Site response in liquefiable layered deposits considering spatial variability in hydraulic conductivity

Previous studies have revealed that evaluation of the liquefaction hazard using mean estimates of soil properties may not adequately capture the response of intrinsically inhomogeneous granular soils, their interactions, and the subsequent inter-layer flow patterns that control the generation and redistribution of excess pore pressure during and after cyclic loading. Hence, they cannot reliably predict the liquefaction hazard and its consequences on site performance in terms of accelerations and deformations. In a parametric numerical study that was initially validated against centrifuge experiments, solid-fluid, fully-coupled, nonlinear dynamic effective stress analyses were performed to evaluate the effects of spatial variability of hydraulic conductivity (k) on seismic site response in layered sand deposits. The pressure-dependent, multiyield-surface, nonlinear, plasticity-based soil constitutive model (PDMY02) implemented in OpenSees was used to simulate the behavior of saturated sand. Analyses were conducted on a high performance computer using the parallel version of OpenSees framework. The Local Average Subdivision method was used to generate and map a stochastic k field over the finite element mesh. The base rock was simulated as an elastic half-space. The influence of k variability on the liquefaction hazard and consequences, including the timing of liquefaction, the resulting accelerations, and key intensity measures (IMs) were evaluated. The results indicate notable differences between the deterministic and median of the stochastic approaches in terms of key surface ground motion IMs as well as the extent and timing of excess pore pressure generation within the liquefiable layer.