(324e) A Novel Systematic Approach to Pressure-Swing Distillation Process Design

Pressure-swing distillation (PSD) is a well-known and widely used process for the separation of azeotropic mixtures. It is based on the pressure dependence of the azeotropic composition. Designing a PSD process is a multi-parameter optimization problem. Even in the most simple two column configuration and on the simplest reasonable level of description by an equilibrium stage model, there is a high number of degrees of freedom which have to be chosen (column pressures, reflux ratios, recycle ratio, overall and feed stage numbers). Finding the optimal numbers for these parameters, for given feed and product specifications, is far from being trivial. Astonishingly, only little work on that widespread and important design problem has been published in the literature so far. None of the existing approaches allows a systematic optimization of all relevant parameters.

Therefore, a new method was developed in the present work that only needs the educt and product specifications as input. The corresponding algorithm couples two modeling levels. On the first level, separation factors which are determined from the knowledge of the fluid property data are used for characterizing the effort for producing desired qualities in the columns. The choice of these factors allows to determine the top and the bottom composition of the columns. For the optimization, they are systematically varied. Furthermore, a range for the column-pressures is defined considering boundary conditions like maximum bottom or minimal top temperatures. Within that range, the column pressures are discretized. For each pair of known column pressures all streams can then be calculated from simple mass-balances. Therefore, on the second level, the two columns can be designed decoupled from each other. The column design itself is based on the equilibrium stage concept using a detailed vapor-liquid equilibrium model and the well-established concept of the N,Q-curve [1] and yields all information for estimating and minimizing the OPEX and CAPEX. The entire algorithm was automated. Due to the decoupling of the column designs the new method is fast and robust. The global economic optimum can be determined swiftly and reliably. This is demonstrated using different industrial relevant examples.

[1] S. Zeck: Einfluß von thermophysikalischen Stoffdaten auf die Auslegung und den Betrieb von Destillationskolonnen, Chemie Ingenieur Technik 62 (1990) 707-717