Horizontal oil-water two-phase flow characterization and identification with pulse-wave ultrasonic Doppler technique

Abstract To characterize and identify the flow patterns of horizontal oil-water two-phase flow, a pulse-wave ultrasonic Doppler technique was adopted to acquire spatiotemporal flowing information non-intrusively. The echogram was firstly constructed from the echo amplitude to analyze the interrelation between the concentration distribution and the echo intensity under different flow patterns. Then the time-averaged velocity profile was calculated from the Doppler frequency shift to investigate the velocity changes at different radial positions with flow patterns. Further, the long-term and dynamic connectivity of time-varying velocity profiles were analyzed through the graph-variate dynamic (GVD) connectivity method to reveal coupling effects in the spatiotemporal velocity field. Based on the above analysis, several features were extracted objectively to characterize flow patterns from different aspects, and their evolutions with different flow conditions were analyzed. The effectiveness of these features was validated through a multi-class support vector machine model with the total identification accuracy of 93.33%.