thread stress concentration in torsion

[Tmoose]

I found Plenty of references to bending and tension but nothing yet for torsion.
I'm not convinced that the factors for a single groove are correct when there are about 20 of them stacked along the shaft

These threads are for a bearing lock nut and are machined on a ~ 5 inch diameter shaft.

kind of like this, but the drive diameter is closer to the bearing size.

http://www.skf.com/images/cat/images/1/1_16/2300f26.gif

 

[rb1957]

you have a fatigue situation with cyclic torque loads ?

i'd've thought that the peak Kt was at the first thread, so number of threads is less critical; i'd've thought the key issue was the load transfer at the thread. Lord knows how this could be accurately known (as there are so many variables that would guessed at.

 

[rb1957]

how does the Kt due to torsion for a shoulder compare to the Kt due to tension ?

you could model the thread as a shoulder ... but then you're assuming a whole bunch about all those variables (fit, lube, ...)

 

[Tmoose]

"you have a fatigue situation with cyclic torque loads ?"

No, a supplier is asking for forgiveness in the 11th hour (again), when they should have asked for permission ( or better yet, read the danged drawing ). So we're stuck reviewing the design every which way. Our practice for decades has been to assume ~constant torque load and evaluate that portion of the shaft against a yield criterion with a few safety factors tacked on so only geometric stress concentrations are applied for keys and shaft transitions and the like.

 

[rb1957]

yeah, and so my question ... how much torque is in the bolt when the wrench is removed ? my answer (FWIW) is none ... torque was applied to extend the bolt.

i'm not sure why you need what you've asked for, or how you would apply it.

but i hear your pain ('bout people not reading drwgs) ... too often guys here just look at drwgs quickly (don't read the notes) "see" what's req'd and off they go ... when they go wrong, you have to fix it ! oh, yes, and the project weenies start in with their carping 'cause you're holding the project up !

 

[electricpete]

I'm thinking the torque of interest is the shaft-transmitted torque.

The stress from torque transmitted in a cylindrical shaft is highest on outside surface if memory serves me right. But it's hard for me to visualize that. And even harder to figure what happens when you put threads on that outer surface.





=====================================
(2B)+(2B)' ?

 

[Tmoose]

The geometrical stress concentration factors for a shaft with a groove in torsion are all over the place.
like page Fig C11 page 779 here -

http://www.mae.ncsu.edu/eischen/courses/mae316/docs/Appendix_C.pdf

The radius (r) at the bottom of a thread is undefined but at best tiny, and not well known, and in our case divided by 5 cause it's a 5 inch shaft, thus r/d is way to the left, and off all the curves. Granted our D/d is somewhere around 1.02, so on the graph the stress concentration might appear to end up between 2 and 3, but there is also the situation that there are 20 grooves side by side. Multiple entities (defense holes, stress foolers) tend to reduce the stress concentration of a single groove and hole situation for bending and tension loading ( curiously making it possible to make a part "stronger" be removing metal ) .

http://engj.org/index.php/ej/article/view/152/126

I need to make time to do a little FEA to back up the pronouncement we can use 100 k$ of shafts as-is.

Dan T

 

[Screwman1]

The root radius is not undefined, it's well defined. The difference in root radius in one of the primary differences between UNR and UNJ thread forms. Better check your thread specs.
I agree with RB that there will be no torsion on the threads once the wrench is taken off. Even in service the torque should not make it to the threads, the clamp load should be adequate to freeze the joint and prevent the threads from seeing service torque.