This repository contains the code and resources related to solving Poisson's equation through deep learning for computational fluid dynamics (CFD) applications. It is organized into two main folders:
This folder contains the work from a master thesis. It includes the implementation, experiments, and results related to solving Poisson's equation using deep learning techniques. For more details, please refer to the Master Thesis folder. You can also access the thesis PDF for a comprehensive overview of the research.
This folder consists of later works focused on improving the surrogate model developed in Chapter 4 of the master thesis. It includes additional experiments, optimizations, and enhancements to the existing model. Additionally, there are two additional frameworks introduced in this folder, as described in the README.md file inside the improved_sm folder. These frameworks aim to improve generalization and accuracy.
Although this repository focus on the surrogate models developed, the DLpisoFoam repository contains an implementation of an OPENFOAM CFD solver capable of using these surrogate models. You can find more information about this implementation in the DLpisoFoam repository.
To use the code in this repository, follow the instructions in each folder's respective README file.
1 - Application of machine learning to model the pressure poisson equation for fluid flow on generic geometries: https://link.springer.com/article/10.1007/s00521-024-09935-0
- This paper contains a deep description of the ML Surrogate Model developed to solve the Pressure Poisson Equation.
2- Enhancing CFD solver with Machine Learning techniques: https://www.sciencedirect.com/science/article/pii/S004578252400389X
- This paper contains the CFD solver benchmark
This project is licensed under the MIT License.
For any questions or inquiries, please contact [email protected].