Uncertainty, Sensitivity Analysis, and Design of Experiments in Flutter Testing

Clear Science Corp. proposes to develop and demonstrate the technical merit and feasibility of software that quantifies uncertainty in predictions of the aeroelastic stability of aircraft. The computational tool will accommodate three main sources of uncertainty in the aeroelastic response of an airframe design: 1) uncertainty or variability in structural properties, 2) uncertainty or variability in flight conditions, and 3) uncertainty or errors in the computational and experimental models that are used to determine aeroelastic response. The Uncertainty Analysis (UA) software will be designed to extract an optimal amount of information from a limited set of simulations, facilitating the use of high-dimensional, high-fidelity aeroelastic models in determining flutter boundaries and nonlinear responses. The proposed tool will consist of three sequential components or steps: 1) design-of-experiment (DOE) algorithms that optimize the space-filling properties of initial samples, 2) construction of metamodels that represent input-output relations in compact forms, and 3) global sensitivity analysis techniques that quantify uncertainty propagation and rank the importance of all sources of uncertainty, based on the metamodels. Development and testing of the UA algorithms in Phase I will utilize CFD-based aeroelastic simulations of flutter associated with a two-dimensional airfoil with two degrees of freedom (pitching and plunging).