Probabilistic dynamics of mistuned bladed disc systems using subset simulation

The work describes an assessment of subset simulation (SubSim) techniques to increase the computational efficiency for the predictions of probabilistic dynamic behaviour in mistuned bladed disc systems. SubSim is an adaptive stochastic procedure to efficiently compute small failure probabilities, which are expressed as a product of large conditional failures probabilities by introducing intermediate failure events. The original version of SubSim with a classical modified Markov chain Monte Carlo (MCMC) method is used in this work to generate samples related to intermediate failure events. A 2-DOFs model with lumped parameters identified from a high-fidelity finite element model is used to represent a bladed disc. The statistics associated to the maximum forced frequency response amplitudes are evaluated from different levels of the blade mistuning using stiffness perturbations of the blades. Direct Monte Carlo simulations (MCS) are used to benchmark the results from the SubSim. The proposed methodology is shown to capture efficiently the statistical properties of the mistuned blades with less than 5% samples compared to the direct MCS method. Trade-off parametric studies of the SubSim method indicate that 2000 samples at each level yield an overall good computational efficiency and accuracy for the bladed disk system considered in this work. The study confirms that SubSim techniques can be effectively used in stochastic analysis of bladed disc systems with uncertainty related to the blade configurations.