Predict Flow Field and Mixing Time for High-Shear Mixers Using CFD

Abstract: Batch high-shear mixers are used in a variety of different processing industries,such as the pharmaceutical, chemical and food industries. One company that utilize this type of mixers is Tetra Pak. High-shear mixers can be used for many different mixing scenarios, but only liquid-liquid mixing is discussed in the thesis. The batch high-shear mixer is constructed with a rotor-stator at the bottom of the tank. The rotor stator structure is a rotor which rotates inside the stationary stator. The fluid effected by the rotor is pushed through small holes on the stator and out into the tank region. The flow field generated inside the vessel is both complex and highly turbulent. With the technology of today it is highly interesting to use Computational Fluid Dynamics, CFD, simulations to determine these performances, from both an economically and time-dependant perspective. One property that is interesting is the time the mixing tank needs to achievea homogene mixture, i.e. the mixing time. However, simulations regarding rotor-stator mixers are in general very computational demanding. It is therefore of interest to improve the methodology to decrease the simulation time while still giving realistic results. This thesis evaluates Tetra Paks current methodology and also suggests possible improvements. The methodology can be divided into two parts, predicting the flow field inside the tank and predicting the mixing time.To decrease the simulation time, Tetra Pak uses a time-scale separationmethod in the current methodology. The separation utilizes that the flow around the rotor-stator converges to a stable state quicker than the rest of the vessel.The stabilized flow is then approximated by boundary conditions. Different boundary conditions have been tested. Of all methods, a transient boundary condition which imitates the motions of the rotor-stator was deemed generate the most realistic flow field. The second best option was a time-averaged boundary condition, simulated with a LES turbulence model. To determine the mixing time has also different methods been tested. The named "homogeneity"-method was the one that gave results most close resembling real life experiments. However, this method did this with both the transient and static time-averaged boundary conditions, indicating that the need of a transient boundary condition might not be as big as believed.However, all simulations and validations were only performed on one tank geometry and also for cases with relative low tip speeds. This implies that thereis still a need to validate the methodology with more experimental cases before a final conclusion of how well the new methodologies for CFD simulations can be drawn.