Health monitoring technologies and data analytics are increasingly widespread in the aviation industry following the growth in the capacity and speed of abundant and accurate data generation and transmission from the aircraft systems. These advances are fueling a change process in aircraft maintenance strategy towards a more proactive, precise, and effective approach consolidated in the concepts of Integrated Vehicle Health Monitoring (IVHM) and Prognostics and Health Management (PHM). Following that, several model-based and data-driven prognostics methods for Remaining Useful Life (RUL) estimation have been developed in the pursuit of improving predictive maintenance interventions for different types of components. Recent papers showcased the significant challenges faced to achieve forecast accuracy as posed by the inherent uncertainty involved in the functional dynamics of complex systems. This work acknowledges these difficulties and tackles variability by embracing it in the methodology deployed by means of considering in its framework the confidence intervals associated with the estimates for a predefined level of confidence.
Nevertheless, the ability to pinpoint times-to-failure by itself is arguably not enough to yield better operational results and improve support levels of service given that scattered standalone interventions may even cause failure occurrences and total downtime to increase. This study demonstrates the rationale behind those effects and exposes the necessity for a method for achieving a compromise to optimally accommodate the concurrent economic, reliability and maintainability goals which are, respectively, the maximization of component useful life expenditure, the minimization of the running-into-failure risk and the minimization of total downtime.
Further on, the article explores the problem in detail identifying the key parameters pointed out in the literature that need to be addressed by the modelling process to ensure the soundness of the method. The text then proposes a solution consisting of an innovative analytical model that optimizes operational availability through the dynamic allocation of flight-hours to each aircraft part of a fleet based on the integration of predictive and scheduled maintenance, minimizing total downtime, while accounting for prognostics uncertainty and the associated the risk of failure and incurring in corrective maintenance. The intended outcome is the capability of providing dynamic maintenance plans specially adjusted to each tail number according to its assessed health status and a calculated prognostic that considers predetermined future flights specifically attributed to optimize the overall availability of the fleet.
An illustrative case study involving multiple components with different aging parameters equipping the aircraft of a small military fleet operating from a single base was used to test the solution and produced results that corroborate the validity of the approach adopted and demonstrate the model’s value and effectiveness. The results also indicate there is significant potential to expand the study and encourage its further development to contemplate multiple-base scenarios and incorporate more detailed aspects such as tasks location within the aircraft, availability of spare parts and resources in general, out-of-phase items and its implementation together with a simulation tool to generalize its application.
The main contributions of the study are twofold. It adds on the theoretical complexity by tackling systems of systems instead of the predominant single component approach, and it provides a model with an optimizing objective function to improve maintenance planning in real-life.