Turbulent transport by diffusive stratified shear flows: from local to global models. III. A closure model

TL;DR

This paper introduces a second-order closure model for turbulent transport in stratified, thermally diffusive shear flows relevant to stellar radiative zones, improving predictions over local models and calibrated against numerical simulations.

Contribution

It develops a new closure model for shear-induced turbulence in low Péclet number regimes, addressing limitations of previous local models near turbulent region edges.

Findings

Model accurately predicts vertical flow and stress profiles in DNS.

Predictions are within a factor of two in worst cases.

Model is simple enough for implementation in stellar evolution codes.

Abstract

Being able to account for the missing mixing in stellar radiative zones is a key step toward a better understanding of stellar evolution. Zahn (1974) argued that thermally diffusive shear-induced turbulence might be responsible for some of this mixing. In Part I and Part II of this series of papers we showed that Zahn's (1974, 1992) mixing model applies when the properties of the turbulence are local. But we also discovered limitations of the model when this locality condition fails, in particular near the edge of a turbulent region. In this paper, we propose a second-order closure model for the transport of momentum and chemical species by shear-induced turbulence in strongly stratified, thermally diffusive environments (the so-called low P\'eclet number limit), which builds upon the work of Garaud \& Ogilvie (2005). Comparison against direct numerical simulations (DNSs) shows that the model is able to predict the vertical profiles of the mean flow and of the stress tensor (including the momentum transport) in diffusive shear flows, often with a reasonably good precision, and at least within a factor of order unity in the worst case scenario. The model is sufficiently simple to be implemented in stellar evolution codes, and all the model constants have been calibrated against DNSs. While significant limitations to its use remain (e.g. it can only be used in the low P\'eclet number, slowly rotating limit), we argue that it is more reliable than most of the astrophysical prescriptions that are used in stellar evolution models today.