• Study of a Feynman-α experiment with a Cf-252 source and He-3 detector, using measurements and MCNP6.2 simulations. • The one-region Feynman model fails to fit the data, confirmed by LOOCV cross-validation. • Model error is introduced using Bayesian inference and represented with a Gaussian autocorrelation function. • The improved fit preserves Y ∞ while shifting the time constant to the neutron lifetime in the detector. Within the framework of point kinetics, the Feynman model is utilized to fit Y-statistic curves, yielding estimations of the model’s optimal parameters. The context of neutron source characterization establishes experimental configurations that deviate from the point kinetics assumptions, resulting in discrepancies between observations and model predictions. A proposed methodology to address this issue involves the modeling of these discrepancies through the implementation of a Gaussian process, which serves to represent the model inadequacy. This paper presents a calibration of the Feynman model in a simple experimental configuration, introducing model error into the calibration process via Bayesian inference. It is corroborated by a simulation-based study using MCNP6.2 to reproduce the detection timeline of the experiment. The Bayesian calibration has the potential to enhance the prompt neutron decay constant estimation, as the correction term effectively handles the effects of the scattering environment, thereby reflecting the main behavior of the neutron fluctuations.
Chartier et al. (Fri,) studied this question.