Fatigue testing of non-preloaded connections

[digfrack]

Hi Guys,

Our customer has requested us to perform a resonance fatigue test on a non-preloaded threaded collar type connection. My conecern is that if we do not preload the connection in excess of the cyclic fatigue loads it will fret, rattle, make a great noise and heat up very quickly (and possibly not resonate successfully). Does anyone have any experience with this type of testing.

Thanks,
Jon

 

[Tmoose]

In service is it installed without preload?

 

[rb1957]

in service, could the installed preload be relaxed ?

 

[digfrack]

They are not able to preload it in service which is why they want to test it. I have proposed to design a 4 point bending test rig and test the connection in this, but unfortunately they do not have a budget for that.

They do have locking mechansim which prevents the collar from unscrewing but it does not improve the fatigue performance.

 

[Screwman1]

Changing the preload is going to change the fatigue response greatly. Depending on the preload level and the mean stress, you could go from failure to no failure as preload increases. I know it is going to be a major PIA, but you need to duplicate the preload condition to get valid results.

 

[rb1957]

"They are not able to preload" ... the installation doesn't allow preload ... there is no preload.

it sounds like both you and your customer have the same sense that the fatigue performance of the connection may not but acceptable.

maybe it'll show obvious signs of distress before it breaks, allowing the imminent breakage to be prevented (by inspecting for the sound, etc).
but maybe this'll require a short (unacceptable) replacement interval.

 

[CoryPad]

Thread adhesive or thermoplastic pellet/patch would be options for this application.

 

[Tmoose]

I'm guessing they break in service. is the break consistently in one location, for instances, in the male thread where it blends into the main thing, or the male thread where engagement with the internal thread begins, or maybe even not in the male thread at all?
testing with the "problem" section in an artificial state won't prove much


A company I worked for made machine tool drawbars made of some fairly exotic steel with a male thread the engaged an internally threaded sleeve that was rotated to achieve some set length then secured with a torqued jam nut. Every time the tool was engaged a portion of the thread on the drawbar was subjected to a tensile load. After a few zillion clamp/unclamp cycles the drawbar would snap in the transition between thread and no thread. The shape of the thread transition was modified to include a large radius and a necked down section. The thread profile was modified to have a genuine radius at the root, and the entire thing was shotpeened to that old MIL spec. I don't think they ever broke another one.
Just Another case where (bad) Geometry beat great metallurgy