Recent advances in Artificial Intelligence (AI) has led to various explorations on their utilization as surrogate models for rapid inference-based design iterations in the field on automotive aerodynamics. The surrogate models promise orders-of-magnitude speed up over Navier-Stokes solvers – an attractive prospect for vehicle design and performance optimization. However, given the large design parameter space, the only feasible solution for training such models is to use synthetic data generated using Navier-Stokes solvers. This short video demonstrates the use of ScaLES to generate space-time resolved volume and surface datasets with a variety of outputs including turbulence characteristics such as Reynolds stresses, generated for a vast sequence of perturbed geometries.
