Related Experiment Video
Updated: Jun 13, 2025
Asymmetrical Flow Field-Flow Fractionation for Sizing of Gold Nanoparticles in Suspension
Published on: September 11, 2020
Volume fraction detection in multiphase systems using neutron activation analysis and artificial neural network
R S F Dam1, W L Salgado1, C C Conti2
1Universidade Federal do Rio de Janeiro, Programa de Engenharia Nuclear (UFRJ/PEN), Avenida Horácio de Macedo, 2030, Bloco G, sala 206, 21941-914, Cidade Universitária, RJ, Brazil; Instituto de Engenharia Nuclear (IEN), Rua Hélio de Almeida 75, 21941-906, Cidade Universitária, RJ, Brazil.
This study applies Artificial Neural Networks (ANN) with Prompt-Gamma Neutron Activation Analysis (PGNAA) to identify fluids in oil exploration. The method accurately predicts fluid volumes in multiphase systems, achieving over 92% accuracy.
Area of Science:
- Nuclear Physics and Engineering
- Artificial Intelligence and Machine Learning
- Petroleum Engineering
Background:
- Multiphase flow analysis is critical in oil exploration.
- Accurate fluid identification is essential for process optimization and safety.
- Traditional methods may face challenges in complex annular flow regimes.
Purpose of the Study:
- To develop and validate an Artificial Neural Network (ANN) model for fluid detection in annular flow.
- To integrate ANN with Prompt-Gamma Neutron Activation Analysis (PGNAA) for enhanced multiphase system analysis.
- To assess the accuracy and generalization capability of the ANN-PGNAA approach for saltwater, oil, and gas mixtures.
Main Methods:
- Utilized MCNP6 Monte Carlo simulations to generate gamma-ray spectra from neutron interactions.
- Simulated an Americium-241/Beryllium (Am-Be) neutron source for spectral data generation.
- Trained an Artificial Neural Network (ANN) on simulated spectral data representing various fluid fractions.
- Evaluated the ANN's performance on independent datasets to test generalization.
Main Results:
- The ANN model accurately predicted the volume fractions of saltwater, oil, and gas.
- The model demonstrated robust generalization, performing well on unseen data combinations.
- Over 92% of predictions showed less than 5% error.
- The combined ANN and PGNAA approach proved effective for multiphase fluid analysis.
Conclusions:
- Artificial Neural Networks combined with PGNAA offer a highly accurate method for fluid detection in multiphase annular flow.
- This integrated approach enhances the analysis of complex systems relevant to oil exploration.
- The study validates the potential of AI-driven nuclear analysis techniques in the energy sector.
