Multi-objective evolutionary optimization for multi-period heat exchanger network retrofit
Jan A. Stampfli, Benjamin H.Y. Ong, Donald G. Olsen, Beat Wellig, René Hofmann
Abstract
Increase in energy efficiency and reduction in greenhouse gas (GHG) emissions in industry are important steps toward a more sustainable economy. Due to the growing share of high value-added industries multi-period operation becomes more common in process industry. Therefore, retrofit of existing multi-period production plants is a key aspect towards more sustainable production processes. Hence, in this work, an existing two-level evolutionary algorithm using a genetic algorithm and a differential evolution for multi-period heat exchanger network retrofit is extended to consider GHG emissions as a second objective to the total annual cost (TAC). The multi-objective problem is addressed by incorporating the non-dominated sorting genetic algorithm (NSGA-II) and hypervolume indicators into the algorithm. By analyzing an industrial case study of a potato chips production, the results of the multi-objective optimization shows that GHG emissions can be reduced by 50%. However, compared to the single-objective optimization, TAC is increased by 27%. By selecting capital costs and operating costs as objectives instead, similar results to the single-objective optimization are achieved showing that the results are highly dependent on the selection of the objectives. Further, changes in utility costs and caused emissions have a high impact on the results.