Do turbulence models deteriorate solutions using a non-oscillatory scheme?

• Homogeneous isotropic turbulence is simulated to compare explicit and implicit LES.
• Contributions from an explicit modelling is small compared with the explicit part.
• Explicit SGS model does not deteriorate the quality for solutions from implicit LES.
• A RANS/LES turbulence model can be used with a non-oscillatory numerical scheme.

In many wind engineering applications such as wake developments in a wind farm, dynamic loads on bridges and pollutant dispersions in urban streets, a RANS/LES hybrid turbulence model is a good compromise between fidelity of solutions and computational costs. When such a model is used with non-oscillatory schemes, an extra caution is needed, especially, where the modelled contribution of turbulence is small and thus it is comparable with numerical dissipations, i.e. LES branch. To quantify effects of an explicit SGS model on solutions using a non-oscillatory scheme, decaying homogeneous isotropic turbulence is simulated with and without an SGS model (explicit and implicit LES respectively) on coarse, medium and fine meshes. All statistical results, such as a decay rate of the turbulence kinetic energy, energy spectra and probability density functions, show that the contribution of the explicit SGS model is small, and thus it does not modify the statistical solutions. For further analysis, the effective numerical viscosity νeffνeff is calculated, and it is one order of the magnitude higher than the SGS viscosity νt except with the finest mesh. At the end, some practical guidelines are drawn for the current method as a generation of inflow turbulence in wind engineering applications.