Nuclear power plants are exposed to a variety of hazards, which may result in risks (the product of the likelihood of the hazard and resulting consequence). One of the key objectives of the NARSIS project is to improve the integration of external hazards and their consequences with existing state-of-the-art risk assessment methodologies in the industry. Accordingly, the main goals of this deliverable are to: - Review the various aspects of risk integration and associated methodologies - Review case histories of accidents in complex industrial set-ups, both nuclear and non-nuclear, and highlight prevalent ‘latent weaknesses’ that eventually led to these accidents - Review deterministic and probabilistic methods to identify latent weaknesses - Review risk integration methods currently used in high-risk industries such as nuclear, chemical and aviation - Review accident investigation procedures and international initiatives associated with major nuclear accidents - Discuss specific risk integration method(s) that are relevant to the NARSIS project@en

Eighteen academic, research and industrial European institutions from Slovenia (GEN, JSI), Croatia (APOSS), Italy (ENEA, UNIPI), France (CEA, BRGM, IRSN, EDF, Framatome – ex Areva NP), Austria (NUCCON), Poland (NCBJ, WUT), Germany (KIT, Framatome - ex. Areva), Finland (VTT), The Netherlands (TU Delft, NRG), United Kingdom (EDF Energy) formed a consortium and applied to the H2020-Euratom call. The main ambitions of the consortium are to fill some gaps identified in existing external events probabilistic safety analyses (PSA) and to improve parts of the existing methodologies by 3 points: (1) to adapt most up to date frameworks and methodologies already existing or under development outside of nuclear community; (2) to use knowledge and experience on recent national and international projects; (3) to develop demonstration cases at the real NPP scale. Interactions are envisaged with on-going international initiatives and with the International Advisory Board, which will follow and discuss the project results with the aim to propose recommendations for future regulations. The main expected results are the development of an integrated risk framework for safety analyses and the development of a decision-making tool for demonstration of nuclear facility management. The integrated risk framework consists of: • Scenarios comprising single or multiple external hazards. Hazards can be combined or 165-2 cascading and include earthquake, flooding, extreme weather and others, • The physical and functional fragilities and interdependencies between systems/equipment are taken into account, • Human factors are taken into account and may play important role during severe accidents. • A support decision-making tool will be developed to demonstrate nuclear facility management during severe accidents due to external natural events. The project is structured into seven work packages (WP): • WP1: External hazards characterisation, • WP2: Fragility assessment of main NPPs critical elements, • WP3: Integration and safety analysis, • WP4: Applying & comparing various safety assessment approaches on a virtual reactor, • WP5: Supporting tool for severe accident management, • WP6: Dissemination, recommendation, and training, • WP7: Project management and coordination. The NARSIS project started in autumn 2017, with the duration of 4 years. @en

Bayesian networks are proposed as a tool to integrate reliability and influential variables relating to the slope stability of an idealized embankment. The site investigation (extent) and slope geometry, as well as the material properties and their spatial variability, are considered within a Bayesian network. The random finite element method (RFEM) is used to quantify the slope reliability and demonstrate the overall methodology. Prior probabilities of geometry, material parameters and their heterogeneity are obtained from ‘initial’ site investigation data. Probabilistic distributions of slope performance (factor of safety) are obtained by Bayesian inference in the network to investigate the impact of additional site investigation. The amount of additional site investigation required to increase the geotechnical reliability is assessed. This work illustrates the applicability of Bayesian networks as an effective reliability and uncertainty assessment tool that can aid decision making for site investigation and during maintenance, where new observations can be readily integrated to obtain updated reliability estimates.@en