How to Size a Bearing from Load Rating and L10 Life

Don't size from the peak load

The number that goes into the life formula is the equivalent dynamic load P, not the worst instant in the duty cycle.

A press or conveyor that spends most hours at 2 kN and only seconds at 5 kN has a P closer to the 2 kN band, because the L10 formula consumes the load the spectrum averages to^1–3,10,11.

Sizing on the 5 kN peak overstates the required C and pushes you to a larger, slower, more expensive bearing than the machine needs. Sources name this as a common selection error: sizing off peak load instead of equivalent load^1,9,19.

The mechanism is in the exponent. Life scales with (C/P) raised to p, so a P that is too high shrinks the calculated life and forces a bigger C to compensate^2,10,11.

That is why the load spectrum has to be worked before any catalog page is opened. The sources do not give a method for combining a multi-level load and speed spectrum into a single P, so ask the supplier to show how they reduced your spectrum.

If your duty cycle is mostly 2 kN with brief 5 kN peaks, do not hand the supplier the 5 kN figure as the design load.

Pick the family before the size

Family sets the exponent p, and because C/P is raised to p, a roller bearing's 10/3 turns the same rating and load into longer life than a ball bearing's 3^2,10,11.

At 5 kN and 1800 RPM, one source calculates about 34,000 h for a ball 6205 and about 71,000 h for a roller NU205 — same load, same speed, different exponent². The ball bearing keeps the speed and quiet-running envelope the roller cannot match, so the trade is real^2,3.

Decide the family on load and speed first, then compare sizes within it.

Screen out sizes on speed and envelope

A size that cannot survive the operating speed is eliminated regardless of its C, so the speed and envelope check runs alongside the L10 calculation as a filter, not after it as a confirmation^1,3,5. If you skip it, you can spend hours on life arithmetic for a candidate that will never be quoted.

The speed check also settles the family question early: roller bearings are generally not first choice for high speed, while ball bearings are preferred there^2,3. Work these checks in order before you touch the life formula.

• ✓Confirm the candidate's limiting speed and dN rating cover your operating rpm
• ✓Check the cage design supports that speed
• ✓Confirm the boundary envelope fits your shaft, housing and mounting
• ✓Decide which side is fixed and which is free
• ✓Confirm misalignment tolerance, lubrication and sealing against the environment

Back-calculate the C you must ask for

Rearranging L10h = 10^6/(60n) × (C/P)^p for C turns your target life into a single number you can hold against a catalog page¹⁰. The size decision becomes a threshold test: find the smallest catalog bearing whose published C is at or above the required value¹⁰.

For a heavy industrial gearbox, one source gives a typical target of 50,000–100,000 h¹⁴. With that target, your P and your speed, you solve for C, then look for a catalog bearing meeting it.

The same source describes obtaining the required C from L10h, P, n and p, then selecting from the specification tables¹⁰. Because C sits in the numerator raised to p, life scales with the p-th power of C/P, so a small shortfall in C costs more life than it looks like^2,10,11.

One source notes a bearing should generally not be subject to a maximum operating load above half its dynamic capacity¹¹. Write the required C into the RFQ as a minimum, not a preference.

Correct for reliability above 90%

L10 is a 90%-reliability statistical rating, not a promised service interval for one unit in one plant^8,11,17. If your duty must hold at higher reliability, the life you get from a given C shrinks.

One source gives the a1 factor as 0.62 at 95% reliability, falling to 0.21 at 99%¹². That means a bearing that just met the target at 90% now needs a substantially larger C to hold the same target.

The corrected life is the L10 multiplied by a1, which is below 1, so the same C yields less life¹². At exactly 90% reliability, a1 = 1 and the plain formula applies without correction¹².

The sources do not give a1 values beyond 99% reliability, so if your requirement is tighter than that, ask the supplier for the factor they used. If failure is costly or safety-critical, target a higher reliability than the default 90%¹.

Decide the reliability figure before you back-calculate C, because it changes the answer.

Check whether static capacity governs

Below roughly 2 RPM the bearing spends its life nearly stationary under load, and static capacity is claimed to be the limiting criterion rather than dynamic C⁸. Engineers who size only on dynamic C values routinely underestimate the required safety margin at low speed⁸.

The static check uses the basic static load rating C0 and the safety coefficient fs = C0/P0⁴ (supplier-reported figure). That is a separate calculation from the L10 work, and it can set the size on its own.

A crane slewing bearing turning 15,000 rotations a year is checked against a ten-year minimum before scheduled replacement, not only against an L10 hour count⁸. When a supplier's datasheet quotes only a dynamic C value, ask explicitly for the C0 figure and verify it is ISO 76-compliant⁸.

If your application turns slowly or holds position under load, run the static check before you finalise the size.

Set the target life from the duty class

The target life is an input you choose, and choosing it from the duty class is what makes the back-calculated C meaningful instead of arbitrary. The sources give example ranges rather than a selection rule.

For 24-hour operation with no failure allowed — water supply, power stations, mine drainage — another gives 100,000–200,000 h¹⁰. A crane slewing bearing at 15,000 rotations per year is specified at a minimum of ten years before scheduled replacement⁸.

Those are three different duty classes and three different targets, so the figure you write into the RFQ depends on which one you are in. The sources do not give a rule for choosing the target life for a given application, so state your basis explicitly.

If your duty is continuous with no failure allowed, the 100,000–200,000 h band is the one to argue from¹⁰. Write the target life and its basis into the RFQ so the supplier sizes to the same number you did.

Demand the calculation and the provenance

A counterfeit or unverified bearing made from inferior steel can cut fatigue life by up to 80%, which breaks the link between the catalog rating and the life you calculated¹⁸ (supplier-reported figure).

A supplier who cannot produce the L10 calculation cannot be checked at all, and one source treats that as a red flag⁸. Provenance is not paperwork for its own sake: it is what keeps the C you sized on equal to the C that arrives.

OEM components carry a 20–40% price premium and longer lead times over distributors and the aftermarket¹⁸ (supplier-reported). Confirm these before the order is placed.

• ✓Ask the supplier to provide the L10 calculation for your P, n and target life
• ✓Ask for the C0 figure and confirmation it is ISO 76-compliant
• ✓Verify ISO 9001 manufacturing certification
• ✓Request material test reports confirming steel cleanliness and alloy composition
• ✓Confirm the source is an authorized distributor, not the gray market

Where the sources disagree

Where two sources give different figures for the same decision input, the buyer who picks one silently inherits that source's assumption, so the disagreement has to stay visible. Grease fill is given as 30–50% by one source and 25–35% of free space for high-speed greases by another^9,16.

Low speed is governed by dynamic C in one workflow and by static C0 below 2 RPM in another^1,3,8,10. Target life ranges differ by duty class^8,10,14.

And ABEC class is a selection step in one source but only a tolerance band in another^1,16,20. Ask which assumption each figure rests on before you treat any one as the rule.

What the sources do not establish

• No source provides actual dynamic load rating C values for specific bearing sizes or part numbers
• No source provides boundary dimensions (bore, OD, width) or static load rating C0 values for candidate sizes
• No source provides numeric limiting speed or reference speed values for candidate sizes
• No source gives a method for combining a multi-level load spectrum and speed profile into a single equivalent dynamic load P and equivalent speed
• No source gives a rule for choosing the target life (hours or revolutions) for a given application; only example ranges are given
• No source gives unit price by size, volume price breaks, tooling or setup costs, or lead time for specific candidate sizes
• No source provides material/heat treatment certificates, dimensional or runout tolerance data, or test reports for a specific size
• No source states how shock or vibration loads are quantified or incorporated into P
• No source states how operating temperature modifies C, life or lubrication choice quantitatively
• No source gives the a1 values beyond 99% reliability or the a2/a3 factors' numeric treatment for the buyer's case