A Unified Mechanistic Model to Classify Flow Patterns and Predict Pressure Drops in Oil–Water Pipeline Systems

Abstract The intricate dynamics of oil–water (OW) flow, driven by diverse flow patterns (FPs) and phase interactions, present a significant challenge for accurately classifying FP and predicting pressure drops in multiphase pipelines. This paper proposes a novel, unified mechanistic model that accurately predicts OW FPs, inversion, and pressure drop in pipelines. The OW FPs, categorized into primary (annular, stratified, and distributed) and secondary types (dispersed and nondispersed), capture a broad range of flow behaviors. The proposed FP classification uniquely identifies key physical mechanisms, such as stratification and dispersion, using dimensionless parameters, including the Weber and Froude numbers, with their ranges identified through genetic optimization. Using data from 16 experimental studies spanning a wide oil viscosity range (1 to 120 mPa·s), the FP classification model achieved accuracies of 80.1% for primary and 77.8% for secondary patterns, outperforming prior mechanistic methodologies, despite covering a broader oil viscosity range. The inversion model, based on oil-to-water density and viscosity ratios, demonstrated a mean absolute percentage error (MAPE) of 15.3%, significantly lower than existing models, which often exhibited errors more than twice as large, even at lower viscosity ranges. By utilizing predictions from the FP classification and inversion models, and incorporating a two-fluid or homogeneous framework, robust pressure drop predictions were achieved with a MAPE of 24.4%, marking an improvement of over 50% compared to previous studies. The proposed unified model eliminates reliance on conventional geometry-specific FP maps, providing a reliable and comprehensive tool for optimizing pipeline design and operational efficiency in multiphase liquid–liquid flow systems.