Round Key vs. Square Key for disk/shaft fit

[poetix99]

(This will be posted identically on Mech. Engr'g (#404) and also on Gears and Pulleys forum.)

I am trying to get any design references about the use of shaft keys with a round cross-section.

We are now trying to analyze this configuration because the original design work is 30+ years old; anything that was written down (if anything WAS written down) is gone. We are pushing the envelope of our known applications, and would like to better know where we're going.

It seems clear that the original rationale was the idea of a lower stress concentration that one would obtain with a semi-circular key slot in the disk. I have questions (suspicions?) about torque capacity and any tendency for the disk to want to "cam" over the key as it transmits torque to the shaft.

We use a fairly heavy shrink fit in addition to the key, so the shrink is really intended primarily for torque transmission. The key is, in a meaningful sense, a back-up.

There are also some thermally induced (hoop) stresses in the disk during normal operation.

Any written design calculations have long since disappeared, and we are re-inventing the wheel, so to speak.

Thanks in advance.

 

[gbent]

I am somewhat vague on the actual configuration you are describing. Is this disk mounted on the end of the shaft, and is the round key installed axially at the split line between the hub and shaft?

If this is the situation, was the key installed after assembly? If so, it is probably what I have always refered to as a "blind dutchman" (blind because you can only get to one side). It could be drilled and reamed with hand tools at assembly, and was cheaper than machining proper keyways during manufacture. In your situation, it was probably used because aligning keyways on heavy press fits is almost impossible. It is also used as a repair for fits that have loosened. The press on the round key tightens up the fit and locks up rotation.

As you said in your post, it is only a backup for a shrink fit that should be sufficient by itself.

 

[poetix99]

Thanks, "gbent".

It is good to get some feedback in order to know what is not clear.

The disks are mounted along the length of the shaft. The key slots are ball-end milled into the shaft. The mating disk slot can be broached, or machined in some other (conventional) way. As I've implied above, there may be several disks on a shaft. The round keys aren't improvised repairs; they're as-designed, but the original design documentation does not exist. The shrink is not a press fit; it is a "shrink", with heated disk and/or cooled shaft. Therefore, keyway alignment is not a particular issue. Torque is developed on the disk(s) rim and transmitted to the shaft.

I'd like to relax the shrink that we use and therefore would impose some torque transmitting "responsibilities" upon the round key.

Therefore, I would like to know of any application experience and/or design consideration that you folks might be willing to share.

Thanks.

 

[gbent]

I took a WAG on the situation, and missed by a mile. The keyway you describe is a lot of work to manufacture, so I would guess they went to that design after either testing or field failures.

The round keyway lowers stress concentrations in the shaft a large amount. Someone seems to have borrowed my machine design book, so I can't look up the exact amount. Another WAG would be this is a relatively slender shaft, with relatively high stress loads, and tended to have fatigue failures due to stress concentrations. The round keyways would have eliminated much of the stress concentrations without requiring a complete redesign of the mechanism.

What are the loads on the shaft? They can be from driven load, input power fluctuations (such as 1 or 2 cylinder engine), inertia of the disks, misalignment, etc. For a redesign, you need to start with the loading and work from there. You can use existing design and any field failures to get a baseline.