Prediction of Air and Water Flow-Rates Independent of Flow Regimes Using Gamma-Ray Attenuation Technique and Artificial Neural Network

Gas-liquid two-phase flow is probably the most important form of multiphase flows and is found widely in the oil industry. The accurate prediction of the air and water flow-rates are important in two-phase flow. Nowadays, multiphase flow-rates measurement by gamma-ray attenuation technique is known as one of the most common precise methods. In this work, the air and water flow-rates independent of flow regime changes were accurately predicted within a two-phase flow loop in the laboratory. For this purpose, a combination of single beam gamma-ray, single detector and artificial neural network (ANN) were used in order to predict the flow-rates in the bubble, plug, slug, annular and dispersed regimes of gas-liquid two-phase flows. Two different types of neural networks (GMDH) were developed. The networks were developed based on four features extracted from recorded pulse height distribution in a dynamic condition. The result shows, air, and water flow-rates were measured with an average of Mean Relative Error (MRE) less than 4.5%. Overall results revealed that using the proposed method, gamma-ray attenuation technique combined with an ANN model can be efficiently used to predict the flow-rates. Furthermore, in this study, a new method based on a single beam, single energy, and the single detector was proposed in order to solve this problem, without any recalibration