Application of artificial intelligence and gamma attenuation techniques for predicting gas–oil–water volume fraction in annular regime of three-phase flow independent of oil pipeline’s scale layer

To the best knowledge of the authors, in former studies in the field of measuring volume fraction of gas, oil, and water components in a three-phase flow using gamma radiation technique, the existence of a scale layer has not been considered. The formed scale layer usually has a higher density in comparison to the fluid flow inside the oil pipeline, which can lead to high photon attenuation and, consequently, reduce the measuring precision of three-phase flow meter. The purpose of this study is to present an intelligent gamma radiation-based, nondestructive technique with the ability to measure volume fraction of gas, oil, and water components in the annular regime of a three-phase flow independent of the scale layer. Since, in this problem, there are several unknown parameters, such as gas, oil, and water components with different amounts and densities and scale layers with different thicknesses, it is not possible to measure the volume fraction using a conventional gamma radiation system. In this study, a system including a²⁴¹ Am-¹³³ Ba dual energy source and two transmission detectors was used. The first detector was located diametrically in front of the source. For the second detector, at first, a sensitivity investigation was conducted in order to find the optimum position. The four extracted signals in both detectors (counts under photo peaks of both detectors) were used as inputs of neural network, and volume fractions of gas and oil components were utilized as the outputs. Using the proposed intelligent technique, volume fraction of each component was predicted independent of the barium sulfate scale layer, with a maximum MAE error of 3.66%.