Forced response of mistuned bladed disks: Identification and prediction

The present investigation focuses on both the estimation of the parameters of a structural model to represent `at best' a set of measurements of the response of a mistuned bladed disk (identification) and the determination of the statistical distribution of the amplitudes of blade vibration given the model parameters (prediction). A mixed least squares-maximum likelihood formulation is presented which is shown to recover well the true model parameters from noisy simulated response data. The prediction aspects of the problem are then investigated by relying on the newly devised combined Closed Form - Perturbation (CFP) approach. The formulation of this technique is extended to the case of both stiffness and damping mistuning and new numerical results are presented that confirm the usefulness of the CFP method in predicting the statistical distribution of the amplitudes of blade response.