The international fusion materials irradiation facility (IFMIF) is aimed to provide an intense neutron source by a high current deuteron linear accelerator and a high-speed lithium flow target, for testing candidate materials for future fusion reactors. An activity aimed at the safety assessment of the IFMIF plant as a whole has been carried out applying the probabilistic risk assessment (PRA) approach to identify and quantify in terms of expected frequencies, the dominant accident sequences related to the plant operation, and define the reference accident scenarios to be further analyzed through deterministic transient analysis, in order to verify the fulfilment of the safety criteria. The accident sequences have been modeled through the event tree technique, which allows identifying all possible combinations of success or failure of the safety systems in responding to a selection of initiating events. The identification of accident initiators, provided by the failure mode and effect analysis (FMEA) procedure, is followed by the systems analysis based on fault tree technique, for the unavailability assessment of the safety systems: finally the accident sequence scenarios are assessed by RISK SPECTRUM software. The study has allowed for the development of all accident sequences resulting from selected initiators relative to IFMIF plant and their grouping within sequence families, denoted as plant damage states, on account of the plant response and expected consequences. The frequency assigned to each family sequence is the sum of the contributors relative to all sequences ending into that particular plant state. The outcome of the analysis shows that IFMIF plant is quite safe and presents no significant hazard to the environment: in fact all the sequences implying potential undesired effects as radioactive release to the outside, show very low frequencies, well below the limit for credible accident (1.0E−6/year). In addition, due to the novelty of the design and the large spreading assigned to the failure parameter probabilistic distributions (data utilized in the probabilistic analysis of this one of a kind plant are largely of a generic nature), an uncertainty analysis has been performed to add credit to the model quantification and to assess if the sequences have been correctly evaluated on the probability standpoint.
Read full abstract