Pilots’-Eye View


When a big jet crashes, pilots analyze and debate the causes. Naturally, professional aviators are scrutinizing American Airlines Flight 587, the European Airbus A300 that on the morning of November 12 had barely left the ground when it broke apart and plummeted into the Rockaways. All 260 people aboard died, as well as five people on the ground.

In this case, the debate focused on pilot training. “I’m very upset by the training,” one retired airline pilot told the Voice. (None of the four pilots interviewed for this article wished to be identified.) “I don’t think we’re turning out nearly as good pilots as before; even in the military you’re [more] a computer programmer” than an aviator. Referring to the violent shaking and loss of control that apparently happened after the Airbus encountered the wake left by a Japan Airlines 747, he said, “When things go to hell, it’s still something you’ve got to control in three axes—pitch, roll, and yaw.”

Another pilot agreed, complaining that pilots, trained by rote, are unprepared for sudden emergencies. He also cited what he saw as a disturbing trend toward the increasingly automated environment in contemporary cockpits. “They’ve thrown all the basic flying skills out the window—young pilots today aren’t trained to solve problems,” he said. “We had detailed engineering courses—we could build the thing—but today it’s a comic book.” Several examples came up during pilot interviews of tragedies that should have been avoided—for instance, Swissair 111, which crashed off Nova Scotia in 1998 after the pilot delayed an emergency landing to dump fuel because the plane exceeded the recommended maximum landing weight. “He should have landed heavy,” said one pilot in astonishment at such an error.

The pilot in control of the American Airlines jet was 34-year-old Sten Molin, the designated co-pilot on the routine run to Santo Domingo. The Connecticut Post has reported that the Greenwich resident trained at a university flying school, was used to the route, and loved to fly. On September 11, he witnessed the second jetliner crash into the World Trade Center from his cockpit seat as he waited in line at JFK to take off for that morning’s flight to Santo Domingo.

“The complaint is they’re trained to the instruments, and when something goes wrong, they aren’t used to the feel of the aircraft,” said Mary Schiavo, former inspector general of the U.S. Department of Transportation and a fierce critic of airlines and federal regulators. Schiavo allowed that the pilots the Voice interviewed have a point—for example, these days major airlines will hire pilots with significantly fewer hours in the air than they once did. (Molin had been flying for American for 10 years.) But as to automation in the cockpit, Schiavo said, “Aircraft are changing, and it won’t be long before it is a flying computer.”

In evident frustration at yet another tragedy in the air and, presumably, at another blow to consumer confidence in airline safety, the retired pilot said air traffic control was partly to blame, for imposing on pilots what he called a double standard. He explained that commercial jet pilots concentrate on their instruments, especially on takeoff and landing; that’s part of what’s called Instrument Flight Rules (IFR), under which pilots are guided by air traffic control. Yet in clear weather they are also expected to look out for obstacles, such as other aircraft. This is called “see and be seen.” “You can’t always be ‘see and be seen’ and fly IFR departures at the same time—you just can’t do that,” he said. Regarding Flight 587, he said, “Controllers have a responsibility to see that you stay away from that turbulence.” On this point Schiavo agreed, suggesting Kennedy controllers should have ensured there was more space between the two aircraft. She said, “Probably the FAA will have some explaining to do.”

At this early stage in the investigation it’s not known if the American Airlines tragedy could have been avoided. But after taking off that morning in clear, calm weather, the aircraft apparently survived two wake-turbulence episodes with all its major structure intact. Then came violent side-to-side movements (yaw), and the tail fin and rudder, used to keep the aircraft pointed forward like an arrow and to control yaw, ripped away from the fuselage.

Reports in The New York Times have suggested that the pilots may have overstressed the rudder in an attempt to control these movements, helping to cause the structural failure. But other reports have made clear that the tail fin itself is suspect. It was first repaired when a weakness was noted by American after the plane was delivered from the French factory in 1988. Six years later it sustained possible damage after an encounter with severe turbulence near Puerto Rico. Schiavo had no doubts where investigators should be looking. “There’s no way that anything the pilots could have done would have ripped off the tail,” she said. Noting that Airbus has had problems with a dangerous phenomenon called “flutter,” in which the tail fin vibrates uncontrollably, Schiavo wondered whether a spasm of flutter in the fin might have exacerbated pre-existing flaws. “I think we’d better focus on the tail,” she said.

Leaving aside questions about possible fatal structural weaknesses, in which case even a Chuck Yeager couldn’t have saved the plane, what would the pilots interviewed by the Voice have done differently to control the craft? When the plane was first buffeted by wake turbulence, the retired pilot said, “I like to think I’d have gone to the right—the wind was from the right.” Such a move, he said, would have taken the plane through the turbulence to calm air, leaving enough air space to turn left again to avoid flying over the heavily populated areas of Brooklyn, thus following a standard noise-abatement departure from runway 31 L at Kennedy Airport. In fact, the Airbus did not turn after the first jolts from turbulence.

Other questions may yet assume importance in the investigation. For example, on one occasion an Airbus rudder did not respond properly. And there have been episodes in which software was suspected of causing catastrophic problems. Most notable is the crash of a Kenya Airways A310 whose engines inexplicably failed when the plane was only 200 feet in the air, moments after taking off from Abidjan in the Ivory Coast. All but 10 people onboard died. That accident happened in January 2000, yet is still shrouded in mystery. According to an article in the Kenyan Daily Nation, witnesses saw the plane “wobble” in the air before breaking up; some witnesses to Flight 587 also saw the plane wobble. The article quoted pilots who speculated that the Airbus may have had a software problem that sent uncoordinated signals to various systems. The aircraft had been in service 14 years, just a year longer than the A300 that went down in Queens.

The National Transportation Safety Board has not raised the question of a computer malfunction in the case of American Flight 587, and last week NTSB spokesman Ted Lopatkiewicz told the Voice that the autopilot was switched off for the entire 103-second flight. But Airbuses in any case are highly automated, ironically in order to guard against misjudgments by pilots. A computer glitch would presumably absolve the pilots of blame for the crash.

Then there’s the far-out crash scenario of sabotage, which has not been ruled out.

Last week, NTSB investigators focused on the visible evidence: the severed tail fin and the possibility that the non-metallic composite materials used in its manufacture had failed, and how that failure may have related to the wake turbulence the plane encountered.

The FAA wasted no time in issuing an order called an emergency airworthiness directive to airlines to inspect the tails of their Airbus A300 and A310 fleets. The order went out on Friday; by Tuesday, the Times reported that American had finished the inspections. Schiavo, noting that the inspections can be “extremely cursory,” said, “It’s hard to see weaknesses in the structure with these inspections.”

The inspections were, of course, too late for the Flight 587 Airbus, a plane that already had a history of problems with its tail fin. The Times has reported that repairs were made to strengthen the area where the fin, or vertical stabilizer, attaches to the fuselage when the plane was first delivered. Then in November 1994 the same aircraft came to FAA notice when it flew through severe turbulence near Puerto Rico, causing dozens of injuries and possible damage to the tail fin.

According to Alison Duquette, an FAA spokeswoman, the routine in such an incident is to conduct a one-time inspection of the systems and structures. “They look at it, basically, and if something’s wrong, it’s taken out of service and fixed,” she said. We do not know what, if anything, American found when it examined the tail, or what action it took; the plane’s maintenance records are now part of the NTSB investigation.

But it looks as if whatever was done, it was not enough. “Vertical tails just don’t come off,” said Bill Kauffman, professor of aerospace engineering at the University of Michigan and an air-safety activist. In the case of Flight 587, he said, “the implications are enormous. Either the [FAA’s] certification or inspection process was faulty, and these are fundamental things.” Kauffman said the crash appears to mark the first structural failure involving composite material. “This was the first major application of composites to a big chunk of structure, other than flaps or slats,” he said.

Boeing now makes the entire tail section of its 777, which first went into service in 1995, out of composites. The company makes grandiose claims for its composites, which are plastics reinforced with carbon fibers: that they are as strong as metal, more resistant to fatigue, and easily repairable. But this is a relatively new material with, presumably, new ways of failing—for example, the composite is made in layers that can come unstuck in a process called delamination.

“When it comes apart, those problems are difficult to fix,” said John Cataldo, formerly a crash investigator at the Naval Safety Center. “We had that on navy planes.” Regarding Flight 587, he said, “I’d be suspicious of the previous damage.” Indeed, it is almost unheard of for an aircraft to lose a tail just because of turbulence.

Pilots know they have to be wary of turbulence from large jets. Light planes are most at risk, but larger passenger aircraft are not immune—the Aircraft Owners and Pilots Association manual points to the example of a DC-9 that got too close to a larger DC-10 on takeoff, caught a wing tip on the ground, and cartwheeled, killing everybody aboard.

Yet pilots disagree on whether the turbulence could have been sufficient to fatally damage any part of American 587’s structure. One pilot, concurring with Kauffman and Schiavo, said, in effect, no way. “It’s not uncommon to encounter wake turbulence,” he said. “But what occurred here defies my understanding of aerodynamics and structure—for the tail to come off under relatively benign conditions.” But for the retired airline pilot, there is no question that it is possible. “I’m not a bit surprised it did shake the fin loose,” he said. A former air force pilot who flew B-52 bombers, he remembers a similar incident in which a B-52 crew managed to land the plane safely despite the loss of its vertical stabilizer.

As it happens, NASA has developed software to assist pilots in controlling an airplane solely by varying the thrust from engines. The software project began at the NASA Dryden research center in Edwards, California, after a DC-10 crash-landed in Sioux City, Iowa, having been flown for miles after an engine explosion severed all the hydraulic lines, so none of the control surfaces could respond to commands from the cockpit.

“We developed the software and flew it on two different types of aircraft with control surfaces locked,” said Dryden spokesman Alan Brown. Brown said the program, called PCA for “propulsion-controlled aircraft,” was offered to manufacturers, but there were no takers. He cautioned that PCA was designed for use on an aircraft with an intact fin. Taking the fin off would make the plane even harder to control. Still, Brown said, Dryden might have continued the research if the industry had shown any enthusiasm. “We were interested in taking it further,” he said, “but it’s up to the manufacturer to apply the technology.”