It is well known that the boiling temperature of a heteroazeotropic mixture is always lower than that of its constituents. That effect has been used for a long time in carrier steam or heteroazeotropic distillation. Three phase distillation (cf. Figure) uses the same physical principle as these processes, but differs from them in the way the distillation is carried out: it is a continuous distillation in which two liquid and one vapor phase are present over the entire height of the column. An entrainer (W) is chosen so that it forms heteroazeotropes with the organic compounds to be separated, C1 and C2. This ensures the desired low temperature in the entire process, which is always lower than the boiling temperature of the lowest boiling pure component C1, C2, or W. Decanters are used both at the top and at the bottom of the distillation column for recovery and recycling of the entrainer. The separated products are withdrawn as the second phase from the bottom and top decanter, respectively.
Continuous three phase distillation was studied in a laboratory scale distillation column (diameter: 50 mm) equipped with 2.9 m of Sulzer CY structured packing arranged in 6 sections. Because of the comparably small hold up and thus shorter exposition time, packing is preferred over trays for the considered process. Two different test systems were investigated: Ethyl acetate - 1-butanol - water and 2-methyl-4-pentanone - 1-butyl acetate - water. In both cases, water is the entrainer, whereas the two organic components are to be separated. While these substances are by no means thermally instable, they form good test systems that allowed studying the basic features of the investigated process because they show the desired phase behaviour. The two test systems differ mainly in the boiling point difference between the two heteroazeotropes and the width of the miscibility gap. In the distillation experiments, the influence of all important process parameters like feed composition, heat duty and entrainer hold up was studied for both test systems. It is shown, that efficient separation of the organic components is possible. The resulting concentration profiles are presented and discussed in ternary diagrams including the underlying three-phase equilibria.
The distillation experiments are accompanied by separate investigations of fluid dynamics of the three phase flow in the column. It is shown that the process can successfully be modeled using an equilibrium stage approach that accounts for the presence of one or two liquid phases. Convergence of the simulation model is not trivial. The homotopy continuation method was used for solving the model equations. The simulation results are compared with the experimental data and interpreted on the background of the knowledge on fluid dynamics.
The experimental and theoretical studies presented show that the new process is feasible and represents a promising alternative or addition to state of the art processes like vacuum distillation.