Deterministic Global Optimization of Multistage Membrane Gas Separation Using Surrogate Models

Abstract This paper deals with deterministic global optimization of multistage membrane gas separation processes that are described by spatially distributed models. The computational tractability of the optimization problem is improved by approximating the spatially distributed models with data-driven surrogates. The resulting problems are solved globally using BARON/GAMS. The binary separation of a mixture containing CO2 and CH4 is considered as a case study for multistage membrane gas separation processes. The influence of the feed composition on the globally optimal multistage configurations is studied.