Predictions of External Car Aerodynamics on Distributed Memory Machines

This chapter focuses on the overall solution around a car when an unsteady solution method is applied. The RANS model chosen for the simulation is the Reynolds–stress model, proved to be superior over the eddy–viscosity models for the calculation of unsteady flows. It is also evident that the Reynolds–stress model is superior for the steady state simulation of flows around a car. The Reynold–stress model (RSM) used for the present study is based on the proposal of Speziale, Sarkar and Gatski for modeling the fluctuating pressure strain correlations. The SSG model is also very robust and therefore, it was possible to apply this model successfully for the calculations of flows in complex geometries. The conclusions from these works were used to set up an unsteady simulation around a car. The experimental data used for the model assessment was provided by PSA Peugeot Citroen. The complete velocity field and pressure distribution on the car surface were measured and used, together with some “known” physical aspects of the flow around car, for the validation of the predictions. The time-dependent simulations with the k-Є model again obtain the steady state results, very similar to the calculations around, for example, circular cylinder where vortex-shedding is damped by the model's poor performances.