(58d) Prediction of Main Transition Velocities and Gas Holdups in a Bubble Column Operated with Aqueous Alcoholic Solutions

Bubble columns (BCs) are the most important gas-liquid contactors. They are frequently used in the chemical industry (for absorption, oxidation, chlorination, etc.) because these reactors are capable of providing excellent mass and heat transfer characteristics. Depending on their specific application, BCs could operate either in the homogeneous flow regime (FR) (mainly used in biotechnological applications) or heterogeneous FR (used in 80 % of the industrial applications). In between, the transition FR exists, which is characterized with single coalesced bubbles but without the presence of liquid circulation. That is why two transition velocities U_trans, distinguishing the boundaries of the three main FRs, are usually identified. The first one is associated with the onset of bubble coalescence, whereas the second one distinguishes the onset of liquid macro-circulation.

The operation of BCs with foaming systems has not been exhaustively investigated. In this work, a new reliable FR identification approach as well as reasonable prediction of gas holdups in the homogeneous and transitions FRs will be demonstrated for conditions with low foaming.

Gauge pressure fluctuations have been recorded in a BC (0.1 m in ID) equipped with a perforated plate gas distributor (96 orifices × ∅ 1.0 mm). The BC operated with mixtures of deionized water (DW) and 2-pentanol with different volume concentrations (0.5, 1.0, 1.5 and 2.0 vol. %). The signal consisted of 10,000 points, which were subdivided into 10 intervals (each consisting of 1000 points). The local average absolute deviation (AAD_i) in each interval was estimated together with the mean AAD (AAD_mean) from all ten divisions. Based on this approach, the average relative error (ARE) in each interval was calculated. It was found that the dimensionless ratio of maximum ARE to total ARE (sum of all ten ARE) is capable of identifying reliably the main transition velocities. In other words, a modified error analysis was performed. For instance, in case of DW + 1.0 vol. % 2-pentanol three transition velocities U_trans (at 0.044, 0.055 and 0.077 m/s) were identified. The third U_trans value distinguished the onset of the complete foaming state. This way, a successful FR identification can be performed by means of only one statistical parameter (AAD is a robust statistical parameter about the data width around the mean of the signal) and error analysis. It is noteworthy that the first U_trans value practically coincides with the prediction by the empirical correlation of Im et al. (2019). In a similar way, the three U_trans values were successfully identified at the other volume concentrations of 2-pentanol in DW (0.5, 1.5 and 2.0 vol. %).

The empirical fitting (average relative error < 5 %) of 40 experimental overall gas holdups to the superficial gas velocity in the homogeneous and transitional FRs of the BC (0.1 m in ID) operated with the same gas-liquid mixtures was compared with the predictions of the empirical correlation of Sotello et al. (1994). The result shows that for foaming systems the gas holdups can be predicted by previously developed correlations in non-foaming systems, however, a small modification should be applied.

References

Im, H., J. Park and J. W. Lee, ACS Omega 4, 1329-1343 (2019)

Sotello, J. L., F. J. Benitez, J. Beltran-Heredia, C. Rodriguez, Int. Chem. Eng. 34, 82-91 (1994)