Turbulence: An Entropic Approach

Abstract

We show that maximizing the generalized entropic functional $S_{q,δ}$ subject to standard kinetic energy constraints provides generalized canonical distributions that agree perfectly with measured probability densities of velocity differences at distance $r$ in highly-turbulent Taylor-Couette flow. The end point of the turbulent cascade is described by $δ=\frac{3}{2}$, a parameter value that also plays an important role in black-hole physics. At this point the Kolmogorov length scale $r=η$ is reached and all observable eddy structures of the turbulent flow disappear, in certain analogy to what is observed for black holes at the event horizon. Our approach generalizes statistical mechanics to more general nonadditive entropic functionals $S_{q,δ}$ such that it is applicable to turbulent flows. This approach asymptotically generates stretched $q$-exponentials as generalized canonical distributions relevant for turbulent flow, with a particular dependence of the stretching exponent $δ^{-1}$ on $q$ that follows from the well-known escort formalism in nonextensive statistical mechanics. Along this particular line in the parameter space, the physics can be described by $S_q$ on its own with suitable escort constraints, leading to the prediction $δ^{-1} (r) =2-q(r)$, thus allowing for a consistent thermodynamic description since $S_q$ is both trace-form and composable. We show that the above theoretically derived relation is well satisfied by measured high-precision experimental data for Taylor-Couette flow. At the Kolmogorov length scale $r=η$, the endpoint of our scenario, one has $δ=\frac{3}{2}$ and at this point the third moment of velocity differences ceases to exist and all eddies disappear. We point out various analogies with thermodynamic entropic approaches to black hole physics.