ABSTRACT

The method for coupling between nested grids is proposed for LES turbulent flows. In this method fluctuating velocity simulated in a coarse grid is imposed to a fine grid and the grid-scale velocity fluctuation of the fine grid is generated based on the scale similarity model. The a-priori test of a turbulent boundary layer flow over a rough surface is conducted to validate this method. In order to fulfill spatially developing turbulent boundary layer flows with no pressure gradient we apply the quasi-periodic boundary condition to the streamwise direction. In the test coarsely resolved velocity data which is generated filtering finely resolved LES data are applied for reproducing subgrid-scale components of the coarsely resolved LES. In this a-priori test the technique for the immersed boundary method is applied to force the averaged fluctuation velocity component to zero. The reproduced fluctuation velocity agrees well with the true value which can be derive by subtracting the generated coarsely resolved velocity data from the finely resolved LES data and the kinetic energy spectra of the reproduced streamwise fluctuation velocities fit to the -5/3 power law for the inertial subrange.

The one-way coupling method is also applied for a free convective boundary layer flow with no heat flux from the ground nesting coarse grid WRF-LES and fine grid LES. In this method velocity data simulated in WRF-LES is imposed to a fine grid LES and the grid-scale velocity fluctuation of the fine grid LES is generated based on the scale similarity model. Although the total kinetic energy spectra of the fine grid LES streamwise fluctuation velocity slightly underestimates the -5/3 power law for the inertial subrange in the high wavenumber region, the connection between the coarse grid WRF-LES and the fine grid LES works well without large interpolation errors.