| he state of the art of analyzing the seismic behavior of earth dams is briefly reviewed. Two model embankments containing stratified soils, i.e., alternating layers of clay and sand are subjected to base shaking in a 1 m radius centrifuge using a servo-hydraulic shaker. The model embankments contained three sand layers in the upstream and downstream sides alternating between clay layers, a central clay core, an upstream clay blanket and a downstream berm. The same base motion is applied to both the models and similar failure pattern, pore pressure time-histories and acceleration-time histories are obtained. Triggered by loosening and weakening of sand close to the bottom clay/sand interfaces, model dams failed, in general, with layers moving downward and outward from the centerlines.;Predictions made by DYSAC2 are first verified using the results from centrifugal testing of a relatively homogeneous and simple clay embankment and then using the centrifuge test results of heterogeneous, complex stratified soil embankments described above. In both cases, reasonable comparison is obtained between measured and predicted results. In clay embankment, high rate of straining experienced by the soil during centrifuge testing compared to low rate of straining in triaxial tests seems to have affected the predictions made by DYSAC2. Lack of interface elements and large displacement theories in DYSAC2 affected the predictions for the stratified soil embankment.;An analysis procedure which treats seismic loading of geotechnical engineering structures as a boundary value problem is developed. This analysis procedure is based on the finite element solution of the fully coupled dynamic governing equations of saturated porous media. In this analysis procedure, the stress-strain behavior of soils is described by bounding surface plasticity models. This analysis procedure is implemented as a computer code named DYSAC2. In DYSAC2 a three parameter time integration scheme named... |