Hydrodynamics and mass transfer in bubble column: Influence of liquid phase surface tension

According to literature, few experiments are performed in organic solvents which are mostly used in commercial gas-liquid reactors. However, it is commonly accepted that data obtained in aqueous solution allow to predict the surface tension effects, and to model the behaviour of organic solvents. In this work, we examine the validity of this approximation. In this objective, the flows observed in two pure media having similar viscosity but different surface tension-respectively, water (reference) and cyclohexane (solvent)-are successively compared at two scales: in a bubble column and in bubble plumes. In bubble plumes, as expected, the mean bubble size is smaller in the medium having the smallest surface tension (cyclohexane), but for this medium the destabilisation of flow is observed to occur at smaller gas velocity, due to break-up and coalescence phenomena. In bubble column, these phenomena induce the bubbling transition regime at lower gas velocity, whatever the operating conditions for liquid phase: batch or continuous. Consequently, when the two media are. used at similar gas superficial velocity, but in different hydrodynamic regimes, greater gas hold-up and smaller bubble diameter can be observed in water; the interfacial area is then not always higher in cyclohexane. This result differs from the behaviour observed in the literature for aqueous solutions. The analysis of bubble plumes in aqueous solutions of butanol shows that this difference is due to a fundamental difference in coalescent behaviour between pure solvents and aqueous mixtures: the surface tension effect is less important in pure liquid than in aqueous solutions, because of the specific behaviour of surfactants. It is then still difficult to predict a priori the bubbling regime or the flow characteristics for a given medium, and all the more to choose an appropriate liquid as a model for industrial solvents. (C) 2007 Elsevier Ltd. All rights reserved.