In this repository, you will find different python scripts to run incompressible or compressible Reynolds-Averaged-Simulations over NACA airfoils.
To launch a simulation, enter your parameters in the params.yaml
file and run:
python main.py -i 0 -g 1 -v 1 -f 1
usage: main.py [-h] [-i INIT] [-g GRADIENT] [-v VTK] [-f FIGURE]
optional arguments:
-h, --help show this help message and exit
-i INIT, --init INIT Only generate the mesh (default: 0)
-g GRADIENT, --gradient GRADIENT
Compute the term of the RANS equations as a post-processing (default: 0)
-v VTK, --vtk VTK Generate the VTK files from the simulation (default: 1)
-f FIGURE, --figure FIGURE
Save an image of the airfoil in the simulation folder (default: 1)
Those scripts have been used to generate the AirfRANS dataset proposed at the NeurIPS 2022 Datasets and Benchmarks Track conference. In particular, the script dataset_generator.py
run multiple simulations by sampling Reynolds number and Angle of Attack as explained in the associated paper.
This script can be re-used to run multiple new random simulations.
The mesh is generated with the blockMesh utility available in the OpenFOAM suite. The block definition is given in the following .
Some of the parameters contained in the params.yaml
file are for the mesh generation. Parameters are defined as:
L
: Size of the domain in metersy_h
: Heigth of the first cell of the boundary layer y_hd
: Heigth of the furthest first cell of the trail (at vertex 1 in the scheme)x_h
: Width of the smallest cell at the leading edge (at vertex 8 in the scheme)y_exp
: Expansion ratio in the y-directionx_exp
: Expansion ratio in the x-direction on the airfoil (edge between vertices 8 and 11)x_expd
: Expansion ratio in the x-direction behind the airfoil (edge between vertices 1 and 10)The original paper accepted at the 36th Conference on Neural Information Processing Systems (NeurIPS 2022) Track on Datasets and Benchmarks can be found here and the preprint here. Disclaimer: An important update correcting an inconsistency in the Machine Learning experiments proposed in the main part of the NeurIPS version of the paper has been done. Please refer to the ArXiv version for the up to date version.
Please cite this paper if you use this dataset in your own work.
@inproceedings{
bonnet2022airfrans,
title={Airf{RANS}: High Fidelity Computational Fluid Dynamics Dataset for Approximating Reynolds-Averaged Navier{\textendash}Stokes Solutions},
author={Florent Bonnet and Jocelyn Ahmed Mazari and Paola Cinnella and Patrick Gallinari},
booktitle={Thirty-sixth Conference on Neural Information Processing Systems Datasets and Benchmarks Track},
year={2022},
url={https://arxiv.org/abs/2212.07564}
}