Frequency-domain modal analysis with aeroelastic applications

The vibration of a mechanical structure is usually stable, i.e. the oscillatory motion dies out if no excitation is present. If a structure is exposed to a fluid flow, its vibration characteristics will change depending on the speed of the flow. There exists a critical speed (the flutter speed) above which the vibration amplitude starts to increase indefinitely. The flutter speed is an important design parameter for an aircraft's wing and tail surfaces, the blades of a wind turbine or a compressor, etc. All newly developed aircraft must for instance be tested in flight to verify that its vibration is stable for all combinations of altitude and speed it is designed for. Since the vibration's amplitude grows violently when flutter occurs, these tests are very dangerous, time-consuming, and costly. In this dissertation we improve existing methods to deal with the requirements posed by ground and flight flutter testing. On the one hand, the computational speed is important since flutter can occur suddenly. On the other hand, the vibration parameters must be estimated accurately, even though the measurements are perturbed by turbulence and other noise sources. We provide a compromise between computational speed and accuracy, derive uncertainty intervals, and show how one can use the effect of turbulence to advantage. Finally, we introduce a new method to predict the flutter speed based on measurements at lower speed.