Probabilistic analyses of landslide hazards and risks: Bridging theory and practice

Slope Engineering is perhaps the geotechnical subject most dominated by uncertainty. The impact of uncertainty on the quality of slope performance predictions is often substantial. Current slope design practice based on the factor of safety cannot directly address uncertainty. Probabilistic slope stability analysis is a rational means to incorporate uncertainty in the design process. It is also the most suitable approach for estimating hazard frequency for site-specific quantitative risk analyses. Unfortunately, the geotechnical profession has been slow in adopting such techniques.; The objective of this work is to integrate probabilistic slope stability analysis into geotechnical practice as a practical design and decision-making tool. A spreadsheet approach for probabilistic slope analysis is developed. The methodology is based on Monte Carlo simulation using the commercial software @Risk and Excel. The analysis accounts for the spatial variability of the input variables as well as the various sources of systematic uncertainty. The output of the analysis is presented as the probability of unsatisfactory performance. It is a measure of the likelihood of the slope failing.; The methodology is tested through the analysis of 10 case studies. It proved practical and flexible in handling a wide variety of real slope problems including effective and total stress analyses, complex stratigraphy, circular and non-circular slip surfaces and different slope analysis methods.; The study indicates that the factor of safety alone can give a misleading sense of safety and is not a sufficient safety indicator. The probability of unsatisfactory performance is a more consistent safety measure. Current slope design practice is calibrated probabilistically through the analysis of case studies of failed and safe slopes. A comparison of the computed probabilities indicates that acceptable slope design practice is equivalent to a probability of unsatisfactory performance not exceeding 2 × 10−2, which could be regarded as an upper design threshold. Stability assessments based on the results of both deterministic and probabilistic analyses provide greater insight into design reliability and enhance the decision-making process. The study also shows that probabilistic slope analyses ignoring spatial variability of input parameters significantly overestimate the probability of unsatisfactory performance. Other conclusions regarding the implementation and practical value of probabilistic slope analyses are also reached.