Reference and tools / Build your own

Datasets

Public simulation datasets to benchmark on before you have your own.

• AirfRANS (Bonnet et al., NeurIPS 2022): Over 1,000 2D airfoils simulated with steady RANS in OpenFOAM: full volume and surface fields.
• DrivAerML (Ashton et al., 2024): 500 parametric variants of the DrivAer car simulated with high-fidelity hybrid RANS-LES.
• DrivAerNet++ (Elrefaie et al., MIT): 8,000 diverse car designs with CFD results, geometry and performance data.
• AhmedML (Ashton et al., 2024): 500 variants of the Ahmed body, the classic bluff-body benchmark, with hybrid RANS-LES.
• PDEBench (Takamoto et al., NeurIPS 2022): A benchmark suite of PDE datasets and baselines for scientific machine learning.
• The Well (Polymathic AI, NeurIPS 2024): A very large collection of physics simulation datasets spanning fluids, plasmas, astrophysics and biology.

Related

• Frameworks: Where to start in code: open neural frameworks, plus classical surrogate and UQ toolkits.
• Get started: From open frameworks and datasets to a surrogate trained on your own simulations.
• Automotive: Drag, lift, cooling, thermal and aeroacoustic design of vehicles, evaluated in seconds instead of overnight CFD.