Torsional vibration excitation in rotating machinery can cause system reliability issues or even catastrophic failures. Torsional vibration detection and monitoring becomes an important step in rotating machinery condition monitoring, especially for those machines driven by a variable frequency drive (VFD), a pulse width modulation motor (PWM), or a synchronous motor (SM), etc. Traditionally, the torsional vibration is detected by a phase demodulation process applied to the signals generated by tooth wheels or optical encoders. This demodulation based method has a few unfavorable issues: the installation of the tooth wheels needs to interrupt the machinery normal operation; the installation of the optical barcode is relatively easier, however, it suffers from short term survivability in harsh industrial environments. The geometric irregularities in the tooth wheel and the end discontinuity in the optical encoder will sometimes introduce overwhelming contaminations from shaft order response and its harmonics. In addition, the Hilbert Transform based phase demodulation technique has inevitable errors caused by the edge effect in FFT and IFFT analyses. Fortunately, in many industrial rotating machinery applications, the torsional vibration resonant frequency is usually low and the Keyphasor® and/or encoder for speed monitoring is readily available. Thus, it is feasible to use existing hardware for torsional vibration detection.
In this paper, we present a signal processing approach which used the Keyphasor/encoder data digitized by a high sampling rate and high digitization resolution analog-to-digital (A/D) convertor to evaluate the torsional vibration directly. A wavelet decomposition (WD) based method was used to separate the torsional vibration from the shaft speed, so that the time history of the torsional vibrations can be extracted without significant distortions. The developed approach was then validated through a synchronous motor fan drive and an industrial power generation system. Detailed results are presented and discussed in this paper.
Faults in Oceanic Crust Contribute to Slow Seismic Waves