Identification of Aerodynamic Forcing Mechanisms from Experimental Measurements of Compressor Blade Vibration

Compressors in gas turbine engines experience a wide variety of unsteady aerodynamic forcing mechanisms that can cause blade vibration. High-amplitude blade vibration can lead to failure of compressor blades due to high-cycle fatigue. The mitigation of unsafe blade vibration relies heavily on component and engine tests to identify operating conditions at which blade vibration amplitudes exceed allowable limits. When unacceptable blade vibration amplitudes are observed, it is important to correctly identify their cause. However, ascertaining cause-and-effect relationships from experimental measurements is extremely challenging, due to the complex flow physics and sparse measurements within typical compressor tests. Therefore, the limited experimental measurements of unsteady pressure and blade vibration need to be interpreted in a manner that allows for identification of the aerodynamic forcing mechanism.

Aerodynamic forcing mechanisms originate from the complex, unsteady flow physics that occur within a gas turbine engine. Examples include wakes or potential effects from other components, rotating stall, secondary flows, tip leakages flows, unsteady shock motion, acoustics, turbulence, vortex shedding and self-induced excitation. These aerodynamic effects can occur simultaneously and are highly dependent on compressor design and operating conditions. This dissertation focuses on identification of aerodynamic forcing mechanisms from experimental measurements of compressor blade vibration. Better understanding of the physics relating the aerodynamic forcing to blade vibration allows for improved accuracy when applying mathematical models and analyses. Therefore, the forcing identification methodology presented in this work allows for experimental measurements to be understood in a manner consistent with the mathematical theory.

The literature describes aerodynamic forcing mechanisms and the ensuing compressor blade vibration in a variety of ways. Examples include forced response, synchronous forcing, acoustic resonance, non-synchronous vibration (NSV), convective NSV, and flutter. The terminology is inconsistent, and often it is unclear whether a term refers to the vibration or to the aerodynamic forcing. Also, the connection between the terminology and the mathematical models relating the aerodynamic forcing to blade vibration is often lacking. As such, methods for identifying forcing mechanisms from experimental data rely heavily on analyst subjectivity and expertise. This work addressed this subjectivity by first developing a categorization of aerodynamic forcing mechanisms into the three general categories of 1) external forcing 2) blade-row aerodynamic forcing and 3) motion-dependent forcing. Then, using cyclic symmetry modal coordinates, mathematical models describing the modal force for each category were proposed. It was demonstrated that each category led to distinct temporal characteristics of the vibration response due to their distinct mathematical forms.

The first category, external forcing, is related to synchronous vibration and therefore is well understood. However, for experimental measurements of non-synchronous vibration, it can be extremely difficult to distinguish between blade-row aerodynamic forcing and motion-dependent forcing. This work developed a method for distinguishing between these two categories of forcing mechanisms based on spectral characteristics of the unsteady pressure and blade vibration data. Additionally, the identification method was supported by mathematical analyses for each category. For motion-dependent forcing, a commercially-available analysis tool was implemented: an aerodynamic damping calculation from computational fluid dynamics results in ANSYS CFX. The computational results were shown to be in good agreement with the experimental data. In contrast, no commercially-available analysis tool exists for blade-row aerodynamic forcing.

The blade-row aerodynamic forcing was related to rotating disturbances observed in the unsteady pressure data. There are many examples in the literature of rotating pressure disturbance causing NSV. These are described by terminology such as "rotating instabilities'" or "convective NSV;" however, the flow physics and how they relate to compressor blade vibration remains poorly understood. In this work, the spatial and temporal scales of the flow physics were investigated through 1) detailed analysis of the unsteady pressure data and 2) a novel empirical model for the rotating pressure disturbances. The relationship between the rotating pressure disturbances and blade vibration was then investigated, and blade vibration amplitudes were found to be linearly related to the unsteady pressure data. Specifically, it was found that forced response mathematical analysis could be used to relate the experimentally-observed blade vibration to the rotating pressure disturbances. The analysis demonstrated that the pressure disturbances were independent of blade vibration, thus supporting the identification as blade-row aerodynamic forcing.