Assessment of the impact of nuclear data uncertainties on the core design of ALFRED

In a context of concerns for climate changes mainly linked to the consumption of fossil fuels, nuclear energy could convince of the opportunities that it offers only following a line of lasting and highly sustainable development. In the European context, Italy holds a technological leadership in Lead-cooled Fast Reactor (LFR) systems, which represent a sizeable promise to fulfil the long-term development goals of the so-called Generation IV (GEN IV). With this aim, Italian industry and research pursue all the activities required to support the construction of the Advanced Lead-cooled Fast Reactor European Demonstrator (ALFRED) reactor. During the last decades, the continued efforts in designing fast nuclear reactors raised, among others, the need for more performing and accurate computational tools. Looking to the neutronic design of the reactor core, this requires fine numerical solutions based on complex calculation schemes, as well as precise and accurate nuclear data evaluations, with the former and the latter concurring synergistically in an inseparable binomial. Generally speaking, all the computational tools and nuclear data used in designing current power reactors, have achieved a high degree of reliability, aiming at constantly improving precision and accuracy of calculated integral parameters, according to a policy more focused on safety and sustainability. This achievement is also accomplished thanks to the availability of real-case data from operating power reactors, providing extremely representative information for any calibration purposes. However, this is not the case for innovative reactor concepts (such as LFR systems), for which the nuclear data adjustment technique, proposed in the past and already applied in fast reactor research, can be successfully used for filling this gap. In this PhD thesis, a comprehensive sensitivity and uncertainty analysis for the key safety relevant parameters of the ALFRED reactor is carried out, with the twofold objective of gathering a physical insight into, and identifying the nuclear reactions contributing the most to, such parameters and their combined uncertainties. Leveraging this, a database of well suited integral experiments, conducted on past facilities included in the IRPhE database, is created by selecting the ones specifically representative of the reference reactor (ALFRED, in this work) with respect to the parameters of interest (e.g., reaction rates ratio, spectral indices, etc.), by means of a detailed representativeness analysis. With these preliminary elements, and in order to enable the calculation of any integral parame ter for the ALFRED reactor with the best possible reliability, the above-mentioned adjustment method is applied. This method uses the knowledge on integral quantities stemming from the selected experiments, paired to a nuclear data set, to calibrate (or adjust indeed) the latter for increasing its predictive capabilities when used in a given neutronic code and in a particular application domain. In this work, in order to practically apply the adjustment method, a new version of the AMARA code is also developed. The key merit of the new version, named AMARA+, in extending the former one, is chiefly due to the inclusion of an algorithm which, by checking the statistical coherence of the data set at hand, improves the quality of the adjustment being performed by disregarding any element of the original set which would scatter, rather than focus, the physical information required for an optimized adjustment. Additionally, AMARA+ adds modernization to the former version, by implementing enhanced input capabilities to manage all quantities requested for any generic problem (e.g., sensitivities and uncertainties), flexibly allowing any user-defined selection of integral experiments included in the representative database of the system of interest. Moreover, with the aim of performing sound and coherent library adjustments using the most up-to-date nuclear data evaluations, specific processes which generate, from the most recent evaluations, libraries and their relative covariances in a format useful for the chosen neutronic code (ERANOS, in this case) are also developed in this work, and applied to the ENDF/B-VIII.0 library for the specific problem here solved. The library adjustment performed by means of AMARA+ predicts the relative corrections which are requested to nuclear data for the key isotopes and reactions of the ALFRED core in order to improve the predictive capabilities of ERANOS when applied to ALFRED-like systems. The obtained relative corrections reduce indeed the discrepancies relative to the C/E values of the involved integral quantities, when applied to the ENDF/B-VIII.0 library targeting the ALFRED reactor, supporting the need for modifications to the original evaluation to different extents, depending on the isotope-reaction. In particular, significant relative corrections are proposed for 238U fission, where a negative correction is suggested only in a limited energy range, and capture, where the correction is even more important in the above-resonance region. Both the proposed corrections suggest that further evaluations are needed for such an important isotope, and with high priority, as confirmed also by the value of the relative correction normalized to the standard deviation of the inherent uncertainties, which attains its maximum value in this case. Regarding 239Pu fission, the most important reaction in fast MOX-fueled reactors like AL FRED, the proposed relative corrections are slightly negative from thermal energies up to about 1 keV, while they become positive just after the resolved resonance region, suggesting that an overestimation of the original evaluations has to be considered. Significant positive (about 10%) corrections occur for 239Pu capture reaction, mainly in a narrow energy window in the unresolved energy range. Even though not as significant as those for 238U, the suggested corrections indicate that fission and capture of 239Pu are reactions that should rank high in a priority list for future experimental cross-section refinements. Finally, the relative corrections proposed for the inelastic scattering cross-sections of all lead isotopes are found always positive, suggesting a systematic underestimation of the order of few percents in the ENDF/B-VIII.0 evaluation in the upper part of the energy range. Due to the importance of lead in a lead-cooled fast reactor, a further refinement of the evaluations by new experiments could be considered in a priority list.