Flow rate determination in a two-phase system using radioactive particle tracking and deep learning.

Area of Science:

• Nuclear Engineering
• Petroleum Engineering
• Computational Fluid Dynamics

Background:

• Accurate flow rate determination is vital in the oil industry for operational efficiency.
• Conventional flowmeters often require direct fluid contact, leading to maintenance issues and increased costs.
• Non-intrusive methods are needed to overcome limitations of traditional flowmeters.

Purpose of the Study:

• To propose a minimally intrusive methodology for flow rate determination in two-phase systems.
• To predict fluid volume fractions and calculate superficial velocities using radioactive particle tracking.
• To reduce operational costs and system shutdowns associated with conventional flowmeters.

Main Methods:

• Utilized radioactive particle tracking with a sealed Cesium-137 source.
• Simulated a stratified saltwater-oil flow in a PVC pipe using MCNP6 Monte Carlo code.
• Employed deep neural networks for volume fraction prediction and cross-correlation for velocity calculation.

Main Results:

• Achieved accurate prediction of fluid volume fractions.
• Calculated superficial velocities using time delays from cross-correlation of oil-phase signals.
• Attained a maximum mean absolute percentage error (MAPE) of 2.24% for oil flow rate compared to theoretical values.

Conclusions:

• The radioactive particle tracking technique offers a viable, minimally intrusive solution for oil flow rate determination.
• Deep neural networks and cross-correlation effectively processed the radioactive signals for accurate flow analysis.
• The proposed method demonstrates high accuracy and potential for reducing costs in the oil industry.