This paper summarizes an analysis of data obtained from an instrumented compressor of an operational, heavy duty industrial gas turbine; the goal of the aforementioned analysis is to understand some of the fundamental drivers, which may lead to compressor blade vibration. Methodologies are needed to (1) understand the fundamental drivers of compressor blade vibration, (2) quantify the severity of “events,” which accelerate the likelihood of failure and reduce the remaining life of the blade, and (3) proactively detect when these issues are occurring so that the operator can take corrective action. The motivation for this analysis lies in understanding the correlations between different sensors, which may be used to measure the fundamental drivers and blade vibrations. In this study, a variety of dynamic data was acquired from an operating engine, including acoustic pressure, bearing vibration, tip timing, and traditional gas path measurements. The acoustic pressure sensors were installed on the first four compressor stages, while the tip timing was installed on the first stage only. These data show the presence of rotating stall instabilities in the front stages of the compressor, occurring during every startup and shutdown, and manifesting itself as increased amplitude oscillations in the dynamic pressure measurements, which are manifested in blade and bearing vibrations. The data that lead to these observations were acquired during several startup and shutdown events, and clearly show that the amplitude of these instabilities and the rpm at which they occur can vary substantially.
Optimizing Automated Gas Turbine Fault Detection Using Statistical Pattern Recognition
