Arrival Management

Please don’t let the bridge be open… Dealing with uncertainties in arrival times

Great! The sun’s shining. One more sip of coffee and you jump on your bike. There are many benefits to cycling to work: Fresh air, a healthy dose of exercise and no traffic jams! Yet your arrival time is not 100% predictable. That bridge that has opened to let ships pass or a strong headwind can slow you down. The further you live from your work, the more uncertain your arrival time is. The same applies to the arrival times of aircraft. The uncertainty persists. So what can we do to improve this?

The sooner the better

At the moment we plan the landing times of incoming aircraft about 30 to 40 minutes in advance. But could we do this earlier? If so, the airline could adjust the speed at an earlier stage, and in doing so perhaps save fuel and reduce emissions. Maarten Tielrooij, aviation consultant at To70, and the Netherlands Aerospace Centre (NLR) together examined how we can extend this planning horizon to two hours. Maarten did so as part of his PhD research at TU Delft.

Nothing as unpredictable as the weather

The problem is: the further you look ahead, the more can happen. Just like in weather forecasts. Maarten Tielrooij explains: “At Schiphol it is mainly the departure times that pose a problem. This is because the twenty busiest connections to and from Schiphol are within two hours’ flying time. If an aircraft is on the ground, many things can occur to cause delays. The aircraft has to wait for a late passenger or there is a technical problem. Or the airport experiences very busy periods. Then suddenly the forecast is no longer correct…” 

Incorrect predictions cancel out any benefits

If air traffic control takes decisions based on incorrect predictions, this can cancel out any benefits or even reduce efficiency. “Take an aircraft that has reduced its speed in order to arrive later. That same aircraft might suddenly have to increase its speed in order to arrive earlier after all.”

How can we make errors in predictions as small as possible?

This is something we are currently working hard to do, e.g. by developing improved models or providing more information on the progress of departures. “Yet,” Maarten Tielrooij continues, “we can never remove all of the uncertainty. Such as those late passengers. Or ice on the aircraft. My research therefore focuses on determining the uncertainty of the predictions. This enables the air traffic controller to include that information in his or her decisions.” 

We determine the uncertainties…
At the moment, the Network Management and Operations Center (NMOC) predicts when an aircraft is expected to enter LVNL airspace. LVNL uses this information to decide how much capacity this requires of the airspace and the airport. Maarten Tielrooij analysed nearly 700,000 predictions. “Next I created a model of the errors in these predictions. This model predicts the uncertainty of the NMOC prediction.

…and make these visible in the display concept
“If the uncertainty has been predicted, we can then make it visible. We have developed a display concept to this end. This shows not just the uncertainty, but also the effect of this uncertainty on the required capacity.” Indeed, Maarten Tielrooij’s research takes no chances. “We are far from finished. The next step is to combine the information from the display concept with the data of actual flights.”

The display concept