Reproduce “Physics-informed neural network solution of thermo-hydro-mechanical (THM) processes in porous media” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Danial Amini, Ehsan Haghighat, Ruben Juanes
A Physics-Informed Neural Network (PINN) framework is established to simulate coupled thermo-hydro-mechanical (THM) processes in porous media. This framework utilizes dimensionless equations, a sequential training strategy, and transfer learning to accurately model complex subsurface phenomena, demonstrating alignment with established numerical methods across various 1D and 2D benchmark cases.
Key insights
- Formulating the THM governing equations in a dimensionless form effectively conditions the problem for deep learning algorithms, enhancing training stability and efficiency.
- The sequential training strategy successfully decouples the strongly coupled THM system into more manageable subproblems, improving the convergence and robustness of PINN optimizers.
- Transfer learning proves to be an effective strategy for PINNs to accurately solve problems with sharp gradients and complex boundary conditions, mitigating initial struggles in challenging scenarios.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.