The machinery failure incident reports that cross a maintenance supervisor's desk tend to have a predictable structure: machine ran fine, machine produced a noise or tripped a fault, machine is down. The post-mortem, when it happens, usually finds the bearing had been degrading for days or weeks. The signal was there. Nobody was reading it.
Bearing failures in discrete manufacturing equipment — stamping presses, CNC machines, conveyor drive units, gearboxes — follow one of four primary degradation mechanisms. Each mechanism leaves a characteristic vibration signature at a different stage of the failure progression. Understanding the mechanism matters because it determines how much warning time you have, what monitoring technique is most sensitive, and where in the lubrication or maintenance regime the root cause likely sits.
Surface Fatigue: The Classic Rolling Contact Failure
Surface fatigue, specifically spalling (pitting of the raceway or rolling element surface), is what most maintenance engineers mean when they say a bearing "failed normally." It's caused by cyclic Hertzian contact stress that eventually nucleates a subsurface crack, which propagates to the surface and chips out a spall. This is the failure mode that bearing defect frequencies — BPFO, BPFI, BSF — are designed to detect.
The four-stage ISO progression model for surface fatigue is well established in condition monitoring practice:
- Stage 1: Ultrasonic emission increases (40–50 kHz range). No detectable change in conventional vibration spectrum. This stage may last weeks to months.
- Stage 2: Defect frequencies appear in the envelope spectrum. Crest factor begins to rise. Still no meaningful change in broadband vibration RMS. Alert lead time here is typically 2–6 weeks on slow-to-medium speed bearings.
- Stage 3: Defect frequency harmonics appear and grow. Sidebands emerge. Broadband RMS begins to rise. Crest factor may peak then stabilize as the defect grows. Alert lead time: days to 2 weeks.
- Stage 4: Rapid progression. Broadband RMS elevated. Machine performance degrading. Failure is imminent — hours to days.
The practical implication: if your condition monitoring only triggers on broadband RMS thresholds, you're catching Stage 3 or Stage 4. Envelope analysis with BPFO/BPFI frequency tracking gives you Stage 2 — which is where you still have time to plan maintenance without production disruption.
Abrasive Wear: When the Lubrication Film Breaks Down
Abrasive wear occurs when hard particle contaminants enter the bearing and score the raceway or rolling element surfaces. In discrete manufacturing environments, the contamination sources are ubiquitous: metal swarf from machining, silica dust from grinding or abrasive processes, coolant that carries particles into housing seals.
The vibration signature from abrasive wear is different from surface fatigue. Rather than discrete impulses at bearing defect frequencies, abrasive wear generates broadband elevated noise across the frequency spectrum — a "grass" rise in the FFT baseline. Early-stage abrasive damage raises the noise floor without producing strong periodic components, which makes it harder to detect with standard bearing defect frequency analysis.
The useful indicators for abrasive wear are broadband spectral kurtosis (statistical sharpness of the vibration distribution), the overall noise floor trend in the 2–10 kHz range, and — critically — oil particle count analysis in oil-lubricated gearboxes. On grease-packed sealed bearings where oil sampling isn't practical, the vibration noise floor and temperature rise are the primary early indicators.
A Tier-2 stamping shop in southwestern Michigan running a press line with helical gear reducers found their reducer bearing replacement rate was three times the theoretical design life. Oil analysis revealed ISO 4406 cleanliness codes consistently above 20/18/15 — far above the 17/15/12 or better typically specified for these bearing types. The contamination was entering through a worn input shaft seal. Replacing the seals and instituting a 6-monthly oil change with filtration improved bearing life substantially, and the elevated broadband vibration signature in the reducers dropped markedly within one operating cycle.
Adhesive Wear: Smearing Under Insufficient Load Separation
Adhesive wear — smearing — occurs when rolling elements slide rather than roll against the raceway, causing localized welding and tearing of surface material at the asperity level. This happens when the elastohydrodynamic (EHL) lubrication film is insufficient to fully separate the contact surfaces, typically under one of three conditions: overloading, excessive speed for the viscosity grade in use, or — paradoxically — very low loads that allow ball-skidding in angular contact or deep groove bearings.
Smearing damage tends to appear as irregular patches on the raceway, not the clean pitting of surface fatigue. The vibration signature is irregular and difficult to distinguish from noise at early stages. As damage progresses, broadband RMS rises and you may see sub-harmonic content below 1× shaft frequency from the irregular rolling contact.
The key diagnostic distinction: smearing damage doesn't produce clean BPFO/BPFI signatures, at least not initially. A bearing that looks fine in an envelope spectrum analysis but shows elevated temperature, rising noise floor, and irregular sub-harmonic content may be experiencing adhesive wear rather than surface fatigue. This is a case where monitoring vibration temperature correlation — looking for housing temperature anomalies relative to ambient and load conditions — adds meaningful diagnostic coverage beyond vibration analysis alone.
We want to be clear that this article isn't telling you smearing will always be caught by current-generation monitoring software. Irregular adhesive damage is genuinely harder to detect in its early stages than outer race spalling. The honest position is that condition monitoring provides better coverage for surface fatigue than for smearing — knowing the limitation matters for setting realistic expectations about what a monitoring system can and cannot catch.
Fretting Corrosion: The Slow Creep at the Mounting Interface
Fretting corrosion develops at the interface between a bearing's outer ring and its housing bore, or between the inner ring and shaft, when there is micro-motion between the mating surfaces under cyclic load. The micro-motion — often invisible to inspection — disrupts the protective oxide film, exposing fresh metal that re-oxidizes, producing iron oxide debris (the characteristic reddish-brown "rust dust") and eventual surface pitting.
On CNC machine tools running variable cutting loads, gearboxes with cyclic torque reversals, and stamping press crankshafts with high-frequency impulse loads, fretting at the bearing seat is a legitimate failure mode that accounts for a significant proportion of "premature" bearing replacements — particularly in machines that have had bearings replaced or are subject to vibration during non-operational periods.
The vibration signature from fretting is subtle: slight looseness develops at the bearing seat, which may manifest as non-synchronous sub-harmonic components and a slight increase in the 1× and 2× shaft vibration as the clearance causes intermittent contact behavior. The most reliable early indicator is an inspection finding of reddish discoloration at the housing bore or shaft journal during a planned maintenance window — not a vibration alert.
Fretting corrosion is best addressed through prevention rather than condition monitoring: correct interference fits per bearing manufacturer specification, applying fretting-resistant compounds to housing bores in high-vibration environments, and ensuring machines are not stored or transported with loose pre-installed bearings. Once fretting is advanced, the bearing seat may need metalising or sleeving.
Mapping Failure Modes to Your Monitoring Strategy
The four failure modes call for different monitoring approaches, and not all of them are well-served by the same technique:
- Surface fatigue: Envelope analysis with BPFO/BPFI/BSF frequency tracking. Most detectable failure mode. Weeks of lead time achievable in Stage 2.
- Abrasive wear: Broadband spectral noise floor trend, kurtosis, temperature. Oil analysis where applicable. Best combated by contamination control at the source.
- Adhesive wear: Temperature-vibration correlation, sub-harmonic content, noise floor. Harder to detect early. Review lubrication regime and load conditions.
- Fretting corrosion: Sub-harmonic and 1× amplitude trends, but primarily a maintenance and installation quality issue. Condition monitoring provides supplementary rather than primary detection.
Discrete manufacturers who understand this mapping can set more accurate expectations about what a condition monitoring deployment will and won't flag. Surface fatigue on outer and inner races — the statistically dominant failure mode in well-maintained equipment — is highly detectable with envelope analysis. The other three modes require a combination of vibration monitoring, temperature monitoring, oil analysis, and fundamental maintenance practice improvement to address comprehensively.
If your current condition monitoring is only watching broadband RMS and comparing it to a static threshold, you're detecting failure modes that are already in Stage 3 or beyond. That's useful — it's still better than no monitoring — but it's not what modern vibration analysis is capable of. The equipment is telling you more than you're currently listening to.