Recovery Actions in PRA (Probabilistic Risk Assessment) for the Risk Methods Integration and Evaluation Program (RMIEP). Volume 1. Development of the Data-Based Method.

摘要

In a probabilistic risk assessment (PRA) for a nuclear power plant, the analyst identifies a set of potential core damage events consisting of equipment failures and human errors and their estimated probabilities of occurrence. If operator recovery from an event within some specified time is considered, then the probability of this recovery can be included in the PRA. The report provides PRA analysts with an improved methodology for including recovery actions in a PRA. A recovery action can be divided into two distinct phases, a Diagnosis Phase (realizing that there is a problem with a critical parameter and deciding upon the correct course of action), and an Action Phase (physically accomplishing the required action). In the methodology, simulator data are used to estimate recovery probabilities for the diagnosis phase. Different time-reliability curves showing the probability of failure of diagnosis as a function of time from the compelling cue for the event are presented. These curves are based on simulator exercises, and the actions are grouped based upon their operational similarities. This is an improvement over existing diagnosis models that rely greatly upon subjective judgement to obtain such estimates. The action phase is modeled using estimates from available sources. The methodology also includes a recommendation on where and when to apply the recovery action in the PRA process.