Upgrade of Air Traffic Control System Stuck in Holding Pattern
After an Aeromexico jetliner collided with a light plane over Cerritos in 1986--one of the deadliest air disasters in Southern California history--government safety investigators pinpointed an unlikely cause: the U.S. air traffic control system.
The crash came as the government was in the early stages of developing a state-of-the-art computer system that was supposed to let controllers better handle air traffic in the nation’s increasingly crowded skies.
Today, that computer project is itself a disaster--a case of bureaucratic bumbling, corporate failings, repeated delays that have set back the project’s completion at least two years and one of the biggest potential cost overruns in U.S. government history: $2 billion and counting.
Even the official in charge of the $5-billion project, U.S. Transportation Secretary Federico Pena, recently called the snafus “appalling and unacceptable,” vowing that the Federal Aviation Administration will take “any and all steps necessary to get this project on track,” if it isn’t canceled first.
At issue is far more than another government contract run amok.
More than 500 million airline passengers each year put their lives in the hands of U.S. air traffic controllers and their equipment, a figure the FAA expects to soar to 750 million by the end of the decade. Already, the agency estimates, the U.S. system handles half the world’s commercial airline traffic.
The FAA says the new computer network, known as the Advanced Automation System, is a must if controllers are to handle that surge while maintaining safety. Many of the system’s existing computers are hopelessly antiquated at 20 years old. The improvements are being paid for largely by airline passengers themselves, via an 8% federal ticket tax that goes into an aviation trust fund for capital improvements.
At the moment, however, the overhaul is only getting more muddled.
The FAA just replaced the project’s top managers, and a pending review is likely to bring more big changes to the project’s blueprints to rein in runaway costs. The IBM division building the system was just sold.
All of this is occurring against a backdrop of White House and congressional debate over whether the computer system’s snafus are proof that air traffic control should be turned over to a semiprivate corporation.
The project is in such disarray that FAA chief David Hinson told a congressional hearing earlier this month that simply canceling the program is “an option that’s clearly in front of us.”
But Hinson also said he is pushing ahead to complete the computer overhaul once its problems are fixed--a strategy backed by the nation’s airlines and controllers.
The current system “is being stretched to the margins,” said Will Faville, director of safety and technology for the National Air Traffic Controllers Assn., the controllers union. “We certainly need new equipment, and we need it yesterday.”
To be sure, the existing system is still the safest in the world because the FAA keeps airplanes spaced far apart, according to Robert W. Poole Jr., president of the Reason Foundation, a Los Angeles think tank that advocates privatizing air traffic control.
“The safety valve is that they slow everything down whenever (the system is) stretched,” Poole said. “That more or less works, but we pay a big price in terms of the number of hours of delays that we encounter as air travelers.”
With passenger traffic rising, the delays will only get worse unless the new computers are installed, said Philip M. Condit, president of Boeing Co., the world’s largest maker of jetliners. “The technology is available today . . . to put a lot of capacity in the system. We need to get on and do it.”
According to the National Transportation Safety Board, nearly 1,000 people have died in crashes of major U.S. commercial airlines since 1983. Last year, however, the figure was zero. The FAA began to consider modernizing the air traffic control system 13 years ago, but the plans took on added urgency eight years ago on a quiet Labor Day weekend, as Aeromexico Flight 498 prepared to land at Los Angeles International Airport.
The controller guiding the DC-9’s descent was briefly distracted when a light plane appeared on his radar screen, seeming to threaten the big jet’s airspace. Moments later, the Aeromexico jet collided with a second small plane that the controller later said he never saw.
Both planes plunged 6,500 feet into a residential neighborhood in Cerritos, killing 82 people, including 15 on the ground.
The NTSB concluded that the crash cause was not caused by human error but by “systemic” failures in the air traffic control system.
Among them: equipment limitations and outdated traffic control procedures that forced pilots to rely too heavily on the old-fashioned “see-and-avoid” concept, which presumes them capable of spotting other planes in time to take evasive action.
Had the new system been in place in 1986, there is no guarantee it would have prevented the Cerritos disaster, according to Faville. Yet that crash, and others blamed on air traffic control problems, dramatically illustrated the need for a new system, FAA officials say.
One year after the NTSB determined the cause of the crash, the FAA awarded the initial $4.7 billion in contracts to install the Advanced Automation System. The biggest chunk, $3.6 billion, went to IBM’s Federal Systems division.
That unit--recently sold to Loral Corp., a defense electronics concern--was supposed to put a powerful, full-color computer workstation in front of each controller. The workstations would provide advanced radar, weather and flight data that would be seamlessly fed into the computer’s software.
But it would be one of the most daunting and complex commercial computer tasks ever undertaken--even for IBM.
“It is leading-edge” technology, said one of the managers involved in the program, who asked not to be identified. “It’s a tricky thing to build, and it’s not clear IBM really understood where all of the pitfalls were ahead of time.”
The IBM (now Loral) group, which has 1,000 people working on the system in Rockville, Md., was to have developed the first workstation prototype by now. It has not, and the first machine to be installed, at an FAA site in Seattle, is not scheduled to begin operating until mid-1998. The rest of the system probably won’t be in place until the turn of the century.
Delays, of course, cost money. The FAA now estimates that unless corrective actions are taken, the modernization will cost at least $2 billion more than originally planned only six years ago.
The agency’s chief counsel, Mark Gerchick, vowed that the FAA will fix the project’s woes in time to head off those kinds of cost overruns. “We are not going to have this projection become a reality,” he said. “We could not be more serious about that.”
But the FAA itself has played a sizable role in the delays by making several changes to IBM’s system.
For example, the agency had once hoped to consolidate the more than 200 air traffic control facilities into 22 sites and to upgrade all airport control towers with the new computer system. Both plans were scaled back sharply.
Now the FAA is making more changes to fix AAS. Hinson recently named Robert M. Valone, a former official of the National Oceanic and Atmospheric Administration, to take over the troubled project. Hinson is also demanding that Loral guarantee it will get the project done on time and within budget from now on.
Responds Loral Chairman Bernard L. Schwartz: “We understood the extent of the problems” before buying the division, “and we’re happy to step up to those responsibilities.” (IBM declined to comment.)
In the middle of this morass are Schwartz’s new employees in Maryland who are actually designing the system and struggling to end the controversy. “There’s no question there’s been distractions,” said one source who’s worked with the team.
The computer team is also being shadowed by Hughes Aircraft Co., which, having lost a bitter battle to IBM for the original AAS contract, is poised to rescue the project if Loral stumbles.
At a congressional hearing earlier this month, a Hughes group vice president, Robert H. Kramp, said his company “stands ready to help in any way we can.”
The White House, the airlines and others think part of the solution is to take air traffic control from the FAA and turn it over to a quasi-private company that would run it more efficiently. But some members of Congress, citing safety concerns, object to the idea, despite the AAS debacle.
Rep. William F. Clinger Jr. of Pennsylvania, the ranking Republican on the House aviation subcommittee, said that despite the “horrendous (cost) overruns and time delays,” he is skeptical that taking air traffic control from the FAA would make the system more efficient without sacrificing safety.
The nation’s 14,700 controllers are stationed not only in airport towers, but also in dozens of “terminal radar approach control centers” located near airports and in the 22 “en route” centers in the United States, which guide aircraft over the vast stretches of airspace between airports.
The en route center serving Southern California is in Palmdale, and a visit there reveals why the system needs a face lift.
At peak shifts, as many as 85 controllers sit in a room the size of a high-school gymnasium, its ceiling two stories high. The room is kept dark so controllers can read their monochrome green display screens.
The controllers must also use old-fashioned paper strips that provide flight data on each aircraft they’re guiding--the type of plane, its speed, altitude, planned route and more.
Under AAS, those details--along with weather data and a variety of other information--would be electronically fed into the controllers’ computers and shown in color on their main display screens. AAS is also supposed to give the controllers exceptional reliability. Pieces of the existing system now crash at random, but AAS is expected to fail no more than a total of three or four seconds a year.
But the question now is whether Loral and the FAA can turn the project around--assuming it is not canceled. Clinger, for one, believes the project will be finished, even if it’s a shadow of its original design.
“We have made such an enormous investment in this system,” the congressman said. “The hope is that we can correct it rather than deep-six it.”
Revamping Air Traffic Control
* 1981: The Federal Aviation Administration begins looking at ways to modernize the U.S. air traffic control system.
* 1984: The Federal Systems division of IBM and Hughes Aircraft Co. begin competing to design the key phase of the modernization effort: an upgrade of air traffic control’s computer network, known as the Advanced Automation System.
* 1988: IBM wins a $3.6-billion contract to develop AAS. Hughes protests the selection, but the award stands.
* 1992: The FAA, noting that the project is behind schedule, warns IBM to fix the problems or face the loss of the contract.
* 1993: FAA chief David Hinson announces a complete review of AAS, saying the project is “years behind schedule” and faces cost overruns of at least $1.2 billion.
* Jan., 1994: The Clinton Administration unveils a proposal to turn over air traffic control to a semiprivate corporation.
* March: The FAA replaces AAS’ senior managers and boosts the estimate of AAS’ potential cost overruns to more than $2 billion unless the program is fixed. Loral Corp. buys the IBM Federal Systems unit.
* April: The FAA’s Hinson tells a congressional panel that canceling AAS is an option but that he prefers to see it finished. Decision will come after the FAA review of the project, to be completed in May.
