Keys, fits and getting torque in

This is a static copy of the chapter for search engines. The interactive version has animated figures, check questions, and buttons that load the example into the calculator.

The torque has to get from the hub into the shaft somehow. The parallel key is the most common answer, and it is worth knowing that you do not design the key section — the standard picks it from the shaft diameter. What you design is its length.

The key section comes straight from the shaft diameter (DIN 6885) — you do not choose it. What you choose is the length, and that is what the two checks decide.

The two checks

crushing: p = 2Td · l_eff · k shear: τ = 2Td · l_eff · b

with k = min(h − t₁, t₂), the key height that is actually in contact with the hub. Typical allowables for a steel hub under a steady load are p ≤ 100–150 MPa and τ ≤ 90 MPa; halve them for shock or reversing loads. In practice crushing almost always governs — the key is far more likely to be hammered into the keyway sides than sheared off.

Effective length

A round-ended (type A) key loses one key width of useful length, so l_eff = l − b. A square-ended (type B) key uses its whole length. And a key can never be longer than the hub.

If the key does not fit the torque

Make the hub longer — the cheapest fix, up to about l = 1.5d, after which the far end of the key stops carrying much.
Two keys at 120° — assume they share only about 75 % ideally, i.e. count on 1.5 keys, not 2.
Splines — many small teeth, far more contact area, no notch as sharp as a keyway, and self-centring. The right answer for high torque.
A keyless connection — shrink disc, taper bush, polygon or press fit. No keyway means no K_t = 2.14 notch, which often lets you use a smaller shaft overall.

Press fits

An interference fit transmits torque by friction and needs no notch cut into the shaft. But it is not free of stress concentration: the shaft flexes under the rigid edge of the hub and frets against it, which gives K_f ≈ 2 at the end of the seat. Chamfering or relieving the hub bore at the ends helps a lot.

⚠ Do not put the end of a press-fit seat in the same place as the peak bending moment. A gear seated with an interference fit right at mid-span puts a K_f = 2 notch exactly where M is largest.

💡 The keyway also removes material — but much less than people expect. A standard keyway on a 40 mm shaft takes about 8 % off the section modulus. It is the notch factor of 2.14, not the lost material, that hurts. ShaftLab accounts for the notch; the small section loss is inside the conservatism of the K_t value.