Automatic classification of vertical counter-current two-phase flow by capturing hydrodynamic characteristics through objective descriptions

An experimental investigation has been carried out for vertical counter-current two-phase flow of air and water, and then an attempt has been made to identify and classify the individual flow regime pattern. At first, visual observation-based flow regime clusters have been constructed through digital photography and video recording. As the visual description is subjective in nature, two methodologies (classifiers 1 and 2) have been developed to classify the flow regime patterns using objective descriptions. Two types of conductive probes (that is, parallel and ring probes) have been constructed for this purpose. Clusters of flow regimes have been obtained based on the conductive probe signal features using both the classifiers separately. Clusters generated by the classifiers based on the conductive probe signal features are found to be very similar to the visually observed ones. The parallel probe is found to be very effective to classify the flow regime pattern without any human linguistic descriptions.

► Vertical counter-current two-phase flow.
► Classification of flow regimes.
► Fuzzy C-means clustering.
► Entropy-based clustering.
► Genetic Algorithms.