|
Failure envelope prediction of skirt footings in sand under combined loading using 2D finite element limit analysis and machine learning |
|---|---|
| รหัสดีโอไอ | |
| Title | Failure envelope prediction of skirt footings in sand under combined loading using 2D finite element limit analysis and machine learning |
| Creator | Tawatchai Suesaming |
| Contributor | Suraparb Keawsawasvong, Advisor |
| Publisher | Thammasat University |
| Publication Year | 2568 |
| Keyword | Skirt footing, 2D failure envelope, Dense sand, Finite element limit analysis, Machine learning, ฐานรากแบบสเกิร์ต, วงวิบัติสองมิติ, ดินทรายแน่น, ไฟไนต์เอลิเมนต์แบบลิมิต, ปัญญาประดิษฐ์ |
| Abstract | This thesis investigates the failure envelope behavior of skirt footings embedded in dense sand under combined vertical and horizontal loading conditions using two-dimensional Finite Element Limit Analysis (2D FELA) integrated with the Bolton stress–dilatancy model. Four loading scenarios, including top-center bearing, top-center pullout, side bearing, and side pullout conditions, were considered. The effects of particle crushing strength, relative density, critical-state friction angle, embedment ratio, footing width, and loading inclination angle on the normalized load capacities were systematically examined. The results indicate that the failure envelope is strongly influenced by soil strength characteristics, footing geometry, and loading direction. In general, increases in particle crushing strength, relative density, and critical-state friction angle enhance the load-carrying capacity of the footing, whereas larger embedment ratios and footing widths tend to reduce the normalized failure envelope. The predicted failure mechanisms also demonstrate that the geometry of the failure zone varies significantly with both loading direction and loading location. To improve computational efficiency, an Extreme Gradient Boosting (XGBoost) model was developed using the numerical dataset generated from the FELA simulations. The model successfully captured the nonlinear relationships between the input variables and the failure envelope responses. Interpretability analyses further identified the loading inclination angle as the most influential parameter affecting the predicted responses. The proposed framework provides an efficient and reliable approach for evaluating and predicting the failure envelope behavior of skirt footings in dense sand under combined loading conditions. |
