Machine learning applications to predict two-phase flow patterns

Abstract

Recent advances in artificial intelligence with traditional machine learning algorithms and deep learning architectures solve complex classification problems.This work presents the performance of different artificial intelligence models toclassify two-phase flow patterns, showing the best alternatives for this specificclassification problem using two-phase flow regimes (liquid and gas) in pipes. Flowpatterns are affected by physical variables such as superficial velocity, viscosity,density, and superficial tension. They also depend on the construction characteristicsof the pipe, such as the angle of inclination and the diameter. We selected 12databases (9,029 samples) to train and test machine learning models, consideringthese variables that influence the flow patterns. The primary dataset is Shoham(1982), containing 5,675 samples with six different flow patterns. An extensive set ofmetrics validated the results obtained. The most relevant characteristics for trainingthe models using Shoham (1982) dataset are gas and liquid superficial velocities,angle of inclination, and diameter. Regarding the algorithms, the Extra Trees modelclassifies the flow patterns with the highest degree of fidelity, achieving an accuracyof 98.8%