Selection & Sizing

Shaft bore and keyway

Torque gets from a shaft into a hub by exactly two mechanisms: a key working in shear and bearing, or friction generated by an interference fit. A great many drives use both at once without anyone having decided which is meant to be carrying the load — and that ambiguity is what fails.

Author
Priyansh Thummar, Editor
Dates
Published · Last updated
Reading time
5 minutes

1. Two mechanisms, one joint

A key works in shear and in bearing against the sides of two keyways. An interference fit generates friction across the whole bore surface. Those are the only two ways torque reaches a hub.

A great many industrial drives use both at once without anyone having decided which is meant to be carrying the load — and that ambiguity is what fails, rather than either mechanism on its own.

2. The three fits and what each is for

Table 1 — Fit selection by the character of the load, not its magnitude
FitWhat carries the torqueSuitsCost of choosing it
ClearanceThe key, in shear and bearingUnidirectional duty, modest shock, frequent removalFretts under reversal; fit loosens progressively
TransitionKey, with some friction assistanceMore location than clearance, still removable coldNeither mechanism fully committed
InterferenceFriction over the whole bore surfaceReversing, shock, torsional vibrationHeat to mount and remove; not adjustable once cooled
KeylessFriction only — no keyway at allHigh-torque reversing; keyway is the limiting featureTighter geometry control, hydraulic mounting equipment
The deciding question is whether the load cycles, not how large it is. A modest torque that reverses is harder on a keyed joint than a large torque that does not.

A keyed clearance fit unloads and reloads the same key flank on every reversal. Do that a few million times and the joint fretts, the fit loosens, and the loosening accelerates the fretting.

3. The keyway tolerance stack

Three dimensions rather than one, which is what catches people out. The shaft keyway has a width and a depth tolerance. The key has its own. The hub keyway has a third. Each is individually reasonable and they accumulate.

So a key that is nominally correct can arrive with a fit that is loose on the sides — exactly where it must not be, because the sides are the working faces. Hence the convention: clearance belongs at the top of the keyway, between key and the roof of the hub slot, never on the flanks.

3.1 Why the keyway limits the bore

The constraint most often missed. Every coupling size has a maximum bore set by the material remaining between the bore surface and the root of the hub teeth — and the keyway removes material from that same wall.

A bore near the published maximum, combined with a deeper-than-standard keyway, can leave a wall too thin to carry hoop stress even though the nominal shaft diameter is within the limit. Check the bore against the keyway actually specified, not against the diameter alone — this is why bore governs so many selections.

4. When interference stops being optional

At the point where the duty cycles. That is a more useful criterion than a torque threshold. It covers:

  • Reversing mill drives
  • Reciprocating machinery
  • Engine drives, with a firing pulse every cycle
  • Any drive where a coupling has already been replaced once for a fretted bore
  • Any joint where the hub turning on the shaft is unacceptable, regardless of calculated load

It is not free. Mounting needs controlled heating and a hub driven fully home in one movement, because it cannot be adjusted once cooled. Removal needs heat again. Hoop stress rises with the interference, so the fit cannot simply be increased until the problem goes away. And a hub mounted short leaves a coupling axially wrong from the start, with no correction that does not involve heating it off again.

5. Keyless, and what it removes

A keyless fit removes the keyway rather than the clearance, and that is a different benefit. The keyway is a stress raiser at precisely the point where fretting tends to initiate a fatigue crack, so on high-torque reversing drives it is frequently the limiting feature of the whole joint rather than the fit around it.

A keyless tapered or straight bore transmits torque entirely by friction and has no slot to concentrate stress. It costs more to produce, demands tighter surface finish and geometry control, and needs hydraulic mounting equipment — specified where the duty justifies it, not as a general upgrade.

6. Specifying it without ambiguity

Four lines on a drawing. Most disputes come from one of them being absent.

  1. Nominal bore diameter and the fit class or ISO designation.
  2. The standard those come from. An H7 and an inch-series clearance class are different tolerance grammars, and neither is a rounding of the other.
  3. Keyway dimensions and their standard.
  4. Explicitly, whether an interference fit is required and what is expected to carry the torque.

A drawing that mixes an inch bore call-out with a metric key will be manufactured exactly as drawn, and it will not assemble.

Frequently asked

Should a coupling hub be a clearance fit or an interference fit?
Clearance with a parallel key is the default across general industrial service: quick to fit, quick to remove, and forgiving of a shaft that has seen wear. Interference is required where torque reverses, where the drive is subject to shock, or where torsional vibration is present — because a keyed clearance fit unloads and reloads the same key flank on every cycle and eventually fretts. The deciding factor is the character of the load, not its magnitude.
Why does the keyway limit the maximum bore of a coupling?
Because the keyway removes material from the same wall that the bore limit exists to protect — the material between the bore surface and the root of the hub teeth. A bore near the size's maximum with a deeper-than-standard keyway can leave a wall too thin to carry hoop stress, even when the nominal shaft diameter is within the published limit. Check the bore against the keyway depth actually specified, not against the diameter alone.
Can a hub have both a key and an interference fit?
Very commonly, and it is worth being explicit about why. With a genuine interference fit the friction carries the torque and the key becomes a safety device or an anti-rotation locator rather than the load path. The problem arises when the fit is nominally interference but insufficient — then the key carries the load it was not sized for, in a joint that can still micro-slip. Decide which mechanism is carrying the torque and specify for that one.
When is a keyless fit worth the cost?
When the keyway itself is the problem rather than the fit. A keyway is a stress raiser at exactly the point where fretting damage tends to initiate a fatigue crack, and on high-torque reversing drives it is often the limiting feature of the whole joint. A keyless tapered or straight bore removes it entirely and transmits torque purely by friction. It costs more to produce, demands tighter surface and geometry control, and requires hydraulic mounting equipment — so it is specified where the duty justifies it rather than as a general upgrade.

References

  1. ANSI/AGMA 9002-B04, Bores and Keyways for Flexible Couplings (Inch Series), for the inch-series bore classes and keyway practice.
  2. ANSI/AGMA 9003-C17, Flexible Couplings — Keyless Fits, for the keyless option in §5, including its mounting pressure and allowable stress calculations.
  3. ISO 286-1:2010, Geometrical product specifications (GPS) — ISO code system for tolerances, for the metric limits-and-fits grammar behind an H7 bore.
  4. DIN 6885, Parallel keys and their keyways, for the key dimensions in the tolerance stack of §3.
  5. Editorial note: no numeric tolerances, interference values or maximum bores are published here. Every one of them is a property of a specific size, material and standard revision, and a figure lifted out of context is the error §6 exists to prevent. Per-size maximum bores, keyway dimensions and recommended interference for the Super Mech Industries range are to be added from catalogue data and confirmed at technical review.