Causes of Similar 737 Jet Crashes Still a Mystery

WASHINGTON —  The two plane crashes were eerily similar. And thus far the searches for what caused them have proven similarly frustrating.

The first crash--near Colorado Springs, Colo., on March 3, 1991--involved a Boeing 737 jetliner that was preparing to land. Suddenly and without warning, United Airlines Flight 585 banked steeply to one side and plunged nose-first into the ground, killing all 25 people on board.

The National Transportation Safety Board--which in 27 years has solved more than 1,000 transport-aircraft crash mysteries--supplemented its own personnel with experts from the airline, the pilots’ union, the National Weather Service, the Federal Aviation Administration and the firms that manufactured the plane and its components in an exhaustive effort to figure out what went wrong.

They examined the wreckage, studied readouts from the plane’s two “black box” recording devices, interviewed witnesses and used computer-generated “flights” in cockpit simulators in an effort to re-create the flight path of the doomed jetliner.

As the weeks stretched into months, the NTSB found it was coming up with more questions than answers. And in the end--for only the fourth time in the agency’s history--the crash detectives had to admit defeat.

“It was an equation with too many unknowns,” said John Lauber, the NTSB member who headed the Colorado investigation. “And there will always be the lingering fear that we overlooked something.”

Then, on Sept. 8 of this year, another Boeing 737 crashed while preparing to land--this time near Pittsburgh, Pa. Once again, suddenly and without warning, the jet banked steeply to one side and then plunged nose-first into the ground. USAir Flight 427 carried 132 passengers and crew members, and again there were no survivors.

The NTSB launched another investigation, taking note of the many similarities between the two accidents. Now, more than six weeks into the new inquiry, investigators have come up with a few theories, but they have to admit that the Pittsburgh crash has them stymied.

Carl Vogt, the NTSB member heading the Pittsburgh investigation, is optimistic.

“As of now, it’s an unsolved crash,” he said. “But we’re confident we’ll figure this one out.”

Barry Schiff, a veteran airline pilot and respected air-safety consultant involved in the investigation of a number of crashes, is not so sure.

“Dollars to doughnuts, they won’t find the cause,” he said.

Schiff’s pessimism stems from the obliteration of evidence in the high-speed impact of Flight 427 and from his own studies of the Colorado crash, an accident that still confounds the experts. After more than 20 months of detective work, the NTSB admitted on Dec. 8, 1992, that it “could not identify conclusive evidence to explain the loss of United Airlines Flight 585.”

The best the federal agency could do was identify the “two most likely events” that could have spun the plane out of control and hurled it to the ground.

One of these was a “rotor wind”--a horizontal tornado created by the strong, gusting winds that were spilling over the crests of the Rocky Mountains that day.

However, no such wind ever measured has been strong enough to knock down an airliner, and the board said “too little is known about the characteristics of such rotors to conclude decisively whether they were factors in this accident.”

The other possibility cited by the board was a malfunction of the systems that move the plane’s rudder--the large, hinged slab on the vertical part of the tail that helps a plane turn left or right by pushing the tail in the opposite direction. Full deflection of the rudder could cause a plane to roll and then plunge nose-first toward the ground.

But there are difficulties with this scenario too. The board conceded that although it had identified problems with the 737’s rudder systems, there were no problems that could not “easily” have been countered by the cockpit crew.

More than 3,000 Boeing 737s are flying throughout the world today, carrying more passengers than any other aircraft, so the NTSB’s lack of a firm conclusion about what went wrong in Colorado Springs is of more than casual interest to the aviation industry.

It also worries pilots.

“There’s a lesson to be learned from every crash, and when you don’t find out what happened, you haven’t learned that lesson,” Schiff said. “As a pilot, you think: ‘Gee, that could happen to me.’ It’s kind of spooky.”

Nonetheless, Lauber said air travelers should not be overly concerned about the plane.

“Because of the attention focused on the plane and its systems after Colorado Springs, the public should be reassured,” he said. The (Boeing 737) was examined with great diligence and effort. . . . It’s a good plane.”

The pilots of Flight 427 apparently had no problems with their Boeing 737 as the plane took off from Chicago on the evening of Sept. 8 and headed southeast for Pittsburgh on the first leg of a scheduled trip to West Palm Beach, Fla.

The weather was good, with calm winds, and the jetliner was making a routine approach to Pittsburgh International Airport when one of the pilots, during a conversation with air traffic controllers, suddenly shouted, “Oh God!”

At that same moment, the plane rolled to the left and plummeted straight down, slamming into the ground with such force that the aircraft and its contents were torn to pieces.

Within a few hours, an NTSB team had reached Pittsburgh, and by the next morning the investigators made their first visit to the crash site--a shallow hillside gully near the town of Aliquippa. Joining them there were specialists from the airline, the pilots’ union and other parties with close interest in the crash.

The personnel were divided into subgroups. One group began studying the plane’s engines, another looked at its control systems, a third studied the meteorology, a fourth interviewed witnesses, a fifth talked with air traffic controllers, a sixth analyzed the wreckage to determine how the plane broke up, and others reviewed maintenance records and the history of the crew, the airline and the plane.

As they pursued their tasks, heavy rains pounded the hillside, slowing the gruesome job of extracting the fragmented remains of the victims from the shattered debris.

That night, the investigators gathered at a local hotel to discuss their early findings.

“The leader of each group gets up and says: ‘Today we found this,’ ” Vogt said. “That way, everyone knows what everyone else knows.”

“One person will come up with a scenario, and another person will try to shoot it down,” said a source close to the investigation who asked not to be named. “It’s a process of elimination. If you can shoot one down, then there’s one less avenue that you have to pursue.”

That’s how a scenario involving sudden failure of the left engine--which could have rolled the plane to the left--was ruled out. The recovered engine’s fan blades were bent, and dirt had been sucked into an inner turbine, showing that it was still spinning, under power, when the plane hit the ground.

Inappropriate deployment of an engine thrust reverser was discounted in much the same way when actuators on the left engine’s reversers--the only ones that could have rolled the plane to the left--were found in the stowed position.

Weather problems were ruled out, as were in-flight breakup and the likelihood of a bomb, which leaves tell-tale residue and characteristic metal distortions near the point of explosion.

By the end of the investigation’s second day, Flight 427’s two black box recording devices--which actually are painted orange--had been recovered from the wreckage and shipped to the NTSB laboratories in Washington. Word came back quickly that both had survived the impact in good shape and were providing clear data.

One of the boxes--the cockpit voice recorder--contained a tape recording of all the sounds in the cockpit during the last 30 minutes of the flight. In addition to the discussions and remarks of the pilot and co-pilot, the device recorded their conversations with air traffic controllers and a number of clicks, hums, thumps and other noises that can provide clues to what was going on.

For example, in the investigation of a 1987 Continental Airlines jetliner crash in Denver, engineers determined the takeoff speed by analyzing the sound of the spinning nose wheel.

The voices picked up by the cockpit recorder are often faint, garbled or distorted, requiring painstaking electronic enhancement by NTSB engineers. Listening to the desperate last moments of someone about to die “is not a job for everyone,” said Alan Pollock, an NTSB spokesman. “You often hear things that are pretty unpleasant.”

The other box--the flight data recorder--collected 17 different types of information about the plane’s performance during the final minutes of flight. The data included readouts on the plane’s altitude, heading, airspeed, engine performance, pitch, roll and acceleration, both vertical and horizontal. While the recorder also contained information on the position of the control yoke, it lacked data--available on newer recorders--about other control settings.

The readings from the flight data recorder have been programmed into a 737 cockpit flight simulator at the Boeing Co. plant in Seattle, creating a full-scale replication of how the USAir jet behaved during its last moments of flight.

Engineers are now feeding in every combination of control failure, autopilot malfunction, weather phenomenon, engine failure and pilot error that they can think of in an effort to match the replication generated by the recording.

If they can come up with a match, it would provide strong evidence of what went wrong. Thus far, they haven’t, but they’re still trying.

“So far, there are a number of theories,” another informed source said. “Most of these center around the possibility that the plane encountered severe wake turbulence from another plane, or that something went wrong with the flight controls--perhaps the rudder. Maybe it was some combination of the two. Maybe the pilot just didn’t have time to react properly.”

The scenario involving wake turbulence--cyclonic winds that spiral back from the wingtips of large aircraft--arose after investigators learned that the USAir jetliner was flying below and four miles behind a Boeing 727 as the two planes approached the Pittsburgh airport.

Once again, there’s a hitch.

The phenomenon has caused smaller planes to crash--most recently a small corporate jet that was trailing a Boeing 757 as the two planes approached John Wayne Airport in Orange County last December. But planes the size of a 737 routinely fly behind other jetliners, and wake turbulence has never been known to down any of them.

One source said that for wake turbulence to flip over a 737, the 737 probably would have to encounter the spiraling vortex at a very precise angle that would raise one wing and lower the other--”a very long shot,” as he put it.

“But wake turbulence could be a factor,” Vogt said.

As for rudder system failure, the NTSB noted that after the Colorado Springs crash, regular inspections of 737 rudder systems were mandated by the FAA. The rudder on the plane in the Pittsburgh accident had been checked four times this year--most recently less than a month before the crash--and no problems were found.

Investigators are scrutinizing the rudder components recovered from the wreckage. Thus far, no incriminating evidence has been found, but the suspicions continue and study goes on.

The wreckage of the plane--some of it in pieces no larger than a dime--has been spread out like a mosaic on the floor of a hangar in Pittsburgh to form a two-dimensional reconstruction of the jetliner.

“They recovered 97% to 99% of that plane,” a source said. “They’re looking for anything abnormal.”

H. Keith Hagy, an accident investigator with the Air Line Pilots Assn., said the evidence sometimes is minuscule, such as tiny hairline cracks visible only under a microscope, that are left “like rings in a tree” every time a thin piece of metal is bent.

Cracks like these suggest that a part may have failed from metal fatigue before the plane crashed, Hagy said, while evidence of stretching or tearing indicate that the part failed on impact.

Other, larger cracks may indicate that an entire fuselage is about to burst open under the repeated stress of cabin pressurization. That’s what happened to an Aloha Airlines jet a few years ago in Hawaii, when a section of the fuselage popped off, leaving passengers riding in the open at 300 m.p.h.

Even the bodies of the victims can provide clues to why the crash occurred. If an autopsy shows a man was killed on impact but there is soot in his lungs, that indicates there was a fire on board before the plane crashed, while the victim was still breathing.

Some evidence may not turn up until months after a crash--like the flawed engine shaft from a DC-10 that an Iowa farmer unearthed while plowing. Some--like the ice that doomed the Continental Airlines jet in Denver in 1987--melted away in the morning sun.

And some may turn out to point in the wrong direction. After an American Airlines DC-10 crashed in Chicago in 1979 when an engine fell off, NTSB Vice Chairman Elwood Driver announced that investigators had found a missing bolt on the airfield grass that was thought to be the culprit. It wasn’t.