Machine condition monitoring and predictive maintenance relies upon early detection of changes in the vibration spectrum that are indicative of defects or wear. As market trends lead toward the monitoring of higher-volume and smaller systems, the signal bandwidth of interest also increases. It is now also possible to embed accelerometers directly into equipment for condition monitoring. This white paper will illustrate the requirement of wide bandwidth monitoring with a discussion of possible sources along with both simulation and experimental results of defect conditions.
Most machinery that is driven by large electric motors will have an input drive running at a speed of between 900 and 3600 rpm (15 to 60 Hz). This fundamental frequency can easily be monitored using accelerometers with relatively low bandwidth, but the information that this signal yields may be limited to imbalance condition only.
Rolling-element bearings are a common cause of machine failure. As bearing defects begin to occur, the vibration level produced in the early stages of failure may be very low and could be masked by vibration from other sources. Frequency-domain analysis is very useful for revealing small signals that may be hidden in the time-domain. Bearing characteristic frequencies are generally many times higher than the running speed (around 50% of the number of balls times the running speed) and may be present even when equipment is in good condition. Bearing defects tend to be impulsive, which produces a series of harmonics spaced at the characteristic frequency, extending to very high frequencies.
Gears represent another case where high multiples of running speed/frequency will be generated, although gearboxes can be significantly more complex to analyze. The basic gear-mesh frequency is given by the product of the number of teeth and the running speed in Hz. The magnitude of sidebands or harmonics of this frequency can indicate the onset or severity of various defects. Gears will also possess their own natural frequencies, which may be excited by impulsive defects or by strong vibration at higher frequencies.
Blades and vanes again generate vibration at a multiple of running speed, even when in good condition. The blade passing frequency is given by number of blades times the running speed. A missing blade will generate a large number of harmonics of the blade-pass frequency (as well as increasing imbalance at 1 x running frequency).
Mechanical looseness is also very likely to cause many harmonics of running speed to appear in the spectrum. In general terms, looseness can cause discontinuities in motion, and sudden changes in position or displacement can produce very high levels of acceleration.
With Exosite's Condition Monitoring Solution you can visualize data points with just a few steps. Once your hardware device registered you have access to multiple visualization and analytics tools.
To learn more about wide bandwidh accelerometers, read TE's whitepaper - here is a download link.