Rotor-stator mixers, with their ability to create high shear fields have a broad spectrum of applications in the chemical, petrochemical and pharmaceutical process industries. To better understand device performance and quantify mixing and dispersion capabilities, analyzing the flow field created by the rotor-stator interactions is crucial. We have previously reported 2-D Particle Image Velocity (PIV) measurements, for several rotor speeds and throughput conditions, for an in-line IKA prototype rotor-stator mixer containing single rows of 12 rotor teeth and 14 stator teeth. The working fluid was water in turbulent flow. In this work we report the development and validation of an accurate CFD model and make quantitative comparison between simulation results and the PIV data.
The model geometry and mesh were developed within Ansys Workbench with the fully transient (sliding mesh) 3-D RANS simulations performed with Fluent using the realizable k-epsilon turbulence model. The effect of mesh density and wall treatment were systematically studied to optimize the numerical scheme. With respect to post processing, the numerical data were sampled in a stator slot at 9 rotor tooth positions on a grid that closely mimicked that for PIV data acquisition. Detailed comparisons were made for three different rotor speeds but at the same volumetric throughput.
Our preliminary study of near-wall modelling techniques considered Non-Equilibrium Wall Functions (NEWF) and Enhanced Wall Treatment (EWT). Both produced similar results but EWT showed advantage in computational efficiency. In the mesh independence study, mesh levels of approximately 2, 6, and 16 million cells were considered. The comparison revealed that the model with 6 million cells was sufficient to insure grid independence and solution accuracy.
The results show that the CFD model predicted the location of mixing layer vortices within 12% relative to the stator slot width and accurately captured the general flow movement. However, noticeable differences were observed as well. The CFD model consistently reported higher velocity magnitude (~20% on average) and somewhat different flow direction near the slot exit, where it adjoins the volute region. Despite the differences, the results show that CFD simulations can be used to gain knowledge of flow structure and device performance. The reasons for model and data mismatch are discussed and methods to reduce the mismatch in future studies are suggested.