Bearing Vibration Analysis: From Vague Symptoms to Diagnosis

Bearing Vibration Analysis: From Vague Symptoms to Diagnosis

In daily industrial maintenance, teams often face a frustrating situation: a machine sounds wrong or feels rough to the touch, but there is no obvious visual defect. That subjective sense that something is off is frequently an early warning of bearing or rotor damage. Relying on experience alone is slow and prone to misjudgment. Modern vibration diagnosis changes that by turning vague feel into objective, quantifiable evidence. Through layered measurement and analysis, it can expose faults inside the machine before they become failures.

Why Vibration Analysis Belongs in Bearing Maintenance

Vibration is the language of rotating equipment. Unbalance, misalignment, looseness, gear problems and bearing defects all leave distinct signatures in the vibration signal. The challenge is separating the relevant pattern from normal machine noise. A structured program moves from a quick check of overall condition to detailed frequency analysis, then to trend monitoring. This is especially important for precision ball bearings and precision roller bearings, where small defects can grow quickly under load.

First Line of Defense: Simple Screening

When a plant has hundreds of machines, deep analysis of every unit is not practical. Simple diagnosis based on amplitude parameters acts as a triage desk. It quickly separates machines that need attention from those running normally.

• Overall health thermometer: RMS vibration velocity is the core indicator of overall vibration severity. It reflects vibration energy and is comparable to body temperature. Once the value crosses a threshold defined by ISO 2372 or a similar plant standard, the machine is no longer in a healthy condition and deserves closer inspection.
• Radar for hidden defects: Early pitting and spalling may not move the RMS value much. Peak value and kurtosis are far more sensitive to impact signals. They can catch short, sharp pulses that indicate a developing defect, like detecting a hidden current beneath a calm surface. A rising kurtosis trend is a strong prompt that something has started to break down.

Precision Diagnosis: Finding the Fault Location

Once simple screening raises an alarm, precision diagnosis is needed to locate the problem. Frequency-domain analysis is the main tool. It breaks a complex vibration signal into individual frequency components so engineers can see which mechanical source dominates.

Low-Frequency Spectrum Analysis: Rotor and Bearing Faults

In the low-frequency range, typically below 1000 Hz, spectrum analysis is effective for rotor-related problems. Mass unbalance produces a high peak at 1x shaft speed. Misalignment usually shows elevated 2x and 3x shaft-speed components. Bearing geometry provides another layer of evidence. By calculating the characteristic frequencies of the inner race, outer race and rolling elements, analysts can match peaks in the spectrum to specific bearing components. A clear peak at the outer-race fault frequency is like a fingerprint match: it points directly to the damaged part. Consistent bearing geometry makes these signatures more predictable and easier to trend over time.

High-Frequency Resonance Demodulation: A Microscope for Early Faults

For fine early spalling, the impact energy is often buried under large background noise, and a standard spectrum may not reveal it. High-frequency resonance demodulation, also called envelope analysis, is designed for this case. It uses the natural frequencies of bearing components as a carrier. Through filtering and demodulation, weak impact signals are separated from the noise and amplified. The result is like giving engineers night vision: even in a noisy machine, they can see the early damage pattern inside the bearing.

Trend Analysis and Remaining Life Prediction

Diagnosing a fault is not the end point. Predicting how it will develop is the goal of intelligent maintenance.

• Absolute and relative checks: An absolute limit such as an ISO threshold tells you whether the machine is bad now. Relative comparison with the machine’s own history tells you whether it is getting worse. A vibration trend chart shows the full path from first symptom to severe damage.
• From repair to prevention: Combined with historical trends and machine-learning algorithms, modern diagnostic systems can estimate remaining bearing life. Maintenance then shifts from reactive repair to predictive action. The plant can order parts and schedule work before a fault causes unplanned downtime.

Case Study: Water Pump Motor Bearing Noise

A water pump motor developed an abnormal bearing noise. Operators first noticed it by ear. A vibration analyzer then showed that RMS vibration velocity was above the limit and kurtosis was clearly elevated, confirming an impact-type fault. Spectrum analysis revealed a distinct peak at the bearing outer-race characteristic frequency, with modulation from shaft rotation. Envelope analysis extracted clear impact pulses. The diagnosis was early outer-race spalling. On teardown, the root cause was improper axial clearance adjustment during installation, which had subjected the bearing to extra axial load. The case shows the full value of vibration diagnosis: vague noise became hard evidence of axial clearance error leading to outer-race spalling. Sensing, analysis and maintenance decision formed a closed loop.

Practical Takeaways

Vibration analysis is not a single measurement; it is a layered process. Start with RMS and kurtosis screening. Move to spectrum analysis when the alarm appears. Use envelope analysis for early bearing impacts. Track results over time and compare them with the machine’s own baseline. Choose bearings with controlled geometry and correct installation clearances so that the diagnostic signatures remain readable and the root causes are not built in from the start. For help matching bearing type, clearance and precision grade to the application, review FIJ precision bearing solutions.

← Back to News