Optimization of zigzag parameters in printed circuit heat exchanger for supercritical CO2 Brayton cycle based on multi-objective genetic algorithm

The printed circuit heat exchanger (PCHE) shows promise for applications in the supercritical carbon dioxide (S-CO2) Brayton cycle due to its great resistance to high temperature and pressure. The zigzag channel is the typical channel configuration of PCHEs containing structural parameters such as pitch, bend angle, and bend radius that can significantly influence flow and heat transfer performance. In this study, the pitch, bend angle, and bend radius are selected as the design variables, while the pressure and heat exchange capacity are considered as the optimization objectives. Based on the non-dominated sorting genetic algorithm-II (NSGA-II), optimization work is conducted to improve the performance and size of the zigzag channel over a wide range of structural pa-rameters. To predict local flow and heat transfer performance, the surrogate model embedded in the one-dimensional analysis code is proposed using the artificial neural network (ANN) method and trained by 144 groups of three-dimensional simulation results. The results calculated by the one-dimensional analysis code show that both the bend angle and bend radius can significantly affect the performance of the printed circuit heat exchanger, while the bend radius plays a more important role on the pressure drop. However, the effect of the pitch on the zigzag channel is relatively small. The optimization results indicate that the appropriate core length of the zigzag channel varies with the design of the outlet temperature. For the 0.4 m core length suitable for the 306 K outlet temperature, the size is 20% less than for a 0.5 m core length with similar flow resistance, and the pressure drop is 66.3% less compared to that of the 0.3 m core length. In addition, the zigzag channel with hybrid structural parameters is proposed and compared to the conventional zigzag printed circuit heat exchanger. The optimization results indicate that the maximum difference between the two objectives is only 4%. Thus, the zigzag channel with constant structural parameters can satisfy the demand for optimization with a simpler design.