Direct and Inverse Modeling of a Microwave Sensor Determining the Proportion of Fluids in a Pipeline

A microwave sensor constituted of both low- and high-frequency probes is presented. The role of the sensor is to determine the proportion of mixed fluids in a pipeline used in petroleum engineering to extract oil from an oil layer. In this paper, the annular flow regime in the pipeline is considered. Since the analytical solution for this configuration is not available for high frequencies, a combination of numerical modeling and neural networks (NNs) is used. Two inversion configurations are compared. Good results are obtained with the better one.