Multi-unit (MU) nuclear power plants (NPPs) share safety features. As these may degrade in time, we propose a framework of an integrated deterministic and probabilistic safety assessment to consider their time-dependent fragility. Physics-of-failure models of the fragility of the shared safety features are embedded into a dynamic model of the accident scenario evolution, whose uncertainties are accounted for by Monte Carlo simulation. We apply the proposed framework to a case study concerning a MU NPP composed of four small modular dual fluid reactors sharing the emergency core cooling systems and the emergency diesel generators. The results show the effectiveness of the framework in providing a dynamic risk picture that accounts for the degradation of shared safety features.