Vehicle External-Flow Surrogate & Wind-Tunnel Digitalization
Background
The drag coefficient (Cd) is a key parameter for high-speed energy consumption; each 0.01 Cd reduction saves about 0.5 L/100km. Wind-tunnel tests are costly and a single LES run takes hours, so surrogates have become popular among automotive aero engineers.
Approach
Data sources
- DrivAerML (RWTH Aachen, 31 TB, multiple DrivAer variants)
- HiLiftAeroML (NASA / SAI, 66.9 TB, real-vehicle LES)
Surrogate
- Graph Neural Network (GNN): unstructured surface meshes
- FNO: field-data mapping
- CNN + Attention: volumetric field data
Key results
- GNN surrogate: Cd error < 3% on DrivAer
- FNO: 40% lower MSE than POD on pressure reconstruction
Use cases
- Rapid concept-stage evaluation (replacing some wind-tunnel tests)
- Aero-kit (spoiler, underbody) parameter optimization
- High-speed stability analysis (side force, lift prediction)
Resources
- DrivAerML (apply for access; free for academic use)
- HiLiftAeroML (usage agreement with NASA/SAI)
Tags: AutomotiveDragLESSurrogateExternal Flow
Related entries
- Physics-Informed CNN for Battery Pack Temperature Fields · A physics-informed convolutional neural network whose loss encodes the finite-di
- Airfoil Aerodynamic Design & CFD Surrogate · Airfoil aerodynamic simulation and surrogate study: large parametric CFD dataset
- Wind-Turbine Blade Load Prediction & Surrogate · Data-driven prediction of blade fatigue load and power curve, solving hard-to-ev
- HydroGym: A Reinforcement Learning Platform for Flow Control · An open-source reinforcement learning platform for flow control that couples six
- MACE-MP-0: A Universal Machine-Learned Interatomic Potential · A general-purpose machine-learned interatomic potential built on an equivariant