Convective Turbulence in Liquid Gallium and Sodium

PI Janet Scheel, Occidental College
The field displays the complex dynamics of the velocity field
Project Description

Many turbulent flows found in nature and exploited for heat transfer in technology are driven by sustained temperature differences. Applications range from chip-cooling devices to the study of convection in the sun and Earth’s atmosphere, core, and oceans. Turbulent Rayleigh-Bénard convection (RBC) is a type of heat transfer in which a fluid cell is kept at a constant temperature difference between top and bottom. Because the fluid is confined, thin boundary layers of the temperature and velocity fields form. To better understand whether heat transport is dominated by these boundary layers and their dynamics, researchers are using leadership-class computing to predict the transition to turbulence in the boundary layers of RBC.

A boundary layer transition to turbulence is expected at larger Rayleigh numbers, which characterize the driving of convective turbulence and is proportional to the temperature difference. However, this transition is also affected by a fluid’s Prandtl number, which corresponds to the properties of that particular fluid. Most high-Rayleigh-number experiments have explored fluids with high or moderate Prandtl numbers, such as gases like helium or air. However, fluids with high Rayleigh numbers but very low Prandtl numbers are of interest for their applications in both nature and technology.

On Mira, researchers are using a highly efficient, parallel spectral element code, Nek5000, to simulate turbulent RBC in a regime with low Prandtl numbers and high Rayleigh numbers, relevant to liquid gallium and sodium. With resolutions on Mira of approximately 10 billion grid points and 100,000 time steps for each RBC simulation, researchers are characterizing the global structure of convective flow and the boundary layer dynamics for this new regime. Furthermore, this study is determining the exact relationship between heat transport and Rayleigh number for low Prandtl numbers. This information has not been obtained before and would be fundamentally useful for both theory and experiments.