Study Of Heat Exchanger Network Synthesis With Improved Particle Swarm Optimization

Heat exchanger network is an important part of energy recovery in chemical, oil refining etc process industry. Organizing heat transfer among streams effectively and improving heat recovery capacity have grate significance for economizing energy and cost saving. After years of research, heat exchanger network synthesis methods have made much progress, but most are confined to traditional algorithm. Intelligent stochastic optimization algorithm provides a promising way for this problem.Aimed at limitations of existing heat exchanger network synthesis method, based on nonlinear inertia weight reduction, an improved particle swarm optimization with no evolutionary iteration of particles as convergence criterion is presented. This thesis verified the feasibility and effectiveness of the proposed method through test function, and applied the method to heat exchanger network without split and with splits in order to improve intellectuality, rapidity of the process and validity of the result.For heat exchanger network without split, based on stage-wise superstructure and mathematical model proposed by predecessors, solving strategies of model with improved particle swarm optimization is put forward. Optimizing investing cost and operating cost simultaneously, design optimum heat exchanger network with best network structure and minimum utilities consumption. The feasibility and effectiveness of the proposed algorithm in heat exchanger network without split is verified through several examples.For heat exchanger network with splits, based on stage-wise superstructure, unreasonable assumption of isothermal mixing in split one of original mathematical model is improved and non-isothermal mixing synthesis model is established. With least annual cost as objective function, solving strategies of model with improved particle swarm optimization is put forward. The strategies can improve the calculation speed, avoiding local optimal. At last, the feasibility and effectiveness of the proposed model and algorithm in heat exchanger network with splits is verified through several examples.