A statistical evaluation of risk priority numbers in failure modes and effects analysis applied to the prediction of complex systems

Abstract

Complex systems such as military aircraft and naval ships are difficult to cost effectively maintain. Frequently, large-scale maintenance of complex systems (i.e., a naval vessel) is based on the reduction of the system to its base subcomponents and the use of manufacturer-suggested, time-directed, preventative maintenance, which is augmented during the systems lifecycle with predictive maintenance which assesses the system's ability to perform its mission objectives. While preventative maintenance under certain conditions can increase reliability, preventative maintenance systems are often costly, increase down time, and allow for maintenance-induced failures, which may decrease the reliability of the system (Ebeling, 1997). This maintenance scheme ignores the complexity of the system it tries to maintain. By combining the base components or subsystems into a larger system, and introducing human interaction with the system, the complexity of the system creates a unique entity that cannot be completely understood by basing predictability of the system to perform tasks on the reduction of the system to its subcomponents. This study adds to the scholarly literature by developing a model, based on the traditional failure modes and effects analysis commonly used for research and development projects, to capture the effects of the human interaction with the system. Based on the ability of personnel assigned to operate and maintain the system, the severity of the system failure on the impact on the metasystems ability to perform its mission and the likelihood of the event of the failure to occur. Findings of the research indicate that the human interaction with the system, in as far as the ability of the personnel to repair and maintain the system, is a vital component in the ability to predict likelihood of the system failure and the prioritization of the risk of system failure, may be adequately captured for analysis through use of expert opinion elicitation. The use of the expert's opinions may provide additional robustness to the modeling and analysis of system behavior in the event that failure occurs.