Airfoil2Vec: Spectral Geometry-Conditioned Neural Surrogate Models for Airfoil Aerodynamics and a Downforce-Generating CFD Dataset
Abstract
We introduce a dataset of approximately 10,000 Reynolds-Averaged Navier-Stokes (RANS) simulations of steady, incompressible, two-dimensional subsonic flow around downforce-generating NACA 4-digit airfoils, targeting aerodynamic regimes relevant to automotive and motorsport applications (openly available on https://huggingface.co/datasets/ratiolabs/downforce-airfoils). Using this resource, we study geometry-conditioned neural surrogates for fast flow prediction, comparing neural fields with neural ODEs, MLPs with graph-based models, and several spectral geometry-conditioning methods. We further propose Airfoil2Vec, an airfoil-specific spectral geometry encoder that combines the joint contour spectrum with separate spectral representations of camber and thickness, for predicting continuous pressure and velocity fields. We evaluate generalization through angle-of-attack interpolation, interpolation and extrapolation to unseen NACA 4-digit geometries, and generalization to unseen non-NACA airfoils. The resulting surrogate accurately captures aerodynamic quantities and qualitative flow features while providing orders-of-magnitude speedups over conventional computational fluid dynamics.
Related
- Microstructure-Conditioned Surrogate Models for Graded Multiscale Optimization of Mycelium Composites
- Extensible membranes in inviscid flow: aerodynamics and singular limits
- HiLiftAeroML: A High-Fidelity Computational Fluid Dynamics Dataset for High-Lift Aircraft Aerodynamics
- On the aerodynamics of an accelerating and decelerating propeller in hover at low Reynolds numbers
- Unsteady airfoil aerodynamics in attached flow: From unsteady thin airfoil theory to wind turbine application
BibTeX
@misc{borde2026airfoil2vec,
title = {{Airfoil2Vec: Spectral Geometry-Conditioned Neural Surrogate Models for Airfoil Aerodynamics and a Downforce-Generating CFD Dataset}},
author = {Haitz Sáez de Ocáriz Borde and Flavio Savarino and Andrei Cristian Popescu and Pietro Innocenzi and Pantelis Papageorgiou and Xerxes Xian Chong},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.38213},
eprint = {2609.38213},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.38213},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Airfoil2Vec: Spectral Geometry-Conditioned Neural Surrogate Models for Airfoil Aerodynamics and a Downforce-Generating CFD Dataset
AU - Haitz Sáez de Ocáriz Borde
AU - Flavio Savarino
AU - Andrei Cristian Popescu
AU - Pietro Innocenzi
AU - Pantelis Papageorgiou
AU - Xerxes Xian Chong
PY - 2026
DA - 2026/09/25
PB - arXiv
DO - 10.48550/arxiv.2609.38213
UR - https://arxiv.org/abs/2609.38213
AB - We introduce a dataset of approximately 10,000 Reynolds-Averaged Navier-Stokes (RANS) simulations of steady, incompressible, two-dimensional subsonic flow around downforce-generating NACA 4-digit airfoils, targeting aerodynamic regimes relevant to automotive and motorsport applications (openly available on https://huggingface.co/datasets/ratiolabs/downforce-airfoils). Using this resource, we study geometry-conditioned neural surrogates for fast flow prediction, comparing neural fields with neural ODEs, MLPs with graph-based models, and several spectral geometry-conditioning methods. We further propose Airfoil2Vec, an airfoil-specific spectral geometry encoder that combines the joint contour spectrum with separate spectral representations of camber and thickness, for predicting continuous pressure and velocity fields. We evaluate generalization through angle-of-attack interpolation, interpolation and extrapolation to unseen NACA 4-digit geometries, and generalization to unseen non-NACA airfoils. The resulting surrogate accurately captures aerodynamic quantities and qualitative flow features while providing orders-of-magnitude speedups over conventional computational fluid dynamics.
ER -