During the past decades airports have been steadily confronted with the problem of scaling their infrastructure. On one hand, in times of growing demand, it has been difficult to keep pace with the growth. On the other hand, in case of dropping demand and unforeseeable events, it has been difficult to scale down. Building the infrastructure for additional runways is costly and has a very long lead time. Furthermore, the physical footprint of the airport increases considerably, with all its negative environmental and logistical consequences. A better utilization of capacity by packing traffic more densely decouples the demand from the infrastructure and makes the airports more agile, more resilient, and responsive to unforeseen demand changes. Therefore, prioritizing better infrastructure utilization to new infrastructure projects can be the better choice in terms of return on investment.
However, increasing runway utilization brings new challenges, since traffic will be less predictable due to queues and interdependencies of traffic – an enormous increase of complexity that humans cannot handle. During the past decade we have worked on how to solve this problem, and building on our experience in the implementation of effective optimization algorithms since the 1990s we have built a solution for air traffic management that is beyond state-of-the-art, bringing together the practitioners and end users with top-level experts in classical and new AI methods.
The design of IRSM is a result of studying how to best optimize the runway use in the highly dynamic environment of an airport and with the air traffic controllers in the decision loop. IRSM is therefore tailored for high-speed information-sharing already in the fundamental design. We call our approach Incremental Optimization (IO). With IO, several streams of information regarding the aircraft and runway status are processed and analysed in real-time to pinpoint exactly which parts of the information that have changed. Only the changed information is then forwarded to the core optimization algorithm, where it immediately is incorporated into the runway sequences. Hence, the core algorithm benefits from knowing exactly what has changed and can avoid to re-compute the parts of the solution that has been kept stable. During idle periods between reception of new information, the IO algorithm searches for any possible improvements of the solution. Hence, IRSM contains an anytime algorithm able to provide complete solutions without time-delay and to handle the trade-off between execution time with result quality.
Our IO algorithm is purely sequence based (as opposed to more traditional flow-based approaches) by optimizing the exact mix of arrivals and departures on mixed-mode runways without the need for arrival-departure patterns or maximum arrival and/or departure rates. Instead, the physical capacity of the runway is handled through minimum pairwise time separation constraints that cannot be infringed. These separation constraints take specific information regarding the involved aircraft and the current meteorological condition into consideration. In situations where the airport capacity is limited due to other factors than runway capacity, the flows can be reduced by adding flow restrictions into the separation constraints. By calculating the arrival-departure patters dynamically, the IRSM will reduce the human workload while allowing for better solutions to be identified, enabling the controller to concentrate on core tasks while leaving the task of making the flows efficient to IRSM. During periods of high traffic, the optimal solutions will be those where the runway capacity is fully utilized, while during periods of lower traffic, the optimal solutions will be those where the balance between punctuality and fuel efficiency is maximized. Since arrivals and departures are planned simultaneously and by the same algorithm, the importance of avoiding delays of departing aircraft can be given the same priority as delays of arriving aircraft. The inherent uncertainty of when departures will be ready to start taxiing from the stands or deicing areas is tackled by adding automatically managed time buffers to the taxi time, where the optimal amount of time buffering will depend on traffic load.
A positive surprise to us while developing IRSM was how easy it was to implement general support for what-if analysis capability into IRSM. Whenever controllers want to analyse decisions before implementing them, they can start what-if scenarios as separate branches of changes. Specific assumptions (e.g., sequence modifications, balancing of flights between runways, insertion of runway closures, or changes of runway configuration) can be added and studied off-line. When a what-if scenario leads to an unacceptable solution, the branch can be deleted. However, if the controllers want to implement the new assumptions into the live operations, the assumptions of the branch are automatically merged into the live scenario. In this way IRSM provides general support for what-if probing. In principle any decision can be tested prior to implementation.
While IRSM is able to produce optimal and stable sequences for mixed-mode runways, it can also sequence flights on runways operated in segregated mode. Normally the demand for departure capacity is different from the arrival capacity. In these situations, IRSM can advise the air traffic controller on flights that can be moved to an alternative runway to minimize overall delay. The planning horizon for the IRSM can be set to several hours. Hence, IRSM is able to detect these imbalances early, providing the stakeholders with enough time to find the solution that is the most optimal for any given situation. IRSM supports different levels of automation, ranging from manual intervention of the user to high levels of automation where sequences are maintained without user input.
In validations with practitioners and end users the IRSM solution has shown that it is fit for purpose.1 In the SESAR project PJ.02-08 "Integrated runway sequence for full traffic optimisation on single and multiple runway airports" it passed V3 level validation showing increased airport capacity, more predictable and punctual flights as well as improved fuel efficiency for airspace users. The runway capacity was increased by 5.1% and the predictability of flights by 60.8% while the level of safety was maintained with Stockholm-Arlanda Airport operating on independent parallel runways. In the large-scale demonstration project SESAR VLD03 W2 SORT the IRSM passed V4 level (TRL 7), demonstrating the benefits at Stockholm Arlanda in shadow mode. This closed the gap between concept development and deployment by demonstrating operational and technical readiness. The IRSM is therefore now at a maturity where it is ready to enter the industrialization stage.
IRSM is generic and configurable to most airports regardless of size, layout, and preexisting system solutions (e.g. AMAN, DMAN, A-SMGCS, A-CDM, etc.).
Some expected benefits of IRSM
To our knowledge, SINTEF's solution to Integrated Runway Sequencing is unique among solutions currently available on the market.
- Relying on pure mathematical optimization, IRSM will provide better operations, improving throughput, predictability and punctuality and reduce the overall environmental impact.
- By requiring less manual input and by providing more stable runway sequences, IRSM will reduce the workload for air traffic controllers and pilots.
- By providing a shared plan for the runway use, IRSM will increase the situational awareness of involved stakeholders.
- By providing general support for what-if analysis, air traffic controllers can analyse the overall impact of decision prior to implementation and making it visible for other stakeholders.
- By supporting various levels of automation, IRSM can facilitate a stepwise transition towards full automation.
- By allowing the air traffic controllers to adjust the optimization parameters, IRSM can be configured and tuned for the environment of your airport.
1. The validation projects have received funding from the SESAR Joint Undertaking and European Union under grant agreement No. 731781, PJ02-08 EARTH and No. 874520, VLD3-W2 SORT.