Quantitative Risk Assessment of Seismically Loaded Slopes in Spatially Variable Soils with Depth-Dependent Strength

Abstract

Risk assessment of seismically loaded slopes is often a prerequisite for guiding decision-making and mitigation, which could be more difficult to quantify when taking the spatial variability of material properties into account. In this study, a random finite-element limit analysis (RFELA) is developed to assess the risk of a seismically loaded slope where the undrained shear strength of soils is spatially variable with depth. The nonstationary random field is first introduced to model the linearly increasing undrained shear strength. Then, finite-element limit analysis (FELA) is employed to evaluate the seismic stability and consequence, in which the seismic loading is characterized by the pseudostatic approach with a range of horizontal seismic coefficients. Finally, the quantitative risk assessment is conducted based on Monte Carlo simulations. The results show that using the nonstationary random field to model the soil spatial variability could significantly reduce the risk compared with the stationary random field, which matches better with the site-specific data. The risk of the slope failure increases with the increase in the seismic coefficient. In addition, the effects of the correlation structure of the undrained shear strength, which includes the coefficient of variation (COV) and autocorrelation distance, on the risk assessment are studied by parametric analyses.