It is nowadays common that when we fly somewhere we have to take multiple flights, connecting between flights at hub airports. These stops, although in some cases inevitable or even convenient, are always a cause of stress. In case of a delay in the inbound flight, our connection can be compromised, resulting sometimes in a struggle to get a seat in the next flight and in a long wait for the replacement flight.
Airlines try to avoid this hassle by managing the delays as soon as they receive information about them. This usually happens prior to passengers landing at their hub airport. The task is to decide whether to delay subsequent flights in order to wait for the delayed passengers (and bags) or to have the flights departing on time. The decision must be carefully balanced, as imposing a delay on departing flights can have a major impact on the operating schedule of the airline and on the experience of other passengers also booked to the departing flight.
TU Delft helped Kenya Airways (KQ) to develop a decision making tool to support the daily delay management process. The results are promising, suggesting that KQ can reduce up to 10% their current delay related costs and a general reduction of passengers experienced delays.
Network carriers or hub-and-spoke carriers, like KQ, KLM or British Airways, are airline companies that rely on one or more hub airports to consolidate passengers from multiple flights, connecting them between flights. For these airlines, it is critical to have effective short duration connections, guaranteeing a smooth experience to passengers without spoiling the resources of the airline. The daily flight schedules are already planned according to this goal — flights are scheduled together, in waves, linking the inbound flights in the beginning of the wave with the outbound flights departing at the end of the wave. However, during operations airlines experience recurrent delays in their flights. According to recent statistics, around 20% of all schedule flights in Europe suffered from delays. This means that more than 5 500 flights per day are delayed causing the lost of connections to thousands of passengers.
The context in Africa is not different from Europe. Kenya Airways, as one of the largest african carriers, is concerned with the delay propagation in their network and with the inconvenience that this cause to their passengers. The airline operates from its hub airport in Nairobi, linking passengers to and from Africa (Figure 2). The Operation Control Centre (OCC) of the airline meets every day a couple of times to analyze the delays of incoming flights and to decide if it is beneficial to postpone the departure times of some outbound flights. To do this, currently the airline makes use of a set of static rules to support these decisions. Better and more precise analysis of costs involved are still needed.
A master student from TU Delft – Faculty Aerospace Engineering, Maarten Wormer, spent one year in Kenya working on this problem, developing an optimization tool to support KQ decisions during delay management.
The project resulted in an optimization decision support tool that can be used to manage a full day of operations, determining:
- which flights to delay and for how long they should be delayed;
- in which flight passengers will be reallocated in the case they miss their connections;
- a priority list of flights, sequencing them for better usage of the airport capacity based on delay management needs;
- the total delay caused to passengers.
The objective is to minimize the total costs of delay, involving the the additional costs of operations, the costs related with the delay compensations given to passengers and the costs associated with the inconvenience cause to the passengers. This costs of inconvenience reflect the fact that given a bad experiences passengers can decide not to fly again with the same airline. The model is solve dynamically, as soon as new information is provided. It takes into consideration information about:
- flights delays;
- passengers connectivity, for business and economy passengers;
- airport capacity constraints, including runway, taxiway and bay capacities at multiple periods of the day (Figure 3);
- and slots allocated to flights from other airlines (which also make use of the capacity of the airport).
The tool has been tested in a set of limited delay scenarios, representative of five days of operations at Nairobi. The results, although preliminary, suggest that the potential total delay related costs from the airline can be reduced by almost 10%. For the scenarios tested, the number of passengers delayed was reduced by more than 70%, resulting also in lower average delay times and missed connections.
The version developed by TU Delft still needs to be improved, but the results are promising. In collaboration with KQ, this delay management platform is being further developed. The airline is currently considering the implementation in practice of this decision support tool to support their operations.