Full-range two-group drift-flux model for two-phase flows in vertical pipes

Accurate prediction of void fraction is critical for the safety and efficiency of thermal, nuclear, and chemical engineering systems. Among various approaches, the drift-flux model (DFM) is widely used for predicting void fraction. The DFM relies primarily on two critical parameters: the distribution parameter and the drift velocity, both of which are essential for predicting void fraction and can also be used to formulate the interfacial drag force for the interfacial momentum transfer terms in the two-fluid model. The one-group ( 1 G) DFM, which treats all bubbles as one group regardless of the bubble size, is widely used. The 1G DFM does not account for the differences in bubble size and drag coefficients between small and large bubbles. To address this, a two-group ( 2 G) approach has been employed to distinguish between distorted bubbles (Group 1) and larger cap, slug, or churn-turbulent bubbles (Group 2) in dispersed flows. Despite these advancements, a comprehensive 2G drift-flux correlation (DFC) for distribution parameter and drift velocity that accounts for 2G bubble characteristics while covering the full void fraction range from dispersed to separated flow remains unavailable. To bridge this gap, this study introduced a novel full-range 2G DFC capable of predicting both adiabatic and boiling flows across the entire void fraction range in pipes. Constitutive correlations for the 2G DFC were analytically derived based on physical considerations, with modeling parameters introduced where necessary, and incorporated nucleate boiling suppression while ensuring mathematical continuity. The developed correlations were indirectly evaluated against an extensive experimental database consisting of steam-water, air-water, and air-aqueous glycerol flows under both adiabatic and boiling conditions. The evaluation demonstrated that the developed formulation could predict liquid film characteristics in annular flows, achieving mean absolute percentage errors of 16.1% for the average liquid film thickness and 30.2% for the maximum liquid film thickness.