Digital image analysis of gas-liquid flow in a cross-corrugated plate heat exchanger channel: A feature-based approach on various two-phase flow patterns

• The algorithm is applied to various flow patterns without pre-classification. • The method is suitable to analyze inhomogeneous and fluctuating two-phase flows. • The effects of flow direction and maldistribution are quantified with key features. • Measured key features are compared to a flow pattern map for the same test channel. • The maximum bubble diameter model is used to validate the measured bubble sizes. High-resolution photographs of the air-water flow in a transparent cross-corrugated channel are analyzed using digital image processing. Gas-liquid distribution is controlled via several inlet nozzles. Flow direction is varied between upward, downward, and horizontal flow. The image analysis algorithm is applied to a wide range of void fractions and flow patterns including bubbly, intermediate and film flow. Large bubble clusters are separated using edge detection. Local film and bubbly flow are identified with texture analysis. The method is suitable for locally inhomogeneous two-phase flow which is difficult to analyze visually. Objective and reproducible results are obtained by measuring features in the processed images: Mean and maximum bubble diameter, and the ratio of local film flow are determined. For non-uniform gas injection, expansion of two-phase flow across channel width is measured. Impact of flow direction and maldistribution on two-phase flow is demonstrated by calculating differences of the features. Uncertainties of image processing results are quantified. Bias is discussed based on visual analysis of the photographs in a previous study. Models for maximum bubble diameter and local film flow ratio, used to model flow pattern transitions in a previous study, are verified: For uniform two-phase distribution, models generally agree with measured values.