Radiation-induced microstructure evolution in UO2, including radiation swelling and fission gas release, drastically affects fuel performance and can eventually result in fuel failure. Traditional fuel performance codes consider these microstructure effects using empirical fits to experimental data. While these empirical models accurately predict material behavior within tested conditions, they cannot predict the behavior outside of those conditions. In order for a fuel performance code to be predictive in a range of operating conditions, it must consider atomistic and microstructure effects. Here, we present the multiscale methodology that is the basis of the MOOSE-BISON-MARMOT suite of fuel performance codes under development at Idaho National Laboratory. MOOSE, the Multiphysics Object-Oriented Simulation Environment, is the basis of the MARMOT fuel microstructure code and the BISON fuel performance code. Here, we show how atomistic simulations are used to develop quantitative models of the effect of radiation damage on microstructure in MARMOT. From these models, mechanistic material constitutive models are developed for the use with BISON. This approach is demonstrated by developing a model of the impact of grain boundary (GB) fission gas bubbles on the effective GB thermal resistance.