Modeling the impact of maintenance on naval fleet total ownership cost

Abstract

The US Navy is making a concerted effort to use total ownership cost (TOC) as a metric for decision-making about the various systems needed to perform the Navy's missions. System Total Ownership Cost seeks to combine aspects related to acquisition costs, operating costs, maintenance costs, and manpower costs (both staffing and training) over the lifecycle of the system. Here, this paper presents initial efforts to consider deferred maintenance and its impact on TOC for long-lived systems, like the DDG-51 class destroyers. Near-term cost pressures often result in decisions that defer maintenance to a later time than scheduled or well after first notice of a maintenance need. Deferring maintenance allows the costs of performing maintenance to be postponed, saving short term costs, but the choice to defer maintenance may also result in the system moving to a state of further degradation. If this is true, later maintenance tasks needed to restore the ship's capability or reliability may become more costly. While these trade-offs are conceptually well understood, they have not been adequately quantified to allow decision makers to make the best decisions when funds are constrained. One reason such quantification has been lacking is that the necessary data is often not available. This paper presents initial work aimed at using data recorded by the Navy to construct a model that could allow for quantitative decision support. The principal challenge is that most of the recorded data is at the system level, implying that the ship must be modeled as a single unit. This assumption results in an underestimation of the impact on reliability of deferring corrective maintenance. Our results show that given the data available, a stochastic renewal process can model the Arleigh Burke (DDG-51) class guided-missile destroyers, implying that the ship returns to a “like new” condition following successful maintenance. The stochastic renewal process model provides a first step in using reported data to develop a model of delayed maintenance and its effect on TOC.