Search

Accurate estimation of slope stability based on numerous candidate estimation methods is difficult as different results may be yielded. It becomes even more challenging when only limited data of geotechnical parameters (e.g., shear strength parameters) are available to evaluate slope reliability. Based on the Bayesian sequential updating technology, a hybrid framework for slope reliability was proposed in this study, through which prior knowledge, multiple estimation methods, and corresponding model uncertainties could be integrated to estimate slope reliability using a small amount of geotechnical data. Three slope examples with various stratigraphic configurations and soil properties were used to illustrate the accuracy and efficiency of the proposed framework, during which the Bishop's simplified method, the upper bound limit analysis method, and the finite element method were adopted. The results showed that with results of direct Monte Carlo simulation based on each method as the benchmark, a compromised mean of the factor of safety (μFS), and conservative standard deviation of the factor of safety (σFS) and failure probability (Pf) were yielded through the proposed framework. When the sample size of geotechnical parameters was greater than a threshold, the estimated μFS was stable, while the σFS and Pf synchronously varied within a small range with the increase in sample size. Demonstrations of the three examples indicated that the proposed hybrid framework can provide reliable and accurate estimations of slope reliability. The proposed framework may serve as a promising vehicle for slope/landslide engineering including failure and preventative mechanisms, movement prediction, and back analysis of geotechnical parameters in a probabilistic context, and big data analysis of geological and geotechnical problems as well. [ABSTRACT FROM AUTHOR]