Boundedness of the velocity derivative flatness factor in a turbulent plane jet
S. L. Tang,R. A. Antonia,L. Djenidi,L. Danaila,Y. Zhou
2016-01-01
Abstract:This paper focuses on the statistics of normalized fourth-order moment of the longitudinal velocity derivative, ∂u/∂x, i.e. the flatness factor S4 = (∂u/∂x)4/(∂u/∂x)2 2 on the axis of a plane jet over a range of Taylor microscale Reynolds number varying between Rλ ' 500 and 1100. The aim is to determine the dependence of S4 on Rλ. Different tests on the jet axis show that local isotropy is closely satisfied, allowing the use of εiso, the locally isotropic form of the mean turbulent kinetic energy dissipation rate ε. The measurements show that S4 remains approximately constant when Rλ ≥ 500. This is inconsistent S4 ∼ Rλ , where α is a small positive number, as predicted by various internal intermittency models. The constancy of S4 is in full agreement with the relatively recent results (6) showing that S3, the skewness of ∂u/∂x, also tends to a constant when Rλ increases. The present results conform with the original similarity hypotheses of Kolmogorov (1). Introduction There is no doubt that the first two similarity hypotheses of Kolmogorov(1; 3), widely known as K41, and Kolmogorov’s (1962) refined similarity hypothesis (2), or K62 the latter was introduced to account for the so-called ”internal intermittency” -, have had a huge impact on turbulence research. Various quantities, such as energy spectra and velocity structure functions, can be used to test either K41 or K62. According to the K41 first hypothesis, these quantities adopt particular universal forms when the Taylor microscale Reynolds number, Rλ (=u′λ/ν, where λ is the longitudinal Taylor microscale u′/(∂u/∂x)′ and a prime denotes a rms value) is very large. Interestingly, the Kolmogorov-normalized onedimensional velocity spectra φu(k 1) (the asterisk denotes normalization by the Kolmogorov length scale, η = (ν3/ε)1/4, where ν is the kinematic viscosity of the fluid; ε is the mean turbulent energy dissipation rate; the overline denotes time averaging, and/or Kolmogorov velocity scale, uK = (νε)1/4) collapse in the high wavenumber region even when Rλ is as small as about 40 (4; 5). In terms of the velocity structure functions, a major outcome of K41 is the prediction (δu∗)n = fun(r), (1) where the velocity increment δu = u(x+ r)−u(x) between two points separated by a distance r along x, (hereafter x is taken in the flow direction); fun is a universal function when normalized by η and/or uK for each value of n. When r→0, expression (1) yields the normalized moments of streamwise velocity derivatives, i.e. Sn = (∂u/∂x)n (∂u/∂x)2 n/2 , (2) which, according to K41, should be constant for each value of n at large Rλ. Following (7) and (8), many studies have focused on the evolution of Sn with Rλ with the view to testing K41 and K62. The majority of the work supports the argument that |Sn| (n ≥ 3) increases continuously with Rλ, viz. |Sn| ∼ R α(n) λ (α > 0), (3) e.g. (7; 8; 9; 10; 11). However, it appears now that not only the small-scale statistics are affected by Rλ (this is the so-call Finite Reynolds number effect, or FRN effect), when the latter is not large enough (12), but the approach towards an asymptotic state as Rλ increases differs from flow to flow (13; 6). These results indicate that the Rλ dependence on Sn should be revisited. And in particular, it should be assessed separately in each flow. Such attempts have been already initiated (6; 13; 14; 15). The latter authors derived the locally isotropic form of the transport equation for ε, directly from the Navier-Stokes equations, in various turbulent flows, i.e. grid turbulence, along the axis in the selfpreserving far-field of a round jet, along the centreline of a fully developed channel flow and a far-wake of a circular cylinder. They showed that, in each flow, the transport equation for ε can be expressed in the form S3 +2 G Rλ = C Rλ , (4) where G is the non-dimensional enstrophy destruction coefficient of ε defined by G = u2 (∂2u/∂x2)2 (∂u/∂x)2 2 . (5) In Eq. (4), analytical expressions for C differ from flow to flow. For example, in grid turbulence, C is equal to 90 7(1+2R) ( n+1 n ) with R = v2/u2 and n is the power-law decay exponent for the longitudinal velocity variance, viz. u2 ∼ x−n (6; 13) whereas, along the axis in the self-preserving far-field of a round jet, C = 90 7(2+R) (6; 13). References (6; 14; 15) showed that since 2G/Rλ is found to be very nearly constant for Rλ ≥ 70∼ 100, S3 approaches a universal constant, with a value of about 0.53, when Rλ is sufficiently large, but the way this constant is approached is flow dependent. In general, Rλ only needs to exceed about 300 for S3 to become universal for all flows considered by Refs. (6; 14; 15). For Eq. (1), Pearson and Antonia (16) showed that (δu∗)2 collapses in 10 1 10 2 −0.5 0 0.5 1 1.5 2 2.5