Study Of The Measurement Technology Of Two-phase Flow Based On Coriolis Flowmeter

Coriolis mass flowmeters are widely used for industrial flow measurements due to their high performance in the flow rate metering of single-phase fluids. However, the accuracy of the meters degrades when two-phase flows are measured. Some exploratory work was finished in the thesis in order to improve the measurement accuracy of Coriolis meter for two phase flow metering。The main content of the thesis is as follows:1. A moving resonator model is established for two-phase flow with low liquid fraction. The complex behavior of the two-phase fluid in the measuring tube of a Coriolis mass flow meter is simulated by a moving resonator model. Combined with beam model and modes analysis, the measurement errors of the Coriolis mass flow meter can be quantitatively estimated. Experimental date validation of the moving resonator model is finished and the results agree with the theoretical analysis.2. A beam model of two-phase flow based on Coriolis meter with correction term has been established. In consideration of additional hydrodynamic pressure, a beam model of two-phase flow based on Coriolis meter with correction term was established on the basic of the research of research group. Experimental date validation of the model with correction term and without correction term was finished and the result shows that the model with correction term will reduce the error.3. A large number of experimental data had been analyzed and the relationship between the CMF signals and parameters of two-phase flow had been explored. In order to analyze the relationship between the inner signal(such as oscillation frequency, tube damping, etc) and parameters of two-phase flow(liquid flow rate, differential pressure, etc),the experiment has been carried out. The relationship between the output signals of CMF and theoretical values has been discussed.4. A neural network soft measurement model was set up. Based on relevant references, six parameters are selected as inputs of the neural network and gas/liquid individual phase flow rates are taken as outputs of the network and then a Elman network is built. The results show that the prediction error is so large that doesn’t meet the need of industrial production and further work is needed.