Comparison of FEA and analytical methods for determining stability of a RAP supported MSE wall

Global stability is one of the failure modes that must be analysed for retaining walls. Limit equilibrium analysis of walls using slope stability software tends to result in a factor of safety that is either too high (circular surfaces) or too low (V-shaped non-circular surfaces). Finite element analysis (FEA) of walls provides a better solution but can be time-intensive and expensive. The primary aim of this project is to compare the results of FEA models with a simpler analytical bearing capacity method that uses Meyerhof’s load inclination correction factors. In particular, cases were examined where Rammed Aggregate Pier reinforcing elements (RAPs) support a mechanically stabilised earth (MSE) retaining wall. For this project, several FEA models replicating these cases were created. Geometric parameters included the area ratio of RAP to matrix soil, or “replacement ratio”, and the dimensions of the MSE wall. Each geometric configuration was then iterated over a range of undrained strength for the matrix soil, resulting in a different factor of safety for each model. A spreadsheet was also created containing the necessary calculations for the Meyerhof bearing capacity method. The factor of safety from the Meyerhof method was compared to the factor of safety computed for each corresponding FEA model. The results show an excellent relationship between the computed factors of safety for FEA models and the bearing capacity method, especially for factors of safety ranging from 1 to 1.5. At factors of safety above about 1.5, the critical failure mode becomes sliding rather than global stability, and the two methods diverge. The major implications of this research are that a complex FEA model can potentially be replaced by the simpler analytical Meyerhof bearing capacity method. Wall designers will benefit from a quick check on the global stability of a retaining wall without having to spend the time and money on more expensive FEA modelling.