Bearing Life: ISO 281 and the L10 Rating

The L10 life is the number every catalogue page is built around: the life that 90 % of a population of apparently identical bearings will reach or exceed under a defined load. It is a statistical statement, not a promise, and it only means something when the load and the conditions used in the calculation are the real ones.

This guide works from the basic life equation through the equivalent load, gives a complete example from load to hours, then adds the modified life factors that ISO 281:2007 introduced. It ends with a calculation sequence and the mistakes that make a life figure meaningless.

What L10 means

Bearing fatigue is a statistical process, so life is expressed as a probability rather than a single number. L10, defined by ISO 281, is the life in millions of revolutions that 90 % of a large group of identical bearings will achieve or exceed at a given load and speed. It is not the life of any individual bearing, and ten bearings running side by side under identical conditions will not all fail together.

The companion figures are L50, the median life at which half the population has failed, and L1, the life reached by 99 % of the population. The mean, or L50, is roughly five times the L10 in the standard model, which is why a machine that is rebuilt on L10 intervals rarely sees a bearing at the end of its true life.

That five-to-one rule matters as soon as a maintenance interval is set. If a position is scheduled for replacement at its L10, say 4,000 hours, then roughly half the bearings removed are still well short of their average life and the other half are being changed unnecessarily. L10 works as a conservative planning figure, while the modified life described below is the engineering target.

The life equation

The basic rating life uses two quantities: the basic dynamic load rating C from the catalogue, and the equivalent dynamic load P from the application.

L10 = (C / P) raised to the power p, where L10 is in millions of revolutions. The exponent p depends on the bearing type, because roller contacts spread load differently from ball contacts.

The exponent is what makes life sensitive. If the real load is 20 % higher than assumed, a ball bearing life falls by roughly 42 %, because 1.2 cubed is 1.73. Estimating the load generously is not a safety margin when the rating is a measured catalogue value; it is the largest single source of error in the whole calculation.

Calculating the equivalent load P

Few bearings see a purely radial load. When both radial and axial loads act, they are combined into the equivalent dynamic load P = X × Fr + Y × Fa, where Fr is the radial load, Fa the axial load, and X and Y are factors that depend on the bearing type and on how much axial load there is relative to the radial load.

Ball bearing factors

For a deep groove ball bearing the factors depend on the ratio of axial load to the static rating, which sets a threshold e. Below the threshold the axial load is carried cheaply; above it, the axial load adds fully.

Roller and tapered factors

Tapered roller bearings and angular contact bearings handle axial load by design, so their Y factors are larger and the threshold behaves differently. Always take X and Y from the bearing type and size you are actually using.

Representative X and Y factors. The exact values depend on the bearing type, the contact angle and the ratio of axial load to static rating; take them from the manufacturer's catalogue for the specific part. The combined load guide works through the selection logic.

A worked example from load to hours

Take a 6205 deep groove ball bearing with a basic dynamic rating C = 14 kN. A first check with a pure radial load of 2 kN gives P = 2 kN and L10 = (14 / 2) cubed = 343 million revolutions. At 1,500 rpm that is 343,000,000 / (60 × 1,500) = 3,811 hours.

Now add an axial component. Suppose the real loads are Fr = 2.2 kN and Fa = 0.8 kN, with e = 0.26. The ratio Fa / Fr = 0.36, which is above e, so X = 0.56 and Y = 1.7. Then P = 0.56 × 2,200 + 1.7 × 800 = 1,232 + 1,360 = 2,592 N, or 2.59 kN. The life becomes (14 / 2.59) cubed = 158 million revolutions, which at the same speed is 158,000,000 / 90,000 = 1,756 hours.

The lesson is in the comparison: adding 0.8 kN of axial load cut the calculated life from 3,811 hours to 1,756 hours. A life calculation that treats the bearing as radially loaded would have been out by more than a factor of two.

How sensitive is the answer to the load you assume?

Because L10 carries the load ratio to the third power, an error in the assumed load moves the result much further than an error in any other input. Sweeping the same bearing across a plausible load range shows the shape of that sensitivity.

Ball bearing, exponent p = 3; hours = L10 × 10⁶ / (60 × rpm). A roller bearing with p = 10/3 falls faster still.

Doubling the load ratio from 0.10 to 0.20 divides the life by eight. That is why an honest estimate of the real load is worth more than a precise exponent: the load enters the equation three times over, while the exponent is a fixed property of the bearing type.

Modified rating life: what ISO 281:2007 added

The basic L10 assumes clean lubrication, a correct fit and 90 % reliability. Real machines rarely meet all three, so the standard defines a modified life, Lnm, that applies correction factors to the basic value.

The reliability factor a1

The factor a1 scales the life to a reliability other than 90 %. For higher reliability it falls below one, because demanding that fewer bearings fail means accepting a shorter life: the factor runs from 1.00 at L10 down to 0.25 at L1, the 99 % point. The full table is tabulated in the ISO 281 rating life standard, and for most industrial positions the practical step is simply to decide whether the position needs ordinary 90 % reliability or something stricter.

Lubrication and contamination factors

The 2007 edition replaced the older life adjustment factor with a system factor that reflects the lubrication condition, expressed through the viscosity ratio and the contamination level. A bearing running with a thick, clean oil film can achieve a life multiplier above one; a bearing running with a thin film or in contaminated oil can drop well below one, and no amount of rating headroom compensates for debris in the contact. The lubrication side is covered in the grease lubrication intervals guide.

The practical consequence is that a catalogue rating is a starting point. Two identical bearings in the same machine can have the same L10 and very different actual lives, depending on how well lubricated and how clean they run. The related ISO 281 standard overview covers the evolution of the standard.

Static rating and the P0 / C0 rule

Dynamic rating governs fatigue life, but it does not cover the case of a bearing that is heavily loaded at low speed, or that takes a shock while stationary. For those positions the static rating C0 is the governing figure, defined by ISO 76. The working rule is that the equivalent static load P0 should not exceed half the static rating C0, so P0 / C0 ≤ 0.5. Above that ratio the risk of permanent raceway deformation - brinelling - rises quickly, and the bearing will be noisy even before it fatigues.

Hours to kilometres for wheel and hub duty

For a rotating wheel the life in kilometres is the life in revolutions multiplied by the rolling circumference. A hub with a 600 mm rolling diameter travels about 1.885 m per revolution, so a life of 40 million revolutions is roughly 75,400 km. Always state the rolling radius used, because hub life claims without a radius are not comparable.

How to calculate bearing life

1. Establish the real loads - radial, axial and any shock - at the bearing position, not at the machine input.
2. Compute the equivalent load P = X × Fr + Y × Fa, taking X and Y from the correct bearing type.
3. Select the basic dynamic rating C for the exact bearing from the catalogue, allowing for any difference in internal design between makers.
4. Apply the correct exponent - 3 for ball bearings and 10/3 for roller bearings - and compute L10 = (C / P)^p.
5. Convert to hours using the operating speed: hours = L10 × 10⁶ / (60 × rpm).
6. Apply the reliability factor a1 if the position needs better than 90 % reliability.
7. Adjust for lubrication and contamination using the system factor, and check the static P0 / C0 ratio for slow or shock-loaded positions.
8. Sanity-check the inputs. Confirm the load distribution against the fit and clearance using the clearance guide, and check the speed against the speed limits guide.

Common mistakes

• Using a catalogue P instead of the real equivalent load. The most common cause of an optimistic life figure.
• Ignoring shock and vibration. A nominal steady load misses the peaks that actually drive fatigue and static deformation.
• Forgetting the axial component. As the worked example shows, a modest axial load can halve the life.
• Treating a catalogue rating as a clean-condition promise. Lubrication and contamination move the real life well away from the basic figure.
• Assuming clearance and fit do not change the load distribution. They set how many rolling elements share the load, which changes P itself.
• Mixing L10 with L50. A statement about L10 is a statement about 90 % reliability, not the average life.
• Using a rating from a different maker's part. C and C0 are not fixed by the boundary standard, so the life figure belongs to the specific bearing, not to the size.
• Applying a reliability factor but not a lubrication factor. Demanding 99 % reliability and then ignoring a contaminated oil film gives a precise answer to the wrong question.

The bearing types in this guide are all covered in the catalogue: the deep groove ball bearings for the ball exponent, the tapered roller range and the cylindrical roller range for the roller exponent, with the ratings from a TIMKEN or SKF listing checkable against the equations above. The bearing type selection guide helps you choose the family first, the terms are defined in the glossary, and the selector filters the catalogue by size and rating.