A reassessment of film thickness prediction in upward annular flow: The role of local parameters and entrainment

Accurate prediction of liquid film thickness in upward annular flow remains an important challenge in multiphase-flow modelling, particularly for high-heat-flux applications such as steam generators, evaporators, and nuclear reactor systems. Despite numerous empirical and semi-empirical correlations, no universally applicable model exists that accounts for the combined effects of flow development, entrainment, and gas compressibility. In the present study, film-thickness data from multiple literature sources spanning a wide range of fluids, geometries, operating pressures, and measurement techniques were assembled and analysed. Three representative approaches were evaluated: the local shear-stress-based model of Anderson and Mantzouranis (1960), the inlet-parameter-based correlation of Ju et al., (2015), and the multi-property correlation of Zhang et al., (2023). The database was classified into three groups according to the availability of local measurements and the degree of entrainment. Analysis of datasets with local film-loading measurements demonstrated that boundary-layer theory successfully predicts film thickness when local liquid loading and interfacial shear stress are used. In contrast, correlations based solely on inlet conditions were found to systematically overpredict film thickness when significant entrainment occurred, owing to the progressive reduction in wall-film loading downstream. Examination of measurements obtained at multiple axial locations further showed that gas compressibility increases local gas velocity and interfacial shear stress, promoting entrainment and modifying film thickness even when inlet conditions remain unchanged. Motivated by the boundary-layer formulation, an empirical power-law correlation was developed using Reynolds, Weber and Ohnesorge numbers and validated using datasets representative of fully developed or weakly entraining annular flow. The correlation achieved a mean absolute percentage error of 12.7% and a root mean square percentage error of 16.1% across the combined Group 1 and Group 2 database. The resulting exponents show good agreement with those proposed by Ju et al., (2015) and Zhang et al., (2023), indicating that these correlations capture the dominant effects of gas inertia, liquid inertia and viscosity. However, deviations from inlet-based correlations were shown to be strongly linked to the presence of entrainment. The results indicate that local flow parameters are important for reliable prediction of film thickness in developing annular flows and suggest that annular-flow regime maps should be formulated using local rather than inlet conditions. The proposed framework provides a physically motivated basis for improved annular-flow modelling and may be extended in future work to include heat-transfer effects.