Human Error Considerations in Determining the Optimum Test Interval for Periodically Inspected Standby Systems

Abstract

This paper incorporates the effects of two types of human error in a model for determining the optimal time between inspections for a safety system. The possibility that a bad safety system is undetected upon inspection (Type B human error), as well as the possibility that a good safety system is inadvertently left in a bad state after the inspection (Type A human error), are considered. We develop a Markov model for the steady-state availability of the safety system which is then used to determine the optimum time between inspections which either maximizes the availability or minimizes the combined inspection and unavailability costs. The safety system failure (hazard) rate need not be constant. The optimum time between inspections increases as the probability of a Type A error increases and a Type B error decreases. The optimum availability decreases and the optimum total cost increases as the error probabilities increase.