Globally optimal distillation column design using set trimming and enumeration techniques

In this article, we present an alternative approach to the use of metaheuristic methods (GA, PSO, SA, etc.) or mathematical programming (MINLP solvers) for the optimal design of one-feed-two-products distillation col-umns. We propose the use of Set Trimming followed by candidate enumeration. For the evaluation of the per-formance of each candidate solution, the method relies on solving the associated system of equations. Three different enumeration procedures are tested: Exhaustive Enumeration, Smart Enumeration, and Segmental Smart Enumeration. Smart Enumeration is an optimization procedure that identifies the solution through a search in the set of candidates organized in ascending order of the objective function lower bound, while Segmental Smart Enumeration is introduced in this article. We compare the results of the proposed procedure with the results using an MINLP approach with different solvers. Numerical results indicate that the best alternative of the enumeration algorithms can identify the global optimum faster than a global solver of mathematical optimiza-tion. Numerical tests also showed local solvers which attained optimal solutions quickly but may be trapped in a local optimum.

Automated summary

In this article, an alternative method is proposed for the optimal design of one-feed-two-products distillation columns. The method utilizes Set Trimming followed by candidate enumeration to evaluate each candidate solution. Three different enumeration procedures are tested and compared with MINLP approach. The numerical results show that the proposed method can find the global optimum faster than global solvers of mathematical optimization.