Assessment of viscosity models that incorporate lag parameter scaling

Abstract

Elliptic blending lag models combine a ‘stress-strain misalignment’ parameter with conventional k−ω or k−ε based linear eddy viscosity models. The lag parameter is effective in scaling the behavior of the eddy viscosity, particularly in the near wall region. This proves to be critical for prediction of non-equilibrium flows. Flow configurations demonstrating such regimes are evaluated with the new class of lag models and their performances are compared with frequently used RANS formulations (k−ωSST,k−ω-2006). A two-dimensional bump geometry that creates different pressure gradient regimes in the flow field was chosen as the first test case as RANS models often perform poorly in adverse pressure gradient flows. A swept configuration of similar bump geometry was also chosen to assess the performance of models for flows with transverse pressure gradients. Separation inside a 3D diffuser of varying inlet aspect ratios expanding in both wall-normal and spanwise directions was studied. Such a configuration incorporates the effect of multi-directional pressure gradient on a flow and usually proves challenging for linear eddy viscosity models. Finally, the models were tested on a complex 3D test case with a variation of NACA 0020 airfoil attached to a plate. This junction flow has been widely examined both experimentally and numerically in the past with emphasis being on the prediction of flow around the junction and subsequent downstream development. The lag models were observed to be successful in reproducing the mean flow quantities when compared to high fidelity/experimental solutions for the 2D and 3D flows.