Issues and Approaches Regarding Success Terms for Probabilistic Risk Assessment Models

Solving event tree accident sequences in probabilistic risk assessments (PRAs) involves assumptions about the success of systems, i.e., event tree top events. The primary assumption is that system failure is a rare event; therefore, the system success probability is very close to 1.0. Under most conditions, this assumption is valid. However, when systems have higher failure probabilities, the success probability is not close to 1.0 and this primary assumption causes sequences with success branches to be overestimated. This paper presents an approach to quantify event tree accident sequences when high system failure probabilities are part of the logic. This approach employs two methods to quantify the sequence success system cut sets, which will be converted to a single recovery basic event and then multiplied back into all the cut sets within that specific sequence. This recovery adjustment will be based on the quantified success system cut sets. One approach will quantify the success system cut sets via the minimal cut set upper-bound (MinCut) approximation, and the second will quantify these cut sets using the binary decision diagram (BDD) quantification.