Key takeaways
Pull a year of failed bearings from a plant and a striking number carry marks made before the machine ever turned, such as brinelling dents at ball pitch on a stationary ring. They were damaged during mounting, then ran on borrowed time until the damage propagated into spalling; the initiating defect is invisible on the day and undetectable in a hand-turn check.
So bearing replacement needs a written method with recorded measurements: mounting steps inside a structured preventive maintenance plan, with fields for measured shaft diameter and mounting temperature, get repeated the same way.
Work on a clean bench with lint-free wipes, never blow the seat down with shop air, and keep the bearing box closed until the shaft is ready. Before it opens:
A plant working to an ISO 4406 oil cleanliness code for its circulating systems should hold the assembly bench to the same standard.
The ring that rotates relative to the load direction needs an interference fit; the ring stationary relative to the load can take a looser fit. With a rotating shaft and a stationary radial load: interference on the inner ring, a slightly loose to transition fit on the outer ring.
Typical shaft designations are j6 for light loads, k5 or k6 for normal industrial duty, and m5 or m6 for heavy or shock loading; housings are commonly H7, or J7 where the outer ring must not creep. A tighter fit resists creep and fretting but costs mounting force and clearance.
Pressing an inner ring onto an oversized seat stretches it: the bore grows, the raceway with it, and the space the rolling elements had shrinks. About 80 percent of the interference appears as lost radial internal clearance on a solid steel shaft, so 25 micrometres of interference spends roughly 20 micrometres of clearance before the machine has turned once.
Fit and clearance class are therefore one decision: a tight fit with a standard CN clearance bearing on a hot application can reach negative operating clearance, meaning preload, friction, heat and short life. Choosing C3 or C4 over CN is covered in the companion piece on bearing internal clearance classes.
Cold mounting suits small bearings, roughly up to 60 to 70 mm bore with modest interference. Use a fitting tool whose impact ring contacts only the ring being fitted: on a shaft fit, the inner ring alone. A sleeve bearing on both ring faces is correct only where the bearing is driven onto the shaft and into the housing in the same stroke, so that each ring receives its own force directly. Drive it with a dead-blow hammer or, better, a press, which adds feedback a hammer cannot: force still rising after the ring should have seated means something is misaligned. Mounting force must never cross the rolling elements.
Heat mounting is normal practice from roughly 70 mm bore upward, and below that where the interference exceeds what a press can deliver. Use an induction heater with a temperature probe, or a thermostatically controlled oil bath. Heat the inner ring to 80 to 90 K above the shaft temperature, which from a 20 C workshop means about 100 to 110 C. Standard bearings are dimensionally stabilised to 120 C, so treat 120 C as the hard ceiling and 110 C as the working target: above the ceiling you risk dimensional change in the rings, and heat degrades the grease in prelubricated units well before that. Never use a flame, and never heat a sealed or greased bearing in an oil bath.
Tapered seats take an adapter sleeve on a cylindrical shaft, or the shaft's own taper, driven up with a locknut or a hydraulic nut; a withdrawal sleeve is the removal counterpart, not a mounting device. The control variable is drive-up distance or the reduction in measured internal clearance, never torque: interference is set by how far the ring goes up the taper.
At mounting, fill the bearing cavity completely and the free housing space to roughly 30 to 50 percent for normal speeds. Overfilling a high-speed bearing is a classic cause of a first-day temperature spike, because churned grease generates heat faster than it can be shed. Record the initial fill, which is not the running refill quantity. Intervals are covered in the guidance on bearing relubrication intervals. Execution belongs on scheduled lubrication routes.
Once mounted, turn the shaft by hand for roughness or a cyclic catch, then take vibration and temperature baselines in the first hour, at four hours and at 24 hours, reading the vibration figure against the ISO 10816-3 vibration severity zones for the machine class rather than against a number remembered from another asset. A rise over one to two hours that peaks and falls to a stable plateau is grease redistributing. One still climbing at hour four, or that never plateaus, means insufficient operating clearance, excess grease or a cocked ring: stop and investigate rather than running it in. Judge the plateau against this machine's baseline or the manufacturer limit, since many motor and fan housings sit steadily 30 to 40 K above ambient.
Fabrico is where this record lives: recurring PM tasks carry the mounting and inspection checklists, measured values and results are stored against the asset so the next technician sees what the shaft measured last time and each asset carries its own record an auditor can open, a failed check can trigger a follow-up task, and machine stop data from PLC or OEE can become a work order. To see it against your asset list, book a short demo.
A plant replaces the drive-end bearing on a 75 kW centrifugal fan motor: a 6312 deep groove ball bearing, 60 mm bore, 130 mm outside diameter, C3 clearance, on a shaft drawn 60 k5, running continuously with the inner ring near 70 C and the outer ring near 50 C.
Step 1, measurement. The shaft seat measures a mean diameter of 60.014 mm over three positions. The k5 band for this size is plus 2 to plus 15 micrometres: in tolerance, but near the top. The bearing bore measures 59.996 mm, 4 micrometres below nominal, inside the Normal class mean bore deviation of 0 to minus 15 micrometres for a bore over 50 up to 80 mm (that tolerance applies to the mean bore diameter; out-of-roundness is limited separately, which is why the bore is measured in two orientations).
Step 2, actual interference. Interference equals shaft diameter minus bore diameter, less the flattening of surface asperities during mounting: 60.014 minus 59.996 equals 0.018 mm, that is 18 micrometres of measured interference, a micrometre or two less in effect on a ground seat, more on a turned one. A shaft at the bottom of the k5 band with a bore at nominal would have given only 2 micrometres, and a shaft at the top of the band against a bore 15 micrometres under nominal would have given 30: the drawing alone leaves a fifteen-fold uncertainty.
Step 3, clearance lost to the fit. Applying the 80 percent rule: 18 times 0.8 equals 14.4 micrometres of radial internal clearance lost at mounting.
Step 4, starting clearance. This C3 bearing measures 38 micrometres of radial internal clearance before mounting; after mounting, 38 minus 14.4 equals 23.6 micrometres.
Step 5, thermal effect. The inner ring runs 20 K hotter than the outer ring. That ring-to-ring difference is what the closure formula needs; a housing surface reading is lower than the outer ring behind it, so do not use it directly. Closure equals the mean bearing diameter, 0.5 times (60 plus 130) equals 95 mm, times the coefficient of expansion times the temperature difference: 95 mm times 0.0000115 per K times 20 K, which is 0.0219 mm. Radial internal clearance is a total-play figure, so the whole of that differential growth is lost: roughly 22 micrometres.
Step 6, verdict. Operating clearance is approximately 23.6 minus 22, so roughly 2 micrometres, positive but only just: a shaft at the top of the k5 band, 60.015 mm, against a bearing bore at the bottom of its Normal class band, 59.985 mm, would give 30 micrometres of interference, 24 micrometres lost at mounting and an operating clearance of 38 minus 24 minus 22, that is minus 8 micrometres: preload before the machine has turned. Decision: 2 micrometres is smaller than the uncertainty in the inputs that produced it, so treat it as zero. Record the measured 60.014 mm and fit the C4 variant: a C4 unit measuring 45 micrometres leaves 45 minus 14.4 minus 22, about 9 micrometres of operating clearance, which is a real margin rather than a rounding artefact. Keep the C3 only for a duty where the ring-to-ring difference is verified below about 15 K. Heat it on an induction heater to 100 C: 60 mm times 0.0000115 times 80 K is roughly 55 micrometres of bore growth, ample to push on against the shoulder. Had the shaft carried a standard CN bearing measuring 15 micrometres of clearance, the arithmetic would read 15 minus 14.4 minus 22, deeply negative: hard preload. Decision: do not fit; obtain the larger clearance variant.
Yes, through a fitting tool, on small bearings with light interference only. For a shaft fit the tool must bear on the inner ring alone, since that is the ring being driven; a sleeve contacting both ring faces belongs only to the case where the bearing enters the shaft and the housing in one stroke. Never strike a ring face directly, and never strike one ring while the other reacts the force: that path runs through the rolling elements and dents the raceway.
Aim for 80 to 90 K above the shaft temperature, which in a workshop at 20 C means roughly 100 to 110 C. Treat 120 C as a firm ceiling for standard bearings: beyond it you risk dimensional instability in the rings. Use an induction heater with a probe or a thermostatically controlled oil bath, and never heat a sealed or prelubricated bearing in oil.
Subtract roughly 80 percent of your measured interference from the bearing's measured radial internal clearance, then subtract the thermal closure from the inner ring running hotter than the outer. If the remainder approaches or crosses zero, you need a larger clearance class, not a looser fit.
In practice no, for any bearing pressed or pulled off a shaft: dismounting loads it in a direction it was not designed for, and the seat interference is already worn. The one defensible reuse is a bearing removed intact from a tapered adapter sleeve with a hydraulic tool, refitted with a fresh clearance measurement.
For small standard bearings on well-controlled seats the catalogue range is enough. Measure the actual clearance when the bearing is large, the fit tight, the temperature differential significant, or a previous unit in the same position failed early: a feeler gauge for cylindrical and spherical roller bearings, a dial indicator under reversed load for deep groove ball bearings, where a feeler gauge cannot reach the contact.
Login
Blog