Availability analysis of a generalized maintainable three-state device parallel system with human error and common-cause failures

Purpose - The purpose of this paper is to study the combined effect of human error, common-cause failure, redundancy, and maintenance policies on the performance of a system composed of three-state devices.Design methodology approach - Generalized expressions for time-dependent and steady state availability of a generalized maintainable three-state device parallel system subjected to human errors and common-cause failures are developed in the paper under two maintenance policies: Type I repair policy (i.e. only the completely failed system is repaired); and Type II repair policy (i.e. both partially and completely failed system is repaired). The Markov method is used to develop general and special case expressions for state probabilities, and system time-dependent and steady state availabilities.Findings - In the case of three-state devices, it is demonstrated that by increasing the number of redundant devices in parallel do not necessarily lead to the improvement in the system availability. In fact, the availability of the system depends significantly on the dominant failure mode of the devices (i.e. short-mode or open-mode). When comparing the effect of maintenance policies on the system availability, it is observed that the Type II repair policy does not lead to an improvement in the system availability. Furthermore, it is observed that both human error and common-cause failure independently lead to lower system availability.Practical implications - This study will help maintenance engineers and reliability practitioners to become aware of the combined impact of redundancy, human error, common-cause failure, and maintenance policies on the performance of the three-state device systems. Consequently, they will make better maintenance related decisions in organizations such as oil refineries and power stations that use three state devices quite extensively.Originality value - Most of the past models have independently studied the effects of redundancy, human error, and common-cause failure on maintainable system made up of three-state devices. This effort is one of the first attempts to study the combined effects of all these factors in a parallel system composed of three state devices.