A study of the impact of prognostic errors on system performance

Condition-based maintenance is growing in popularity as a means of improving equipment maintenance efficiency. Unfortunately, the prognostic tools associated with condition-based maintenance are subject to statistical error. These errors can lead to unnecessary preventive maintenance due to underestimation of system remaining life and unnecessary system failures due to overestimation of system remaining life. What is not clear is if these statistical errors outweigh the benefits of a just-in-time maintenance philosophy. This study attempts to address this concern through the evaluation and comparison of three maintenance policies for a simple system. The maintenance policies are run-to-failure, scheduled preventive maintenance and condition-based maintenance. A discrete-event simulation model is used to estimate the average time between successful missions for the system under each of these policies. An extensive set of numerical experiments is used to analyze system performance under a wide variety of operating conditions. The results suggest that condition-based maintenance can improve system performance as much as 10% to 15% beyond that achieved using scheduled preventive maintenance. However, the results also suggest that moderate statistical error can render condition-based maintenance inferior to scheduled maintenance and severe statistical error can render condition-based maintenance inferior to run-to-failure