Vibration as Measurement Object in Predictive Maintenance
And
Predictive Maintenance Basic Procedure Based on Vibration Measurement
Predictive maintenance (PdM) is maintenance method which is based on actual machine condition. In this maintenance method, there are several measurement objects which are commonly used such as vibration, pressure, sound, oil viscosity, temperature, etc. However, measurement object selection is really depended on the type of machine. For example, measurement objects for common rotary machine are different from those which are used for steam boiler. For common rotary machine, measurement objects which are commonly used are vibration, sound, and performance. For steam boiler, measurement objects which are used are leakage, temperature, pressure, and spark detection.
Now, let’s take an emphasis on rotary machine since this type of machine is commonly used in industry. Based on practical experience and many research results, vibration is the best PdM measurement object for rotary machine. In rotary machine, dynamic forces, which act on the machine, generate machine vibration response. These dynamic forces are generated by fault which is inside the machine. Because of that, machine vibration response can be used for machine fault early detection since its level correlates with the machinery fault severity.
Basically there are several particular steps which should be done in predictive maintenance based on vibration level monitoring. First step is vibration data collection process. The vibration data is measured in time domain (waveform). The vibration measurement can be done either periodically or continually. After vibration measurement is done, the second step is vibration data analysis using FFT algorithm. FFT (Fast Fourier Transform) is algorithm which converts vibration waveform (time domain) into vibration frequency spectrum in frequency domain. By having vibration frequency spectrum, we can continue to the next step which is vibration signature observation. Vibration signature is the vibration response frequency component which has high magnitude in frequency domain. This frequency component represents the type of machinery fault that exists in the machine. Based on this observation result, machine condition diagnostic (the third step) is done. This diagnostic encompasses fault source prediction. Furthermore, the prediction result is compared with other analysis results which come from other PdM object measurement such as temperature and pressure. The last step is breakdown time prediction. If the maximum overall vibration level (OVL) is known, the breakdown time can be predicted using overall vibration level curve trending. From this information, we can calculate the available time which can be used to prepare reparation procedure and spare parts procurement and hence the reparation process can be well prepared.