Practical availability analysis of standby systems

Comprehensive availability models are developed in this paper for analyzing redundant standby safety systems. The models, which include both periodic inspections and continuous monitoring for discovering failures, incorporate human errors associated with testing, monitoring, and repair as well as spurious alarms and failures after a demand during the mission time. Equations are developed for calculating optimum inspection intervals and average steady state availabilities of standby m-out-of-n:G systems (with several schemes of staggering the tests), as are techniques for including undetected and common-cause failures. Analytical models provide a basis for many generic conclusions. Numerical examples representative of safety systems in nuclear power plants illustrate, for example, the importance of developing monitoring techniques to detect errors made in testing and returning components to service after maintenance.