Flight delays are a fact of life. MITRE and the FAA are working together to ensure air traffic controllers and operators have the right information and analysis tools to mitigate delays with understanding—all provided by a new computer model.
At any given moment, there could be 5,400 planes crisscrossing the U.S. National Airspace System (NAS), with more readying for takeoff. The slightest hiccup can cause a delay—resulting in missed connections, disrupted commerce, disgruntled airlines, and unhappy travelers.
For air traffic managers, though, purposeful delays are vital tools to keep the NAS operating safely and smoothly. Without the ability to delay planes on the ground, a single incident could cause safety issues in the sky. Intentional delays give operators and managers breathing room to address problems and get back on track.
The Federal Aviation Administration manages delays through procedures called Traffic Management Initiatives (TMIs) but using them effectively requires data analytics. Delay too little, and the original problem may be worsened. Delay too much, and you cause other hardships for no real reason. Choose wrong, and costs and risks increase rapidly.
Flights can also be subject to multiple delays simultaneously: a ground stop here, a weather delay there, all adding up to headaches, complications, and questions for the FAA from the airlines: Why create this delay for this long? What was the impact on other flights? Was it necessary? Was it efficient?
The FAA leveraged MITRE to develop the insights and answers that could be derived from analyzing flight delay data.
A New Model for Understanding Flight Delays
Currently, air traffic management decisions are mostly based on operators’ field experience. The FAA needed a centralized system to collect information on the reasons for—and impact of—flight delays and evaluate their effectiveness.
As part of that effort, MITRE researcher Ehsan Esmaeilzadeh and his team created a new modeling capability called Layered TMI designed to help the FAA understand the delays associated with each flight.
“We developed a logic to identify the flights subjected to TMIs, and a logic to estimate the delay received from each,” Esmaeilzadeh explains. “At the end, this tells you: Here’s the decisions the FAA made today; here’s the affected flights; and this is the impact those decisions caused in the National Airspace System (NAS).”
Understanding and measuring delays has multiple benefits for the FAA, and for airlines. The new Layered TMI model is already in use by some analysts providing the FAA and airlines with detailed understanding they previously lacked.
“We have found the Layered TMI capability extremely useful and impactful,” says Martin Durbin, a manager with the FAA’s Operations Performance Assessment branch, “and we’ve asked the University of California, Berkeley NEXTOR research group to leverage this data to support airport-TMI impact analysis.”
Helping the FAA Turn TMI into A-OK
Esmaeilzadeh and his team aim to bring the modeling capability to greater prominence as part of MITRE's CAASD Transportation Data Platform (TDP).
They are also hosting the capability in a cloud-based Collaborative Research Environment (CRE) to enable cross-organization collaboration and accelerate the transition from research into operations.
And while flight delays may be unavoidable, MITRE and the FAA will work to ensure they at least won’t be unnecessary.