Analytical and Experimental Evaluation of Multivariable Stability Margins in Active Flutter Suppression Wind Tunnel Tests

This work addresses control law synthesis for active flutter suppression followed by the design and development of an active flexible wind tunnel model representative of high aspect ratio commercial aircraft. An initial early-design math model of the structural dynamics, unsteady aerodynamics, sensing, and actuation of the system was used to synthesize three control laws that would stabilize the system against flutter over a range of speeds below and above the passive open-loop flutter speed. Three methods for establishing closed-loop robustness measures were used to quantify the robustness of the system based on its initial mathematical model. The system was then tested with these control laws, and their robustness to system variations in the wind tunnel was studied. Such variations included speed as well as gain and phase in the control loops, representing gain and phase uncertainties in the system. This was followed by revisiting the robustness of the control laws with mathematical testing, this time with a more accurate mathematical model of the system. The work highlights the importance of (a) working with as high accuracy as possible math models of the aeroservoelastic system, (b) understanding the key sources and types of uncertainties possible, (c) evolving the control laws during the vehicle development process to account for initial uncertainties in the models that can be large, (d) reaching final control laws that demonstrate sufficient robustness (as required by certification agencies) with the most accurate math models available, and (e) confirmed robustness with tests in flight at selected critical conditions subject to selected variations in the system. The work adds insight regarding the kind of model variations that the control designer needs to consider and the way different measures of system robustness compare to one another and relate to the capacity of the control design to permit model uncertainties.