Demonstrating a Quantitative and Systematic Approach to Reducing Excess Conservativism in Nuclear Criticality Safety Analyses

Although reducing conservatism would alleviate unnecessary constraints in processing, storage, transportation, and disposal of nuclear materials, excessively conservative approaches are still utilized in many safety analyses. Criticality safety limits are put in place to reduce the likelihood of having a nuclear criticality accident to a value that is deemed incredible but often utilize parameters that are conservative to the point of becoming incredible themselves. The analyses that determine criticality limits are supposed to be based on credible instead of incredible events and circumstance, highlighting the need to be able to distinguish between what is in the realm of possibility and what is not. This paper provides a quantitative approach for reducing unrealistically conservative parameters by recalculating limiting factors in a state that deviates from the worst-case scenario and assigning probability distributions to these systematic deviations. This provides a technical basis for replacing excessively conservative values with something that is both objective and reasonably bounding, which may be systematically utilized in any criticality safety analyses. The assumption of “perfect sphericity” in the TRUPACT-II package’s fissile contents model was used as an example case to demonstrate the proposed approach for replacing qualitative reductions to conservatism with quantitative reductions. Through a series of Monte Carlo calculations and statistical analyses, it was shown that conservative deviations from sphericity will provide lower k_eff values, where the magnitude and impact of this deviation is system specific. The statistical significance from applying probabilistic conservatism will be dependent on the chosen κ value and integration limit for the exponential distribution, as it varies the degree of conservatism applied to any parameter of interest. This approach is not limited to geometric assumptions and may be applied to a variety of conservative parameters. In an effort to move toward a standard method for reducing conservatism, this objective approach may be used in lieu of or in conjunction with subjective methods for relaxing constraints.