48 PSA Level 2 — NPP Ringhals 2 (PSAM-0156)

A comprehensive PSA Level 1 and Level 2 has been performed from 2001 to 2004. After this there has been some minor updates by RAB. In this paper the Level 2 work and results will be presented.

The Fault-tree (RiskSpectrum) model is an integrated level 1 and level 2 model. Level 1 event tree sequences were grouped into particular plant damage states (PDS) for which accident progression is similar based on the information available from level 1. Emphasis is on the consideration of those functions in the level 1 event that significantly influence the quantification of the containment event trees (CET). Phenomenological issues, which depend on parameters already known from level 1 were considered in the grouping process, such as pressure in the reactor coolant system and atmospheric condition in the containment. In particular, special PDSs were defined to distinguish containment bypass sequences, as SGTR and interfacing system LOCA (V-LOCA), from non-by-pass-sequences.

The accident progression of core melt is modelled in PDS specific containment event trees (CET). The end states of the CETs represent directly release categories (RC). 19 RCs were defined to characterise the fission product retention potential of the plant, ranking from low retention and thus high potential source term to high retention and thus low potential source term. With the CET analysis the key events that determine the accident progression had been evaluated: isolation of the containment, depressurisation of the RCS (including subsequent events as core recovery, induced SGTR and passive failure of the reactor coolant system), in-vessel steam explosion, hydrogen combustion with failure of the containment (repeatedly questioned in different phases of the accident), failure of the containment due to high pressure breach of the RPV (DCH and rocket mode failure), ex-vessel steam explosion, melt concrete interaction and long-term pressurisation. Operator actions have been considered according to the Ringhals accident management guidelines. The CET questions had been answered using fault trees providing the link to level 1 information (system availability, likelihood for operator action) and dedicated FORTRAN codes providing the link to accident progression information for relevant sequences (period for operator action, amount of hydrogen etc) derived from the MAAP4 calculation. When quantifiying the CET with the RiskSpectrum code a reasonable uncertainty range for the branch probability had been assumed.

The source terms had been calculated with MAAP4. Dedicated modelling had been performed for the steam generator secondary side and the auxiliary building. For each release category a MAAP4 calculation based on a representative accident sequence was performed. In addition, sensitivity calculations were carried out to account for the influence of other accident sequences and of MAAP4 uncertainties.

An important result of the level 2 PSA is the frequency for large (> 0.1% of the core inventory) early (in a time scale where no evacuation of the population is possible) release. Nine RC contribute to “large early release”: containment rapture at different moments (before, at, after RPV failure), unisolated SGTR RCs, large interfacing system LOCA, pre-accidental containment leak.

Using the product of the source term, in terms of Bq of a particular nuclide, and the frequency of the release category as a risk measure for this nuclide a ranking of the contribution of the different release categories to the “source term risk” is possible. The highest risk contribution results from containment bypass via SGTR, followed by containment rupture and containment bypass via V-LOCA. Intact containment results, as expected, in the lowest risk contribution. The contribution by basemat penetration is marginal.