Consistent quantification of energy uncertainties in experimental time-of-flight nuclear data across resolved, unresolved, and fast regimes

Abstract

A major task for nuclear data evaluations is the curation of measurement data to be used in the inference of nuclear data quantities. The quality of the resulting measurement data uncertainty depends on the rigor of the evaluator reviewing the data set and the uncertainty information provided by the experimenter. Nuclear data in the fast and resonance regions have fundamental differences which has led to separate experimental UQ and evaluation methodologies, despite the fact that some measurements often span both regions. Uncertainties associated with the time-of-flight technique are particularly disjoint as a model of the underlying nuclear data quantity is needed to propagate uncertainty in the independent variable (energy) to an effective uncertainty/covariance in the dependent quantity (cross section or other nuclear data quantity). This frequently leads to inconsistent, incorrect, or missing covariance components. Such deficiencies propagate into key application domains, where uncertainties can influence safety, design, and operational decision-making. This article explores the misalignment that arises from different assumptions made across evaluation regimes, offers a consistent set of terminology, and demonstrates the necessary approach for consistent quantification of time-of-flight uncertainties across both fast and resonance regions. The approach requires the evaluator to adopt a unified experimental model, with the principal inconsistency between the two regimes arising from the treatment of experimental resolution. Consistently applying a unified experimental model becomes challenging when fluctuations from unresolved resonances are present, due to the lack of an explicit fluctuating cross section representation. To address this, two strategies are suggested: further averaging or smoothing of the experimental data to mitigate fluctuations, or employing statistical resonance models to estimate an effective derivative for use in time-of-flight uncertainty propagation.