How come BPVC S-N curves show stress amplitude greater than UTS

[owenhale]

Hi there,

I've been racking my brain on this for a while so it's time to seek some advice from others in the field. I'm getting up to speed with using the BPVC fatigue calculations in BPVC Section VIII Div 2. One thing I can't reconcile is how come the S-N curves in the BPVC show a stress amplitude that's greater than the UTS of the material? I would expect that at low cycles, the curve would be capped at the UTS but that doesn't seem to be the case. My first thought was it that maybe the stress amplitude was defined peak to peak for these curves but from what I read that doesn't seem to be the case. What am I missing here?

Thanks,
Owen

 

[TGS4]

These stresses are pseudo-elastic stresses. That means that they are stresses that exceed yield, based on an elastic analysis/calculation. In reality, the component will experience plastic deformation, that could have a plastic strain that could be on the order of the true ultimate plastic strain (or even higher, depending on the magnitude of the compressive triaxiality (although that can the topic for a different post)). But take a typical carbon steel - the true ultimate plastic stain could be on the order of 40%. Multiply that by E, and you could have these pseudo-elastic stresses on the order of 12E6 psi.

Another way to state this is that although it is plotted as an stress-life curve, it’s really a strain-life curve multiplied by a standard value of Young‘s Modulus. And of course shifted downwards and to the left to account for mean stress effects and for the design margins.

Does that make sense?

 

[owenhale]

Hi TGS4,

Thanks for your reply and thank you for taking the time to explain it in multiple ways, that definitely helped. It seems like this first section of the curve where stress is above yield just wouldn't be applicable since the elastic relationship wouldn't hold here. Do engineers typically just ignore this portion of the curve?

Thanks,

 

[mfgenggear]

These stresses are pseudo-elastic stresses. That means that they are stresses that exceed yield, based on an elastic analysis/calculation. In reality, the component will experience plastic deformation, that could have a plastic strain that could be on the order of the true ultimate plastic strain (or even higher, depending on the magnitude of the compressive triaxiality (although that can the topic for a different post)). But take a typical carbon steel - the true ultimate plastic stain could be on the order of 40%. Multiply that by E, and you could have these pseudo-elastic stresses on the order of 12E6 psi.

Another way to state this is that although it is plotted as an stress-life curve, it’s really a strain-life curve multiplied by a standard value of Young‘s Modulus. And of course shifted downwards and to the left to account for mean stress effects and for the design margins.

Does that make sense?

I not sure I follow your explanation. To my experience with gear design, the object is to obtain material with the correct tensile properties. starting with the y or with a safety factor. Interpolate the the number of cycles required to sustain the number of cycles.
With torque and rpm vs tensile properties required?
I would like to learn more.
Thank you

 

[TGS4]

Hi TGS4,

Thanks for your reply and thank you for taking the time to explain it in multiple ways, that definitely helped. It seems like this first section of the curve where stress is above yield just wouldn't be applicable since the elastic relationship wouldn't hold here. Do engineers typically just ignore this portion of the curve?

Thanks,

Not elastic - but pseudo-elastic. And yes, this portion of the curve is used.

 

[mfgenggear]

Here is extensive white paper on the matter