(324b) Design of Efficient Systems for Multicomponent Distillation

In the traditional industrial applications of multicomponent distillation, a relatively limited number of distillation column arrangements are repeatedly applied to a wide variety of separations.  However, there are many separation process solutions differing significantly in cost and energy consumption that carry out the same overall process. Therefore a systematic method to identify and design optimal multicomponent separation sequences is needed that does not rely on the inventive activity of a few experienced engineers. Even for a commonly perceived “mature” technology such as distillation, a method to elucidate all possible multicomponent separation schemes has only been developed recently.

This talk will focus on two aspects of multicomponent continuous distillation.  First, a method to draw multicomponent distillation configurations will be introduced. It is known that the number of possible distillation configurations increases rapidly in hundreds and thousands as the number of components in an n-component mixture increase beyond three (n > 3).  Therefore, the first challenge for a process engineer is to be able to draw all feasible configurations and then narrow down the search to a set of suitable candidates.  A systematic procedure to draw distillation column configurations to separate an ideal to near ideal n-component mixture into product streams each enriched in one of the components will be presented.  The method is simple and easy to use.

The second aspect of a system for multicomponent distillation design is a screening tool that allows rank listing of distillation configurations generated by the above procedure. While it is possible to compare systems based on rigorous modeling and simulation, it becomes impractical for the large number of possible configurations generated when n becomes large. Instead, the configuration’s vapor duty as calculated by the Underwood equations is minimized. The total vapor flow in a column is closely related to both the reboiler heat requirement (a principal component of operating cost for distillation) and column width (a major component of capital cost). Thus, a nonlinear program written to minimize total vapor duty is able to produce a ranked list of energy-efficient distillation columns that is very similar to a ranked list generated by rigorous simulation, and takes much less computation. This program can be extended to design of configurations with less than n-1 columns; in addition, a study will be presented on whether the Underwood equations can be used as an accurate predictor of energy efficiency in real systems.