Efficient Optimization-Based Design of Distillation Columns for Homogenous Azeotropic Mixtures

Design of distillation columns using mathematical programming has been a research topic for over a decade. However, a lack of robustness and high sensitivity toward the initial solution have prevented the widespread use. To overcome the numerical problems of the optimization algorithms, a combination of shortcut calculations with the rectification body method (RBM) and rigorous optimization is proposed. All potential splits can be generated and ranked based on the minimum energy demand calculated in the shortcut step. The least attractive split alternatives can be discarded to reduce the complexity of the rigorous model, while the results of the shortcut calculations for the promising alternatives can be used further for the initialization and bounding of the rigorous optimization. The rigorous optimization problems are then optimized as successive relaxed mixed-integer nonlinear programming (SR-MINLP) problems:? instead of a combinatorial expensive MINLP problem, a succession of NLP problems with decreasing bounds is solved. An integer solution is achieved in all case studies considered. The robustness of the algorithm is improved, and the calculation times are reduced compared to existing methods. The design methodology is reliable and yields good results. It is illustrated with several case studies involving azeotropic mixtures.