An Efficient Procedure for Seismic Slope Stability Analysis Considering Input Uncertainties and Soil Spatial Variability

This paper presents a probabilistic slope stability analysis considering soil non-homogeneity and seismic loadings. One objective is to propose an efficient procedure for such analysis at least in a preliminary design stage of practical engineering. The procedure includes the Discretization Kinematic Approach (DKA) which can evaluate the non-homogeneous slope stability, the Pseudo-Dynamic Method (PDM) which allows considering the time and space variations of seismic loadings, and an adaptive Polynomial Chaos Kriging (PCK)-based probabilistic analysis method, which permits to construct a fast-to-evaluate metamodel and can provide failure probabilities, sensitivity index and other interesting results within limited simulations. By benefiting from the high computational efficiency of the proposed procedure, a variety of hypothetical cases are analyzed to give some insights into the following three issues: (1) comparison of PDM and the traditional pseudo-static method in a probabilistic framework; (2) sensitivity analysis of seismic intensity parameters (seismic coefficients, period and amplification factor) and soil parameters (unit weight, friction angle, cohesion and non-homogeneity coefficients); (3) discussion of incorporating random fields into this procedure aiming a better soil spatial modeling. The proposed procedure is expected to be practical for seismic landslide risk analysis due to its computational efficiency and versatility. Some insights about the seismic method selection and uncertainty modeling in the probabilistic analysis are provided.

Keywords: Probabilistic seismic analysis; Discretization kinematic approach; Pseudo-dynamic approach; Non-homogeneity; Random field.