Data-driven surrogate modeling of seismic uplift capacity for strip anchors in frictional–cohesive soils with surcharge using isogeometric analysis
Article
Nguyen-Minh, Toan, Bui-Ngoc, Tram, Shiau, Jim, Nguyen, Tan and Nguyen-Thoi, Trung. 2025. "Data-driven surrogate modeling of seismic uplift capacity for strip anchors in frictional–cohesive soils with surcharge using isogeometric analysis." Ocean Engineering. 336. https://doi.org/10.1016/j.oceaneng.2025.121769
Article Title | Data-driven surrogate modeling of seismic uplift capacity for strip anchors in frictional–cohesive soils with surcharge using isogeometric analysis |
---|---|
ERA Journal ID | 4710 |
Article Category | Article |
Authors | Nguyen-Minh, Toan, Bui-Ngoc, Tram, Shiau, Jim, Nguyen, Tan and Nguyen-Thoi, Trung |
Journal Title | Ocean Engineering |
Journal Citation | 336 |
Article Number | 121769 |
Number of Pages | 26 |
Year | 2025 |
Publisher | Elsevier |
ISSN | 0029-8018 |
1873-5258 | |
Digital Object Identifier (DOI) | https://doi.org/10.1016/j.oceaneng.2025.121769 |
Web Address (URL) | https://www.sciencedirect.com/science/article/pii/S0029801825014751 |
Abstract | The behavior of strip anchors under uplift loads in complex conditions, such as frictional–cohesive soils with surcharge and seismic loading, poses significant engineering challenges. This study presents a data-driven surrogate modeling framework to predict seismic uplift capacity, combining high-fidelity simulations with machine learning. A large dataset is generated using Upper Bound Limit Analysis (UBLA) implemented via Isogeometric Analysis (IGA) and Second-Order Cone Programming (SOCP). Key input variables i.e., cohesion factor (c'/γB), internal friction angle (φ′), embedment ratio (H/B), seismic coefficients (Kh, Kv), and surcharge factor (q/γB), are evaluated to explore their impact on uplift capacity (Pu/γB). Sensitivity analysis highlights embedment ratio and internal friction angle as major factors enhancing uplift, while seismic coefficients reduce it, especially in low-cohesion soils. The Levenberg-Marquardt Neural Network (LMNN) outperforms other machine learning models, achieving a mean squared error (MSE) of 0.006 ± 0.000 on training and 0.005 ± 0.001 on testing data, with an R2 of 1.000. This integration of IGA-UB with LMNN provides reliable upper-bound estimates for uplift capacity, particularly in high-cohesion soils. The development of a closed-form solution ensures practical relevance, making this hybrid approach a powerful tool for engineering design under varied loading scenarios. |
Keywords | Closed-form solution; Uplift capacity; Strip anchors; Frictional-cohesive soils; Isogeometric analysis; Machine learning; Sensitivity analysis |
Contains Sensitive Content | Does not contain sensitive content |
ANZSRC Field of Research 2020 | 400502. Civil geotechnical engineering |
Public Notes | Files associated with this item cannot be displayed due to copyright restrictions. |
Byline Affiliations | Ton Duc Thang University, Vietnam |
Van Lang University, Viet Nam | |
School of Engineering |
Permalink -
https://research.usq.edu.au/item/zyq36/data-driven-surrogate-modeling-of-seismic-uplift-capacity-for-strip-anchors-in-frictional-cohesive-soils-with-surcharge-using-isogeometric-analysis
19
total views0
total downloads7
views this month0
downloads this month