This study investigates the bearing capacity of ring footings on frictional–cohesive soils under surcharge by analyzing three stability factors i.e., F c , F q , and F γ , based on Terzaghi’s principle of superposition. These factors are governed by two key inputs: the soil’s internal friction angle ( ϕ ) and the geometric ratio of the inner radius to the footing width ( r ₁/ B ). Advanced predictive models are developed by integrating finite element limit analysis (FELA) with an adaptive meshing technique and a data-driven Kolmogorov–Arnold Network (KAN). Building upon recent applications of KAN in geotechnical modeling, this study advances its use by demonstrating superior predictive accuracy and interpretability compared with traditional Artificial Neural Networks. A closed-form representation of the stability factors is derived from the trained KAN model, offering an efficient and transparent means for estimating bearing capacity. The optimized KAN framework achieved high coefficients of determination and low root mean square errors for both training and testing phases. Sensitivity and feature-importance analyses confirmed that ϕ exerts the dominant influence on stability, whereas r ₁/ B has a secondary effect. The results enhance the mechanistic understanding of ring-footing behavior on frictional–cohesive soils under surcharge and provide practical guidance for foundation design across diverse soil conditions.
Vu-Hoang et al. (Sun,) studied this question.