| The blades of a turbine engine rotor always have small, random variations in their dynamic properties due to manufacturing tolerances, wear, and other factors. These structural variations, or mistuning, can lead to localization of the vibration energy in certain blades. As a result, the maximum blade vibration and stress levels may be much higher for an actual, mistuned system than would be predicted based on the ideal, tuned design. Therefore, there has been a large amount of research devoted to understanding and predicting the structural dynamic behavior of mistuned rotors. Almost all of the work to date has been limited to single stages, or isolated bladed disks, with the assumption that the single stage vibration can be analyzed independently from the other rotor stages. However, it is shown in this dissertation that a single-stage model may provide dramatically different response predictions, compared to a true multistage model, for some operating conditions. Furthermore, multistage rotors exhibit certain types of modes and response patterns that are extended across multiple stages, which cannot be captured at all by single-stage models. The purpose of this research is to develop efficient modeling methods that extend the predictive capabilities of mistuned bladed disk simulations to multistage rotors. A key challenge in modeling multistage rotors is that the coupling among stages destroys the cyclic symmetry of the system, even for the tuned case, due to the different number of blades on each stage. In this work, this obstacle is overcome by adopting a componentbased reduced-order modeling approach, in which each stage is treated as a substructure. The new modeling technique requires only single-sector finite element models of the individual stages in order to construct a multistage model. The resulting reduced-order model is extremely compact, yet it captures the effects of blade mistuning on any or all stages. In addition, methods for system identification of multistage rotors are developed, and important applications to structural health monitoring and damage detection are explored. |
