A keyway that fretted on a 30 mm shaft. The coupling was a rigid flange, driving a conveyor pulley. The shaft was keyed with a 8×7 mm square key. After 6 months, the keyway wall showed fretting cracks — fine red powder between the key and the hub bore. The torque was well within the key’s capacity. The issue wasn’t torque — it was the key clearance. The key was a loose fit in the keyway. Under reversing torque (conveyor starts and stops), the key rocked in the slot. The fretting wear followed. This is about key fit and shaft-hub connections.
The three key fits
A square key in a keyway can have three fits:
-
Clearance: the key slides into the keyway by hand. There’s 0.05-0.1 mm clearance on the sides. Under load reversal, the key rocks. Fretting and wear follow. Use only for steady, unidirectional torque.
-
Transition: the key needs a light press to install. The key has 0-0.02 mm interference. It stays centered. Under reversal, the key doesn’t rock. This is the standard fit for most power transmission.
-
Interference: the key needs a hydraulic press or a cold-shrink installation. 0.05-0.1 mm interference. The key is locked. Used for heavy shock loads. But it can crush the keyway if the hub is thin.
The failed keyway had a clearance fit. The 8×7 mm key was 7.95 mm wide in a 8.00 mm keyway. The 0.05 mm clearance let the key rock. Every conveyor start (shock load) slammed the key from one side to the other. After 6 months, the fretting started.
What I changed
1. Re-machined the keyway to transition fit. I bored the keyway to 8.00 mm (using a keyway broach). The new key was 8.02 mm (oversize by 0.02 mm). It pressed in with a soft hammer. No rocking. No fretting. The conveyor has run 2 years without issue.
2. Added a split-hub clamping coupling. For new designs, I use a split-hub clamping coupling (like a Ruland or Zero-Max). The coupling has two clamping screws that squeeze the hub onto the shaft. No key needed. The friction transmits torque. There’s zero backlash and zero fretting. The split hub costs 3x a keyed coupling but eliminates keyway failures entirely.
3. Used a Woodruff key for small shafts. For shafts under 20 mm, I use Woodruff keys (half-moon). The Woodruff key seats itself in a semicircular pocket. It can’t rock. It can’t axially slide. It’s self-aligning. For small pulleys and pinions, the Woodruff key is the default. The downside: the keyway is weaker (no full depth). But for small torques, it’s fine.
The key length rule
A key is sized for the torque, but the length must be checked. The shearing area is key_width × key_length. For an 8 mm wide key in a 30 mm shaft: recommended key length is 1.5 × shaft diameter = 45 mm. The shearing area is 8 × 45 = 360 mm². At 100 MPa shear stress: torque capacity = 360 × 100 × 0.015 (shaft radius) = 540 N·m. That’s enough for a 5 kW motor at 1000 RPM (48 N·m). But if the hub is only 30 mm wide, the key is 30 mm long. Capacity drops to 360 N·m. Still fine. But the key must not exceed 2/3 of the hub width — otherwise it protrudes and interferes with the bearing.
Key material
Standard keys are C1018 cold-drawn steel (yield 350 MPa). For heavy loads, I use 4140 alloy steel (yield 650 MPa). The key should be softer than the shaft and hub — it’s the sacrificial element. If the key shears, you replace a $5 key instead of a $500 shaft. Always use a key that’s weaker than the shaft. A hardened key that doesn’t shear destroys the shaft keyway.
The fit I require: transition (0.02 mm interference) for square keys. The fretted keyway wasn’t overloaded — it was a loose clearance fit that rocked under reversing torque. For new designs, use a split-hub clamp instead of a key. Use the key as the sacrificial element: softer than shaft and hub.