The Mystery Of Usair Flight 427 Continues -- Inquiry Goes On; More Troubles Reported With Boeing 737'S Controls

AN UNUSUAL HEARING will convene next week as federal officials make a second review of the 1994 crash of a USAir Boeing 737 in which 132 people died. Meantime, airline pilots continue to report trouble with the flight-control mechanisms on the 737. -----------------------------------------------------------------

Why USAir Flight 427 suddenly flipped upside down and screamed to earth in 1994 remains one of aviation's most disturbing mysteries.

Authorities think the Boeing 737-300 jetliner's rudder swung to the extreme left and froze, tossing the plane into a tumbling, high-speed dive. The accident unfolded in clear, calm skies as Flight 427 descended routinely into Pittsburgh on Sept. 8, 1994. The crash killed all 132 people on board.

Despite 40,000 investigative man-hours and millions of dollars spent on lab and flight tests, investigators still can't explain why the rudder moved.

The National Transportation Safety Board (NTSB) next week will convene an unusual second hearing in Springfield, Va., to review evidence about the Pittsburgh crash.

Since the first NTSB hearing last January, 737 pilots have continued reporting instances of their flight controls malfunctioning in flight, often making uncommanded movements. (See chart on A 20.)

Records obtained by The Seattle Times show at least 45 reported instances of flight controls moving inadvertently - including 19 incidents on USAir 737s - in the 14 months since the Pittsburgh crash.

Safety agencies generally have characterized the reports as of little significance. Malfunctions that come and go are nothing new on commercial jetliners, and authorities expected a spurt in reports involving 737s because of questions raised by the Pittsburgh crash and the unsolved crash of a United Airlines 737-200 in Colorado Springs in 1991.

A source at USAir said the airline thinks its pilots are "over-reporting" flight-control problems they normally would treat as minor nuisances.

But some independent experts say the reports about uncommanded 737 control-surface movements should be viewed with concern.

Does it add up to a safety issue?

"How many nuisances equal a hazard?" said Chuck Miller, an aviation-safety consultant from Sedona, Ariz. "It's a highly subjective thing, but somebody has got to make a decision as to how many of these things add up to a safety issue."

Such decisions are among the challenges that 400 international aviation officials are tackling in a safety seminar at the Westin Hotel this week: how to shift the industry's focus from investigating crashes to recognizing safety problems as they emerge.

Boeing and safety-board officials declined to be interviewed for this story, saying the crashes and 737 questions will be discussed at Wednesday's hearing.

The 737, the most widely used commercial jet in history, is counted by Boeing as one of the safest aircraft in the world, with accidents every 2 million flights, half the industry average.

Its crash record is somewhat better than average. Since the model entered service in 1967, 61, or 2.3 percent of the fleet of 2,600 planes, have crashed, including 21 in the past five years.

By comparison, 4.7 percent of all McDonnell Douglas DC-10s, 3.3 percent of McDonnell Douglas' DC-9/MD-80s fleet, 2.2 percent of Boeing 747s and 3.4 percent of Boeing 727s have been lost. Nevertheless, the unsolved Pittsburgh and Colorado Springs crashes have kept the 737 in the spotlight, and another safety question has emerged since the first hearings on the Pittsburgh crash.

An engineering-standards group has asked for more thorough follow-up investigation of findings suggesting that many 737s may be flying with debris-contaminated hydraulic fluid.

Aviation experts have known for 50 years that microscopic debris can wreak havoc with delicate flight-control systems. Though maintenance procedures and filters have evolved to minimize contamination, little scientific study of the problem has been conducted.

Evidence of highly contaminated hydraulic fluid turned up in the ill-fated United and USAir jets. Later, in random spot checks of hydraulic fluid on 737s, 22 percent of the samples were found to have what the engineering-standards group considers highly contaminated fluid.

The group, the Society of Automotive Engineers (SAE), establishes fluid standards for auto, factory, aircraft, weapons and space systems.

SAE now is seeking a more definitive estimate of how many of the 2,600 737s in commercial service might be flying with dirty fluid - and weighing whether it should write tougher standards for jetliner flight-control components.

Paul Knerr, a member of SAE's aircraft system's committee, said he would like to see authorities routinely measure hydraulic-fluid contamination levels on any 737 that reports a control problem.

Erratic symptoms - or a seizure

Knerr said such sampling would reveal if there's any basis for a widely held but undocumented belief that, as he puts it, "The dirtier the fluid, the more likely the anomaly."

Aircraft flight-control surfaces move in response to the smooth flow of pressurized fluid in various chambers, directed by a series of valves and levers. Dirty fluid can cause rogue movements of flight-control panels on the wing and tail.

"It's like clogging an artery," said Karun Nair, a Houston-based hydraulics-systems designer. "It can create erratic symptoms that come and go, or cause a seizure."

One case where a correlation was established between dirty fluid and a rogue flight-control movement involved a U.S. Air Force F-4 fighter jet in the early 1970s.

The F-4 flew briefly out of control and nearly crashed when its flight controls moved without being commanded to do so. A subsequent investigation led by Dwayne Rubertus, a civilian engineer at the Wright-Patterson Air Force flight-dynamics lab in Dayton, Ohio, found that contaminants had jammed the flight controls for several seconds.

Since then, there has been little additional study correlating dirty fluid and flight-control glitches, partly because contamination is very difficult to quantify.

"The particles can take many different shapes, round, long, oblong and slender," Rubertus said recently. "The effect of going through orifices is such that a round particle may pass through, but the long slender one may get caught edgewise, just like a bone a lot of times passes through your throat, but sometimes gets stuck.

Most incidents fit pattern

"Probably 99 percent of the time, nothing will happen, but a small percentage of the time, the thing gets turned crosswise and gets stuck. That can cause a lot of things to happen."

Nair, the hydraulics designer, said most of the incidents 737 pilots have reported in recent months fit the pattern of what can happen when dirty fluid circulates through a jet's flight-control system.

Knerr and Nair said uniform sampling of 737 fluid could help industry better understand whether dirty fluid may be triggering or compounding flight-control glitches.

The glitches usually involve rogue movements of the rudder, the movable part of the vertical tail section that controls the aircraft's left/right heading, or the ailerons, wing panels that flip up and down to put the plane in a bank.

Hydraulics experts say debris characteristically causes momentary disruptions, then tends to recirculate with the fluid, often leaving no trace. In many reports, pilots describe sudden rudder or aileron movements that seem to go away after they shut down certain systems. Often the assumption is made that one of the shut-down systems caused the problem.

Last March 29, a Southwest Airlines 737-300 was cruising into San Jose, Calif., at 37,000 feet when the rudder moved, causing the jet to suddenly swerve and bank 25 degrees to the right. Pilots Richard Duke and Randall Walter disengaged the autopilot and turned off another computer called the yaw damper, which automatically makes small rudder adjustments during flight. The pilots then manually steered the aircraft back to level flight.

Duke and Walter told a Federal Aviation Administration (FAA) official the jet came "within 10 seconds" of rolling upside down, according to a report on the incident.

Federal agencies unconcerned

Such reports aren't limited to recent months. Pilots have reported several hundred cases of flight controls moving inadvertently on 737s since the early 1970s.

"Pilots feel there is obviously something happening. What it is, nobody knows," said a veteran 737 pilot. "We're trying to be more aware of flight-control anomalies."

The official agencies, however, have not declared any safety concerns.

"Nobody seems to think this is a big deal," said a source close to the NTSB. "If there is a problem, they believe pilots will disengage the autopilot or yaw damper and fly the airplane."

Investigators say they have been unable to turn up any evidence that debris recovered from the hydraulic systems of the USAir and United jets had anything to do with them crashing.

Yet the crashes precipitated several theories about how contamination might present a danger, and safety officials subsequently took action to address those dangers.

In the months after the Colorado Springs crash, authorities learned debris could jam a 737's rudder-control valve in a variety of ways. Under certain conditions a rudder could reverse itself. Or it might swing to an extreme position in response to a slight command from the pilot or yaw-damper computer.

Investigators found moisture, traces of a white powder and bronze-like chips near the Colorado Springs jet's rudder valve, but the particles were never analyzed.

Without solving the Colorado Springs crash, the safety board, nonetheless, called for two improvements in the 737's rudder system.

The FAA subsequently ordered airlines to implement one of the changes - making it harder for the rudder to reverse - but ultimately ignored the other. The FAA proposed that airlines be required to regularly inspect for corrosion in a standby rudder valve but dropped the proposal last year after airlines protested.

Clues to a broader problem

When the USAir jet crashed in Pittsburgh, investigators immediately suspected a rogue rudder movement. Again, despite months of exhaustive testing, which included rebuilding and testing a heavily damaged rudder-valve unit, investigators failed to to come up with any evidence implicating the rudder.

But to some, there were indications of problems.

Hydraulic-fluid samples from the crashed USAir contained a "level 10" concentration of contaminants, NTSB records show. Level 3 is considered very clean fluid. The concentration of impurities doubles at each higher level. Most jetliner flight-control components are designed to operate with up to level-6 contamination, said SAE's Knerr.

More than 100 hydraulic fluid samples were then drawn from 21 737s operated by USAir, United, Southwest, China Air, Boeing and Qantas. The result: 22 percent of the samples were found to have contaminant levels 8 through 12, NTSB records show.

Investigators interpreted that finding to mean Flight 427's dirty fluid was no worse than fluid being used in one-fifth of the 737 fleet.

"They simply said, `Oh, that's a typical airplane, don't worry about it,' " said SAE's Knerr, whose committee sets jetliner-fluid standards.

But Knerr saw it differently. "My reaction was just the opposite: `Oh, my gosh! How can we be flying that dirty, guys?' "

NTSB records show the 737's hydraulic system may be particularly susceptible to jams because debris can pass through comparatively coarse filters and clog microscopic clearances between valve surfaces.

In a review of the 737 flight-control systems published last May, the FAA also pointed out that Boeing does not require or recommend ever checking or changing the 737's hydraulic fluid. The company does recommend changing filters on intervals established by the airline.

The review pointed out that no one knows how varying levels of debris affect the 737's hydraulic system, and called for investigating how sensitive the 737's flight controls are to dirty fluid.

Last January, Knerr, vice president of engineering at Valencia, Calif.-based Canyon Engineering, approached the NTSB to urge follow-up fluid sampling of the 737 fleet. He argued that samples of fluids circulating during flight would be more characteristic than samples the safety board had drawn from parked airplanes.

Knerr said a safety-board investigator initially seemed interested in the idea, putting him in contact with Boeing, which, in turn, assigned an engineer to design equipment needed to take in-flight samples.

But the board's interests subsequently waned, Knerr said, as it shifted its focus to a completely different theory: that wing tip turbulence from a jet flying four miles ahead of Flight 427 may have somehow thrown the USAir jet into a dive.

That theory, which a number of pilots and experts consider an improbable explanation for Flight 427 flipping out of control, was the subject of $1 million in testing on the East Coast this fall. It will be among the topics reviewed at next week's NTSB hearing. ----------------------------------------------------------------- Problem 737 flights. . Since the unsolved September 1994 crash of a USAir 737 near Pittsburgh, U.S. aviation-safety investigators have studied these crashes or instances of erratic flight involving a Boeing 737. The causes of most incidents are unsolved. Authorities say 737s remain safe to fly in.. . Tail no. date airline place . ----------------------------------------------------------------- n/a 10/22/95 British Airways n/a . Uncommanded rudder oscillations.. . N266AU 10/5/95 USAir Indianapolis . On takeoff, uncommanded hard swing left.. . N39343 10/3/95 Continental Cleveland . At cruise, uncommanded side-to-side skidding followed

by roll. . N590US 9/30/95 USAir Indianapolis . At cruise, 5 to 6 quick bumps followed by abrupt 25-degree

left roll.. . N563AU 9/29/95 USAir Indianapolis . Uncommanded full left rudder movement believed caused by

faulty rudder-trim switch.. . n/a 9/29/95 USAir Charlotte . On descent, uncommanded roll left, 15 to 20 degrees. . . N977UA 9/25/95 United Denver . Autopilot initiates uncommanded bank.. . N303AW 9/23/95 America West Phoenix . On approach, at 2,000 feet, uncommanded hard right yaw. . . N695SW 9/18/95 Southwest Phoenix . On takeoff, violent jolt right, with sensation of being

thrown left. . N280AU 9/12/95 USAir Dallas-FW . Uncommanded elevator kick pitches nose up. . N266AU 9/10/95 USAir Indianapolis

Climbing at 16,500 feet, nose pitches up, steering

column jumps back, then forward. Yoke jumps back and

forth three times.. . PK759/KYJ 9/10/95 British Airways Birmingham, UK . On takeoff, hard rudder swing left; full right rudder deflection required to keep level flight.. . . N303AW 9/10/95 America West Phoenix . On takeoff, uncommanded sharp right yaw.. . . N14325 8/30/95 Continental Cleveland . On descent, uncommanded 40 degrees left roll, then 20 degrees right. Leaky lavatory water suspected of corroding autopilot circuits.. . . N373US 8/25/95 USAir Fort Lauderdale . On approach, uncommanded roll 30 degrees right.. . . N303AW 8/23/95 America West Phoenix . On approach, felt like airplane had been hit with a broadside.

Rumble heard in aft cabin.. . . PK801 8/21/95 Ryanair n/a . Violent yaw on takeoff requires full opposite rudder deflection

to regain control. . . . N434US 8/17/95 USAir Phoenix . Uncommanded rudder kick prior to touchdown.. . . N125G4 8/12/95 Aviateca El Salvador . Crash. On approach in rainstorm, plane hits volcano. 65 killed.

Pilot error suspected.. . . N287US 8/10/95 USAir MYS . Uncommanded rudder kick; autopilot responds and commands right roll.. . . N287US 8/10/95 USAir New Orleans . On approach, uncommanded roll and yaw; autopilot switched off.. . . N396US 8/5/95 USAir Albany, N.Y. . Shudder, then uncommanded roll left.. . . N390US 7/25/95 USAir Charlotte . On climb, slight bump, uncommanded 10-degree roll.. . . N448US 7/25/95 USAir Richmond, Va. . On approach, uncommanded 30-degree right roll. Problem

persists after autopilot, yaw damper disconnected.. . . N390US 7/18/95 USAir Orlando . Uncommanded left roll, twice. Leaky lavatory water suspected

of causing autopilot malfunction. . . PW238 7/16/95 n/a n/a . Autopilot commands excessive right roll; unable to duplicate in

lab test. . . N392US 6/26/95 USAir Washington, D.C. . Uncommanded left roll; autopilot could have contributed to event. . . n/a 6/26/95 USAir n/a . Uncommanded left bank requires wheel and rudder input to level

plane. . N322SW 4/26/95 Southwest Phoenix . Abrupt uncommanded right rudder input. . . . n/a 4/7/95 VASP Sao Paolo, Brazil . Uncommanded yaw, vibrations. Power blackout.. . . N9355 3/30/95 United Indianapolis . Uncontrollable 30-degree left roll caused by malfunctioning

wing spoiler. . . N673AA 3/29/95 Southwest San Jose . At cruise, uncommanded roll from rudder kick lasts 20 seconds.. . . VH-TJK 2/20/95 Qantas Sydney, Aus. . On approach, violent uncommanded right roll.. . . N164AW 2/3/95 America West Ontario . On approach, uncommanded small rudder movements.. . . N207AU 1/30/95 Frontier Bozeman, Mont.. On approach, uncommanded left and right rudder movements. . . N155AW 1/26/95 America West Boston . On takeoff, right roll. Cause determined to be related to

out-of-adjustment landing gear. . . N155AW 1/23/95 America West Albuquerque.

On takeoff, uncommanded right roll.. . . N155AW 1/23/95 America West Albuquerque . On test-flight takeoff, uncommanded right roll.. . . N563AU 1/22/95 USAir Pittsburgh.

On takeoff, No. 1 generator lost. At cruise, uncommanded

extreme hard left rudder movement. Rudder pedals move. . . n/a 12/29/94 Turkish Airlines Van, Turkey. Crash. Jet hits ground on third attempt to land in snowstorm.

At least 53 of 69 killed.. . . 7TVEE1 12/22/94 Air Algerie Coventry, Eng. . Crash. On approach, plane nearly clips housing development,

crashes short of runway. Wrong altimeter setting

believed made by pilot. . . N583AU 12/20/94 USAir Raleigh . On approach, level flight, no turbulence, very sharp

uncommanded rudder input; plane yaws left.

Rudder pedals do not move. . . n/a 12/15/94 Air France n/a . Two uncommanded rolls within 30 seconds. . . N681MA 12/11/94 Mark Air Anchorage. During climb, uncommanded pitch up, then down. Control yoke

moves 8 inches. . . N688SW 12/3/94 Southwest Sacramento . On takeoff, hard right pull.. . n/a 12/1/94 Air France n/a . Two uncommanded rolls within two minutes. . N583AU 12/20/94 USAir Philadelphia . On approach, uncommanded yaw damper input; rudder pedals move. . N978UA 9/21/94 United n/a On takeoff, veers right. . N513AD 9/8/94 USAir Pittsburgh Crash. On approach, clear, calm skies, plane jostles, inverts,

nose dives into ground. 132 killed. . n/a = not available. . . Source: National Transportation Safety Board; Federal Aviation. Administration service-difficulty reports and accident and incident reports..