A sensitivity study of support system modeling

In published probabilistic risk assessments, two principal approaches have been used to evaluate the frequencies of accident sequences. The first method uses large fault-tree models of the frontline systems and includes the support systems indirectly through fault-tree linking. The other approach delineates the support systems explicitly in the event trees. The industry degraded core rulemaking individual plant evaluation methodology uses the second approach to avoid the need to use complex computer codes to link large fault trees. In principle, both approaches should yield the same results. In practice, because of the complexities of the systems, approximations have to be made. This paper presents a sensitivity study of two approximations commonly made in the support state approach: (1) modeling on a system level versus train level for support systems and (2) including only the key support systems versus all support systems