The objective of the present document is to provide guidance on the implementation of external events into an “extended” L2 PSA. It has to be noted that L2 PSA addresses issues beginning with fuel degradation and ending with the release of radionuclides into the environment. Therefore, the present document may touch upon, but does not evaluate explicitly issues that involve events or phenomena which occur before the fuel begins to degrade.
Following the accident at Fukushima Dai-ichi, the nuclear safety community has realized that much attention should be given to the areas of operator interventions and accidents that may develop at the same time in more than one unit if they are initiated by one or more common external events. For this reason and to fulfill the PSA end-users’ wish list (as reflected by an ASAMPSA_E survey), the attention is mostly focused on interface between L1 and L2 PSA, fragility analysis, human response analysis and some consideration is given to L2 PSA modeling of severe accidents for multiple unit sites, even though it is premature to provide extensive guidance in this area.
The following recommendations, mentioned in various sections within this document, are summarized here:
Vulnerability/fragility analyses should be performed with respect to all external hazards and all structures, systems and components potentially affected that could be relevant to L2 PSA,
Importance should be given to the assessment of human performance following extreme external events; for extreme circumstances with high stress level, low confidence is justified for SAM human interventions and for such conditions, human interventions could be analyzed as sensitivity cases only in L2 PSA,
Results presentation should include assessment of total risk measures compared with risk targets able to assess all contributions to the risk and to judge properly the safety (see document  for recommendations on PSA results presentation),
Total risk measures shall be associated to appropriate information on all uncertainties, simplifications and conservatisms that appear today to be inherent to any extended PSA,
Because NPPs on multi-units sites are in general not fully independent, verification and reassessment of current single PSAs is needed before developing multi-units PSA,
Because established methodologies for multi-unit sites L1-L2 PSA analysis are not yet available (even if multi-unit sites L1-L2 PSA analysis are now on-going in some countries), it is recommended to use first a simplified method. The boundary between L1 and L2 PSA shall be defined appropriately and some relevant adaptations/simplifications in both L1 and L2 PSA may be considered (in a first step) to limit the complexity of the multi-unit sites L1-L2 PSA development.