Fatigue Strain-Life curve conversion to Stress-Life curve

[MihaelaN]

Hello,

Is there any methodology to convert a strain-life curve into stress-life curve even though it is meant to be used in low-cycle fatigue analysis?

Thank you in advance!

 

[rb1957]

for composites ?

If metal, then stress = E*strain ... no?

For composites, maybe use FEA to infer stress for strain.

"Hoffen wir mal, dass alles gut geht !"
General Paulus, Nov 1942, outside Stalingrad after the launch of Operation Uranus.

 

[LiftDivergence]

If metal, then stress = E*strain ... no?

In this case, no. The whole point of the strain-life methodology is the recognition that there is localized plasticity at a notch.

Strain controlled cyclic data should be indexed using a magnitude from a cyclically stable stress-strain curve. The size / shape of the hysteresis loop will change based on cyclic hardening / softening and potential relaxation. All of this is accounted for in a strain-life analysis usually by calculating an equivalent notch strain. Even MMPDS has mean stress relaxation and equivalent strain formulae.

If you want to "convert" one data set to another you would at least need to account for that. But there are other issues that make it problematic as well.

My question is why would you want to do so? Strain-life data is much less abundant so if you have a data set for some material, it is highly likely that there will be stress-life data for the same material readily available.

If you are using strain-life data because you have high strain rate / predicting low-cycle fatigue, it wouldn't make sense to "switch" anyway.

What exactly are you trying to accomplish?

Keep em' Flying
//Fight Corrosion!

 

[rb1957]

is it the strain over the concentration, or the "far field" strain ?

"Hoffen wir mal, dass alles gut geht !"
General Paulus, Nov 1942, outside Stalingrad after the launch of Operation Uranus.

 

[MihaelaN]

@LiftDivergence, thank you for your reply.

I have done some strain-controlled fatigue testing on welded test specimens and therefore I have the available strain-life fatigue data. This data will be used to analyze a welded pressure vessel using an S-N fatigue curve, so I need either to convert the experimental data from strain to stress-life or to find a method to use the current strain data in the analysis (since the software doesn't allow strain-life fatigue curve).

Any suggestions? Thank you!

 

[Sparweb]

I am pretty sure there's a methodology for doing that detailed in "Fatigue of Structures and Materials" by Jaap Schijve

 

[rb1957]

I think your fundamental problem is you're using an allowable strain/life curve but trying to analyze with stress/life analysis.

I think you need to be consistent ...
either get strain output from your FEM (directly or indirectly from stress/E) or
convert your strain /life curve to stress/life (by strain*E).

Now the problem would be if you have plastic strains, but then your life will be very short (< 1000 cycles ?), then I'd go with the strain approach.

Why do you say "the software doesn't allow strain-life fatigue curve" ? Do you mean your in-house fatigue analysis is stress based. ok, write a new analysis procedure that is strain based, because strain is the correct data for your specific job. If mgmt question this, ask them what to do. Reckon they'll say "convert strain data to stress".

"Hoffen wir mal, dass alles gut geht !"
General Paulus, Nov 1942, outside Stalingrad after the launch of Operation Uranus.

 

[LiftDivergence]

Strain life analysis should be performed using the stabilized local notch strain.

You don't necessarily need software to do this analysis.

You have e-N data from test. I would use that. From FEM/ classical analysis, predict the local strain. Then apply a method of correcting for cyclic changes to the stress/strain relation. If you are unsure how to do this by hand I would look and some textbooks. Chapter 14 of Dowling's mechanics of materials is exceptional. You can apply mean strain correction using smith-watson-topper or the like. Damage accumulation can be done with palmgren-miner just like S-N approach.

Keep em' Flying
//Fight Corrosion!

 

[LiftDivergence]

Any update on this, OP?

so I need either to convert the experimental data from strain to stress-life or to find a method to use the current strain data in the analysis (since the software doesn't allow strain-life fatigue curve)

Basically what I (and I think others) were saying is... just do a strain-life check by "hand".



Keep em' Flying
//Fight Corrosion!