A rolling-element bearing is a bearing which carries a load by placing
      round elements between the two pieces. The relative motion of the pieces
      causes the round elements to roll (tumble) with little sliding.
      One of the earliest and best-known rolling-element bearings are sets of
      logs laid on the ground with a large stone block on top. As the stone is
      pulled, the logs roll along the ground with little sliding friction. As
      each log comes out the back, it is moved to the front where the block then
      rolls on to it. You can imitate such a bearing by placing several pens or
      pencils on a table and placing your hand on top of them. See "bearings"
      for more on the historical development of bearings.
      A rolling-element rotary bearing uses a shaft in a much larger hole, and
      cylinders called "rollers" tightly fill the space between the shaft and
      hole. As the shaft turns, each roller acts as the logs in the above
      example. However, since the bearing is round, the rollers never fall out
      from under the load.
      Rolling-element bearings have the advantage of a good tradeoff between
      cost, size, weight, carrying capacity, durability, accuracy, friction, and
      so on.Other bearing designs are often better on one specific attribute,
      but worse in most other attributes. Only plain bearings have as wide use
      as rolling-element bearings.

      Design
      Typical rolling-element bearings range in size from 10 mm diameter to a
      few metres diameter, and have load-carrying capacity from a few tens of
      grams to many thousands of tonnes.
      A particularly common kind of rolling-element bearing is the ball bearing.
      The bearing has inner and outer races and a set of balls. Each race is a
      ring with a groove where the balls rest. The groove is usually shaped so
      the ball is a slightly loose fit in the groove. Thus, in principle, the
      ball contacts each race at a single point. However, a load on an
      infinitely small point would cause infinitely high contact pressure. In
      practice, the ball deforms (flattens) slightly where it contacts each
      race, much as a tire flattens where it touches the road. The race also
      dents slightly where each ball presses on it. Thus, the contact between
      ball and race is of finite size and has finite pressure. Note also that
      the deformed ball and race do not roll entirely smoothly because different
      parts of the ball are moving at different speeds as it rolls. Thus, there
      are opposing forces and sliding motions at each ball/race contact.
      Overall, these cause bearing drag.
      There are many types of rolling-element bearings, each tuned for a
      specific kind of load and with specific advantages and disadvantages. For
      example:
      
        Ball bearings use spheres instead of cylinders. Clever use of surface
        tension allows balls of high accuracy to be made much more cheaply than
        comparable cylinders. Ball bearings can support both radial
        (perpendicular to the shaft) and axial loads (parallel to the shaft).
        For lightly-loaded bearings, balls offer lower friction than rollers.
        Ball bearings can operate when the bearing races are misaligned.
        
      
       Caged radial cylindrical bearings
      
        Common roller bearings use cylinders of slightly greater length than
        diameter. Roller bearings typically have higher radial load capacity
        than ball bearings, but a low axial capacity and higher friction under
        axial loads. If the inner and outer races are misaligned, the bearing
        capacity often drops quickly compared to either a ball bearing or a
        spherical roller bearing.
        
        Needle roller bearings use very long and thin cylinders. Since the
        rollers are thin, the outside diameter of the bearing is only slightly
        larger than the hole in the middle. However, the small-diameter rollers
        must bend sharply where they contact the races, and thus the bearing
        fatigues relatively quickly.
        
        Taper roller bearings use conical rollers that run on conical races.
        Most roller bearings only take radial loads, but taper roller bearings
        support both radial and axial loads, and thus have some of the same
        advantages as ball bearings. Taper roller bearings are used, for
        example, as the wheel bearings of most cars, trucks, buses, and so on. A
        disadvantage is that the tapered roller is like a wedge and thus bearing
        loads try to eject the roller; the force which keeps the roller in the
        bearing adds to bearing friction.
        
        sphcriel roller bearing use rollers that are thicker in the middle and
        thinner at the ends; the race is shaped to match. Spherical roller
        bearings can thus adjust to support misaligned loads. However, spherical
        rollers are difficult to produce and thus expensive. And, the bearings
        have higher friction than a comparable ball bearing since different
        parts of the spherical rollers run at different speeds on the rounded
        race and thus there are opposing forces along the bearing/race contact.
        
      Most rolling-element bearing designs are for rotating or oscillating
      loads, but there are also linear bearing designs. A common example is
      drawer-support hardware. Another example is a bearing for a shaft which
      moves axially in a hole. Axial-motion bearings often work like the
      stone-and-log example, with a pathway so rolling elements that fall off
      the end are pushed around to the other end, and the load rolls on to it.
      These are called recirculating bearings.

      Bearing failure
      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.
      There are three usual limits to the lifetime or load capacity of a
      bearing: abrasion, fatigue and pressure-induced welding. Abrasion is when
      the surface is eroded by hard contaminants scraping at the bearing
      materials. Fatigue is when a material breaks after it is repeatedly bent
      and released. Where the ball or roller touches the race there is always
      some bending, and hence a risk of fatigue. Smaller balls or rollers bend
      more sharply, and so tend to fatigue faster. Pressure-induced welding is
      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 tiny 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.

      Constraints and trade-offs
      All parts of a bearing are subject to many design constraints. For
      example, the inner and outer races are often complex shapes, making them
      difficult to manufacture. Balls and rollers, though simpler in shape, are
      small; since they bend sharply where they run on the races, the bearings
      are prone to fatigue. The loads within a bearing assemble are also
      affected by the speed of operation: rolling-element bearings may spin over
      100,000 rpm, and the principal load in such a bearing may be centrifugal
      force rather than the applied load. Smaller rolling elements are lighter
      and thus have less centrifugal force, but smaller elements also bend more
      sharply where they contact the race, causing them to fail more rapidly
      from fatigue.
      There are also many material issues: a harder material may be more durable
      against abrasion but more likely to suffer fatigue fracture, so the
      material varies with the application, and while steel is most common for
      rolling-element bearings, plastics, glass, and ceramics are all in common
      use. A small defect (irregularity) in the material is often responsible
      for bearing failure; one of the biggest improvements in the life of common
      bearings during the second half of the 1900s was the use of more
      homogeneous materials, rather than better materials or lubricants (though
      both were also significant). Lubricant properties vary with temperature
      and load, so the best lubricant varies with application.
      Although bearings tend to wear out with use, designers can make tradeoffs
      of bearing size and cost versus lifetime. A bearing can last indefinitely
      -- longer than the rest of the machine -- if it is kept cool, clean,
      lubricated, is run within the rated load, and if the bearing materials are
      sufficiently free of microscopic defects. Note that cooling, lubrication,
      and sealing are thus important parts of the bearing design.
      The needed bearing lifetime also varies with the application. For example,
      Harris reports on an oxygen pump bearing in the U.S. Space Shuttle which
      could not be adequately isolated from the liquid oxygen being pumped, but
      all lubricants reacted with the oxygen leading to fires and other
      failures. The solution was to lubricate the bearing with the oxygen.
      Although liquid oxygen is a poor lubricant, it was adequate, since the
      service life of the pump was just a few hours.
      The operating environment and service needs are also important design
      considerations. Some bearing assemblies require routine addition of
      lubricants, while others are factory sealed, requiring no further
      maintenance for the life of the mechanical assembly. Although seals are
      appealing, they increase friction, and a permanently-sealed bearing may
      have the lubricant contaminated by hard particles, such as steel chips
      from the race or bearing, sand, or grit that got past the seal.
      Contamination in the lubricant is abrasive and greatly reduces the
      operating life of the bearing assembly.