Effective grid scale of large-eddy simulations in various advection schemes using airflow around 1:1:2 block model

Abstract

Large-eddy simulation (LES) is widely used for predicting turbulent flows. The accuracy highly depends on the reproduction of resolved grid scale turbulences determined by grid resolution and numerical discretization. However, the actual grid scale is not specified in models except for designing the numerical grid. Therefore, this study explicitly determines the effective grid scale in LES by comparing power spectral densities (PSDs) of LES and wind-tunnel experiment (WTE) data for the airflow around a 1:1:2 building model. A frequency-domain filtering approach is used to determine the cut-off length scale, analyzing various advection schemes. Results show that numerical diffusion significantly impacts the effective grid scale, with upwind schemes leading to larger cut-off lengths due to excessive dissipation of high-frequency turbulence structures. Spatial and statistical analyses reveal that the resolution of grid scale turbulence varies across different flow regions, particularly in areas of strong flow separation. These findings highlight the importance of selecting appropriate advection schemes to reproduce the grid scale turbulence in LES specified by numerical mesh. The concept of the effective grid scale provides a refined metric for assessing LES resolution, contributing to better turbulence modeling in wind engineering.