SAS-SST Model Assessment and Improvement

Abstract The SAS-SST model originally proposed by Menter et al. [1], is able to decrease the eddy viscosity level when instabilities are present. This allows the growth of instabilities and the development of coherent structures on the contrary to classical eddy viscosity models. Although this model evidences many satisfactory results, it fails to predict an unsteady solution on the backward-facing step (BFS) flow where the Kelvin-Helmoltz (KH) instability occurs. In the SAS formulation, eddy viscosity level is ruled by the ratio of the turbulent length scale over the von Karman length scale. Nevertheless, von Karman length scale is infinite at velocity profile inflexion point, therefore this ratio becomes null at that point leading to SAS term inactivity. To avoid such behavior we suggest a new formulation that bounds this ratio and protects the boundary layer where SAS activation is irrelevant. This paper depicts the new SAS-SST formulation derivation and provides a comparison between the two SAS-SST formulations as well as classical SST in URANS mode on three canonical test cases thanks to ONERA Navier-Stokes solver: elsA . Firstly, the transonic flow over the M219 cavity [2] is considered. Pressure signals were recorded (after transient phase) at ten positions on the cavity floor. Sound Pressure Level and RMS have been compared to de Henshaw’s experimental database. SST-URANS model exhibits good predictions that are improved by both SAS-SST formulations. Moreover new formulation improvements are noticeable at the end of the cavity. Then, the cylinder in a crossflow at high Reynolds number (ReD = 3,6x10^6) is considered. KH instabilities are observed as soon as separation occurs thanks to the new SAS-SST formulation which is consistent with a better instabilities detection and as a consequence a faster eddy viscosity level adaptation that favors unsteady solutions. Statistics in the cylinder wake are under progress and primary results compared to the LES database of Catalano et al. [3] are promising. The Driver and Seegmiller BFS [4] flow, that motivated this survey, is then considered. While classical SAS-SST model predicts a steady solution, two-dimensional KH rolls convected downstream are observed with the present formulation; however they do not turn three-dimensional. Second-order scheme used here is overly dissipative and may be responsible for the damping of structures therefore computations involving higher-order space discretization schemes are ongoing. Considering these favorable results, a hot jet in a crossflow is to be computed and compared to a LES database. This configuration that couples turbulence and thermal heat transfer, is not well predicted by RANS models, however improvements are expected thanks to SAS-SST. References [1] Menter, F. and Egorov, Y., “The Scale-Adaptive Simulation Method for Unsteady Turbulent Flow Predictions. Part 1: Theory and Model Description,” Flow Turbulence Combustion, Vol. 85, 2010, pp. 113–138. [2] de Henshaw, M. J. C., “M219 Cavity Case : Verification and Validation Data for Computational Unsteady Aerodynamics,” Tech. rep., QinetiQ, 2002, RTO-TR-26 AC/323(AVT)TP/19. [3] Catalano, P., Wang, M., Iaccarino, G., and Moin, P., “Numerical simulation of he flow around a circular cylinder at high Reynolds numbers,” International Journal Of Heat and Fluid Flow, Vol. 24, 2003, pp. 463–469. [4] Driver, D. M. and Seegmiller, H. L., “Features of a Reattaching Turbulent Shear Layer in Divergent Channel Flow,” AIAA Journal, Vol. 23, No. 2, 1985, pp. 163–171.