An improved model for gas-liquid flow pattern prediction based on machine learning

Abstract

The determination of flow patterns is a fundamental problem in two-phase flow analysis, and an accurate model for gas-liquid flow pattern prediction is critical for any multiphase flow characterization as the model is used in many applications in petroleum engineering. We developed a new model based on machine learning techniques via dimensionally analyzing more than 8000 laboratory multi-phase flow tests. As shown in the test results, the flow pattern is affected by fluid properties, in-situ flow rates of liquid and gas, flow conduit geometry and mechanical properties. Applying hydraulic fundamentals and dimensional analysis, three upscaling numbers are developed to reduce the number of freedom dimensions. These dimensionless variables are easy to use for upscaling and have physical meanings. Machine learning techniques on the dimensionless variables significantly improved their predictive accuracy. Until now the best matching on these laboratory data was approximately 80% using the most recently developed semi-analytical models. The quality of the matching is improved to 90% or greater on the experimental data using machine learning techniques.