The First Order Stability (FOS) concept
Every airport has to deal with the problem of allocating flights to airport gates, under a set of objectives and constraints. In practice, this problem is solved one day in advance, based on the schedule arrival- and departure-times information provided by airlines. This is a hard problem to solve, due to the common size, the number of constraints involved and the conflicting objectives. For example, at Amsterdam Airport Schiphol (AAS), on a typical day, more than a thousand flights have to be assigned to over hundred gates and remote stands. In addition, the strong competition between airlines and the ever increasing passenger demand for comfortable and reliable travel experiences make this a relevant topic for air transport, and for airport authorities in particular.
Reliability is a key work in this problem. Not only for passengers, but also for all the stakeholders involved in the operations at the airports. A flight allocation solution which is unable to accommodate disturbances, such as late arrivals or aircraft changes, will cause major delays in the operations and potentially cause major losses of revenues. In addition, it will negatively impact on passengers satisfaction. It is estimated that currently 40% of the flights at AAS have at least one gate change compared to the original solution, and that around 20% of the passengers have a wrong gate number on their boarding pass when they arrive at the airport.
To solve this challenging problem the First Order Stability (FOS) concept was developed, based on the idea that flights should be allocated to gates according to their updated sequence of arrival and departure times and not just by following their scheduled times. The goal is to increase flexibility and robustness in the allocation process by following two key strategies: to postpone the gate scheduling to a later moment, when the level of uncertainty in the gate assignment process is reduced significantly; and to stabilize the order of flights arriving and departing, reducing the probability of swapping during operations.
A simulation tool using the FOS concept was designed and implemented. The tool uses a stochastic model, based on extensive analysis of historic flight data, and a dynamic model that incorporates all new information from flight information systems as soon as it becomes available. As a result, the model is able to deal with delays in a flexible way, easily solving small disturbances, without compromising airline goals or jeopardizing passengers experiences. A two stage system was developed in order to fulfill the requirements from some airport operators that need to have their workload planned few hours in advance. During the first stage flights are divided in sets and allocated to groups of airport gates based on their risk of conflict. In the second stage the dynamic information is used to optimize the allocation of flights in each gate group, assigning each flight to specific gates based on their forecasted arrival and departure sequence.
The C-pier at AAS was used as a first case study to illustrate the practicability and the potential of the concept. A case study, using the context of AAS and KLM, showed the increase in robustness and flexibility the FOS method can provide. Ground conflicts are almost completely reduced, eliminating a major source of operational disturbances. Despite the high uncertainty regarding scheduled times, FOS provides solutions that can make equal or higher usage of the capacity available, even for reasonable levels of confidence. When allocating the same number of flights as AAS is currently doing, during second stage allocation the average swap probability of flights can easily be reduced to less than 10%.
Future tests need to be performed – for instance, the application of this concept to the entire AAS airport in a wide range of operating days. But the first results illustrate that the FOS method indeed increases the robustness and flexibility in the daily process of allocating flights to airport gates. Less gates changes in the future will mean smoother airport experiences for passenger and more stable airport operations for airlines.
This work was developed under the collaboration between the Faculty of Aerospace Engineering at TU Delft and Air France-KLM. The work was realized by the former TU Delft – Aerospace Engineering Master student Dennis Buitendijk as part of his graduation thesis. He was supervised by Bruno F. Santos (TU Delft), John-Paul Clark (Georgia Tech, USA) and by Joris de Kaey and Bernard Vroom from the Air France-KLM Decision Support group.