Computationally Optimised Manpower Rosters for Airline Short-Term Layover Maintenance.

Abstract

Abstract Introduction: Short-term layover maintenance activities at an airline base generally involve the execution of significant numbers of high-frequency, short-duration routine inspections in between flights on aircraft that are deployed to ongoing flight operations. The problem of scheduling labour supplies for this type of maintenance activity requires an operations manager to consider issues such as the number of scheduled flights each day, the timing constraints for the tasks that must be implemented on aircraft that are departing from or arriving at the base of operations, the employee skills and certifications that are needed to implement the required maintenance actions, as well as the potential for the planned flight schedule and associated maintenance schedule to be disrupted by unforeseen events such as flight delays. Published literature in the operations research subject area of ‘personnel scheduling’ covers a range of computational and mathematical techniques that can be adapted to solve or optimise this type of scheduling problem. This study has explored whether there is a significant knowledge gap between state-of-the-art developments in published research in this subject area and practical application at aircraft maintenance organisations on the African continent. Furthermore, the study has sought to establish whether computational models for the optimisation of an airline’s short-term layover maintenance labour resources that are based on a fixed flight schedule and thus a fixed task demand profile are sufficiently robust to the effects of flight schedule disruptions (e.g. delays) that occur routinely in real-life operations to support their practical application. Methods: A case study method has been employed in this study. In this regard, a case airline company based in Johannesburg, South Africa (with an in-house maintenance division) has been used for data collection purposes and for the implementation and testing of an experimental labour scheduling model for the short-term layover maintenance staffing problem. This labour scheduling model has been developed by adapting modelling concepts and techniques from contemporary operations research literature to the case setting. Initially, the experimental model has been used to generate an optimised staff roster for short-term layover maintenance activities under the ideal condition of 100% on-time performance of the case airline’s flight schedule. This optimised roster has then been compared to the existing staff roster and associated labour planning model that are in use at the case organisation for the given category of maintenance activity. Performance criteria such as total direct labour costs and the stability of each staff roster have been considered in this comparison. Following on from the initial computational modelling exercise, the optimised staff roster has then been subjected to sensitivity testing in a simulation environment in order to assess its robustness in scenarios where the input flight schedule and associated task demand profile for short-term layover maintenance activities is disrupted from the original plan. To ensure high levels of realism, schedule disruption profiles based on actual flight operations records in the case setting have been used in this testing process. Page III Results: The initial study results show that the experimental model for the short-term layover maintenance labour scheduling problem uses 33% fewer AMEs and 48% fewer rostered manhours to meet the service demands of the case airline under fixed flight schedule conditions than the equivalent labour force specified by the organisation’s existing labour planning model. However, it has been noted that a significant portion of this resource saving relates to conservative demand modelling by the case organisation, particularly with regards to the quantities and standard time inputs that have been imposed for after-flight checks in the organisation’s existing planning model. Adjusting for these conservative inputs, it has been shown that the experimental model provides a 22.0% reduction in rostered manhours and associated labour costs in comparison to a modified version of the case organisation’s existing model, but with no reduction in the number of AMEs employed. Although not as substantial as the original finding, this more realistic outcome remains positive from the perspective of the operations manager and it has been estimated that associated savings in direct labour cost savings could amount to almost 1.3 million rand per annum. The robustness testing of the output labour roster from the experimental labour scheduling model has also yielded satisfactory results. In this regard, the output roster has been shown to require only 3.75 manhours of unplanned AME overtime in order to manage the modified demand profiles for short-term layover maintenance tasks relating to simulated flight schedule disruptions over a 20 day testing period in the case setting. This level of unplanned overtime adds only 0.41% in supplementary charges to the case organisation’s labour bill at an overtime rate of 1.5 times the normal AME labour rate. Furthermore, no requirements for AME callouts and no conflicts in terms of induced flight delays due to missed maintenance tasks have been identified over the robustness testing period. These findings suggest that the impacts of unplanned flight schedule disruptions on the experimental labour planning model are largely insignificant. Conclusions: The outcomes of the study support the assertion that a significant knowledge gap exists between state-of-the-art research in the subject area of personnel scheduling and practical application at African-based aircraft maintenance organisations. In addition, preliminary evidence has been obtained to suggest that computational models for the optimisation of short-term layover maintenance labour resources at an airline under fixed flight schedule conditions are sufficiently robust to the effects of schedule disruptions to support their practical application. However, these conclusions must be considered within the context of the case-based research methodology that has been employed. In this regard, it is recommended that further research be conducted in order to examine the transferability of research findings at the case organisation to a broader population of aircraft maintenance organisations on the African continent. Furthermore, it is acknowledged that only a limited scale of robustness testing has been implemented in this study. As a result, additional robustness testing over an expanded set of test days is recommended as an extension to the study.