Study of fuel evolution in fast reactors of the 4. generation. Impact of the nuclear data on their performance

The objective of this PhD topic is to contribute to the understanding of the variations of the core neutronic characteristics of the 4. generation reactors (Sodium Cooled Fast Reactors (SFR) and Gas Cooled Fast Reactors (GFR)) during fuel depletion. The neutron characteristics of interest are of course the burn up reactivity swing and the breeding gain but also the Doppler effect and the coolant void effect. Fuel depletion leads to a degradation of the core safety parameters. The study of these variations and their associated uncertainties contributes to justify 4. generation reactor core designs as envisaged in their last developments. These last developments concerned Sodium Cooled Fast Reactors (SFR) and Gas Cooled Fast Reactors (GFR) which were reshaped in order to meet Generation IV goals on economics, safety and reliability, sustainability and proliferation resistance. They exhibit very innovative characteristics compared to the European Fast Reactor (EFR) whose design was very much in line with those of Phenix and Super Phenix. Recent CEA studies had led to large 3600 MWth SFR cores using oxide fuel and to large 2400 MWth GFR cores using carbide fuel. Since the designs have to balance between positive breeding gain and safety characteristics such as rather low void reactivity effects (SFR) or rather small core pressure drop (GFR), scoping studies for breakthrough SFR cores were performed using dense fuels either carbide (already taken as a reference for the GFR core) or metal. These preliminary breakthrough SFR images are characterized by high power density and highly positive breeding gain (Breeding Gain = 0.17). As a first step towards the development of GFR plants, a low power experimental GFR called ALLEGRO is being envisaged and has been studied for its peculiar characteristics. To study the main neutronic characteristics of these cores, one can use analyses based on the sensitivity methods of the deterministic computer code ERANOS (neutronic code system). These methods are available in statics without the possibility of taking into account fuel depletion. In order to mitigate this insufficiency, a subsequent part of the thesis consisted in developing the depletion perturbation theory which requires to couple Boltzmann and Bateman equations and allows a more precise understanding of the behaviour of the previous cores. The method is now able to calculate the sensitivity of the actinides and fission products concentrations and of neutron characteristics of interest such as breeding gain, Doppler reactivity effect and the coolant void reactivity coefficient effect. In order to illustrate these sensitivity developments, uncertainties of the neutron characteristics have been calculated using a preliminary variance covariance matrix called BOLNA. The uncertainty analyses highlight the contribution of each isotope to the neutron characteristics of the various core designs. This determination has given relatively small uncertainty variations with burn up when possible modifications of nuclear data are applied. The in-depth study performed on sodium nuclear data evaluations (ENDFB-VII, JEFF-3.1, JENDL-3.3) highlight the difficulty of creating accurate enough nuclear data and their associated covariance matrix. It appears hence that although the feasibility of these core designs are not questioned (relatively optimistic values being calculated are within the target value of 700 pcm for the reactivity swing and 7% for the reactivity coefficient), their performance will require integral experiments both to confirm what has been evaluated with nuclear data covariance matrices and to reduce nuclear data uncertainties. Lastly, the sensitivity methods are used to explain the peculiar behaviour of integral characteristics like the void effect or the Doppler effect with depletion in the GFR and ALLEGRO cores. One reason was track back to the difference in size of the two cores but also to the different structural materials being used. Furthermore, the building up of Pu239 fission products and the change in Pu239 and Pu241 isotopes being different induce divergent behaviour of both Doppler with time. For the SFR, the distribution of the void effect in the various core zones which present different fuel depletion histories is finally analyzed to be compared to that of the power distribution and finally to that of the breeding gain. It appears that the SFR core design with a rather flat internal breeding gain has, as a consequence, a rather flat void effect which is another nice feature. One concludes on the advantages resulting from the last core designs as well as their degree of performance from the view point of computational tools very dependent at first on the nuclear data knowledge. (author)