Assessment of Subgrid-Scale Models by Direct Numerical Simulation of a Temporally Developing Turbulent Mixing Layer.

Assessment of subgrid-scale (SGS) models in large eddy simulation was conducted using direct numerical simulation data base of a temporally developing turbulent mixing layer. The direct numerical simulation data base carefully calculated by means of a spectral method was used for comparison of SGS models. In this paper, various SGS models for incompressible turbulence were tested to clarify the characteristics of the models. The Smagorinsky model, the Bardina model and the GNS model predict the mean value of Reynolds stress very well, whereas the local structure of the Reynolds stress cannot be well represented by these models. Among the three models, the Bardina model predicts the local structure best. As for the Cross terms, the Bardina model shows high correlation with DNS data. The dynamic eddy viscosity model for total SGS stress predicts very small stress which is close to the Reynolds stress.