Is turbulent mixing a self convolution process ?

TL;DR

This paper investigates whether turbulent scalar mixing can be modeled as a self-convolution process by analyzing experimental PDF evolution, revealing non-universality and dependence on flow geometry.

Contribution

It provides experimental evidence and discusses the applicability of self-convolution models for turbulent scalar mixing, highlighting their limitations and conditions.

Findings

PDF evolution from skewed to Gaussian is non-universal

Mixing route depends on injector-to-channel cross section ratio

Self-convolution models have specific advantages and limitations

Abstract

Experimental results for the evolution of the probability distribution function (PDF) of a scalar mixed by a turbulence flow in a channel are presented. The sequence of PDF from an initial skewed distribution to a sharp Gaussian is found to be non universal. The route toward homogeneization depends on the ratio between the cross sections of the dye injector and the channel. In link with this observation, advantages, shortcomings and applicability of models for the PDF evolution based on a self-convolution mechanisms are discussed.