ANN-PSO aided selection of hydrocarbons as working fluid for low-temperature organic Rankine cycle and thermodynamic evaluation of optimal working fluid

Organic Rankine cycle (ORC) has been demonstrated to extract useful work output from low-grade heat sources like solar-thermal, biomass/biofuel combustion, geothermal, and waste heat. However, working fluid selection for ORC is a complex process and calls for careful optimization. To address this problem, the current work constitutes a design of experiments approach with a full-factorial design. A heat source temperature of 150 C is selected, and a list of 11 possible candidates of working fluid mixtures (among hydrocarbons) is taken. Work output and efficiencies from each fluid are determined based on the design of experiments, and the results are used to model an artificial neural network (ANN). Equations for work output and first law efficiency are developed using tan sigmoid function and ANN constants which act as objective functions that are maximized using multi-objective particle swarm optimization (PSO). The results of the ANN-PSO model is validated with the values from thermodynamic analysis with less than 2% error. The optimal working fluid obtained for maximum work output is R600a operating at an evaporator pressure of 1.88 MPa without any superheating. The resulting maximum work output is 7.15 kW at 8.05% thermal efficiency and an exergy efficiency of 38.13%.

Automated summary

In this study, a design of experiments approach and artificial neural network modeling were used to optimize the working fluid selection for organic Rankine cycle (ORC), resulting in the identification of the optimal working fluid and its corresponding performance.