The seat-of-the-pants approach to flutter testing is possible for the same reason that an automobile tire that vibrates most severely at 60 mph starts shaking at 55 and stops at 65 or 70. The tendency to flutter does not usually rise instantly to a maximum when one parameter or another--say, airplane speed--reaches a critical value. It normally ramps up gradually enough for speed increments of one or two mph to give the pilot warning of impending trouble. The pilot relies on the feel of the stick to warn him of diminished damping. The dead beat response softens; the stick wobbles once or twice before returning to center. Sometimes a control surface actually flutters with a low intensity, so the structure flexes but does not fail and the pilot, by slowing the airplane down, can arrest the flutter.
Still, "usually" and "normally" must be added. On occasion, catastrophic flutter occurs at a speed only very slightly above a speed that was considered "safe," and sometimes even below it. Today, except in the lowest-budget flight test programs, sensitive electronic motion sensors and strain gauges measure and record vibrations that are triggered by a mechanical shaker rather than from the pilot's hand or foot. The slightest decrease in damping is instantly detected. Nevertheless, testing for flight flutter is no test pilot's favorite activity.
Until after World War II, structures were built as lightly as possible, and balanced and stiffened later as needed. Flutter problems were mainly attacked with empirical methods, often crude and sometimes quite drastic. Before the National Air Races in 1934, Steve Wittman's midget racer Chief Oshkosh repeatedly encountered severe but non-destructive flutter of a wingtip. It happens that short beams, like short piano strings or organ pipes, have higher natural frequencies than long ones. Therefore, after each test flight on which flutter was encountered, Wittman chopped a few more inches off the wing. Eventually the span was reduced to 16 feet, and the wing area from a minimal 78 square feet to a microscopic 42. The airplane landed awfully fast, but at least its wing was free of flutter.
William R. Laidlaw, former chief of the Structural Dynamics Section at North American Aviation, describes in The Revolution in Structural Dynamics a similar approach to controlling stabilizer flutter in the 1950s. During a test flight, an FJ-4 Fury, a carrier-based derivative of the F-86 Sabre Jet, lost more than half of both horizontal stabilizers to a flutter incident (the pilot managed to land safely). Engineers mounted a Fury tail assembly on a rocket sled at the Navy's test facility at China Lake, California, and the flutter was duplicated. North American engineers considered various modifications to solve the problem, but finally settled on the crudest one: They sawed 12 inches off each end of the tail. A few months later, however, an FJ-4 shed its tail during a pullout from a high-speed dive, killing the pilot. The experts went back to work. They then discovered that putting a heavy load on the tail reduced its flutter margins. Another amputation, this time of a mere six inches, solved the problem for good.
Today, the practice of carrying armaments and auxiliary fuel on external pylons, and the vast variety of possible combinations of external loads, make flutter analysis of modern fighters especially difficult. On the other hand, composite structures using carefully controlled arrangements of graphite and other exotic fibers are much stiffer than aluminum or steel, and can even be made to deform under load in such a way as to reduce aerodynamic loads and therefore the chance of flutter.
Increasingly, new fighter and transport designs rely on electronics for stability and control, and in some cases for flutter prevention as well. The F-16, for example, is prone to a non-destructive wing flutter when carrying certain combinations of external loads. The wings flap out of phase--the left wing goes down while the right one goes up--imparting a rocking motion to the fuselage. Rather than modify the wing, researchers at the U.S. Air Force flight testing facility at Edwards Air Force Base in California programmed the fighter's electronic flight control system to sense the flutter and use the ailerons to oppose the wing's flexing. The fix, which will be incorporated in a flight control software upgrade scheduled for 2002, is indicative of what electronic flight controls can do. But they are not a panacea for flutter; the number of control surfaces available on an airplane's wings and tail falls far short of the number of possible flutter modes they can exhibit.
Discernible in the future are "smart" materials that expand or contract slightly in response to electrical signals. "They're like muscles," says Tom Noll, head of the aeroelasticity branch at NASA's Langley center. "They're normally in a neutral state, but they can be 'flexed' when extra stiffness or resistance to deformation is needed." Another possible weapon against flutter comes from the new field of MEMS--micro electro-mechanical systems. Thin surface overlays could raise thousands of tiny spoilers on an electrical command, disrupting airflow and preventing the aerodynamic augmentation that is fundamental to flutter.
Flutter analysis is often called a black science. Even though flutter is today well understood and largely preventable, it is still as formidable a foe as ever, and its malevolent unpredictability remains. "Some fear flutter because they do not understand it," said the famous aerodynamicist Theodore von Karman. "And some fear it," he added, "because they do."