Roller Bearings

What is a Roller Bearing

A roller bearing is a rolling-element bearing in which the load between a rotating shaft and a fixed housing is transmitted by hardened steel rollers running between two raceways. The defining feature is the contact geometry: a roller touches each raceway along a line, whereas a ball touches at a point. Line contact spreads the same load over a much larger area, so contact stress is lower for a given load and the bearing can carry more before the raceway plastically deforms. For a comparable envelope, a line-contact roller bearing typically carries on the order of two to three times the radial load of a ball bearing of the same bore and outside diameter, which is why heavy and shock-loaded machinery uses rollers almost universally.

Every rolling bearing, roller or ball, is built from four functional parts: the inner ring with its raceway, the outer ring with its raceway, the rolling elements (here, rollers), and the cage that separates and guides those rollers at equal spacing. Some designs add integral seals or shields, and some, such as needle roller bearings, omit one or both rings to save radial space. The rollers and raceways carry the load; the cage carries none, but it governs the bearing's speed, temperature and noise behavior.

Rolling bearings replaced plain (sliding) bearings across most of industry because rolling friction is roughly one to two orders of magnitude lower than sliding friction at startup, which slashes wear and energy loss. The modern caged rolling bearing dates to the late nineteenth century: Friedrich Fischer of Schweinfurt, Germany devised a machine to grind steel balls to high precision in 1883, and Sven Wingqvist of Sweden patented the self-aligning multi-row ball bearing in 1907, founding SKF. The Timken Company, founded by Henry Timken in 1899, commercialized the tapered roller bearing for carriage and automobile axles, establishing the inch-series geometry still in use today.

The application scale of roller bearings spans an enormous range. Miniature needle bearings a few millimeters across guide gearbox shafts and automotive transmissions, while slewing and spherical roller bearings several meters in diameter support wind-turbine main shafts, ship rudders, excavator turntables and rolling-mill rolls. A single large bearing can carry hundreds of tonnes of static load. Because boundary dimensions, the bore, outside diameter and width that define how a bearing fits a shaft and housing, are standardized in ISO 15 and ISO 355, bearings from competing manufacturers are dimensionally interchangeable, which is central to how procurement works in this category.

Four engineering quantities determine whether a roller bearing is the right one for a duty: the basic dynamic load rating C, the basic static load rating C0, the rating life L10 under the actual load and speed, and the tolerance and internal clearance class that match the shaft fit. These four, together with material and lubrication, set both the reliability and the total cost of ownership across the bearing's service life.

The Four Roller Bearing Families

Roller bearings divide into four mainstream families by roller geometry: cylindrical, tapered, spherical and needle. Each geometry resolves load in a characteristic direction and tolerates a characteristic degree of misalignment, so choosing the family is the first and most consequential selection step. The table below summarizes the load and misalignment behavior of each family before the detailed discussion.

Cylindrical roller bearings use straight rollers that contact the raceway along their full length, giving the highest radial capacity of any single-row design and good high-speed performance. Their axial capability depends on the rib configuration, which the designation encodes: NU and N types have no ribs on one ring and float axially, suiting the non-locating end of a shaft where thermal growth must be absorbed; NJ and NF types add one rib and locate the shaft in one axial direction; NUP and NH types locate it in both directions. Double-row NN and NNU designs serve machine-tool spindles where high radial stiffness is needed.

Tapered roller bearings use conical rollers running on conical raceways whose apexes meet on the bearing axis, so true rolling occurs across the contact. This geometry carries combined radial and axial load simultaneously, which is why it dominates automotive wheel hubs, gearbox shafts and machine-tool spindles. Because a single row produces an induced thrust under radial load, tapered bearings are mounted in opposed pairs and set to a defined end-play or preload at assembly. The inner ring (cone) and outer ring (cup) are separable, which simplifies mounting. Inch-series geometry follows the Timken convention; metric boundary dimensions follow ISO 355.

Spherical roller bearings use barrel-shaped rollers in two rows running on a common spherical outer raceway. The spherical raceway lets the inner ring and rollers swivel relative to the outer ring, so the bearing self-aligns and tolerates shaft deflection or housing misalignment of roughly 0.5 to 2 degrees without edge loading. They carry very high radial load plus axial load in both directions and absorb heavy shock, making them the standard for vibrating screens, crushers, conveyors, paper machines and wind-turbine gearboxes. Many are supplied with a tapered 1:12 bore (1:30 for the heaviest 240 and 241 series) for mounting on an adapter sleeve.

Needle roller bearings are cylindrical roller bearings whose rollers have a length three to ten times their diameter. The slim rollers give very high radial capacity in a small radial cross-section, so needle bearings save space where shaft diameter is fixed but radial envelope is tight, as in automotive transmissions, universal joints, pumps and small gearboxes. Drawn-cup needle bearings use a thin pressed outer shell and run directly on a hardened shaft as the inner raceway, which minimizes outside diameter. The trade-off is that needles do not tolerate misalignment and carry no axial load.

Load Ratings and Rating Life

Two load ratings appear on every bearing data sheet, and they answer different questions. The basic dynamic load rating C governs fatigue life of a rotating bearing; the basic static load rating C0 governs permanent deformation of a stationary or slowly oscillating bearing. Confusing the two is a frequent and expensive error, so this chapter separates them and shows how each enters a calculation.

The basic dynamic load rating C, defined in ISO 281, is the constant radial load (or axial load for thrust bearings) under which a group of identical bearings attains a basic rating life of one million revolutions. It is a calculated reference value, not a load the bearing should actually run at. The basic static load rating C0, defined in ISO 76, is the load that produces a total permanent deformation of rolling element and raceway equal to 0.0001 of the rolling-element diameter at the most heavily loaded contact. Static selection compares C0 with the equivalent static load P0 through the static safety factor s0 = C0 divided by P0.

The basic rating life L10 is the life that 90 percent of a population of identical bearings will reach or exceed before the first sign of fatigue spalling; equivalently, 10 percent are expected to have failed. For roller bearings the ISO 281 relation is L10 = (C divided by P) raised to the power 10/3, expressed in millions of revolutions, where P is the equivalent dynamic load. The exponent is 10/3 for roller bearings and 3 for ball bearings, reflecting line versus point contact. The table below works a representative example to make the sensitivity concrete.

Two consequences follow from the 10/3 exponent. First, life is extremely sensitive to load: doubling P cuts roller-bearing life to roughly one tenth, so a modest load reduction or a one-size-larger bearing yields a large life gain. Second, life hours depend on speed, since L10 in hours equals L10 in million revolutions divided by 60 times the rpm. For combined loads, the equivalent dynamic load is P = X times Fr plus Y times Fa, where the radial factor X and axial factor Y come from the specific bearing table and depend on the ratio of axial to radial load.

The basic L10 assumes ideal lubrication and cleanliness, which real installations rarely achieve, so ISO 281 also defines the modified rating life Lnm = a1 times aISO times L10. The reliability factor a1 raises the target above 90 percent: it equals 1.0 at 90 percent, about 0.64 at 95 percent and about 0.25 at 99 percent reliability. The systems life modification factor aISO combines the lubricant viscosity ratio kappa, the contamination factor eta-c keyed to ISO 4406 oil-cleanliness classes, and the ratio of the fatigue load limit Cu to the load P. Clean oil with a thick elastohydrodynamic film can push aISO above 1, extending life beyond L10; contamination, water and thin films can drive aISO below 0.1, which is why field life so often falls short of catalog L10 when lubrication is neglected.

Bearing Steel, Cages and Lubrication

The fatigue life predicted by ISO 281 is achievable only if the steel, the cage and the lubricant are matched to the duty. Material choice sets the ceiling on contact-fatigue resistance and operating temperature; the cage sets speed and quiet-running behavior; and lubrication, often called the fifth bearing component, is the single largest determinant of real-world life. This chapter covers all three.

Bearing steel. The classic raceway and roller material is high-carbon chromium through-hardening steel, designated SAE 52100 in the United States, DIN 100Cr6 in Europe, JIS SUJ2 in Japan, and EN material number 1.3505. It contains about 1 percent carbon and 1.4 to 1.6 percent chromium, and after hardening and tempering it reaches roughly 58 to 65 HRC uniformly through the section, the hardness needed to resist rolling-contact fatigue, indentation and wear. For large bearings, heavy shock or contaminated service, makers use case-carburizing steels such as 20MnCr5 or 18CrNiMo7-6: carburizing builds a hard wear-resistant case (about 58 to 64 HRC) over a tough, ductile low-carbon core that resists fracture under shock. Cleaner melts (vacuum-degassed, VIM-VAR) and surface engineering raise fatigue life further for aerospace and high-reliability duty.

Cage materials. The cage carries no load but its material decides the speed, temperature and noise envelope. The table below compares the three mainstream choices.

Glass-fiber-reinforced polyamide 66 gives the lowest friction and the quietest, fastest running, but the standard grade is generally limited to about 120 degrees Celsius continuous and can be embrittled by aged oil, high water content or some aggressive additives, so above that range or in hostile chemistry a brass or steel cage is specified. For the highest-speed precision spindle bearings, PEEK cages extend both the temperature and the speed limit beyond polyamide. Where a polymer cage is at risk, brass remains the conservative default for large industrial and heavy-shock bearings.

Lubrication. The lubricant separates roller and raceway with an elastohydrodynamic film a fraction of a micron thick; whether that film fully forms is captured by the viscosity ratio kappa in the aISO life factor. Grease is the default for most sealed industrial bearings because it is simple, retains itself, and helps seal against contamination; it suits speeds up to a moderate fraction of the bearing's limiting speed and is specified by a base-oil viscosity and an NLGI consistency grade (commonly NLGI 2 or 3). Oil bath, oil circulation and oil-air or oil-mist systems are used for higher speeds, higher temperatures or where heat must be carried away, as in gearboxes and high-speed spindles. Insufficient or contaminated lubrication, not fatigue, is the leading practical cause of premature roller-bearing failure.

Key Specification Parameters

Reading a bearing data sheet is a core procurement skill. A single page may list two dozen numbers, but only a handful drive the decision. This chapter decodes the parameters that matter, in the order they typically appear on a manufacturer table.

Boundary dimensions are bore d, outside diameter D and width B (plus the cup width and overall width for tapered bearings), all in millimeters and standardized by ISO 15 (radial bearings) and ISO 355 (metric tapered). Because they are standardized, a bearing of a given designation has the same envelope across brands. In a metric basic designation, the last two digits encode the bore: for codes 04 and above, multiply by 5 to get the bore in millimeters (so a 6308 or NU308 has a 40 mm bore), while codes 00, 01, 02 and 03 mean bores of 10, 12, 15 and 17 mm.

Basic dynamic load rating C and basic static load rating C0, both in kilonewtons, are the load anchors discussed in Chapter 3: C feeds the L10 fatigue-life formula for rotating bearings, while C0 feeds the static safety factor s0 = C0 divided by P0 for stationary or slowly oscillating bearings. A common guideline is s0 of at least 1.5 for roller bearings under normal load, rising toward 3 or more where shock load or quiet running is required. The fatigue load limit Cu is the load below which, under clean lubrication, no fatigue is expected; it feeds the aISO modified-life factor.

Limiting and reference speeds, in rpm, bound how fast the bearing can run. The reference speed is a thermal limit derived under defined cooling and lubrication; the limiting speed is a mechanical limit set by cage strength, roller dynamics and seal friction. Roller bearings generally have lower speed limits than ball bearings of the same size because of higher friction and heat generation. The table below decodes the parameter set with representative meaning.

Internal radial clearance is the play between rollers and raceways before mounting, graded C2 (less than normal), CN or C0 (normal), C3, C4 and C5 (progressively greater). Press-fitting the inner ring onto an interference-fit shaft expands the raceway and consumes some clearance, and thermal expansion in service consumes more, so a heated or tightly fitted shaft is usually paired with a C3 or C4 bearing to preserve a small running clearance and avoid preload-induced overheating.

Tolerance class, per ISO 492 (equivalently DIN 620 and JIS B 1514), defines how tightly bore, outside diameter, width and running accuracy are held. From loosest to tightest the sequence is Normal (P0), P6, P5, P4, P3 and P2; the smaller the number, the tighter the bearing. Normal class suits general machinery, P6 suits motors and gearboxes, P5 suits machine-tool spindles, and P4 and P2 serve high-speed precision spindles. The ABEC scale runs opposite, with ABEC 1 mapping to P0, ABEC 5 to P5, ABEC 7 to P4 and ABEC 9 to P2.

Selection Decision Factors

To turn the preceding chapters into a specific part number, work through the decision sequence below. Most selection failures come not from a single wrong number but from settling a downstream choice before an upstream one. These eight steps double as a fixed RFQ template.

1. Load direction and magnitude: Resolve the load into radial Fr and axial Fa components and identify shock content. Pure heavy radial points to cylindrical; combined radial-plus-axial points to tapered; misalignment plus heavy radial points to spherical; a tight radial envelope points to needle.
2. Misalignment and deflection: Estimate shaft slope under load and the alignment achievable between the two bearing seats. If misalignment can exceed a few arc-minutes, choose a self-aligning spherical roller bearing; rigid cylindrical or tapered bearings edge-load and fail early under misalignment.
3. Size for life: Pick a candidate, read its C, and compute L10 = (C divided by P) to the 10/3 power; convert to hours at the operating speed. If L10 falls short of the required service interval, step up one size or family rather than accept the shortfall, since the 10/3 exponent makes a small size increase very effective.
4. Static check: For stationary, slowly oscillating or heavily shock-loaded duty, verify s0 = C0 divided by P0 meets at least 1.5 for normal roller-bearing load and 3 or more under shock.
5. Speed and lubrication: Confirm the operating speed sits within the limiting speed, then choose grease (sealed, simple, moderate speed) or oil (high speed, high temperature, heat removal) and specify viscosity and consistency to achieve a viscosity ratio kappa near or above 1.
6. Fit, clearance and tolerance: Set shaft and housing fits per ISO 286 to the load and rotation pattern, pick internal clearance (C3 or C4 for interference fits or hot running), and select the tolerance class (Normal for general duty, P5 or better for spindles).
7. Sealing, material and environment: Specify seals or shields against contamination and water, choose through-hardened 52100 or case-carburized steel by shock and contamination level, and pick a cage material by temperature and speed.
8. Total cost of ownership: Weigh purchase price against expected life, relubrication interval, downtime cost and the standardized interchangeability that lets a second-source brand of the same designation drop in. A bearing chosen one size up or one cleanliness class better often repays its premium within the first overhaul cycle.

One last dimension is frequently overlooked: manufacturer serviceability and interchangeability. Because boundary dimensions are standardized in ISO 15 and ISO 355, a designation such as NU308 or 22215 carries the same envelope across SKF, Schaeffler (FAG and INA), NSK, NTN, Koyo (JTEKT), Timken and Chinese makers ZWZ, HRB, LYC and C&U, so a worn unit can be second-sourced by part number. What differs between brands is internal geometry optimization, steel cleanliness, cage design, quality consistency and the availability of local stock, mounting tools and engineering support, which is why critical drives favor the established brands while non-critical auxiliaries can use cost alternatives of the same designation.