Physics-Informed Gaussian Process Regression for the Constitutive Modeling of Concrete: A Data-Driven Improvement to Phenomenological Models
Chenyang Li, Himanshu Sharma, Youcai Wu, Joseph Magallanes, K.T. Ramesh, Michael D. Shields · 2026
Read Physics-Informed Gaussian Process Regression for the Constitutive Modeling of Concrete: A Data-Driven Improvement to Phenomenological Models by Chenyang Li, Himanshu Sharma, Youcai Wu, Joseph Magallanes, K.T. Ramesh, Michael D. Shields (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-03367.txt.
Description
Chenyang Li and five co-authors improve the standard Karagozian and Case concrete model by letting data, guided by physics, replace its weakest part. That widely used model depends on an empirically calibrated failure surface with little flexibility and no uncertainty estimates. The authors keep its modular elastoplastic structure but swap the failure surface for a constrained Gaussian process regression surrogate, learned from experimentally accessible measurements.
Trained on triaxial compression data at several confinement levels, the surrogate is tested at confinement levels it never saw. A plain Gaussian process interpolates well near training conditions but deteriorates under extrapolation and breaks physical constraints, even with simulated data added. The physics-informed version, which builds derivative information and constraints into the regression, holds up far better, pointing to safer data-driven constitutive models for civil and defense work.
Catalog record
| CREATOR | Chenyang Li, Himanshu Sharma, Youcai Wu, Joseph Magallanes, K.T. Ramesh, Michael D. Shields |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 66 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2601-03367