American Nuclear Society
Home

Home / Publications / Journals / Nuclear Science and Engineering / Volume 194 / Number 1

Derivation of the Two-Exponential Probability Density Function for Rossi-Alpha Measurements of Reflected Assemblies and Validation for the Special Case of Shielded Measurements

Michael Y. Hua, Jesson D. Hutchinson, George E. McKenzie, Tony H. Shin, Shaun D. Clarke, Sara A. Pozzi

Nuclear Science and Engineering / Volume 194 / Number 1 / January 2020 / Pages 56-68

Technical Paper / dx.doi.org/10.1080/00295639.2019.1654327

Received:March 4, 2019
Accepted:August 6, 2019
Published:December 13, 2019

Rossi-alpha measurements of fissionable assemblies are used to estimate the prompt neutron decay constant α. Reactivity can be inferred from α if the values of the neutron generation time and effective delayed neutron fraction are assumed. If multiple measurements are performed on an assembly near delayed critical, one can determine α at delayed critical and directly infer reactivity (without needing to assume values for the neutron generation time or effective delayed neutron fraction). Previous works have demonstrated that two-exponential fits for Rossi-alpha measurements of reflected assemblies have better fit metrics than those of one-exponential fits; however, the two-exponential probability density function that is needed to obtain α from the fit parameters has not been derived. This paper derives the two-exponential fit based on a two-region point kinetics model for Rossi-alpha measurements of reflected assemblies, a generalization of the current, one-region model (one-exponential fit). The new model is validated for shielded assemblies, a special case of reflected assemblies where the reflector-to-core leakage is negligibly small. The validation is performed using shielded, fissionable assemblies (highly enriched uranium with keff ≈ 0.95 and weapons-grade plutonium with keff > 0.77). The results show that the two-exponential model can (1) predict the constant α within two standard deviations, and (2) deconvolve α and the time a neutron spends in the reflector region, neither of which is possible with the one-exponential model.