(132c) The Effect of Particle Shape On the Scale Up of Just Suspended Impeller Speed

Authors: 
DiMattesa, M., Rowan University
Cendo, D., Rowan University
Boesch, S., Rowan University
Etchells, A. W. III, AWE3 Enterprises
Hesketh, R. P., Rowan University



The goal of this study is to determine the minimum stirrer speed required to generate suspensions of spheres, rods, and plates.  Most of the previous work that has been done on just suspended impeller speed has been done for spherical particles and there are only a small number of research studies on the effect of particle shape on the just suspended impeller speed.  As a result of this uncertainty in the effect of particle shape on just suspended impeller speed it is estimated that industries use more energy than is required to reach this state. In this study experiments were conducted with 3 shapes of particles over a range in particle sizes and solid mass fractions.  It was found that at equivalent solid-liquid mass fractions and particle volumes, spheres were the most difficult to suspend requiring about 1.1 times higher stirrer speeds than both rods and plates.  In addition, the required speed to just suspend rods and plates of equal particle volume was very similar.  Rods with diameters from 0.8 to 2 mm and lengths from 10 to 30 mms were tested, and it was concluded that the minimum dimension of the rod has the most significant impact on predicting just suspended impeller speed. Plates with various dimensions were tested, and it was concluded that both the side length and thickness have an impact on the just suspended impeller speed. Correlations of just suspended impeller speed and a volume/surface area shape factor show a maximum for plates and sphere at the 0.33 mm value, but further tests are required to confirm this behavior for rods.  Scaled up testing was done using a geometrically similar tank that was 2.5 times larger than the small scale tank.  It was concluded from these results that the best method of predicting scale up of just suspended speed for our system was to use a constant Froude number.  This method resulted in an overall error of less than 3%.