The Calibration of the Sellafield Drypac Plant Cf Shuffler

The Sellafield Drypac Plant (SDP) is a new facility for the treatment of non-contact handling Intermediate Level Waste (ILW) prior to long term storage. Waste from MAGNOX (low burn-up, natural uranium fuel) reprocessing activities is recovered in 200-l drums, dried and super-compacted into “pucks”, which are placed in concrete lined 500-l storage drums. The fissile content of each drum must not exceed a criticality limit set by the regulatory requirements. The plutonium content is measured by the Passive Neutron Coincidence Counting (PNCC) technique, which makes use of a plant supplied Pu/240Pueff ratio and a measured Cf Add a Source (AAS) matrix correction. Because the SDP may accept waste from waste vaults, which might also have contained enriched uranium, it is necessary in such cases to supplement the PNCC measurements with a direct total fissile assay. For this reason one of the two Passive Neutron Monitors (PNM’s) has been fitted with an integral californium shuffler. In this paper we describe the experimental calibration of the SDP Cf shuffler using a 500-l drum of simulated corroded MAGNOX fuel cladding. Measurements were made using a stringer of four plates of Highly Enriched Uranium (HEU) in various radial, annular and vertical positions within the drum. Emphasis is given to the interpretation of the experimental data in order to derive a volume averaged absolute calibration coefficient (counts.s-1.235Ueq.g-1) along with a justifiable uncertainty budget. The method of estimating the self-attenuation correction in the HEU calibration plates is described in detail. Two methods used in reducing the spatial response data are also described. The first method is the conventional weighting factor approach in which each individual measurement point is scaled according to an effective volume assigned to it. The weighted summed over all points yields the Volume Weighted Average (VWA) calibration coefficient. In the second approach, the discrete measurement points are smoothly interpolated and spatial integration is performed explicitly. The difference between the two approaches is an empirical way of assessing the small systematic uncertainty associated with the calibration that purely arises as a result of the procedure used to process the experimental data.