Blade tip timing method: advanced measurement data processing

Abstract

This report is done as a part of my thesis work concerning development and application of different advanced data processing methods for blade tip timing experimental data. Initial work involves understanding the basics behind blade tip timing measurement method, critical issues affecting the implementation of these methods and the challenges faced in data processing. Then, the existing auto-regressive, sine fitting and spectral data processing methods are applied to an industrial test case to investigate the implementation of these advanced methods and to examine their feasibility towards industrial data. Operation of these different methods are studied and the frequencies/engine orders computed from each method are compared with the Campbell data (FEM) for validation. Feasibility and limitations of different methods are analyzed and discussed. Then, different windowing techniques are applied to the basic spectral methods such as non-uniform Fourier transform and Lomb-Scargle periodogram to study the effects of windowing the data on frequency spectrum results. Cross spectrum is applied to non-uniform Fourier transform and Lomb-Scargle periodogram to study its impact on identifying true frequencies in case of severe corruption of replicas. Next, sine fitting method is developed for two frequency identification i.e. to detect simultaneous resonances. This method is tested and validated using the data generated from multi-mode tip timing simulator which is updated from the existing blade tip timing simulator. Finally, the feasibility of this methods is analyzed and discussed.