Vortex filtering for turbulence models applied to crossflow separation

Abstract

The application of turbulence models to the flow about slender bodies at high angles of attack poses many difficulties. Regions of threedimensional crossflow separation and strong vortices outside the boundary layer usually force the modification of standard models which have been developed for attached boundary layers and wake flows. This work examines the performance of three turbulence models for predicting crossflow separation; the Baldwin-Lomax model with corrections due to Degani-Schiff, the Spalart-Allmaras model, and Menter's shear-stress transport (SST) model. These models are applied to the crossflow separation from a smooth tangentogive cylinder at 20 • angle of attack. The current work utilizes a scheme for generating gridindependent solutions for these types of flows. From this it's found that the production of turbulent eddy-viscosity in the leeward vortices must be reduced for the one-and two-equation models to provide accurate results. A vortex filter which is Galilean-invariant and computationally efficient is developed for this purpose. Computed results indicate that this filter can substantially improve the prediction of the surface pressure and crossflow separation locations for the one and two-equation turbulence models considered.

Computational challenges in high angle of attack flow prediction

“…However, in spite of these improvements the RANS-based equations still lead to questionable predictions at high to very high angles of attack, where the flow is unsteady and the time-averaged equations are no longer capable of properly modeling the flowfield. This has led to further developments in equation and turbulence modeling, including vortex filtering [25] and the DES method.…”