Improved performances of the fast reactor calculational system ERANOS-ERALIB1 due to improved a priori nuclear data and consideration of additional specific integral data

A single consistent scheme of calculational methods and nuclear data called ERANOS-ERALIB1 was produced in 1996 to calculate fast reactor neutronic parameters. It represents a significant improvement on previous schemes such as CARNAVAL-IV, PROPANE and VASCO, each of which were required in order to calculate one specific application. The nuclear data library ERALIB1 has been obtained by a consistent statistical adjustment based on 355 integral data from 71 different systems. The performance of ERALIB1 is excellent, as demonstrated during its validation for which all the k_eff SUPER-PHENIX data were reproduced to within 70 pcm.

The only restriction on this satisfactory performance is related to the rather poor prediction of the sodium void reactivity effect. This was due to very bad nuclear data for ²³Na, and the unsatisfactory methods used to calculate the sensitivity coefficients for the sodium void reactivity variation Δρ_Na. To improve the performance relative to this point and to enlarge the domain of validation several actions have been undertaken:

•a revision of the formalism and algorithms used to calculate the derivatives of Δρ_Na to the sodium cross section data,

•a significant enlargement of the integral data base related to this aspect of the sodium void effect. Compared to the initial data base established in support of ERALIB1, several additional (18) sodium void configurations corresponding to voids of different volumes at different core locations have been studied.

In order to broaden the range of application of the improved library, which will be called ERALIB1.A, significant effort has been devoted to additional configurations which have firstly been evaluated, and then if judged suitable, included in the adjustment process. They are related to two specifically targeted experimental programmes:

•a study of neutron deep penetration. Several configurations of the JANUS experimental programme (shielding constructed of separate steel, iron and sodium plates) have been analysed. With this complementary information ERALIB 1A becomes applicable for accurate predictions of shielding configurations,

•a study of steel reflectors for a fast reactor of the SUPER-PHENIX type. The measurements performed in the MASURCA (the CIRANO experimental programme) and FCA facilities include spectral indices (F25(r)/F25(0),...) at different positions in the reflector. As a consequence of these measurements, important information has been obtained for additional “secondary” structural material isotopes, such as ⁵⁷Fe, ⁶⁰Ni and ⁵³Cr.

Significant effort has also been devoted to the analysis of 29 β_eff experiments. The result of this is an improvement of the uncertainty on ν_d(E) which guarantees a prediction of β_eff with the required accuracy (3% for critical configurations, and 5% for power reactors).

The consistent statistical adjustment method by Gandini et al. (1973) has been completed. Rigorous criteria have been introduced to identify any data which are suspect and/or inconsistent in the integral data base. These data may introduce additional bias in the adjusted library, and for that reason they must be discarded before adjustment.