Multiscale mechanics of bentonite-based superabsorbent polymer improved silt under freeze-thaw cycling: from cyclic damage mitigation to machine learning prediction
Accurate prediction of dynamic resilient modulus (MR) for seasonally frozen subgrades remains challenging because existing models rarely capture coupled freeze-thaw cycling (NFT) and moisture sensitivity in high-plasticity silts. This study investigated bentonite-based superabsorbent polymer (BT-SAP)-treated silt under freeze-thaw conditions. Prepared 225 repeated-load triaxial specimens with five initial moisture states, three BT-SAP contents (BC), and five NFT sequences (0–7 cycles); each underwent a 16-stage stress sequence, yielding 1200 averaged MR data points. BT-SAP significantly mitigated MR degradation; 0.25% content maintained MR > 55 MPa at NFT = 7. A six-parameter physical model and machine-learning models (GEP/ANN) outperformed classical frameworks (R2 > 0.88, RMSE < 10). Sensitivity analyses identified NFT and BC as dominant. BT-SAP lowered the freezing point (−4°C to −1°C) and reduced pore-water phase-change damage via hydrogel moisture regulation. The six-parameter model is recommended for field evaluation; 0.25% BC gave the most efficient laboratory improvement. This supports durable frost-region infrastructure.
Authors
- Jianwen Hao
- Yefeng Du (ORCID: https://orcid.org/0009-0008-3746-8228)
- Xiaoning Zhang (ORCID: https://orcid.org/0000-0002-7525-553X)
- Qing Jin (ORCID: https://orcid.org/0000-0003-3013-4205)
- Xinmei Li (ORCID: https://orcid.org/0009-0001-2768-3011)
- Wei Lv
- Shang Gao
- Xinzhuang Cui
- Xiongying Ma
Institutions
- Shandong University (CN)
- Chongqing University (CN)
Publication Details
- Journal
- Road Materials and Pavement Design
- Published
- 2026-11-17
- DOI
- https://doi.org/10.1080/14680629.2026.2761643
- Primary Topic
- Geotechnical Engineering and Soil Stabilization
- Type
- article
- Field-Weighted Citation Impact
- 0.00