Improvement of INVS measurement uncertainty for Pu and U-Pu nitrate solution

Makino, Risa; Swinhoe, M. T.*; Suzuki, Yoshimasa ; Mukai, Yasunobu ; Menlove, H. O.*; Marlow, J. B.*; Nakamura, Hironobu 

The Inventory Verification Sample systems (INVS) is a non-destructive assay (NDA) system for small inventory samples to quantify the Pu amount in Pu and U-Pu nitrate solutions and MOX powder with 31.3% counting efficiency (for Cf). It has been used for IAEA safeguards measurement for many years at the Plutonium Conversion Development Facility (PCDF) for the samples taken at the time of Random Interim Inspection (RII) and Physical Inventory Verification (PIV), and is currently used as partial defects verification equipment with uncertainty about 35% accuracy for solution samples. The INVS is a inline detector therefore it is very convenient to use since it can be used for a sample inside the glove box without bag-out. If the measurement uncertainty can be improved (to 1%), it is expected that the usage can be extended to the operator's own measurements in MC&A to reduce effort and the number of destructive analyses (DA). In order to improve the measurement uncertainty, we have performed multiplicity measurements using a variety of solution samples (concentration, acidity and volume) to determine the Pu amount and uncertainty, although the multiplicity analysis is challenging. To control the measurement quality, a few MOX ceramic fuel pellets with known Pu amount are used. The results are evaluated by three kinds of methods (calibration curve, known- and multiplicity) to determine the best measurement method. In the range of concentration of typical samples, we could confirm good correlations between measured doubles and Pu effective mass in the three methods. Especially, the conventional calibration curve method showed a good performance with short measurement time and small uncertainty when the calibration standards were a good match to the unknown samples. In this paper, we present the INVS system design and measurement results with uncertainty using solution samples and comparison results among the three calibration methods.