ARTIFICIAL NEURAL NETWORK AND GAMMA RADIATION APPLIED TO OIL PIPELINE SCALE CALCULATION

Abstract

This report investigates a technique for detecting and measuring barium sulfate (BaSO₄) scale deposits, which are common in the oil sector and reduce internal diameter, impeding flow. A gamma-ray source (¹³⁷Cs) and three NaI(Tl) detectors were used to measure the maximum thickness of eccentric scale. An artificial neural network (ANN) was trained with detector data to predict maximum scale thickness, regardless of the presence of gas, oil, saltwater, or scale inside the tube. The MCNP6 code generated a dataset for training and assessing the ANN's generalization capabilities, considering various maximum scale thicknesses and locations. Over 90% of the patterns had relative errors within ±10%.