Simulating corrective maintenance: Aggregating component level maintenance time uncertainty at the system level

The corrective maintenance process can be decomposed into failure and repair processes. Creating a model to capture the corrective maintenance process then requires an accurate estimate of the behavior of these constituent processes. For systems composed of many individual parts, information about failure and repair behavior is more likely to be available at the component level than the system level. Depending on the number of components that comprise the system, modeling each part may become computationally burdensome; in addition, some few components may account for a large portion of the overall system failures. In such a situation, one solution to the modeling burden is aggregation: the mathematical assimilation of many component distributions into a single representative distribution for the group. This paper describes how aggregation may be performed for such a system and how an algorithm may be developed to automate the process. Next, it describes how to simulate an aggregated distribution using a pseudo-random number generator and finally demonstrates these concepts for a sample problem. The first section of the paper introduces corrective maintenance modeling and aggregation; the second section describes aggregation for corrective maintenance; the third explains how to simulate the aggregated distribution; the fourth demonstrates aggregation; and the fifth discusses limitations of the method and concludes. (C) 2013 The Authors. Published by Elsevier B. V. Selection and/or peer-review under responsibility of Georgia Institute of Technology