Bayesian-Based Probabilistic Limit Cycle Oscillation Prediction of Aircraft with Stores

The overall technical objective of the Phase II effort is to extend the Phase I developments and achieve a fully predictive Bayesian-based, data driven, probabilistic prediction tool of the F-16 limit cycle oscillation (LCO). This tool will rely on the available flight test data on some configurations to predict the LCO amplitude on different configurations within an uncertainty band that account for all uncertainties. These predictions will be achieved using a calibrated finite element model of the F-16 and either ZONA Euler Unsteady Solver (ZEUS) or NASA’s FUN3D code as aeroelastic analysis software whichever is faster while providing the necessary accuracy. Both of these codes will be enriched with the nonlinear structural damping (NSD) model developed earlier by the ZONA/ASU Team which captures the changes in damping taking place when the deformations of the structure become large. A separate set of comparisons of the F-16 behavior predicted by ZEUS with NSD and FUN3D with NSD will elucidate the selection of one aeroelastic software over the other. The NSD model includes two tunable damping parameters a and g. Extending the Phase I efforts, the mean values of these two parameters will be correlated to the flight conditions and F-16 with stores configuration to first permit a deterministic computation of the LCO amplitudes. Next, the damping parameters will be considered as random to model the epistemic and aleatoric uncertainties in the predictions/validations with flight test data. Further, their joint distribution will be estimated in a Bayesian format using available historical flight test data on 10 different configurations. Finally, the posterior distribution and uncertainty bands of the LCO amplitude will be predicted for any new configuration/flight test conditions which could then be used to assess, probabilistically, whether the particular stores configuration can be cleared for flight. The Monte Carlo simulations involved in this last phase will be greatly sped up through the use of the fast LCO amplitude surrogate developed and successfully validated in Phase I in the context of ZEUS with NSD. The validity of this surrogate with FUN3D with NSD data is expected and will be confirmed in Phase II.