A vibration spectrum plot looks like noise until you know which frequency band to look at, and by the time overall vibration amplitude alone trips an alarm, the underlying fault, whether it's a worn bearing, a bowed rotor, or a rubbing seal, has usually been developing for weeks. Reading the frequency domain correctly turns a vague high-vibration alert into a specific, actionable finding pointed at the right component. See how iFactory automates this diagnosis across your fleet with a Book a Demo.
Overall Vibration Amplitude Tells You Something Is Wrong. Frequency Analysis Tells You What.
Bearing wear, rotor imbalance, misalignment, and seal rub each produce a distinct frequency signature, but most control room alarms only trend overall amplitude. iFactory automatically decomposes vibration signals into their component frequencies and matches the pattern to the most likely fault, cutting diagnosis time from days of manual spectrum review to minutes.
Reading A Vibration Spectrum Like A Diagnostic Fingerprint
Every rotating component generates vibration energy at frequencies tied to its rotational speed, the number of blades or gear teeth it has, and the specific way it's failing. Plotting amplitude against frequency, rather than watching a single overall number, is what turns raw vibration data into a diagnosis.
Illustrative spectrum showing relative amplitude across common diagnostic frequency bands used in gas turbine rotating equipment analysis
Which Frequency Band Points To Which Fault
Matching an observed frequency peak to its most likely mechanical cause is the core skill of vibration diagnostics, and it is exactly the kind of pattern matching that benefits from being automated across hundreds of measurement points rather than reviewed manually one spectrum at a time.
| Frequency Signature | Likely Fault | Typical Symptom |
|---|---|---|
| 1X Running Speed | Rotor Imbalance | Smooth radial vibration, rises with speed |
| 2X Running Speed | Shaft Misalignment | High axial vibration alongside radial |
| Ball Pass Frequency Outer (BPFO) | Outer Race Bearing Defect | Sharp discrete peak, often with harmonics |
| Ball Pass Frequency Inner (BPFI) | Inner Race Bearing Defect | Sideband peaks around running speed |
| Sub-Synchronous (0.4-0.48X) | Oil Whirl Or Seal Rub | Unstable, intermittent amplitude spikes |
| Gear Mesh Frequency (GMF) | Gear Tooth Wear | High-frequency peak with sidebands at gear speed |
A single frequency peak rarely confirms a fault on its own; sideband patterns and harmonic content around the primary peak are usually needed to distinguish between similar-looking signatures.
Stop Reading Spectrums By Hand
iFactory automatically flags the fault signature hiding inside your vibration data and routes it to the right maintenance action.
Vibration Severity Zones And What They Actually Mean
ISO 10816 and equivalent standards define severity zones based on overall vibration velocity, but the zone alone doesn't tell you how urgently to act. Combining severity zone with frequency content and trend direction is what separates a genuine escalation from a stable, tolerable condition.
A Repeatable Process For Turning Vibration Data Into A Work Order
A consistent diagnostic workflow keeps vibration analysis from depending entirely on one experienced analyst's judgment, and it gives less experienced technicians a structured path to the same conclusion.
Where Vibration Monitoring Programs Miss Early Warnings
Alarming On Overall Amplitude Only
Setting alarms purely on overall vibration velocity misses early-stage bearing defects that show up as a discrete high-frequency peak long before overall amplitude rises.
Inconsistent Sensor Mounting
Comparing readings from sensors mounted at slightly different locations or orientations between inspections introduces enough variation to obscure a genuine developing trend.
Ignoring Phase Data
Skipping phase measurement makes it far harder to distinguish between imbalance and misalignment, since both can produce similar amplitude at 1X running speed.
Reacting To A Single Elevated Reading
Acting on one high reading without checking for a sustained trend can lead to unnecessary teardown for what was actually transient process-driven vibration.
Where Sensor Placement Changes What You Can Actually Diagnose
The quality of a vibration diagnosis is limited by where the sensors are mounted and what type of measurement they take, not just by how sophisticated the analysis software is downstream. A proximity probe measuring shaft relative displacement answers different questions than an accelerometer measuring casing vibration, and mixing the two up during interpretation leads to false conclusions even when the underlying data is perfectly accurate.
Radial proximity probes mounted near each bearing, typically in an X-Y configuration at ninety degrees apart, are what make orbit and phase analysis possible, which in turn is what separates a confirmed imbalance diagnosis from a misalignment diagnosis that happens to produce a similar-looking overall amplitude. Axial position probes catch thrust bearing wear and rotor position drift that radial sensors alone would miss entirely. Casing-mounted accelerometers, meanwhile, are generally better suited to picking up the higher-frequency content associated with bearing defects and gear mesh issues, since these signals attenuate quickly through the oil film that a shaft-relative probe measures across.
Radial Proximity Probes
Measure shaft displacement relative to the bearing housing, essential for orbit analysis, phase measurement, and distinguishing imbalance from misalignment.
Axial Position Probes
Track thrust bearing wear and rotor axial position, catching a failure mode that radial vibration monitoring alone would not detect.
Casing Accelerometers
Best suited for high-frequency bearing defect and gear mesh detection, capturing signals that attenuate too quickly to register on shaft-relative probes.
Frequently Asked Questions
Q: How early can a bearing defect be detected through vibration analysis?
Bearing defects typically show up first as a discrete peak at the ball pass frequency, either outer or inner race depending on defect location, well before overall vibration amplitude rises enough to trip a standard alarm threshold. This early-stage signal can appear weeks to months ahead of a bearing failure that would otherwise be caught only through temperature rise or overall vibration trending. Reach out through Support Contact to see how frequency-based bearing detection compares to your current alarm strategy.
Q: What is the difference between rotor imbalance and shaft misalignment in a vibration spectrum?
Imbalance typically shows a dominant peak at 1X running speed with relatively low axial vibration, while misalignment usually produces significant vibration at both 1X and 2X running speed along with elevated axial readings that imbalance alone would not produce. Phase measurement between bearings on either side of a coupling adds further confirmation, since imbalance and misalignment produce distinct phase relationships that a frequency spectrum alone cannot fully separate.
Q: Can seal rub be detected before it damages the seal or shaft?
Seal rub often produces an unstable, intermittent sub-synchronous vibration signature that can be caught in its early stages if the monitoring system is set up to flag irregular, non-repeating patterns rather than just a fixed frequency threshold. Because rub events can be sporadic depending on thermal growth and clearance conditions, continuous monitoring generally catches this fault earlier than periodic route-based vibration readings taken only once or twice a month.
Q: How often should vibration spectrums be reviewed for a gas turbine in continuous service?
Continuous online monitoring systems typically capture and screen spectrums automatically on a near-real-time basis, with a detailed manual review triggered whenever a trend or a new frequency peak crosses a defined threshold. For units without permanent online monitoring, a route-based reading every one to two weeks is a reasonable baseline, though this interval should tighten to daily or continuous monitoring once any fault signature has been identified and is being tracked toward a planned repair. A Book a Demo session can show how automated screening fits your current monitoring interval.
Q: Why do two machines with the same overall vibration reading sometimes have very different actual conditions?
Overall vibration velocity is a single summed number across the entire measured frequency range, so two very different frequency distributions, one broad and low-level, the other a single sharp defect peak, can sum to the same overall value while representing very different levels of actual mechanical risk. This is the core reason frequency domain analysis exists as a discipline separate from simple overall vibration trending, and why relying on overall amplitude alone for alarm decisions creates blind spots for specific, well-defined fault types.
Turn Every Spectrum Into A Clear Diagnosis
iFactory screens vibration data continuously so bearing, rotor, and seal faults surface as specific findings, not vague alarms.
