Objective Flow Regime Identification for Inclined Two-Phase Flows

Two-phase flow regimes are categorizations of flow patterns based on the geometric characteristics of the interface between phases. Flow regimes are widely implemented in thermal-hydraulics and reactor safety simulation tools. It is therefore crucial to properly classify flow conditions into their respective regimes to design and operate multiphase flow systems. The traditional approach to flow regime classification entails recording two-phase flows in transparent pipes with high-speed video cameras and making subjective judgements about the flow based on visualization. This subjectivity is susceptible to error and disagreement between researchers. Additionally, pipe inclination introduces additional complexity that makes visual classification of flow regimes challenging. Recent work with machine learning has endeavored to reduce the subjectivity involved in flow regime identification and classification efforts.

ART has developed a modular structure of k-means algorithms designed to objectively organize horizontal, inclined, and vertical flow conditions into flow regimes using an experimental database obtained with the separate effects test facility.

Images of flow conditions at different angles are shown in Figure 2, highlighting the changes to the interfacial structure that can occur with respect to angle.