ABSTRACT

This study aims to evaluate the performance of γ transition model (versions with and without crossflow instability extension) using unstructured mesh on 6:1 prolate spheroid at 6.5 x 106 Reynolds number and 5o angle of attack. The performance of γ transition model is evaluated by an available experimental study and compared with the results of SST k-ω turbulence model, and γ-〖Re〗_θ transition model, which is the most popular transition model. The effect of transition model is shown with axial force, normal force, surface pressure and surface friction coefficients. The grid convergence index (GCI) study is performed with three different mesh levels for axial and normal force coefficients to find out the discretization uncertainty band around them. The γ transition model estimates force coefficients, approximately 58% less than fully turbulent results with higher GCI values. While transition models do not much change the surface pressure coefficients, the significant differences are seen in the surface friction coefficients. While there are not seen any dramatic changes in surface friction coefficients with fully turbulent analyses, this model captures drastic changes in the surface friction coefficients at the top surface which are sign of the transition locations. On the other hand, there is no sign of any transition phenomenon at the spheroid bottom, contrary to the observations of experimental measurements. Therefore, the γ transition model is not able to estimate the complete transition front geometry correctly. In addition, γ transition model is more sensitive to the surface mesh size with respect to γ-〖Re〗_θ transition model. The solution time is another disadvange of γ transition model. Even though the model has more simpler than the γ-〖Re〗_θ transition model, the computation time took 3.8 times more, since the iteration convergence rate of force coefficients is slower than the convergence rate of γ-〖Re〗_θ transition model solution. Eventhough including the crossflow extension into γ transition model enlarges the transition region on the top of the geometry, it does not still create transition at the bottom. In addition, crossflow extension does not change the solution time.