An Efficient Solution for Probabilistic Slope Seismic Stability Analysis Based on Polynomial Chao Kriging Metamodel

Abstract

Slope stability analysis plays a crucial role in geotechnical engineering, particularly in regions susceptible to seismic activity. The inherent non-homogeneity and uncertainty of soil properties pose significant challenges in assessing slope stability under seismic conditions. To address these complexities, a novel and efficient methodology named DUBLA-PDM-PCK is proposed. In this methodology, the effects of soil non-homogeneity and uncertainty, along with the time and spatial variations of seismic loading, are systematically considered. The deterministic framework integrates discretized upper bound limit analysis (DUBLA) to accommodate soil non-homogeneous characteristics, and the pseudo-dynamic method (PDM) to model seismic loading variability. Then, a robust and efficient probabilistic analysis method, PCK-MA, is implemented utilizing adaptive Polynomial Chaos Kriging metamodeling, Monte Carlo Simulation, and Analysis of Covariance to investigate the uncertainty of the parameters. This approach treats nine key parameters, including soil cohesion, friction angle, non-homogeneous coefficients, horizontal and vertical seismic coefficients, period, and amplification factor, as random variables to assess their uncertainty effects on failure probability (stability level) and sensitivity indices. The DUBLA-PDM-PCK methodology offers a streamlined and reliable tool tailored for assessing slope stability in seismic environments, demonstrating notable efficiency in addressing soil variability and seismic loading uncertainties. Its application holds promise for guiding engineering practices and enhancing understanding of slope behavior in regions prone to seismic hazards.