Metamodelling Based on High and Low Fidelity Model Interaction for UAV Gust Performance Optimization

Abstract

The size of small flexible winged Unmanned Air Vehicles (UAVs) makes them particularly susceptible to gusts. Accordingly, a gust performance evaluation problem is formulated where the simulation of the gust response is treated as an aeroelastic problem. Generally, a high-fidelity simulation can result in a large computing effort that, being multiplied by a large number of calls for the aeroelastic simulation in the design optimization process, could make the latter prohibitively computationally expensive. Therefore, a multifidelity technique is employed, where a metamodel of the high-fidelity simulation response is built based on a tuned lower fidelity aeroelastic model. In order to address and validate such a multifidelity modelling approach, the linear aeroelastic equations of a 2D airfoil employing quasi-static aerodynamics are used for the lowfidelity model and solved analytically, whereas the nonlinear aeroelastic equations of a 2D airfoil employing unsteady aerodynamics are used for the high-fidelity model and solved numerically. Three different approaches to the low-fidelity model tuning are investigated and compared, both explicit and implicit. An application of the Moving Least Squares Method (MLSM) to the low-fidelity model tuning is also investigated. Good agreement between the high-fidelity model and the corrected low-fidelity model shows that a multifidelity model-based strategy is suitable for use in optimising the gust performance of small UAVs even in the presence of large structural deformations of their wings.