Assessment of Uncertain External Store Aerodynamics Using mu-p Flutter Analysis

Abstract

F LUTTER testing is rarely performed on full-scale aircraft, due to the high risk for structural damage and failure. Instead, flutter boundaries are computed based on a numerical model of the aircraft, and flight testing under less critical conditions can be performed to collect data for validation of the numerical model. Clearly, problems arise when the flight-test data show deviation from the numerically predicted data. Deviations are likely to occur for any aircraft configuration, because model imperfections and simplifications always lead to some uncertainty in the numerical model, and the question arises as to the impact that these uncertainties have on the predicted flutter boundaries. Inmost cases, some parts of the numerical model, such as themass properties of different components, are well known, whereas other elements, such as the aerodynamic loads in some region of a wing with complex geometry, are known to be subject to uncertainty. Earlier studies [1–3] have shown how aerodynamic uncertainties can be introduced in the numerical model based on physical reasoning and known modeling difficulties. The same technique for modeling of aerodynamic uncertainty will be applied in this study, but it will also be shown how data points collected at subcritical conditions can be used to establish an uncertaintymodel that is capable of producing reliable flutter boundaries. In the present study, a wind-tunnel model was specifically designed to demonstrate and evaluate uncertainty modeling approaches. A fairly simple wing geometry and structure is chosen to minimize the errors introduced by modeling simplifications. An external store in the form of a wing-tip missile is used to increase the model complexity gradually, and the impact on the flutter behavior is investigated both numerically and experimentally. In cases in which the numerical predictions deviate from the experimental results, an uncertainty description is developed and -p flutter analysis and model validation [4] are applied to compute bounds on the flutter speed.