The Australian Defence Force is in the process of receiving several new capabilities following recent acquisition decisions. In an era of declining government - and thus military - spending, there is an increasing imperative to maximize performance within existing Defence budgets. Therefore there is a desire - indeed obligation - to seek to manage these new aircraft fleets in the most efficient way possible. In practice, this translates into areas such as better maintenance processes and increased operational availability. In this work we consider a new fleet of 24 MH-60R Seahawk naval combat helicopters being delivered to the Royal Australian Navy. The fleet is required to maintain eight aircraft embarked on ships at sea every day, and to achieve a certain number of ashore flying hours (across two squadrons) and embarked flying hours each year, over a 30 year life. Good fleet management would suggest that not only should these three primary requirements be met, but also the flying hours across the fleet should be somewhat balanced. In that way, no individual aircraft is too far ahead or below the rest of the fleet regarding total flying hours. Aircraft too far ahead would ultimately have to be retired early, reducing the fleet size. Similarly, each aircraft should have a reasonable split of embarked and ashore hours, as aircraft wear more rapidly when embarked due to increased exposure to corrosive salt-laden environments and the higher deployed operational tempo. A fleet simulation model has been developed to test the capability of the fleet to meet these requirements. The model represents individual aircraft moving between various states, such as between ashore and embarked, and between serviceable (able to fly) and different types of maintenance, from regular short inspections to depot-level maintenance. Unscheduled maintenance is included as a random effect, with both the time between failures and the repair time represented as probability distributions. The model also incorporates various fleet management policies, to see which are most effective in meeting the three main requirements as well as the desire to provide a reasonably balanced fleet. Six policy types are tested: daily flying allocation; daily maintenance allocation; maintenance crew rotations within squadrons; tail rotation between squadrons; balancing the flying hours across the fleet; and sharing squadron resources. There are a large number of inputs in the model, and while many are fixed, others are more ambiguous, such as the range of flying hours that an aircraft and the fleet may fly each day. When combined with the uncertainty in the values of the unscheduled maintenance distributions, as well as the range of fleet management policies described above, there is a large parameter space that can be explored. Given this, a simulation experimental design approach has been applied to this problem. This allows a thorough exploration of the parameter space in fewer runs than would otherwise be required. We apply this design to eleven continuous variables, two discrete variables, and the six fleet management 'policies' listed above. Each of these policies may have between two and five options. Ultimately, 1040 design points are used, with 50 replications at each point -by contrast, a full factorial design would require around 1E33 points. The results show the strong influence of varying unscheduled maintenance on the ability of the fleet to meet the ongoing requirements, particularly if the frequency is greater than anticipated. It also demonstrates that particular fleet management policies, such as those that share squadron resources and rotate maintenance crews within squadrons, can be used to counteract an increase in unscheduled maintenance. For the requirements to balance the embarked, ashore and total flying hours across the fleet, tail rotation policies that enable more frequent rotations are most effective overall, especially for balancing embarked hours. Overall, the experimental design technique provides insights to decision makers. In this fleet management example, it identifies the variables and policies of greatest influence. It thus assists decision makers in determining the best policies, and targeting the areas that may negatively impact on meeting requirements.
