Use of the nuclear data in the reactor dosimetry: limitations, improvements and perspectivesC. Destouches1, C. Domergue1, D. Beretz1, G. Gregoire1 and B. Habert2
1 Commissariat à l'Énergie Atomique, Centre de Cadarache, DER/SPEx/LDCI, 13108 Saint-Paul-lez-Durance, France
2 Commissariat à l'Énergie Atomique, Centre de Cadarache, DER/SPRC/LEPH, 13108 Saint-Paul-lez-Durance, France
Published online: 21 May 2008
One of the main objectives of reactor dosimetry is the determination of the physical parameters characterizing the neutron field in a specific place. Reaction rates and fluence rates for example, are derived from activity measurements of irradiated samples using a neutron interpretation process, which needs the knowledge of appropriate nuclear cross sections. For reactor dosimetry, activation cross sections come from international dosimetry activation files, IRDF, EAF, themselves retrievals from international nuclear libraries as ENDF, JEFF, JENDL... These nuclear data are used not only for the activity measurement and the reaction rate derivation (mean values) but also to perform spectrum unfolding and to assess output uncertainties (covariance). Thus, their accurate knowledge is a significant stake for reactor dosimetry improvement, especially for the industrial applications as shows the high uncertainty value of 15% (k = 1) commonly encountered for the calculated neutron flux (E > 1 MeV) on the vessel and internal structures of a power reactor. After a schematic presentation of the reactor dosimetry process, each step, activity measurement, reaction rate derivation and spectrum unfolding, is analysed in order to point out where and how the nuclear data are used. A critical review of the available nuclear data drawn from the new release of the international dosimetry file IRDF2002 is made and improvements and lacks, from a reactor dosimetry point of view, are listed. Then the benchmark concept is analysed in terms of representativeness and use in the dosimetry interpretation: How to use a cross section, which has a C/E significantly different from unity? Finally, the paper concludes with a synthesis of the needs in evaluation improvements for nuclear data in terms of new cross sections estimations, covariance matrices improvements, and experimental benchmark database extension.
© CEA 2008