Global optimization of distillation columns using explicit and implicit surrogate models

Abstract Surrogate-based optimization of distillation columns using an iterative Kriging approach is investigated. To avoid suboptimal local minima the focus lies on deterministic global optimization. The determination of optimal setups and operating conditions for ideal and non-ideal distillation columns, leading to mixed-integer nonlinear programming (MINLP) problems, serve as case studies. To cope with output multiplicities of the model an implicit surrogate formulation is proposed.