Solving Heat Exchanger Network Synthesis Problems with Tabu Search.pdfVIP

Solving Heat Exchanger Network Synthesis Problems with Tabu Search * B. Lin and D. C. Miller Department of Chemical Engineering Rose-Hulman Institute of Technology Terre Haute, IN 47803 {bao.lin | ler}@ Abstract: This paper describes the implementation of a meta-heuristic optimization approach, Tabu Search (TS), for Heat Exchanger Networks (HEN) synthesis and compares this approach to others presented in the literature. TS is a stochastic optimization approach that makes use of adaptive memory in the form of tabu lists. Both recency-based and frequency-based tabu lists are used to provide short-term and long- term knowledge of search history. TS is shown to locate the global optima with a high probability and low computation times, demonstrating the algorithm ’s potential for solving a variety of other mixed integer nonlinear programming (MINLP) problems. 1. Introduction Heat exchanger network synthesis problems (HENS) have attracted significant research because of the large saving that can be achieved in energy costs. As energy costs continue to increase, industry will have greater incentive to apply heat integration as broadly as possible in their facilities. Recent work by Furman and Sahinidis (2001) has shown that solving HENS either sequentially or with a stagewise-simultaneous formulation are NP- hard. This limits the usefulness of deterministic methods, such as generalized benders 1 decomposition (GBD) (Ciric and Floudas, 1991) and outer-approximation (OA) (Yee and Grossmann, 1990), since their computation time increases exponentially with problem size. Thus, stochastic methods, such as simulated annealing (SA) (Dolan et al., 1989; Athier et al., 19