Performance Analysis of Immersed Boundary Method for Predicting External Car Aerodynamics

This paper presents calculations of external car aerodynamics by using the finite volume (FV) immersed-boundary method. The FV numerical codes primarily employ Reynolds-Averaged Navier-Stokes (RANS) models. In recent years, and due to possibility to run very large computational meshes, these models are usually used in conjunction with the advanced near-wall models. Moreover, it has been often demonstrated that the accuracy of RANS near-wall models relies on the mesh quality near the wall so by the rule, larger number of wall body-fitted cell-layers are employed. An immersed boundary (IB) method becomes an attractive alternative to the ‘standard’ FV approaches especially when applied to low quality CAD data. In general, the IB method is less investigated and validated for the car aerodynamics, particularly in conjunction with advanced near-wall turbulence models and an adaptive mesh refinement (AMR). In this ‘meshing-less’ approach, there is a background volume mesh and a surface mesh intersecting (being ‘embedded’ in) the background mesh. Adaptive mesh refinement must be applied on the cells cut with the surface to properly capture geometry details and improve accuracy of the predicted flow features. RANS turbulence model adopted in this work is the k-ζ-f model which is less sensitive to the wall-normal distance than most of RANS models, being therefore well suited for the IB approach. The results obtained by the IB method are compared to the experimental data of Hupertz et al. [1] obtained for the well-known DrivAer notchback model and to the reference calculation using the body-fitted mesh. This paper demonstrates predictive capability of the IB method.