Multi-objective optimization research of printed circuit heat exchanger based on RSM and NSGA-II

The printed circuit heat exchanger (PCHE) is widely recognized for its high efficiency and compactness, making it a valuable component in the supercritical carbon dioxide (S-CO2) Brayton cycle. In this study, we established an experimental platform for the PCHE and developed a computational fluid dynamics (CFD) model based on experimental data. We combined machine learning and genetic algorithms (NAGA-II) to investigate the influence of different design variables (ellipticity, Reynolds number, distance between the hot side channels, distance between the hot side and cold side channels) on the friction factor (f), performance evaluation criteria (PEC), and entropy production (S) of PCHE. The results indicate that the proposed RSM model can accurately predict the f, PEC, and S, with prediction coefficients of 0.9009, 0.8591, and 0.9365. Subsequently, the NAGA-II optimization method was employed to perform multi-objective optimization of PCHE properties. The TOPSIS decision method was used to select the optimal case from the Pareto frontier. The best performance was obtained for the PCHE with an ellipticity of 0.99, a distance of 0.53 mm between the hot-side channels, a distance of 0.74 mm between the hot-side and cold-side channels, and a Re of 15230.4. Compared to the original case, the optimal case increased the PEC by 8.5 % and reduced the S by 9.98 %. This study presents a promising method for predicting and optimizing the characteristics of PCHE.