抄録
LES and DNS of accelerating plane channel flows were conducted to investigate the validity of LES to such a flow. From the DNS results, it was observed that a rather rapid acceleration process consists of two parts: the former is the increase of the streamwise turbulent intensity, or corresponding elongation of the sublayer streaks, and the latter is the re-distribution of the turbulence to the other two directions. It was found that LES with a simple eddy viscosity model could not reproduce the latter re-distribution process at both coarser and finer grid resolutions tested here, despite the better prediction of the former process regardless of the adopted grid resolution.
