Aerodynamic performance of the “DrivAer” car model under the combined influence of rear-end and underbody geometry: A PANS computational study

The combined effects of rear-end configuration and underbody design on the aerodynamic performance of the “DrivAer” vehicle model are examined through computational modeling. The study makes use of available experimental data for the purpose of a comparative assessment. The numerical analysis employs the Partially-Averaged Navier–Stokes (PANS) approach, which constitutes a variable-resolution hybrid turbulence modeling framework bridging Reynolds-Averaged Navier–Stokes (RANS) and Large-Eddy Simulation (LES) methodologies. The investigation focuses on the fastback, notchback, and estate-back “DrivAer” variants, considering both detailed and simplified underbody geometry. The dynamics of the unresolved turbulence within the PANS framework is captured using a suitably modified three-equation, low–Reynolds-number RANS model founded on the eddy-viscosity hypothesis. Within this formulation, a transport equation is solved for the variable ζ u . This quantity, defined as ζ u = v u 2 ¯ / k u , corresponds to the normalized wall-normal component of the turbulence intensity and plays a central role in the model. Specifically, it accounts for the effect of Reynolds-stress anisotropy on the characteristic velocity scale employed in the formulation of turbulence viscosity. The PANS approach adopted in this study is implemented in the commercial CFD code AVL-FIRE following the formulation of Basara et al., which was used to perform all simulations. In addition to the PANS approach, all configurations under consideration are also simulated using both steady RANS and unsteady RANS (URANS) frameworks, while employing the same underlying turbulence model formulation that forms the basis of the PANS methodology. Unlike conventional RANS-based approaches, the PANS technique is capable of resolving a portion of the turbulence spectrum in accordance with the temporal and spatial resolution in relation to the underlying computational grid, while modeling only the remaining unresolved scales. As a result, PANS offers a distinct advantage in accurately representing flow unsteadiness associated with boundary-layer separation and subsequent vortex shedding from the vehicle surface, which leads to wake structures that vary significantly with rear-end shape and underbody configuration. This enhanced predictive capability is particularly evident in the computation of the aerodynamic properties of the “DrivAer” car model, where the PANS results closely reproduce experimentally observed trends.