Understanding materials degradation is critical to determining nuclear power plant lifespans, but analyses overlook a key factor for nuclear plant applications: how the sensors used to assess the materials also degrade. A new paper shows how sensor degradation can be explicitly determined from real nuclear plant data and used to determine more accurate remaining useful and efficient lifetimes for nuclear reactor pressure vessels.
Researchers at the University of Illinois Urbana-Champaign, led by NPRE assistant professor Syed Bahauddin Alam, have implemented a data-driven model for sensor degradation for reactor pressure vessels based on nearly 26 years of data from the Ameren Missouri Callaway Reactor 1. As reported in the journal npj Materials Degradation, the model allowed the researchers to considerably refine calculations of the nuclear reactor pressure vessel’s lifetime.
“Despite the importance of understanding sensor degradation in assessing quality, reliability and safety of nuclear reactor pressure vessel, there’s very little data-driven work on the nuclear plant and asset monitoring,” Alam said. “Ours is one of the first works to incorporate nuclear plant data into existing sensor models and calculations for reactor pressure vessel, and we’re continuously expanding it to include more data and create more powerful models.”
Materials in nuclear reactors are subjected to extreme heat and radiation, so they naturally degrade. Assessing this degradation is critical to determining when reactor components have reached the end of their useful lives. While this is a major research focus in nuclear power, most efforts do not consider how the degradation of sensors used to monitor the reactor pressure vessel can impact assessments.
“Degradation of reactor components and pressure vessel is itself a statistical process, but the degradation of the sensors adds another statistical process on top of that,” said Raisa Hossain, a graduate student in Alam’s group and the study’s lead author. “We used an established and proven algorithm called the Kalman filter to simultaneously extract the details of both the component and sensor degradations from the final sensor readings, where the two degradations compound.”
When the researchers simulated pressure vessel degradation with their model, they found that excluding sensor degradation led to noticeably different estimates for the remaining useful life. Alam and his group plan to build on this result and develop even more sophisticated methods for safety and reliability analysis.
“We demonstrated that data-driven degradation models are possible, but we’re in the process of building machine learning models that can very accurately capture degradation effects,” Alam said. “We’re even having distributed fiber optic sensors in our lab to test out our ideas.”
The study, “Sensor Degradation in Nuclear Reactor Pressure Vessels: The Overlooked Factor in Remaining Useful Life Prediction,” is available online. DOI: 10.1038/s41529-024-00484-4
Kazuma Kobayashi, a graduate student in Alam’s group, also contributed to this work.
Alam is also a faculty affiliate of the National Center for Supercomputing Applications at Illinois.