Why Did Jetliner Crash Near Pittsburgh? Check Incident Over Honduras

Copyright 1994, The Seattle Times Company

While flying a Boeing 737-300 passenger jetliner from Houston to Honduras last April, Continental Airlines Captain Ray Miller heard a muffled boom.

In the same instant, Miller felt the aircraft suddenly twist and roll violently to the right.

Miller disengaged the autopilot and turned the control wheel sharply to the left, holding it firm against stiff resistance. This deployed wing panels, called ailerons, to turn the plane left and thus counter the mysterious, insistent pull to the right.

For the next 18 minutes, as the 737 descended from 37,000 feet to a safe emergency landing, Miller fought to control the airplane. On board were more than 120 passengers and crew.

"Fortunately it was a very clear day and smooth. I feel if I had the normal low-level bumpy air I would have possibly gotten behind the control response and lost control of the aircraft," Miller wrote in a special report to the National Aeronautics and Space Administration.

Five months after Miller's April 11 flight, USAir Flight 427, also a 737-300, suddenly rolled over and dove about 4,000 feet into a rugged ravine near Pittsburgh, killing all 132 people on board.

The Sept. 8 Pittsburgh crash remains under investigation by the National Transportation Safety Board with no focus on a possible cause. Yet there appears to be little urgency on the part of the safety board to examine the strikingly similar Honduras incident for clues.

"If you substitute Honduras for Pittsburgh, you'd have the same event, only it happened at a higher altitude and this time the pilot lived to tell how it happened and what he did to save the aircraft," said Dennis Lods, a jetliner-systems expert with Sterns, Walker & Lods, a San Francisco law firm which specializes in aviation accidents.

The Continental jet and the USAir jet rolled off Boeing's Renton factory floor within months of each other in 1987.

Boeing investigated the Honduras incident and found problems related to two parts: a "yaw damper," which has disrupted 737 flights since the late 1960s, and a rudder-control actuator that the Federal Aviation Administration five weeks earlier had ordered to be gradually replaced in all 737s. The actuator can jam or move the rudder in a direction opposite to the pilot's command.

The NTSB took no active role in the Honduras investigation, even though the safety board has been closely tracking 737 rudder problems since the March 3, 1991 crash of a United Airlines 737-200 near Colorado Springs.

The NTSB's most experienced 737 control-systems expert was briefed on the Honduras incident by Boeing and Continental, but did not see Miller's report until provided with a copy by The Seattle Times, said an NTSB source, who asked to remain anonymous.

"We really haven't gotten far enough into the Pittsburgh investigation to characterize what it (the Honduras incident) may be related to," the NTSB source said.

The committee of industry experts investigating the Pittsburgh crash, including representatives from USAir, Boeing, the FAA and the Air Line Pilots Association, is concentrating on analyzing the wreckage of USAir 427, the source said.

The committee will gather for a closed-door meeting in a few weeks and may discuss the Honduras incident at that time, the NTSB source said. It is not expected to arrive at a consensus about what caused the Pittsburgh crash until late next year.

"I don't have the feeling that it (the Honduras incident) is any more urgent than the course we're taking now," the NTSB source said.

Some safety experts disagree.

The 737, the most widely used jetliner in the world with more than 2,400 jets in service, has a long history of autopilot and rudder-system malfunctions. Miller's account, some experts contend, should have been the catalyst for issuing a formal warning to pilots that a 737 can suddenly go out of control.

"To me, when the pilot says he could have lost the airplane if he had been in less than clear weather, this is the kind of information that should have been put out through pilot-alert bulletins," said Leo Janssens, president of Aviation Safety Institute, a Worthington, Ohio-based aviation safety advocacy group.

Janssens believes USAir 427 Capt. Peter Germano and First Officer Charles Emmett III "might have reacted faster and perhaps better" had they been made aware of Miller's experience.

The low-key handling of Miller's report highlights how airplane manufacturers and airlines, rather than oversight agencies, can control the thrust of investigations. This approach often increases public risk, said Mike Hynes, an independent aviation-accident investigator from Oklahoma City.

"I'm a pilot, so I do take and live with calculated risks," said Hynes. "But I don't think the average person on the street would accept the risk the industry is asking him to accept."

Hynes pointed out that NASA Air Safety Reporting System (ASRS) reports are submitted voluntarily, typically by pilots, with no provisions to regularly circulate the documents among safety authorities, much less among other pilots.

Miller did not respond to interview requests made through the International Association of Continental Pilots.

Officials from Boeing and Continental, which operates a fleet of 105 737s, declined to be interviewed about the Honduras incident.

The 737 has a history dating back two decades of uncommanded movements of the rudder, the large movable slab on the vertical tail section that controls the left/right position of the aircraft.

Two areas of the rudder system have been identified as causing the rogue movements. The first is an electronic device called the yaw damper, which automatically compensates for the tendency of the tail to oscillate during flight by continually adjusting the rudder slightly one direction or the other.

The other is the rudder actuator Power Control Unit (PCU), the main mechanical device used to move, or deflect, the rudder.

Cables connect the PCU to foot pedals in the cockpit, which a pilot depresses to operate the rudder; electrical wire links the PCU to the yaw damper, which deflects the rudder by sending electronic signals directly into the PCU.

After Miller's successful emergency landing, Continental removed the jet's cockpit-control panel, accessories unit and flight computers, along with the yaw-damper coupler and rudder PCU. The parts were then turned over to a Boeing investigative team, led by technical-support engineer Larry Moore, according to a source close to the investigation.

On May 11, Moore gave Continental an engineering report essentially clearing the autopilot and cockpit controls and blaming the incident on a malfunction involving the yaw damper and the PCU.

However, Boeing's analysis showed the malfunction should have caused an uncommanded rudder movement lasting no longer than 110 seconds, the source said. That conflicts with Miller's statement that he fought for control of the airplane for 18 minutes and believed he was fighting the autopilot.

FAA spokesman Dave Duff said Boeing's tests showed the incident was caused by hydraulic fluid leaking from the PCU and disrupting the electronic operation of the yaw damper, which deflected the rudder "3 degrees." Duff could not explain the 16-minute discrepancy.

According to Duff, the PCU leak identified in the Honduras incident is "not related" to the PCU leak described in the FAA's order, effective March 3, 1994, requiring airlines to repeat a special, extensive test on 737 rudders every three months to check for signs of a malfunctioning PCU. Airlines are required to install an improved PCU within five years.

Duff said that Boeing's findings showed the Honduras incident "does not appear to be a rudder hard over," referring to a maximum rudder deflection of 26 degrees, which would roll a 737 into a dive.

By attributing the Honduras incident to a yaw-damper malfunction, Boeing is able to fall back on one theory it successfully advanced in the 1991 Colorado Springs crash probe, aviation sources said.

That case involved a United jet that flipped and dove into the ground, killing all 25 on board; after a 22-month investigation, the safety board ruled the crash unsolvable.

Twice in the week before that crash, pilots flying the ill-fated 737 reported an uncommanded yaw to the right. In each case the pilots switched off the yaw damper and landed safely, and mechanics removed and replaced the yaw-damper coupler.

After retrieving the yaw damper from the crash wreckage, the safety board ruled that a loose wire actually caused those two earlier incidents and could have come into play on the day of the crash.

But even if the yaw damper suddenly deflected the jet's rudder as it approached Colorado Springs, the safety board agreed with the Boeing analysis that the deflection would have been limited to 2 degrees, and that there would have been "little or no effect on airplane controllability."

However, The Seattle Times has obtained documents indicating Boeing at the time was in possession of reports from at least one other airline indicating uncommanded yaw-damper deflections could, under some circumstances, cause serious control problems.

The documents - which describe 737 yaw-damper problems experienced by Frontier Airlines from 1969 through 1975 - lend weight to Continental pilot Miller's assessment that the yaw damper might be to blame for the serious control problems he experienced on the Honduras flight. "The yaw damper is exceeding its design limits on its own initiative and putting inputs into the system," wrote Miller in the ASRS report.

The Frontier documents have never been made available to the NTSB, according to the NTSB source.

Manufacturers and airlines generally consider their transactions proprietary, and can be selective about divulging information during safety probes.

In one 1971 internal memo, Frontier's manager of jet training advises pilots that the airline "has experienced a few cases of malfunctioning yaw dampers on B-737 aircraft which have caused serious control problems."

The memo makes reference to ongoing Boeing research into the problem and instructs pilots to switch off the yaw damper before takeoff and landing and "any other time an abnormal yaw develops."

In a 1972 Frontier memo, the airline's manager of avionics makes reference to Boeing instructions on how to prevent moisture from seeping into the electronics bay area where the yaw-damper computer is housed.

The memo describes how an open electrical circuit caused the yaw damper to steadily deflect the rudder on a 737 flight. The pilot canceled out that deflection by adjusting a device called the rudder trim in the opposite direction.

When the yaw-damper open circuit disappeared, the rudder returned to neutral, but the rudder-trim adjustment "caused a sudden yaw in the aircraft."

In a 1973 letter from the FAA to then-airline President A. L. Feldman, the agency cites a Aug. 16, 1973, flight over Montana in which a "severe yaw" was encountered on a 737 flight. "This yaw was so severe that one of the cabin attendants was injured and required hospitalization."

The United jet that crashed was purchased from Frontier on June 6, 1986.

-------------------------- 737 INCIDENT OVER HONDURAS --------------------------

Continental Airlines Capt. Ray Miller filed this report with the NASA Air Safety Reporting System three days after making an emergency landing in Honduras in April. Submittal of the reports are voluntary. Pilots usually file them in routine matters like straying off course. "In this case, there was no threat of any violations, if anything this guy was a hero," said Mike Hynes, an aviation consultant. "I think he felt a moral obligation to put it into the system . . . so that the next guy could learn something from it." Miller's account:

Aircraft made violent yaw to the right. Simultaneous muffled boom, and instant roll to the right.

Autopilot/autothrottles manually disengaged immediately. Control column forces and displacement to regain control of the aircraft indicated a very serious problem with controllability.

To maintain wings level took 40 degrees left aileron (marked by the index on the wall) rudder forces normal. Aileron forces to the left very heavy. Forces to the right extremely light (or more like normal). As slow as I could fly and maintain control about 160 knots.

Emergency landing made about 18 minutes after onset at San Pedro Sula, Honduras. I had quite a lot of difficulty with roll control in that for corrections to the left would take a lot of pressure (for illustration say 20 lbs.) for corrections to the right all I had to do was ease off (for illustration say 2 ounces).

Fortunately, it was a very clear day and smooth. I feel if I had the normal low-level bumpy air I would have possibly gotten behind the control response and lost control of the aircraft. The company, etc. (now three days after the incident) inspecting the aircraft. It has been relayed to me that they have been able to duplicate the problem on the ground, but do not know why.

The "A" autopilot paddles disengage but the autopilot itself will not disengage until all power is removed from the aircraft. The yaw damper is exceeding its design limits on its own initiative and putting inputs into the system. After the fact, it appears I was fighting with the autopilot for control of the aircraft. This was a Boeing 737-300.