Mixed integer optimization of combined solvent extraction and distillation processes

Abstract

Liquid solvent extraction with a distillative purification of its products is frequently used to separate close boiling or azeotropic mixtures. In this paper a methodology for the synthesis of superstructures for the mixed integer optimization of such processes is presented. A rather specific and small superstructure is generated in order to keep the rigorous process model solvable. The highly nonideal behavior of partially inmiscible mixtures makes a rigorous modeling using excess Gibbs energy (g E) models for phase equilibrium calculations necessary. For the synthesis of the superstructure only knowledge of the phase equilibrium (i.e. vapor-liquid and liquid-liquid equilibrium) is necessary. Two strategies make the combination of a mixed integer optimization and a rigorous simulation of the process possible: Firstly, a rather specific superstructure for a liquid solvent extraction process is generated. Secondly, the difficult liquid-liquid phase equilibrium calculation is separated from the process optimization task. The capabilities of the model are shown with an example process consisting of a liquid solvent extraction column, two distillation columns, a decanter and several heat exchangers. It is possible to perform simultaneously an optimization of the continious operation variables as well as in some structural variables like the overall number of equlibrium stages, the feed locations or the existance or non-existance of complete columns.