groundup.ai
9/9/2026Long before a bearing gets noisy or trips an overall vibration alarm, its lubricant film has already thinned to the point where metal touches metal. The machine still runs. The route sheet still says the grease is fine. And the data has already started to change. That gap, between the moment the film fails and the moment anyone can see it, is where poor lubrication either becomes a two-minute re-grease or a bearing replacement. This piece is about reading that gap.
Most bearings that fail early do not wear out. They dry out. Lubrication is the single largest cause of premature bearing failure, at 36% by SKF's own damage analysis, and once you count contamination and the wrong grade, lubrication-related issues drive up to 80% of failures. Yet long before a bearing gets noisy or trips an overall vibration alarm, its lubricant film has already thinned to the point where metal touches metal. The machine still runs. The route sheet still says the grease is fine. And the data has already started to change.
That gap, between the moment the film fails and the moment anyone can see it, is where poor lubrication either becomes a two-minute re-grease or a bearing replacement. This piece is about reading that gap.
A lubrication failure is any condition where the lubricant can no longer keep load-bearing surfaces separated by a film. Common forms:
Unlike an imbalance or a misalignment, poor lubrication is not really a fault in its own right. It is the condition that produces the next fault. That distinction shapes everything about how it reads in the data.
Here is the sequence that matters, and it is specific to lubrication.
When the film thins, the first thing that happens is friction. Asperities on the two surfaces begin to graze each other, generating high-frequency energy and a little heat. There is no crack, no spall, no defect frequency yet, just contact where there should be a film.
If nothing changes, that friction starts to remove metal. Micro-welding and abrasion score the surfaces. Now you have damage, and the classic bearing defect tones begin to appear.
So a lubrication problem and a bearing problem are the same story at two different times. The whole economic case for catching it early lives in that timeline. At the friction stage, the fix is lubricant. A few stages later, the fix is hardware, a shutdown, and whatever the failure took out with it.
The friction from a lost film is detectable well before the bearing is audible or trips an overall alarm. The window between the two is where a re-grease still beats a replacement.
Overall velocity RMS averages energy across roughly 10 to 1000 Hz, a band ruled by low-frequency, high-displacement events. Early lubrication friction is the opposite: high-frequency and low-amplitude. It barely registers in that average.
This is why a machine can be running dry and still report normal. The number most routes rely on is simply not measuring in the band where a lost film first speaks.
Three signals shift while the overall reading holds steady, and the order they move in is itself diagnostic.
Friction and enveloped impact energy move first. Demodulation and enveloping lift the small, repetitive metal-to-metal contacts out of the high-frequency carrier, where a raw spectrum would miss them.
Temperature follows. A lost film generates heat, so a bearing or housing running warmer than its own norm supports the picture.
Discrete bearing defect frequencies come last. When outer-race, inner-race, or ball-spin tones finally appear under enveloping, the lubrication problem has already crossed into bearing damage.
Read together, rising friction and temperature with no defect tones yet is the signature of a machine that is still savable with lubricant.
Illustrative development of a poor-lubrication signature. Enveloped impact energy and temperature climb well ahead of overall velocity, which stays inside the normal band. Illustrative normalised data, not customer measurements.
This is the call that decides the repair, and it is unique to lubrication.
If the high-frequency and enveloped energy is rising but the spectrum shows no discrete defect frequencies, you are most likely looking at a film problem: starvation, contamination, or the wrong lubricant. The surfaces are stressed but not yet broken.
Once clear outer-race or inner-race tones appear, the window has closed. The lubrication condition has become mechanical damage, and re-greasing will no longer save the bearing.
That single distinction, friction without defect tones versus friction with them, is what separates a maintenance win from a replacement. It is also why a generic high-vibration alarm is useless here. It fires, if it fires at all, only once the damage is already done.
Vibration tells you the film has failed. It does not always tell you why. This is where lubrication diagnosis has a second evidence stream that other faults do not.
Water ingress, common on outdoor assets, wash-down environments, and anything near process fluid, destroys film strength and often shows as a rising friction floor with a temperature climb but no mechanical defect. Particle contamination from dust or upstream wear does the same. A lubricant sample closes the loop: water content, particle count, viscosity, oxidation, and wear-metal ferrography turn a suspected film problem into a confirmed cause. Pairing the machine data with an oil or grease sample is often the fastest route from anomaly to certainty, and it is specific to this failure mode.
Lubrication demand is not constant. Friction and temperature rise naturally with load and speed, so a reading that is normal under full load can look alarming at idle, and the reverse. A machine watched against a single fixed threshold will either miss a developing film problem or raise false alarms every time the duty changes.
The same is true across assets. Two identical gearboxes do not share a friction and temperature signature once lubricant type, re-greasing history, and load differ. A per-asset baseline, captured against operating state, is what makes a rising friction trend mean something rather than nothing.
Poor lubrication is one of the root causes the Groundup.ai Asset Library™ identifies, alongside bearing outer-race and roller faults, gear failure, coupling failure, seal failure, and looseness.
That is what turns a rising friction band into a decision. The Groundup.ai Asset Library™ holds the confirmed lubrication signatures across fleets and sectors, so the system does not just show elevated energy and leave the interpretation to whoever is on shift. It recognises the pattern, separates a film problem from actual bearing damage, estimates how much runway remains against the P-F interval, and points the team at the fix while it is still lubricant, not steel.
That is the difference between an alert and #CognitiveMaintenance: reading the data, and understanding it.
Yes. Poor lubrication is within the set of root causes Groundup.ai identifies, because the evidence lives in the friction, enveloped, and thermal signals the system already captures.
The strength is in combining them. Rising enveloped impact energy, a climbing temperature, and the absence of discrete defect tones, all measured against a machine's own baseline and operating state, is a far stronger indication of a film problem than any single value crossing a limit. And because the diagnosis arrives at the friction stage, it arrives while a re-grease is still the answer.
The data tells you which machine and how urgently. The lubricant sample tells you why.
What are the early signs of poor lubrication in machine data?
Rising high-frequency friction and enveloped impact energy, usually with a small temperature increase, while overall vibration stays normal and before any bearing defect frequencies appear.
How do I know if it is still a lubrication problem or already bearing damage?
If friction and temperature are up but there are no discrete defect tones, it is most likely still a film problem you can fix with lubricant. Once outer-race or inner-race frequencies appear, it has become bearing damage.
Why does the overall vibration reading stay normal?
Because early lubrication friction is high-frequency and low-amplitude, and the overall value averages a lower-frequency band where that energy barely registers.
Does lubricant analysis still matter if the data already flags a problem?
Yes. The data pinpoints the machine and urgency. An oil or grease sample confirms the cause, water, particles, oxidation, or wear metals, so you fix the source, not just the symptom.
Can Groundup detect poor lubrication?
Yes. It is one of the root causes the Groundup.ai Asset Library™ identifies, using friction, enveloped impact energy, and temperature together against the machine's own baseline. On a rail drivetrain deployment, this surfaced as a repetitive impact pattern on a differential gearbox that pointed to a lubrication condition, later confirmed by contamination and wear indicators. Read the Etihad Rail case study here.
Poor lubrication is the cheapest fault to fix and one of the most expensive to miss, and the difference between the two is time.
Caught at the friction stage, it is a re-grease, a sample, and a note in the log. Missed until the overall alarm finally trips, it is a seized bearing, an unplanned shutdown, and a teardown to find out what a starved film quietly destroyed.
The signal is there early, in the friction band, in the enveloped impacts, in a bearing running a few degrees warm against its own baseline. The only question is whether anything is set up to read it while a re-grease still saves the machine.
That is the job predictive maintenance was supposed to do. It is the job #CognitiveMaintenance actually does.
Bergabunglah dengan Wall of ❤️ kami — bagikan dan tag kami!
Kami menyoroti para pemimpin yang mendorong inovasi dan kinerja, memberi Anda visibilitas sambil membentuk masa depan AI
Artikel Terkait
