^{a}and Chung-Kung Lo

^{b}

^{a}elyphonic@iner.gov.tw

^{b}cklo@iner.gov.tw

Taiwan has developed Probabilistic Reliability Assessment (PRA) in evaluation of the risk assessment of nuclear power plants for more than 30 years. In power operation of the PRA model, the mitigation to recovery is usually not considered after the Initiating Event (IE) occurred. However, the Loss of Offsite Power (LOOP) would cause many mitigation systems to fail at the same time. Therefore, it is the only exception to consider the offsite power recovery in the IE of LOOP.

The recovery rate of the offsite power is calculated based on statistical records of the recovery time after the highvoltage of the nuclear power plant's offsite-grid has tripped. In the past, simple statistical analysis method was used to evaluate the recovery rate of the offsite power. This method count the number of times that failed to recovery within the available time window under the incident scenario directly, the probability of fail to recover is the percentage of the counts to the total number, and assumed the data distribution is Lognormal distribution with EF = 5. Nonetheless, this method would obtain 0 (or 1) if the available time window defined by the incident scenario is larger (or smaller) than the longest (or shorter) recovery time of the statistical data. In order to solve this problem, this paper constructs a process to determine the distribution of the probability of fail to recover. This process assumes that each recovery time record is a normal distribution; first establish the distribution curve of the recovery time (T), then convert the curve of the recovery time into a curve of recovery rate (μ), substitute the available time window under the incident scenario into the curve of recovery rate to calculate the probability of fail to recovery.