Random field failure and post-failure analyses of vertical slopes in soft clays

Abstract

This research investigates the spatial heterogeneity of cohesion within soft clay and its implications for slope stability and post-failure analysis. In-situ cone penetration tests were conducted in alluvial soft clays to calibrate probabilistic strength properties. Slope stability analyses employed deterministic, semi-deterministic, and comprehensive probabilistic approaches, while post-failure analysis utilised the nodal integration-based particle finite element method. The undrained shear strength (cu) demonstrated a log-normal distribution (mean: 19 kPa, standard deviation: 3 kPa), with correlation lengths modeled through Bayesian inference. Treating correlation lengths as distributions resulted in a negligible 2% difference compared to using a single value for the probability of failure. Semi-deterministic analyses exhibited results similar to probabilistic analyses, offering computational advantages. Nevertheless, probabilistic analysis, considering spatial variability, provided more comprehensive insights for post-failure analysis. For a vertical slope of critical height in the studied soft clay, probabilistic analyses predicted a range of runout distances from 0 m to over 125 m. Specifically, 89% of these distances were less than 80 m, and 82% were less than 40 m. The findings contribute to an enhanced understanding of spatial variations in soil strength within soft clay slopes, providing valuable insights for future geotechnical assessments and design considerations.