On the Effect of an Anisotropy-resolving Subgrid-scale Model on Turbulent Vortex Motions

where ( ) denotes a filtered value. In Eq. (1), ρ, P , U i, ν and Sij denote the density, filtered static pressure, filtered velocity, kinematic viscosity and the strain-rate tensor, respectively. The SGS-stress tensor τij is originally expressed as τij = UiUj − U iU j . LES has long been recognized as a promising way to predict complex turbulence in engineering applications. Since the success of LES depends strongly on the accurate prediction of the SGS stresses, a number of research groups have proposed several kinds of SGS models for τij (see for example, Germano et al., 1991; Lilly, 1992; Zang et al., 1993; Vreman et al., 1994; Salvetti and Banarjee, 1995; Horiuti, 1997; Sarghini et al., 1999; Morinishi and Vasilyev, 2001). Although these models have provided encouraging results, there still remain several aspects to be further improved. Among them, an important concern may be in the reduction of the prediction accuracy, when they are applied to engineering applications using coarse grid resolution in the near-wall region. To overcome this difficulty, Abe (2013) recently proposed a new anisotropy-resolving SGS modeling concept, where the SGS-stress expression is constructed by combining an isotropic eddy-viscosity model (EVM) with an extra anisotropic term (EAT). This SGS model successfully improved the prediction accuracy, particularly with a coarse grid resolution in the near-wall region, while maintaining computational stability. Although the application of the model to several test cases indicated the basic capability of this SGS modeling concept (Abe, 2013; Abe, 2014), it had not been made clear how the EAT worked for improving the predictive performance. To investigate this issue, Ohtsuka and Abe (2013) compared the simulation results obtained by this anisotropic SGS model with those by a linear isotropic SGS model. They found that this anisotropy-resolving SGS model enhanced unsteady motions in the near-wall region. Based on the above background, the objective of the present study is to elucidate in more detail how the SGS models influence turbulent vortex motions. For this purpose, we perform a detailed investigation of the model performance by means of an a priori test using the direct numerical simulation (DNS) data of a plane channel flow. We make several reduced velocity fields from the DNS data with different filter widths. Applying some representative SGS models to these filtered velocity-field data, we evaluate the SGS stresses. The results obtained are compared with the true values estimated directly from the DNS data and the performance of the SGS models is then discussed.