Mass transfer area in different gas‐liquid reactors as a function of liquid properties

Abstract

Using pilot-scale test plant, the interfacial area per unit volume was investigated in different gasliquid reactors, i.e. packed column, bubble column and free jet reactor. The interfacial area was studied as a function of liquid viscosity and of operating parameters. As a rule, identical test conditions were maintained in all the measurements, in order to obtain comparable results. The interfacial area-was determined by chemical means using the sulphite system (a solution of sodium sulphite in water as model liquid and air as gaseous medium). The viscosity of the solution can be increased by adding carboxymethyl cellulose without significantly affecting the reaction kinetics. The addition of a surfactant to the sulphite system allowed comparative measurements at reduced surface tension. Based on a large number of measurements, the correlations of the interfacial area are expressed as power laws. The inclusion of experiments with a jet tube reactor and a stimng vessel allows an extensive comparison of all reactors. All tests were carried out with the same material system and the same method was used to determine the interfacial area per unit volume. Therefore, a comparison with respect to mass transfer is possible.

Influence of Surface Tension on Effective Packing Area

“…Nevertheless, the number of studies that have focused on the relationship between surface tension and effective area (unconfounded by variation of other physical properties) in packed columns is quite sparse. Sedelies et al 2 employed an air-aqueous sodium sulfite system to characterize effective areas of random packing (polypropylene pall rings). To test the effect of surface tension, surfactant (Tween 20) was added to solutions, resulting in surface tension values of 35 mN/ m. Unfortunately, massive foam formation was encountered, which severely limited the feasibility of the reduced surface tension studies.…”

Correlating Interfacial Area and Volumetric Mass Transfer Coefficient in Bubble Column with the Help of Machine Learning Methods

“…But the rest of the models were able to correlate the data accurately. From Figure10, prediction by SVR for the data by Voyer and Miller;72 Sedelies et al;71 Grund et al; 67 and Wilkinson et al70 (black circles) in Figure10[a] have crossed ±20% margins. RF and EXT were able to reduce the error of the overunder predicted data.…”

“…For such a mixture less data is present in the test data set, thus resulting in lower correlative ability. In Figure10, Sedelies et al71 estimated mass transfer area as a function of different liquid properties for which SVR and ANN showed poor correlative capability giving 11 and 10 overunder predictions. For the same,58 reported data for a̲ > 800 m 2 /m 3 , ANN was not able to learn the relation of higher values a̲ with predictors and the predictions are in a straight line on training data.…”