SEISMIC-INDUCED CYCLIC MOBILITY IN DENSE SAND STRATUM

Abstract

Five constitutive models are used to study the validity of dilation front hypothesis proposed by Lee & Schofield (1988) in a saturated dense sand stratum undergoing the behaviour of cyclic mobility. These include models which are curve-fitted to produce experimentally observed strain hardening behaviour of saturated dense sand, strain softening behaviour of loose sand as well as a semi-empirical model. These models are incorporated into a one-dimensional dynamic finite element program using Newmark' s time integration scheme to study the problem of an infinitely long, saturated sand stratum subjected to base excitation. Two types of base excitation are used, viz a sinusoid and a time history from a centrifuge model "earthquake". Using this analytical approach, the interaction between the non-linear cyclic mobility behaviour of the soil and the propagating shear waves were studied in detail. The results of the analyses show that strain hardening models are capable of generating high spiky acceleration peaks very similar to those observed experimentally. On the other hand, such peaks were not predicted by strain softening models. This supports Lee and Schofield's suggestion that the high acceleration peaks arise as a result of the recovery of stiffness in the sand. Results obtained by using the semi-empirical model also suggest that the recovery of stiffness arises due to the tendency of the dense sand to dilate. Lee and Schofield's postulation is shown to be largely correct but the effect of shear wave reflect ion at the ground surface may require further consideration. The mechanisms highlighted by this study is very different from those conventionally assumed in the analyses of dense sand embankments. This implies that new analytical methods may be required to account for these recently observed and verified mechanisms.