What a bearing rating actually promises

A rolling bearing doesn't have a single "load it can take". Roll the same load through it long enough and the races eventually fail by rolling-contact fatigue — tiny cracks under the surface that spall into a pit. The question is never "will it hold?" but "how long will it last?" — and because fatigue is statistical, even that has to be a probability. ISO 281 is the standard that turns a load and a speed into a life, and it's the number stamped implicitly into every bearing catalogue's dynamic load rating C.

L10 and the C/P law

The headline result is the basic rating life, written L10: the number of revolutions (in millions) that 90% of a large batch of identical bearings will reach before the first sign of fatigue. The 10 means 10% are allowed to have failed by then — it is a design target, not a guarantee for any single bearing. It comes from a simple power law:

L₁₀ = (C / P)ᵖ

C is the bearing's dynamic load rating (the load that would give exactly one million revolutions of L10 life — a catalogue number), P is the actual equivalent load the bearing sees, and the exponent p is 3 for ball bearingsand 10/3 for roller bearings. That exponent is why bearing life is so sensitive to load: halving the load on a ball bearing multiplies life by 2³ = 8. To get hours from revolutions you just divide by the speed: L₁₀ₕ = 10⁶·L₁₀ / (60·n), with n in rpm.

Equivalent load: P = X·Fr + Y·Fa

Real bearings rarely see pure radial load. A bearing carrying both a radial force Fr and an axial (thrust) force Fa is rated on an equivalent dynamic load — the pure radial load that would do the same fatigue damage:

P = X·Fr + Y·Fa

The factors X and Y depend on the bearing type and on how much thrust it's carrying relative to its capacity. Below a threshold ratio e, a small amount of thrust is essentially free (X = 1, Y = 0, so P = Fr); above it, the thrust starts to count and X drops while Y rises. A deep-groove ball bearing's e and Y even slide with the ratio f₀·Fa/C₀; an angular-contact bearing's factors depend on its contact angle. Bearing types that can't react thrust at all (plain cylindrical and needle rollers) simply take P = Fr and need a separate locating bearing for the axial load.

The other check: static safety

Fatigue life is about motion. A bearing also has to survive standing still under a peak or shock load without the rolling elements brinelling (denting) the races. That's a separate check against the static load rating C₀: the static safety factor s₀ = C₀ / P₀, where P₀ is the equivalent static load. Typical targets are around 1–2 for smooth running, higher for shock; roller bearings (line contact) are held to a higher target than ball bearings (point contact). A bearing can easily pass its life target and still fail this one under a rare peak.

Speed and lubrication set the real ceiling

L10 assumes the bearing is properly lubricated. In practice the speed is what caps a selection: every catalogue lists a limiting speed, and the speed factor n·dm(speed × mean diameter) has to sit inside the band the chosen lubrication method allows — grease for low-to-moderate, oil for higher, with cooling for the extreme end. Housing temperature limits the grease and the seals, and a shaft running much hotter than the housing eats into the bearing's internal clearance through differential expansion — enough to preload and cook a bearing that passes every load check on paper. These are the limits that usually decide the design, not the fatigue sum.

Worked selection & a design checklist

How the C/P law runs backwards to pick a bearing for a target life, why the plain-bush PV method is a completely different animal, and a checklist for a selection that survives in service — not just on the fatigue sum.

Sizing backwards from a target life

Selection inverts the rating law. Fix a target life in hours, convert it to millions of revolutions at the operating speed, and solve for the required dynamic rating: C_req = P·(L₁₀)^(1/p). Then walk the catalogue for the chosen type from the smallest bore that fits the shaft, and take the first bearing whose C exceeds Creq andwhose C₀ passes the static check. That order matters — sort by capacity alone and you can end up proposing an oversized-bore bearing for a small shaft. An application/shock factor inflates P before sizing, and the ISO 281 reliability factor a₁ adjusts the life if you need better than 90% (a₁ < 1 for higher reliability).

Plain bushes play by different rules

A plain (sleeve) bush has no rolling fatigue, so ISO 281 doesn't apply at all. It's sized by the pressure-velocity (PV) method: the projected bearing pressure P = Fr/(d·L), the rubbing velocity V = π·d·n/60000, and their product PV, each checked against the bush material's limits. Exceed the pressure limit and it extrudes; exceed velocity and it overheats; exceed PV and it wears out. A longer bush (higher L/d) spreads load and drops the pressure — but too long edge-loads under shaft deflection.

Selection checklist

  1. Resolve the real radial and axial loads at the bearing location, and the operating speed.
  2. Pick a bearing type suited to the thrust: deep-groove or angular-contact ball for combined loads, cylindrical/needle roller for heavy radial-only, tapered/spherical roller for heavy combined.
  3. Set a target L₁₀ life (hours) and reliability, apply a shock factor, and size C_req = P·(L₁₀)^(1/p).
  4. Check the static safety factor s₀ = C₀/P₀ against a target for the duty (higher for shock and roller bearings).
  5. Confirm the speed factor n·dm and the catalogue limiting speed suit the lubrication method, and the housing temperature suits the grease and seals.
  6. Check the shaft-to-housing temperature difference won't close the internal clearance; step up the clearance class (C3/C4) if it will.
  7. Confirm the final designation's dimensions, ratings and clearance against the current manufacturer datasheet before production.

← All guides · Open the Bearing Calculator