Sensitivity analysis of rod rearrangement in criticality safety for PWR fuel assemblies under transportation accidents

This study conducted a sensitivity analysis of the criticality safety of Pressurized Water Reactor (PWR) fuel assemblies under transportation accidents using the Monte Carlo method, with a focus on the impact of non-uniform fuel rod arrangements on the neutron effective multiplication factor keff. The dataset generation process involved the following key steps: 1.Model Construction:   A simplified Monte Carlo computational model was developed based on the actual parameters of the AFA3G fuel assembly (e.g., fuel rod diameter: 8.192 mm, active zone length: 3657.6 mm). Conservative assumptions were adopted, such as setting the fuel enrichment to 5 wt% and assuming full submersion in water. To improve computational efficiency, the model omitted detailed structural features of the transport container, retaining only a single fuel assembly. 2.Data Generation Methods: Uniform and Non-Uniform Arrangement Simulations: Four methods were employed to generate fuel rod arrangement configurations: Uniform Arrangement by Blocks, Layer-by-Layer Determination, and Old/New Birdcage Deformation. Custom Python scripts were used to automatically generate input files for the calculations. Parameter Scanning: For the Uniform Arrangement by Blocks, the fuel rod pitch in the central region was adjusted. For the Layer-by-Layer Determination, the number of deformed layers and their spacing were varied. For the Old/New Birdcage Deformation, the curvature radius of the damaged nozzle deformation was adjusted to simulate non-uniform fuel rod arrangements. Statistical Settings: Each simulation generation included 10,000 initial neutrons, with a total of 1,000 generations run (800 of which were active). A single calculation took approximately 40 minutes (using an Intel i7-13700KF CPU and NVIDIA RTX 3080 GPU). 3.Errors and Missing Data: Statistical Errors: Inherent fluctuations due to the Monte Carlo method caused variations in results. 4.Data Files and Tools: Input files were generated using Python scripts. The Monte Carlo calculation program was based on the ENDF/B-VI nuclear database.