Industrial & Engineering Chemistry Research, Vol.49, No.9, 4383-4398, 2010
"Profile Method" for the Measurement of k(L)a and k(V)a in Distillation Columns. Validation of Rate-Based Distillation Models Using Concentration Profiles Measured along the Column
The "profile method" enables the determination of vapor- and liquid-side volumetric mass-transfer coefficients k(V)a and k(L)a by comparing the concentration profiles measured in both phases in a distillation column with the simulated profiles. The method is based on the fact that the shape of the concentration profiles along the column depend significantly on the distribution of the mass-transfer resistance between the liquid and vapor phases. This makes it possible to measure k(L)a and k(V)a directly in the distillation columns, in the same manner that is possible in absorption columns, and therefore to validate the distillation models more conclusively than by means of comparing the experimental and calculated height equivalent to a theoretical plate values. The coefficients were measured on structured packing Mellapak250Y. A small axial mixing of phases on this type of packing enables the use of a plug-flow model for both phases. Three distillation systems, methanol-ethanol, ethanol-propanol, and methanol-propanol, were used under total reflux and atmospheric pressure. Samples of the phases were withdrawn directly from the column packing by means of special sampling devices: The dependence of the individual transport coefficients on the phase flow rate was obtained by changing the reboiler duty. The dependencies were compared with those predicted by the following three mass-transfer correlations developed for packed columns: RBF (Rocha et al. Ind. Eng. Chem. Res. 1996, 35, 1660), BS (Billet, R.; Schultes, M. Trans. Inst. Chem. Eng. 1999, 77, 498), and DELFT (Olujic et al. Chem. Eng. Proc. 1999, 38, 683). The coefficients obtained by the profile method differ substantially from those calculated from the original models. Specifically, all of the original models considerably overvalue the k(L)a coefficients (RBF by up to 2 times, BS by up to 1.5 times, and DELFT by up to 6.3 times) and undervalue the k(V)a coefficients (all by up to 2 times). The dependence of the mass-transfer coefficients on the rate of the phases predicted by any of the original models is not consistent with that of the profile method. It has also been shown that the experimental concentration profiles strongly differ from the profiles calculated from the Aspen Plus simulation program. The correlations of the k(L)a and k(V)a coefficients with improved flow dependencies according to the profile method are presented. The concentration profiles calculated from the improved correlations fit the profiles accurately.