EDIM Local Power History and Burnup Inference Based on Destructive Assay Data

This study proposes a method to infer local burnup and power history for a measured fuel sample by leveraging correlations between nuclide concentrations and power history. Accurate modeling of fuel samples with available radiochemical assay (RCA) measurements is essential for code validation and bias prediction in spent-fuel nuclide inventory calculations. Our method aims to reconstruct a sample’s power history and offers an alternative to sample power calibration based on the ¹⁴⁸Nd burnup indicator fission product. To accomplish this, as part of our entropy-based debiasing inference methodology, we developed a perturbation and inference procedure using a power history decomposition method. The procedure represents power as a time series broken down into multiple fundamental components, termed power modes. We then used SCALE/Polarissimulation data to establish correlations between the discharged fuel isotopics and the power modes, which allowed us to explore the inherent correlations within the simulation cloud. Additionally, the method identified burnup indicators beyond traditional ¹⁴⁸Nd and ¹³⁷Cs that can be utilized to adjust sample power in burnup calculations. For demonstration, we modeled a fuel sample selected from a Three Mile Island Unit 1 spent fuel rod, for which RCA measurements were performed at Oak Ridge National Laboratory and the measurement data are publicly available. A simulation cloud of predicted nuclide concentrations was generated by perturbing the decomposed power modes. The simulation cloud was then used to reconstruct the power history and to provide a more accurate sample burnup estimate. The nuclide concentrations derived from the improved model were compared with the corresponding measured values.