The impact of nuclear data uncertainty on identifying plutonium diversion in liquid-fueled Molten Salt Reactors

Molten salt-fueled reactors (MSRs) have gained popularity recently, but there are challenges in accounting for nuclear material due to the inability to count discrete items (e.g., fuel assemblies). The focus of this work is to calculate what changes would occur due to diversion of fissile material, and to evaluate the effect of nuclear data uncertainties on these quantities. The impact of other sources of uncertainty (e.g., measurement error) is out of the scope. A thermal-spectrum MSR was modeled using Serpent 2 and SCALE 6.3.b12 (beta), including plutonium diversion scenarios. The feed and removal capabilities were recently added into the depletion schemes of these codes. Protracted and abrupt diversion scenarios were developed for the removal of 1 and 10 significant quantities (SQs) of plutonium. The SCALE module Sampler was used to perform nuclear data uncertainty propagation along the depletion interval from neutron cross-sections, fission product yield and decay data. Results showed that along with plutonium species, other actinides and fission products presented significant changes between the reference scenario (i.e., no diversion) and the 10 SQ plutonium protracted diversion scenario. Considering their propagated nuclear data uncertainty from Sampler simulations, some actinides such Am and Cm showed changes higher than their nuclear data uncertainty, which goes from 2.23% for ²⁴¹Am up to 4.88% for ^242mAm. Some fission products also presented notable changes, Sr and Y isotopes are examples with positive changes, while Cd, Eu and Sm isotopes showed more prominent changes on the negative side, with uncertainties in the range of 1.89% for ¹⁴⁹Sm to 3.31% for ¹¹³Cd. The analysis extended to nuclides regularly moved to other modeled inventories, showing that ^105,106Ru can be good candidates given their low uncertainty that stands within the 1% range. Many other isotopes with considerable changes presented high uncertainty values and methods to improve their nuclear data uncertainty estimation will be sought in future work.