Slug-to-churn or churn-to-slug: revisiting the flow patterns transition debate

Abstract

Perhaps one of the most contentious yet long-lasting debates in slug and churn flow literature revolves around the directional nature of the transition between these two flow patterns. Which terminology truly captures its nature—slug-to-churn or churn-to-slug transition? The present study is tackling this debate through an experimental investigation by leveraging high-speed flow visualization and a synergistic combination of advanced signal processing techniques. The analysis is performed for void fraction waves recorded at Z/D = 10, 25, 40, and 60 in an air-water flow along a vertical pipe under gravity-driven conditions at an elevated inlet superficial gas velocity. Visual insights from high-speed imaging conducted at the same spatial positions, combined with statistical analysis and a spatiotemporal-spectral framework incorporating Recurrence Quantification Analysis (RQA), Power Spectral Density (PSD), and Direct and Continuous Wavelet Transforms (DWT and CWT), provided a multidimensional, cross-validated approach—both qualitative and quantitative—to conclusively determine the transition mechanisms and direction. The findings establish churn flow as a spatial precursor to slug flow, unfolding through four distinct regimes: semi-annular, churn, churn-slug transition, and unstable slug flow at Z/D = 10, 25, 40, and 60, correspondingly. The churn-slug transition emerged as a gradual process, wherein diminishing phase interaction-induced instabilities allow slug flow characteristics to take hold. A previously unnoticed mechanism, termed liquid phase penetration, was uncovered as a fundamental driver of churn flow. Propelled by momentum transfer from incoming gas plugs, this mechanism destabilizes leading gas plugs, amplifies huge wave formation, and reinforces flooding dynamics, propagating churning behaviour in upward direction. Its role is pivotal, making its incorporation into slug/churn transition models—especially those based on the entrance effect theory—imperative. Moreover, the study confirmed the exceptional performance (∼99.85 % accuracy) of a novel AI-based diagnostic tool, integrating the CWT framework with a CNN, offering a real-time, scale-independent data-driven solution for axial flow pattern identification (i.e., static instability diagnosis), promising enhanced operational reliability and safety in systems of varying dimensions operating under developing two-phase flow conditions. Nonetheless, this study offers a preliminary contribution, aiming to ignite discussion, encourage future endeavours, and shape the trajectory of future investigations into the slug/churn transition. To solidify the present findings, further experimentation in long pipes and across a broad range of inlet superficial gas velocities is essential.