A fan bearing assembly that kept failing every nine months. The fan was a 37 kW centrifugal unit running at 1480 RPM. The bearings were 6314 deep groove, mounted in a split pillow block housing. The failure was always the same: the outer ring had rotated in the housing, the housing bore was polished to a mirror finish, and the bearing had overheated. The bearing itself was fine — the races were still good. The housing was the problem.

The technical name for the outer ring rotating in the housing is bearing creep. It’s different from slip. Slip happens at the contact between the rolling elements and the raceway. Creep happens between the ring and its seat. The outer ring creeps when the interference fit between the ring and the housing is too loose.

Why the outer ring shouldn’t move

In a standard fan arrangement, the outer ring is a press fit in the housing and the inner ring is a slip fit on the shaft. The shaft rotates, the inner ring rotates with it, the rolling elements roll between the races, and the outer ring stays stationary in the housing. That’s the design. The outer ring only stays stationary because the interference fit generates enough friction to hold it against the tangential force from the rolling elements.

That tangential force is small. It’s roughly 5% of the radial load, acting at the pitch diameter. For a 6314 bearing under 12 kN radial load, the tangential force is about 600 N. The friction capacity of the fit needs to be higher than that. If the fit is too loose, the ring moves.

What the fit actually was

The housing was a standard split pillow block with an H7 bore. The bearing outer ring was a nominal 150 mm. The H7 tolerance for 150 mm is +0.040 / 0 mm. That means the bore could be up to 0.040 mm larger than the ring. With zero interference, the ring sits loose. It can rotate. The measured bore was 150.035 mm — close to the maximum. The ring had about 0.03 mm clearance. That’s not a press fit. That’s a loose fit.

The bearing catalog says the outer ring should be mounted with an interference of 0.01 to 0.03 mm for this load. The housing provided zero interference. The ring crept. Every nine months, the ring rotated a few degrees, wore the bore, and the bearing ran hotter. The plant replaced the bearing every time without checking the housing bore. Nine months later, same failure.

The fix

Two options. The right one for a pillow block: bore the housing to a controlled interference. The pillow block was machined to accept a 150 mm ring with 0.02 mm interference. That means the bore was cut to 149.98 mm. The ring pressed in with a light press. It hasn’t moved since.

The second option, used when the housing can’t be machined: apply Loctite 620 (a high-temperature retaining compound) to the outer ring seat. The compound fills the clearance gap. It cures to a solid filler that holds the ring. This works when the clearance is under 0.1 mm. Above that, the compound layer is too thick and it squeezes out. For the 0.03 mm clearance in this housing, the Loctite fix would have worked. We machined instead, because the housing was off the machine anyway.

The inspection checklist for repeated bearing failure

Before replacing the same bearing a third time, check the seats. A bearing that fails identically every time is telling you something about the mounting, not the bearing.

  • Outer ring seat: measure the housing bore. If it’s bigger than the ring OD, the ring can creep. Fix the bore.
  • Inner ring seat: measure the shaft. If the shaft is worn (smaller than the ring ID), the inner ring spins on the shaft and wears it. A shaft that’s scored under the inner ring seat is a classic failure.
  • The bearing itself: after removal, check if the outer ring has a polished band. That’s creep evidence. Check if the rolling elements have a wear pattern from misalignment.
  • The housing: check for fretting corrosion (a reddish-brown dust) at the ring seat. Fretting means the ring has been micro-moving for a long time.

The plant had all the evidence: polished bore, fretting dust, overheated bearing. They just didn’t look at the housing until we pulled the micrometer out.

When creep is actually normal

Some applications deliberately run with the outer ring loose in the housing. Cam followers and some idler pulleys are designed to creep. The ring rotates slowly in the housing so the load zone moves around the ring. This distributes the wear. It’s called “self-aligning creep” and it’s intentional. The difference: in those designs, the housing is hardened and the creep is slow and controlled. In a pillow block with a soft cast iron housing, creep is never acceptable. The cast iron wears, the clearance grows, and the creep accelerates.

Bearing creep is a fit problem, not a bearing problem. If the outer ring moves in the housing, measure the bore and fix the interference. 0.01 to 0.03 mm interference for a 150 mm ring under fan loads. If you can’t machine, use retaining compound. And check the fretting dust — it’s the evidence the ring has been moving for months.