CFD methods for liquid mixing with the presence of a vortex. A comparative study of simulational methods with experimental validation

Abstract

Mixing liquids is a fundamental process in many industrial and laboratory appli cations, extensively studied through experimental and numerical approaches. Ob taining certain information through experiments can however be challenging and expensive, while easily obtained through simulations with Computational Fluid Dy namics (CFD). It is therefore of interest to identify cost-effective CFD methods for accurate mixing prediction. This thesis will further investigate practices when sim ulating stirred tanks in CFD by focusing on effects that appear with the presence of a vortex. Through simulations and experiments, power number, pumping number, and mixing time are determined and analysed. This is done at different operational conditions with and without a vortex. The stirred tank is simulated both as steady state and with the free surface fixed and as unsteady with the free surface being solved at operational conditions with a vortex present. Treating the flow as steady and neglecting the vortex is found to have a minor impact on power number, pumping number, and mixing time, while significantly reducing computational costs. However, the presence of a vortex affects the fluid’s velocity field, thus also affecting the transportation of the species that is to be mixed. The flow is therefore to be treated as unsteady and the free surface should be solved in order to accurately predict the transient mixing behaviour. Both the steady and transient approach correspond well with experiments over all, but at operational conditions with vortex, both methods overestimate mixing time by up to 30%. The results indicate that the power number, pumping number, and dimensionless mixing time are independent of Reynolds number under fully turbulent conditions. Changing the water level in the tank, leading to the presence of a vortex, has a minor impact on the pumping number but a significant effect on the others. Estimation of the mixing time based on the pumping number and water volume yields consistent results across varying water levels, making it especially practical due to the limited variation of pumping number.