The first phase of the ATW program was pre-flight ground testing. This testing was performed in consideration of both static and dynamic properties of the ATW. Deflection tests were performed to determine the sizes of static loads that could be borne by the ATW. Vibration tests were also performed to determine the dynamic modal characteristics of the ATW. It was shown that the first bending and torsion modes were at frequencies of 14.05 and 22.38 Hz. The data from these tests were used to generate computational models for predicting the onset of flutter.

The second phase of the ATW program was a flight test for envelope expansion. In this phase, it was required to analyze experimental data acquired at a series of test points with increasing velocity and dynamic pressure. At each test point, the ATW was excited with a frequency-varying input and its responses were measured by sensors. The resulting flight data were telemetered to a control room and analyzed by the flutter-prediction methodologies. A total of five flight tests were performed in April 2001.

Figure 2. This Accelerometer Response shows the onset of flutter.
A flutter instability of the ATW was encountered at approximately mach 0.83 at an altitude 10,000 feet (≈3 km). The wing was broken such that the boom and roughly 30 percent of the wing were lost. The pieces fell to the ground without striking the F-15B aircraft or FTF-II testbed. The flutter incident was quite demonstrative of the phenomenon. The instability was encountered during a slow acceleration from mach 0.825 to mach 0.830. The damping changed dramatically during this acceleration. The ATW went from stable, with accelerometer responses going from approximately 3 g (where g = the standard gravitational acceleration at the surface of the Earth) to unstable during a time interval of only 5 seconds (see Figure 2).

The data from the ATW were analyzed to evaluate the flutter-prediction methodologies. The results indicated that computational models were able to predict the onset of flutter reasonably well, but that small errors in the models could cause large errors in the predictions. The traditional approaches for analyzing flight data were shown to afford a capability to predict the onset of flutter only during operation at flight conditions near the instability. The flutterometer was shown to be somewhat conservative in the worst-case estimates of flutter, but it presented a reasonable prediction of flutter at flight conditions that were far from the instability.

This work was done by Rick Lind, David F. Voracek, Tim Doyle, Roger Truax, Starr Potter, Marty Brenner, Len Voelker, and Larry Freudinger of Dryden Flight Research Center and Cliff Sticht of Ames Research Center. For further information, contact the Dryden Commercial Technology Office at (661) 276-3689. DRC-01-37.