Using Physics-Informed Neural Networks for Solving Navier-Stokes Equations in Complex Scenarios
| Title | Using Physics-Informed Neural Networks for Solving Navier-Stokes Equations in Complex Scenarios |
| Publication Type | Journal Article |
| Year of Publication | Submitted |
| Authors | Bottarelli T, Fanfani M, Nesi P, Pinelli L |
| Journal | Engineering Applications of Artificial Intelligence |
| Keywords | Physics-Informed Neural Networks; artificial intelligence; Fluid Dynamic; Partial Differential Equations; Navier-Stokes |
| Abstract | Physics-Informed Neural Networks (PINNs) offer a promising approach for solving Navier-Stokes equations in fluid dynamics by embedding physical laws directly into the training process. Traditional Computational Fluid Dynamics (CFD) methods, like those using finite volume or element approaches, require extensive computational resources and dense mesh, which are iteratively used for each time step of the solution. PINNs bypass some of these limitations by using neural networks to produce solutions based on the governing equations, reducing the need for large datasets, dense meshing and continuous estimation over time. This paper evaluates the application of PINNs in near real-world scenarios with various geometries, comparing their performance against CFD solutions obtained via fluid-dynamic simulators. The study focuses on the accuracy of results and training times across different neural network architectures, activation functions, and numbers of sampling points. Additionally, several training strategies such as fine-tuning, multi-resolution learning, and parametrized training are proposed to enhance efficiency, and obtaining speed up. Results demonstrate that PINNs can achieve comparable accuracy to CFD methods while significantly reducing computational costs. The findings suggest that with appropriate training techniques, PINNs can be effectively used in industrial applications requiring rapid and accurate fluid dynamic simulations, paving the way for their broader adoption in practical engineering problems. |
| Refereed Designation | Refereed |