摘要

The common practice for the estimation of the probability of a component failure to start on demand is to lump together standby failures and failures due to the stress induced by the demand itself. In this paper, a failure by demand stress as well as a standby failure is considered separately to estimate the probability of a failure on demand when a Surveillance Test Interval (STI) is modified. The proposed estimation method is based on alpha, the ratio of the probability of a failure by demand stress to that of a failure on demand. The estimate of alpha, however, should be changed when an STI is modified. An estimation method for alpha', the changed parameter corresponding to alpha under an STI modification is also suggested. With the proposed estimation method and the estimate of alpha' from a plant specific database, a risk impact analysis due to an STI modification for the Emergency Diesel Generator (EDG) of UCN units 3 and 4 is performed. From the sensitivity analysis, the following conclusions can de derived: if it is assumed that all failures occur during the standby time, as it is the case in the common practice, the results can overestimate the risk impact. And it is required to construct a component reliability database containing detailed information on the failure types to recognize whether a component failure occurs during a standby time or a test.

  • 出版日期2010-10