Turbulence and Near-Wall Modeling for Commercial Vehicle Aerodynamics: A Comparative Study of Scale Resolving Approaches

Abstract

This thesis presents a comparative study of turbulence and near-wall modeling for aerodynamic simulations of commercial vehicles. The work addresses a central challenge in industrial computational fluid dynamics, namely how to obtain reliable flow predictions without the prohibitive cost of high-fidelity scale-resolving methods. Two geometries were investigated: a simplified truck model and a detailed production-scale Volvo truck. The simplified case was used to evaluate wall-modeled LES, hybrid RANS–LES methods, Scale-Resolving Hybrid methods, and selected RANS formulations against a wall-resolved LES reference and experimental data. The production truck case was used to assess model behavior under full-scale wind-tunnel conditions. Additional simulations investigated model sensitivity to numerical choices such as time step, inner iterations, and prism-layer design. Across both configurations, the turbulence model had a strong influence on predicted separation behavior, wake structure, and drag. For the simplified truck, DDES Elliptic Blending k-ε and wall-modeled LES showed the best overall agreement with the wall-resolved LES. For the full-scale truck, DDES Elliptic Blending k-ε gave the most accurate absolute drag predictions and the best agreement for yaw-weighted drag. However, DDES k-ω SST captured the relative drag differences between geometry variations more accurately. The sensitivity studies showed that the current numerical setup is generally robust. Timestep reduction and prism–layer variation showed only limited effects in aerodynamic drag. Furthermore, it was shown that 6 to 8 inner iterations generally resulted in convergence within each timestep. Overall, the results show that no single turbulence model is optimal for all purposes, and that model selection in industrial vehicle aerodynamics should be guided by the specific engineering objective.