Efficient algorithm for simultaneous synthesis of heat exchanger networks

Abstract

The mathematical programming based simultaneous approaches for heat exchanger network synthesis (HENS) result in large, complex, non-convex mixed-integer nonlinear programming (MINLP) models, for which finding even a feasible solution is a challenge. We propose a tailor-made search strategy that repeatedly revives the outer approximation (OA) algorithm of Viswanathan and Grossmann (1990), which in its original form is mostly ineffective for solving large HENS problems. We propose three smaller and simpler perturbations of the master problem in the OA algorithm by prioritizing, fixing, eliminating, or limiting exchangers in various ways to avoid premature termination. Our approach needs no feasible starting point, and solves much faster and better than some commercial MINLP solvers. We illustrate the application of our strategy with two recent HENS models on seven literature examples with up to 39 process streams. The algorithm solves them very efficiently and obtains solutions as good or better than those reported in the literature. Its robustness and effectiveness are exemplified by a large literature problem involving 39 process streams, where it obtains a 0.32% better solution than the best reported in the literature via a genetic algorithm. © 2013.