Experimental and numerical investigations on thermal-hydraulic characteristics of supercritical CO 2 flows in printed circuit heat exchangers
Zengxiao Han,Jiangfeng Guo,Junlin Chen,Xiulan Huai
DOI: https://doi.org/10.1016/j.ijthermalsci.2023.108573
IF: 4.779
2023-08-02
International Journal of Thermal Sciences
Abstract:In the supercritical CO 2 (SCO 2 ) Brayton cycle, the pre-cooler is employed to cool SCO 2 to be near the pseudo-critical point state to minimise compression work and achieve high cycle efficiency. The dramatic variation in thermophysical properties of SCO 2 makes heat transfer and flow very complex in the pre-cooler, bringing a lot of difficulty to the design of the related components and system. This study attempts to address this issue. The experiment and simulation are carried out to analyse the influences of mass flux, inlet temperature, and pressure on the thermal-hydraulic characteristics of SCO 2 flows in a straight-channel printed circuit heat exchanger (PCHE) employed as a pre-cooler, and the underlying heat transfer mechanism is analysed as well. The local heat transfer coefficient ( h ) is higher when the phenomenon that the local temperature of SCO 2 ( T f ) is equal to the pseudo-critical temperature ( T pc ) appears in the laminar sublayer than when this phenomenon appears in the buffer layer. The new heat transfer and flow correlations are presented, and the genetic algorithm and back-prediction (GA-BP) neural network is employed to predict the local heat transfer and flow performance. The maximum relative error between the Nusselt number ( Nu ) predicted by the new heat transfer correlation and Nu obtained via experiment and simulation is within ±20%. The maximum relative error between the Nu predicted by the GA-BP neural network and the Nu obtained through experiment and simulation is within ±15%. The maximum relative error between the Fanning friction factor ( f ) predicted by the new friction factor correlation and f obtained via experiment and simulation is within ±5%. The maximum relative error between f predicted by the GA-BP neural network and f obtained through experiment and simulation is within ±4%. The present investigations analyse the thermal-hydraulic characteristics of SCO 2 in the pre-cooler prototype and could provide useful guidance for the design of pre-coolers for the SCO 2 Brayton cycle.
engineering, mechanical,thermodynamics