Lifetime of a torsional support? 2

[logbook]

I just had a disagreement with a colleague on a mechanical issue which frankly neither of us is competent to give a definitive opinion on. There is a torsional support for a mass, the mass being held at the top and bottom and rocked to and fro by some external force. The details are not important.

My colleague argues that provided the torsional support is not taken past its elastic limit it will have an essentially unlimited life. My argument is that even for small motions it will fail eventually due to fatigue or hysteresis. Obviously the support may eventually deteriorate due to decomposition of the support, but not withstanding that I would expect that after a few years of rocking backwards and forwards at a 50Hz rate it will break. I suspect that the support is some sort of rubber/plastic/elastomer but the same effect presumably occurs with torsional support wires.

Does anyone have any expert knowledge in this field to resolve the issue?

 

[btrueblood]

Your colleague is at least partially correct - as long as the material is steel, and operating (well) below its fatigue limit (which is somewhat correlated to the yield strength) it can survive in excess of 108 cycles without breaking. Some interpret the available data (for steel) to indicate theoretically infinite fatigue life is possible, for the most part most fatigue testing is halted after 107 to 108 cycles for economic reasons. For other materials, the existence of a "knee" in the fatigue life vs. stress level plot is not as evident as it is for steel.

At 50 Hz, operating 24/7, you'll get to 108 cycles in about 23 days. I know of a lot of pieces of equipment that have survived cyclic motion for much longer than that.

 

[btrueblood]

Whoops, those should have been 10⁷ and 10⁸; should have previewed that post.

 

[logbook]

The material is much more likely to be something soft like rubber or plastic. The rate is probably 5E8 cycles per year.

I know cam belts in cars break after 60,000 miles or so and I assumed that a similar failure mechanism would take place in a torsional vibration situation.

Thanks for the response. I have now googled on the idea of fatigue level vs stress plot and got some useful data.

It is surprising how a little push in the right direction can be such an asset. Thanks

 

[IRstuff]

Cam belts get affected by heat and atmosphere in addition to flexure. Both heat and atmosphere tend to cause stiffening of the rubber, which is why you see cracking on the belt.

TTFN

 

[btrueblood]

Also, those belts take a pretty high degree of bending strain, and carry significant tensile loads simultaneously, and both of these stresses are reversing/alternating. If you do the design right, the stress (esp. the alternating component of it) in the flexures can be (hopefully) kept very low. The lower the alternating component, the higher the fatigue life. It's a little bit funny, and maybe a little bit sad, that steel has a fatigue "knee" at 10⁷ or so cycles. This results in most fatigue testing, for steel and for any other material, to be stopped after that many cycles. Thus, the design life for a lot of common machinery is right about that point too (100,000 miles in a car going 55 mph is about 2000 hrs. 2000 hrs at 2000 rpm gives about 10⁸ cycles on engine components). I.e. does it seem like the endpoints of the fatigue data graphs limit our ability to design for longer life?