An innovative approach to predict the bearing capacity of ring foundations in dense sand using Bolton failure criterion and hybrid XGBoost algorithms

This study investigates the bearing capacity of ring foundations resting on dense sand using two-dimensional finite element limit analysis (2D FELA) under axisymmetric conditions, incorporating the Bolton failure criterion. Six key design parameters – external radius (ro), radius ratio (ri/ro), soil unit weight (γ), critical-state friction angle (ϕcv), particle crushing strength (Q), and relative density (DR) – are examined for their influence on the bearing capacity factor (Nγ) and failure mechanism. The predicted bearing capacity (qu) from 2D FELA aligns well with experimental data from previous studies. A hybrid machine learning framework, integrating eXtreme Gradient Boosting (XGBoost) with Genetic Algorithms (GA), is proposed to enhance prediction accuracy. The GA-XGBoost model achieves high predictive performance (R² = 86.29%, RMSE = 0.2594). Shapley analysis identifies the radius ratio (ri/ro) as the most influential factor affecting Nγ. These findings contribute to more reliable engineering calculations for a ring footing on dense sand.