Hide/Show Apps

On the non-linear vibration and mistuning identification of bladed disks

Yümer , Mehmet Ersin
Forced response analysis of bladed disk assemblies plays a vital role in rotor blade design and has been drawing a great deal of attention both from research community and engine industry for more than half a century. However because of the phenomenon called ‘mistuning’, which destroys the cyclic symmetry of a rotor, there have been several difficulties related to forced response analysis ever since, two of which are addressed in this thesis: efficient non-linear forced response analysis of mistuned bladed disks and mistuning identification. On the nonlinear analysis side, a new solution approach is proposed studying the combined effect of non-linearity and mistuning, which is relatively recent in this research area and generally conducted with methods whose convergence and accuracy depend highly on the number of degrees of freedom where non-linear elements are attached. The proposed approach predicts nonlinear forced response of mistuned bladed disk assemblies considering any type of nonlinearity. In this thesis, special attention is given to the friction contact modeling of bladed disks which is the most common type of nonlinearity found in bladed disk assemblies. In the modeling of frictional contact a friction element which enables normal load variation and separation of the contact interface in three-dimensional space is utilized. Moreover, the analysis is carried out in modal domain where the differential equations of motions are converted to a set of non-linear algebraic equations using harmonic balance method and modal superposition technique. Thus, the number of non-linear equations to be solved is independent of the number of non-linear elements used. On the mistuning identification side, a new method is enclosed herein which makes use of neural networks to assess unknown mistuning parameters of a given bladed disk assembly from its assembly modes, thus being suitable for integrally bladed disks. The method assumes that a tuned mathematical model of the rotor under consideration is readily available, which is always the case for today’s realistic bladed disk assemblies. A data set of selected mode shapes and natural frequencies is created by a number of simulations performed by mistuning the tuned mathematical model randomly. A neural network created by considering the number of modes, is then trained with this data set for being used to identify mistuning of the rotor from measured data. On top of these, a new adaptive algorithm is developed for harmonic balance method, several intentional mistuning patterns are investigated via excessive Monte-Carlo simulations and a new approach to locate, classify and parametrically identify structural non-linearities is introduced.