Study of flow fields in mixing tanks with particles using CFD

Abstract: The demand for food products with particles has increased in the last decades. Among them, yoghurt with fruit pieces or cereals can be found. Thus, the food processing industry is facing important challenges to offer these products to meet the quality requirements. Tetra Pak creates then this project to research and increases the knowledge of how these fluid-particle interactions behave. The main goal of this line of work is to correctly create good mixing and thus, particle distribution. By doing so, the final product can have the correct amount of cereals within the yoghurt, for example. This study develops a CFD methodology to predict the behavior of flows with particles. This methodology is applied on two production scaled tanks with different shapes and stirrer geometries. Experiments are conducted on the tanks to obtain validation data to compare with the simulations. The study investigates the ways of modeling the stirrer prescribed motion (SM and MRF), and the use of RANS and URANS models to predict the flow behavior. The particle modeling is conducted by 1-Way Coupled Lagrangian Particle Tracking (LPT) and Discrete Element Modeling (DEM). The implications of these models are assessed. The findings show that the turbulence model SST k−ω is able to model the flow behavior. The prescribed motion can be modeled using an initial solution computed by MRF and continuing the simulation using SM to capture some unsteadiness. The particle studies suggest that 1WC LPT is not fully recommendable when sedimentation exists. 1WC DEM results offer more physical results. However, none of them is fully capable to predict the results obtained in the experiments. As a side goal, the two tank stirrers are assessed. The stirrers are named in this project ViscoJet and Magnetic. The ViscoJet presents both in the experiments and in the simulations lousy results. A lack of bottom-to-top movement is predicted and therefore no good mixing and particle distribution are found. On the other side, the Magnetic stirrer creates a better bottom-to-top flow field, but the particle distribution is still poor.