Adaptive filtering for the simulation of turbulent flows with lattice Boltzmann methods

This paper presents an extension of a Large Eddy Simulation based approach in the framework of Lattice Boltzmann Methods. An extended filter technique is introduced based on selective viscosity filters. In contrast to previous approaches we omit adjusting parameters, a technique frequently used in literature to achieve good fit with turbulence benchmarks. Such parameters have to be adapted manually for different Reynolds numbers and grid resolutions. The new filter approach is based on a time dependent coupling between resolved scales and a phase averaged strain-rate. We demonstrate that our method is able to recover turbulent statistics of the Taylor–Green vortex benchmark, representing Homogeneous Isotropic Turbulence for varying mesh resolutions and Reynolds numbers. Beyond that, we show the applicability to wall-bounded turbulence represented by the turbulent channel flow at friction Reynolds numbers Re_τ=180, Re_τ=395 and Re_τ=590. A spectral viscosity analysis is performed in order to demonstrate the methods ability to adapt the numerical dissipation according to the mesh resolution.