Abstract:
Carbon dioxide (CO2) has been recognised as a low-cost, high efficiency and non-toxic natural refrigerant suitable as a substitution for CFCs and HCFCs in the low-temperature circuits of cascade or volatile secondary refrigeration systems. Due to the unique thermophysical properties of CO2 when working in the high reduced pressure conditions of these systems, existing correlations fail to predict flow condensation heat transfer rates accurately, and new models need to be developed for the design and optimisation of CO2 condensers. An experimental study was conducted to measure the heat transfer rate of CO2 flow condensation for mass fluxes ranging from 100 to 500kg/m2-s, at saturation temperatures of -10, -5 and 0˚C under a wide range of vapour qualities inside a horizontal 4.73mm inner diameter tube. An open-loop test rig was used which featured a high stability of the coolant temperature. Experimental results showed that the effects of mass flux and vapour quality on the rate of heat transfer were more significant in annular flows than in stratified flows, and that high mass fluxes and vapour qualities resulted in high heat transfer rates. Experimental results also showed that decreasing the saturation temperature significantly increased the heat transfer rate in annular flows compared to flow condensation in stratified flows. A new model for CO2 flow condensation inside horizontal tubes was proposed which included criteria for flow regime transition and empirical correlations to predict the heat transfer rate in each regime. The transition was predicted by Soliman’s Froude number which was fitted to the published experimental observations of transition between CO2 two-phase flow regimes. Two semi-analytical correlations were fitted to the experimental data, one for annular and the other for stratified flows. The proposed model was evaluated against a CO2 flow condensation databank, which included the heat transfer measurement points from references in the open literature and the present experimental study. Comparison results showed the current model successfully predicted the data points with an average absolute deviation of 7%.