An analysis of sensor effectiveness to inform a predictive maintenance policy

摘要

Joint Vision 2020 presents a plan for military dominance over the spectrum of military operations. One program that allows this to happen is Performance Logistics, which intends to increase availability and lower life cycle costs for weapon platforms. The ability to sense impending failures plays an important role in Performance Logistics. This thesis studies how sensor performance, as a tool of Condition Based Maintenance, affects the availability and cost of a generic component. Different types of maintenance policies are evaluated and compared using mathematical models. The maintenance protocols considered are reactive and proactive, namely: run to failure, scheduled inspection times, sensor based, and a combined inspection and sensor policy. Given parameters such as time and repair cost due to warnings or failures and frequency of inspection, it's found that a sensor influences the benefits of implementing a Condition Based Maintenance policy. In this thesis, results show improvement in availability and a reduced long-run average operating cost when the median of the random ratio of warning to failure time is roughly 0.8, the standard deviation is less than 0.1, and the mean time of maintenance for failure is greater than three times the mean time of repair due to warning.