ISO 281 — Rating life, and why no ratings appear here
The calculation behind every load rating: L10 is a formula, and the number it needs from the catalogue is not one this standard supplies.
Basic rating lifemillions of revolutions, at 90 % reliability
| L10 = (C / P)^p |
| L10h = 10^6 / (60 n) × L10 |
| Lnm = a1 × aISO × L10 |
C is the basic dynamic load rating, P the equivalent dynamic bearing load, n the speed in revolutions per minute, and p the life exponent. L10 is the life reached by 90 % of a population of identical bearings under identical conditions — it is a statistic about a population, not a service life for one bearing. The third line is the modified rating life introduced with the 2007 edition, which brings in a reliability factor a1 and a life modification factor aISO covering lubrication, contamination and the fatigue load limit.
The life exponentpoint contact against line contact
| Bearing type | p |
|---|---|
| Ball bearings | 3 |
| Roller bearings | 10/3 ≈ 3.333 |
The exponent is not a fudge factor. It encodes the contact geometry: a ball touches its raceway at a point and a roller along a line, and the load-life relationship differs between them. Using 3 for a roller bearing underestimates its life by roughly a third.
The standard defines the calculation, not the rating
This is the reason no dynamic load rating appears anywhere in this library. ISO 281 says how C is calculated from a bearing's internal geometry — ball diameter, ball complement, raceway conformity, and how those interact under load. The plan in ISO 15 deliberately does not fix any of those, so two bearings can carry the same designation and different ratings. For a 6205, NSK publishes 15 400 N and SKF 14.8 kN — 14.0 kN against 14.8 kN for the same boundary dimensions. For a 6212 the same two makers give 52.5 kN and 55.3 kN. Neither figure is wrong, and neither is a standard value. The same applies to limiting speed, where the figure moves with lubrication and sealing as well.
Why the reliability factor is not tabulated
The a1 values for reliabilities other than 90 % changed between editions. The 1990 edition ran 1.00, 0.62, 0.53, 0.44, 0.33 and 0.21 for 90 % down to 99 %; the 2007 edition runs 1.00, 0.64, 0.55, 0.47, 0.37 and 0.25 for the same reliabilities. Both sets are published, both are quoted without their edition, and the difference is large enough to matter — 0.21 against 0.25 is a 19 % change in predicted life. Rather than pick a set, this page records that the choice of edition is part of the answer.
What a rating life is not
It is a fatigue life under stated conditions. Wear, corrosion, electrical erosion, inadequate lubrication, and the static brinelling that happens when a bearing is dented while stationary are all outside it, and in most failed bearings one of them is the cause. A calculated life longer than the application survives is not evidence that the calculation was right.