Accelerated Gradient-based Design Optimization Via Differentiable Physics-Informed Neural Operator: A Composites Autoclave Processing Case Study
Janak M. Patel, Milad Ramezankhani, Anirudh Deodhar, Dagnachew Birru · 2025
Read Accelerated Gradient-based Design Optimization Via Differentiable Physics-Informed Neural Operator: A Composites Autoclave Processing Case Study by Janak M. Patel, Milad Ramezankhani, Anirudh Deodhar, Dagnachew Birru (2025) 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-2502-11504.txt.
Description
Janak Patel and co-authors speed up engineering design with a physics informed neural operator in this 2025 paper. Optimizing a process by simulation is slow when each trial needs a heavy finite element run. Neural operators can replace those runs, but older ones fail on complex, high dimensional designs. Their case study is curing composite parts in an autoclave.
The new architecture, called PIDON, extends a deep operator network with physics constraints. It models nonlinear behavior across wide design ranges more accurately. Gradient based optimization can then run through the surrogate directly. The authors report accuracy beyond prior state of the art models. Faster trials mean more design options can be explored for curing temperature and pressure cycles.
Catalog record
| CREATOR | Janak M. Patel, Milad Ramezankhani, Anirudh Deodhar, Dagnachew Birru |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 47 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-11504