Experimental investigation on transition mechanisms and modal decomposition analysis of gas–liquid flow patterns in horizontal–vertical elbow

A series of experiments are conducted to investigate the transition mechanisms and characteristics of six typical gas–liquid flow patterns in a horizontal–vertical elbow using electrical capacitance tomography and high-speed camera. The dominant modes and corresponding time coefficients are obtained by performing proper orthogonal decomposition on the pulsating gas holdup (GHU) distribution data to explore their physical mechanisms and correlations. Reduced-order descriptions for different flow patterns are discussed. The results show that after passing through the elbow, the horizontal slug or bubble flow turns into vertical bubble flow due to the small gas volume content and the mixing effect of secondary flow, accompanied by a swirl-switching phenomenon. A slug flow forms at the elbow outlet when there is a stratified flow comes from the horizontal pipe, and changes in flow conditions will affect the generation frequency and stability of Taylor bubbles. The horizontal annular or mist flow with high gas volume content will be transformed into churn flow in the vertical pipe. The modal decomposition analysis indicates that, for all the investigated conditions in the present study, mode 1 represents the mean distribution of GHU fluctuations, and there is a pair of modes representing the dominant swirling features. For the slug and churn flows, mode 2 characterizes the features of gas slug or large bubbles, the time coefficient of which is highly connected with that of mode 1. Meanwhile, it is also shown that the obtained low-dimensional descriptions of different flow patterns using the dominant modes are able to reconstruct most of the GHU distribution features in gas–liquid flows with the reconstructive loss less than 3%.