Differential Pressure/ultrasonic Fusion Method For Identification Of Oil-gas-water Three-phase Flow

Gas-liquid two-phase flow is widely encountered in industrial processes,and accurate online identification of three-phase flow patterns is an important prerequisite for monitoring mine production conditions,optimizing processes and improving production efficiency.Traditional flow pattern identification methods use complex multimodal sensors to obtain several features,which are inputted into a classifier to identify all flow patterns.However,the three-phase flow pattern is affected by both gasliquid structure and oil-water structure,and there is oil-water continuous phase transition phenomenon in gas-liquid structure,which leads to complicated flow pattern classification,and it is difficult to find several features that are sensitive to all flow patterns.This research takes the oil-gas-water three-phase flow in horizontal pipe as the research object,and proposes a flow pattern identification method of oil-gas-water three-phase flow with the fusion of ultrasonic/pressure drop dual modal information.Decoupling oil-gas-water three-phase flow pattern into gas-liquid structure and oilwater structure.Using ultrasonic reflection data to characterize the position of gasliquid interface and oil-water interface in the pipe to identify the gas-liquid structure.And using local Doppler technique to calculate the average fluid flow velocity,fusing the fluid pressure drop signal to identify the oil-water continuous phase information.Then we combine the oil-water interface information provided by ultrasonic sensor to identify the oil-water structure,and identify the flow pattern with decision classification tree model.The main research works completed include:(1)Under the premise of treating oil and water phases as a single liquid phase,we propose the interface search algorithm and the concept of partition interface polarization points based on the characteristics of pulsed-wave ultrasonic sensors that can provide position information,and according to the differences in the propagation behavior of ultrasonic waves at different media partition interfaces.We use the partition interface polarization points to characterize the location of different phase partition interfaces,and then characterize the phase distribution structure of the fluid in the pipeline to achieve accurate identification of the gas-liquid structure in the case of unknown oil-water structure.(2)In view of the large difference of oil-water dynamic viscosity,the differential pressure sensor is used to obtain the three-phase flow pressure drop information,extract the frequency domain characteristics of the pressure drop data to characterize the degree of dispersive phase aggregation in the oil-water structure,and calculate the friction factor of different continuous phase oil-water structures according to the DarcyWeisbach equation with the local Doppler technique to achieve the accurate identification of oil-water structures on the basis of known gas-liquid structures.(3)In order to solve the characteristics of complex three-phase flow structure,a decision classification tree model is proposed to identify three-phase flow patterns.The oil-gas-water three-phase flow pattern is decoupled into a combination of gas-liquid structure and oil-water structure,and two decision classification trees are established for two types of structure,which simplifies the multi-classification problem into multiple binary classification problems and selects the features for each binary classification problem to achieve accurate identification of the three-phase flow pattern.In order to verify the effectiveness of the above method,experiments were carried out on the oil-gas-water three-phase flow device of Tianjin University and data were collected.A total of 176 groups of experimental data of various flow patterns were obtained.The accuracy rate of identifying oil-gas-water three-phase flow patterns by using the above method reached 92.3%.