Bearing Capacity Enhancement of Hexagonal Skirted Footings: Numerical, Regression, and ANN-Based Prediction

This paper presents a comprehensive numerical analysis of the improvement in bearing capacity and settlement performance of hexagonal shallow footings with inclined skirts. Various numerical analyses were conducted using PLAXIS 3D to investigate the influence of skirt length-to-footing width (L/B) ratios and skirt inclination angles (θ) on hexagonal footings in loose sand. The models showed very good agreement with experimental data reported in previous studies, with an R² value of 0.996 and a maximum error of less than 4.31%. It was concluded that the inclusion of inclined skirts has a positive effect on bearing capacity, increasing it by up to approximately 2.97 times compared to non-inclined configurations, while significantly reducing settlement. In addition to numerical simulations, an empirical formula for bearing capacity and settlement was developed using multiple regression based on geometric and inclination parameters. The model demonstrated a good fit (R² = 0.993). Furthermore, an Artificial Neural Network (ANN) model with a 4-10-10-1 architecture was proposed to predict bearing capacity using normalized input parameters, including skirt depth, inclination angle, stress, and settlement ratio. During training, validation, and testing, R² values greater than 0.998 were achieved, indicating a high level of accuracy with low prediction error. These findings highlight the importance of skirt inclination in enhancing foundation design, providing an efficient and cost-effective approach to increase the safety factor of foundations constructed on weak soils without the need for additional structural elements such as panels or strips.