Work smarter, not harder: New technology takes flight

Runways may rank last on your list of considerations when purchasing plane tickets, but have you ever considered how much their maintenance affects your wallet?

With more than 2 million passengers each day in U.S. airports, according to the American Society of Civil Engineers, it’s no wonder America’s infrastructure is simply “not keeping up.”

But what if we could predict issues in our infrastructure well before they rise to the surface?

That’s a question Imad Al-Qadi, University of Illinois at Urbana-Champaign Bliss Professor of Engineering and Illinois Center for Transportation director, and Erman Gungor, UIUC doctoral student, asked themselves with their research effort, “Temperature Responses of Partially Restrained Airfield Rigid Pavement.”

Sensors collect the necessary data for machine learning at JFK International Airport in 2013. The researchers’ data will improve future airfield design and lengthen life cycles.

Here the two observed the daily use of John F. Kennedy International Airport in Queens, New York with the help of the Federal Aviation Association’s Navneet Garg, an UIUC alumnus. The researchers were able to take a closer look at perhaps the biggest culprit of airfield wear and tear — the airplane’s gear tires.

“The tire pressure and weight of these airplane gears (are) impacting the performance of the pavement,” Al-Qadi said.

In an effort to design better runways in the future, Al-Qadi found it vital to try and understand the “stresses and strains” of pavements at the U.S.’ sixth busiest airport, according to the U.S. Department of Transportation’s Bureau of Transportation Statistics.

Researchers had to get creative, thinking outside of the box. It was out with the old way of digging through data and in with the new machine learning. Here computers learn similarly to that of humans, building knowledge based on prior experiences.

With machine learning, however, those prior experiences are acquired through a nontraditional means with a programmer initially inputting those experiences via algorithms.

Gungor said machine learning helped the research team “organize the data in different structures,” which in turn was “much easier and intuitive to use and process.”

The result?

The team was able to understand what causes the curling or the bending of pavement in the airfield, attributing to the need for repairs or even replacement.

“We were able to see the impact (given) the change of temperature,” Al-Qadi said, “because the temperature will make a slab curl upwards during the day when the temperature is high and the opposite (occurs) when the temperature is low during the night.”

Researchers agree that machine learning will allow for better design of airfields and longer life cycles in the future.

“In order to maintain our airports’ functionality and serve the millions of people who are traveling every year by air, we need to maintain our runways and keep our taxiways functional,” Al-Qadi said. “Using (machine learning) will allow us to predict the accurate life of these pavements. This will reduce the downtime of airports significantly, and it will be more efficient and more cost effective.”

Written by: Emily Jankauski

Posted: Sept. 25, 2019