FLUID MECHANICS AND TRANSPORT PHENOMENA Extensive Validation of Computed Laminar Flow in a Stirred Tank with Three Rushton Turbines J. M. Zalc, M. M. Alvarez, and F. J. Muzzio Dept. of Chemical and Biochemical Engineering, Rutgers University, P.O. Box 909, Piscataway, NJ 08855 B. E. Arik Dantec Dynamics, Mahwah, NJ 07430 High-resolution CFD results, supplemented by extensi®e experimental ®alidation, are presented for Newtonian laminar flow fields in a stirred tank equipped with three Rush- ton turbines. Flow fields are computed using the ORCA software suite for Reynolds numbers ranging from 20 to 200 with an unstructured tetrahedral mesh containing roughly 2 million tetrahedra. Each of the flow solutions takes less than 8 h to con® erge when running in parallel on eight desktop workstations. Excellent agreement is obtained between computed ®elocity fields and planar ®elocity ®ectors obtained using particle image ®elocimetry. Planar laser-induced fluorescence was used to expose persistent poor-mixing regions, in excellent agreement with numerical results. The computational results used illustrate strong flow compartmentalization and significant spatial hetero- geneity with respect to local deformation rates within the ®essel. Introduction Stirred tanks are widely used in the chemical processing Ž . industries for blending Godfrey, 1992 , liquid liquid disper- Ž . Ž sion Wichterle, 1995 , gas liquid dispersion Hudcova et al., . Ž . 1989 , suspension formation Saravanan et al., 1997 heat Ž . Ž . transfer Xu et al., 1997 , mass transfer Montes et al., 1999 , Ž and reactions Ali and Menzinger, 1997; Bourne and Yu, . 1994 . Efficient mixing is crucial to the outcome of all of these processes. Poor mixing can have a negative impact, such as inconsistent product quality in blended products, excessive byproduct formation in cases involving multiple reactions Ž . Zalc and Muzzio, 1999 , wide particle-size distributions in crystallization processes, and slow mass-transfer rates in bioreactors and multiphase reactors. Turbulent flows are of- ten used to achieve high mixing rates, but there are many instances in which working in the laminar regime is unavoid- able. For instance, turbulence is practically unattainable when blending viscous doughs or pastes. Turbulence is undesirable in fermentors where the high shear rates and random velocity fluctuations associated with turbulence can lead to excessive Correspondence concerning this article should be addressed to F. J. Muzzio. Ž rates of cell damage and death Cherry and Papoutsakis, 1988; . Croughan et al., 1987 . Surprisingly, few data exist regarding flow and mixing in stirred-tank reactors in the laminar regime. Early experimental work on stirred tanks yielded gross measurement of power requirements at different agitation Ž . rates Aiba, 1958; O’Connell and Mack, 1950 . Studies re- porting power requirements of stirred tanks tacitly expected that the more energy drawn from the impeller, the better the mixing, assuming that energy would be dissipated uniformly and effectively. Experimental techniques developed later for investigating flow systems locally were intrusive in nature. The insertion of probes into the flow was used to determine veloc- Ž ity components and illustrate bulk-flow patterns Desouza and . Pike, 1972; Norwood and Metzner, 1960 . Unfortunately, only a small number of such probes could be used, because the presence of the probes extensively perturbed the flow of in- terest. Mixing times have been studied based on conductivity or pH measurements using several probes inserted into the Ž flow Biggs, 1963; Hayes et al., 1998; Jahoda and Machon, . 1994 . The results obtained depend strongly on the number and location of the probes. October 2001 Vol. 47, No. 10 AIChE Journal 2144