Flow Pattern Identification of Oil–Water Two-Phase Flow Based on SVM Using Ultrasonic Testing Method

Abstract

A flow pattern identification method combining ultrasonic transmission attenuation with an ultrasonic reflection echo is proposed for oil–water two-phase flow in horizontal pipelines. Based on the finite element method, two-dimensional geometric simulation models of typical oil–water two-phase flow patterns are established, using multiphysics coupling simulation technology. An ultrasonic transducer test system of a horizontal pipeline with an inner 50 mm diameter was built, and flow pattern simulation experiments of oil–water two-phase flow were carried out in the tested field area. The simulation results show that the ultrasonic attenuation coefficient is extracted to identify the W/O&O/W dispersion flow using the ultrasonic transmission attenuation method, and the identification accuracy is 100%. By comparison, using the ultrasonic reflection echo method, the echo duration is extracted as an input feature vector of support vector machine (SVM), and the identification accuracy of the stratified flow and dispersed flow is 95.45%. It was proven that the method of the ultrasonic transmission attenuation principle combined with the ultrasonic reflection echo principle can identify oil–water two-phase flow patterns accurately and effectively, which provides a theoretical basis for the flow pattern identification of liquid–liquid multiphase flow.