On stochastic inlet boundary condition for unsteady simulations

Abstract

The paper deals with the stochastic generation of synthesized turbulence, which may be used for a generating of an inlet boundary condition for unsteady simulations, e.g. Direct Numerical Simulation (DNS) or Large Eddy Simulation (LES). Assumptions for the generated turbulence are isotropy and homogeneity. The described method produces a stochastic turbulent velocity field using the synthesis of a finite sum of random Fourier modes. The calculation of individual Fourier modes is based on known energy spectrum of turbulent flow, and some turbulent quantities, e.g. turbulent kinetic energy and turbulent dissipation rate. A division of wave number range of the energy spectrum determines directly the number of Fourier modes, and has a direct impact on accuracy and speed of this calculation. Therefore, this work will examine the influence of the number of Fourier modes on a conservation of the first and second statistical moments of turbulent velocity components, which are prespecified. It is important to ensure a sufficient size of a computational domain, and a sufficient number of cells for meaningful comparative results. Dimensionless parameters characterizing the resolution and size of the computational domain according to a turbulent length scale will be introduced for this purpose. Subsequently, the sufficient values of this parameters will be shown for individual numbers of Fourier modes.The paper deals with the stochastic generation of synthesized turbulence, which may be used for a generating of an inlet boundary condition for unsteady simulations, e.g. Direct Numerical Simulation (DNS) or Large Eddy Simulation (LES). Assumptions for the generated turbulence are isotropy and homogeneity. The described method produces a stochastic turbulent velocity field using the synthesis of a finite sum of random Fourier modes. The calculation of individual Fourier modes is based on known energy spectrum of turbulent flow, and some turbulent quantities, e.g. turbulent kinetic energy and turbulent dissipation rate. A division of wave number range of the energy spectrum determines directly the number of Fourier modes, and has a direct impact on accuracy and speed of this calculation. Therefore, this work will examine the influence of the number of Fourier modes on a conservation of the first and second statistical moments of turbulent velocity components, which are prespecified. It is important to ensure a sufficient size of a computational domain, and a sufficient number of cells for meaningful comparative results. Dimensionless parameters characterizing the resolution and size of the computational domain according to a turbulent length scale will be introduced for this purpose. Subsequently, the sufficient values of this parameters will be shown for individual numbers of Fourier modes.