Structures that consist of a cyclic
assembly of identical substructures are said to possess cyclic
symmetry. The design of a single stage of a turbomachinery rotor,
or bladed disk, typically features cyclic symmetry. Using cyclic
symmetry routines in commercial finite element software enables
one to analyze the entire bladed disk using a model of only one
sector, which results in considerable reduction of computational
costs. However, there are always small structural deviations among
blades, called mistuning. Although mistuning is typically small
(e.g., blade-alone natural frequency variations of 1-5%), it can
lead to a dramatic increase in maximum blade stress and is a major
driver for high cycle fatigue in turbine engines. From a modeling
perspective, mistuning destroys the cyclic symmetry, such that the
full bladed disk needs to be modeled. To address this issue, we
develop efficient approaches for reduced-order modeling of
mistuned bladed disks.
These reduced-order models treat
mistuning as a variation in natural frequencies for one or more
modes of an isolated blade. For bladed disks with inserted blades,
one can measure the natural frequencies of manufactured blades
directly. However, this is not possible with integrally bladed
disks (blisks), which are becoming more prevalent in newer and
next-generation engines. For blisks, we have developed several
mistuning identification methods that use experimental response
data for the full blisk in order to extract individual blade
mistuning parameters. The combination of work on modeling and
identification of mistuning in blisks in our research group has
made it possible to accurately model blisks at a fraction of the
computational cost compared to finite element analysis.
|