Abstract
The Digital Cherenkov Viewing Device (DCVD) [1] is a tool used by nuclear safeguards inspectors to verify irradiated nuclear fuel assemblies in wet storage based on the Cherenkov light produced by the assembly. Verifying that no rods have been substituted in the fuel, so-called partial-defect verification, is done by comparing the intensity measured with a DCVD with a predicted intensity, based on operator fuel declaration. The prediction model currently used by inspectors is based on simulations of Cherenkov light production in a BWR 8x8 geometry. This work investigates prediction models based on simulated Cherenkov light production in a BWR 8x8 and a PWR 17x17 assembly, as well as a simplified model based on a single rod in water. Cherenkov light caused by both fission product gamma and beta decays was considered. The simulations reveal that there are systematic differences between the model used by safeguards inspectors and the models described in this publication, most noticeably with respect to the fuel assembly cooling time. Consequently, if the intensity predictions are based on another fuel type than the fuel type being measured, a systematic bias in intensity with respect to burnup and cooling time is introduced. While a simplified model may be accurate enough for a set of fuel assemblies with nearly identical cooling times, the prediction models may differ systematically by up to 18 % for fuels with more varied cooling times. Accordingly, these investigations indicate that the currently used model may need to be exchanged with a set of more detailed, fuel-type specific models, in order minimize the model dependent systematic deviations.