Ultrasound Measurement for Prediction of Gas-Liquid Two-Phase Flow Patterns*

A method for identifying the flow patterns of gas-liquid two-phase flow in a vertical pipe using ultrasound was developed in this study. The method of analyzing the gas-liquid two-phase flow rising in a vertical pipe can be applied to an immediate measurement method in an airlift pump which is potentially used for pumping resources such as rare earth mud and cobalt rich crusts existing in the deep-sea bottom. Two-phase flow pattern identification method was realized using ultrasound echo intensity and Doppler velocity distribution by unsupervised machine learning. Measuring flow patterns makes it possible to predict the subsequent flow based on a two-phase flow model in a vertical pipe and control a pump system. By statistically processing both distributions in the distance and time directions in this order several features such as the mean of all echoes and frequency components of the temporal variation of the flow velocity fluctuation were extracted. Flow characteristics were visualized by applying dimensionality reduction method to these features. Moreover, we compared the result with the existing flow pattern maps and human sensory identification and evaluated the performance.