A novel two-step synthesis method with weakening strategy for solving large-scale heat exchanger networks

Efficiently solving large-scale heat exchanger network (HEN) synthesis problems remains a challenging task mainly due to the presence of discontinuity caused by integer variables as well as nonlinearity of the objective cost function. This paper focuses on efficiently solving large-scale HEN synthesis problems based on a chessboard presentation with no splits, and aims to achieve globally near-optimal solutions by using a two-step synthesis method along with a weakening strategy for fixed investment costs that is conducive to bypass the discontinuity. The first step is to generate the most promising network structure by using an improved cuckoo search algorithm to optimize integer and continuous variables simultaneously. In the second step, a cuckoo search algorithm is utilized to directly optimize the heat exchanger area of the optimal structure obtained in the first step. Meanwhile, fixed-cost weakening strategies are applied in the above two steps for improving the optimization efficiency. The proposed method has been tested on several HEN cases, and two of which, namely the 20- and 39-stream problems, are presented in this paper. For these two cases, although the solutions obtained do not contain stream splits, their total annual costs are close to best solutions with stream splits and lower than those best solutions with no-splits reported in the literature, and their chessboard network configurations are clear and simple. Based on these results, the developed method demonstrates the applicability to handle large-scale HEN problems, and is also useful for guiding the grassroots or retrofit design of large-scale HENs or other engineering process syntheses with less cost and less pollution.