Parameter Variability in Viscous Convection

Abstract

For the optimal design of cooling and heating devices, the properties of the included fluids are crucial. The temperature dependence of viscosity deserves attention, as changes can be one order of magnitude or more. Here we examine the influence on convective motions by simulating a heating and cooling experiment with a vertical cylinder by finite element computational fluid dynamics (CFD) models. Such an experimental setup in which flow patterns are determined by transient viscous convection has not been simulated before. Evaluating the general behavior of the experiment in 2D, we find a dynamic phase after and before phases with moderate changes. Flow patterns in the dynamic phase change significantly with the temperature range of the experiment. We compare the outcome of the numerical models with results from laboratory experiments, finding major discrepancies concerning the flow patterns in the dynamic phase. 3D modeling shows weaker dynamics but does not show good timing with the experiment. The study depicts the importance of parameter dependencies for convective motions and demonstrates the capabilities and limitations of models to reproduce details of viscous convection.