Quantitative assessment and multi-objective optimization of supercritical CO2 cycles with multiple operating parameters

To clarify the importance degree and assess the effect of key parameters on comprehensive performance during operational adjustment, the artificial neural network (ANN) method is employed to quantitatively analyze the supercritical CO2 (sCO2) cycle performance for the advantage of accurate prediction and regression capabilities. The effects of seven operating parameters on the performance of the simple sCO2 cycle, recompression sCO2 cycle, and partial cooling sCO2 cycle are discussed. The test results obtained through the ANN method depict that the key parameters are turbine inlet temperature (TTIT) and pressure ratio (Pr) according to importance degree ranking. The maximum relative errors observed in the quantitative analysis of thermal efficiency and input heat are 2.48% and 3.47%, respectively. Additionally, the performance comparison of the quantitative analysis shows that the R2 values of thermal efficiency, output work, and input heat in this research are 0.057025, 0.0001381, and 0.019063 higher than those in the MATLAB software. Furthermore, the recommended operating parameters for TTIT and Pr are 500/900/500 degrees C and 2.266/3.378/2.276 in the three sCO2 cycles to achieve multi-objective optimization. The corresponding values for thermal efficiency, output work, and input heat are 44.91%/57.7%/ 44.05%, 104.8761/278.3495/109.5893 kJ/kg, and 190.3585/198.9562/198.8478 kJ/kg, respectively. This research can contribute to expanding future investigations on adjusting experimental parameters to enhance the performance of sCO2 cycles.