Expert Panel May Have Key To Which 737S Are Most At Risk

There is new evidence that certain Boeing 737s may be at increased risk for the kind of rudder malfunctions widely suspected of causing two catastrophic crashes and a near-crash since 1991.

A panel of experts reviewing the National Transportation Safety Board's investigations of the crashes has concluded that the rudder-control units of two of those 737s share common characteristics.

That discovery has raised the likelihood that there are other 737s that have the same higher-risk rudder-control-system traits: tighter slide clearances inside a key rudder-control valve, and other control-system alignments.

The work of the expert panel last year led Boeing to acknowledge potential problems with the rudder-control systems on as many as 3,000 older-model 737s in service around the world. Boeing plans to replace those systems with ones believed to solve the control problems, but that could take at least two years.

The panel has now found that the rudder-control units of a USAir 737-300 that crashed in Pittsburgh on Sept. 8, 1994, killing all 132 aboard, and an Eastwind Airlines 737 that nearly crashed near Richmond, Va., in June 1996, share the characteristics that make them more likely than others to cause a malfunction.

The most dramatic malfunction is what is known as a rudder "hardover," in which the rudder - the movable panel on an aircraft's vertical tail that controls the direction of travel - swings uncommanded by the pilot all the way to one side. If it happens at low altitude and slow speed, it can snap the plane into a dive the pilot may not be able to stop.

A rudder hardover is widely believed by investigators to have caused the USAir crash. It also is suspected of causing the 1991 crash of United Airlines Flight 585, a Boeing 737-200 that suddenly dived on final approach to Colorado Springs, Colo., killing all 25 on board. The rudder-control unit was so mangled after that crash that investigators were unable to accurately measure for the suspect characteristics.

The NTSB investigation of the Colorado Springs crash ended without establishing a probable cause. While no probable cause has been declared yet in the Pittsburgh crash, the NTSB has made 14 recommendations to the Federal Aviation Administration that focus on fixing 737 rudder-control problems.

The FAA has rejected seven of the NTSB's recommendations, prompting an increasingly heated debate between the nation's two top aviation-safety agencies.

In a letter to the FAA last February, NTSB Chairman Jim Hall said the 737 rudder system is not as safe as the systems on other passenger jets. He implied that the FAA should never have certified the 737 rudder design.

Hall urged the FAA to order airlines to speed up the installation of an improved rudder part and to warn pilots about the potential for dangerous rudder movements.

FAA officials countered that their agency has been taking all appropriate steps to make sure the 737 is as safe as any other jet.

Last month, Hall said the FAA's refusal to carry out seven of the 14 recommendations was unacceptable.

Now, the NTSB appears to be nearing another recommendation, this one designed to quickly find and fix 737s with the rudder-control systems that may pose a higher risk.

Two troubling discoveries

In a July 10 letter to the expert panel, Hall noted that investigators were collaborating with Parker Hannifan, the Irvine, Calif.- based supplier of 737 rudder power-control units, or PCUs, to "develop techniques" for identifying jets with "characteristics similar to the USAir Flight 427 and Eastwind rudder control units."

Those characteristics were isolated last November when members of the expert panel monitored a test in which Boeing and Parker engineers measured the slide clearances and alignments on PCUs from the Pittsburgh crash and the Richmond near-crash, as well as those on a new PCU.

The panel was interested in whether dirty hydraulic fluid could create a jam in the dual servo valve, a device about the size of a soft-drink can at the heart of a PCU. The servo contains two pencil-like slides, one nestled inside the other, which work together in a delicate sequence to direct pressurized hydraulic fluid to move the rudder.

Two troubling discoveries were made. The first was that the servo valves in both the USAir and Eastwind control units had slide clearances tighter than those on the new PCU valve. The USAir jet's was significantly tighter. Though the clearances of all three were within allowable limits, the valves with tighter clearances seemingly would be more susceptible to jamming.

The second important finding showed that slides from the USAir and Eastwind control-unit valves were aligned in such a way that they tended to trigger rudder reversals, a type of hardover, much quicker that those on the new PCU.

These two discoveries prompted members of the safety panel to call for steps to identify other 737s with the higher-risk characteristics.

"This is still an active issue and the investigative team is considering a recommendation to the FAA to identify all suspect units," Hall wrote.

The FAA would not discuss its position on identifying the higher-risk rudder-control units, other than to repeat that it was taking the steps needed to assure the safety of 737s.

Boeing safety officials, meanwhile, said they don't believe it's necessary to try to identify 737s with the tighter slide clearances and alignments in the control units.

While acknowledging some PCUs were shown to be more prone to jamming because of the narrower space between the housing and a slide in the unit's servo valve, the Boeing officials said that "thorough testing has shown that a secondary slide jam can not occur in flight, (and) therefore, neither can a rudder hardover or reversal." They said the current steps being taken are adequate to assure the safety of 737s.

Phil Condit, chairman and chief executive officer of Boeing, reiterated the company's position this week in a question-and-answer piece in the New York Times Magazine.

Asked why Boeing hasn't recalled the 737 fleet while it works on the rudder problem, Condit said: "There is zero evidence" that a rudder problem caused a crash or near-crash.

"But we did find a condition that could produce a problem," he said, referring to an October lab test that prompted Boeing to replace the rudder-control part on 737s. "We went out to the fleet and said, `Go check for this.' We checked every single airplane and didn't find any of them with that problem. We changed the (PCU) design so that won't happen."

The "thermal-shock" test

Flight crews have reported inadvertent movements of the 737 rudders since the plane entered service in the late 1960s. But the rudder-control system didn't come under intense scrutiny until the 1991 Colorado Springs accident. Concern grew after the Pittsburgh crash, in which radar and the plane's flight-data recorder traced a sudden plunge to the left.

Frustrated by the inconclusive Colorado Springs investigation and the lack of clear progress in the Pittsburgh investigation, Hall created the six-member panel of independent experts in February 1996 to review the NTSB's work in both investigations.

He picked David Horn, crash investigator and reliability analyst from the U.S. Air Force; Werner Koch, hydraulics expert from the FAA; Ralph Vick, aeronautical engineer from Moog Aircraft Group; K.C. Shih, research engineer from NASA's Ames Research Center; Paul Knerr, hydraulics-systems engineer from Canyon Engineering; and, David Haasl, a risk analyst from Oregon. Greg Phillips, the NTSB's resident rudder expert, was assigned to direct the panel.

The panel soon focused on the rudder power-control unit. Early in both the Colorado Springs and Pittsburgh crash investigations, Boeing engineers had conducted tests for the NTSB which seemed to absolve the PCU from any blame in either accident. Yet, if rudder movements had anything to do with either crash, then the PCU had to have been a factor.

The expert panel turned its attention to re-examining whether dirty hydraulic fluid could have jammed the USAir jet's PCU.

The panel devised an experiment to assess whether hot, dirty fluid directed into a very cold servo could cause the slides to jam, misdirect pressure and trigger a rudder hardover.

In August 1996, the panel placed a new PCU inside a large Coleman ice chest rigged with hydraulic lines and tubes connected to a tank of liquid nitrogen used to cool the PCU to minus 40 degrees Fahrenheit. Hydraulic fluid heated to 170 degrees Fahrenheit was then pumped into the new PCU, and it worked smoothly.

Then the USAir jet's PCU was placed inside the Coleman cooler and the experiment repeated. The USAir PCU jammed.

During the week of Oct. 7, 1996, Boeing repeated the "thermal shock" experiment in its Seattle systems-integration lab using more sophisticated equipment. Boeing's results were identical to the expert panel's. Computer data generated by the Boeing test was then used to assess what was going on inside the servo.

The data indicated the servo's outer slide could be jammed without pilots or mechanics noticing. If a pilot stomped on a rudder pedal, and the outer slide was jammed, the servo could misdirect fluid and cause the rudder to reverse itself.

Investigators wondered if something similar to the lab experiments had unfolded on the Pittsburgh flight.

Perhaps the jet's PCU had been cooled by inactivity at high altitude, they reasoned. Then, when one of the USAir pilots depressed the right rudder pedal to correct for a jolt of turbulence, hot fluid could have shot through the servo.

Under this scenario, the rapid temperature change would have caused the servo slides to expand to the point of jamming - as demonstrated in the thermal-shock experiments - and misdirecting fluid to reverse the rudder.

This would have made the jet swerve and roll left. Instinctively, the pilots would have tried to regain control by pushing down sharply on the right rudder pedal, a move that would only exacerbate the leftward dive as the rudder swung further left.

Boeing disputed this rudder-reversal scenario with research showing that the temperature changes created in the lab settings were unlikely to have occurred under the flight conditions in Pittsburgh.

The stomp test

However, Boeing engineers began a series of experiments using a mechanical device to jam the servo slides. The tests confirmed the capacity for a jammed PCU to reverse the rudder. Boeing then devised a simple test to gauge whether PCUs on in-service 737s might be susceptible to jamming.

The rudder test, made mandatory by the FAA early last November, calls for a mechanic to stomp on the rudder pedals of a parked jet to assure that the pedals move smoothly and freely. So far, no problem PCUs have turned up.

The tests do not measure whether the clearances between the slides and between the outer slide and the servo-valve housing might be as tight as those on the Eastwind or 427 jets. Expert-panel members believe the narrower gaps could be significant during flight where temperature changes, vibrations, the ebb and flow of dirty hydraulic fluid and other variables could come into play.

The stomp tests also don't measure whether the alignment of the outer slide and associated linkage parts make it easier for them to lock into a position that could quickly cause a rudder reversal.

On Nov. 20, after the rudder stomp tests had begun throughout the 737 fleet, the NTSB's Phillips and members of the expert panel arrived at the Parker lab in Irvine, Calif., to watch Parker and Boeing engineers measure the slide clearances and alignments on the two PCUs used in the experiments. Phillips brought the Eastwind jet's PCU to Irvine to undergo the same measurements.

That is when the troubling discoveries were made on the tighter clearances and alignments of the servo-valve slides.

The NTSB is pushing for the overall 737 rudder-control retrofit to get under way this fall and be completed by January 1999. But given the complexity, it could take a year or two longer.

In the meantime, many safety experts believe, each unmodified 737 will continue to fly with the risk of a rudder malfunction. Based on the expert panel's discoveries, that risk may be greater for some 737s than others.

Links to Byron Acohido's earlier stories on the Boeing 737 are on The Seattle Times Today's News Web site at: http://www.seattletimes.com