A comparison of methods for introducing synthetic turbulence

Abstract

Scale resolving simulations are indispensable to provide in-depth knowledge of turbulence in order to improve turbulence modelling approaches for turbomachinery design processes. However, the in flow conditions of spatially evolving turbulent flow simulations are of utmost significance for the accurate reproduction of physics especially for Large Eddy Simulations. The present paper compares two approaches to introduce synthetically generated velocity fluctuations for Large Eddy Simulations: an in flow boundary condition and a source term formulation. In case of the boundary condition the velocity fluctuations are added to the mean velocity components at the inlet panel of the computational domain. The source term formulation uses an additional volume term in the momentum equation to add the fluctuating velocity field at an arbitrary location. The functionality of these methods in combination with the synthetic generation of fluctuations is validated in the generic test case of spatially decaying homogeneous isotropic turbulence. Furthermore, the spatial variation and anisotropy of turbulent statistics in a turbulent boundary layer as well as the development length of the different combinations, needed to reach a fully developed flow, are analysed for a turbulent channel flow.