Distillation column optimization: A formal method using stage-to stage computations and distributed streams

This work addresses the complexities of optimizing the number of stages in a distillation column, which typically lead to challenging non-linear mixed-integer optimization problems. To simplify this, we employ distributed streams, thereby eliminating discrete degrees of freedom. To avoid sophisticated initialization procedures, the optimization problem is reformulated by employing a sequence of stage-to-stage calculations, each reduced to maintaining only the MESH (mass, equilibrium, summation, heat) equations for a single stage.
Our numerical experiments show the efficiency and stability of solving the simplified optimization problem in various scenarios, including single and multiple distillation columns. For a single column scenario, we compare the accuracy of our optimization method with a full enumeration approach. Additionally, for a pressure swing flowsheet designed to separate an azeotropic mixture, we illustrate potential energy savings by optimizing a stage distribution versus using a predetermined stage distribution.