Numerical-based seismic displacement hazard analysis for earth slopes considering spatially variable soils

Abstract

This paper presents a numerical-based seismic displacement analysis for slopes considering the spatial variability of soils. Two generic slope models are developed based on a finite-difference approach, and the soil parameters such as cohesion and friction angle are modelled as random fields. Dynamic analyses are conducted to estimate the slope displacements based on both deterministic and random slope cases. Dozens of scalar intensity measures (IMs) and vector-IMs are utilized for developing the 1-IM and 2-IM-based displacement models. It is found that the scalar-IMs of spectral acceleration at a degraded period of the whole slope, spectrum intensity, and peak ground velocity, and the Arias intensity (IA) or cumulative absolute velocity-containing vector-IMs such as [IA, mean period] are the most efficient scalar- and vector-IMs for developing the displacement models, respectively. In addition, compared to the deterministic cases, the median predictions by the random-based displacement models are notably larger, attributing to the existence of relatively weak zones for the random slope cases. Based on the displacement models developed, two strategies are proposed to conduct the slope displacement hazard analysis considering the spatial variability of soils. Comparative results indicate that neglecting the soil spatial variability significantly underestimates the slope displacement hazard, and the proposed strategy 1 results in slightly greater displacement hazard than strategy 2. Such difference is mainly caused by the treatment of the aleatory variability within the two strategies.