Large Scale Multi-Stream Heat Exchanger Network Synthesis Based On The Stream Effective Temperature Level

Multi-stream heat exchangers are widely used in process industries such as gas processing and petrochemical industries to exchange heat energy among more than two streams because of their higher efficiency, more compact structure and lower cost compare with two-stream heat exchangers. In the last few decades, much research work focused on conventional two-stream heat exchanger networks, while investigation on synthesis of multi-stream heat exchanger network (MSHEN), especially large scale MSHEN, is limited due to its complexity. In this thesis, the studies on synthesis methods of MSHEN are supported by German Research Council Foundation (DFG), and main research improvements are as follows:(1) Bring forward an effective method of generating initial near optimum feasible solution. A new calculation method for stream heat transfer temperature difference contribution value is proposed based on the thermodynamic principle of equal exergy loss distribution. Thereby, stream pseudo-temperature, that is, the stream effective temperature level, is determined, and composite curves are constructed on the T-H diagram, a reasonable feasible network structure is attained by vertical match in enthalpy intervals. In principle countercurrent heat exchange in network is achieved based on the stream effective temperature level, so probability of attainment of near optimum feasible solutions is high and these solutions are original feasible solutions and ordinal optimal solutions in evolution.(2) A new and effective mathematical model is presented for synthesizing MSHEN. A NLP model of MSHEN synthesis is proposed based on the stream effective temperature level. The decision variables of network synthesis in the model are stream heat transfer temperature difference contribution values, hence the number of dimensions is evidently reduced. Moreover, no binary variables exist in the model, the vertical match approach on the T-H diagram can deal with complex constraints and assure all solutions in evolution procedure feasible, so large scale MSHEN synthesis problem complexity is remarkably reduced by using the new NLP model. Furthermore, the model can also be used to solve heat exchanger network problems with unequal heat transfer film coefficients and different construction materials.(3) The problem of large scale MSHEN synthesis is solved. Some improvements of GA(genetic algorithm) are made, for example, design of crossover and mutation operators for synthesizing MSHEN, OCX operator for preserving good paternal information and EC operator for holding the diversities of population individuals. Moreover, the idea of parallel algorithm is introduced to advance the search capability and calculation efficiency of GA. Finally, improved GA/SA is successfully used to solve a large scale MSHEN synthesisproblem including 100*100 splitting streams, and the optimal solution can be attained.