Workforce scheduling of airport ramp service employees under the impact of flight delays
Abstract
The efficient scheduling of airport ramp service employees (RSEs) is vital for the smooth operation of airlines, since it will satisfy the manpower demand while minimizing the manpower cost. Airlines use established flight schedules, which by nature are deterministic in terms of arrival and departure times, as the primary input to calculate the manpower demand for RSEs. However, in practice, airline operations are subject to many uncertain factors, such as weather, air traffic control routing, and non-routine maintenance. Flight delays, caused by these factors, are usually not considered during RSE scheduling. Therefore, with random flight delays, the theoretically “optimal” scheduling plan will not remain optimal in terms of satisfying manpower demands and saving manpower costs. This study used historical on-time performance as well as flight schedules as inputs to a demand simulation program, which was used to project manpower demands during the planning period. Working schedules of RSEs to cover these demands were generated by the shift scheduling model using mixed integer programming and robust optimization. The scheduling plan generated by the robust scheduling model was compared with the plan generated by the deterministic model in terms of the overall cost, the number of demand violations, and the penalty cost to evaluate the effectiveness of the approach proposed by this study. Findings suggested that the scheduling plan generated by the robust scheduling model saved more on the overall cost and the penalty, and had less demand violations than the scheduling plan generated by the deterministic model. Through the method proposed by this study, airlines may be able to save on overall manpower costs, respond faster and better to manpower demand variations caused by unexpected flight delays, and decrease the number of manpower demand violations. Also, a scheduler will be able to adjust the scheduling plan according to the anticipated on-time performance by simply changing the value of the overall demand satisfaction rate in the scheduling model, which could reduce the workload of the scheduler.
Degree
Ph.D.
Advisors
Fanjoy, Purdue University.
Subject Area
Management|Industrial engineering|Operations research|Artificial intelligence
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.