Investigating mixer-viscometer techniques for partially filled stirred tanks

Abstract

Optimising the mixing stage of formulated product manufacture would help with resource utilisation, reducing cost, and developing superior products. However, accurately measuring and characterising mixing dynamics remains challenging. This study explores the use of traditional mixer-viscometer techniques, the Couette analogy and torque curve method, to infer viscosity from torque-speed data during batch mixing processes with partial fill levels (25 %, 62.5 %, 100 %) and varying fluid rheology. The methods yield comparable results in determining the mixer constant, k’, and consistency index, K, and flow index, n, at 100 % fill level but show limitations at lower fill levels due to non-linearity in torque-speed data. Changes in the torque-speed relationship were seen to occur around the point of vortex formation for lower fill levels (Fr > 1). Thus, these data points were excluded in the determination of k’ for 25 % and K and n for 25 % and 62.5 %, but this didn’t improve estimates for apparent viscosity and shear rate for any of the fluids or fill levels. To overcome these challenges, non-linear modelling techniques or data driven models may be required for inferring viscosity from torque at partial fill levels.