Search

• Explored the suitability of power-law type correlation, k L a = α U s g β , in estimating the gas-liquid mass transfer coefficient during bubble column scaling up. • The existing correlations are found to adequately describe our experimental results, suggesting such correlations are scale insensitive. • Using a dimensionless mass transfer efficiency, we show that β = 1 is appropriate and present a method to calculate a once the type of sparger (coarse or fine bubble distributor) is chosen. The productivity of a bubble column reactor (BCR) critically depends upon gas-liquid mass transfer coefficient, k L a. The prediction of k L a as function of design and operating conditions is central to BCR scale up. A large number of researchers have successfully characterized k L a experimentally in terms of superficial gas velocity, U sg using the power law relation k L a = α U sg β , with α , β as fit parameters. We probe the applicability of such correlations to the design of a scaled up BCR, which differs from laboratory BCR in two important aspects: (i) the scale of operation, which can be O (10 2 - 10 3) times larger, and (ii) the type of sparger used. To this end, experiments were performed with air and water in a pilot scale ( D C = 1.6 m diameter) BCR using a coarse bubble sparger. We found that the existing correlations do, indeed, describe k L a over a wide range of BCR sizes, suggesting that these correlations are fairly scale insensitive. However, the correlations provide no means to capture the role of sparger explicitly. We cast our experimental k L a values in terms of a mass transfer efficiency and independently recover the power law relation with β = 1. We suggest that the role of sparger design can be incorporated in the definition of α through the well-documented sparger efficiency factors. The α and β estimates thus obtained are in good agreement with the literature. [ABSTRACT FROM AUTHOR]

The present article focuses on the hydrodynamics of gas-liquid flow through a packed bubble column (PBC) reactor. Typically used in heterogeneously catalyzed gas-liquid reactions, such reactors employ a stationary packing of solid particles (catalyst) through which gas and liquid phases flow concurrently upwards. Designing the reactor to operate in a regime where the overall reaction rate is independent of mixing and mass transfer phenomena is crucial, and an important part of the design is the choice of the packing particle size, D P . It is well known that liquid-solid mass transfer and intraparticle (void) diffusion limitations drive the choice of the largest acceptable mean particle diameter. In the present work, we provide a condition for smallest acceptable mean particle size based on the consideration of gas-liquid mass transfer. Using experiments in a 10” diameter PBC apparatus and computational fluid dynamics (CFD) simulations, we demonstrate that this condition is related to a balance between bubble buoyancy and surface tension and is given by D P > K σ / β 2 ρ l g with K ≈ 1.6 and β ≈ 0.84 . [ABSTRACT FROM AUTHOR]

The present article focuses on the quantification of the depth of a depression (i.e. a vortex) formed when stirring liquids in an unbaffled stirred vessel. Based on several experiments, we show that while the vortex depth may be described very well by Nagata’s (1975) inviscid model for large Reynolds number ( Re D = ND 2 ν ≳ 10 4 ) this model does not apply for smaller Reynolds numbers. A number of researchers (e.g. Zlokarnik (1971), Rieger et al. (1979)) have addressed this difficulty by including Reynolds number in their correlations. However, those correlations produce unphysical estimates for Re D ≲ 10 3 – a situation still of considerable industrial relevance. To this end, we have developed a new correlation based on 100 experiments where viscosity, impeller size, agitation speed, and impeller submergence were independently varied. The new correlation significantly extends the range of Reynolds numbers over which it can be applied ( Re D ∈ ( 10 2 , 10 5 ) ), and has the appropriate inviscid behavior, unlike the prior models. Using validated computer simulations in OpenFOAM, and additional experiments the correlation has been validated at various scales (0.046 m to 0.92 m tank diameters). The computational results also provide insights into the effect of fluid viscosity on the overall flow structure within the tank. In particular, the flow velocity magnitude rapidly decays away from the impeller for Re D ≲ 10 3 , and the flow is no longer dominated by tangential motion of the fluid. Consequently, under viscous conditions, the surface motion and vortexing diminish rapidly as impeller submergence is increased. [ABSTRACT FROM AUTHOR]