Stress and metal fatigue 1

[eltooon]

When talking about metal fatigue, we know that number of loading cycles to failure is related to stress. But the stress we talk about here refers to max. stress or stress range?

I'd expect both max stress and stress range affect the number of cycles. Yet in a typical S-N curve I see only the max stress. Information regarding stress range is missing? Then how is that S-N curve useful to us?

Anybody help me clear my doubt?

 

[swall]

First of all, S-N curves are only applicable to metals, such as steel, that exhibit a fatigue endurance limit. If your stresses are below the fatigue limit, it doesn't matter if you are talking max stress or stress range. If your stresses are above the fatigue limit, then it matters. One example I can think of is fatigue loading in springs. At the same max stress,a sinusoidal load cycle will exhibit fatigue failure earlier than a saw tooth load pattern, as the spring is exposed to stress for a longer period during each load cycle.

 

[metengr]

eltooon;
Here is a web site that you will find useful regarding a tutorial on fatigue.

https://www.efatigue.com/fatiguecalc.html

 

[davefitz]

eltoon:
stress range, but for low cycle fatigue, the test data used to generate the fatigue curve is usually strain range data , multiplied by a constant youngs modulus, to generate an apparent stress range. Since the alternating range in low cycle fatigue is often in excess of yield, one would need to adjust the stress value to account for plastic yielding.

swall:
I hear that the latest theories ( from people that work with aluminum) are that there is no such thing as a fatigue limit , that the values often used for steel are a result of an inaccurate test protocol or limited instrument accuracy.

 

[TVP]

There has been a great deal of research, both academic and industrial, in the very high cycle range (also called ultra high cycle and gigacycle) that shows even steels do not necessarily possess an endurance limit. Here are some previous threads with links to references on the subject:

http://www.eng-tips.com/viewthread.cfm?qid=242798

http://www.eng-tips.com/viewthread.cfm?qid=270181

 

[swall]

Davefitz--yes, I was aware of the "giga" high cycle fatigue behavior, as pointed out by TVP. My answer to eltoon was based on the known state of the art. After all, we got lots of steel articles out there designed around endurance limits.

 

[eltooon]

Thank you everyone. I've learnt a lot from the discussion, still trying to appreciate the complication of matel fatigue.

Now I come across DNV-RP-C203 and BS 7608B "curves", what are these two standards all about, and any copies available on-line?

 

[TVP]

There are copies of DNV-RP-C203 available for download. Here is one link:

DNV-RP-C203

http://www.advancepipeliner.com/Resources/Pipeline/Fatigue/RP-C203.pdf

BS 7608 is a standard published by the British Standards Institute (BSI), and is not frely available. You can pay to download a copy from the following website:

http://shop.bsigroup.com/ProductDetail/?pid=000000000000295202

 

[eltooon]

TVP,

Thanks for the links.
Btw, I found the 2010 edition of DNV-RP-C203:

http://exchange.dnv.com/publishing/Codes/download.asp?url=2010-04/rp-c203.pdf

 

[PriamEngineering]

When looking at an S-N curve it’s important to understand how the graph is generated. Historically a rotating beam specimen is used for the test. The beam is loaded to a certain stress and that stress is cycled by rotating the beam. The number of times the sample rotates before failure is recorded. This is done a number of times for a variety of stresses. A mean line is then drawn through the points and that’s how the S-N curve is created. Note that the mean stress cannot (in general) be used for general engineering components. You have to modify the endurance limit for amongst other things, reliability, size and surface finish. The Abscissa records number of cycles and the ordinate records stress (not stress range). The problem is that most S-N curves are only produced for zero mean stress. If you have a high mean stress you have to account for this using EFR stress or replicate the test at that stress state. This is how I do that:

http://www.eng-tips.com/viewthread.cfm?qid=263887&page=1

good luck!

http://www.priamengineering.co.uk