472373 Simultaneous Optimization of Heat Exchanger Network Synthesis and Heat-Exchanger Design Using Genetic/Simulated Annealing Algorithm

Monday, November 14, 2016
Grand Ballroom B (Hilton San Francisco Union Square)
Wu Xiao, Kaifeng Wang, Xiaobin Jiang, Xiangcun Li, Xuemei Wu and Gaohong He, Dept. of Chemical Engineering, Dalian University of Technology, Dalian, China

Heat exchanger network synthesis is a complex mixed-integer nonlinear programming (MINLP) problem, numerous assumptions for simplifying the heat exchanger network, involving the neglect of the impact of heat exchanger fouling and pumping duties, as well as the assumption of constant stream heat-transfer coefficients, can lead to a local or infeasible result, or make the result failed to be applied in the industrial condition.

A novel approach of simultaneous optimization for heat exchanger network synthesis and detailed design of heat exchangers is proposed, and Genetic/Simulated Annealing algorithm (GA/SA) is used for searching global optimal solutions. The model combines the operational conditions for each match in heat exchanger network with the main geometric constraints of each heat exchanger. Two alternative connection schemes of heat exchangers are considered, which are referred to as one-stream series-wound and one-stream parallel in the stage-wise superstructure of heat exchanger network. Thus, the influence of the pressure drop for branches on the main stream can be considered. GA/SA has been applied to minimize the total annual cost which is based on the area of heat exchangers and pumping duties, as well as the energy costs for utilities. Two loops are used in the algorithm, in the inner loop, the geometric parameters of each heat exchanger are treated as iteration variables to provide feasible heat exchangers, or infeasible heat exchangers with a penalty, the TEMA standards are considered so that the results are realistic. In the outer loop, GA/SA with crossover operation and mutation operation is used to update the heat exchanger network structures and the geometric parameters of each heat exchanger to expand the scope of the search results.

An example is used to illustrate the availability of the proposed model and algorithm. The total annual cost of the optimal network is $58540.23, which is much lower than that obtained by Mizutani (202920 $/yr) obviously, and it is also 6.45% less than that obtained by Serna (62314.46 $/yr). Inconstant overall heat-transfer coefficients are the interconnecting parameters, which combine heat exchanger network synthesis and detailed design of each heat exchanger. Our method performed better at balancing the capital cost for exchangers and pumping cost, as well as utility cost, and therefore by searching the global optimal solution it can offer options for obtaining main structure parameters of each heat exchanger with minimum total annual cost.

Extended Abstract: File Uploaded
See more of this Session: Interactive Session: Systems and Process Design
See more of this Group/Topical: Computing and Systems Technology Division