Parametric Reduced Models for Limit-Cycle Oscillation Prediction

Predicting the limit–cycle oscillations of large–scale aeroelastic systems comes at significant cost. This high cost prohibits use of traditional limit–cycle analysis methods in parametric searches, for example, for multidisciplinary design optimisation, or for design search exploration. This work proposes use of a parametric nonlinear reduced order model to enable rapid envelope exploration of limit–cycle oscillation amplitudes. Source–transformation automatic differentiation is used to compute parametric nonlinear reduced model codes of the full order model code. The method exploits information readily available from a flutter search, and does not require the generation of new data or a new reduced basis. The resultant reduced models are used to define a flight envelope of non–destructive limit–cycle oscillations. Reduced model predictions are consistently conservative and the defined envelope has errors that are at most 5%, and all computed at a fraction of the cost of the full model.