← Papers

PhysicsNeMo-CFD benchmark

Engineering metrics for benchmarking car-aero surrogates

Paper title A Benchmarking Framework for AI models in Automotive Aerodynamics

Tangsali et al. (NVIDIA) · Aero & CFD · Public: code

Paper arXiv 2507.10747PDF

An open evaluation layer in NVIDIA PhysicsNeMo-CFD that scores any model's predicted surface and volume fields with the checks aerodynamicists use: integrated drag and lift, design-trend ranking (Spearman), centreline pressure, field errors, residuals of the governing equations, and prediction on a point cloud instead of the mesh. It also fixes a DrivAerML train/validation split.

Share on LinkedIn, X or Bluesky, or or .

Know related work that belongs here?

Something wrong on this page?

More papers

All papers
  1. PhysicsNeMo-Mesh — GPU-Native Meshes and Simulation Cases for Physics AI
  2. Multi-fidelity scaling — Compute-optimal data scaling via multi-fidelity training
  3. CAE-UQ — Predictive Uncertainty for Neural CAE Surrogates