Operation characteristics and performance prediction of a 3 kW organic Rankine cycle (ORC) with automatic control system based on machine learning methodology

Yong-Qiang Feng,Qiang Zhang,Kang-Jing Xu,Chun-Ming Wang,Zhi-Xia He,Tzu-Chen Hung
DOI: https://doi.org/10.1016/j.energy.2022.125857
IF: 9
2023-01-15
Energy
Abstract:Automatic control system enables the laboratory organic Rankine cycle (ORC) to adapt to variable operating conditions of industrial application. In this study, the operation characteristics of a 3 kW ORC with automatic control system applied to a chemical plant, as well as the performance prediction and optimization using machine learning methodology, are addressed. The dynamic behaviors for startup, operating and stop stages are discussed. The BP-ORC neural network model is established based on 3400 sets of experimental data, while the prediction accuracy is analyzed based on the errors of the training and test samples. The effects of six operation parameters on system performance are examined, while the bi-objective optimization for maximum thermal efficiency and maximum net output work is investigated. Results indicate that the component response times for startup stage and stop stage are 90s and 300s, respectively. Increasing the mass flow rate, decreasing the expander outlet temperature and increasing the expander inlet temperature ensure a higher net output work, while increasing the expander inlet temperature, decreasing the expander outlet temperature and increasing pump outlet pressure enable a higher thermal efficiency. The optimum net output work and thermal efficiency from Pareto-optimal solution are 2.87 kW and 8.855%, respectively.
energy & fuels,thermodynamics
What problem does this paper attempt to address?