Securing the safeguardability of a reprocessing process for spent nuclear fuels (SNFs) is imperative. Particularly, the quantity of special nuclear materials inside SNFs must be estimated with the highest achievable precision. Unlike aqueous reprocessing, pyro-processing involves handling input materials in a solid state. Hence, partially extracted samples analyzed by destructive assay (DA) should maintain an acceptable level of representativeness. In this study, a representative sampling method widely applied in the pharmaceutical industry was adopted for homogenization in the head-end process of pyro-processing. By employing representative sampling, specifically based on the mechanism of the rotary riffler, the overall process of homogenization prior to DA analysis was simplified, and less probable hold-up that could contribute to materials unaccounted for (MUF) would be expected. The resulting Pu sampling uncertainty was confirmed to be less than 1% (for ≥ 1,000 μm particle size and ≤ 5 kg sample mass), ensuring sufficient control of Pu accounting uncertainty at a reasonably low level (≤ 1%). Thus, representative sampling can be a competitive alternative to previously suggested methodologies.

Spent filters contained in drums of radioactive waste generated from nuclear power plants are contaminated with various radioactive isotopes due to their use in various water purification processes in the system. Radiation doses from the spent filters can vary from low to high levels. To dispose of drums containing spent filters as radioactive waste, the inventory of radioactive isotopes in the filters must be determined. Two methods for determining the inventory are indirect measurement using scaling factors and direct analysis of filter samples. This study suggests a method to determine the appropriate sample size for each drum based on the number of filters stored in the drum, when direct analysis is used to determine the inventory of radioactive isotopes. In particular, Visual Sample Plan (PNNL) software’s Item Sampling function was used to calculate the sample size, considering the confidence level and minimum acceptable coverage rate. As a result, assuming that the number of filters packed per drum ranges from a minimum of 1 to a maximum of 30, the study suggests that a full inspection is required for drums containing 9 or fewer filters, while drums containing 10 filters should be sampled with 9 samples, 11 filters with 10 samples, 12-13 filters with 11 samples, 14-16 filters with 12 samples, 19-22 filters with 14 samples, 23-26 filters with 15 samples, and 27-30 filters with 16 samples.