Although CFD is considered a well-established tool in the development of car aerodynamics it is obvious that there is still a strong demand on verification and validation of schemes and models involved in the numerical codes. This is reflected by the constantly high number of publications dealing with the matter. One of the test cases most often adopted for the purpose is the generic car model proposed by Ahmed et al. [1]. This is because of its geometrical simplicity and the realistic complex flow field which is reproducibly generated in its wake and makes it a selective test case. Another reason is the availability of comprehensive experimental data sets, amongst others the one we reported in an earlier paper [2]. The velocity data covers the complete flow field around the model but focuses on the wake region. Besides mean values it provides turbulence quantities such as Reynolds stresses and other higher order statistical moments measured using laser-Doppler anemometry (LDA). After reviewing the results of numerous CFD approaches and comparing those to the experimental data it turns out that most deficiencies arise from the failing prediction of detachment and reattachment of the flow on the rear body at a slant angle of 25°. A closer analysis reveals a severe underestimation of the fluctuating quantities in the wake and most notably in the shear layers between the outer flow and the recirculation region. This fact in turn can be directly associated with the difficulties in predicting reattachment. Because of these findings further LDA measurements with high spectral resolution were performed at selected locations. One major aim of the study was to identify the time scales of the fluctuations and in this way to come to suggestions what measures are advisable to reduce the uncertainties both in CFD and experimental results.
展开▼
