In this project, our team experimented Physics-Informed Neural Network approach to stimulate a 2D model in a naive Catalytic Reactor. The approach's result is competitive to traditional Computational Fluid Dynamics method, but is less computational & data-intensive, further showcasing the potential of applying PINNs into solving fluid dynamics and heat transfer problems.
Temperature simulation in a 2D Catalytic Reactor by PINNs
The implementation code is presented in both Tensorflow and DeepXDE (a library for applying PINNs at ease) in 2D_deepxde.ipynb and 2D_tensorflow.ipynb.
Thesis PINNS in Catalytic Reactor
The thesis is submitted and graded to Institut National des sciences appliquées de Toulouse (INSA Toulouse) for final project. Special thanks to my collaborators Mai Dinh Nam and Nguyen Phuc Luan, with the advise of Mrs Marie and Ms Liantsoa.