In conducting Probabilistic Risk Assessments (PRAs) for Nuclear Power Plants (NPPs) two key mathematical terms are used in computing the average unavailability for Standby Safety System components. The two terms include: 1) a component’s demand failure probability, and 2) a component’s standby failure rate, λ. Here, p_d represents the proportion of the component’s failures resulting from demand related stresses (i.e., failures that occur when the component is activated or a demand is placed upon it) and λ embodies the fraction of failures occurring from standby stresses (i.e., failures that happen when the component is in a quiescent state). Currently, it is not known whether these components have a propensity to fail predominantly from one or the other stress types. Typically PRAs are performed by using generic data for the failure mode estimators. Unfortunately, the generic data only provide an estimate for the component’s dominant failure mode. Hence, no estimate is provided for the component’s other failure mode. The ramification for using the inaccurate estimates for the failure mode measures is not isolated to the obvious misinterpretation of the plant’s risk assessment that could result, but also could lead to wrong decisions being made regarding component test intervals since test intervals for NPP components are usually established based on reliability considerations (i.e., p_d versus test interval and λ versus test interval).

The implications for using generic data, as well as the development of a technique for using plant specific data to estimate component stress measures is examined. The use of the refined stress measure estimators to better prescribe standby safety system component test intervals is also studied. As an aid in examining the responses of the stress measures to a variety of combinations of actual demand failure probabilities and standby failure rates, a Monte Carlo simulation program was developed to model a variety of mixed stress load conditions. The program will serve as an evaluation tool for studying the implications associated with using incorrect stress indicator estimates (i.e., p_d and λ) obtained from generic failure rate values and for establishing a method for applying plant specific data to estimate PRA unavailability parameters to improve PRA predictions and establishing proper component test intervals.