Keys and Keyseats

§1The sacrificial link

A key does three jobs at once — transmits the torque, indexes the hub’s angular position, and, sized with intent, volunteers as the drive’s cheapest casualty.

The torque path from shaft to pulley, gear or coupling flange passes through the key’s flanks, so every newton-metre of the drive crosses this one small block — job one, and the arithmetic of §2. Job two is timing: the key fixes the hub’s clocking on the shaft, which is why cam sprockets, impellers and anything phase-critical is keyed even when friction alone could carry the torque. Job three completes the previous page’s fuse philosophy: a key is cheap, accessible and replaceable in minutes, so it is often deliberately the weakest member — a jam shears a five-dollar key instead of twisting a shaft or stripping gear teeth, exactly as the shear pin was placed to do, and “sheared its key” is among the happiest failure reports a drive can file. All three jobs assume one geometric fact worth stating plainly: standard parallel keys drive on their sides — the flanks bear, the top runs clearance — so the working faces are the narrow strips §2 now checks, and the whole component tradition (sizes, fits, even its failure etiquette) is organised around those flanks.

§2The parallel key, sized

Two checks size a key — shear across its waist, crushing on its flank — and because the flank is the smaller face at the lower allowable, crushing nearly always writes the length.

§3The family

Around the plain parallel key stand four relatives, each answering one situation the standard block handles badly.

§4The keyseat’s price

The slot that houses the key is a notch machined into the shaft’s most stressed skin — and a shaft’s fatigue account pays for it at roughly double the nominal stress.

Torsional shear in a shaft peaks at the surface, and the keyseat cuts two sharp internal corners exactly there: a standard profiled (end-milled) seat concentrates the local stress by a factor of around two, and every fatigue lesson of the shafts and fasteners pages applies — under reversing or fluctuating torque, cracks in keyed shafts start, with monotonous reliability, in a keyseat corner or at its abrupt end. The mitigations are all geometric. Radius the corners: standards specify a fillet in the seat’s floor corners, and the matching chamfer on the key that lets it seat over the fillet — a filed-square “fitted” key that forces the corner sharp undoes the standard’s protection. Soften the run-out: a sled-runner seat, milled in on a ramp, ends gently where a profiled seat ends in a stress-raising step — worth specifying on fatigue-critical shafts even at the cost of a longer slot. And place the seat kindly: never let it run through a shoulder fillet or coincide with a bending hot-spot, because stress concentrations multiply where features overlap. The keyseat is a permanent toll on the shaft; §2 sized the key, but this section is why the shaft around a keyed connection is often the member the calculation really protects.

§5Fitting practice

A keyed joint is one of the last routinely hand-fitted connections in machinery, and the fitting has a definite standard: snug on the flanks, clear on top, and no rock anywhere.

The working fit is side-tight: the key should enter its seats with firm hand or light tap pressure and sit without shake, because §1’s flanks can only share load they actually touch — a slack key lets the hub accelerate across the clearance and hammer the flanks at every torque reversal, opening the wallowing spiral §6 describes. Fitting is therefore filing and trying: dress the key (never the seats) to a sliding-tight fit, ease its edges with the chamfer §4 required, and confirm the hub seats fully with the key in place — a key standing proud of a shallow hub seat cocks the hub and fakes a tight fit. The top face runs clearance; it is not a bearing surface, and a key fitted tight on depth simply jacks the hub eccentric like §3’s taper key. Retention completes the job: a parallel key holds nothing axially, so the hub is located by its own means — circlip, shoulder, or the classic set screw bearing on the key’s flat, which pins key and hub together without bruising the shaft. What a set screw must never be asked to do is carry the drive alone on a bare round shaft: that is not a keyed joint but a slipping one with delusions, and every worn, spun, screw-scarred shaft in the repair pile says so.

§6When a key is the wrong answer

Keys assume torque that mostly pushes one way. Give them full reversals, impacts or precision-balance duty, and the honest move is a different connection entirely.

The keyed joint’s characteristic decline is mechanical, gradual and self-accelerating: reversing torque works the flanks across their fit clearance, fretting polishes and pits the contact, the seat wallows wider, the impacts grow with the growing clearance — and a joint that began snug ends as a hammer mill, destroying seat, key and finally hub. Where that duty is the duty, the alternatives take over. An interference or shrink fit drives by friction over the whole bore — no notch, no backlash, the fatigue-cleanest connection there is, at the price of assembly by heat or press and a fight at removal. Taper-lock and QD bushings package the same friction grip in a splittable, field-serviceable form — the pulley trade’s standard answer. Clamp and shrink-disc elements do it externally with a ring of screws, keyless and infinitely re-clockable. And when the torque is simply beyond any single block’s flanks, the spline — §3’s “many keys at once” — spreads the load around the full circumference, as its own page in this library details. The parallel key remains the right default for one-directional, moderate, serviceable drives — which is most of machinery — and the mark of judgement is knowing the day the default expires.

§7Quick reference

The working core of the page on one card rack.