Maximum Entropy in Turbulence

TL;DR

This paper applies the maximum entropy principle to turbulence, deriving energy spectra and spatial distributions that align with experimental data and DNS results, providing insights into turbulence behavior across scales.

Contribution

It introduces a maximum entropy framework for modeling turbulence energy spectra and spatial distributions, connecting physical constraints with observed data.

Findings

Energy spectra follow a log-normal distribution consistent with experiments.

Shannon's entropy and viscous dissipation increase with Reynolds number in DNS data.

The approach offers a solution to the channel flow problem using entropy-based turbulence modeling.

Abstract

Turbulence may appear as a complex process with a multitude of scales and flow patterns, but still obeys simple physical principles such as the conservation of momentum, of energy, and the maximum entropy principle. The latter states that the energy distribution will tend toward the maximum entropy under physical constraints, such as the zero energy at the boundaries and viscous dissipation. For the turbulence energy spectra, a distribution function that maximizes entropy with the physical constraints is a log-normal function, which agrees well with the experimental data over a large range of Reynolds numbers and scales. Also, for channel flows DNS data exhibits an increase in the Shannons entropy and total viscous dissipation as a function of the Reynolds number in a predictable manner. These concepts are used to determine turbulence energy spectra and the spatial distribution of turbulence kinetic energy. The latter leads to a solution to the channel flow problem, when used in conjunction with the expression for the Reynolds stress found earlier.