D/FW AIRPORT -- At 8 a.m. today, a computer programmed to think like an air traffic controller will make history at Dallas/Fort Worth Airport by deciding when and where an incoming plane should land.
This is the first test in real air-traffic conditions of a new NASA software program that will calculate -- much more accurately than a human controller -- exactly when an approaching aircraft will arrive over the end of one of D/FW's runways.
Today's test will begin with D/FW's big morning "push" of arriving aircraft, when more than 100 flights will land in less than an hour, said Jo Ellen Casilio, D/FW control-tower manager.
Although live controllers make the final decisions, the computer will, for the first time, choose the runway for arriving aircraft and suggest the order in which approaching planes should land.
Having computers juggle the flow of incoming flights is expected to reduce flight delays by 20 percent and increase D/FW's capacity to land airplanes by up to 30 percent, said Tom Davis, the aerospace engineer leading development of the program for the National Aeronautics and Space Administration.
The system could cut up to two minutes off the flight time of every airliner coming to D/FW, where the average flight delay last year was about six minutes. That two minutes' savings equates to almost $80 per flight for American Airlines, said Bill May, air traffic control coordinator for American, D/FW's largest carrier.
With more than 100 arrivals an hour during D/FW's busiest flight times, that is a savings of $8,000 an hour and potentially more than $100,000 a day for all airlines, officials say.
The computer will allow controllers to safely move aircraft closer together in the crowded skies over the Metroplex. The system's greatest benefit will come during inclement weather, when arriving flights are restricted to simultaneous landings on two runways. In good weather, three runways are used simultaneously for landings.
"This system is going to have the accuracy of the computer, with the common sense of the controller," May said. "With the two mixed together, you have a winner."
What the computer won't do is replace controllers. "Maybe that could come some day, but I don't see it happening in my lifetime," said Mark Pallone, president of the National Air Traffic Controllers Union at the D/FW Terminal Radar Approach Control facility in the base of the central tower. "This is an aid to controllers. It's not set up to take a controller's job."
Full installation and 24-hour use of the system at D/FW is several months away, Davis said. The so-called FAST program, Final Approach Spacing Tool, will be used by D/FW's approach controllers. They are responsible for tracking aircraft by radar within a 75-mile radius of D/FW and sequencing them to D/FW and other local airports for landing.
D/FW Airport was chosen as the pilot site for the system because of its phenomenal growth, Federal Aviation Administration officials said.
But the system, now written only for D/FW, is the wave of the future, NASA officials said. "We expect to have this system installed in 15 to 20 airports nationwide by 2000," Davis said. He declined to speculate on the cost but said the FAA has spent up to $2 million a year for five years for NASA's development of the program.
The system does not rely on the problem- plagued aging computers at the FAA's Fort Worth Air Route Traffic Control Center. The center, just south of D/FW, controls aircraft in flight over much of Texas and five other states until the planes get close enough to the Metroplex for D/FW approach controllers to take over. D/FW's approach computers are newer and have no history of failure, unlike their counterparts at the Fort Worth center.
The trial run of the FAST software comes after months of simulation testing at the FAA's Tech Center in New Jersey and NASA's Ames Research Center in California. Nine controllers specially trained on the new system will work with it for about two hours today, then meet with system installers to evaluate the test.
FAST keeps continuous track of the progress of arriving flights, said Ron Nichol, site coordinator for the project at D/FW.
The computer reads data from the approach controllers' radars for each plane when the aircraft get within 75 miles of D/FW. It then uses specific information, including each aircraft's speed and the winds it will encounter, to calculate when it will arrive for landing.
Then, comparing that information with the same data about all other planes en route to D/FW, the computer will decide which runway each plane should use and in what order.
Now, controllers have to make these calculations quickly in their heads, based on their own experience. "What we're giving them that they don't currently have is predictability," Nichol said. "It's a level of accuracy they can't have now."
But controllers ultimately stay in control. "The controllers don't have to do what the computer tells them. If they don't like what they see, they can turn it off," Pallone said. "But I don't foresee that happening.
. . . It seems to work real well."
Controllers have been included in the planning and helped design the software.
"The whole idea was to write the software to emulate the controllers' thought processes," Casilio said.
The software being tested today is designed for a southerly flow of landing aircraft, which occurs about 70 percent of the time at D/FW. The flow is determined by wind direction, because aircraft generally must land into the wind. If the wind is blowing from the south, aircraft approach the north end of the airport and land heading south.
Software is being written for a northerly traffic flow at D/FW and should be ready for testing in late March, Nichol said.