AI Tool Accelerates Nuclear Material Inspections
A software algorithm developed by the US Oak Ridge National Laboratory (ORNL) has reduced the time needed to inspect 3D-printed parts for nuclear applications by 85 %. The software is being supported through the US Department of Energy’s (DOE’s) Advanced Materials & Manufacturing Technologies (AMMT) program to accelerate commercialization of new materials and manufacturing technologies through demonstration and deployment.
Researchers are now training the algorithm for Idaho National Laboratory (INL) to apply similar methods for irradiated materials and nuclear fuel. Additive manufacturing (3D-printing) can enable the domestic fabrication of complex nuclear parts in a short time. The quality of such products is typically verified through computed tomography or CT scans, which use X-rays to capture images of any weakness or errors in the internal structure.
INL researchers often delay examining materials removed from a nuclear reactor for the safety of lab technicians. Radiation accrued during long X-ray CT scans can also affect the detector, limiting its operating life and the accuracy of its images. Shorter scans would mean less radiation dosage per scan and less waiting, while enabling higher-quality data and faster feedback to performance models.
ORNL’s software algorithm uses machine learning to rapidly reconstruct and analyze the images to significantly cut down the cost, time and number of scans needed to perform an inspection. Researchers at INL applied ORNL’s new algorithm to analyze more than 30 3D-printed sample parts in less than five hours of scan time. Without the software, it would have taken more than 30 hours to complete each scan, paving the way for potential applications with radioactive materials and fuels.