Seismic Stability of Reinforced Soil Retaining Walls: Summary of Discussions for the Technical Sessions at the Seventh International Conference on Geosynthetics
Two 1.5-hour special technical sessions on the Seismic Stability of Reinforced Soil Retaining Walls were organized during the Seventh International Conference on Geosynthetics, which was held in Nice, France (September 22-27, 2002). The Sessions were chaired by the authors. There were a total of 12 presentations, with each speaker giving a 10-minute presentation followed by one or two short questions. The rest of time was made available for free discussion at the end of each Session. A list of presentations is given below followed by the summary of discussions. Note that most presentations were made based upon the papers submitted to the Conference.
J. Koseki, K. Watanabe, M.
Tateyama, and K. Kojima (Japan)
Comparison
of model shaking test results on reinforced-soil and gravity type retaining walls
J. Martin II (US) - absent
Preliminary implications from post-earthquake
investigations of conventional and geosynthetic-reinforced soil walls
C.C. Huang,
L.H. Chou, and Y.H. Chen (Taïwan)
Comparative study on the seismic displacements of
geosynthetic-reinforced modular block walls
K. Watanabe, M. Tateyama, T. Yonezawa, H. Aoki, F. Tatsuoka, and J. Koseki (Japan)
Shaking table tests on a new type bridge abutment
with geogrid-reinforced cement treated backfill
T. Uchimura,
F. Tatsuoka, M. Shinoda, Y. Sugimura, and T. Kikuchi (Japan)
Roles of tie rods for seismic
stability of preloaded and prestressed reinforced soil structures
M. El
Emam, and R.J. Bathurst (Canada)
Shaking
table model study on the dynamic response of reinforced soil walls
E. Guler, and M. Hamderi (Turkey)
Finite
element analysis of reinforced segmental retaining walls with cohesive and
granular backfills
Overall Discussions
Session II
J.H. Greenwood,
and C.J.F.P. Jones (United Kingdom)
Seismic
design and the strength of geosynthetic reinforcements
J. Takemura (Thailand) - absent
Centrifuge modeling of reinforced soil structures
C.C. Ng, S-H. Chew, G. Karunaratne, S.A.H. Tan,
and L. Heng (Singapore)
Geosynthetics reinforced soil wall subjected
to blast load
P. Rimoldi,
D. Peila, C. Castiglia,
and P. Recalcati (Italy)
Testing
and modelling geogrid reinforced soil embankments subject to high energy rock
impacts
Kongkitkul, W., Hirakawa, D., and F. Tatsuoka
Viscous deformation during cyclic loading of geosynthetic
reinforcement
D. Leshchinsky (USA)
Closure of Sessions
Jones questioned Huang about the motivation of studying the full-height wall in his shaking table tests, whereas the walls that failed in Taiwan were of segmental block facing. Huang indicated that the shaking table study he reported in the Conference were indeed of different study from the failure walls of Taiwan. Tatsuoka made a comment saying that for critical and important structures, local failure is possible through the intermediate height of the wall. Such lateral force is resisted by the rigid walls. Also, the walls in Taiwan were constructed with sloping foundation. For the unstable walls in Taiwan during the earthquake, he believed that the local failure at the facing triggered global failure.
Leshchinsky questioned Bathurst amplification of the wall during an earthquake. How would one extrapolate the results of amplification from 1-m wall to that of 5-m wall. Bathurst answered that scale effect is a concern. He suggested that one calibrates the numerical models and then use the tool for the analysis of a taller wall.
Wakeman asked Bathurst about the amplification with larger shaking intensity. Bathurst indicated that the boundary conditions, etc., in a model test are extremely important. A high wall is more critical and thus issues such as resonant frequency, etc., are important design considerations. Bathurst also mentioned that the stiffness of reinforcement is an important item to be looked at in reduced scale model tests.
An audience questioned about the pore pressure in the cohesive soil wall that Guler analyzed. Guler replied that drainage should be provided in the case of cohesive soil. Purely cohesive soil is practically impossible to construct. What he meant by cohesive soil is really the natural low quality soil that is free draining. The so-called cohesive soil in reinforced soil wall is different from fully saturated clay.
Another audience asked if Guler analyzed cases with shorter reinforcement. Guler replied that his analysis did not go beyond the reinforcement length of less than 70% the wall height. But he found that the length ratio exceeding 130% had no influence on the results.
Jones indicated that the failure mode shown by Guler was fully external. He suspects that multiple mechanisms are involved in a real failure. Guler suggested that his analysis by reducing c and f simultaneously is an extreme case of study. He did notice possible connection strength problem during staged construction due to the settlement of the backfill, but the settlement was not as significant during shaking. He re-emphasized that his results showed that external stability is as important as the internal stability.
An audience asked if the simulation of soil and reinforcement interaction behavior depends on the soil friction angle. Guler replied that sufficiently large number of elements cannot be placed near the reinforcement and soil using Paxis. However, since he reduced f for the soil, the interaction behavior of the reinforcement was affected in a similar way.
Adams asked why the tie rods were put outside the backfill and not inside it. Uchimura replied that it is for the sake of convenience. The manner the tie rods were installed, whether inside or outside the backfill, did not affect the results.
A question was asked if the specimen were shortened after the test. Greenwood replied yes.
Zornberg asked if the same specimen was used in the test 14 years ago and now. Greenwood replied that the specimens used were representative of each other.
Adams suggested that rock impact can be mitigated by creating slope with negative battress. Limited number of tests had not allowed Rimoldi to look at a different option.
Greenwood asked about the type of geosynthetic material used in the creep tests. Kongkitkul replied it was a new material. Greenwood was curious if different results will be obtained with other materials. Tatsuoka, a co-author replied that similar conclusions were obtained in their recent tests using other polymers such as polyester, vinylon, etc.
A professor from India asked if the vertical force was included in the seismic design. Leshchinsky replied FHWA/AASHTO does not consider vertical force. He also replied that the vertical acceleration shows greater effects if the corresponding acceleration is large.
Zornberg asked why the dynamic earth pressure effect was calculated using 50% the wall height. That was the suggestion by Seed and Whitman in a classical paper without any reasoning.
Alexview asked about the compound mode of failure in the reinforced soil structures. Consequently he feels that slope stability approach should be used in the analysis. Leshchinsky agreed because there should only be one critical failure surface, whereas FHWA/AASHTO design considers slope and wall with different approaches.
Another question was asked if AASHTO design guideline is for geosynthetic or it also includes metallic reinforcements. Leshchinsky replied that metallic reinforcement is part of AASHTO design guideline.
An audience asked for the speakers’ opinions about the exponential increase in earth pressure with a sloping backfill.
Leshchinsky suggested to consider f seriously. Most studies do not report if the peak or residual angle of friction was considered in their analysis.
Huang replied that he does not have an answer to the question.
Bathurst commented that the onset of failure should be based on the peak value whereas the displacement should be calculated using the residual value.
Koseki replied that he did not conduct calculation of earth pressure for his model tests, but in current Japanese design standard, both peak and residual values are used.
Tatsuoka made a final comment indicating that the localization issue should be answered using the peak value whereas the stability should be calculated using the residual value. He indicated that the issue is as critical for numerical analysis such as using the finite element method where strain localization has to be considered.
Another audience asked about the seismic design of sloping backfill to the presenters during the final discussion.
Leshchinsky replied that Mononobe-Okabe analysis is simply not the right mechanism for the sloping backfill. He, however, has the solution for design based on slope stability approach.
Tatsuoka suggested the selection of f in Mononobe-Okabe analysis is a key issue. f is not a constant value but varying along the failure plane. The use of residual value of f leads to very flat failure surface that is not observed in real.
A question was asked if the creep rate decreases upon unloading following a cyclic loading test. And if the creep effect is temporary or permanent.
Koseki asked Greenwood if creep behavior is a degradation phenomenon. Greenwood replied that creep behavior is a degradation factor because the residual strain is permanent.
Tatsuoka asked about AASHTO not having seismic design factor considering the creep of geogrid. He also commented that the residual strain curve should be established for different strain rates otherwise it is not unique because different strain rates give rise to different residual curves.
Tatsuoka concluded that based on the presentations, we now agreed that reinforced soil retaining walls performed differently from the conventional type of retaining structures. He also pointed out that we should continue to look at several unresolved issues. For example, the types of f used in design. We now know that rupture strength of polymeric reinforcement is a function of strain rate. He commented that creep reduction factor should not be used for geosynthetic reinforcement for the fact that shear strength of soil is not reduced in geotechnical design. Creep is a response of the material, which does not have a time memory.