Probabilistic stability analyses for sedimentary deposits

Abstract

This paper presents several approaches utilised by the author in assessing slope designs, inclusive of Probabilities of Failure, in sedimentary strata. A common issue seen by the author in probabilistic analyses is the use of population statistics, which honour variability in point sampling, but do not reflect variability at the larger scale. This then results in overestimates of the Probability of Failure. Issues in assessing the variability in inputs for analyses are discussed. Two case studies are presented with focus and discussion on use of the appropriate variability in the respective analyses. The cases have considered the scale at which the data is collected, and, critically, the analysis methodology which influences the approach in selection of variability. The case studies have utilised a Monte Carlo approach and use of limit equilibrium stability analysis software. Recent trends in analysis methodology (surface response methodology) and emergence of improvements in software (which allow generation of random fields) suggest the field of probabilistic analysis has matured. However, without careful consideration to the key design parameters, probabilistic analysis may simply serve to provide what appears as more sophisticated results, but which offer no additional value in managing risk for a project.