Thermal boundary layer structure in low-Prandtl-number turbulent convection

TL;DR

This study investigates the thermal boundary layer structure in low-Prandtl-number turbulent convection using direct numerical simulations, revealing regional differences, scaling behaviors, and deviations from classical profiles across various Rayleigh numbers.

Contribution

It provides detailed analysis of local boundary layer behaviors and their scaling in low-Prandtl-number convection, highlighting regional differences and deviations from classical theory.

Findings

Boundary layer thickness varies by region and decreases with Ra.

Dynamically rescaled profiles align with classical profiles in some regions.

Boundary layers remain transitional, not fully turbulent, at high Ra.

Abstract

We study the structure of the thermal boundary layer (BL) in Rayleigh-B\'enard convection for Prandtl number ($Pr$) 0.021 by conducting direct numerical simulations in a two-dimensional square box for Rayleigh numbers ($Ra$) up to $10^9$. The large-scale circulation in the flow divides the horizontal plates into three distinct regions, and we observe that the local thermal BL thicknesses in the plume-ejection region are larger than those in the plume-impact and shear-dominated regions. Moreover, the local BL width decreases as $Ra^{-\beta(x)}$, with $\beta(x)$ depending on the position at the plate. We find that $\beta(x)$ are nearly the same in impact and shear regions and are smaller than those in the ejection region. Thus, the local BL width decreases faster in the ejection region than those in the shear and impact regions, and we estimate that the thermal BL structure would be uniform throughout the horizontal plate for $Ra \geq 8 \times 10^{12}$ in our low-$Pr$ convection. We compare the thermal BL profiles measured at various positions at the plate with the Prandtl-Blasius-Pohlhausen (PBP) profile and find deviations everywhere for all the Rayleigh numbers. However, the dynamically-rescaled profiles, as suggested by Zhou \& Xia ({\it Phys. Rev. Lett.}, vol. 104, 2010, 104301), agree well with the PBP profile in the shear and impact regions for all the Rayleigh numbers, whereas they still deviate in the ejection region. We also find that, despite the growing fluctuations with increasing $Ra$, thermal boundary layers in our low-$Pr$ convection are transitional and not yet fully turbulent.