Bearing failure analysis

A dead bearing is a symptom, not a diagnosis

When a rolling-element bearing fails, the temptation is to press in a new one and move on. But the bearing didn't fail for no reason, and whatever killed it is usually still there — the wrong grease, a failed seal, a shaft voltage, a misalignment, a standstill vibration. Fit an identical bearing into the same conditions and it fails the same way, on the same timeline, and you are back under the machine in a month. The value is in the autopsy: every failure mode leaves a distinct signature, and reading that signature tells you what to fix so the replacement actually lasts. This trainer walks six bearings across four checks — running temperature, the vibration defect band, the condition of the lubricant, and the marks on the races and rolling elements — five failures and one healthy unit. Read all four together and each bearing names its own cause.

Heat and the grease: starvation versus contamination

Two of the most common failures both show up as wear and heat, and they are constantly confused — but the fix for one does nothing for the other. Lubrication starvation is a bearing running without an oil film: metal-to-metal contact drives friction and temperature up, the bearing runs hot (well over 200 °F), the grease is dry and carbonized, and the races discolor blue-brown from tempering with smeared, adhesively-worn metal. Contamination is the opposite problem — too much of the wrong thing in the contact. A failed seal lets grit reach the raceways, the grease is present but gritty and dark with hard particles, the races are covered in fine scratches and denting, and it runs only moderately warm. Same complaint at a glance; opposite causes. The lubricant tells them apart: gone and cooked, or there and full of grit.

Race defects: fatigue versus electrical discharge

The second look-alike pair is a genuine race defect — the analyst's software flags "outer-race defect" on both — but the mechanism, and the fix, are worlds apart. Rolling-contact fatigue is honest wear-out: after millions of stress cycles a piece of the race spalls off as a single localized pit, and the bearing rings at its outer-race ball-pass frequency (BPFO) with harmonics. Temperature and grease are normal; this bearing simply reached the end of its life. Electrical discharge looks similar on a spectrum but comes from a completely different place: on a variable-frequency-drive motor, shaft voltage builds until it arcs through the bearing to ground, and each spark pits the race until the pits organize into a regular washboard fluting pattern. The giveaways are a measurable shaft voltage (tens of volts peak) on the drive-end shaft, that repeating fluting pattern rather than one clean pit, and grease darkened and burnt by the arcing. A fatigue spall you replace and reset the clock; electrical fluting you replace and break the current path, or the next one flutes in months.

Brinelling — dents without a running defect

The last failure mode fools people because it makes noise but shows none of the heat or lubrication signs of a running defect. Brinelling is denting of the raceway at the rolling-element spacing, pressed in while the bearing is not turning. True brinelling is a single overload event — a shock load, or driving a bearing on with a hammer through the balls instead of pressing on the correct ring. False brinelling is the slow version: vibration transmitted through a bearing that is standing still (a spare on a shelf beside a running press, an idle motor coupled to a running one, a bearing rattled in transport) wears evenly-spaced marks at ball pitch. Either way the tell is the same — dents at exactly the ball spacing, with no rotational smearing and normal temperature and grease. The bearing is noisy from the very first startup because the damage was there before it ever ran.

The discipline — read all four

No single reading names the fault; the pattern across four does. Temperature separates the hot failures (starvation) from the ones that run cool. The vibration band separates a general problem (broadband, no clear tone) from a localized one (a clean BPFO tone) from an electrical one (high-frequency fluting). The lubricant separates dry from gritty from burnt. And the race evidence — blued and smeared, abraded, a single spall, washboard fluting, or dents at ball spacing — usually settles it, with the shaft-voltage probe as the tiebreaker between a fatigue spall and electrical damage. Take all four readings, lay them side by side, and the bearing tells you not just that it failed, but why — which is the only thing that keeps the replacement from dying the same death.