Experimental and numerical study on helical piles in aeolian sand: bearing behavior and design methods

Abstract

In aeolian sand, the mechanical behavior of helical anchors involves complex performance evolution mechanisms that are not yet fully understood. This study employs a multi-scale integrated approach combining field tests, numerical simulations, and machine learning to systematically investigate the evolution laws of the bearing behavior of helical anchors. The results indicate: (1) The critical embedment depth threshold for helical anchors in aeolian sand is H = 5D; beyond this threshold, the load direction effect can be neglected. (2) Multi-plate helical anchors exhibit significant geometrically nonlinear superposition behavior. Dense spacing (S/D < 4) produces notable stress superposition effects (η = 1.15–1.32), whereas wide spacing (S/D ≥ 4) results in independent bearing units (η = 0.97–1.03). (3) The XGBoost machine learning model identifies the internal friction angle, anchor plate diameter, and embedment depth ratio as the most influential features affecting bearing capacity. Based on these control parameters, predictive equations for the bearing capacity coefficient N_q and soil lateral friction coefficient K_u were developed, with predictions showing excellent agreement with experimental data. This provides engineers with a reliable analytical framework for performance-based design. The study not only deepens the understanding of the behavioral mechanisms of helical piles in aeolian sand but also offers practical solutions for geotechnical engineering practice.