Gas-water Two Phase Flow Characterristic And Flow Partten Identification In Horizontal Narrow Rectangular Channel

The development of human society is inseparable from the effective utilization of energy.Fossil energy sources such as natural gas,oil and coal are all non-renewable,and their combustion products produce a certain degree of pollution to the environment.The development of newer,cleaner and more efficient energy sources is imminent.Nuclear energy has high efficiency and low pollution,and is a significant new type of energy.The narrow rectangular channel has been widely used in small reactor cores,highly integrated circuits compact heat exchangers,due to the higher heat exchange efficiency compact structure.Due to the influence of the scale effect,the gas-liquid two-phase flow characteristics in a narrow channel are quite different from that of a conventional channel.The traditional empirical formula for pressure drop is not suitable for narrow rectangular channels with a small aspect ratio.In this paper,air-water is used as the working medium,and the horizontally placed single-phase and two-phase flow of a narrow rectangular channel with a height of 20 mm and a width of 3 mm,and an aspect ratio of 0.15 are carried out under average temperature and normal pressure conditions.Image information and pressure fluctuation signal of the fluid are investigated in the experimental section.First is the study of the flow pattern characteristics of the two-phase fluid.The two-phase fluid flowing through the experimental section is observed via a high-speed camera to observe visually,and the pressure difference signals are collected through the signal acquisition system.It is found through observation that there are five flow patterns of two-phase flow in the experimental section: wavy flow,bubbly flow,slug flow,mixed flow and annular flow.By drawing the flow pattern diagram of the two-phase flow in the horizontal narrow rectangular channel,it can be found that bubbly flow and slug flow are the main flow patterns.According to the collected differential pressure fluctuation signals,the fluctuation curves of the differential pressure signals of various flow patterns are drawn,and their respective change characteristics are analyzed.Secondly,the resistance characteristics of single-phase and two-phase fluids are studied.For single-phase fluid,the critical Reynolds number for the transition from inner layer flow to turbulent flow in a narrow rectangular channel is about 2400,which is slightly larger than that of a circular channel.By revising Sadatomi’s calculation formula,a calculation relation formula that is more in line with this experiment is obtained.In the two-phase flow resistance characteristic experiment,the homogeneous model and the split-phase model are used to predict the pressure drop in the experimental section,and the predicted values are compared with the experimental results.It is found that the prediction result of the homogeneous model is low and thus not suitable for this experiment;the prediction result of the split-phase model is slightly higher than the experimental value.Through the influencing factors of Chisholm c coefficient proposed by Chen,the LM relational expression is revised,and a new calculation relational expression is obtained,which can better predict the frictional resistance pressure drop in the horizontal narrow rectangular channel.The predicted value of the new relational expression the average deviation from the experimental value is within 20%.Finally,the research on flow pattern recognition is carried out.The experimental pressure fluctuation signals are decomposed through the complementary set empirical mode decomposition(CEEMD),then the energy of each component is extracted as a feature parameter by means of Hilbert transform.The normalized feature parameters are put as the input parameters of the XGboost model,and the model is trained to conduct the flow pattern recognition research.The results show that CEEMD can effectively solve the problems of EMD and modal aliasing and over-decomposition,and the overall recognition rate of XGboost convection patterns reaches above 92%.