Deterministic global flowsheet optimization: Between equation‐oriented and sequential‐modular methods

Abstract

Flowsheet optimization problems often exhibit multiple (suboptimal) local solutions. However, deterministic global optimization can get prohibitively expensive with increasing flowsheet size and has been mostly restricted to equation‐oriented flowsheet models. In this work, we demonstrate how a reduced‐space formulation for global optimization (a hybrid between equation‐oriented and sequential‐modular methods) can be applied to flash calculations as well as a process flowsheet combining flash units, reaction, and recycling, and investigate the influence of model formulation on solution time. Both for flash units using ideal or nonideal thermodynamics as well as for the flowsheet, the reduced‐space formulation can be solved over an order of magnitude faster than a fully equation‐oriented formulation. This demonstrates the potential of the approach even for flowsheets with implicit unit operations. However, the results also highlight the importance of identifying suitable model formulations at the unit operation level as well as the flowsheet level. © 2018 American Institute of Chemical Engineers AIChE J, 65: 1022–1034, 2019