Real-Time Oil Pipeline Integrity Assessment Using Hybrid MCNPX-MLP and Na-۲۴ Radiotracer: Corrosion, Leakage, and Material Identification

Maintaining crude oil pipeline integrity requires addressing three challenges: timely leak detection, quantitative corrosion monitoring, and real time identification of internal fluid composition and mineral scale types. Sodium ۲۴ (Na ۲۴), with a ۱۵ hour half life and dual gamma emissions at ۱.۳۶۹ and ۲.۷۵۴ MeV, serves as an effective radiotracer for these tasks due to high penetration through steel walls and strong dependence of gamma attenuation on effective atomic number and density. This study presents a dual modality framework combining MCNPX Monte Carlo simulations with a multi layer perceptron (MLP) neural network for pipeline diagnostics. The geometry of a stainless steel pipeline, concentric and eccentric corrosion scales of variable thicknesses, a point Na ۲۴ source, collimated NaI(Tl) detectors, and diverse internal fluids including oil water gas mixtures and mineral scales such as calcium carbonate, barium sulfate, and iron compounds were precisely modeled. This parametric modeling generated a synthetic dataset of gamma ray spectra and radiographic images under thousands of possible leak, corrosion, and fluid change scenarios. The high fidelity dataset was used for training, cross validation, and optimization of the MLP network to simultaneously predict four parameters: remaining wall thickness, defect location, Na ۲۴ leakage intensity, and internal material type. Cross validation results demonstrated agreement with experimental data. The trained network achieved a mean absolute relative error below ۰.۳% and an R² exceeding ۰.۹۸۸ for corrosion thickness estimation, while attaining perfect accuracy (۹۸.۲%) in defect localization, leakage discrimination, and classification of six distinct fluid and scale compositions. Despite the time consuming nature of Monte Carlo simulations, the optimized network delivers comparable results in under one millisecond. This speed advantage renders the proposed method a powerful tool for real time pipeline inspection, integrity management, and rapid safety decision making during emergencies

کلیدواژه ها:

Pipeline integrity ، Na-۲۴ radiotracer ، MCNPX simulation ، MLP neural network ، Corrosion monitoring