A novel metering system consists of capacitance-based sensor, gamma-ray sensor and ANN for measuring volume fractions of three-phase homogeneous flows

Measuring the volume fraction of different types of fluids with two or three phases is so vital. Among all available methods, two of them, capacitance-based and gamma-ray attenuation, are so popular and widely used. Moreover, nowadays, AI which stands for Artificial Intelligence can be seen almost in all areas, and the measuring section is no exception. In this paper, the main goal is to predict the volume fraction of a three-phase homogeneous fluid which contains water, oil, and gas materials. To opt for an optimised method, a combination of capacitance-based sensors, gamma-ray attenuation sensor and Artificial Neural Networks (ANN) is utilised. To train the proposed metering system which is a MLP type, two inputs are considered. For the first input, the concave sensor is simulated in COMSOL Multiphysics software and different combinations of three phases (different volume fractions) are applied. Then through theoretical investigations of gamma-ray sensor, Barium-133 which radiates 0.356 MeV is used. This way, the second required input is generated. Finally, to implement a new and accurate metering system, a number of networks with different characteristics are run in the MATLAB software. The best structure had a Mean Absolute Error (MAE) equal to 0.33, 3.68 and 3.75 for the water, gas and oil phases, respectively. The accuracy of the presented metering system is illustrated by the received outcomes. The novelty of this study is proposing a new combined method that can measure a three-phase homogeneous fluid’s volume fractions containing water, gas and oil, precisely.