Efficient simultaneous synthesis for heat exchanger network with simulated annealing algorithm

The simultaneous synthesis of heat exchanger network is primarily regarded as mixed integer nonlinear programming models and its combinatorial nature can easily lead to a suboptimal network design and unmanageable computational effort. Based on simulated annealing algorithm, this paper presents an efficient simultaneous synthesis method that provides the optimal networks in a two-level procedure. Two probability models are proposed to generate candidate structures by random perturbations in the upper level. The minimum total annualized cost of each candidate structure is solved in the lower level and then sent to the upper level where different structures are evaluated by simulated annealing mechanism. This two-level method is applied to solve four benchmark cases of heat exchanger network synthesis and the computational results show that satisfactory heat exchanger network designs can be guaranteed with reasonable cooling schedules and two most economical networks with no splits are solved for case 1 and case 3.