Performance of an anisotropy-resolving subgrid-scale model for predicting turbulent channel flow with wall roughness

Large eddy simulation (LES) of a turbulent channel flow with wall roughness on one side was performed by adopting an anisotropy-resolving subgrid-scale (SGS) model at a shear Reynolds number = 395. To investigate the grid dependency of the LES results caused by the SGS model, three grid resolutions were tested under the same definition of a roughness shape by using the immersed boundary method. The obtained results were compared with the direct numerical simulation data without roughness (flat wall) and discussed using general knowledge in this kind of flow fields as reference. The primary concern was focused on how the present SGS model could reduce the grid dependency as well as the affects of roughness on the mean velocity and turbulent stresses.