(690b) Computational Prediction of the Just-Suspended Speed, Njs, in Stirred Vessels Using the Lattice Boltzmann Method (LBM)

Industrial processes such as chemical reactions, dissolution of solids, and crystallization that involve finely dispersed solid particles suspended in a liquid in stirred vessels are widely used in the chemical and pharmaceutical process industries. Suspending all solids in the surrounding liquid is a crucial process objective in process optimization. The minimum agitation speed required to achieve off-bottom solid suspension, N_js, is a critical parameter used to ensure solid suspension in stirred vessels. Previous work has been extensively focused on the experimental determination of N_jsusing approaches that include solids sampling, visual observation, pressure gauge technique, steady cone radius measurements, and, more recently, electrical resistance tomography (ERT). Although each technique has different advantages and disadvantages such as objectivity, unobtrusiveness, visual observation, and applicability to industrial stirred vessels, it is still difficult and impractical for a wide range of industrial applications. Recently, Computational Fluid Dynamics (CFD) has been used to attempt to predict N_js, mainly using the Reynolds Average Navier Stokes (RANS) to simulate turbulence. However, solid suspension is three-dimensional and highly dependent on velocity fluctuations that are space and time dependent. It is thereby essential to introduce a more suitable CFD approach so as to predict N_js more reliably. This could additionally alleviate the need for extensive and expensive experimentation. Therefore, the objective of this study is to quantify N_js in a stirred tank provided with different axial and radial impellers using the Lattice Boltzmann Method (LBM). To do so, the mass fraction of suspended solid particles in a control volume near the bottom of the vessel was computationally predicted over a wide range of agitation speeds. Logistic regression analysis was combined with the computational approach to interpret the measurement data and hence predict N_js using a mathematical approach recently developed by our group to extract N_js from experimental ERT data. We validated the our CFD approach by comparing the predicted N_jsresults with available experimental data. The predicted N_jsresults were found in good agreement with those previously published, indicating that this CFD model combined with the logistic regression analysis proposed here are suitable for the determination of N_js. In turn, this work could help facilitate the solid suspension design and scale-up of industrial stirred vessels.