Usage of two transmitted detectors with optimized orientation in order to three phase flow metering

Abstract

Artificial neural network (ANN) is a good technique used to handle problems of modeling, prediction, control, and classification. In this study, four accurate and precise MLP model were developed. The first one was used in order to determine the type of three phase flow regime. All of the regimes were recognized correctly. Then, according to determined regime, three independent MLP models were used in order to predict the volume fraction percentages of gas, oil and water. The networks were developed based on the validated simulated data from MCNPX code. The volume fractions were measured with Mean Relative Error (MRE) of less than 1.63%.