Physics-based forward model for near-real-time quantitative imaging of spent nuclear fuel assemblies

Area of Science:

• Nuclear Engineering
• Applied Physics
• Nuclear Safeguards

Background:

• Passive Gamma Emission Tomography (PGET) is IAEA-authorized for spent nuclear fuel verification.
• Accurate fuel pin activity quantification is challenging due to self-shielding and scattering.
• Existing PGET methods struggle with precise activity measurements.

Purpose of the Study:

• To develop and validate a linear inverse approach for PGET image reconstruction.
• To improve the accuracy of fuel pin activity quantification in spent fuel assemblies (SFAs).
• To enhance the speed and efficiency of PGET system modeling.

Main Methods:

• Developed a physics-based forward model incorporating collimator septal penetration and detector scattering.
• Framed image reconstruction as a linear inverse problem.
• Validated the enhanced forward model using VVER-1000 and VVER-440 SFA simulations, comparing with MCNP.

Main Results:

• Achieved a 3.7% relative difference in counts between the forward model and MCNP simulations.
• Demonstrated superior image quality and 100% fuel pin identification accuracy in simulations.
• Attained an average uncertainty of 2.3% in activity quantification.
• Successfully reconstructed images from measured VVER-440 SFA data, imaging all pins and the water channel.

Conclusions:

• The enhanced forward model significantly accelerates PGET sinogram simulation and system matrix calculation (>100,000x faster than Monte Carlo).
• The linear inverse approach provides accurate fuel pin activity quantification and improved image quality.
• The developed method shows strong potential for real-world nuclear fuel inspection and future algorithm development.