If you follow an organized process, the time spent analyzing spectra will be far more effective. Here is a brief review:
1. Ski-slope check: Quickly check for ski-slope in case it was a bad measurement. Check the time waveform to see if you can determine what went wrong during the measurement.
2. Noise floor check: Noise can indicate that impacting or rubbing is occurring. It can also be a sign of failed data collection. Check the time waveform for the source of noise. If the noise floor is raised in a limited band, consider resonance.
3. Check twice line frequency: The peak at 120 Hz or 100 Hz is related to electric motor faults (laminations, eccentricity, soft foot, etc.). The presence of a peak is not abnormal; if the amplitude increases it should be checked.
4. Order normalize: Find the shaft turning speed peak (1X); this is the "first order". If you are analyzing spectra from a machine with multiple shafts, identify each turning speed peaks.
a. If the amplitude of the 1X peak is high consider unbalance (if in radial direction), unbalance in an overhung machine (axial direction), misalignment (axial direction), bent shaft/rotor bow (radial), pulley eccentricity (radial), flexibility or weak foundations (horizontal), and consider resonance (typically horizontal or vertical).
b. If you also see 2X, and/or 3X, and/or 4X then suspect misalignment (or cocked bearing).
c. If there are additional harmonics, consider coupling wear, and if the noise floor is raised, consider rotating looseness.
5. Relate peaks and patterns to orders: You should now identify which peaks are exact multiples of running speed (i.e. synchronous, e.g. 6X, 15X, 48X turning speed), which peaks are non-synchronous (i.e. non-integer multiples, e.g. 3.67X, 9.34X, etc.), which are sub-orders (i.e. fractional orders, e.g. 1/2 X, 1/3 X, etc.), and which are simply sub-synchronous (i.e. less that 1X, but not fractional, e.g. 0.48X, 0.76 X, etc.)
a. Synchronous: The vibration is likely to be related to a rotating element: suspect flow problems (orders equal number of pump vanes, fan blades, compressor vanes, etc.), and gear problems (peak at the gear mesh frequency), motor problems (orders equal number of rotor bars, number of stator slots, etc.).
b. Sub-synchronous: Suspect belt faults (there should be harmonics), turbulence (a broad peak), cage faults (rolling element bearings), oil whirl (journal bearing) or vibration from another machine.
c. Non-synchronous: Suspect rolling element bearing damage (expect harmonics, check for 1X or FTF sidebands) or external source of vibration. Could also be due to combinations of other frequencies.
6. Harmonics and sidebands: Both are common and both are clues to the type of fault.
a. Weak harmonics: Suspect "non-linear" vibration; that is, the vibration that is not smooth; something may be restricting the rotation.
b. Strong harmonics: Suspect that impacting is occurring (bearing damage, gear damage, etc.).
c. Sidebands: Any time you see sidebands, it means that vibration (internal forces) are rising and falling periodically, or the speed is fluctuating periodically. There are many possible reasons.
7. Additional tips: Do not rely on linear spectra alone. There are hidden treasures to be found in spectra, waveforms, phase, and high frequency techniques.
a. Logarithmic display: Quickly toggle to logarithmic display, especially if the amplitude of any of the peaks are high - it will help you see sidebands and harmonics that have relatively low amplitude.
b. Time waveform analysis: If you see a high noise floor, harmonics or sidebands, you are advised to look at the time waveform to better understand the nature of the fault.
c. Phase analysis: If you suspect unbalance, misalignment, flexibility, looseness or cocked-bearing, you are advised to use phase for an accurate diagnosis.
d. Gears and bearings: If you suspect bearing or gear damage, study a time waveform (to study what happens as the balls roll around the bearing and to study how the vibration changes as each gear tooth comes into mesh) and utilize enveloping/Shock Pulse/Peak Vue/Spike Energy, etc.
Of course, with more time, and animations and simulations, all of this is much easier to explain, and MUCH easier for you to understand; but I hope it has been helpful. (Please note that there are other possible fault conditions and analysis techniques; this tip is provided as a guide only.)
If you are interested in training, please call 877-550-3400, write email@example.com or visit www.mobiusinstitute.com.
Tip Provided by Mobius