DNS ON MULTISCALE-GENERATED GRID TURBULENCE USING A CLASSICAL GRID

Abstract

We carry out direct numerical simulation of experimental grid turbulence, for which the mesh Reynolds number is 2500. The grid is directly constructed in the computational domain. The streamwise computational domain size is >100 times the mesh size. The value of the decay exponentn is estimated asn ≈ 1.36 by using a ratio defined by the turbulent kinetic energy and its dissipation. The prevailing perspective is that the memory of the turbulence cannot be considered as short. In this paper, we propose a promoter that focuses on the generation of turbulence using the grid, not on the shape of the grid, as addressed in previous studies (Hurst & Vassilicos (2007); Krogstad & Davidson (2011)), from a multiscale perspective and investigate the effects of the turbulence-generating method on the decay exponent n. Specifically,n is increased ton ≈ 1.53 and 1.41 because of the changes in the initial conditions. INTRODUCTION Grid turbulence is the most fundamental type of turbulent flow and has been studied extensively (Pope (2000)). Grid turbulence usually becomes highly homogeneous in the downstream region (Hinze (1975)). Its decay follows a power law, which includes a decay coefficient, the virtual origin, and the decay exponent (Mohamed & LaRue (1990); Pope (2000)) applicable in the region. The decay exponent is directly related to the fundamental characteristics of the decaying grid turbulence. If the grid turbulence is modeled by the Saffman turbulence (Hinze (1975)), the decay exponent will be close to 6 /5 (Krogstad & Davidson (2010)) when the turbulent Reynolds number is sufficiently high. Mohamed & LaRue (1990) proposed a method in which a search is performed for the fit that gives the smallest variance between the data and the form of the decay power law. Wang & George (2002) proposed an indirect means of obtaining the power law through the Taylor microscale. In this work, we employ another method to estimate the decay exponent and the virtual origin by focusing on a principal relation in grid turbulence. Wakes produced by the grid are significantly affected by the grid configuration (Mohamed & LaRue (1990); Lavoieet al. (2005)). The prevailing perspective is that the memory of the generated turbulence cannot be considered to be short (Davidson (2004)). In addition, scatter of the decay exponent among experiments may reflect the dependence of the decay exponent on initial conditions (George (1992)). In recent previous studies (mainly Hurst & Vassilicos (2007)), the turbulence generated by using a grid of a specific fractal shape has been investigated. However, in these previous studies, the concept of multiscale generation of turbulence was added to the generation of turbulence by focusing on the shape of the grid, not on the generation of turbulence by the grid. In this paper, we propose a promoter that focuses on the generation of turbulence using the grid, not on the shape of the grid, and we investigate the effects of the initial conditions on the decay exponent of grid turbulence.