| The vibratory response of a turbine blade is very sensitive to small changes in the blade properties. Therefore, the variations that come from the manufacturing process and wear can significantly increase a blade's vibratory stress at resonance. This phenomenon, referred to as mistuning, can cause blades in a gas turbine engine to fail from high cycle fatigue.; A new reduced order model of mistuned bladed disk vibration is presented. This new approach is shown to accurately represent the response of real turbine geometries when only a single family of modes is excited. Yet its mathematical form is even simpler than that of a mass-spring model. Because it requires only minimal input data, this model is much easier to use than previous reduced order methods. Furthermore, its simplicity allows the fundamental parameters that control mistuning to be readily identified.; This model is then used to develop a set of analysis tools to assess the mistuning in an engine component and predict its vibratory response under rotating conditions. These tools include a completely experimental method of system identification which uses vibratory response measurements of the bladed disk system as a whole to infer its mistuning. As a system-based method, this approach is particularly well suited to integrally bladed rotors, whose blades cannot be removed for individual testing. Next, an analytical method is developed to adjust mistuning for the effects of centrifugal stiffening. The approach allows mistuning measured from a stationary rotor to be used to predict the part's forced response under rotating conditions. Finally, the system identification method is extended to allow forced response measurements from a rotating test to be used as input. The mistuning measured through this approach will reflect all effects present during the operating conditions. One particular advantage of this method is that it can capture the effect of centrifugal loading on the mistuning of conventionally attached blades. These techniques are verified through a series of experimental and numerical benchmarks. |
