Traditional nuclear fuel qualification is a lengthy process challenged by erratic or incomplete irradiation experimental data, leading to many unqualified fuels. In response, this paper presents an accelerated fuel qualification (AFQ) framework that integrates multiscale modeling, machine learning, and legacy data assimilation to inform specific integral testing. The framework leverages atomistic simulations to elucidate fundamental mechanisms, such as xenon diffusion and defect kinetics, which inform mechanistic models of fuel behavior. These mechanistic models are then validated against legacy experimental data, while machine learning is used to refine critical parameters, such as Xe diffusivity, and to further reduce computational uncertainties.

As a demonstration, the framework is applied to characterize uranium mononitride (UN) fuel, resulting in the quantification of swelling, which is a dominant failure mechanism, uncertainty quantification of the swelling process in UN, and the development of performance envelopes as a function of temperature, linear heat generation rate, and burnup. The AFQ methodology outlined here offers a robust proof-of-concept template for qualifying advanced nuclear fuels, supporting regulatory modernization efforts for next-generation reactor technologies.