Comparing 2D and 3D slope stability in spatially variable soils using random finite-element method

Although the spatial variability of soil properties in slope stability analysis has long been recognized, the adoption of three-dimensional (3D) probabilistic slope models has been limited due to various technical challenges. Utilizing the random finite-element method (RFEM), this study presents a comprehensive comparison between two-dimensional (2D) and three-dimensional (3D) probabilistic slope stability analyses. Based on the results of 2000 Monte Carlo simulations, this work contributes valuable insights for understanding both factors of safety (FS) and failure characteristics in slope engineering and safety assessments. The results also reveal a strong correlation between FS in 2D and 3D RFEM analyses, with no significant bias in their means. Although 2D RFEM analyses can generate a much larger number of failure realizations at significantly lower computational costs, they cannot fully envelop the failure samples of 3D RFEM analyses. Moreover, 2D RFEM analyses tend to overestimate sliding volumes. In contrast, the 3D RFEM model excels in capturing complex slope geometries, thus providing more accurate representations of failure characteristics. This study advances the understanding of the strengths and limitations of 2D and 3D probabilistic slope stability analyses using RFEM, offering practical insights for real-world slope engineering applications.