Maximum stress in a cylindrical square drive 2

[random_guy]

Hi All,

I'm trying to determine the maximum stress of a circular part under torsion with a square drive in one side. It is essentially a socket adapter:

I can find the stress of a square shaft, and the stress of a hollow shaft, but a square hole in a circular shaft eludes me, and I do not wish to rely solely on FEA analysis.

I cannot find anything within Roark's, or any of my other references. Anyone have any similar experience or any ideas?

Thanks in advance.

Wise men learn more from fools, than fools do from the wise.

 

[dhengr]

Nsgoldberg:
The max. stress will be at the internal reentrant corners, left end of your photo. It will be a combination of torsional shear stress, concentrated and bending/winding around the reentrant corner; plus bearing stresses near that corner, for example, up on the inside of the upper right corner; plus tension reacting against the bearing load, down along the right side below that corner, but all at the min. section thickness. There have been a couple other threads about this same general stress issue, see if you can find them. This is a tough problem to try to do by hand, so in this case a good/accurate FEA model might be the more enlightening approach. Then see if you can explain the results long hand, with stress concentration factors, with superposition, and max. combined stresses. I doubt that you are going to find a treatise on this exact topic. And, I’ll bet the manufacturers do a lot of testing, and just thin them down until they start breaking them.

 

[desertfox]

Hi

As a very rough estimate, draw a circle inside your square so that the circle touches the sides of the square on all four sides, then do the hand calculation as you would for a hollow shaft, of course this will only provide a approximation of the average stress.

Desertfox

 

[random_guy]

desertfox, that's how I started out. But I don't believe it to be a very accurate representation.

dhengr, that is what I had been coming to conclude. Since I'm only making a handful of these, I don't have the resources to test. So FEA is going to have to do.

Thank you for your input!

Wise men learn more from fools, than fools do from the wise.

 

[rb1957]

Y? ... aren't there enough sockets in the world. if you're making a custom socket, aren't there enough examples to see what works (materials, geometries) ?

Quando Omni Flunkus Moritati

 

[MintJulep]

Gieck presents a solution for an arbitrary shape with non-uniform wall thickness based on the concept of a "median line".

I've got to think Roark has the same.

 

[random_guy]

rb1957, unfortunately no. It is similar, but not similar enough. And it is carrying a load nearly a factor higher than those commercially available, but in nearly the same size.

MintJulep, I do not have Gieck's text. I assume you are referring to this?:

http://books.google.com/books/about/Engineering_Formulas.html?id=oLw2MBxujRwC

I could not find anything in Roark's regarding a median line or diameter.

Wise men learn more from fools, than fools do from the wise.

 

[rb1957]

so i guess it's "just" like a socket drive, and not really a socket drive ? (else why'd you be driving so much more torque into a bolt head ?)

if you're FEAing, why not model a standard socket (as a benchmark) then remodel for the increased loads ?

if you are relying on FEA, the fit of the square drive into the square hole will have a significant effect on your results.

Quando Omni Flunkus Moritati