Experimental determination of the mixing requirements for solid suspension in pharmaceutical stirred tank reactors

Glass and glass-lined, stirred-tank reactors are of significant importance in the pharmaceutical and related industries. Because of fabrication issues, a retreat blade impeller (RBI) with a low impeller clearance off the tank bottom is commonly used in glass-lined reactors, typically combined with a single baffle (providing only partial baffling conditions) mounted from the top of the reactor. In addition, these reactors are often provided with a torispherical bottom. Other configurations are also used, including full baffling or no baffles at all, hemispherical bottoms, and different impeller types. Despite their common use, some of the most important mixing characteristics of this type of reactor have not been fully studied, such as the minimum impeller agitation speed, Njs, to just suspend finely divided solids. In this work, Njs was experimentally obtained for a number of different pharmaceutically-relevant agitation systems including vessels with different types of bottoms (torispherical or hemispherical), impeller types (RBI, disk turbine, 4-blade and 6-blade pitched-blade turbine), baffling conditions (fully baffled, partially baffled, and unbaffled tanks), impeller clearance off the tank bottom, and solid particle size. The power dissipated by the impeller was also measured. A novel experimental approach to determine Njs was developed here. This approach consists of experimentally measuring the size of the circular region of the tank bottom covered by the solids at increasing values of the agitation speed, N, plotting N vs. the size of this region (expressed as either the region’s diameter Ds or its area As), and linearly regressing the data to obtain Njs as the limit of the N value for Ds or As going to zero. The Njs results obtained with the new approach were compared with those obtained using the traditional Zwietering’s approach, visually requiring that the solids do not rest on the tank bottom for more than 1-2 seconds. Excellent agreement was found between the results obtained using the novel approach and those obtained using Zwietering’s method. The novel method proposed here completely eliminates the observer’s bias from the experimental determination of Njs. The results obtained here show that Njs is a strong function of most of the variables listed above, and especially the baffling type