This research develops a simple hybrid two-level approach to designing heat exchanger networks (HENs) comprising non-uniform exchangers, whose governing models are inherently nonlinear and non-convex. Generating HENs and identifying the best ones is handled by a Genetic Algorithm (GA) at the first level based on a modified node representation. While at the second level, the optimal values of continuous variables are determined to reach the network's minimum total annual cost (TAC) by a mathematical optimization model. This model consists of linear programming (LP) based on maximum energy recovery (MER), a linear correction procedure (COP) for modifying the heat load of exchangers and streams’ temperature, and two external search loops. Since the exchangers' area is computed implicitly in this framework, different cost functions can be considered for the exchangers in the TAC function. Investigation of a case study demonstrated that the proposed approach could produce structures with a TAC about 6% lower than the literature.
