Frequency and length of fatigue testing 1

[DuncanGibbons]

Hi all,

I am wanting to calculate how long a fatigue test will take per specimen. ASTM E466 states that up to 100 Hz test frequency for constant amplitude axial fatigue test is acceptable.

If the test is performed at 100 Hz and say for 2 million cycles, am I correct in saying the test will last 2 million / 100 = 20 000 seconds or 5 hours and 34 minutes or are there other factors involved?

Thanks

 

[rb1957]

sure. the question I'd have is do you need 2 million cycles as the life in service, ie the safe life, or is 2 million cycles equivalent to infinite life ?

if it is the service life, then you need to apply factors to it, maybe 5 (so the test would run for a day).

How many specimens are you running ? (ie more than one, yes?)

another day in paradise, or is paradise one day closer ?

 

[DuncanGibbons]

Thanks for the reply.

Where does a factor of 5 come from? I just chose 2 mil for the calculation

I am planning to test 8 specimens. 4 in 10^3 to 10^6 cycle region and 4 in 10^6 to 10 ^7 cycle region.

 

[rb1957]

5 is a typical safe life factor ... can be as low as 3 or as high as 10 (or more !?).

2 million may be a reasonable estimate for life. Are you using a limit load (in which case very conservative) or some lower load, more representative of service loads (50% limit ?)

another day in paradise, or is paradise one day closer ?

 

[rb1957]

what sort of fatigue analysis have you done (to know it should survive 2 million cycles) ?

another day in paradise, or is paradise one day closer ?

 

[DuncanGibbons]

I'm generating material data for a specific material and process, so haven't done any fatigue analysis other than review literature on data generated for similar material and process conditions.

 

[rb1957]

oh, ok, you're making an s/n allowables curve. then 2 million is a number. I'd consider maybe 10 or 100 million ... maybe follow on testing ?

R = 0 (smin = 0) ? or -1 (reversed loading, like on a rotating beam)

Kt ? (1, unnotched) 3? (open hole)

so you're going to run at different load levels, and see when it fails ?

another day in paradise, or is paradise one day closer ?

 

[MintJulep]

Your calculation is correct. If the test starts and runs to completion without interruption.

That often doesn't happen. Some instrumentation stops working, or the test machine breaks, or the power goes out, or the sample slips in the fixture.......

 

[EdStainless]

So are you testing two samples each at four different loads?
It is normal to use slightly higher loads and higher cycle rate to speed tests.
Just make sure that the samples don't heat up.
One problem is that if either of those factors is significantly different from the service conditions you may generate data that isn't applicable.
There aren't very many applications where people test beyond 5 million cycles. Those get more specialized.
In spring wire we commonly tested 200 million cycles.

= = = = = = = = = = = = = = = = = = = =
P.E. Metallurgy, consulting work welcomed

 

[DuncanGibbons]

Thanks for the info and tips.

I am planning fully reversed axial and unnotched (I have read various sources say only full reversed fatigue should be plotted on SN curves).

Yes, two samples at four different loads.

 

[rb1957]

I don't know about those sources. To say "only full reversed fatigue should be plotted on SN curves" seems very restrictive.

Sure, reversed loading is most severe; maybe reversed loading represents loading that "you" mechanicals see in your mechanisms ?

Only Kt = 1 ... again, ok, maybe your mechanical practice (maybe you use neuber to correct for Kt ?). sure Kt = 1 tells you something, so too does Kt = 3 ... but, meh.

another day in paradise, or is paradise one day closer ?

 

[DuncanGibbons]

One source says its recommended, other source says its a must.

Link

Link Page 27, think its specific for this class

SHingley's Mechanical Design textbook also states "The typical S-N diagram is only applicable for completely reversed loading." Page 307

 

[rb1957]

2nd link, page 15 ... "R= -1 and R= 0 are two common reference test conditions used for obtaining fatigue properties"
2nd link, page 28 ... "The S-N diagram is applicable for completely reversed stresses (R=-1) only!" ... never heard this restriction (in some 40 years !!) It might be interesting to talk to the author about this, to see where he's coming from. But it is easy to test (rotating beam) ... though adding tension to a rotating beam test isn't difficult either, so ...
2nd link, page 70 ... "master fatigue diagram" ... seems very odd to show minimum cyclic stress higher than maximum cyclic stress ?
2nd link ... how can a paper on fatigue (ok, after a quick scan) not mention "stress concentration" ??

another day in paradise, or is paradise one day closer ?

 

[DuncanGibbons]

Shingley's Mechanical Design textbook states "Historically, there have been many ways of plotting the data obtained for general fluctuating stresses, with such names as Goodman diagram, modified-Goodman diagram, master fatigue diagram, and Haigh diagram." It also states that Basquin's equation is applicable for fully reversed stress cycles.

So it seems if the SN curve is to be used for calculating other values, fully reversed is recommended (according to Shingley at least). It seems to be the standard for comparison of data, they mention a bunch of other fatigue equations and criteria (Goodman, Morrow, SWT, Walker) for converting alternating stress cycles to fully reversed for comparison purposes I presume?

 

[rb1957]

it could be that the analysis prepares an allowable fatigue curve, that covers a range of R values, based on an R = -1 test.

I would suggest running tests at other R values as a way of validating the analysis.


another day in paradise, or is paradise one day closer ?

 

[rb1957]

still don't understand how you take Kt into account

another day in paradise, or is paradise one day closer ?

 

[SWComposites]

DG -
What material are you testing?
What test specimen type?
100 Hz sounds high.
8 tests for an S-N curve sounds too few.
What is the application for this data? Aircraft?

 

[DuncanGibbons]

The material is Ti64 ELI, specimen type will be the standard round ASTM E466 (Specimens with a continuous radius between ends).

For a research project and the aim is to perform an initial characterization of the production process, but yes within the aviation field.