Mohamed Elrefaie

Building foundation physics models

DrivAerNet: A Parametric Car Dataset for Data-Driven Aerodynamic Design and Graph-Based Drag Prediction


Conference paper


Mohamed Elrefaie, Angela Dai, Faez Ahmed
International Design Engineering Technical Conferences and Computers and Information in Engineering Conference, (Awarded " ASME Papers of Distinction"), 2024


View PDF
Cite

Cite

APA   Click to copy
Elrefaie, M., Dai, A., & Ahmed, F. (2024). DrivAerNet: A Parametric Car Dataset for Data-Driven Aerodynamic Design and Graph-Based Drag Prediction. In International Design Engineering Technical Conferences and Computers and Information in Engineering Conference (Vol. (Awarded " ASME Papers of Distinction"). https://doi.org/10.48550/arXiv.2403.08055


Chicago/Turabian   Click to copy
Elrefaie, Mohamed, Angela Dai, and Faez Ahmed. “DrivAerNet: A Parametric Car Dataset for Data-Driven Aerodynamic Design and Graph-Based Drag Prediction.” In International Design Engineering Technical Conferences and Computers and Information in Engineering Conference. Vol. (Awarded " ASME Papers of Distinction"), 2024.


MLA   Click to copy
Elrefaie, Mohamed, et al. “DrivAerNet: A Parametric Car Dataset for Data-Driven Aerodynamic Design and Graph-Based Drag Prediction.” International Design Engineering Technical Conferences and Computers and Information in Engineering Conference, vol. (Awarded " ASME Papers of Distinction"), 2024, doi:10.48550/arXiv.2403.08055.


BibTeX   Click to copy

@inproceedings{mohamed2024a,
  title = {DrivAerNet: A Parametric Car Dataset for Data-Driven Aerodynamic Design and Graph-Based Drag Prediction},
  year = {2024},
  journal = {International Design Engineering Technical Conferences and Computers and Information in Engineering Conference},
  volume = {(Awarded " ASME Papers of Distinction")},
  doi = {10.48550/arXiv.2403.08055},
  author = {Elrefaie, Mohamed and Dai, Angela and Ahmed, Faez},
  howpublished = {}
}

This study introduces DrivAerNet, a large-scale high-fidelity CFD dataset of 3D industry-standard car shapes, and RegDGCNN, a dynamic graph convolutional neural network model, both aimed at aerodynamic car design through machine learning. DrivAerNet, with its 4000 detailed 3D car meshes using 0.5 million surface mesh faces and comprehensive aerodynamic performance data comprising of full 3D pressure, velocity fields, and wall-shear stresses, addresses the critical need for extensive datasets to train deep learning models in engineering applications. It is 60% larger than the previously available largest public dataset of cars, and is the only open-source dataset that also models wheels and underbody. RegDGCNN leverages this large-scale dataset to provide high-precision drag estimates directly from 3D meshes, bypassing traditional limitations such as the need for 2D image rendering or Signed Distance Fields (SDF). By enabling fast drag estimation in seconds, RegDGCNN facilitates rapid aerodynamic assessments, offering a substantial leap towards integrating data-driven methods in automotive design. Together, DrivAerNet and RegDGCNN promise to accelerate the car design process and contribute to the development of more efficient vehicles. To lay the groundwork for future innovations in the field, the dataset and code used in our study are publicly accessible at: https://github.com/Mohamedelrefaie/DrivAerNet
The parametric DrivAer model is depicted with morphing boxes applied for geometry transformation in ANSA® software using a total of 50 geometric parameters and 32 morphable entities. The morphing boxes are color-coded to highlight the areas susceptible to parametric modifications, facilitating the creation of `DrivAerNet' dataset. Utilizing this morphing technique, we generated 4,000 unique car designs.
Architecture of RegDGCNN for aerodynamic drag prediction.
The model processes a 3D mesh by converting it into a point cloud representation. It takes n input points, calculates an edge feature set of size k for each point at an EdgeConv layer, and aggregates features within each set to compute EdgeConv responses for the corresponding points. The output features of the last EdgeConv layer are aggregated globally to form a 1D global descriptor, which is then used to predict the aerodynamic drag coefficient Cd, enabling direct learning from the 3D geometry of the object. The EdgeConv block ingests an input tensor of dimensions n x f, where it determines edge features for each point utilizing a multi-layer perceptron (MLP). Post-MLP application, the block outputs a tensor of dimensions n x a_n by conducting a pooling operation over the neighboring edge features.

Share



Follow this website


You need to create an Owlstown account to follow this website.


Sign up

Already an Owlstown member?

Log in