdz: (5) forces equivalent to passive and active soil pressures,
0 respectively, and are calculated by multiplying passive
Therefore, the point of application of the resultant or active soil pressures and the spring corresponding
horizontal force is obtained as: area for each spring. A typical force-displacement plot
for soil springs between the retaining wall and the free
M eld soil is shown in Figure 2.
h0 = nt : (6)
Pnt Figure 3 shows the structural model of the soil-
For cohesionless soils, the modulus of elasticity (E ) and wall system. The wall is modeled by using shell
the shear modulus (G) vary increasingly with depth. elements of concrete. The width of the wall and soil
Two common assumptions for this variation are linear shell elements, perpendicular to the paper in Figure 3,
and parabolic. Here, the parabolic assumption is used is considered to be 1 meter.
for the shear modulus as follows [11]: The free eld soil modeling in this study consists
of an in nite half-space elastic layer of dense cohesion-
p less soil with unit weight of . This half-space layer
Gz = GH : z=H; (7)
where Gz and GH are the elastic shear moduli at depths
z and H , respectively.
The soil behind the wall is divided into layers and
the above equation is used to estimate G in the middle
of each layer. The equivalent springs representing the
soil adjacent to the wall are modeled using nonlinear
link elements. The sti ness of each spring, de ned as
the subgrade modulus, is derived as follows [9]:
Gz
ks = Cz : ; (8)
H
where Cz is a constant representing the geometric Figure 2. Typical force-displacement plot for soil springs
properties of the model. The value of Cz is assumed behind a retaining wall.

Figure 3. Soil-wall nite element model.

www.SID.ir
Archive of SID
276 S. Maleki and S. Mahjoubi

is free at the top and considered xed at the bottom structure is not restrained to the abutment, the gap
where similar soil acceleration is applied to the soil and between the backwall and the superstructure is usually
wall during an earthquake. Free eld soil is assumed to closed by the superstructure movements and, most
have a nite length equal to 4 to 5 times its thickness. of the time, the collision of the superstructure into
Choudhury and Chatterjee suggested an in uence zone the backwall is unavoidable in severe earthquakes.
of 10 times the wall height for the soil media behind This collision produces a considerable concentrated
the wall [19]. However, in their study, soil media at force at the top of the bridge abutment. In the
the end is not restrained against vertical displacement, second case where the superstructure is restrained
which is the case in this study (Figure 3). The authors to the abutment, the superstructure inertial force is
found that restraining the end of the soil media against transferred to the abutment through the restraint
vertical displacement decreases the in uence zone and as a concentrated force at the abutment seat level.
makes the model smaller and more ecient. Therefore, abutments have to resist a considerable
The height of the retaining wall and the soil concentrated force at their top during earthquakes in
behind it are assumed to be the same and equal to H . both cases. However, the e ect of a superstructure
The free eld soil layer is modeled using plane strain horizontal force on top of the abutment is usually
elements. Soil layers have di erent elastic properties; ignored in bridge abutment design. To evaluate
however, they are assumed to be constant within each the e ect of this force and determine the part of
soil layer. Shear modulus is calculated by substituting this force taken by the soil behind the abutment, a
the average depth of the layer as z in Equation 7. concentrated force with variable value is considered
Poisson's ratio for all soil layers is assumed to be 0.3. to push the wall against the soil at the assumed
Four di erent cases of rigid retaining wall, exible superstructure center of mass, as is the case during
cantilever retaining wall, bridge abutment and propped earthquakes.
bridge abutment are considered, each with 4 m, 6 m Ground motions of 6 historical earthquakes (the
and 8 m wall heights. In the case of a rigid retaining 1940 El Centro, the 1989 Loma Prieta, the 1992
wall, it is assumed that the wall has unlimited sti ness Landers, the 1986 Palm Spring, the 1971 San Fernando
and so shows no deformation. Although in reality this (at Pacoima dam station) and the 1966 Park eld) are
is impossible, the case of a buttressed wall comes very used in nonlinear time-history analyses. Figure 4 shows
close to this assumption. the pseudo acceleration (PSA) spectrum for ground
The exible wall case is the case of a cantilever motions used in the analyses.
retaining wall with variable thickness. It is assumed
that the thickness varies from 0.3 m at the top to 0:1 H
at the bottom. In the case of a 4 m exible wall, it is ANALYSES RESULTS
not practical to vary the wall thickness from 0.3 m to Nonlinear nite element dynamic time history analyses
0.4 m. Therefore, the wall thickness is considered to be on the soil-wall systems described above are performed.
constant and equal to 0.4 m. Soil pressure is calculated from the spring forces output
In the case of bridge abutments, the wall thickness divided by the distance between the springs and the
is considered to be constant throughout the height.
The thickness is assumed to be equal to 1 m for 6 m and
8 m height walls and 0.8 m for a 4 m height wall. The
case of a propped bridge abutment is a special case of
bridge abutment in which the superstructure restrains
the abutment horizontally at the top. Although in
reality this restraining is limited, the case is valuable
as an extreme boundary condition case.
In all cases considered for analyses, the wall is
fully restrained at the bottom (rotations and displace-
ments). However, the e ect of foundation rotational
sti ness is investigated separately for a 6 m high bridge
abutment and the propped bridge abutment models.
In these models, the restraint against rotation at the
bottom of the wall is removed and a rotational spring is
substituted to model the foundation and soil rotational
sti ness.
In bridge structures, the superstructure connec-
tion to the abutment can be categorized into two Figure 4. Pseudo acceleration (PSA) spectra for selected
main groups: free and restrained. If the super- ground motions.

www.SID.ir
Archive of SID
Seismic Soil Pressure on Retaining Walls 277

Figure 5. Seismic soil pressure for 4 m height wall. (a) Bridge abutment; (b) exible wall; (c) rigid wall.

Figure 6. Seismic soil pressure for 6 m height wall. (a) Bridge abutment; (b) exible wall; (c) rigid wall.
element's thickness (1 m). Plots of soil pressure versus 8 m high walls, the cantilever retaining walls are more
z=H for the 6 ground motions are shown in Figures 5-7. exible and reduce soil pressure. Bridge abutments
As can be seen in Figure 5 for the 4 m high and propped bridge abutments with all wall heights
wall, the seismic soil pressures for di erent cases have considered here (4 m, 6 m and 8 m height) show similar
a similar distribution, but are slightly di erent in soil pressure distributions as rigid walls, but have less
maximum values. However, this is not the case in 6 m maximum seismic soil pressure. It should be noted that
and 8 m high walls (Figures 6 and 7). For these walls, the above seismic soil pressures resulted from analyses
the distribution for exible walls is very di erent from must be added to the active or at rest soil pressures
that of rigid, propped and abutment walls, which have to obtain the total soil pressures behind the retaining
similar distributions. In fact, distribution is dependent walls.
on wall exibility. In other words, for 4 m high walls, all To evaluate the e ects of foundation rotational
cantilever walls act as rigid walls whereas for 6 m and sti ness on seismic soil pressure, the restrained base

www.SID.ir
Archive of SID
278 S. Maleki and S. Mahjoubi

Figure 7. Seismic soil pressure for 8 m height wall. (a) Bridge abutment; (b) exible wall; (c) rigid wall.
of the 6 m high wall is replaced with a rota- force value is great enough, it can push the soil into
tional spring of varying sti ness. Results of the the passive state in that region. By increasing the
nonlinear analyses using the El Centro ground mo- external force, the depth of this passive earth pressure
tion are shown in Figure 8. As can be seen in (passive depth) is also increased. One can conclude
Figure 8, the results show a great dependence on from Figure 9 that the external concentrated force
base rotational sti ness. For instance, variation of a ects taller walls more severely. For instance 4.7% of
base rotational sti ness from 105 kN.m/rad/m to a 1500 kN concentrated force on a 4 m high abutment
106 kN.m/rad/m increases the total soil thrust by wall is taken by the soil, whereas this ratio is 29.1% for
about 103%. an 8 m high abutment wall.
Figure 9 illustrates the soil pressure distribution
due to external concentrated force at the abutment SUGGESTED SEISMIC SOIL PRESSURE
top. This force is applied along the assumed super- DISTRIBUTION
structure's center of mass. This pressure also should
be added to at rest soil pressure to acquire total soil Several seismic soil pressure distributions along the wall
pressure. As can be observed in Figure 9, the external height are suggested by researchers [26]. For example
concentrated force at the top of the abutment increases the M-O method suggests a linear distribution, but the
the soil pressure in the upper part. If the external Seed and Whitman method suggests no distribution
and only de nes 0:6 H as the seismic thrust point of
action.
Based on the results of this study, two ap-
proximate distributions of seismic soil pressures are
proposed, as shown in Figure 10. Type I distribution
is a typical distribution suggested here for rigid walls
and semi-rigid walls, such as bridge abutment and
propped bridge abutment. Type II distribution is
a distribution suggested for exible walls such as
cantilever retaining walls taller than 5 m. It is
suggested that the maximum seismic soil pressure, q0 ,
be obtained from one of the three equations below.
Equation 11 uses the Peak Ground Acceleration (PGA)
of an earthquake. Equation 12 uses the area under
the pseudo acceleration response spectrum (PSA) of
Figure 8. E ect of base rotational sti ness on total an earthquake, whereas Equation 13 uses the peak
seismic soil thrust. spectral acceleration.

www.SID.ir
Archive of SID
Seismic Soil Pressure on Retaining Walls 279

Figure 9. Force-induced soil pressure. (a) 4 m bridge abutment; (b) 6 m bridge abutment; (c) 8 m bridge abutment.

Table 1. Coecient and distribution type in method.

Case Distribution
Bridge abutment (H < 5 m) I 0.31
Bridge abutment (H  5 m) I 0.55
Rigid and propped wall (H < 5 m) I 0.32
Rigid and propped wall (H  5 m) I 0.64
RC retaining wall (H < 5 m) I 0.25
RC retaining wall (H  5 m) II 0.44

Figure 10. Suggested approximate seismic soil pressure

distributions. (a) Type I; (b) Type II. Table 2. Coecient 1 and distribution type in 1
method.
q0 = :kh : :H; (11) Case Distribution 1
Bridge abutment, rigid wall, and
q0 = 1 :A0 : :H; (12) propped wall (independent of height) I 0.13
& RC retaining wall (H < 5 m)
q0 = 2 :S: :H; (13)
RC retaining wall (H  5 m) II 0.07
where:
q0 = maximum seismic soil pressure. Table 3. Coecient and distribution type in
2 2
kh = seismic horizontal acceleration coecient. method.
A0 = the area under the PSA spectrum between
0:2T and 1:5T in which T is the period of the Case Distribution 2
rst mode of vibration (the spectrum Bridge abutment (H < 5 m) I 0.018
must be in units of m/s2 ). Bridge abutment (H  5 m) I 0.03
S = Spectral acceleration in units of m/s2 Rigid and propped Wall (H < 5 m) I 0.019
obtained from the PSA spectrum of an
earthquake at the period of vibration for Rigid and propped Wall (H  5 m) I 0.035
the rst mode of the soil-wall system. RC retaining wall (H < 5 m) I 0.015
, 1 and 2 are coecients from Tables 1 to 3, RC retaining wall (H  5 m) II 0.020

www.SID.ir
Archive of SID
280 S. Maleki and S. Mahjoubi

adjusting the pressure for wall height and boundary

condition.
Distribution type (I or II) shown in Figure 10 is
also de ned in Tables 1 to 3 for each case with di erent
wall heights and boundary conditions.
Seismic soil pressure distributions introduced
above are based on the results of analyses on xed
based models for two reasons. Firstly, to simplify
distributions and their corresponding relationships, so
that they are independent of soil sti ness, foundation
sti ness and foundation width which could complicate
the relationships and make them undesirable for de-
sign. Secondly, to arrive at a conservative estimate
for pressure distributions suitable for design purposes.
A xed-based wall absorbs much more seismic soil
pressure than a wall with nite base rotational sti ness. Figure 11. Period of the rst mode of vibration (T ) for
It should be noted that the base rotational xity is soil-wall systems.
not unreal. For example, the case of pile foundations
(usually used in bridge abutments) is very close to base Distribution II:
rotational xity. 3
Many earthquake records with di erent intensities PST = q0 H: (16)
and frequency contents and di erent soil types may 5
be used along with the proposed method suggested Figure 12 shows the seismic soil pressures against a 6 m
in this paper. However, this is deemed unnecessary high wall subjected to the San Fernando earthquake,
because the suggested method shows good accordance resulted from analyses in di erent cases. Suggested
with other reliable studies in most cases. distributions for seismic soil pressure using the and
The q0 obtained from Equation 11, which is called methods are also shown for comparison. As can
the method herein, is less accurate in comparison be seen in Figure 12, the seismic soil pressure is best
with the other two relationships (Equations 12 and 13). predicted by the 1 method. The method, although
This is because the PGA is the simplest and most showing more dispersion, gives acceptable results for
available parameter of an earthquake and ignores the design purposes. It can also be observed that the
other characteristics of ground motion. Equations 12 suggested distributions are in good agreement with
and 13, herein called the 1 and the 2 methods, give the distributions of seismic soil pressure results from
more accurate results but require the PSA spectrum time history analyses of di erent cases under di erent
of the input ground motion. In using the methods, earthquakes.
one needs the period of the rst mode of vibration The seismic soil pressures introduced above
(T ) for a soil-wall system. Based on the results of should be added to the active or at rest soil pressures to
modal analyses of several soil-wall systems, the authors acquire the total soil pressure. For rigid and semi-rigid
suggest Equation 14 or Figure 11 to rapidly obtain T walls (such as bridge abutment and propped bridge
without analysis: abutment), at rest soil pressure is more appropriate,
whereas for exible walls, such as cantilever retaining
T = 0:0029H 2 0:0062H + 0:142; (14) walls higher than 5 m, active soil pressure should be
added to suggested distributions. Figure 13 shows
where H is the wall height and is restricted between 3 typical total soil pressure distributions for the two cases
to 10 meters. discussed.
Note that for design purposes, the design response For the case of concentrated lateral force at the
spectrum and the PGA of any local seismic design top of the abutment (Figure 14a), which usually occurs
codes can be used as input in and methods, in bridges during earthquakes, a new distribution for
respectively. soil pressure is herein introduced and is illustrated in
Total seismic soil thrust against a retaining wall Figure 14b. This distribution has only one parameter,
(PST ) for proposed distributions can be obtained as: z0 , which is the passive depth de ned above. The
distribution in depths less than z0 follows the passive
Distribution I: soil pressure and at depths more than 0:9 H follows
the at rest soil pressure. Between depths z0 and 0:9H ,
2 soil pressure linearly decreases from passive to at rest
PST = q0 H: (15)
3 soil pressure. The depth ratio, z0 , can be obtained as

www.SID.ir
Archive of SID
Seismic Soil Pressure on Retaining Walls 281

Figure 12. Comparison of seismic soil pressures against a 6 m high wall subjected to San Fernando earthquake, resulted
from analysis versus computed values using the proposed and methods. (a) Bridge abutment; (b) exible wall; (c)
rigid wall.
follows:
0:004H 1:6 P 0:9
z0 = ; (17)
(Kp K0 )
where:
P = external concentrated force at the
abutment top along the assumed
superstructure's center of mass
(for 1 m abutment width) in kN.
H = height of the abutment in meters.
Kp , K0 = passive and at rest soil pressure
coecients, respectively.

Figure 13. Total soil pressure distribution during is in kN/m3 and z0 is in meters.
earthquake. (a) Type I (rigid or semi-rigid wall); (b) Type
II ( exible wall). Figure 15 illustrates the seismic soil pressure
caused by a 2000 kN concentrated lateral force at
the top of a 6 m high abutment wall resulted from
analysis and computed by the above equation. As
can be seen in Figure 15, the seismic soil pressure
can accurately be predicted by the suggested distri-
bution.

FURTHER VERIFICATION OF THE

PROPOSED METHOD
In this section, the proposed seismic pressure values
and distribution (Equations 11-13) are compared with
the past analytical and experimental research data.
Seismic soil pressure for a 4.6 m rigid retaining wall
subjected to ATC S1 motion (The ATC recommended
Figure 14. Concentrated lateral force at the top of the motion for S1 soil conditions, PGA = 0.3 g) is calcu-
abutment. (a) Boundary conditions; (b) suggested soil lated using the proposed simpli ed method and com-
pressure distribution. pared with values computed by the SASSI computer

www.SID.ir
Archive of SID
282 S. Maleki and S. Mahjoubi

Figure 15. Seismic soil pressure caused by a 2000 kN

concentrated lateral force at the top of a 6 m high
abutment wall, resulted from analyses and computed by
proposed method.

program [24], M-O active earth pressure (using kh =

0:3, kv = 0) and the Ostadan proposed method [27].
The results are shown in Figure 16a. As can be seen
in Figure 16a, seismic soil pressure values computed
by the proposed method are in good accordance with
the two other seismic pressure values and are less than
the M-O active earth pressure at depths of more than
0:6H .
In addition, to verify the proposed seismic soil
pressure distribution, normalized seismic soil pressure
(soil pressure normalized to 1 at the bottom of the wall
for pressure distribution type I) is plotted in Figure 16b
against z=H (normalized depth) and compared with
the normalized seismic soil pressures presented by the
M-O, Wood [6], Ostadan [27], Veletsos and Younan Figure 16. (a) Comparison of proposed seismic soil
(rigid wall) [12] methods, and also results from the pressure and computed values with SASSI program and
shaking table tests performed by Ohara and Maehara Ostadan method for a 4.6 m wall, ' = 30 , ATC motion.
(rigid wall- case compound displacement II) [28] and (b) Comparison of proposed seismic soil pressure
recently performed centrifuge tests by Nakamura (rigid distribution with analytical and experimental
wall- case 21) [29]. Figure 16b shows that the proposed distributions [6,12,27-29].
seismic soil pressure distribution is in good agreement
with analytical methods and experimental data pre- sliding on a linear failure surface, can be criticized,
sented in the literature. in most cases. This paper's modeling technique is
applicable to rigid and exible walls with di erent
SUMMARY AND CONCLUSIONS end conditions. Using this model, nonlinear dynamic
nite element time history analyses were performed
In this paper, a method for modeling retaining wall sys- on several soil-wall systems to arrive at the following
tems was proposed. The model includes free eld soil, conclusions:
a retaining wall and springs modeling the interfacing
soil. The method is exible enough to be used under 1. Comparing the results of analyses with the modi ed
di erent soil and wall conditions and has satisfactory M-O [3] method shows that the M-O method is
precision. The popular M-O method's assumptions, not accurate in every case, but, on average, gives
such as a rigid gravity wall and a rigid wedge of soil acceptable results for practical purposes.

www.SID.ir
Archive of SID
Seismic Soil Pressure on Retaining Walls 283

2. The notion that the seismic soil pressure increases 8. Finn, W.D., Yodendrakumar, M., Otsu, H. and Steed-
with increasing PGA of an earthquake is not true man, R.S. \Seismic response of a cantilever retaining
in every case. It is found that seismic soil pressure wall", 4th Int. Conf. on Soil Dyn. and Earthquake
is more closely related to the area under the PSA Engrg., Southampton, pp. 331-431 (1989).
spectrum than the peak ground acceleration of an 9. Scott, R.F. \Earthquake-induced pressures on retain-
earthquake. ing walls", 5th World Conf. on Earthquake Engrg. Int.
Assn. of Earthquake Engrg., 2, Tokyo, Japan, pp.
3. For cantilever walls, seismic soil pressure is greatly 1611-1620 (1973).
in uenced by the value of the base rotational 10. Veletsos, A.S. and Younan, A.H. \Dynamic modeling
sti ness. Maximum soil pressure happens for rigid and response of soil-wall systems", J. Geotech. Engrg.,
xed-base walls. ASCE, 120(12), pp. 2155-2179 (1994).
4. Bridge abutments and propped bridge abutments 11. Richards, R.J., Huang, C. and Fishman, K.L. \Seismic
show similar soil pressure distributions as of rigid earth pressure on retaining structures", J. Geotech.
walls, but the peak seismic soil pressure is less. Engrg., ASCE, 125(9), pp. 771-778 (1999).
5. Three di erent methods were proposed for seismic 12. Veletsos, A.S. and Younan, A.H. \Dynamic response of
cantilever retaining walls", J. Geotech and Geoenviron
soil pressure distribution against retaining walls Engng, ASCE, 123, pp. 161-172 (1997).
(Equations 11-13). These methods are simple to
use and show good agreement with nonlinear nite 13. El-Emam, M.M., Bathurst, R.J. and Hatami, K.
\Numerical modeling of reinforced soil retaining walls
element analyses results, and also past analytical subjected to base acceleration", Proc. 13th World
and experimental research data. Conference on Earthquake Engineering, Vancouver,
6. For the case of a concentrated load applied at the BC, paper No. 2621 (2004).
top of an abutment, a simple pressure distribution 14. Green, R.A. and Ebeling, R.M. \Seismic analysis of
was also presented (Equation 17). cantilever retaining walls", Phase I, ERDC/ITL TR-
02-3. Information Technology Laboratory, US Army
7. An equation for estimating the rst mode period Corps of Engineers, Engineer Research and Develop-
of vibration for the soil-wall system was introduced ment Center, Vicksburg, MS, USA (2002).
(Equation 14). 15. Puri, V.K., Prakas, S. and Widanarti, R. \Retaining
walls under seismic loading", Fifth International Con-
REFERENCES ference on Case Histories in Geotechnical Engineering,
New York, NY, USA (2004).
1. Mononobe, N. and Matsuo, H. \On the determination 16. Green, R.A. and Ebeling, R.M. \Modeling the dy-
of earth pressures during earthquakes", Proc. World namic response of cantilever earth-retaining walls using
Engrg. Conf., 9, paper No. 388, pp. 177-185 (1929). FLAC", 3rd International Symposium on FLAC: Nu-
2. Okabe, S. \General theory of earth pressure", J. merical Modeling in Geomechanics, Sudbury, Canada
Japanese Soc. of Civ. Engrs., 12(1) (1926). (2003).
3. Seed, H.B. and Whitman, R.V. \Design of earth 17. Cheng, Y.M. \Seismic lateral earth pressure coe-
retaining structures for dynamic loads", The Specialty cients for c ' soils by slip line method", Computers
Conference on Lateral Stresses in the Ground and and Geotechnics, 30(8), pp. 661-670 (2003).
Design of Earth Retaining Structures, ASCE, pp. 103- 18. Psarropoulos, P.N., Klonaris, G. and Gazetas, G.
147 (1970). \Seismic earth pressures on rigid and exible retaining
walls", Soil Dyn. and Earthquake Engng., 25, pp. 795-
4. Andersen, G.R., Whitman, R.V. and Geremaine, J.T. 809 (2004).
\Tilting response of centrifuge-modeled gravity retain-
ing wall to seismic shaking", Report 87-14, Depart- 19. Choudhury, D. and Chatterjee, S. \Dynamic active
ment of Civil Engineering, M.I.T. Cambridge, MA earth pressure on retaining structures", Sadhana,
(1987). Academy Proceedings in Engineering Sciences, 31(6),
pp. 721-730 (2006).
5. Sherif, M.A., Ishibashi, I. and Lee, C.D. \Earth 20. Choudhury, D. and Subba Rao, K.S. \Seismic passive
pressures against rigid retaining walls", J. Geotech. resistance in soils for negative wall friction", Canadian
Engrg. Div., ASCE, 108(5), pp. 679-695 (1982). Geotechnical Journal, 39(4), pp. 971-981 (2002).
6. Wood, J.H. \Earthquake induced soil pressures on 21. Subba Rao, K.S. and Choudhury, D. \Seismic passive
structures", Doctoral Dissertation, EERL 73-05, Cali- earth pressures in soils", Journal of Geotechnical and
fornia Institute of Technology, Pasadena, CA (1973). Geoenvironmental Engineering, ASCE, 131(1), pp.
7. Ishibashi, I. and Fang, Y.S. \Dynamic earth pressures 131-135 (2005).
with di erent wall movement modes, soils and founda- 22. Itasca Consulting Group, Inc. \FLAC (Fast La-
tions", Japanese Soc. of Soil Mech. and Found. Engrg., grangian Analysis of Continua) user's manuals", Min-
27(4), pp. 11-22 (1987). neapolis, MN, USA (2000).

www.SID.ir
Archive of SID
284 S. Maleki and S. Mahjoubi

23. Schnabel, P.B., Lysmer, J. and Seed, H.B. \SHAKE- He then pursued his PhD degree in New Mexico
A computer program for earthquake response analysis State University at Las Cruces and nished it in
of horizontally layered sites", Earthquake Engineering 1988. He has many years of experience in structural
Research Center, University of California, Berkeley, design, both in the US and Iran. He has been
California, USA (1972). a faculty member at Bradley University in Illinois
24. Lysmer, J., Ostadan, F. and Chen, C.C. \SASSI2000- and Sharif University of Technology in Iran. He
A system for analysis of soil-structure interaction", has authored and/or coauthored over forty technical
University of California, Department of Civil Engineer- papers and has authored two books and a chapter
ing, Berkeley, California, USA (1999).
in the Handbook of International Bridge Engineer-
25. Huang, C. \Plastic analysis for seismic stress and ing to be published in 2011. His research area is
deformation elds", PhD dissertation, Dept. of Civil
Engrg., SUNY at Bu alo, Bu alo, NY, USA (1996). mainly focused on the Seismic Design of Bridges and
26. Gazetas, G., Psarropoulos, P.N., Anastasopoulos, I.
Buildings. He has also patented a seismic damper in
and Gerolymos, N. \Seismic behavior of exible retain- Iran.
ing systems subjected to short-duration moderately
strong excitation", Soil Dyn. and Earthquake Engng., Saeed Mahjoubi started his university studies in
24, pp. 537-550 (2004). 2000 at Sharif University of Technology (SUT) in Iran
27. Ostadan, F. \Seismic soil pressure for building walls: in the eld of Civil Engineering. Having nished
An updated approach", Soil Dynamics and Earthquake
Engineering, 25, pp. 785-793 (2005). his undergraduate studies in 2004, he continued as a
28. Ohara, S. and Maehara, H. \Experimental studies graduate student at SUT in Structural Engineering.
of active earth pressure", Memoirs of the Faculty of He successfully defended his thesis entitled \Seismic
Engineering, Yamaguchi University, 20(1), pp. 51-64 Behavior of Integral Abutment Bridges" to obtain his
(1969). MS degree in 2006. He was accepted as a PhD student
29. Nakamura, S. \Reexamination of Mononobe-Okabe at SUT in 2008. During his graduate studies he
theory of gravity retaining walls using centrifuge model attended several international conferences to present
tests", Soils and Foundations, 46(2), pp. 135-146 his research which is mainly focused on Bridge Seismic
(2006). Analysis and Retro t (Structural Engineering World
Congress, Bangalore, 2007; 3rd International Bridge
BIOGRAPHIES Conference, Tehran, 2008, and 2 articles presented at
Shervin Maleki obtained his BS and MS degrees the 8th International Congress on Civil Engineering,
from the University of Texas at Arlington with honors. Shiraz, 2009).

www.SID.ir