Optimization of multi-refinery hydrogen networks

Abstract

Hydrogen is an important feedstock and fuel in the refining and petrochemical industries. In this paper, we address the optimal synthesis of inter-plant hydrogen networks by modifying the superstructure-based one-plant model of Elkamel et al. (2011). The bilinear terms involved in the component balance equations of this model give rise to a nonconvex mixed integer nonlinear program (MINLP). A specialized outer-approximation algorithm is developed for solving this optimal synthesis problem to global optimality, where the lower bound on the global optimum is obtained by means of piecewise under-and-over estimators for the bilinear terms. We propose and compare several integration schemes to demonstrate significant cost savings. © 2012 Elsevier B.V.