Experimental and numerical investigations of cabin-flow in a mini-van

Abstract

Passenger comfort becomes more and more important in the automotive industries. Thus it is necessary to investigate the ventilation concepts realized in automotives e.g. inlet flows or interactions between the inlet-flow and linings. The here presented investigations aimed at the determination of the flow field between the left front air inlet and the left sidewindow of a mini-van by solving the Reynolds Averaged Navier Stokes Equation and the validation of these predictions comparison the predictions with flow field measurements. Therefore experimental and numerical investigations for a fixed flow-rate and 4 different inlet configurations were carried out at the facilities of the DLR in Göttingen. The configuration of the mini-van consists of standard interiours with carpets, instrument panels, and linings but without seats and passenger (dummies) for both numerical and experimental investigations. The calculations have been performed using the CFD codes of STAR-CD with the second-order differencing schemes MARS (monotone advection and reconstruction scheme) and CD (for density only) solving the Reynolds-averaged Navier-Stokes equations with various turbulent models and different meshing strategies. The differences of the predicted flow fields have been analysed in order to choose approach suited for such simulations. For a non-intrusive determination of the velocity components of the instantaneous flow-field the particle image velocimetry (PIV) technique in a 2C-2D setup was employed. To illuminate the tracer particles a Nd:YaG Laser has been used. Imaging of the scattered light from the tracer particles was performed using a high-resolution CCD-Kamera with a frame grabbing rate of 0.5 Hz. For any of the four inlet configurations 1000 image-pairs have been recorded. Experimental and numerical results show the interaction of the inflow-jet with the B-pillar of the respective side window for different inlet configuration. A quantitative comparison between numerical and experimental data in terms of the normalized differential vector (U_exp - U_cfd)/U_mean shows a good agreement in general, though a superior agreement if the experimental and numerical results was achieved if the k-omega LowRe turbulence model was used in the simulations.