CFD-Method for 3D Aerodynamic Adjoint Simulations : For External Automotive Aerodynamics

Today’s rules and regulations regarding emissions from combustion vehicles are very strict and the travel range per tank and/or charge, especially for electric vehicles, is a crucial factor which will always be considered by the customers. Hence, automotive manufacturers strive to boost fuel and battery economy. This can, to a great extent, be done by improving the aerodynamics of the vehicle for lower drag. The conventional CFD process for aerodynamic development is relatively time consuming and there is rarely enough timeor resources to find the optimal design in all regions of the vehicle. Hence, the adjoint solver was investigated to make the aerodynamic development process more efficient by providing sensitivities of the geometry with respect to drag force. The adjoint solver was investigated both through a literature review as well as by performing CFD and adjoint simulations. The CFD and adjoint simulations were performed using Fluent 2019 R1 and the realizable k-ε turbulence model. It was found that it is important to monitor surface sensitivities during the solution in addition to the adjoint residuals to assess convergence of the adjoint simulation. It is also recommended to analyse regions of high residuals in the domain to ensure that they are far away from the surface(s) of interest. Investigations regarding different stabilization schemes as well different meshes for the adjoint solver were performed. It was concluded that the residual minimization scheme (RMS) is the preferred stabilization scheme. It was found that a coarser mesh can be used to reduce localized transient behaviour if the adjoint solver has trouble converging. It was found that a simplified model of a fully detailed car geometry is necessary to reduce the complexity and the resolution of the mesh to be able to use the RMS and to avoid local instabilities. A proposed CFD and adjoint procedure with guidelines and recommendation was developed