Effectiveness of various techniques in reducing noise generated in measuring torsional vibration

摘要

Torsional vibration can be characterized as the cyclic variation of shaft speed, which can cause various failures in rotating machines, such as: gear-tooth breakage, blade-off due to blade fatigue in steam turbines, break-off of shafts, and overloading of components fitted onto the shaft. Commercially, there are only a few systems available that measure this type of vibration as compared to lateral vibration measurement systems. Most of these systems required modifications to the rotating machine, which in some cases are unacceptable. Therefore, it has become common practice to develop in-house torsional vibration measurement systems. A common measurement technique, called Time Interval Measurement (TIMS), calculates the instantaneous speed of the shaft from a frequency modulated carrier wave. Since torsional vibration is the cyclic variation of shaft speed, the shaft speed can be used to determine torsional vibration. Noise can be easily introduced into this type of system masking the torsional vibration; this was apparent in the measurement system developed by Kar, which was used as a baseline for the experiments conducted in this thesis. Various techniques were employed to reduce the effects of the noise in the measurement system, such as (1) created an algorithm, different than the one used by Kar, to calculate shaft speed, (2) increased the sampling rate of the data acquisition boards, (3) resampled the shaft speed into the order domain in order to remove harmonic noise, and (4) created an algorithm that corrects the shaft speed calculation to account for unequal spacing of encoder segments. These noise reducing techniques were compiled into a LabVIEW? program in order to develop a robust measurement system. Each technique was tested individually on two test rigs constructed at the Turbomachinery Laboratory. Each technique proved to reduce the noise introduced into the system, but the geometric compensation algorithm proved to be the most effective in reducing the noise. This thesis proved that an in-house measurement system could be developed at a relatively low cost and with relative ease.