737 needs more than just redesign

WASHINGTON - Two puzzling Boeing 737 crashes were probably caused when an abrupt rudder movement sent the planes spiraling into a dive that the pilots were powerless to control, a federal safety panel said yesterday in a report that warned a redesign that previously was ordered for the world's most popular passenger jet is not sufficient.

The National Transportation Safety Board's report came after eight years of painstaking and often frustrating investigations into a series of 737 incidents, including a United crash in Colorado that killed 25 people in 1991, a USAir crash that killed 132 people near Pittsburgh in 1994 and a serious incident involving Eastwind Airlines at Richmond, Va., in 1996.

AP PHOTO John Sherbak and John Kretz take a break from the National Transportation Safety Board hearing yesterday in Sprinfield, VA.

Unlike most crash probes, the board's conclusions yesterday were based only on circumstantial evidence, making officials skittish about demanding specific technical changes in the 737's design.

The board stressed that the 737 has an excellent safety record. Chair Jim Hall said he regularly flies a 737 on one leg of his frequent trips home to Chattanooga, and other board members said they would have no problem putting their families on a 737.

The board report was based on eliminating all the known possibilities until only one plausible cause remained. In extremely rare cases, the board said a malfunction in the valve that directs hydraulic fluid through the rudder system can cause the rudder to move in the opposite direction than commanded by the pilot. A rudder is the flat movable plate at the end of the horizontal tail fin that can direct the plane left or right, although it is usually used only in turbulence or on landing in cross winds to keep the plane on course.

The board said that the specific problem that likely caused the rudder to malfunction - a jammed valve - had been solved by a redesign of the valve, now being implemented on the more than 3,000 of the 737s worldwide. But the board said the redesign was not sufficient to remove the possibility that such a malfunction could happen again in a different fashion.

The board unanimously said the Federal Aviation Administration should require that all 737s have a "reliably redundant rudder actuation system." Sources said the board's staff had wanted to be more specific, but board members were uncomfortable with making a specific recommendation in the absence of hard evidence. Even though investigators recovered the rudder valve in all three cases, it showed no obvious signs of a jam.

Thomas McSweeny, the FAA's associate administrator for regulation and certification, said he had no idea what the board meant by "reliably redundant" and would seek clarification.

The four-member board also called on the FAA to convene a test and evaluation board to isolate a fault that may appear and disappear quickly, leaving no trace. McSweeny said the FAA already had decided to convene a new board to take a look at the 737 rudder.

The probe was hampered because investigators could obtain little information from the planes' flight recorders about the rudder's movements or what the pilots were doing with the controls. Hall said the FAA and Boeing could shortened the investigations if only the planes had been equipped with more sophisticated flight data recorders, one of the board's prime ongoing crusades. The United plane measured only five aircraft movements or systems, called "parameters." The USAir and Eastwind planes measured 11 parameters. In neither case did the recorders measure rudder movement. Modern jets such as the Airbus A320 measure up to 300 parameters.

In response to earlier board recommendations, the FAA has ordered that all older planes be upgraded to 18 parameters, but Hall called the FAA order too little and too slow.

"The FAA failed in its responsibility to the flying public," Hall said.

McSweeny defended the FAA's actions throughout the 737 investigation, noting that the FAA ordered a redesign of the valve that would not allow a rudder reversal to take place under the board's scenario. He said the FAA also plans other actions, including new advice that pilots add speed during certain phases of flight so that wing control surfaces have enough strength to overcome any rudder movement.

The Boeing Company said it has already made numerous improvements to the rudder and has made recommendations on pilot training, but would cooperate fully with the FAA and the safety board in any further improvements.

Charles Higgins, Boeing's vice president of airplane safety, said the 737 has one of the world's best safety records and that the new valve design means "we have completely eliminated any possibility of a (rudder) reversal."

The board's report marks a milestone, but not an end, to the debate over two of jet aviation's rare mystery crashes - the Sept. 8, 1994, roll-and-dive of USAir Flight 427 and the March 3, 1991,

dive of United Flight 585. The United crash was originally thought to be related to some rare weather phenomenon. While deep in the Pittsburgh investigation, the board's attention was called to the June 9, 1996, incident involving Eastwind Flight 517 at Richmond, in which pilots used new training methods developed after the Pittsburgh crash to recover.

"A rudder reversal scenario will match all three events," said Dennis Crider, a safety board engineer who spent months developing computer simulations of the three flights that would fit the scant facts known from flight data recorders.

In each case, the planes were approaching airports for a landing. And in each case, the pilots were at a point that they likely were using the rudder.

The computer simulations raised the possibility that the rudder control valve jammed in flight for some unknown reason. Then, the valve sent hydraulic fluid the wrong direction the first time the crew attempted to use the rudder to line up for a runway or to correct the plane's path because of turbulence.

If the valve jammed as the board concluded, a crew member pushing on one rudder pedal on the floor would actually send the rudder in the opposite direction. What's more, the harder the pilot pushed on the pedal, the more the pedal would push back, likely leading the pilot to push even harder. Ironically, if the pilot did something that was counterintuitive - relax the pressure - the rudder would return to normal neutral position.

03-25-99