AirfRANS (Original Meshes): 2D Airfoil RANS Simulations in PLAID Format

Provider
Florent Bonnet et al. (Extrality, Sorbonne Universite); PLAID version by Safran (Roynard, Casenave, Staber)
License
ODbL-1.0 (source)
Size
54 files
Format
Parquet (in sampled files)
Stored at
Hugging Face (https://huggingface.co/datasets/PLAID-datasets/AirfRANS_original)
Access
Free and public. No account with the data provider is needed.
Category
Physics
Source
https://arxiv.org/abs/2212.07564

About AirfRANS (Original Meshes): 2D Airfoil RANS Simulations in PLAID Format

AirfRANS is a dataset of two-dimensional incompressible steady-state Reynolds-averaged Navier-Stokes (RANS) simulations over NACA airfoils at subsonic flight conditions, introduced by Bonnet, Mazari, Cinnella and Gallinari (Extrality and Sorbonne Universite, NeurIPS 2022 Datasets and Benchmarks) to benchmark machine learning surrogates that predict velocity, pressure and turbulent viscosity fields and the resulting lift and drag. This repository is the PLAID-format version published by Safran and the PLAID team on the original simulation meshes, with an ML4PhySim_Challenge_train split matching the ML4PhySim challenge; a remeshed variant exists separately. Measured 2026-11-14, the data/ folder holds the samples and the repository uses about 19 GB. Samples are PLAID objects whose meshes follow the CGNS standard; load with Hugging Face datasets and convert with the plaid library (plaid.bridges.huggingface_bridge), or turn meshes into Muscat meshes. The AirfRANS data are licensed under the Open Database License (ODbL) v1.0, as stated by the original authors and on the Zenodo record of this version.

How to load this dataset in Python

from datasets import load_dataset
ds = load_dataset("PLAID-datasets/AirfRANS_original")
print(ds)

Data structure

File types in sampled files: Parquet

Files

Example file: data/all_samples-00000-of-00052.parquet

Analyze this dataset without downloading it

On Scigantic, AirfRANS (Original Meshes): 2D Airfoil RANS Simulations in PLAID Format opens in a hosted JupyterLab notebook with the data already mounted, so you can read files where they are stored instead of copying them first. Each notebook includes an AI assistant that can see the mounted files and write and run analysis code. Open the dataset page to launch a notebook, or browse the full catalog.

This page is also available as Markdown and as JSON.

Related Physics datasets