Very high cycle fatigue vs. endurance limit 2

[geesamand]

We build a fair number of horizontal shafts that operate with an overhung load. Materials are 1018 C/S, 304 S/S, and 316 S/S with a well radiused ground finish in critical areas. Total shaft life is required to be 1,000,000,000 to 5 billion cycles.

Does endurance limit still apply here? Do you have real-life experience of shafts designed to endurance limit and operating in "corrosion-free" conditions lasting this long?

I've looked for academic information on these situations but so far I've not found much.

I have read that endurance limit can be interpreted as the limit where stress intensity at the crack tip is not sufficient to cause crack propagation. In that regard it's plausible that endurance limit can work for VHCF situations if the material has been shown to have an endurance limit.

Thanks,

Dave

 

[rb1957]

steels have an endurance limit, whether your particularly S/Ss do is something i'm not sure about (though i'd've thought so). have you looked into Mil HDBK 5 (5H or 5J i think is available on-line, google it).

what is your loading like ... you say an over-hanging load, but does the shaft rotate under this ? (if so easy to build up cycles, spinning the shaft very quickly). or does the over-hanging load (magnitude) cycle ?? (harder to cycle quickly).

you can get some mileage out of the damage tolerance route .. you want to show that a typical flaw (0.005" radius surface flaw) is below the threshold stress intensity ... possibly less well documented than S/N data !

not much help i'm afraid ... good luck !

 

[GregLocock]

Someone with far more experience than me said that the endurance limit seen for steel is more an artefact of the test method than a real limit. Unfortunately I can't remember his reasoning, and I've lost contact with him.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[unclesyd]

You might want to checkout this site.

https://efatigue.com/

GregLocock,

I've heard the same thing from a couple of sources. One was my boss in early 1950's who had worked at Wright Patterson and the other was a Metallurgist from Bethlehem Steel when he was discussing 4140 steel shafts.

 

[geesamand]

Thanks for the responses and links. Yes, this is fully reversed loading due to a rotating shaft under the influence of gravity. That's why over 5-10 years you quickly run into billions of load cycles.

 

[tbuelna]

geesamand,

1018 mild steel and 304/316 cres should theoretically have some fatigue endurance limits. But establishing what those limits are might be tough.

Values like endurance limit and fatigue life for a particular material and structure can vary widely based on many factors. In the end, fatigue life mostly boils down to a statistical value that a certain percentage of representative sample parts of a consistent material, heat treat, manufacturing process, and load life cycle have been tested to.

Establishing S-N values for structural materials requires lots of very time consuming and expensive testing. There is data available for high performance structural metals in texts like MIL-HDBK-5, but I doubt you will find data for 1018 or 304/316 in there.

To be honest, once you apply the appropriate knock-down factors for things like R,Kt and combined stresses, and then analyze for something like 10^9 cycles, you'll likely find that the allowable stress for your chosen materials is very low. Probably less than 5 to 10KSI at best.

Good luck.
Terry

 

[rb1957]

i figure you'd get 1E10 cycles running 32 Hz for 24/7/365 for 10 years. talk to us about the set-up and the operation. you should be able to cycle something like 10 times (100 times) the service rate, but i sense that you don't think you can ? talk to us about the required life 1E9 cycles or 5E9 ? unfactored or safe (factored) ?? somebody somewhere has run very high cycle fatigue tests (at least to come up with the idea of infinite life) so there's some experience out there (and no doubt written about). what sort of srevice stress level has been calculated ? < 10ksi ? < 1ksi ?? what things could happen in-sevice that'll mess with your nice, clean lab test ??

just some points to ponder ...

 

[geesamand]

Testing is always ideal but I cannot support it. The testing would cost more than the profit of any number of products would ever pay for.

I'm hoping simply to find the state of the art in published information to better apply to our practices. We're currently using a modified s-n approach that does not recognize an endurance limit but does apply a single data point for a very high cycle test that effectively lowers the allowable stress below the published endurance limits. We take into account the various form, surface, stress concentration, etc factors. I'm concerned that the one data point we have for low load / high cycle fatigue is not representative. Our shafts are designed and finished for maximum fatigue life and the specimen in the data point was not.

I had a look at eFatigue and saw nothing other than classic fatigue theory for < 10^6 cycles and endurance limit for above that. Perhaps there's more under the hood of their calculation engine but nothing on the site suggests that's the case.

There must be some other data out there whether it's published or proprietary. I will try to dig deeper into MIL and NASA publications.

David

 

[rb1957]

i was with up to "does apply a single data point for a very high cycle test that effectively lowers the allowable stress below the published endurance limits" ...

you're doing analysis after manufacture ... a common enough occurrence, and now you're trying to "show it good".

but you say "published endurance limits", which sounds like you have the answer you were looking for ??

 

[TVP]

David,

Finding data for 1018, 304, and 316 in the gigacycle fatigue region will be difficult. For some general background, take a look at this previous Eng-Tips thread and a presentation from Prof. Socie, one of the foremost experts on fatigue:

thread2-242798

http://fcp.mechse.illinois.edu/media/pdfs/20091027/3 Fatigue How and Why.pdf

For the most contemporary information on this subject, I suggest reviewing articles published in the International Journal of Fatigue. The following Google Scholar keyword searches should be useful, and perhaps you can obtain the actual articles from a nearby University library:

http://scholar.google.com/scholar?s...ernational+Journal+of+Fatigue&as_sdt=80000000

http://scholar.google.com/scholar?a...as_yhi=&as_sdt=1.&as_sdtp=on&as_sdts=23&hl=en

 

[geesamand]

Thank you very much TVP. I will dig into that and possibly order reprints of the relevant articles.

rb1957: yes, we've been manufacturing shafts in this configuration for some years. There is enough conservatism in the calculations and years of experience to know we can stay where we are. We're constantly designing new designs and/or new materials, so a better method is valuable and I'm looking to improve upon it.

Currently we do not use endurance limit. At 10^6 cycles, the endurance limit is where the s-n curve goes horizontal. We have one unofficial data point of some relevance that's in the gigacycle range and we use this point to form a curve on the s-n plot between the 10^6 point and through this data point. Therefore as cycles increase above 10^6 cycles, our allowable stresses decrease. I'm interested in other very high cycle fatigue data or a better calculation method if such a thing exists.

 

[rb1957]

there's an interesting slide in the illinios university article posted showing declining fatgiue life in the giga-cycle rate (above the traditional endurance limit) ... maybe it lends some credience to your mystery data point.

so you've got something that works and you want to make it better. a simple literature search might give you some ideas (though it appears it doesn't) but i doubt it'd be directly applicable to your specific application. if you can't afford testing on your own, why not try to interest a nearby university ? ... this sounds somewhat esoteric, = interesting to academia ??

 

[GregLocock]

Have you considered setting up a very high frequency beam bending test (using a voicecoil for excitation) to give you some additional data points?

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[kevindurette]

If there's little profit in it, perhaps you can over-engineer it with an extra factor on top of what you're already going to use and just use the endurance limit. I'm more of a plastics guy to be honest with you, though...

 

[unclesyd]

From an old DOS program.
Fatigue analysis is based on the Soderberg, Goodman, or
modified Goodman criteria.

NASA Handbook on Shaft Design

http://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/19840018973_1984018973.pdf

Information on Designing for Fatigue.

http://nptel.iitm.ac.in/courses/IIT-MADRAS/Machine_Design_II/pdf/1_7.pdf

http://nptel.iitm.ac.in/courses/IIT-MADRAS/Machine_Design_II/pdf/1_8.pdf

 

[HDS]

Much of the research in fatigue these days is being done by the military as they try to use planes and equipment beyond their original design life. So you are looking in the right place.

Your scenario matches one of the common test setups so I would have thought you could find data that correlates well to the application. My usual suggestion to talk to the expert at the steel supplier because they often know where the data is.

 

[IceStationZebra]

One thing that I didn't see (and pardon me if I missed it) is the loading. For steel and some SS under a constant amplitude cyclical loading you will see an endurance limit - but under variable loading you may not. The variable amplitudes help micro-cracks propagate.

ISZ

 

[unclesyd]

Here is a paper covering some of the work by various authors at the Japanese NIMS. It looks like with a little effort you should be able to find data on material close to your material composition.
I would check IOP Science site. There are links at the top of the page.

http://iopscience.iop.org/1468-6996/8/7-8/A02/pdf/1468-6996_8_7-8_A02.pdf