Combining Shortcut Methods and Rigorous Minlp Optimization for the Design of Distillation Processes for Homogeneous Azeotropic Mixtures

For the separation of azeotropic mixtures with distillation usually a number of conceptual design alternatives are generated. Instead of evaluating the economic viability of the different alternatives by repetitive simulation studies, a combination of shortcut calculations and rigorous MINLP optimization is proposed. In this two step process the shortcut calculations are used to screen and rank many different alternatives. The most promising alternatives are then rigorously optimized with a detailed MINLP model. This methodology is demonstrated for a single column and then extended to a distillation flowsheet with two columns and a recycle stream. In the latter problem, an intermediate NLP optimization on the flowsheet level is employed to generate favorable initial values for the recycle stream. The combination of the two calculation levels increases the problem complexity incrementally. The availability of good initial values on each level improves reliability and solution times of the optimization problems.