UniFoil: A Universal Dataset of Airfoils in Transitional and Turbulent Regimes for Subsonic and Transonic Flows

TL;DR

UniFoil addresses the need for large, diverse CFD datasets that capture multi-regime aerodynamics, including laminar–turbulent transition and shock-wave interactions, which are poorly represented in prior public corpora. It provides over $5\times 10^5$ 2D airfoil simulations across Ma in $[0.1,0.85]$ and Re in $[1,10]\times 10^6$, using SA turbulence and an $e^N$ transition model and offering both fully turbulent and transition cases with Cp fields and aerodynamic coefficients. Key contributions include the first large-scale public dataset that resolves shocks and transition, compressible RANS data with standardized outputs and open reproducibility under CC-BY-SA, and a pipeline for high-throughput generation. The dataset enables rigorous ML benchmarking, evaluation of regime-generalization, and development of physics-aware surrogates for drag and lift across subsonic to transonic flows. Overall, UniFoil is positioned to advance CFD data-driven discovery and open science in aerospace and related fields.

Abstract

We present UniFoil, a large publicly available universal airfoil dataset based on Reynolds-averaged Navier-Stokes (RANS) simulations. It contains over 500,000 samples spanning a wide range of Reynolds and Mach numbers, capturing both transitional and fully turbulent flows across incompressible to compressible regimes. UniFoil is designed to support machine learning research in fluid dynamics, particularly for modeling complex aerodynamic phenomena. Most existing datasets are limited to incompressible, fully turbulent flows with smooth field characteristics, overlooking the critical physics of laminar\-turbulent transition and shock\-wave interactions\-features that exhibit strong nonlinearity and sharp gradients. UniFoil addresses this limitation by offering a broad spectrum of realistic flow conditions. Turbulent simulations utilize the Spalart\-Allmaras (SA) model, while transitional flows are modeled using an e^N\-based transition prediction method coupled with the SA model. The dataset includes a comprehensive geometry set comprising over 4,800 natural laminar flow (NLF) airfoils and 30,000 fully turbulent (FT) airfoils, covering a diverse range of airfoil designs relevant to aerospace, wind energy, and marine applications. This dataset is also valuable for scientific machine learning, enabling the development of data-driven models that more accurately capture the transport processes associated with laminar-turbulent transition. UniFoil is freely available under a permissive CC\-BY\-SA license.

Figures (12)

• Figure 1: The first row shows the pressure coefficient ($C_p$) distributions for a subsonic flow at Mach 0.2 and a transonic flow at Mach 0.8. The subsonic case exhibits a smooth pressure field, whereas the transonic case features two shock waves forming on the suction and compression sides of the airfoil. The second row presents contours of the dimensionless turbulent viscosity, $\mu_t / \mu$, for both a fully turbulent model and a laminar–turbulent transition model under identical flight conditions. The fully turbulent model assumes immediate transition at the leading edge, while the transition model predicts a delayed onset of turbulence, with the transition location indicated by arrows.
• Figure 2: An overview of FT-airfoils geometry sampling.
• Figure 3: An example of the structured mesh around airfoil generated by pyHyp Secco2021.
• Figure 4: An overview of the entire training pipeline. The modal coefficients $a_i$ are obtained from \ref{['eq:modal']}.
• Figure 5: $C_p$ field contours from the ground truth data (first row) and reconstructed data (second row). Absolute error field is plotted in the third row. This is the absolute error between the pressure coefficient values from ADflow and prediction from neural network, as shown in the first formula from top in the third row. In the third row, the numbers in each sub--figure represent the relative $L_2$ norm error as defined in the second formula from top.