Smooth Bar S-N Curves (VIII.2-3-F.1.2)

[AH_AK]

I am trying to reconcile the considerable difference between the ASME S-N curves and what you find in typical machine design textbooks (e.g. Shigley).

In the texts (e.g. Shigley), S-N curves are constructed using S_ut. The endurance limit (N=10^6) is 0.5*S_ut for lower strength steels. The curve goes to S_ut at N=1 (first cycle failure), and down to S_e (0.5*S_ut) at N=10^6. Of course, this is for a smooth bar, neglecting all correction factors.

Looking at the methodology in VIII.2-3-F.1.2 for generating curves, S_ut does not appear to play a role in generating the S-N curve for S_ut <=80 ksi. So all carbon steel with S_ut<=80 ksi would be using the same S-N curve. As a result, the endurance limit (N=10^6) ends up being much lower than the textbook hand calculation. Looking at Figure 3-F.1, the endurance strength for carbon steel with S_ut=80 ksi would be just over 10 ksi. That is nearly a factor of 4 more conservative than the textbook methodology (S_e=40 ksi). Then on the low-N end, the stress amplitude for first-cycle failure appears to be almost 600 ksi! This doesn't make any sense to me. It is worth noting that an 80 ksi stress amplitude corresponds to N=10^3 cycles to failure in the plot.

What am I missing here.

 

[TGS4]

Unfortunately, you are missing quite a lot. Including design margins and mean-stress corrections being built-in.

I recommend reading ASME PTB-1 to better understand where the smooth-bar fatigue curves come from.

 

[AH_AK]

This is exactly what I was looking for. If the design factors and other corrections are built into the curves, then it makes total sense that they are quite different than experimental S-N curves. Thanks for replying and providing a reference to a document that provides additional explanation.