The guidance and support of a rotating shaft requires at least two bearings arranged at a certain distance from each other. Depending on the application, a decision is made between a locating/non-locating bearing arrangement, an adjusted bearing arrangement and a floating bearing arrangement.

On a shaft supported by two radial bearings, the distances between the bearing seats on the shaft and in the housing frequently do not coincide as a result of manufacturing tolerances. The distances may also change as a result of temperature increases during operation. These differences in distance are compensated in the non-locating bearing. ­Examples of locating/non-locating bearing arrangements: see Figure 1 to Figure 4.

Ideal non-locating bearings are cylindrical roller bearings with cage N and NU or needle roller bearings, Figure 1 , , link. In these bearings, the roller and cage assembly can be displaced on the raceway of the bearing ring without ribs.

All other bearing types, for example deep groove ball bearings and spherical roller bearings, can only act as non-locating bearings if one bearing ring has a fit that allows displacement, Figure 2. The bearing ring subjected to point load therefore has a loose fit; this is normally the outer ring, see Conditions of rotation, link.

The locating bearing guides the shaft in an axial direction and supports external axial forces. In order to prevent axial stresses, shafts with more than two bearings have only one locating bearing. The type of bearing selected as a locating bearing depends on the magnitude of the axial forces and the accuracy with which the shafts must be axially guided.

A double row angular contact ball bearing, Figure 3 , link, for example, will give closer axial guidance than a deep groove ball bearing or a spherical roller bearing. A pair of symmetrically arranged angular contact ball bearings or tapered roller bearings, Figure 4, used as locating bearings will provide extremely close axial guidance.

Angular contact ball bearings of the universal design, Figure 5, give particular advantages. The bearings can be fitted in pairs in any O or X arrangement without shims. Angular contact ball bearings of the universal design are matched such that, in an X or O arrangement, they have a low axial internal clearance (design UA), zero clearance (UO) or slight preload (UL).

Spindle bearings of the universal design UL, Figure 6, have slight preload when fitted in an X or O arrangement (designs with higher preload are available by agreement).

In gearboxes, a four point contact bearing is sometimes fitted directly adjacent to a cylindrical roller bearing to give a locating bearing arrangement, Figure 3 , link. The four point contact bearing, without radial support of the outer ring, can only support axial forces. The radial force is supported by the cylindrical roller bearing.

If a lower axial force is present, a cylindrical roller bearing with cage NUP can also be used as a locating bearing, Figure 4 , link.

Fitting is also made easier with a matched pair of tapered roller bearings as a locating bearing (313..-N11CA), Figure 7 , link. They are matched with appropriate axial internal clearance so that no adjustment or setting work is required.

These bearing arrangements normally consist of two symmetrically arranged angular contact ball bearings or tapered roller bearings, Figure 8. During fitting, one bearing ring is displaced on its seat until the bearing arrangement achieves the required clearance or the necessary preload.

Due to this adjustment facility, the adjusted bearing arrangement is particularly suitable where close guidance is required, for example in pinion bearing arrangements with spiral toothed bevel gears and spindle bearing arrangements in machine tools.

A fundamental distinction is drawn between the O arrangement, Figure 8 , and the X arrangement, Figure 8 , of the bearings. In the O arrangement, the cones and their apexes S formed by the contact lines point outwards; in the X arrangement, the cones point inwards. The support base H, in other words the distance between the apexes of the contact cones, is larger in the O arrangement than in the X arrangement. The O arrangement therefore gives the lower tilting clearance.

When setting the axial internal clearance, thermal expansion must be taken into consideration. In the X arrangement, Figure 9, a temperature differential between the shaft and housing always leads to a reduction in the internal clearance (assuming the following preconditions: shaft and housing of identical material, inner ring and complete shaft at identical temperature, outer ring and complete housing at identical temperature).

In the O arrangement, a distinction is drawn between three cases:

• The roller cone apexes R, i.e. the intersection points of the extended outer ring raceway with the bearing axis, coincide: the internal clearance set is maintained, Figure 10 
• The roller cones overlap and there is a short distance between the bearings: the axial internal clearance is reduced, Figure 10 
• The roller cones do not meet and there is a large distance between the bearings: the axial internal clearance is increased, Figure 11.

Adjusted bearing arrangements can also be achieved by preloading using springs, Figure 12 . The elastic adjustment method compensates for thermal expansion. It can also be used where bearing arrangements are at risk of vibration while stationary.

The floating bearing arrangement is an economical solution where close axial guidance of the shaft is not required, Figure 13. The construction is similar to that of the adjusted bearing arrangement.

In the floating bearing arrangement, however, the shaft can be displaced in relation to the housing to the extent of the axial clearance s. The value s is defined as a function of the required guidance accuracy such that the bearings are not axially clamped even under unfavourable thermal conditions.

Suitable bearing types for the floating bearing arrangement include deep groove ball bearings, self-aligning ball bearings and spherical roller bearings.

In each of the bearings one ring, usually an outer ring, has a fit that allows displacement.

In floating bearing arrangements and cylindrical roller bearings with cage NJ, the length compensation takes place within the bearings. The inner and outer rings can have tight fits, Figure 13 .

Tapered roller bearings and angular contact ball bearings are not suitable for a floating bearing arrangement, since they must be adjusted in order to run correctly.

Rolling bearings are located on the shaft and in the housing in a radial, axial and tangential direction in accordance with their function. Radial and tangential location is normally achieved by force locking, i.e. by tight fits on the bearing rings. Axial location of the bearings is generally by geometrical locking.

The following must be taken into consideration in the selection of fits:

• The bearing rings must be well supported on their circumference in order to allow full utilisation of the load carrying capacity of the bearing
• The rings must not creep on their mating parts, otherwise the seats will be damaged
• One ring of the non-locating bearing must adapt to changes in the length of the shaft and housing and must therefore be capable of axial displacement
• The bearings must be easy to fit and dismantle.

Good support of the bearing rings on their circumference requires rigid seating. The requirement that rings must not creep on their mating parts also requires rigid seating. If non-separable bearings must be fitted and dismantled, a tight fit can only be achieved for one bearing ring.

In cylindrical roller bearings N, NU and needle roller bearings, both rings can have tight fits, since the length compensation takes place within the bearing and since the rings can be fitted separately.

With tight fits and a temperature differential between the inner and outer ring, the radial internal clearance of the bearing is reduced. This must be taken into consideration when selecting the internal clearance.

If materials other than cast iron or steel are used for the adjacent construction, the modulus of elasticity and the differing coefficients of thermal expansion of the materials must also be be taken into consideration to achieve rigid seating.

For aluminium housings, thin-walled housings and hollow shafts, a closer fit should be selected if necessary in order to achieve the same force locking as with cast iron, steel or solid shafts.

Higher loads, especially shocks, require a fit with larger interference and narrower geometrical tolerances.

Axial bearings, which support axial loads only, must not be guided radially (with the exception of axial cylindrical roller bearings which have a degree of freedom in the radial direction due to flat raceways). In the case of groove-shaped raceways this is not present and must be achieved by a loose seat for the stationary washer. A rigid seat is normally selected for the rotating washer.

Where axial bearings also support radial forces, such as in axial spherical roller bearings, fits should be selected in the same way as for radial bearings.

The contact surfaces of the mating parts must be perpendicular to the axis of rotation (axial runout tolerance to IT5 or better), in order to ensure uniform load distribution over all the rolling elements.

The conditions of rotation indicate the motion of one bearing ring with respect to the load direction and are expressed as either circumferential load or point load as shown in the table.

If the ring remains stationary relative to the load direction, there are no forces that displace the ring relative to its seating surface. This type of load is described as point load.

There is no risk that the seating surface will be damaged and a loose fit is possible.

If forces are present that displace the ring relative to its seating surface, every point on the raceway is subjected to load over the course of one revolution of the bearing. A load with this characteristic is described as a circumferential load.

As damage to the bearing seating surface can occur, a tight fit should be used.

The fit is determined by the ISO  tolerances for shafts and housings (ISO 286) in conjunction with the tolerances Δ_dmp for the bore and Δ_Dmp for the outside diameter of the bearings (DIN 620).

The ISO  tolerances are defined in the form of tolerance zones. They are determined by their position relative to the zero line (= tolerance position) and their size (= tolerance grade, see ISO 286). The tolerance position is indicated by letters (upper case for housings, lower case for shafts). For a schematic representation of the most common rolling bearing fits, see Figure 14.

The tables on pages link to link contain recommendations for the selection of shaft and housing tolerances that are valid for normal fitting and operating conditions.

Deviations are possible if particular requirements apply, for example in relation to running accuracy, smooth running or operating temperature. Increased running accuracies thus require closer tolerances such as tolerance grade 5 instead of 6. If the inner ring is warmer than the shaft during operation, the seating may loosen to an impermissible extent. A tighter fit must then be selected, for example m6 instead of k6.

In such cases, the question of fits can only be resolved by a compromise. The individual requirements must be weighed against each other and those fulfilled that give the best overall solution.

The numerical values for the fits (link to link) are valid for solid steel shafts and cast iron housings. In the table header, below the nominal diameters, are the normal tolerances for the bore or outside diameters of radial bearings (excluding tapered roller bearings). Below these are the deviations for the most important tolerance zones for fitting of rolling bearings.

In each cell are five numbers in accordance with the following scheme, for example for shaft ⌀40 j5:

Shaft fits: see tables from link.

In each cell are five numbers in accordance with the following scheme, for example for a housing ⌀100 K6:

Housing fits: see link to link.