A novel relatively linear hybrid Three-Level strategy for retrofitting heat exchanger networks.

• A relatively linear hybrid three-level strategy for retrofitting HENs is presented. • A genetic algorithm produces the structures at the first level. • The total annual cost of each structure is optimized at the second and third levels. • These levels consist of LP and ILP formulations with external search loops. • This method can reduce the TAC of retrofitted HENs by about 0.5%-20% compared to others. This paper proposes a three-level straightforward retrofitting approach for heat exchanger networks (HENs) using a genetic algorithm (GA) coupled with linear programming (LP) and integer linear programming (ILP) formulations. At the first level, the GA produces HEN structures. The local total annual cost (TAC) of these HENs is computed at the second level, consisting of LP, ILP, and an external search loop. LP model is formulated based on the minimum utility consumption, whereas ILP determines the minimum investment cost of implementing potential changes to the existing network to turn it into a GA-generated network. The external search loop is also used to handle the stream splitting ratios. Eventually, the final optimal TAC of the retrofitted network is determined by applying a correction procedure at the third level to the second level results. The results demonstrate that this approach could reduce the retrofitted network's TAC by about 0.5%-20% compared to others. [ABSTRACT FROM AUTHOR]