Framework providing pythonic APIs, algorithms and utilities to be used with Modulus core to physics inform model training as well as higher level abstraction for domain experts
1) Improved the MoE for the forward problem solution that couples the dynamic properties map of the pressure, water saturation and gas saturation fields with the peaceman analytical well model in Learn_CCR.py by using a flexible user input decision on the choice of experts and gates. Using polynomial regression degree 3 or XGboost regressor as the choice of experts.
2) Added solution for the forward problem in the CO2-Brine composition model. Improved upon this model by adding a physics-informed loss from EOS and Darcy equations
3) Added a full reservoir characterization workflow for 2D and 3D cases that solves the inverse problem using generative models (VCAE/GAN/DDIM/CAE/PCA/SVD/KSVD/KPCA/AE/Level set) as a form of domain parametrisation.
4) Provided numerical solvers for an active learning approach in model learning during the forward problem for both the black oil and CCUS case
Modulus Pull Request
Description
1) Improved the MoE for the forward problem solution that couples the dynamic properties map of the pressure, water saturation and gas saturation fields with the peaceman analytical well model in Learn_CCR.py by using a flexible user input decision on the choice of experts and gates. Using polynomial regression degree 3 or XGboost regressor as the choice of experts. 2) Added solution for the forward problem in the CO2-Brine composition model. Improved upon this model by adding a physics-informed loss from EOS and Darcy equations 3) Added a full reservoir characterization workflow for 2D and 3D cases that solves the inverse problem using generative models (VCAE/GAN/DDIM/CAE/PCA/SVD/KSVD/KPCA/AE/Level set) as a form of domain parametrisation. 4) Provided numerical solvers for an active learning approach in model learning during the forward problem for both the black oil and CCUS case
Checklist
Dependencies