Hickel, S., Adams, N.A. (2007) ย 
Physics of Fluids 19: 105106.ย doi: 10.1063/1.2773765

Approaches to large eddy simulation where subgrid-scale model and numerical discretization are fully merged are called implicit large eddy simulation (ILES). Recently, we have proposed a systematic framework for development, analysis, and optimization of nonlinear discretization schemes for ILES [Hickel et al., J. Comput. Phys. 213, 413(2006)]. The resulting adaptive local deconvolution method (ALDM) provides a truncation error which acts as a subgrid-scale model consistent with asymptotic turbulence theory. In the present paper ALDM is applied to incompressible, turbulent channel flow to analyze the implicit model for wall-bounded turbulence.

Computational results are presented for Reynolds numbers, based on friction velocity and channel half-width, of Re๐œย = 180, Re๐œย = 395, Re๐œย = 590, Re๐œย = 950.ย All simulations compare well with direct numerical simulation data and yield better results than the dynamic Smagorinsky model at the same resolution. The results demonstrate that the implicit model ALDM provides an accurate prediction for wall-bounded turbulence although model parameters have been calibrated for the infinite Reynolds number limit of isotropic turbulence. The near-wall accuracy can be further improved by a simple modification which is described in the paper.


Grid resolution study for LES of turbulent channel flow at Re๐œย = 590.ย Mean velocity profile and resolved Reynolds stresses for implicit LES using ALDM with van Driest damping on grid โ€” 723, โ€”โˆ™โ€” 483, - - - 243, โ€”โˆ™โˆ™โ€” 123.