On-line decision support for take-off runaway scheduling at London Heathrow Airport

The research problem considered in this thesis was presented by NATS, who are responsible for the take-off runway scheduling at London Heathrow airport. The sequence in which aircraft take off is very important and can have a huge effect upon the throughput of the runway and the consequent delay for aircraft awaiting take-off. Sequence-dependent separations apply between aircraft at take-off, some aircraft have time-slots within which they must take-off and all re-sequencing performed by the runway controller has to take place within restrictive areas of the airport surface called holding areas. Despite the complexity of the task and the short decision time available, take-off sequencing is performed manually by runway controllers. In such a rapidly changing environment, with much communication and observation demanded of the busy controller, it is hardly surprising that sub-optimal mental heuristics are currently used. The task presented by NATS was to develop the decision-making algorithms for a decision support tool to aid a runway controller to solve this complex real-world problem. A design for such a system is presented in this thesis. Although the decision support system presents only a take-off sequence to controllers, it is vitally important that the movement within the holding area that is required in order to achieve the re-sequencing is both easy to identify and acceptable to controllers. A key objective of the selected design is to ensure that this will always be the case. Both regulatory information and details of controller working methods and preferences were utilised to ensure that the presented sequences will not only be achievable but will also be acceptable to controllers. A simulation was developed to test the system and permit an evaluation of the potential benefits. Experiments showed that the decision support system found take-off sequences which significantly reduced the delay compared with those that the runway controllers actually used. These sequences had an equity of delay comparable with that in the sequences the controllers generated, and were achieved in a very similar way. Much of the benefit that was gained was a result of the decision support system having visibility of the taxiing aircraft in addition to those already queueing for the runway. The effects of uncertainty in taxi times and differing planning horizons are explicitly considered in this thesis. The limited decision time available ensures that it is not practical for a runway controller to consider as many aircraft as the decision support algorithms can. The results presented in this thesis indicate that huge benefits may be possible from the development of a system to simplify the sequencing task for the controllers while simultaneously giving them greater visibility of taxiing aircraft. Even beyond these benefits, however, the system described here will also be seen to have further potential benefits, such as for evaluating the effects of constraints upon the departure system or the flexibility of holding area structures.