An unavailability model for errors made during surveillance testing

Periodic testing or examination is conducted routinely for some critical standby equipment or instrumentation, in an attempt to increase its availability. This is often called 'surveillance testing.' There is a finite probability that tested equipment will be left in a degraded or nonfunctioning state because of an error, and this important type of human error is often called a 'restoration error' in probabilistic safety analysis (PSA) parlance. One would expect a trade-off between the reduction in unavailability gained by a surveillance testing and the increase in unavailability incurred by surveillance testing errors. In this paper, a mathematical model of the effect of errors made during surveillance testing on equipment unavailability is described.