Comparing JEFF-4.0 and legacy JEFF libraries in fuel depletion benchmark studies

Accurate nuclear data are essential for reliable predictions of spent nuclear fuel composition. This work evaluates the performance of the recently released JEFF-4.0 nuclear data library in UO2 fuel depletion calculations, focusing on its predictive accuracy for actinide and fission-product inventories. Calculations were performed with the Monte Carlo burnup code SERPENT 2.2 for two representative test cases: a simplified TMI-1 PWR pin-cell model and the REBUS experimental benchmark. The results obtained with JEFF-4.0 were compared against previous JEFF releases (JEFF-3.1.2 and JEFF-3.3) and major ENDF/B evaluations (ENDF/B-VII.1, ENDF/B-VIII.0, and ENDF/B-VIII.1). Compared to previous JEFF releases, JEFF-4.0 shows improved consistency and close agreement with experimental data in this case study for several key nuclides, including burnup indicators 137Cs and 148Nd. Nonetheless, systematic deviations persist for specific fission products and minor actinides, indicating areas where further evaluation refinements may be needed. A statistical normalization method was applied to quantify and visualize discrepancies among isotopic predictions from different libraries, allowing a model-independent assessment of trends across libraries. The methodology developed here demonstrates that standardized comparisons of predicted inventories across nuclear data libraries, even when applied to mock-up pin-cell calculations, provide a practical diagnostic tool to identify and interpret discrepancies prior to experimental validation.