The present study investigates the effect of taking into account the time dependency of the fine (subgrid) scales in a residual-based variational multiscale approach for large eddy simulation. The residual-based variational multiscale method with time-dependent (dynamic) subgrid scales is presented, and the impact of the time dependency is studied for the well-known test case of turbulent channel flow. Results are presented from computations for various values of the Reynolds number, namely Ret =180; Ret=395 and Ret=590, and several time-step sizes. A generalized-a time-integration scheme is employed. Results from our numerical experiments with dynamic subgrid scales are compared to results obtained with an approximation not explicitly taking the time-dependency of the subgrid scales into account. For all Re t-values and time-step sizes under consideration, results for resolved quantities computed by both approaches are very similar. This statement applies to both, mean streamwise velocity and root-mean-square velocity fluctuations. However, it provides a model for the subgrid-scales not depend- ing on the time-step size and enabling a more robust representation of unresolved scales. Thus we conclude that the time-dependent subgrid-scale approximation is not capable of producing more accurate results for this type of flow if the time-step size is chosen within an optimal range. However, we expect it to be more advantageous for more complex problems, since our results indicate that it provides a more reliable representation of unresolved scales.     «

The present study investigates the effect of taking into account the time dependency of the fine (subgrid) scales in a residual-based variational multiscale approach for large eddy simulation. The residual-based variational multiscale method with time-dependent (dynamic) subgrid scales is presented, and the impact of the time dependency is studied for the well-known test case of turbulent channel flow. Results are presented from computations for various values of the Reynolds number, namely Ret...     »