Probability density and scaling exponents of the moments of longitudinal velocity difference in strong turbulence

Abstract

We consider a few cases of homogeneous and isotropic turbulence differing by the mechanisms of turbulence generation. The advective terms in the Navier-Stokes and Burgers equations are similar. It is proposed that the longitudinal structure functions ${S}_{n}(r)$ in homogeneous and isotropic three-dimensional turbulence are governed by a one-dimensional (1D) equation of motion, resembling the 1D Burgers equation, with the strongly nonlocal pressure contributions accounted for by Galilean invariance-breaking terms. The resulting equations, not involving parameters taken from experimental data, give both scaling exponents and amplitudes of the structure functions in an excellent agreement with experimental data. The derived probability density function $P(\ensuremath{\Delta}u,r)\ensuremath{\ne}P(\ensuremath{-}\ensuremath{\Delta}u,r),$ but $P(\ensuremath{\Delta}u,r)=P(\ensuremath{-}\ensuremath{\Delta}u,\ensuremath{-}r),$ in accord with the symmetry properties of the Navier-Stokes equations. With decrease of the displacement $r,$ the probability density, which cannot be represented in a scale-invariant form, shows smooth variation from the Gaussian at the large scales to close-to-exponential function, thus demonstrating onset of small-scale intermittency. It is shown that accounting for the subdominant contributions to the structure functions ${S}_{n}(r)\ensuremath{\propto}{r}^{{\ensuremath{\xi}}_{n}}$ is crucial for a derivation of the amplitudes of the moments of the velocity difference.