Cause Analysis and Improving Safety
- From time to time, I see discussion here regarding whether cause analysis is purely "looking backward" or is it (and should it be) "looking forward" as well. This article from today's New York Times is relevant:
New York Times
October 1, 2007
Fatal Airplane Crashes Drop 65%
WASHINGTON, Sept. 30 After two infamous crashes in 1996 that together killed 375 people, a White House commission told the airline industry and its regulators to reduce the domestic rate of fatal accidents 80 percent over 10 years. That clock ended Sunday.
They have come close to reaching that goal. Barring a crash before midnight Sunday, the drop in the accident rate will be about 65 percent, to one fatal accident in about 4.5 million departures, from one in nearly 2 million in 1997.
There have been no fatal airliner crashes involving scheduled flights this year in the United States and just one fatal accident: a mechanic who was trying to close the cabin door of a chartered Boeing 737 on the ground in Tunica, Miss., fell to the pavement during a rainstorm.
Around the world, airliners continue to crash. There have been 7 crashes this year that killed more than 20 people each.
Even so, there has been strong progress internationally. William R. Voss, president of the Flight Safety Foundation, recently calculated that if the 1996 accident rate had remained the same in 2006, there would have been 30 major accidents last year. Instead, there were 11.
This is the golden age of safety, the safest period, in the safest mode, in the history of the world, said Marion C. Blakey, the administrator of the Federal Aviation Administration, in a speech to an aviation group in Washington on Sept. 11, two days before her five-year term ended.
Some of the improvement may be luck, as there is an element of randomness to crashes. But part of the explanation certainly lies in the payoff from sustained efforts by American and many foreign airlines to identify and eliminate small problems that are common precursors to accidents.
Airlines around the world, even in less-developed nations, have also benefited from equipment improvements, like cockpit instruments that help planes steer clear of mountains when visibility is poor, and jet engines that are so reliable that pilots can go through their entire careers without seeing one fail.
Aviation safety experts have uncovered subtle problems. One oft-cited example is a discovery in the last decade by US Airways (then US Air) that many of its planes approaching Charlotte Douglas International Airport in North Carolina were coming in high and hot, too fast and at a steep angle.
As a result, airplanes were conducting unstabilized approaches, meaning pilots had to fiddle with flaps, throttle and other controls just before landing.
The US Airways discovery at Charlotte was something new because the airline did not demostrate it after a crash or from pilot reports.
The airline instead tapped into the system that feeds information to one of the black boxes, the flight data recorder, and siphoned off a stream of data that went to a removable recording device. Then it analyzed flights by the hundreds and looked for unusual patterns, a technique now common with airlines.
Convinced, the F.A.A. changed the approach procedure there, and the airport installed a system to guide planes at a proper angle.
Nearly all unstabilized approaches end with a safe landing, but a study by Mr. Vosss organization found that such approaches were a factor in two-thirds of 76 accidents and serious incidents worldwide during landing attempts from 1984 to 1997. So one focus of the last 10 years has been to look for air traffic procedures that could cause problems.
The Air Line Pilots Association cited another problem that is now being resolved. The airlines pooled their data an action that was itself an innovation on operations at Reno , Nev. , and found that the cockpit system that warns of imminent flight into a mountain often sounded a false alarm.
Aviation experts say that if safety alarms sound falsely too often, they become like the homeowners smoke alarm that is set off by an egg frying in the kitchen people start ignoring it. As at Charlotte with the high and hot approaches, this was a known glitch in the system that had not caused any crashes, but that might someday contribute to one.
The solution in Reno , which is still being developed, is better guidance for pilots to follow flight paths precisely and stay farther away from mountains in the area.
In other places, improvements have been as simple as better signs on taxiways to prevent planes from moving into the path of other aircraft.
Its not one thing. Its a series of small things, said John Cox, who was an Air Line Pilots Association safety representative for 20 years. Many of those small things were minor problems observed in everyday operations, he said, then counted, scrutinized and eliminated before they caused an accident.
Newer planes are also safer. All American airliners, for example, now have enhanced ground proximity warning systems. These systems use the Global Positioning System to compare the planes position against a database of mountains and buildings, and warn of impending collision.
Analyzing data from safe flights is a reversal of the historic practice, which is to go out and kick the tin after a plane crash, looking for clues. Analyzing safe flights is almost all that is left, experts say, as the accident rate falls and there is less tin to kick.
The sample is so small, you wont have effective data sampling, said Hank Krakowski, a United Airlines executive who served as co-chairman of the Commercial Aviation Safety Team. That team is an outgrowth of the White House commission, and it comprises airlines, aircraft builders and pilot unions. (In October, Mr. Krakowski is to become the F.A.A.s chief operating officer.)
Some unions have complained about trends like maintenance outsourcing, in which an airline pays another airline or an outside shop to do crucial safety work, and some government auditors have echoed the concern.
But there have been no fatal crashes in which maintenance error was a cause since January 2003, when a US Airways Express flight, a Beechcraft 1900, went out of control on takeoff because of an improperly rigged tail. Statistically, the era of outsourcing appears to be safer than when airlines did most of the work themselves, although that does not suggest a cause-and-effect relationship.
The decade-long push to reduce the accident rate began with a safety summit in 1996, after the T.W.A. Flight 800 disaster off Long Island and the ValuJet crash in the Everglades of Florida. The summit was convened by the secretary of transportation at the time, Federico F. Pena, who declared a goal of zero accidents.
In 1997, a national commission on aviation safety and security, led by Vice President Al Gore and known as the Gore Commission, concluded that a more realistic goal would be to cut the rate of fatal accidents by 80 percent. Because crashes are sporadic, the goal was stated as the average of the most recent three years.
Despite the safety improvements since then, not all the trends are positive. Airports have lately recorded a disturbing number of what they call proximity events, in which a plane lands on a runway already occupied by another because someone made a wrong turn or a controller made an error.
On July 11, for example, a United plane in Fort Lauderdale , Fla. , took a wrong turn onto a runway where a Delta Air Lines plane was supposed to land; the two came within 100 feet, according to the F.A.A.
Probably the biggest threat of all, today, many, many people agree, is not so much a midair collision as a runway incursion incident, said Richard Healing, an aviation safety expert and former member of the National Transportation Safety Board.
The F.A.A. has a radar system at many airports to warn tower controllers of conflicts on the airport surface, but the system can be confused by puddles on the pavement, which the radar sometimes misinterprets as airplanes. And it warns only the controllers, not the pilots directly.
The F.A.A. is improving the ability to track airplanes on the ground by gradually installing a system that uses a combination of radar and other means, including one that uses multiple antennas to listen for radio beacons on the plane and, by triangulation, calculate its position.
But the safety board argues that even if the new system works as designed, it is still inadequate because several seconds will elapse from the time the system sounds an alarm to when the controller sees it and issues instructions to pilots.
The F.A.A. is experimenting at Dallas-Fort Worth with runway status lights, embedded in the pavement, that flash at pilots when a runway is occupied.
As the number of flights increases, the rate of crashes has to decline or the absolute number of crashes will rise. And as airports get busier, the risk of a crash on the ground increases.
Adding to the problem is that airliners are getting smaller, and a new class of very light jets, seating four to eight people, is entering service. Some of those may be flown by a single pilot who is not a professional, but they will fly at the same altitudes as airliners.
The F.A.A. is facing challenges as it handles ever more traffic. It wants a new air traffic system that can squeeze planes closer together. It wants more reliance on user fees instead of taxes on passenger tickets, cargo and fuel.
But Congress has not agreed. It has approved only a temporary extension of current taxes. And although the F.A.A. administrators five-year term has expired, the White House has not named a candidate it will try to get through the Democratic Senate.
The aviation system continues to evolve, with new runways, new terminals and new towers.
In mid-September, the F.A.A. opened a new tower at Washington Dulles International Airport . It will handle 25 million to 26 million passengers this year, but the airports managers estimate that traffic will double by 2025. The number of runways will go to five from three, and midfield concourses will double to four.
The new tower replaces the signature Eero Saarinen model of the early 60s, which is perched next to the sweeping roof line of the terminal. It can house up to a dozen working controllers comfortably; the old one was a squeeze for nine.
At 25 stories tall, it lets controllers see even small jets between the terminals. The older, shorter tower required them to strain to see some planes taxiing between terminals.
With the regional jets, wed see the top of the tail through the air-conditioners, said David Bridson, a controller.