Research On Natural Circulation Flow And Heat Transfer Of Supercritical Carbon Dioxide

Supercritical carbon dioxide（S-CO₂）is an ideal working fluid for energy conversion systems,and it is of great practical value and research significance to realize S-CO₂ natural circulation.However,the abrupt variations of thermo-physical properties in the pseudo-critical region make the flow and heat transfer characteristics of S-CO₂ quite complicated in that region,and the perfect analysis methods and theories have not yet been developed.By using the methods of experimental research,theoretical analysis and machine learning modeling,this paper conducts comprehensive research on S-CO₂ natural circulation flow and heat transfer.Design and construction processes and operating parameter ranges of S-CO₂ natural circulation experimental facility have been introduced in the thesis,and a steady-state characteristic study of S-CO₂ natural circulation has been carried out.The results show that,depending on the interplay of buoyance force and flow resistance,there is a peak in the steady-state flow-power curve.But flow instabilities would occur when mass flow rate approaches the curve peak under relatively lower inlet temperature.The influences of different system parameters on the S-CO₂ natural circulation steadystate flow characteristics are also analyzed.The parametric analyses of S-CO₂ natural circulation flow instability have been performed through experiments,and the law of different system parameters affecting the natural circulation flow instabilities is summarized.Based on the experimental data,it is proved that the flow instability is related to the transition of S-CO₂ convective heat transfer mode.The experiments show that the system parameters affect the power thresholds of flow instabilities,and the presence of pressurizer amplifies the vibration amplitude of the flow rate.The S-CO₂ convective heat transfer is obviously deteriorated and enhanced in the pseudo-critical region.The BuAc dimensionless number,which characterizes the combined effects of buoyancy and thermal acceleration,and the corresponding threshold value have been derived in this research.The theoretic analysis implies that the heat transfer phenomenon of S-CO₂ in pseudo-critical region is highly related to the effects of average specific heat capacity,buoyancy and thermal acceleration.To develop a reliable and practical method for predicting S-CO₂ convective heat transfer,13116 experimental data of S-CO₂ convective heat transfer have been collected in this paper,and a heat transfer correlation is fitted based on experimental databank,which is assessed and compared with the typical correlations.The results indicate that the direct calculation performance of empirical correlations can be directly improved by the introduction of dimensionless numbers of buoyancy and thermal acceleration effects.However,the accuracy of all empirical correlations will be significantly reduced after iterative calculations.This thesis introduces a machine learning method to solve the prediction problem of S-CO₂ convective heat transfer.The data-driven models of S-CO₂ convective heat transfer have been established,and the predictive performance of different artificial neural network models with different input and output characteristics have been compared in detail.In view of practical engineering needs,a "two-step method" for heat transfer prediction has been proposed,which provides a practical and feasible prediction scheme of supercritical fluid convective heat transfer.