The literature on reduced-order modeling, simulation, and analysis of the vibration of bladed disks found in gas-turbine engines is reviewed. Applications to system identification and design are also considered. In selectively surveying the literature, an emphasis is placed on key developments in the last decade that have enabled better prediction and understanding of the forced response of mistuned bladed disks, especially with respect to assessing and mitigating the harmful impact of mistuning on blade vibration, stress increases, and attendant high cycle fatigue. Important developments and emerging directions in this research area are highlighted. I. Introduction T URBINE engine rotors, or bladed disks, are rich dynamical systems that are known to suffer from severe vibration problems. Although a bladed disk is typically designed to have identical blades, there are always random deviations among the blades caused by manufacturing tolerances, wear, and other causes. This is called mistuning. Even though mistuning is typically small (e.g., blade natural frequency differences on the order of a few percent of the nominal values), mistuned bladed disks can have drastically larger forced response levels than the ideal, tuned design. The attendant increase in stresses can lead to premature high cycle fatigue (HCF) of the blades. HCF is a major cost, safety, and reliability issue for gas-turbine engines. For example, in 1998 it was estimated by the U.S. Air Force that about 55% of fighter jet engine safety Class A mishaps (over $1 million in damage or loss of aircraft) and 30% of all jet engine maintenance costs were due to HCF. 1 It is clearly of great interest to be able to predict-and, ultimately, to reduce-the maximum blade response as a result of mistuning. The comprehensive modeling, analysis, and understanding of bladed disk vibration is thus critical to reducing the occurrence of HCF and improving the performance and reliability of turbine engines. Bladed disk vibration first received significant attention from the research community in the late 1960s and the 1970s. Notable early work was done by Whitehead, 2 Wagner, 3 Dye and Henry, 4 and Ewins. 5−8 The bladed disk vibration literature has been surveyed by Srinivasan 9,10 and Slater et al. 11 The 1997 survey by Srinivasan 10 Matthew P. Castanier is an Associate Research Scientist in the Department of Mechanical Engineering at the University of Michigan. He received his Ph.D. in Mechanical Engineering from the University of Michigan in 1995. His research interests are in the area of structural dynamics and vibration, including reduced-order modeling, low-to mid-frequency vibration and power flow in complex structures, localization and related phenomena in periodic or cyclic structures, and vibration of mistuned bladed disks in turbine engines. Christophe Pierre is Dean of the Faculty of Engineering at McGill University in Montréal, where he is also Professor of Mechanical Engineering and holds the Canada Research Chair in Structural Dynamics and Vibration. He ...

No abstract

Component mode synthesis (CMS) techniques are widely used for dynamic analyses of complex structures. Significant computational savings can be achieved by using CMS, since a modal analysis is performed on each component structure (substructure). Mistuned bladed disks are a class of structures for which CMS is well suited. In the context of blade mistuning, it is convenient to view the blades as individual components, while the entire disk may be treated as a single component. Individual blade mistuning may then be incorporated into the CMS model in a straightforward manner. In this paper, the Craig–Bampton (CB) method of CMS is formulated specifically for mistuned bladed disks, using a cyclic disk description. Then a novel secondary modal analysis reduction technique (SMART) is presented: a secondary modal analysis is performed on a CB model, yielding significant further reduction in model size. In addition, a straightforward non-CMS method is developed in which the blade mistuning is projected onto the tuned system modes. Though similar approaches have been reported previously, here it is generalized to a form that is more useful in practical applications. The theoretical models are discussed and compared from both computational and practical perspectives. It is concluded that using SMART, based on a CB model, has tremendous potential for highly efficient, accurate modeling of the vibration of mistuned bladed disks.

New techniques are presented for generating reduced-order models of the vibration of mistuned bladed disks from parent finite element models. A novel component-based modeling framework is developed by partitioning the system into a tuned bladed disk component and virtual blade mistuning components. The mistuning components are defined by the differences between the mistuned and tuned blade mass and stiffness matrices. The mistuned-system model is assembled with a component mode synthesis technique, using a basis of tuned-system normal modes and attachment modes. The formulation developed is general and can be applied to any mistuned bladed disk, including those with large geometric mistuning (e.g., severe blade damage). In the case of small (i.e., blade frequency) mistuning, a compact reduced-order model is derived by neglecting the attachment modes. For this component mode mistuning model, the blade mistuning is projected first onto the component modes of a tuned, cantilevered blade, and then projected again onto the tuned-system normal modes via modal participation factors. In this manner, several natural frequencies of each mistuned blade can be used to capture systematically the effects of the complex physical sources of mistuning. A numerical validation of the developed methods is performed for both large and small mistuning cases using a finite element model of an industrial rotor. Nomenclature c = aerodynamic coupling damping matrix projected onto tuned-system normal modes E = Young's modulus F = real Fourier matrix f = excitation force vector j = 1 p K = stiffness matrix in physical coordinates M = mass matrix in physical coordinates N = no. of blades N h = no. of the retained tuned-system normal modes corresponding to harmonic h p = modal coordinates q,q h = set of tuned-cantilevered-blade mode participation factors for the blade motion in the retained tunedsystem modes in physical coordinates and in cyclic coordinates U = set of the retained tuned-cantilevered-blade normal and boundary modes v, v = mistuned-cantilevered-blade normal mode participation factor for a tuned-cantilevered-blade normal mode and set of the factors x = physical coordinates = structural damping coefficient = nondimensional mistuning parameter = reduced stiffness matrix or stiffness projection to the retained component modes , = eigenvalue and diagonal matrix of eigenvalues of the retained component normal modes = reduced mass matrix or mass projection to the retained component modes , = set of the retained component normal modes in physical coordinates and in cyclic coordinates = component interface modes ! = frequency Subscripts b, i = finite element degrees of freedom (DOF) of the cantilevered-blade boundary and interior H = maximum harmonic number h = harmonic number n = blade number o = tuned blade R = set of the retained cantilevered-blade normal mode numbers r = rth cantilevered-blade normal mode = finite element DOF of the blades = finite element DOF of the disk = generalized coordinates for the retained component normal m...

The analysis of the response statistics of mistuned turbomachinery rotors requires an expensive Monte Carlo simulation approach. Simple lumped parameter models capture basic localization effects but do not represent well actual engineering structures without a difficult parameter identification. Current component mode analysis techniques generally require a minimum number of degrees of freedom which is too large for running Monte Carlo simulations at a reasonable cost. In the present work, an order reduction method is introduced which is capable of generating reasonably accurate, very low order models of tuned or mistuned bladed disks. This technique is based on component modes of vibration found from a finite element analysis of a single disk-blade sector. It is shown that the phenomenon of mode localization is well captured by the reduced order modeling technique.