How ISO 281 Defines Bearing Life

Two buyers can look at the same application, the same catalogue and the same load, and reach different answers about how long a bearing will last - unless both of them mean the same thing by "last". ISO 281 is the standard that fixes that meaning. It defines the rating life L10, it supplies the load-life equation that converts a load into that life, and since its 2007 revision it adds modification factors for lubrication and contamination that move the answer a long way from the simple catalogue figure.

This article explains the definition, works the arithmetic for a common 6205, and shows how the numbers change when the load doubles, when reliability is raised above 90 % and when lubrication is poor. It is written for buyers and maintenance engineers who have to compare two ratings and decide which one is honest for their duty.

Why the industry needed one definition of life

A bearing does not fail on a calendar date; it fails after a number of stress cycles under a given load. Metallurgically, fatigue begins at a subsurface inclusion and grows until a spall forms, and because inclusions are distributed randomly, the cycles-to-failure vary from bearing to bearing even within a single production lot. Any useful definition of life therefore has to be statistical, and any useful rating has to be built on a stated probability of survival.

Before a common standard, each maker rated its bearings on its own basis, so two catalogue figures were not comparable. ISO 281 fixed the basis: one probability of 90 %, one equation form and one set of exponents, so a rating can be compared across brands and converted into hours.

The L10 definition and the load-life equation

L10 is the basic rating life: the number of revolutions that 90 % of a large population of apparently identical bearings will complete, or exceed, under identical conditions before the first evidence of fatigue. It is a statement about a population, not a promise for one bearing. One in ten is expected to reach fatigue before L10, and that tenth bearing is by definition a survivor at the short end of the distribution, not a defective one.

The basic rating life comes from a single equation. In words: L10 equals the basic dynamic load rating C divided by the equivalent dynamic load P, raised to the life exponent p.

What the exponent does

The exponent is the most consequential number in the whole calculation and the easiest to overlook. Because a ball bearing's life goes with the cube of C over P, doubling the load cuts the life by a factor of eight, and a 10 % error in the load estimate moves the life by roughly 30 %. Roller bearings use 10/3 instead of 3, which is why they tolerate a load change slightly more gracefully and why the two families are never interchangeable on a load-life basis: the same envelope can carry two different exponents.

Rating life versus application life

The value from (C/P) to the power p is the basic rating life, and it assumes ideal conditions: correct lubrication, a clean environment, a fully developed oil film and a load distributed as the catalogue assumes. Real applications rarely match, and the gap between the basic rating life and the life actually achieved is exactly what the 2007 revision was written to close.

That gap is not academic: two bearings with identical envelopes can share a catalogue rating and still deliver different service lives, because the rating describes the bearing and the achieved life describes the bearing in your machine.

The 2007 revision: reliability, lubrication and contamination

ISO 281:2007 generalised the basic rating life into a modified rating life by multiplying it by two factors: a reliability factor a1 and a life-modification factor aISO that carries the lubrication and contamination conditions. Three inputs drive the result, and all three are properties of the application rather than the bearing:

• Reliability. The required probability of survival above 90 %, expressed through a1.
• Lubrication. The viscosity ratio kappa, the actual operating viscosity against the viscosity the bearing needs for a full film.
• Contamination. The cleanliness of the lubricant, which limits how good the film can be.

The reliability factor a1

a1 converts the 90 % rating life into a life for a higher reliability. It falls quickly, which is the price of certainty: designing for one failure in a hundred rather than one in ten removes three quarters of the basic life.

The viscosity ratio kappa

aISO is driven mainly by the viscosity ratio kappa, the actual operating viscosity divided by the viscosity the bearing needs for full-film lubrication at that speed and temperature. A kappa above about 1 gives a factor above 1; a kappa well below 1, as happens in a slow or hot bearing running on thin oil, pushes aISO below 1 and can erase half the catalogue life. Viscosity grades are standardised by ISO 3448, and the required viscosity follows the speed and the pitch diameter, so kappa is a calculation rather than an assumption. The tolerance class behind the rating also sets the internal geometry that the film has to protect, as the P6, P5 and P4 tolerance guide explains.

Worked example: a 6205 at load and speed

Take a 6205 deep groove ball bearing (see the deep groove ball range) with a basic dynamic rating C of about 14.0 kN, rotating at 3,000 rpm and carrying an equivalent dynamic load P of 1.5 kN. Then C over P is 9.33, and L10 is 9.33 cubed, about 813 million revolutions. To convert revolutions into hours, divide by 60 times the speed in rpm: 813,000,000 / (60 × 3,000) gives about 4,500 hours. The L10 life and load rating guide works the same equation from first principles.

Two things stand out from the table. The double-load case loses seven eighths of the life, which is the cubic penalty in a single row of numbers. And the last row shows what a modest lubrication penalty does: an aISO of 0.5, which is ordinary in a dirty or slow application, halves the life without changing the load or the speed at all.

L10 versus L50, and when static rating governs

L10 is the tenth-percentile life; L50 is the median, and for ball and roller bearings it is conventionally about five times L10. That factor is the origin of the 5× rule of thumb - half of a population outlives five times the L10 figure - and it is a useful sanity check on a replacement interval, provided it is not confused with a design target. When a machine is a wind turbine, a gearbox or anything whose failure stops production, L50 is irrelevant and L10 is too optimistic; you design to L5 or L1 and accept the cost.

When static rating governs

Fatigue life assumes rotation. When a bearing is slow, heavily loaded or subject to shock, it can fail by plastic deformation of the raceway instead of by fatigue, and that regime is governed by the basic static load rating C0 defined in ISO 76, not by C. The usual check is the ratio of the static equivalent load P0 to C0. For very slow rotation, oscillating motion or a stationary load, keep P0 over C0 at or below about 0.5 to limit permanent deformation. For the 6205 above, C0 is about 7.8 kN, so a static load beyond roughly 3.9 kN sits in the region where the static check governs and the life equation no longer describes the risk.

How to read a catalogue page now

A catalogue rating table is a floor, not a promise. The C and C0 columns tell you how one maker's design compares at the reference condition; they say nothing about your lubrication, your contamination or your fit. Read the rating row, then read the page for the assumptions behind it: the reference speed, the limiting speed, the assumed film condition and the tolerance class.

Ratings are a floor, not a promise

If your operating viscosity ratio kappa is well under 1, or your environment is dirty, treat the printed life as an upper bound and apply aISO before you commit. The thermal and speed limits behind the rating are covered in the reference and thermal speed guide, and ISO/TS 16281 extends the calculation to the modified reference life when you need to account for the internal load distribution in a misaligned or heavily loaded bearing.

Step-by-step: calculating life

1. Establish the load case. Convert radial and axial loads into an equivalent dynamic load P, using the load factor for the application, not the nameplate load.
2. Take the real rating. Use C and C0 from the catalogue of the brand you will actually fit, at the bore and series you intend to order.
3. Compute the basic life. Apply L10 = (C/P) to the power p for the bearing type, and convert to hours with 60 times the speed.
4. Apply the reliability factor. Choose a1 for the required reliability, remembering that L5 is 0.64 times L10 and L1 is 0.25 times L10.
5. Estimate the life-modification factor. Work out kappa and the cleanliness, take aISO, and multiply.
6. Check the static case. Confirm P0 over C0 stays at or below about 0.5 where rotation is slow or the load is shock-like.

If you only have the application data and no number, the bearing selector filters the catalogue by envelope, brand and load case so a duty can be turned into a shortlist of candidates in one step.

Common mistakes

• Comparing two brands on C alone. The same envelope from two makers can differ by a few percent in C, and because the ratio is cubed, a small rating difference becomes a noticeable life difference at the duty limit.
• Ignoring the viscosity ratio. Reading a catalogue life without checking kappa treats a full film as guaranteed when it may be thin, which is how a bearing misses its interval in a slow or hot drive.
• Ignoring contamination. Particle count drives aISO as much as viscosity does; a clean grease and a dirty one at the same consistency give different lives.
• Assuming mounting does not matter. Fit, shoulder support and misalignment change the load distribution inside the bearing and therefore the load the equation should use.
• Treating the exponent as a detail. Using 3 for a roller bearing, or ignoring it when the load changes, misplaces the answer by a factor that grows with the load.
• Using L10 as a design target for a critical position. For a machine that must not stop, L10 is the wrong percentile; design to L5 or L1 and accept the reduced life for the higher reliability.

For a cross-check on how the same numbers behave across bearing families, compare the ratings on the FAG and rival listings, or read how the same equation is applied to a roller application in the tapered roller range. The same dimensional envelope that carries an L10 of thousands of hours in one position can be marginal in another, which is why the standard requires the two factors rather than a single number. The interchange caveats in ISO 15 and the interchangeable bearing explain why an envelope match is not a life match.