CAE ML Datasets
Large-scale Computational Fluid Dynamics (CFD) and Finite-Element Structural Analysis (FEA) datasets to aid the research and development of data-driven and physics-driven Machine Learning methods
Announcements
| Oct 31, 2024 | “WindsorML has been accepted at NeurIPS 2024! The preprint will be updated to reflect the latest version submitted to the conference” |
|---|---|
| Aug 23, 2024 | “DrivAerML: High-Fidelity Computational Fluid Dynamics Dataset for Road-Car External Aerodynamics” paper is now available on arxiv |
| Aug 02, 2024 | “WindsorML: High-Fidelity Computational Fluid Dynamics Dataset for Automotive Aerodynamics” paper is now available on arxiv |
| Aug 02, 2024 | “AhmedML: High-Fidelity Computational Fluid Dynamics Dataset for Incompressible, Low-Speed Bluff Body Aerodynamics” paper is now available on arxiv |
Aims & Scope
There is a shortage of datasets containing 3D realistic geometries simulated using high-fidelity CFD and FEA approaches. To address this we have created a community-driven project to develop large-scale CFD and FEA based training datasets using high-fidelity simulation data to aid the research and development of data-driven and physics-driven Machine Learning approaches. Three datasets are currently available (AhmedML, WindsorML and DrivAerML), focused on high-fidelity CFD within the automotive domain, but in future this will be expanded to other domains.
This website aims to help users to find the latest datasets, including how to access and download them as well as follow their use in publications.
Contact
The datasets are the result of a community effort from academia and industry. You can find the full list of contributions on the individual dataset pages. For general information please contact Neil Ashton who can route your question to the relevant person.