Prediction of Volume Fractions and Flow Regime Using Nuclear Technique and Deep Rectifier Neural Network

Abstract

During the extraction of oil, from submerged rocks, it is necessary to know the flow of oil, water, and gas that flow through the pipes of the oil plant. This real-time knowledge allows for better performance management in plant production. Therefore, this report presents an approach to the use of artificial neural network techniques to obtain, simultaneously, the three-volume fractions in three-phase flows (gas, water, and oil), and the active flow regime (Annular, Stratified and Homogeneous), through the interpretation of the height of pulses collected through radiation detectors.