Comparison of Natural Frequencies for Detection of Cracked Rotor Wheels
Abstract
High cycle fatigue, regarding turbine engines, is one of the main causes of rotating component failure. Specifically, the blades of the wheels in the fan, compressor, and turbine sub-assemblies. Traditionally strain gauges are employed as a means of measuring blade vibration during component or full engine development testing. For rotating machinery, strain gauges require the use of a slip ring or a telemetry package. This becomes increasingly complicated as the number of strain gauges increase, thus the need for a more non-intrusive measurement capability for discernment of blade stress responses. Non-Intrusive Stress Measurement Systems (NSMS) allow engineers to detect high cycle fatigue (HCF) issues prior to component failure. It is important for the turbine engine industry to monitor for high cycle fatigue issues to maintain a fleet readiness. When unexpected HCF causes component or system failure the potential consequences are grounded fleets, cancelled flights, monetary loss, and loss of life. Once these issues occur an investigation is initiated and could take a few weeks to several months or more to resolve. This time impacts the engine companies as well as the people dependent upon functional engines. HCF monitoring processes and techniques are crucial to preserving fleet maintenance. One of the ways to prevent premature HCF failure is by detecting cracks in the blades or the wheels of the rotor. It is the subject of this thesis to determine whether the static deflection of the blade as it rotates will begin to grow independent of rotational changes experienced by the rotor for an internal crack in the wheel as opposed to the blade of a rotor. Should a crack in the wheel occur, the stiffness should decrease, which would manifest when testing the rotor’s natural frequencies as a decrease in the natural frequency compared to an un-cracked rotor. The experiment was conducted using analysis tools for predicting blade natural frequencies of the pre-cracked rotor as well as physical experiments to determine the natural frequencies of the post-cracked rotor. The spin facility set up, data acquisition, data reduction, experiment details and results are provided. Both strain gauges and NSMS techniques were used to measure the natural frequencies of the rotor, and detection of damage while mounted in the spin facility. This research effort concluded it is possible to detect a crack in the wheel of a rotor using the NSMS blade stack capability. It is necessary to have a baseline vibration survey to understand the pre-damaged static deflection of each blade. This research also concluded that a comparison of the pre-cracked and post-cracked natural frequencies manifested roughly a 5% decrease. With a crack in the wheel, the expected stiffness of the wheel would decrease, thus, causing a decrease in the natural frequency of the component. This is evident in the comparison of the pre-cracked ping test data and the post-crack bench test data. In summary, it is possible to detect an internal crack of a rotor and the natural frequencies of the blades can change with an internally cracked wheel.
Degree
M.Sc.
Advisors
Key, Purdue University.
Subject Area
Design|Electromagnetics|Industrial engineering|Physics
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.