Synthetic turbulence inflow conditions for large-eddy simulation

Abstract

Direct numerical or large-eddy simulations of the majority of spatially inhomogeneous turbulent flows require turbulent inflow boundary conditions. A potential implication is that any results computed may be strongly influenced by the prescribed instantaneous inlet velocity profiles. Such profiles are practically never available, and a usual practice is to generate synthetic inflow data satisfying certain statistical properties, which may, for example, be known from experimental data or empirical correlations. The present paper describes a new method for generating turbulent inflow data based on digital filters that is capable of reproducing specified statistical data. Two variants of the approach are presented: a simple method in which the Reynolds stresses and a single length scale are prescribed, and a more detailed approach that is able to reproduce the complete Reynolds-stress tensor as well as any given, locally defined, spatial and temporal correlation functions. The application of the methods to a plane jet flow and to a developing wall boundary layer serve to demonstrate the applicability of the approach.

Efficient Generation of Inflow Conditions for Large Eddy Simulation of Street-Scale Flows

“…(7) even numerically (di Mare et al, 2006). However, for a Gaussian (Klein et al, 2003) or an exponential function, it is possible to obtain a very simple but approximate solution.…”

“…This is a more general approach in terms of using the correlation functions, which is based on solving an array of equations of the unknown filter coefficients and the correlation coefficients. However, as stressed in the paper (di Mare et al, 2006) the equations are incompatible and the solution is not unique. Most important, there is no guarantee that the equations has a solution for any given correlation coefficients.…”

“…The method allows spatially varying turbulence scales on non-uniform grids to be imposed at the inlet. It was developed independently of the work of di Mare et al (2006) and Veloudis et al (2007), whose method is similar in some respects, and has already been used elsewhere (Xie & Castro, 2005;Xie & Castro, 2006b). This paper describes the method in detail for the first time (in §3); it is shown to be very much more efficient than its nearest equivalents above -those of Klein et al(2003) and Veloudis et al (2007) -and also inherently more suitable for coupling to a large-scale weather code.…”

“…di Mare et al(2006) constructed a digital filter based on a relatively arbitrary correlation function, which could be simply assumed in a Gaussian shape, an exponential shape, or be obtained from experiments or precursor numerical simulations. This is a more general approach in terms of using the correlation functions, which is based on solving an array of equations of the unknown filter coefficients and the correlation coefficients.…”

Large-eddy simulation of particle-laden turbulent flows

“…However, the specification of turbulent inflow conditions cannot be avoided in the present case. For this reason, the proposed digital filter methodology of Klein, Sadiki & Janicka (2003) and di Mare et al (2006) is used to generate turbulent inflow profiles. A digital filter is applied to maps of random numbers to construct turbulence data which are spatially correlated (the method also allows the generation of temporally correlated data).…”