A comprehensive prediction of stability against cantilever retaining walls using finite element limit analysis and neural networks model

• The finite element analysis method was verified by a series of comprehensive comparisons with existing studies, and it was employed to established a numerical model for stability against cantilever retaining walls. • This study investigated the effect of various parameters on the failure mechanism and safety factor against cantilever retaining walls. • A surrogate model for a convenient prediction of the stability of cantilever retaining walls was proposed. Cantilever retaining walls are commonly used due to their lightweight and self-stabilized soil mass above base slabs. Although several studies have extended earth pressure theories for cantilever retaining walls, these theories have limited applicability to further predicting the stability against cantilever retaining walls. In this study, the finite element limit analysis method is verified through comparisons with existing experimental tests, and then a comprehensive numerical model for the stability of cantilever retaining walls is established. The model combines the finite element limit analysis method with the strength reduction method to cover most practical project situations. Compared with the limit equilibrium method, the finite element limit analysis model provided a prediction of the safety factor based on the most probable failure mechanism considering soil properties, wall geometries, and interface properties. Furthermore, a series of parameter studies are conducted to investigate the effect of various parameters on the stability of cantilever retaining walls. Based on a large dataset obtained from the model, a neural network model is trained and validated to provide a surrogate model for a convenient prediction of the stability of cantilever retaining walls.