Gene Vogel
EASA Pump & Vibration Specialist
The most basic tool a vibration analyst uses is the vibration spectrum. The spectrum is a graphic illustration of the frequencies present in a vibration signal and their relative amplitudes. A good way to understand the spectrum is that it is a “bar graph” of the frequencies, with hundreds of individual vertical frequency “bars” across a range of frequencies. Most spectra are displayed with only a single dot for the highest amplitude in each frequency bar, so the graph appears as a jagged line reflecting those highest amplitudes for each bar. The highest frequency in the graph is called the fmax and the number of bars in the graph is known as the “number of lines of resolution.”
Setting the fmax and the number of lines of resolution is a key initial step for the vibration analyst (see Figure 1.)
For most common analysis tasks, 400 lines of resolution is adequate. Higher resolution may be needed for some more specialized tasks. Setting the proper fmax is a little more complex. To evaluate common faults other than rolling element bearing and gear mesh, an fmax of 400 Hz (24,000 CPM) is adequate; however, rolling element bearing and gear mesh can generate vibration at much higher frequencies, up to 5 kHz. A good practice is to survey the machine with a 5 kHz fmax; if there are no significant amplitudes present at the higher frequencies, set the fmax to 400 Hz for analysis of the common machine faults.
Many vibration analysis instruments provide a special high frequency detection mode for evaluating vibration energy from rolling element bearing, gears and similar mechanisms where friction is a factor. Some common proprietary detection modes include gE, gSE, PeakVue and SPM. Some analysts will use one of these detection modes to assess the need for recording higher fmax spectra.
Several vibration standards, ISO 10816, IEC 60034, NEMA MG1 and EASA AR100 use velocity amplitude units for general acceptance criteria. Amplitude units of acceleration or displacement may be applicable for some special application, but unless there is a special concern for looking at high frequency or low frequency signals, velocity units are best for general spectrum analysis. Velocity units may be in/sec pk, mm/sec pk or mm/sec rms. Be sure to note the proper units on any printed data.
Most vibration analysis instruments and software will by default auto-range spectra displays. While this makes it easier to distinguish individual peaks in the spectrum, it makes it more difficult to visually compare spectra from different locations on the machine to assess machine condition. A good practice for overall evaluation of machine condition is to manually set the same amplitude scale for all spectra. Also, since it is common to record spectra for the horizontal (H), vertical (V) and axial (A) directions at each bearing location, it is helpful to standardize the display format of spectra to three spectra per page with consistent H-V-A orientation (see Figure 2).
Printed, or digital image, spectra can be useful for preliminary evaluation of machine condition or for report documents, but vibration analysis instruments and software have powerful cursor tools that hold the secret to accurate assessment of specific machine faults. Most systems have both a harmonic cursor and a sideband cursor tool. These are primary tools used for spectrum analysis. By identifying families of peaks in the spectra, the fundamental frequency of each family of peaks will strongly suggest possible fault sources. Integer harmonics of rotating speed (1xrpm, 2xrpm, 3xrpm, etc.)kl,m often indicate mechanical looseness or shaft alignment faults, depending on the pattern. Half orders are typical of a rub. Peaks related to line frequency are obviously electrical in nature, and of course, a dominant 1xrpm peak results from rotating imbalance (see Figure 3).
Sidebands around a dominant peak in the spectra indicate modulation of that dominant peak. A classic example is a cracked inner race of a low-speed bearing. The amplitude of the vibration at the ball pass inner race (BPFI) bearing fault frequency is modulated as the crack rotates in and out of the load zone. Some instruments and many software analysis programs have sideband cursor tools to help identify sideband frequencies.
Beyond an initial assessment of machine condition, it is often necessary to take a more accurate look at specific frequencies in the spectrum. Expanding the display (zooming in) does not increase the resolution of a spectrum. To more accurately determine the frequency of a peak, a new spectrum must be recorded from the machine. The fmax and the lines of resolution settings directly affect the resolution of a spectrum; that is, how accurately the frequency of a peak is indicated. Most spectrum analyzers have 3200 lines of resolution available; some have even more. High resolution spectra are often used to separate 2x vibration on 2 pole induction motors, where 2xrpm and 2x line frequency (2xlf) are very close together.
An understanding of spectrum analysis, how to set up for data acquisition and how to use tools available in an instrument or software program are fundamental skills for vibration technicians. An understanding of the basics is essential for anyone in the service center who may have to work with customers on machine performance issues. EASA has several technical resources that address those member needs. The Vibration Analysis for Service Centers webinar series recordings is a training program suitable for any interested service center member. The accompanying handbook is an excellent reference source, along with more than a dozen related technical articles in the EASA Resource Library at easa.com. There are some complex aspects of vibration analysis, but an understanding of the basics will help everyone who has concerns for rotating equipment condition.
AVAILABLE IN SPANISH
ANSI/EASA AR100
More information on this topic can be found in ANSI/EASA AR100
Print