Multisensor Fusion Based Measurement On Two-phase Flow Parameters

Two-phase flow is commonly encountered and of paramount significance in both natural and industrial processes, and yet its rheological complexity makes the accurate measurement on its process parameters difficult to achieve and hence presents an urgent issue to scientific researchers and industrial engineers. Multi-sensor fusion techniques combine data and related information from spatially and temporally distributed sensors to achieve more complete, specific and reliable inferences than could be achieved by using a single, independent sensor. The integration or fusion of data from each local sensor eliminates possible redundancy and contradictions and consequently improves accuracy and uncertainty of the measurement.In light of the parametric measuring techniques and flow mechanism of two-phase flow, the measuring models of V-cone meters and electrically sensitive sensors are established through numerical simulations; the process parameters of two-phase flow, such as flow regime, mass flowrate, and flow velocity as well as the phase concentration are jointly determined with data fusion and feature fusion methods. The main works accomplished are as follows:1. Based on the parametric measuring techniques and flow mechanism of two-phase flow, typical flow conditions of oil-water and gas-water two-phase flow were studied through CFD unstable simulations. The simulation results show that the flow condition is disturbed by the V-cone meter but will recover in a short distance; the pressure difference due to tapping position can be compensated with a constant ratio; the turbulence occurred at the cone tail has no influence upon the pressure tapping means; the segregated flow condition of gas-water two-phase flow involves fluctuations of velocity and differential pressure, which must be taken into consideration in velocity measurement; the velocity ratio between gas and water unavoidably influences the flow model development. The simulation work lays a theoretical foundation for the development of V-cone measuring correlation and cross-correlation model, and the measuring systems of electrically sensitive sensors were built on simulation basis.2. The influence of viscosity of oil-water two-phase flow upon the differential pressure model is derived in view of the existing models for two-phase flow, from which a correlation of measuring the mass flowrate of oil-water two-phase flow is presented in consideration of the viscosity character. Besides, a slip ratio based measuring correlation for gas-water two-phase flow mass flowrate measurement is presented by introducing the slip ratio with conductivity rings. The results indicate that a better accuracy is achieved by introducing viscosity character of oil phase into the measuring model, and the velocity ratio based homogeneous model is more stable in parametric measurement than the normal model, in addition, the velocity ratio based Chisholm correlation in conjunction with de Leeuw's selection on Blasius parameter evidently improves the measuring accuracy. This work provides a theory and experience integrated model for the differential pressure meters in two-phase flow measurement.3. With the analysis on flowing properties of two-phase fluids, a method of dynamically seeking suitable signal segment for cross-correlating measured signals from two-phase flow is proposed, in addition the prediction models of mixture velocity and phase concentration by using electrically sensitive sensors were established to achieve the simultaneous measurement on these two parameters. Cross-correlation results show that the cross-correlated velocity is a kind of structural velocity within two-phase flow, and can be piecewisely fitted into the mixture velocity within different range of Froude number. The prediction of phase concentration indicates that the flow condition of oil-water mixture meets the assumption of Maxwell equation and hence by using this equation achieves 8.1% average relative error in water holdup measurement of oil-water two-phase flow; the gas-water two-phase flow can be treated as the alternatively flow of full pipe and stratified flow and the average relative error is 10%.4. Based on the independent measuring model and methodology of above measuring systems, the information from each sensing plane of a Dual-plane ERT system is fused at data level and feature level respectively, moreover, the measurement data from a V-cone meter and an ERT system is also fused at feature level. And the performance of each fusion method is analyzed and discussed with the flow regime identification on gas-water two-phase flow in a horizontal pipe. At last, the flow velocity, phase concentration and slip ratio as well as the flow rate and superficial velocity of each phase are measured in acceptable accuracy by fusing the information from a V-cone meter and the electrically sensitive sensors. The results show that the data fusion with adaptive weighted estimation optimizes the measuring performance and feature fusion achieves higher accuracy by introducing additional information, and that the fusion based results have more accurate measurement than each independent sensor.