Multi-scale Models and Simulations of Nuclear Fuels

In nuclear reactors, severe radiation environments continuously alter thermo-mechanical and chemical properties of nuclear fuel materials (actinide based alloys and ceramics) [1]. The physics and chemistry of such materials increase in complexity due to irradiation effects. To address these issues, several projects have been developed all over the world to assess the properties of multi-component actinide based ceramics. However, most assessments still involve fitting of data followed by extrapolations or interpolations into new temperature, composition or pressure regimes. One of the most challenging aspects of developing a comprehensive understanding of nuclear fuels is their complex, evolving composition. The study of multicomponent systems containing U, Np, Pu, Am, Cm and their oxides, nitrides, and alloys, is considerably complicated by the presence of fission products such as Xe, Cs, Sr, He, I, and Tc. Most commercial reactor and fuel performance codes assume the fuel to be a homogenous material and average the properties over the computational domain. By including the heterogeneous character of the fuels, the precision and accuracy of predictions can be significantly improved. This is even more important for simulations of dispersed, inert matrix, and micro(nano)structured fuels. Most of the theoretical and computational results published over the last few decades refer to actinide based oxides; much less attention was paid to nitrides and carbides, and very little to metallic fuels. That is a reflection of the overwhelming presence of oxide fuels in existing nuclear reactors. However, part of the new generation of fast nuclear reactors, as well as the reactors for space propulsion, require non-oxide fuels. In this work we submit that a multi-physics approach to developing a fundamental understanding of properties of complex nuclear fuel materials in the reactor environment, will lead to improved tools for predicting phenomena such as heat transfer, phase stability, species diffusion, and fission products retention. Recently, combined theoretical, computational, and experimental efforts have provided valuable information about materials properties and important phenomena associated with nuclear fuels. However, as of today, models and simulations are not regarded as critical tools for fuel design and optimization. One of the reasons is the lack of predictability, often associated with the empirical correlations used in computational the models. The applicability of such correlations is limited to a regime Theory-based models and high performance simulations are briefly reviewed starting with atomistic methods, such as Electronic Structure calculations, Molecular Dynamics, and Monte Carlo, continuing with meso-scale methods, such as Dislocation Dynamics and Phase Field, and ending with continuum methods that include Finite Element and Finite Volume. Special attention is paid to relating thermo-mechanical and chemical properties of the fuel to reactor parameters. By inserting atomistic models of point defects into continuum thermo-chemical calculations, a model of oxygen diffusivity in UO2+x is developed and used to predict point defect concentrations, oxygen diffusivity, and fuel stoichiometry at various temperatures and oxygen pressures. The simulations of coupled heat transfer and species diffusion demonstrate that including the dependence of thermal conductivity and density on composition can lead to changes in the calculated centerline temperature and thermal expansion displacements that exceed 5%. A review of advanced nuclear fuel performance codes reveals that the many codes are too dedicated to specific fuel forms and make excessive use of empirical correlations in describing properties of materials. The paper ends with a review of international collaborations and a list of lessons learned that includes the importance of education in creating a large pool of experts to cover all necessary theoretical, experimental, and computational tasks.