If it ain't necessarily broke, should they fix it?
On Oct. 16, even before the announcement that the rudder control on the Boeing 737 could be made to seize up, the National Transportation Safety Board made a long series of draft recommendations about the jetliner's rudder system. The recommendations arose from the Colorado and Pittsburgh disasters, and from the report of an Eastwind Airlines 737 that landed safely -- without injuries -- at Richmond, Va., in June, 1966, after going through a yaw-and-roll that was suspiciously akin to the disaster at Pittsburgh.

US Air 737 plane crash in New York,1989
© AFP/M. Caldwell

On closer inspection, it turned out that these were not isolated incidents. Starting with the 1991 United 737 crash in Colorado, the Safety Board says it "has been informed of numerous uncommanded roll and yaw events involving the Boeing 737 series." Two occurred in France -- and in those cases, the pilots were able to shut off the autopilot and get the plane back under control, says Paul Czsyz (pronounced sizz), an experienced aircraft engineer who now teaches at St. Louis University. That's because the rolls happened at 25,000 feet, high enough to allow plenty of time for recovery.

As best experts understand it, the Pittsburgh roll seems to have originated with a movement of the rudder that was not commanded by the pilot. Normally, slight rudder movements can happen under the control of a device called the yaw damper. This system automatically controls yaw by nudging the rudder back and forth, keeping the nose pointed ahead without constant pilot attention. But the damper is supposed to move the rudder only 3 degrees, and the accident probably grew from a 20-degree movement of the rudder -- called a "rudder hardover."

A rudder hardover is an emergency measure -- used only to balance out the flight if one engine goes out.

Like any mechanism, the yaw damper can malfunction. If it does, the pilots -- assuming they recognize the problem -- can just shut it off and fly manually. But the USAir 737 had no indicator for rudder position, and in the 20-odd seconds between the start of the accident and its grisly outcome, there is no sign the pilots suspected a rudder problem.

Rudder problems are most serious when a plane is close to the ground, like preparing for landing, says Curtis Meisenheimer, an aerospace analyst with Frost & Sullivan, a consulting firm in Mountain View, Calif. "Anything that happens then is extremely dangerous, because you are getting bounced around enough as it is" and there is not much room to maneuver and regain flight speed.

In the Pittsburgh accident, investigators believe that a wake vortex (defined) may have contributed to the instability that started the 737's roll, but there is no proof. Here's some Federal Aviation Administration (FAA) information on wake turbulence.

Czsyz calls this explanation "grasping at straws," arguing that the 737 could not have been turned over by the wake of a plane that was no bigger that it was and was five miles ahead. Pilots, he says, are routinely trained to respond and right planes that have rolled over due to turbulence, but whatever the USAir plane experienced "was something the pilots had never encountered before. That's what made it so hard for them."

In the suggestion box
On Oct. 16, the Safety Board passed to the FAA a long list of recommendations designed to "confirm the continued operational safety of the B-737." If the FAA approves, Boeing will be required to suggest and carry out changes that would prevent potential loss of control. That means, Meisenheimer says, "to work out what might have caused the rudder hard-over, and to work out maintenance procedures, in-flight recovery procedures, testing procedures, and possible redesign of the rudder system to keep the aircraft flying." The 737s rudder, he points out, is one panel, with one control device, whereas "other aircraft use two panels, and two actuators," which provides redundancy to reduce the chances of failure.

Twenty nine years after Boeing began making them, 737s are still rolling off the assembly lines. Today's models have a cockpit indicator for the rudder position, and Boeing could be required to install an indicator in older models, which lack them. (All 737s have an indicator for the effects of the yaw damper.)

Boeing might also be required to re-engineer the yaw damper system and reinstall it on all existing 737s. Meisenheimer says it's "similar to an auto recall. ... Now, we're talking about real money. Redesign can get expensive."

Other recommendations covered pilot training and maintenance procedures, both aimed at cutting off the unwanted yaw and roll motion immediately. The message to Boeing, says Meisenheimer, is that "we need you to develop emergency procedures to recover the aircraft from any condition."

Military pilots, notes Meisenheimer, who used to train U.S. Navy aviators, are taught to recover an airplane from "stalls (defined), rolls and spins" -- training civilian pilots don't necessarily receive. "The present issues involve recognition and recovery from unusual attitudes," he says, "and there is much industry debate about how to keep airline pilots current in these procedures. You've got to recognize the problem quickly, and get out of the situation quickly," and that calls for more training in conditions no civilian pilot wants to experience.

Although the 737 has proven itself to be a safe plane, Meisenheimer observes, the three incidents noted by the Safety Board are "just the tip of the iceberg" that point to a systematic problem that must be addressed.

The decision on rudder safety rests with the FAA, which has 90 days to respond to the Safety Board's recommendations. There is no indication how the news that cold controls and hot, contaminated hydraulic fluid could combine to cause a seize-up will affect the FAA's action.

Want to hear a contrary analysis of the Pittsburgh accident?

There are 1 2 3 4 5 6 7 8 9 10 11 12 documents. (Glossary | Bibliography)