Numerical prediction of the aerodynamics around cars has long been one of the rudimentary needs in automotive engineering. Despite all of the recent developments in Computational Fluid Dynamics (CFD) and its constituent technologies, however, it’s still not an easy task to accurately predict the flows around and the aerodynamic forces and moments acting on ground vehicles. The complex configurations and flow physics involved in ground-vehicle aerodynamics require, among other things, sophisticated geometry modeling and meshing tools, advanced turbulence models and an efficient solution algorithm.
This paper discusses various aspects of applying modern Computational Fluid Dynamics (CFD) to the prediction of aerodynamic flows around ground vehicles. The discussion will be in the framework of an unstructured mesh finite-volume method applied to the steady-flow Reynolds-averaged Navier-Stokes (RANS) equations, which has been established as a standard CFD approach for external aerodynamics applications. The main issues in this paper include mesh, numerics, and turbulence modeling. The issue of mesh will be discussed with the primary focus on unstructured meshes that allow the use of cells of arbitrary topology, including hexahedra, tetrahedral, pyramids, prisms, and any combination of these. Special emphasis will be placed on viscous-hybrid meshes, which combine prism layers in near-wall regions (for better resolution of strong gradients in wall boundary layers) and a tetrahedral mesh in the far field. Solution adaptive mesh refinement will be discussed in view of its ability to resolve economically a wide range of length scales in the flow. We’ll discuss the impact of various turbulence models on the prediction of the aerodynamics of ground vehicles. We’ll also consider the great challenges posed by the salient features of the subject flow, including strong streamline curvature, crossflow, various types of flow separation on the body, and the ensuing shear layers and vortices. A second-moment turbulence closure, employing Reynolds-stress transport equations, is proposed as a viable turbulence model that can accurately model the salient features of the flow around ground vehicles. The issue of near-wall treatment will be addressed with an emphasis on the use of wall functions. This paper describes all of these issues in detail and it showcases simulations for a selected number of vehicle shapes and related configurations.