Characterizing the development of gravity-driven slug flows using high-speed imaging and PIV-PLIF techniques

Abstract

Although developing gravity-driven slug flow frequently occurs in oil and gas, and energy systems, its development dynamics remain underexplored; this gap, in turn, has left the underlying relationships between flow evolution and transport phenomena in these applications inadequately characterized as well. The present study experimentally investigates the spatiotemporal-spectral development of gravity-driven air–water and CO₂-water slug flows in a vertical 25.4 $\mathrm{mm}$ ID pipe. Enhanced flow visualization techniques, utilizing high-speed imaging and particle image velocimetry-planar laser induced fluorescence (PIV-PLIF), were employed to determine the behaviors of gas and liquid phases and interactions at four positions along the pipe axis. A machine vision-based algorithm was employed to extract slug unit cell characteristics and instantaneous void fraction signals, allowing for a comprehensive statistical analysis of gas phase behavior across the flow domain. A novel algorithm was also developed to preprocess raw PIV-PLIF images, facilitating phase discrimination and noise reduction before PIV cross-correlation analyses are conducted. The results showed a logarithmic growth in the lengths of Taylor bubbles, liquid slugs, and slug unit cells along the pipe, with liquid slugs constituting nearly 60 % of slug units downstream. Taylor bubble length distributions correlated well with log-normal fits, while liquid slug and unit cell lengths transitioned from log-normal patterns upstream to near-normal distributions downstream with broader and less peaked shapes. Taylor bubble velocities and appearance frequencies of the flow structures declined exponentially along the pipe, with Taylor bubble velocities showing narrower and more peaked near-normal distributions downstream. Instantaneous void fraction signals exhibited fewer, wider peaks and troughs with reduced small-amplitude oscillations downstream. The analysis of the signals indicated a complete bubbly-to-slug transition at $Z/D=30$. Gas-liquid phase interactions, classified as almost-zero, weak, and strong, impacted liquid phase velocity profiles and the behavior of Taylor bubbles, with minimum stable liquid slug lengths of 8–9 $D$ and a wake length of 1.8 $D$ observed. Empirical correlations were developed to represent the spatiotemporal-spectral aspects of flow development, with spectral parameters, particularly liquid slug frequency, identified as the most reliable indicators of the fully developed region, predicting entrance lengths of 114.0 $D$ and 113.4 $D$ for air–water and CO₂-water, respectively. Gas density was found to strongly influence flow characteristics and transition, accelerating the approach to the fully developed region.