Characteristics of the Velocity Power Spectrum as a Function of Taylor Reynolds Number

An understanding of the wide range of scales present in a turbulent flow as well as the turbulence kinetic energy associated with those scales can provide significant insight into the modeling of such flows. Since turbulence is a stochastic process, statistical quantities such as mean, root mean square, correlations and spectra are used to identify and understand the evolution of turbulent flows. Time-resolved velocity measurements presented herein are obtained using hot-wire anemometry in nearly homogeneous, isotropic and moderately high Taylor Reynolds number, Rlambda , flow downstream of an active grid. Velocity power spectra presented herein are show that the slope, n, of the inertial subrange, where the inertial subrange is defined as the wavenumber range where the power spectrum scales as kappa--n, varies with R lambda as n = 1.69 -- 5.86 Rlambda--0.645. This variation in the slope of the inertial subrange is consistent with measurements presented by Mydlarski and Warhaft (1996) in an active grid flow and Saddoughi and Veeravalli (1994) in a turbulent boundary layer. The effectiveness of velocity power spectrum normalizations proposed by Kolmogorov (1963), Von Karman and Howarth (1938), and George (1992) are compared qualitatively and quantitatively. The effectiveness of these normalizations suggests how the turbulent scales make specific portions of the velocity spectrum self-similar. It is found that the relation between the large and small scales is also shown by the normalized dissipation rate, which is defined as the dissipation rate normalized by the ratio of the turbulence kinetic energy to the time scale of the large scale structure is shown to be a constant with respect to R lambda for Rlambda ≥ 450. A modified model of the one-dimensional velocity power spectrum is proposed that is based on a model proposed by Pope (2000), which has been demonstrated to model power spectra at high value of Rlambda where the slope of the inertial subrange is very close to --5/3. This modification takes into account the varying inertial subrange slope found in the data presented herein.