Link to the paper - PINN-Poroelasticity
If you are interested in physics informed neural networks (PINNs) and coupled single and multiphase flow in porous media, please check out our work below:
- We find it challenging to solve coupled poroelasticity relations using PINNs (data-free).
- To resolve this issue, we propose a stable training strategy based on the sequential fixed-stress-split algorithm.
- We also explore the role of adaptive weights in training such relations simultaneously and sequentially.
| Attachment | Size |
|---|---|
| pinn_poroelasticity.jpg | 299.97 KB |