Editing Rolling-element bearing (section)

==Bearing failure==
[[File:FailedBearing.jpg|thumb|A prematurely failed rear bearing cone from a [[mountain bicycle]], caused by a combination of [[pitting]] due to wet conditions, improper lubrication, improper pre-load adjustment, and fatigue from frequent shock loading.]]

Rolling-element bearings often work well in non-ideal conditions, but sometimes minor problems cause bearings to fail quickly and mysteriously. For example, with a stationary (non-rotating) load, small vibrations can gradually press out the lubricant between the races and rollers or balls ([[false brinelling]]). Without lubricant the bearing fails, even though it is not rotating and thus is apparently not being used. For these sorts of reasons, much of bearing design is about failure analysis. Vibration based analysis can be used for fault identification of bearings.<ref>{{cite journal|last=Slavic|first=J |author2=Brkovic, A |author3=Boltezar M|title=Typical bearing-fault rating using force measurements: application to real data.|journal=Journal of Vibration and Control|date=December 2011|volume=17|issue=14|pages=2164–2174|doi=10.1177/1077546311399949|s2cid=53959482 |url=http://lab.fs.uni-lj.si/ladisk/?what=abstract&ID=60|url-access=subscription}}</ref>

There are three usual limits to the lifetime or load capacity of a bearing: abrasion, fatigue and pressure-induced welding. 
*Abrasion occurs when the surface is eroded by hard contaminants scraping at the bearing materials. 
*Fatigue results when a material becomes brittle after being repeatedly loaded and released. Where the ball or roller touches the race there is always some deformation, and hence a risk of fatigue. Smaller balls or rollers deform more sharply, and so tend to fatigue faster. 
*Pressure-induced welding can occur when two metal pieces are pressed together at very high pressure and they become one. Although balls, rollers and races may look smooth, they are microscopically rough. Thus, there are high-pressure spots which push away the bearing [[lubricant]]. Sometimes, the resulting metal-to-metal contact welds a microscopic part of the ball or roller to the race. As the bearing continues to rotate, the weld is then torn apart, but it may leave race welded to bearing or bearing welded to race.

Although there are many other apparent causes of bearing failure, most can be reduced to these three. For example, a bearing which is run dry of lubricant fails not because it is "without lubricant", but because lack of lubrication leads to fatigue and welding, and the resulting wear debris can cause abrasion. Similar events occur in false brinelling damage. In high speed applications, the oil flow also reduces the bearing metal temperature by convection. The oil becomes the heat sink for the friction losses generated by the bearing.

ISO has categorised bearing failures into a document Numbered ISO 15243.

=== Life calculation models ===
The life of a rolling bearing is expressed as the number of revolutions or the number of operating hours at a given speed that the bearing is capable of enduring before the first sign of metal fatigue (also known as [[spalling]]) occurs on the race of the inner or outer ring, or on a rolling element. Calculating the endurance life of bearings is possible with the help of so-called life models. More specifically, life models are used to determine the bearing size – since this must be sufficient to ensure that the bearing is strong enough to deliver the required life under certain defined operating conditions.

Under controlled laboratory conditions, however, seemingly identical bearings operating under identical conditions can have different individual endurance lives. Thus, bearing life cannot be calculated based on specific bearings, but is instead related to in statistical terms, referring to populations of bearings. All information with regard to load ratings is then based on the life that 90% of a sufficiently large group of apparently identical bearings can be expected to attain or exceed. This gives a clearer definition of the concept of bearing life, which is essential to calculate the correct bearing size. Life models can thus help to predict the performance of a bearing more realistically.

The prediction of bearing life is described in ISO 281<ref name="ISO281:2007" >{{Cite web
 |title=Rolling bearings -- Dynamic load ratings and rating life
 |publisher=ISO
 |year=2007
 |id=ISO281:2007
 |url=http://www.iso.org/iso/catalogue_detail.htm?csnumber=38102
}}</ref> and the [[ANSI]]/American Bearing Manufacturers Association Standards 9 and 11.<ref name="STLE, Zaretsky" />

The traditional life prediction model for rolling-element bearings uses the basic life equation:<ref name="MD, Bearing life" >{{Cite web
 |title=The meaning of bearing life
 |author=Daniel R. Snyder, SKF 
 |website=Machine Design
 |date=12 April 2007
 |url=http://machinedesign.com/bearings/meaning-bearing-life
}}</ref>

<math display="block">
L_{10} = (C/P)^p
</math>

Where:
* <math>L_{10}</math> is the 'basic life' (usually quoted in millions of revolutions) for a reliability of 90%, i.e. no more than 10% of bearings are expected to have failed
* <math>C</math> is the dynamic load rating of the bearing, quoted by the manufacturer
* <math>P</math> is the equivalent dynamic load applied to the bearing
* <math>p</math> is a constant: 3 for ball bearings, 4 for pure line contact and 3.33 for roller bearings

Basic life or <math>L_{10}</math> is the life that 90% of bearings can be expected to reach or exceed.<ref name="ISO281:2007" /> The median or average life, sometimes called [[Mean Time Between Failure]] (MTBF), is about five times the calculated basic rating life.<ref name="MD, Bearing life" />
Several factors, the '[[ASME]] five factor model',<ref name="STLE, ISO281" /> can be used to further adjust the <math>L_{10}</math> life depending upon the desired reliability, lubrication, contamination, etc.

The major implication of this model is that bearing life is finite, and reduces by a cube power of the ratio between design load and applied load. This model was developed in 1924, 1947 and 1952 work by [[:sv:Arvid Palmgren|Arvid Palmgren]] and Gustaf Lundberg in their paper ''Dynamic Capacity of Rolling Bearings''.<ref name="STLE, ISO281" /><ref name="eBearing, ISO281" /> The model dates from 1924, the values of the constant <math>p</math> from the post-war works. Higher <math>p</math> values may be seen as both a longer lifetime for a correctly-used bearing below its design load, or also as the increased rate by which lifetime is shortened when overloaded.

This model was recognised to have become inaccurate for modern bearings. Particularly owing to improvements in the quality of bearing steels, the mechanisms for how failures develop in the 1924 model are no longer as significant. By the 1990s, real bearings were found to give service lives up to 14 times longer than those predicted.<ref name="STLE, ISO281" /> An explanation was put forward based on [[fatigue life]]; if the bearing was loaded to never exceed the [[fatigue strength]], then the Lundberg-Palmgren mechanism for failure by fatigue would simply never occur.<ref name="STLE, ISO281" /> This relied on homogeneous [[vacuum-melted steel]]s, such as [[AISI 52100]], that avoided the internal inclusions that had previously acted as stress risers within the rolling elements, and also on smoother finishes to bearing tracks that avoided impact loads.<ref name="STLE, Zaretsky" /> The <math>p</math> constant now had values of 4 for ball and 5 for roller bearings. Provided that load limits were observed, the idea of a 'fatigue limit' entered bearing lifetime calculations. If the bearing was not loaded beyond this limit, its theoretical lifetime would be limited only by external factors, such as contamination or a failure of lubrication.

A new model of bearing life was put forward by [[Schaeffler Group|FAG]] and developed by [[SKF]] as the Ioannides-Harris model.<ref name="eBearing, ISO281" >{{Cite web
 |title=ISO Adopts SKF Bearing Life Calculations 
 |website=eBearing News
 |date=28 June 2006 
 |url=http://www.ebearing.com/news2006/062801.htm
}}</ref><ref name="Ioannides-Harris" >{{Cite journal
 |first1=Stathis |last1=Ioannides |first2=Ted |last2=Harris 
 |publisher=SKF 
 |year=1985
 |title=A New Fatigue Life Model for Rolling Bearings
}}</ref> ISO 281:2000 first incorporated this model and ISO 281:2007 is based on it.

The concept of fatigue limit, and thus ISO 281:2007, remains controversial, at least in the US.<ref name="STLE, Zaretsky">{{Cite journal
 |title = In search of a fatigue limit: A critique of ISO standard 281:2007
 |author = Erwin V. Zaretsky
 |date = August 2010
 |journal = Tribology & Lubrication Technology
 |pages = 30–40
 |url = http://www.stle.org/assets/document/8-10_tlt_commentary_ISO_Standard_281_2007_Part_II.pdf
 |publisher = Society of Tribologists and Lubrication Engineers (STLE)
 |url-status = dead
 |archive-url = https://web.archive.org/web/20150518100537/http://www.stle.org/assets/document/8-10_tlt_commentary_ISO_Standard_281_2007_Part_II.pdf
 |archive-date = 2015-05-18
|author-link = Erwin V. Zaretsky
 }}</ref><ref name="STLE, ISO281">{{Cite journal
 |title = ISO 281:2007 bearing life standard – and the answer is?
 |date = July 2010
 |journal = Tribology & Lubrication Technology
 |pages = 34–43
 |url = https://www.stle.org/assets/document/tlt_July_cover_story_article.pdf
 |publisher = Society of Tribologists and Lubrication Engineers (STLE)
 |url-status = dead
 |archive-url = https://web.archive.org/web/20131024084224/http://www.stle.org/assets/document/tlt_July_cover_story_article.pdf
 |archive-date = 2013-10-24
}}</ref>

=== Generalized Bearing Life Model (GBLM) ===
In 2015, the SKF Generalized Bearing Life Model (GBLM) was introduced.<ref>{{cite journal
 |last1=Morales-Espejel |first1=Guillermo E.
 |last2=Gabelli |first2=Antonio
 |last3=de Vries |first3=Alexander J. C.
 |title=A Model for Rolling Bearing Life with Surface and Subsurface Survival—Tribological Effects
 |journal=Tribology Transactions
 |volume=58
 |issue=5
 |pages=894–906
 |doi=10.1080/10402004.2015.1025932
 |year=2015
 |s2cid=137670935
 }}</ref> In contrast to previous life models, GBLM explicitly separates surface and subsurface failure modes, making the model flexible to accommodate several different failure modes. Modern bearings and applications show fewer failures, but the failures that do occur are more linked to surface stresses. By separating surface from the subsurface, mitigating mechanisms can more easily be identified. GBLM makes use of advanced tribology models<ref>{{cite journal
 |last1=Morales-Espejel |first1=Guillermo E.
 |last2=Brizmer |first2=Victor
 |title=Micropitting Modelling in Rolling–Sliding Contacts: Application to Rolling Bearings
 |journal=[[Tribology Transactions]]
 |volume=54 |year=2011 |issue=4 
 |pages=625–643
 |doi=10.1080/10402004.2011.587633
 |s2cid=137662003
 }}</ref> to introduce a surface distress failure mode function, obtained from the evaluation of surface fatigue. For the subsurface fatigue, GBLM uses the classical Hertzian rolling contact model. With all this, GBLM includes the effects of lubrication, contamination, and race surface properties, which together influence the stress distribution in the rolling contact.

In 2019, the Generalized Bearing Life Model was relaunched. The updated model offers life calculations also for hybrid bearings, i.e. bearings with steel rings and ceramic (silicon nitride) rolling elements.<ref>{{cite journal
 |last1=Morales-Espejel |first1=Guillermo E.
 |last2=Gabelli |first2=Antonio
 |title=A model for rolling bearing life with surface and subsurface survival: Sporadic surface damage from deterministic indentations 
 |journal=[[Tribology International]]
 |volume=96
 |date=April 2016
 |pages=279–288
 |doi=10.1016/j.triboint.2015.12.036
 }}</ref><ref>{{cite journal
 |last1=Morales-Espejel |first1=Guillermo E
 |last2=Gabelli |first2=Antonio
 |title=Application of a rolling bearing life model with surface and subsurface survival to hybrid bearing cases
 |journal=[[Proceedings of the Institution of Mechanical Engineers, Part C]]
 |volume=233
 |issue=15
 |pages=5491–5498
 |doi=10.1177/0954406219848470
 |year=2019
 |s2cid=164456996
 }}</ref> Even if the 2019 GBLM release was primarily developed to realistically determine the working life of hybrid bearings, the concept can also be used for other products and failure modes.