Global and flexible models for Sodium-cooled Fast Reactors in fuel cycle simulations. (June 2019)

• Record Type: Journal Article
• Authors: Ernoult, M.; Doligez, X.; Thiollière, N.; Zakari-Issoufou, A.A.; Bidaud, A.; Bouneau, S.; Clavel, J.B.; Courtin, F.; David, S.; Somaini, A.
• Is Part Of: Annals of nuclear energy. Volume 128(2019)
• Page Start: 69
• Page End: 76
• Publication Date: 2019-06
• Subjects: Sodium-cooled Fast Reactor -- Fuel cycle -- Breeding ratio -- Meta-model; Nuclear energy -- Periodicals; Nuclear engineering -- Periodicals; 621.4805
• DOI: 10.1016/j.anucene.2018.12.037
• Languages: English
• ISSNs: 0306-4549

Abstract

Highlights: New flexible model for a wide range of Sodium cooled Fast Reactors inside CLASS. Simplifications options lead to bias lower than 1% on actinides composition. Precision of the meta-model better than 4% on major actinides in spent fuel. Breeding ratios of designs covered go from 0.88 (burner) to 1.41 (breeder). Biggest impact on breeding from blankets, major impact from isotopic composition. Abstract: Since Sodium cooled Fast Reactors are present in many scenarios and strategies for the future of nuclear energy while not having a specific design established yet, we created a new fast and flexible model for the dynamic fuel cycle simulation tools CLASS. It includes a depletion meta-model and a fuel loading method based on artificial neural networks. It is able to represent a wide range of Sodium cooled Fast Reactor designs using oxide fuels and a wide range of fuel management strategies within fuel cycle simulation tools. A comprehensive analysis of simplification options has been made in order to choose the right level of complexity for the reference full core depletion calculations performed with the MURE code used for the training of the meta-model. The process from these reference calculations to the final meta-model is explained and a specific focus is given to the operations going from detailed full core depletion results to global results suitable for neural networks training. Details on the creation process for neural networks based predictors, one for each average cross-section, and their training on full core depletion calculations are given as well as the implementation within the CLASS code. The irradiation meta-model achieves good precision on all major and minor actinides present in spent fuel. The designs and loaded fuel covered by the model allow significant burner to strong breeder strategies. A sensitivity analysis shows that the number of fertile blankets is the primary contributor for breeding capabilities, but effects of isotopic composition are also significant. A test scenario illustrates the model capacity to simulate burner and breeder designs.