Experiment investigation and machine learning prediction of a biomass-fired organic Rankine cycle combined heating and power system under various heat source temperatures and mass flow rates

The operation characteristics of a small biomass direct-fired organic Rankine cycle combined heating and power system are investigated experimentally. The thermal balance test and start-stop characteristics is examined, while the effects of key operation parameters on the system performance are addressed. A back-propagation artificial neural network model is developed to predict the system behavior, while the uncertainty analysis using Monte Carlo algorithm is analyzed. A further tri-objective optimization for maximizing expander shaft work, maximizing ORC thermal efficiency, and maximizing system energy efficiency simultaneously is discussed. Results indicate that the ORC thermal efficiency reaches its peak at 12.1 % when the heat source temperature is 121 °C and the heat source mass flow rate is between 0.26 and 0.3 kg/s. Both the heating coefficient and system energy efficiency increase with the cooling water mass flow rate. The system exergy destruction increases from 8.45 kW to 10.43 kW as the heat source temperature rises from 121 to 127 °C. Evaporator 2 contributes the most to exergy destruction, with a value of 6.595 kW, accounting for 69.7 % of the system exergy destruction. The optimal solutions derived from the bi-objective optimization of the ORC loop thermal efficiency and the energy efficiency are 11.635 % and 89.467 %, respectively. In contrast, the optimal solutions from the tri-objective optimization, which includes energy efficiency, ORC loop thermal efficiency, and expander shaft power, are 89.075 %, 11.729 %, and 1.801 kW, respectively.