Professional Documents
Culture Documents
Seismic Earth Pressures On Cantilever Retaining Structures 2010
Seismic Earth Pressures On Cantilever Retaining Structures 2010
Abstract: An experimental and analytical program was designed and conducted to evaluate the magnitude and distribution of seismically
induced lateral earth pressures on cantilever retaining structures with dry medium dense sand backfill. Results from two sets of dynamic
Downloaded from ascelibrary.org by UNIVERSITY OF NEW ORLEANS on 10/09/17. Copyright ASCE. For personal use only; all rights reserved.
centrifuge experiments and two-dimensional nonlinear finite-element analyses show that maximum dynamic earth pressures monotoni-
cally increase with depth and can be reasonably approximated by a triangular distribution. Moreover, dynamic earth pressures and inertia
forces do not act simultaneously on the cantilever retaining walls. As a result, designing cantilever retaining walls for maximum dynamic
earth pressure increment and maximum wall inertia, as is the current practice, is overly conservative and does not reflect the true seismic
response of the wall-backfill system. The relationship between the seismic earth pressure increment coefficient 共⌬KAE兲 at the time of
maximum overall wall moment and peak ground acceleration obtained from our experiments suggests that seismic earth pressures on
cantilever retaining walls can be neglected at accelerations below 0.4 g. This finding is consistent with the observed good seismic
performance of conventionally designed cantilever retaining structures.
DOI: 10.1061/共ASCE兲GT.1943-5606.0000351
CE Database subject headings: Retaining walls; Earth pressure; Seismic effects; Earthquakes; Centrifuge models; Numerical
models; Simulation; Seismic design; Backfills.
Author keywords: Retaining walls; Earth pressures; Earthquakes; Centrifuge models; Numerical models; Simulation; Seismic design;
Backfills.
Introduction 共1970兲 are widely used in practice and are generally used as a
standard to which other solutions are compared. Nevertheless,
The seismic response of retaining structures is a complex soil- over time, there have been studies raising questions about the
structure interaction problem. Wall movements and dynamic earth general applicability of this methodology. Some analytical studies
pressures depend on the response of the soil underlying the wall, suggested that the MO method may lead to unconservative esti-
the response of the backfill, the inertial and flexural responses of mates of the dynamic earth pressures 共e.g., Green et al. 2003;
the wall itself, and the nature of the input motions. In engineering Ostadan 2005兲, while other analytical studies and field observa-
practice, the current state of the art 共Anderson et al. 2009兲 is tions suggested that the MO method is conservative, if not overly
largely founded on the pioneering work performed in Japan fol- conservative 共Gazetas et al. 2004; Psarropoulos et al. 2005兲.
lowing the great Kanto Earthquake of 1923 by Okabe 共1926兲 and Most important, with the increased awareness of seismic risks
Mononobe and Matsuo 共1929兲. The method generally known as and a better understanding of the magnitude of potential ground
the Mononobe-Okabe 共MO兲 method is based on Coulomb’s motions, designers of retaining structures in regions in close prox-
theory of static soil pressures and was originally developed for imity to seismic sources have increasingly found that the applica-
gravity walls retaining cohesionless backfill materials. Later stud- tion of the MO method, as later modified by Seed and Whitman
ies mostly adopted the same general approach and were similarly 共1970兲, leads to very large seismic forces that make the resulting
based on analytical solutions assuming ideal cohesionless backfill structures prohibitively expensive. Yet, while damage to cantile-
or on experimental data from relatively small scale shaking table ver retaining structures supporting above-grade embankments has
experiments with cohesionless backfill. Whereas uncertainty was been reported in recent large earthquakes 共Tatsuoka et al. 1996兲,
acknowledged regarding the position of the point of application of there is no indication that conventionally well-designed cantilever
the resultant dynamic earth pressure, most sources tend to agree retaining structures with level nonliquefiable backfill experience
that the MO method gives adequate results 共e.g., Prakash and significant failures due to seismic loading 共e.g., Lew et al. 1995;
Basavanna 1969; Seed and Whitman 1970; Bolton and Steedman Gazetas et al. 2004; Lew 2009兲. In fact, some well-documented
1982; Sherif et al. 1982; Ortiz et al. 1983; Steedman and Zeng case histories show that retaining structures designed only for
1990兲. The MO method and its adaptation by Seed and Whitman static loading perform reasonably well under seismic loading with
peak ground acceleration up to 0.5 g 共Clough and Fragaszy 1977;
1
Postdoctoral Scholar, Dept. of Civil and Environmental Engineering, Gazetas et al. 2004兲.
Univ. of California, Berkeley, CA 94720 共corresponding author兲. To address this apparent discrepancy between analysis and
2
Professor, Dept. of Civil and Environmental Engineering, Univ. of performance, we embarked on a series of dynamic centrifuge ex-
California, Berkeley, CA 94720. periments in order to study the seismic behavior of the retaining
Note. This manuscript was submitted on August 31, 2009; approved
wall-backfill systems and to evaluate the magnitude and distribu-
on February 27, 2010; published online on March 6, 2010. Discussion
period open until March 1, 2011; separate discussions must be submitted tion of seismic earth pressures on cantilever retaining walls. A
for individual papers. This paper is part of the Journal of Geotechnical two-dimensional 共2D兲 finite-element 共FE兲 model was also devel-
and Geoenvironmental Engineering, Vol. 136, No. 10, October 1, 2010. oped and calibrated against the centrifuge results to assess the
©ASCE, ISSN 1090-0241/2010/10-1324–1333/$25.00. ability of numerical models to capture the essential features of the
The MO method is the earliest and most widely used method for Two series of dynamic centrifuge experiments were performed on
estimating the magnitude of seismic forces acting on a retaining cantilever walls connected with a stiff floor slab 共open channel
Downloaded from ascelibrary.org by UNIVERSITY OF NEW ORLEANS on 10/09/17. Copyright ASCE. For personal use only; all rights reserved.
wall. The MO theory assumes that the retaining wall yields suf- structures兲 designed to represent prototype structures under de-
ficiently that a soil wedge at the point of incipient failure develops sign for the Bay Area Rapid Transit System 共BART兲 and Valley
behind the wall and that this wedge behaves as a rigid body. The Transportation Authority 共VTA兲 共Al Atik and Sitar 2008兲. The
MO method gives the total active thrust acting on the wall by experiments were carried out on the dynamic centrifuge at the
applying pseudostatic equilibrium to the forces acting on the soil Center for Geotechnical Modeling at the University of California,
wedge with the point of application of the thrust at 1/3 the height Davis. Technical specifications for the centrifuge and shaking
of the wall above its base 共Mononobe and Matsuo 1932兲. table are available in the literature 共Kutter et al. 1994兲. All results
After reviewing the results of experimental work based on are presented in terms of prototype units unless otherwise stated.
small 1 g shaking table experiments, Seed and Whitman 共1970兲
suggested that the point of application of the dynamic increment
thrust should be between one half to two thirds the wall height Model Test Configurations and Preparation
above its base. However, they observed that the peak ground ac-
celeration occurs for only one instant of time and does not have The models were constructed in a flexible shear beam container
sufficient duration to cause significant wall movements. Thus, and the centrifugal acceleration used in the two experiments was
Seed and Whitman 共1970兲 concluded that “many walls adequately 36 g. A profile view of the full centrifuge model configuration is
designed for static earth pressures will automatically have the shown in Fig. 1. The model structures used in the experiments
capacity to withstand earthquake ground motions of substantial were open channel structures with cantilever walls and a stiff base
magnitudes and in many cases, special seismic earth pressure pro- constructed of aluminum. In prototype scale, the retaining struc-
visions may not be needed.” tures were 5.67 m high. The difference in the structures was the
While many additional analytical and experimental studies flexibility of the walls, with the walls of the stiff structure being
have been conducted in the last 80 years 共e.g., Bolton and Steed- about 6 times stiffer than those of the flexible structure. The es-
man 1982; Sherif et al. 1982; Ortiz et al. 1983; Steedman 1984; timated natural periods of the stiff and flexible walls were 0.11
Steedman and Zeng 1990; Stadler 1996; Madabhushi and Zeng and 0.24 s, respectively. In plan view, both structures spanned the
2007兲, most of the basic assumptions inherent in the MO ap- width of the container. The sand used in both centrifuge models
proach have remained unchallenged. The latest NCHRP report was fine, uniform, angular Nevada sand with minimum and maxi-
共Anderson et al. 2009兲 recommended the use of the MO method mum dry densities of 14.50 and 17.49 kN/ m3, respectively.
for the design of nongravity cantilever walls and the current The first centrifuge experiment, LAA01, was performed on a
AASHTO LRFD Bridge Design Specifications suggest that a seis- two-layer sand model. The stiff and flexible structures sat on dry
mic coefficient equal to half the peak ground acceleration is to be medium-dense sand 共Dr= 73%兲 of approximately 12.5 m height
used in the design of nongravity cantilever walls ignoring the wall and supported a dry medium-dense sand backfill 共Dr= 61%兲.
inertial forces. The estimated natural period of the base soil prior to shaking was
Yet, cantilever retaining walls supporting a variety of non- 0.19 s. The natural period of the backfill soil could not be esti-
liquefiable soils have performed reasonably well during past mated due to a glitch in the instrumentation. The second centri-
earthquakes. It is important to note that some of these retaining fuge experiment, LAA02, was performed on a uniform density
walls were not designed for seismic loading and others were de- sand model 共Dr= 72%兲. The estimated natural periods of the base
signed for base accelerations not more than 20% of the peak and of the backfill soil prior to shaking were 0.19 and 0.15 s,
accelerations that they actually experienced during shaking respectively.
共Gazetas et al. 2004兲. Clough and Fragaszy 共1977兲 investigated The sequence of the model construction consisted of first plac-
the seismic performance of open channel floodway structures in ing the sand underneath the structures. The stiff and flexible struc-
the Greater Los Angeles area during the 1971 San Fernando tures were then placed at their appropriate locations. The backfill
earthquake. They concluded that “peak accelerations of up to 0.5 sand was later placed behind the walls. The sand was placed using
g were sustained by the floodways with no damage even though dry pluviation in different layers. The height of each layer corre-
no seismic loads were explicitly considered in the design.” De- sponds to a horizontal array of instruments. The soil density was
spite some damage to cantilever retaining walls during the 1995 produced by calibrating the drop height, mesh opening, and speed
Kobe earthquake 共Tatsuoka et al. 1996兲, well-documented case of drop for the pluviator. After placement of each layer, the sand
histories from the 1971 San Fernando earthquake 共Clough and surface was smoothed with a vacuum and instruments were
Fragaszy 1977兲, 1994 Northridge earthquake 共Lew et al. 1995兲, placed at their specific positions. Industrial grease was placed
1995 Kobe earthquake 共Koseki et al. 1998兲, and the 1999 Athens between the structures’ walls and the sides of the container to
earthquake 共Gazetas et al. 2004兲 show that cantilever retaining provide a frictionless walls/container boundary and prevent sand
structures with nonliquefiable level backfill perform well under from passing through. Lead was added to the structures in small
seismic loading, even if they were not specifically designed to pieces of 6.45 cm2 共1 in.2兲 each in order to match the masses of
handle dynamic loads. Most recently, Nakamura 共2006兲 per- the reinforced concrete structures.
Fig. 6. Back-calculated dynamic earth pressure coefficients at the time of maximum dynamic wall moments on the stiff and flexible walls as a
function of peak ground acceleration measured at top of soil in free field
Numerical Analyses adding the retaining structures and the soil-structure springs and
applying the structures’ weight; and 共4兲 updating the soil behavior
A 2D plane strain FE model of the channel structures and the to the plastic stage and applying the input ground motion. Static
backfill and base soil was developed on the OpenSees platform to and transient system responses were recorded throughout the
simulate centrifuge experiment LAA02, presented in Fig. 1. analysis.
Boundary conditions of the 2D FE mesh consisted of 共1兲 base
nodes of the soil continuum fixed horizontally and vertically to
Soil Constitutive Model
reproduce the fixed-base conditions of the model container; 共2兲
displacement degrees of freedom of the lateral boundary nodes of The uniform density dry sand of experiment LAA02 was modeled
the soil continuum tied together horizontally and vertically, 共3兲 by single-phase, four-node, quadrilateral elements. Pressure-
traction free surface; and 共4兲 dynamic excitation defined as the dependent multiyield 共PDMY兲 material was used to simulate the
recorded base acceleration. The simulated input earthquakes in- nonlinear sand response. This constitutive soil model is based on
cluded input acceleration time series recorded during the Loma the framework of multiyield surface plasticity, in which a number
Prieta-SC-1, Kobe-PI-2, and Loma Prieta-SC-2 shaking events in of conical yield surfaces with different tangent moduli are em-
centrifuge experiment LAA02. Peak accelerations varied from ployed to represent shear stress-strain nonlinearity and confine-
0.49 to 0.8 g. ment dependence of shear strength. Detailed description of the
The sequence of the FE analysis consisted of 共1兲 generating PDMY soil material and its parameters can be found in Yang
the soil mesh and specifying the boundary conditions; 共2兲 setting et al. 共2003, 2008兲 and Elgamal et al. 共2002, 2003兲.
the soil behavior to linear elastic and applying its self-weight; 共3兲
Model Calibration
Table 2. Main Modeling Parameters for Dry Medium-Dense Nevada Table 2 presents the main modeling parameters for the dry
Sand 共Dr= 72%兲 medium-dense sand used in experiment LAA02. The initial dry
Model parameter Parameter value soil mass density, void ratio, and reference shear modulus 共ob-
tained from shear-wave velocity兲 were based on centrifuge mea-
Initial mass density 共kg/ m 兲
3
1692 surements. Initial friction angle and phase transformation angle
Reference shear modulus, Gr 共kPa兲 5.30⫻ 104 were determined based on calibration against centrifuge results
Poisson’s ratio 0.3 and on values presented in Arulmoli et al. 共1992兲. Poisson’s ratio
Reference bulk modulus, Br 共kPa兲 1.15⫻ 105 was defined to result in a friction angle of 35° for normally con-
Reference confining stress, Pr⬘ 共kPa兲 54 solidated sand. The low strain bulk modulus was determined
Peak shear strain 0.1 using the elastic relation with the low strain shear modulus and
Pressure dependent coefficient 0.5 Poisson’s ratio. Peak octahedral shear strain and contraction and
Shear strain and G/Gmax pairs Based on Fig. 7 dialation parameters were defined based on recommendations
Friction angle 共deg兲 35 provided in Yang et al. 共2008兲. Liquefaction-induced strain con-
Phase transformation angle 共deg兲 27 stants were set to zero to deactivate the liquefaction mechanism.
Contraction constant 0.05 The yield surfaces in OpenSees were based on the shear modu-
Dilation constants d1 = 0.6, d2 = 3.0
lus reduction curve specified as G/Gmax and shear strain pairs.
According to the procedure outlined in Zeghal et al. 共1995兲 and
Liquefaction induced strain constants 0
Elgamal et al. 共2005兲, shear stress and shear strain responses at
Number of yield surfaces 11
different depths along the centerline of the soil were estimated
Void ratio 0.566
based on the one-dimensional shear beam idealization using the
Analytical Results
on the stiff and the flexible retaining walls observed in the cen-
trifuge experiment were reasonably simulated by the FE model.
Fig. 12 presents the maximum total earth pressure distributions
Fig. 10. Comparison of computed and recorded total wall moment interpreted from the strain gauge measurements, computed in
time series at different locations on the south stiff and north flexible OpenSees and estimated by the MO method using the peak
wall during Loma Prieta-SC-1 ground acceleration at the top of the soil in the free field for Loma
Prieta-SC-1 for both stiff and flexible walls. Both experimental
and analytical results show that maximum total earth pressures
Fig. 10 shows that the computed and recorded total wall mo- increase with depth. The trend and magnitude of total earth pres-
ments are in good agreement. The computed moments well repro- sures observed during the centrifuge experiments and computed
duced the phase and magnitude of the moment responses of the in OpenSees agree reasonably well. The computed profiles in gen-
stiff and flexible walls. Moreover, the FE model successfully eral fall somewhat below the values back-calculated from strain
simulated the gradual increase in static moment as a result of gauge measurements and above the values obtained from the tac-
shaking and soil densification and the cubic distribution of static tile pressure sensors. The numerical solution departs from the
and total moments along the depth of the walls. measured values at the base of each wall because the wall-base
connections were modeled in OpenSees as rigid moment connec-
tions, which do not reflect the true behavior at the joints where
Lateral Earth Pressures
some moment flexibility is present.
Fig. 11 presents a comparison of the computed and recorded total Overall, while the results of the numerical analyses confirmed
共static plus dynamic兲 lateral earth pressures at various locations the trends observed in the centrifuge experiments, the calibration
on the south stiff and north flexible walls during the Loma Prieta- of the numerical model was not trivial even with an extensive
SC-1 shaking event. Recorded earth pressures in Fig. 11 were
measured using the tactile pressure sensors while computed earth
pressures were obtained in OpenSees using spring force record-
ers. Lateral earth pressures in Fig. 11 were corrected such that
positive earth pressure corresponds to a force acting on the wall in
the direction away from the backfill.
As shown in Fig. 11, the degree of agreement between the
computed and recorded total earth pressures on the stiff and the
flexible walls is quite variable. Despite the reasonable phase
agreement between the recorded and computed total earth pres-
sure time series, computed values usually overestimated the
magnitude of the recorded total earth pressures. It should be noted
that the accuracy of the earth pressure magnitudes measured by
the tactile pressure sensors is limited due to drift and conditioning
problems encountered with these sensors during the centrifuge
experiments. Moreover, we found that the computed earth pres-
sures in OpenSees were sensitive to small variations in the prop- Fig. 12. Maximum total earth pressure profiles interpreted from the
erties of the wall-soil springs and the type of soil-structure strain gauge measurements, computed in OpenSees and estimated
interaction. Overall and despite the mentioned limitations, the using the MO method on the south stiff and north flexible walls for
main characteristics of the static and total earth pressures acting the Loma-Prieta-SC-1 shaking event
Conclusions References
The results of our experimental and numerical studies lead to Al Atik, L., and Sitar, N. 共2008兲. “Experimental and analytical study of
several important conclusions: the seismic performance of retaining structures.” PEER Rep.
1. The experimental and numerical analysis results consistently 2008104, College of Engineering, Univ. of California, Berkeley, Calif.
show that the maximum dynamic earth pressures increase Anderson, D. G., Martin, G. R., Lam, I., and Wang, J. N. 共2009兲. “Seis-
Downloaded from ascelibrary.org by UNIVERSITY OF NEW ORLEANS on 10/09/17. Copyright ASCE. For personal use only; all rights reserved.
with depth and can be reasonably approximated by a trian- mic analysis and design of retaining walls, buried structures, slopes,
gular distribution analogous to that used to represent static and embankments.” NCHRP Rep. 611, Transportation Research
earth pressures. Consequently, there seems to be no basis for Board, Washington, D.C.
the currently accepted position of the point of application of Arulmoli, K., Muraleetharan, K. K., Hosain, M. M., and Fruth, L. S.
the dynamic earth pressure force in dynamic limit equilib- 共1992兲. “VELACS laboratory testing program, soil data report.”
rium analyses at 0.6 to 0.67 H and, instead, the point of Project No. 90-0562, The Earth Technology Corporation, Irvine,
Calif.
application should be at 1/3 H, as originally suggested by
Bolton, M. D., and Steedman, R. S. 共1982兲. “Centrifugal testing of mi-
Mononobe and Matsuo 共1932兲.
croconcrete retaining walls subject to base shaking.” Proc., Conf. on
2. An important aspect of the dynamic interaction between the
Soil Dynamics and Earthquake Engineering, Vol. 1, Balkema, Rotter-
cantilever retaining walls and retained soils is the fact that dam, The Netherlands, 311–329.
that maximum dynamic earth pressures and maximum wall Clough, G. W., and Fragaszy, R. F. 共1977兲. “A study of earth loadings on
inertial forces do not tend to occur simultaneously. As a re- floodway retaining structures in the 1971 San Fernando Valley earth-
sult, the current design methods based on the MO theory quake.” Proc., 6th World Conf. on Earthquake Engineering. Vol. 3,
were found to significantly overestimate the recorded dy- Sarita Prakashan, Meerut, India.
namic earth pressures and moments. Elgamal, A., Yang, Z., Lai, T., Kutter, B., and Wilson, D. 共2005兲. “Dy-
3. The relationship between the back-calculated seismic earth namic response of saturated dense sand in laminated centrifuge con-
pressure increment coefficient 共⌬KAE兲 at the time of maxi- tainer.” J. Geotech. Geoenviron. Eng., 131共5兲, 598–609.
mum dynamic wall moment and peak ground acceleration Elgamal, A., Yang, Z., and Parra, E. 共2002兲. “Computational modeling of
obtained from our experiments suggests that seismic earth cyclic Mobility and post-liquefaction site response.” Soil. Dyn. Earth-
pressures on cantilever retaining walls can be neglected at quake Eng., 22, 259–271.
Elgamal, A., Yang, Z., Parra, E., and Ragheb, A. 共2003兲. “Modeling of
accelerations below 0.4 g.
cyclic mobility in saturated cohesionless soils.” Int. J. Plast., 19,
4. The analytical results show that the FE analysis is able to 883–905.
capture quite well the essential system responses observed in Gazetas, G., Psarropoulos, P. N., Anastasopoulos, I., and Gerolymos, N.
the centrifuge experiments. However, the veracity of the nu- 共2004兲. “Seismic behavior of flexible retaining systems subjected to
merical analyses is strongly dependent on access to high- short-duration moderately strong excitation.” Soil. Dyn. Earthquake
quality experimental or field performance data for model Eng., 24, 537–550.
calibration and, therefore, field performance predictions Green, R. A., Olgun, C. G., Ebeling, R. M., and Cameron, W. I. 共2003兲.
using numerical models should be approached with caution. Seismically induced lateral earth pressures on a cantilever retaining
Finally, while it is important to note that our results strictly wall, Proc., 6th U.S. Conf. and Workshop on Lifeline Earthquake
apply to dry cohesionless medium dense materials, even better Engineering, Long Beach, Calif.
performance can be expected in denser and moderately cemented Koseki, J., Tatsuoka, F., Munaf, Y., Tateyama, M., and Kojima, K. 共1998兲.
“A modified procedure to evaluate active earth pressure at high seis-
or cohesive materials. In addition, cantilever walls on individual
mic loads.” Special Issue on Geotechnical Aspects of the January 17,
footings could experience rocking and translation away from the
1996 Hyogoken-Nambu Earthquake, Soils and Foundations, 2, 209–
retained soil mass which, although mostly undesirable, would 216.
further decrease the seismic earth pressures. At this point, more Kutter, B. L., Idriss, I. M., Kohnke, T., Lakeland, J., Li, X. S., Sluis, W.,
experimental work and well-documented case histories are Zeng, X., Tauscher, R. C., Goto, Y., and Kubodera, I. 共1994兲. “Design
needed to fully explore the range of potential soil conditions and of a large earthquake simulator at UC Davis.” Proc., Int. Conf., Cen-
types of retaining structures and to further develop methods of trifuge 94, Balkema, Rotterdam, The Netherlands, 169–175.
analysis that are consistent with the actual dynamic behavior of Lew, M. 共2009兲. private communication.
these systems. Lew, M., Simantob, E., and Hudson, M. E. 共1995兲. “Performance of
shored earth retaining systems during the January 17, 1994,
Northridge earthquake.” Proc., 3rd Int. Conf. on Recent Advances in
Acknowledgments Geotechnical Earthquake Engineering and Soil Dynamics, Vol. 3, St.
Louis.
Madabhushi, S. P. G., and Zeng, X. 共2007兲. “Simulating seismic response
The writers gratefully acknowledge the support and valuable
of cantilever retaining walls.” J. Geotech. Geoenviron. Eng., 133共5兲,
input provided by Professor Bruce Kutter, Dr. Dan Wilson, and all 539–549.
the staff at the Center for Geotechnical Modeling at the Univer- Mononobe, N., and Matsuo, M. 共1929兲. “On the determination of earth
sity of California, Davis. This research was supported by a grant pressures during earthquakes.” Proc. World Engrg. Congress, 9, 179–
from the San Francisco BART and the Santa Clara VTA to the 187.
Pacific Earthquake Engineering Research Center at UC Berkeley. Mononobe, N. and Matsuo, M. 共1932兲. Experimental investigation of lat-
The writers also received valuable input from Ed Matsuda, Dr. eral earth pressure during earthquakes, Earthquake Research Institute
Jose Vallenas at BART, and James Chai at VTA. Any opinions, and Research Office of Public Works, 884–902.
Earthquake Engineering, Santiago, Chile. Yang, Z., Elgamal, A., and Parra, E. 共2003兲. “A computational model for
Psarropoulos, P. N., Klonaris, G., and Gazetas, G. 共2005兲. “Seismic earth liquefaction and associated shear deformation.” J. Geotech. Geoenvi-
pressures on rigid and flexible retaining walls.” Int. J. Soil Dyn. ron. Eng., 129共12兲, 1119–1127.
Earthquake Eng., 25, 795–809. Yang, Z., Lu, J., and Elgamal, A. 共2008兲. “OpenSees manual for PDMY
Seed, H. B., and Whitman, R. V. 共1970兲. “Design of earth retaining struc- and PIMY materials.” 具http://cyclic.ucsd.edu/opensees/典 共August
tures for dynamic loads.” Proc., ASCE Specialty Conf. on Lateral 2008兲.
Stresses in the Ground and Design of Earth Retaining Structures, Vol. Zeghal, M., Elgamal, A. W., Tang, H. T., and Stepp, J. C. 共1995兲. “Lotung
1, Cornell Univ., Ithaca, N.Y., 103–147. downhole array. II: Evaluation of soil nonlinear properties.” J. Geo-
Sherif, M. A., Ishibashi, I., and Lee, C. D. 共1982兲. “Earth pressure against tech. Engrg., 121共4兲, 363–378.