Thermally coupled distillation columns represent a very interesting option for the intensification of distillation systems in order to reduce the energy consumption, and, as a consequence, the environmental impact of the separation process. Several thermally coupled distillation schemes can be generated for the separation of multicomponent mixtures. This fact is an advantage, since a wide portfolio of alternatives can be used to separate a specific mixture; however, this is also a disadvantage since a lot of alternatives must be explored in order to find the optimal one. The optimal configuration, for a given mixture, depends on the nature of the mixture, usually quantified for ternary mixtures through the ease of separation index (ESI), and also on the feed composition. As can be noticed, the size of the design and optimization problem increases when these variables are considered in the generation of the solutions space. For the separation of ternary mixtures, Tedder and Rudd (1978) presented a composition map for which thermally coupled systems allowed energy savings. However, the scenario is different for quaternary mixtures, since no similar information is available. Therefore, in this work, energy consumption data for five feed compositions for a mixture near to ideality are presented. The quaternary sequences studied are: conventional direct (three columns), conventional indirect (three columns), thermally coupled direct (main column and two side rectifiers), and thermally coupled indirect (main column and two side strippers). The design and optimization of the distillation sequences is performed through a multiobjective genetic algorithm with constraints handling, coupled to the commercial process simulator Aspen Plus, and enhanced through the use of neural networks.
|Konference||26th European Symposium on Computer Aided Process Engineering|
|Periode||12/06/2016 → 15/06/2016|
|Navn||Computer Aided Chemical Engineering|