Why Bearings Fail: The L10 Life Idea Explained Simply

A machine can run smoothly for months, make a little extra noise for one week, and then suddenly start shedding shiny metallic flakes into the grease. That feels mysterious until you realize a rolling bearing is not failing from one dramatic overload most of the time. It is usually failing because the same tiny contact patch has been stressed millions of times.

The idea engineers use to talk about that fatigue process is L10 life. It sounds abstract at first, but it answers a practical question: if I know the bearing type, the load it sees, and the shaft speed, what fatigue life should I roughly expect before a noticeable fraction of similar bearings begin to fail? Once that idea clicks, catalog ratings, load margins, and early bearing failures make much more sense.

Most bearing life problems start below the surface

When a rolling-element bearing carries load, the ball or roller does not touch the raceway at a mathematical point or line forever. The steel deforms elastically by a microscopic amount, which creates a small real contact patch. The pressure inside that patch is very high, and beneath it the material experiences repeated changing shear stress as each element rolls past.

If the load, lubrication, and cleanliness are reasonable, the bearing can survive an astonishing number of cycles. But the stressed volume under the surface slowly accumulates fatigue damage. A tiny crack starts below the running track, grows toward the surface, and eventually breaks out as a spall, which is a small flake or pit in the raceway. Once that happens, vibration and noise rise quickly because every ball or roller now strikes the damaged area on each revolution.

Figure 1: A bearing may look healthy for many cycles, then show a small crack, and finally break out a spalled area that quickly raises noise and vibration.

This matters because many beginners imagine a failed bearing as a lubrication-only problem or a simple wear problem. Those certainly happen, but the basic rating life approach is mainly about rolling-contact fatigue. L10 life is not trying to predict every possible failure. It is trying to predict when fatigue damage becomes likely in a population of similar bearings.

What L10 actually means

The most useful plain-language definition is this: L10 life is the life that 90% of a large group of apparently identical bearings are expected to exceed under the same operating conditions. That also means about 10% of the group may fail from fatigue before reaching that life.

That wording is important. L10 is not the life of one guaranteed bearing. It is not the average life, and it is not the minimum life. Real bearings have tiny differences in material cleanliness, heat treatment, surface finish, mounting accuracy, and operating contamination. So a batch of bearings spreads out in service life even when they came from the same catalog page.

Figure 2: L10 is a population idea: one bearing in a group can show fatigue damage while most of the similar bearings still continue running.

That is why L10 is a conservative but useful language for design. If a machine builder needs long, predictable service, they do not size the bearing so the estimated L10 life barely matches the planned service interval. They usually want a healthy margin because the real machine may run hotter, dirtier, or less perfectly aligned than the ideal catalog case.

The core equation and where it comes from

For a beginner, the most important bearing-life equation is:

L10 = (C / P)^p

Here L10 is the basic rating life in millions of revolutions, C is the bearing's basic dynamic load rating, P is the equivalent dynamic bearing load, and p depends on bearing type. For ball bearings, engineers commonly use p = 3. For roller bearings, the exponent is usually a little larger, often written as 10/3.

The exact exponent is not a magic number pulled from nowhere. It comes from the way contact stress, material fatigue, and life statistics behave in rolling contact. The full derivation used by bearing standards is more involved than a first course needs, but the practical message is simple: life changes much faster than load. Because the load term sits in the denominator and is raised to a power, even a moderate load increase can shorten life sharply.

To convert from millions of revolutions to hours, use shaft speed n in rpm:

L10h = (10^6 / 60n) x (C / P)^p

That step is what makes the catalog number useful to a working engineer. Bearings do not fail after counting only revolutions on paper; they fail after operating hours on an actual machine.

Worked example 1: turning catalog life into operating hours

Suppose a blower shaft uses a deep-groove ball bearing with a basic dynamic load rating of C = 35 kN. After doing the shaft reaction calculation and combining radial and small axial effects, you estimate an equivalent dynamic bearing load of P = 4.5 kN. The shaft runs at 960 rpm.

For a ball bearing, use p = 3. First calculate life in millions of revolutions:

L10 = (35 / 4.5)^3 = 470.5 million rev

Now convert that to hours:

L10h = 470.5 x 10^6 / (60 x 960) = 8169 h

So the basic rating life is about 8200 hours. Interpreting that result correctly is the important part. It does not mean the bearing will definitely fail at 8200 hours. It means that under basic rating assumptions, about 90% of similar bearings should exceed that life before fatigue failure. Some will run much longer. A few may fail earlier.

Worked example 2: a small load reduction can nearly double life

Keep the same bearing and speed, but imagine the machine team improves pulley alignment and reduces belt tension. The equivalent dynamic load falls from 4.5 kN to 3.6 kN.

The new basic rating life becomes:

L10,new = (35 / 3.6)^3 = 918.4 million rev

Convert to hours:

L10h,new = 918.4 x 10^6 / (60 x 960) = 15944 h

The life has risen from about 8200 hours to about 15900 hours. That is a life ratio of:

15944 / 8169 = 1.95

So a 20% load reduction nearly doubled fatigue life. This is the engineering reason experienced designers care so much about shaft layout, belt tension, gear overhang, and misalignment. The bearing may look unchanged from the outside, but the life math changed dramatically.

Figure 3: The same bearing arrangement can see very different fatigue life as radial load rises, which is why even a modest load increase matters.

Why real bearings still fail earlier than the catalog life suggests

The basic L10 equation assumes the bearing is properly mounted, adequately lubricated, reasonably clean, and not suffering from severe misalignment or shock. Real machines do not always cooperate. That is why an apparently "well-sized" bearing can still die early.

Contamination is one of the most common life killers. A hard dirt particle dents the raceway and leaves a local stress raiser. The fatigue crack then starts from that damaged spot far sooner than the clean-raceway calculation assumed.

Poor lubrication is another frequent culprit. If the oil film or grease film becomes too thin, the surfaces see more direct metal-to-metal distress, more heat, and more surface smearing. The fatigue model assumes a healthier surface condition than that.

Misalignment changes the load distribution across the rolling elements. Instead of sharing load neatly, the bearing loads one edge or one part of the raceway too heavily. The equivalent dynamic load on paper may no longer represent the real stress state inside the bearing.

Shock loading creates dents or unusually high short-duration contact stress. That can push the problem away from normal fatigue life and toward brinelling or early crack initiation. In other words, some bearings fail because they were overloaded once, not merely because they were cycled many times.

What L10 does not tell you

L10 is powerful, but it has limits. It does not directly tell you about grease replacement interval, seal life, corrosion, cage failure, electrical pitting, or transport damage. It also does not tell you the exact life of one specific bearing sitting on your bench.

That is why bearing engineers separate questions. One question is, "Is the fatigue rating life reasonable for this load?" Another is, "Will the machine actually give the bearing a fair chance to reach that life?" A design that passes the first question but ignores the second one may still fail in service.

How standards and catalogs handle the topic

Modern bearing standards and catalog methods package a complicated rolling-contact fatigue problem into a practical design workflow. The catalog gives you the basic dynamic rating C. You calculate or look up the equivalent dynamic load P from the radial and axial loading case. Then the basic life equation gives a first fatigue-life estimate.

For more serious work, the standards framework goes further. Reliability adjustments beyond the basic 90% level, lubrication quality, contamination level, and material improvements can all matter. Manufacturers often provide application factors and adjusted life methods because the same nominal load means very different things in a dusty conveyor, a sealed electric motor, and a precision spindle.

The practical lesson is that standards do not remove engineering judgment. They organize it. They give you a common language so one engineer's "comfortable margin" is not another engineer's guess.

Practical rules of thumb

• Always calculate shaft reactions before selecting the bearing. Guessing load is usually more dangerous than choosing the wrong catalog series.
• Read L10 as a population-based fatigue measure, not as a promise to one bearing.
• Remember the cube law for ball bearings. Small load increases hurt life more than intuition suggests.
• Protect the bearing during assembly. Cleanliness and alignment are often as important as catalog rating.
• If the machine is shock-loaded or badly contaminated, basic L10 alone is not enough for selection.
• When a bearing fails early, inspect the raceway pattern before blaming the catalog. The damage shape usually tells a story.

Engineering judgment: what an experienced designer notices

When engineers say "the bearing failed," the useful follow-up question is "failed from what?" If the raceway shows classic spalling after long service, L10 thinking probably applies directly. If the raceway is dented, blue from heat, edged-over from misalignment, or dirty with hard particles, then the machine environment changed the problem.

That is why good designers use L10 as a starting point, not a final excuse. They choose a sensible dynamic rating, reduce unnecessary load where they can, keep the shaft and housing aligned, and give the bearing clean lubricant. The bearing life equation rewards all of those decisions.

If you want more machine-design explanations in the same style, continue through the EnggTools Articles library.