Calcium carbonate scales occur on the inner walls of the pipes used in oil and gas production, reducing the internal diameter and obstructing fluid flow. This study proposes a method for predicting of scale thickness using a neutron source and gamma radiation analysis with a deep neural network (DNN). The detection system includes a 241Am-Be neutron source and 72 detectors that count particles and record photons. A measurement setup was designed using the MCNP6 code to optimize the detection of scale thickness, utilizing a flat neutron source and strategically positioned detectors. The gamma-ray spectra generated from this setup were used as input data for the DNN to predict scale thickness within a three-phase water–gas-oil annular system. The results indicated that the DNN accurately predicted scale thickness with a mean relative error of 3.01 %, achieving a relative error below 5 % for 92.68 % of the dataset, demonstrating the potential for reliable real-time monitoring in industrial settings.
Read full abstract