New Metrics to Assess Reliability and Functionality of Repairable Systems

The classical definition of reliability may not be readily applicable for repairable systems. Commonly used concepts such as the Mean Time Between Failures (MTBF) and availability can be misleading because they only report limited information about the system functionality. In this paper, we discuss a set of metrics that can help with the design of repairable systems. Based on a set of desirable properties for these metrics, we select a minimal set of metrics (MSOM) which provides the most information about a system, with the smallest number of metrics. The metric of Minimum Failure Free Period (MFFP) with a given probability generalizes MTBF because the latter is simply the MFFP with a 0.5 probability. It also generalizes availability because coupled with repair times it provides a clearer picture of the length of the expected uninterrupted service. Two forms of MFFP are used: transient and steady state. Other metrics of cost, planning horizon, number of failures within the planning horizon are also used. We also present in this paper the metric of effective age which has been used in the literature for remanufactured products. It uses a relatable unit of years to report the performance of a repaired system. In doing so, the metric results in the effective age of the system as a linear weighted sum of the ages of the components. The system topology information (e.g. reliability block diagram) is then used to calculate the coefficients of the weighted sum associated with each component, leading to a very simple and useful model to assess system functionality. We demonstrate the value of the MSOM using the design of a four-joint unmanned ground vehicle.