Flutter stability prediction using the unsteady aerodynamic reduced-order model

Abstract

CFD-base aeroelastic simulations are recognized as the most accurate schemes because they have significant advantages in terms of accuracy of nonlinear flow calculation especially in transonic and low supersonic regimes. However, in spite of their accuracy, application of this approach to an actual aircraft model is extremely time consuming. Moreover, because the aeroelastic instabilities can only be predicted by simulating whether the response is decaying or diverging, the aeroelastic analysis should be conducted several times until the neutral response is found. Therefore, to overcome these difficulties, this paper has introduced an efficient aeroelastic analysis methodology based on reduced-order modeling (ROM), whereby unsteady aerodynamic forces are determined without direct aerodynamic computations. In this study, two reduced-order models (ROM) for unsteady aerodynamics are formulated and the aeroelastic analysis codes using the ROM were developed. System identification and proper orthogonal decomposition (POD) methods are applied to construct the reduced order models of unsteady aerodynamics. In the system identification approach, the aerodynamic system is considered as a black box and the modal responses and generalized aerodynamic signals are captured to construct the Auto Regressive Moving Average (ARMA) model. This model offers full information of the relationship between the wing motion and the unsteady aerodynamic loads in the aeroelastic analysis instead of the aerodynamic solver. This approach is very simple and significantly faster than the CFD-based aeroelastic analysis. However, it shows distinct limitation when the aerodynamic system has an uncertainty, such as control surface motion. In the POD method, the aerodynamic system is projected into a K-L basis or optimal basis set. Thus the aerodynamic system can be expressed by the modal superposition such that the deformation of the structure can be described by a linear combination of these mod...