Configuration and working fluid are the two important factors that influence Organic Rankine Cycle (ORC) performance. Superstructure technology, which contains as much as possible ORC architecture features, is considered as cutting-edge artificial design technology to ORC configuration, however, it leads to intrinsic difficulty for modelling and calculation. In order to overcome the weakness and to achieve structure construction and fluid selection simultaneously, a nested three-level algorithm employing evolutionary algorithm, sequence quadratic programming (SQP) and 0-1 programming is proposed. Based on HEATSEP simplification method developed by Lazzaretto, ORC structure with different architecture features is decomposed into several elementary cycles sharing one to three of four thermodynamic processes, and this form of combination for elementary cycles is called topology combination. By using real number coding for design parameter and binary coding for topology parameter, evolutionary algorithm based on computational intelligence is adopted to generate ORC structure without artificial design. Using 0-1 programming with working fluid selection coefficient in the outer-level algorithm, simultaneous achievement of fluid optimization and structure design of ORC is realized. A case study is done after verifying accuracy of the algorithm by reference data, with a predefined set of 13 organic fluids. The results show that both intelligent construction for ORC structure and selection of optimum pure organic fluid can be achieved by the algorithm preliminarily in terms of maximum net power output.
Keywords Simultaneous optimization, intelligent construction, fluid selection, evolutionary algorithm, SQP, 0-1 programming