An investigation of higher-order closures in the computation of the flow around a generic car

• Published in: Journal of mechanical science and technology 2008, 22(5), , pp.1019-1029
• Main Author: Jang, Yong-Jun
• Format: Article
• Language: English
• Subjects: Automotive bodies; Boundary layer and shear turbulence; Circulation; Computational fluid dynamics; Control; Dynamical Systems; Engineering; Exact sciences and technology; Fluid dynamics; Fundamental areas of phenomenology (including applications); Industrial and Production Engineering; Mathematical models; Mechanical Engineering; Nonlinearity; Physics; Roofs; Separation; Turbulent flows, convection, and heat transfer; Vibration; Wakes; 기계공학
• ISSN: 1738-494X; 1976-3824
• DOI: 10.1007/s12206-008-0205-3

The ability of non-linear eddy-viscosity and second-model-closure models to predict the flow around a simplified three-dimensional car body, known as the “Ahmed body,” is investigated with a steady RANS scheme. The principal challenge is to predict the separation from and reattachment onto the slanted rear roof portion at the slant angle 25°, which is close to the critical value at which separation is just provoked from the roof surface. At these conditions, it has been conjectured that separation is intermittent, with periodic flapping being a highly influential process. This is thus an exceptionally challenging case, especially for low-Re models, as the geometrical complexity occurs together with high-Re conditions (Re = 768,000) and highly complex flow features in the wake of the body. A 1.89M-node mesh containing 44-blocks was employed for one half of the spanwise symmetric body. The results demonstrate that the Reynolds-stress-transport model employed is able to reproduce, in contrast to all other models, the reattachment of the flow on the slanted rear surface. As a consequence, the strong streamwise vortices emanating from the sides of the body and associated with lift and circulation are also reproduced in good agreement with experimental data. The physical processes at play and the reasons for the predictive differences are discussed.