The proposed project intends to translate research to practice through rigorous and efficient methodologies and tools to assess seismic safety of NPP. It also has the aim to innovate current practice by supporting simulation results with experimental data and experience feedback in the framework of Bayesian approaches and machine learning. The research will develop methods to improve the predictability of (non linear, best-estimate) beyond design analyses (design extension earthquakes). The refined seismic PSA provides meaningful support in the decision making process and could be useful for real time expertise of plant safety in case of temporary unavailability of safety relevant equipment or structures. It is also proposed to develop efficient tools to identify major contributors to risk such that efforts to increase safety and resistance are focused on relevant equipment. The outcome will thus increase the reliability of the analyses and in turn increase confidence in the probabilistic and deterministic safety assessment results. The results of this project will then help nuclear operators in their periodic safety reviews and to respond to the high-level EU-wide safety objectives of the amended EURATOM nuclear safety directive (stress tests). The considered accident scenarios will provide input for updating severe accident management guidelines (SAMG).