Constructing response surfaces with Kriging models for quick estimation of radionuclide activity in spent nuclear...

Spent nuclear fuel discharged from the reactor core of a nuclear power plant at end of each operating cycle is usually temporarily stored in cooling pools and eventually will be placed in waste canisters buried 500 m below the surface. Over time, radioactive isotopes within these fuel assemblies continue to decay and release radiation, potentially posing risks to the biosphere. To address this concern, monitoring the variation in radioisotope activity with cooling time is essential. This study explores an indirect method employing a Kriging model to construct response surfaces to perform a quick assessment of activity of key radioactive nuclides in spent nuclear fuel at various cooling times. Using data from Taiwan's Chinshan Nuclear Power Plant Unit 1 as an example, 1855 spent nuclear fuel assemblies were selected from a pool stored with 3390 fuel assemblies, based on parameters such as initial uranium-235 enrichment, total burnup, and specific power. The selected assemblies were then analyzed using SCALE 6.2 to calculate the activities of radioisotopes critical to safety, forming a comprehensive database. Properly selected parameters enabled the establishment of an appropriate Kriging model. The process involved three major steps: initial sampling, Kriging model establishment for response surface estimation, and optimization based on maximum errors allowed. This iterative process resulted in the optimal Kriging model for estimating the response surface. The response surface considered four variables: specific power, initial uranium-235 enrichment, total burnup, and cooling time. Validation of the Kriging models using spent nuclear fuel from Taiwan's Chinshan Power Plant and Kuosheng Power Plant showed that the response surfaces well-predicted the nuclide activity as it varies with cooling time, implying that the method may be used by the utility for safe and efficient management of nuclear waste.