A Tragic Echo

By Fenella Saunders|Saturday, November 01, 1997
When a small commuter plane crashed off the Rhode Island coast in 1991, the National Transportation and Safety Board blamed the accident on pilot error. However, on conducting its own investigation, the Airline Pilots Association found evidence that the plane’s right engine had broken off during flight and smashed through the tail. To prove its case--and exonerate the dead pilot--the association turned to Ronald Stearman, an aerospace engineer at the University of Texas at Austin. Stearman has recently managed to vindicate the pilot through an innovative analysis of the plane’s cockpit voice recorder. His methods may help maintenance workers spot problems in aircraft before a crash.

Although the voice recorder was crucial to Stearman’s investigation, he learned nothing from listening to the pilot and copilot-- their conversation gave no hint of any problems. So Stearman analyzed the voice recorder for background noises made by the plane itself. If the study by the pilots’ association was right and the engine did fall off, the plane’s mechanical problems should have affected the recordings.

Looking at Federal Aviation Administration inspection records, Stearman discovered that this type of airplane, a twin-engine Beech Aircraft 1900C, had a history of engine-mount cracking, which had prompted six redesigns. When Stearman used a computer to simulate how the engine would act with a cracked mount, he found that the damage would almost certainly have caused a deadly phenomenon known as whirl flutter.

First described some 40 years ago, whirl flutter is the violent shaking of an aircraft that results from the airstream’s battering of the propellers. This happens either because the plane’s speed exceeds design limits or because hidden structural damage weakens the plane to such an extent that normal speeds stress the craft to the breaking point. Flutter is catastrophic, says Stearman. To an observer it might look like an explosion.

After studying the computer simulation, Stearman analyzed three of the recorder’s four tracks--one track each for the pilot, copilot, and general cabin area. (The fourth track, hooked up to a mike in bigger planes for a second copilot, was unused.) Stearman found a warbling effect on various cockpit noises. The warbling probably occurred because the recorder was vibrating, making the tape slip. These vibrations, Stearman says, match recorded lurching changes in the rotation of the propeller blades, a clear sign of whirl flutter.

The most telling information, however, came from the unused fourth track on the recorder. When Stearman analyzed the tape, he found a signal on this silent track. The wires to the tape recorder, because they were attached to the plane’s superstructure, acted as a simple microphone and picked up the aircraft’s vibrations, he says.

This makeshift microphone, says Stearman, managed to record signals that were masked by noise on the other tracks. The vibrations he detected on the fourth track exactly matched the plane’s engines’ vibrations. The track culminates with a huge signal spike, caused, Stearman realized, by the engine’s crashing into the tail, where the voice recorder was housed. (The microphones for the recorder are in the cockpit, but the recorder itself is in the rear of the plane.)

Stearman says that an analysis like his should be part of routine aircraft maintenance. Clear signals of mount damage occur throughout the 30-minute tape, he says, and he suspects that if the tape were longer, such signals could have been detected hours before the damage became critical. Recorders could show other types of problems as well, Stearman says, such as engine gear failures, or cabin decompression because of a faulty seal. We are suggesting that those who maintain the airplanes leave a track blank, then check that silent track. If you see things like this, you’d better go back and check the engine mounts. This could give you enough lead time so that nothing happens before you can get it fixed.
Comment on this article