Numerical Analyses of Slope Stability and Post-failure Behavior with Quantitative Uncertainty Assessment


Student thesis: Doctoral Thesis

View graph of relations



Awarding Institution
Award date20 Nov 2020


Slope failure or landslide is a common and severe geo-hazard that can cause significant casualties and economic losses. Therefore, landslide hazard analysis and subsequent risk assessment are of substantial importance, which require reasonable prediction of both occurrence of slope failure (e.g., slope failure probability) and failure consequence (e.g., post-failure behavior). Slope failure probability and post-failure behavior are significantly affected by spatial variability of in situ soil properties, which is considered as a major source of geotechnical uncertainties. During the past decades, geotechnical design codes worldwide have been migrating toward reliability-based design to provide a probabilistic and systematic means of accounting for geotechnical uncertainties encountered by practitioners. However, practitioners are unwilling to adopt probabilistic design in engineering practice. This is probably owing to the lack of real case histories of slope stability analysis using probabilistic methods that demonstrate the value of probabilistic analysis in engineering practice. Furthermore, the link between probabilistic slope design and conventional design practice is weak. Another likely reason is the lack of a user-friendly and non-intrusive numerical tool for geotechnical engineers who have difficulties in modifying numerical programs for probabilistic analysis and limited knowledge of probability theory and statistics.

To address the above-mentioned issues, a numerical framework called Numerical Analysis with Quantitative Uncertainty Assessment (NAQUA) is proposed in this study for probabilistic slope stability and post-failure behavior analysis. A graphical user interface (GUI) is developed using Python for facilitating the implementation of the proposed NAQUA framework. It provides a user-friendly and practical tool for geotechnical engineering practitioners to perform probabilistic slope analysis. Previous studies mostly focused on probabilistic slope stability analysis, and those on probabilistic analysis of the post-failure behavior of slopes are limited. In particular, the existing numerical tools in engineering design practice do not include any for conducting probabilistic post-failure behavior analyses. Therefore, a Monte Carlo simulation (MCS) based method called random smoothed particle hydrodynamics (RSPH) is proposed under the NAQUA framework. Herein, the random field theory is used in combination with smoothed particle hydrodynamics (SPH). The proposed RSPH method is capable of simulating the whole process of slope failure and providing reasonable estimates of both failure probability and post-failure behavior. With the proposed RSPH method, a sliding probability is proposed for a soil element within a slope, and the spatial-temporal variation in the sliding probability within a cohesive slope is explored to investigate the evolution of the spatial distribution of the failure slip surface during the entire process of a landslide. The soil sliding probability at each location within a slope is estimated at each stage of the failure process and used subsequently in the assessment and management of the landslide risk to adjacent structures.

NAQUA is implemented to perform probabilistic slope stability analyses on a series of real case histories of failed and stable slopes, which cover typical situations encountered in engineering practice such as fill slopes, slopes with soil nails, and slopes with retaining walls. Different failure modes are also investigated in the case studies, including the global sliding failure of slopes and the sliding and overturning failure of retaining walls. Based on the analysis results of these slope case histories, the value of probabilistic approach in slope analysis and design is demonstrated, thereby facilitating geotechnical practitioners to implement probabilistic slope analysis in engineering practice.