Validation Study of the Jaguar Land Rover Aerodynamic Computational Fluid Dynamics Process Using Metric-Based Assessment Criteria

Abstract

Credibility of simulation data has always been fundamental in aerodynamic vehicle development, as a significant amount of early design phase work is conducted virtually before a physical test property is made. As the automotive industry pivots toward artificial intelligence and machine learning techniques to assist in aerodynamic development, training these models with simulation data requires a comprehensive understanding of the accuracy and validity of the underlying simulation. It is critical these systems are trained from reliable data with a full understanding of both the limitations and predictive performance of the computational fluid dynamics (CFD) process and the wind tunnel facility it is benchmarked against. Validation and verification studies have been a long-established set of guidelines to determine if the simulation model appropriately reflects reality (validation) or if it has been set with robust numerical schemes, mesh settings, or boundary conditions (verification). The work presented here shows a comprehensive validation study with more than 400 test configurations and 18 vehicle properties. It evaluates Reynolds-averaged Navier–Stokes (RANS) and detached eddy simulation (DES) approaches using moving reference frame (MRF) and rigid body motion (RBM) to account for wheel rotation and comparing STAR-CCM+ CFD process and the FKFS Aeroacoustic Wind Tunnel (AAWT). The results demonstrate that DES—particularly when wheel rotation is modeled using RBM—provides the highest overall predictive performance, with a drag accuracy from −2% to +4% for 80% of cases with corrections applied, which gets to ±2% for over 95% cases with an additional calibration step. A metric-based assessment criterion that combines key performance metrics into a single detection event (DE) score derived from failure mode effects analysis (FMEA) principles is proposed with an example shown for the 2021 Range Rover Velar. The benefit being that it removes a more judgement-based, qualitative approach, aiding toolset selection and methods development gaps.