Image processing techniques for high-speed videometry in horizontal two-phase slug flows

Abstract

Two-phase flow measurements are very common in industrial applications especially in oil and gas areas. Although some works in image segmentation have analyzed gas–liquid slug flow along vertical pipes, few approaches have focused on horizontal experiments. In such conditions, the detection of the Taylor bubble is challenging due the great amount of small bubbles in the slug area and, thus, requires a special treatment in order to separate gas from liquid phases. This article describes a new technique that automatically estimates bubble parameters (e.g. frequency, dimension and velocity) through video analysis of high-speed camera measurements in horizontal pipes. Experimental data were obtained from a flow test section where slug flows were generated under controlled conditions. Image processing techniques such as watershed segmentation, top-hat filtering and H-minima transform were applied to detect and estimate bubble contour and velocities from the observed images. Finally, the estimated parameters were compared to theoretical predictions, showing good agreement and indicating that the proposed technique is a powerful tool in the investigation of two-phase flow.