Overview
Bearing fault diagnosis commonly uses five methods: audible inspection, vibration analysis, temperature monitoring, lubricant analysis, and acoustic emission testing. Each method offers different sensitivity and diagnostic information. The following sections summarize practical techniques and typical fault signatures for each method.
01 Abnormal rotating-sound analysis
Audible inspection uses a stethoscope-like technique to monitor bearing condition. Common tools include a long-handled screwdriver used as a listening rod or a rigid plastic tube about 20 mm in diameter. Electronic stethoscopes improve monitoring reliability. A normal bearing runs smoothly and evenly with a continuous, relatively uniform whooshing or low roaring sound. Typical abnormal sounds and their causes are explained below.
- Uniform, continuous hissing: Caused by rolling elements rotating in the inner and outer raceways, producing irregular metallic vibration noise independent of speed. This usually indicates insufficient grease; replenish lubricant. If the equipment has been idle for a long time, especially in cold conditions, a hissing or rustling sound may appear due to reduced radial clearance or a harder grease consistency. Adjust radial clearance appropriately and replace with grease of a slightly higher penetration/softness.
- Periodic hollow or rumbling within a continuous whoosh: Caused by dents, grooves, or corrosion pits on the rolling elements or raceways. The sound period is proportional to shaft speed. Replace the bearing.
- Irregular, uneven scraping: Caused by foreign particles such as metal chips or sand entering the bearing. The sound intensity is low and not related to speed. Clean the bearing and re-grease or change the oil.
- Continuous irregular rasping: Often related to inadequate fit, e.g., inner ring loose on the shaft or outer ring loose in the housing. If the sound is severe, inspect and repair the fitting between bearing and mating parts.
02 Vibration signal analysis
Bearing vibration is sensitive to defects such as spalling, dents, corrosion, cracks, and wear. Frequency analyzers and dedicated vibration meters can measure vibration amplitude and frequency content; the frequency distribution helps identify specific fault types. Measured values depend on operating conditions and sensor location, so establish baseline measurements for each machine before setting diagnostic thresholds.
Common bearing condition-monitoring techniques include vibration analysis, lubricant analysis, temperature measurement, and acoustic emission. Vibration-based diagnosis is the most widely used and is divided into simple diagnostics and precision diagnostics.
2.1 Simple diagnostic methods
Simple diagnostics compare measured vibration values (peak, mean, rms, etc.) to predefined thresholds. If measured values exceed limits, proceed to precision diagnostics. Three general types of judgment standards are used:
- Absolute standard: Fixed limit values used to judge whether measured vibration exceeds acceptable bounds.
- Relative standard: Periodic measurements at the same location on the same component, compared over time; diagnosis is based on the ratio of current vibration to baseline vibration under no-fault conditions.
- Comparative standard: Measurements of several bearings of the same model under similar conditions at the same location, compared against each other.
Absolute standards require specified measurement methods and applicable frequency ranges. No single absolute standard fits all bearings, so combine absolute, relative, and comparative standards for reliable diagnosis.
Common simple diagnostic techniques include:
- Amplitude-value diagnosis: Amplitude metrics include peak value XP, mean X (for harmonic vibration the average over half a period; for impact-type vibration the average of absolute values), and root-mean-square Xrms. This is the simplest and most common method, comparing measured amplitudes with thresholds.
- Peak value is suitable for faults with instantaneous impacts, such as surface pitting.
- Mean value gives similar diagnostic utility to peak but is more stable; generally applied at higher speeds (e.g., above ~300 rpm).
- RMS value is time-averaged, suitable for faults like gradual wear where amplitude changes slowly over time.
- Probability-density diagnosis: Amplitude probability-density for a healthy rolling bearing is typically Gaussian. Faults cause skewness or dispersion in the density curve.
- Kurtosis diagnosis: A healthy bearing with Gaussian amplitude distribution has a kurtosis near 3. Kurtosis rises with developing impact-type faults. This method is largely independent of speed, size, and load, and is suitable for detecting pitting-type damage.
- Waveform factor diagnosis: Defined as peak divided by mean (XP/X). It is an effective indicator for simple diagnostics.
- Crest factor diagnosis: Defined as peak divided by RMS (XP/Xrms). Crest factor is insensitive to bearing size, speed, load, and to changes in sensor or amplifier sensitivity. It is useful for detecting pitting-type faults. Monitoring XP/Xrms over time enables early warning and tracks fault progression:
- Healthy bearings show a small, stable XP/Xrms.
- When damage begins, impact signals increase peak amplitude while RMS remains relatively unchanged, so XP/Xrms increases.
- As the fault grows, RMS begins to rise and XP/Xrms decreases toward the healthy level.
2.2 Precision diagnostic methods
Bearing vibration contains rich frequency components, both low and high. Specific faults correspond to specific frequency components. Precision diagnostics use signal processing to separate those components and identify faults. Common precision techniques include:
- Low-frequency analysis: Low-frequency vibration is considered below 8 kHz. Vibration velocity is typically analyzed for low frequencies. Acceleration sensors are commonly used, but acceleration signals must be converted to velocity using a charge amplifier and integrator, then low-pass filtered with an upper cutoff of 8 kHz to remove high-frequency content before frequency analysis.
- Mid- and high-frequency demodulation analysis: Mid-frequency ranges from 8–20 kHz and high frequency from 20–80 kHz. Acceleration can be analyzed directly in these bands. After the sensor and charge amplifier, a high-pass filter removes low-frequency components, then the signal is demodulated and subjected to frequency analysis to extract characteristic frequencies.
03 Temperature analysis
Bearing temperature is commonly inferred from the bearing housing surface; measuring the outer-ring temperature via an oil hole gives a more direct reading. Temperature typically rises slowly after startup, reaching a steady state in 1–2 hours. Normal temperature depends on machine heat capacity, cooling, speed, and load. Improper lubrication or installation can cause rapid temperature rise and abnormal high temperatures, requiring immediate intervention.
High temperature damages bearing lubricant and shortens bearing life. Continuous operation above about 125°C for extended periods reduces bearing lifespan. Causes of high bearing temperature include insufficient or excessive lubrication, contaminated lubricant, excessive load, bearing damage, insufficient clearance, and high friction from seals.
Continuous temperature monitoring of bearings or other critical components is necessary. Any temperature change under stable operating conditions may indicate a fault. Periodic temperature measurement can be done with digital thermometers that display °C or °F. Critical bearings—those whose failure would cause equipment downtime—should be fitted with temperature sensors for continuous monitoring. Note that bearings will show a natural temperature rise immediately after relubrication that may persist for a day or two.
04 Lubricant analysis
Lubricant analysis often uses ferrographic techniques to identify and predict rolling fatigue. An oil sample is passed through a high-gradient magnetic field so solid particles deposit onto a glass slide in size-dependent distribution. Observing particle shape, size, color, and composition helps determine wear types and predict machine condition.
Ferrographic methods target ferrous particles but also identify nonferrous metals, sand, organic debris, and seal fragments. Specific indicators:
- Spherical steel particles 1–5 μm in diameter indicate the onset of fatigue microcracks.
- Fatigue spalling particles with length-to-thickness ratio about 10:1 and length greater than 10 μm indicate abnormal fatigue wear; lengths greater than 100 μm indicate bearing failure.
- Thin fatigue flakes with length-to-thickness ratio about 30:1 and lengths of 20–50 μm, often with voids, increase in number when fatigue begins and can be used together with spherical particles as fatigue markers.
05 Acoustic emission testing
Acoustic emission (AE) refers to elastic waves released when a material deforms or cracks propagate. AE testing detects and analyzes these signals to infer source events and internal conditions. AE signals are categorized as burst-type and continuous-type. Burst-type signals are discrete pulses distinguishable from background noise. Continuous-type signals consist of many small pulses so dense they are not individually resolvable; they are effectively many small bursts occurring in rapid succession.
In failing bearings, both burst and continuous AE signals can occur. Relative motion and rubbing among bearing components (inner ring, outer ring, rolling elements, cage) produce Hertzian contact stresses; surface cracks, wear, dents, scoring, seizure, poor lubrication leading to surface roughening, contaminant particles creating hard edges, and electrical current pitting can all generate burst-type AE signals.
Continuous AE signals typically stem from global faults such as lubrication failure (loss of oil film, ingress of contaminants into grease), high temperatures, or frequent localized faults that produce a high rate of burst events in a short time. Bearings emit rich AE information during contact impacts, so AE monitoring is effective for bearing condition monitoring and diagnosis.
