7075-T6511 Extrusion S/N Curves

[Stephan Nelle]

I am looking for S/N curves of 7075-T6511 extrusion. The closest I have found is the first S/N curve in the 7075 section of MIL-HDBK-5J (Figure 3.7.6.1.8(a)or MMPDS-01 (Figure 3.7.6.1.8(a) MMPDS-08 (Figure 3.7.9.1.8(a) and MMPDS-13 (Figure 3.7.10.1.8(a)...these are the documents I have access to. It is "...for unnotched 7075-T6 aluminum alloy, various product forms...". Included in the product forms is 1.25 extruded bar and 1.25 rod. Where might I find S/N curves for exactly what I need (7075-T6511 extrusion)? Or, is it likely they don't exist? Is there likely to be an appreciable difference in the curve given vs the curve I am after?

 

[LiftDivergence]

Here: Properties of Aluminum Alloys: Fatigue Data and the Effects of Temperature, Product Form, and Processing by J.G. Kaufman. I own this book and I can confirm it is very useful.

Kind of expensive but you might be able to find a used copy. Alternatively check libraries

https://www.worldcat.org/advancedsearch

Enter the ISBN 978-0871708397 and search. You can also enter your ZIP to see what's around you.

There's also an eBook but it looks like it is just as expensive as a new print copy. No idea why, that's annoying.

Keep em' Flying
//Fight Corrosion!

 

[Stephan Nelle]

LD, thank you very much. I will find a copy.

 

[rb1957]

the MMPDS graph does cover a range of Kt. Not sure why the title says "unnotched".

the material form may not be exactly what you have, but it'll be near enough. I think you'll find more scatter between different data sets than between different forms.

Remember you'll divide the life by a safe life factor (like 5, maybe 3 if undercarriage), which gives lots of room for these little differences.

If you're really concerned then use a higher safe life factor (8?) or test.

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

 

[Stephan Nelle]

RB, valid points. I had thought about the scatter factor covering small differences. Just didn't know how small (or not small) the differences would be, though I initially guessed pretty small. I am new at this so glad to hear advice from experienced guys.

My intent is to compare a Miner's fatigue analysis using the fatigue curve given in Figure 11 of AFS-120-73-2 (reproduced as Figure A2-1 in AC23-13A) to the same analysis, but using a fatigue curve specific to the material in question. This is part academic exercise (practice for me) and part trying to get better/more realistic results.

 

[WKTaylor]

Hmmm... SN curves for aluminum alloys is a dated/low-value method for fatigue analysis.

Modern durability and damage tolerance [fracture mechanics] in a load-spectrum environment is more appropriate.




Regards, Wil Taylor
o Trust - But Verify!
o We believe to be true what we prefer to be true. [Unknown]
o For those who believe, no proof is required; for those who cannot believe, no proof is possible. [variation,Stuart Chase]
o Unfortunately, in science what You 'believe' is irrelevant. ["Orion", Homebuiltairplanes.com forum]

 

[LiftDivergence]

An S-N curve is not really a "method" per se, it is just a tool which can be used. It is completely possible (recommended even) to employ stress-life methodology with real fluctuating stress history, spectra, edited spectra, and damage accumulation. The spectrum tells you how the quantity and magnitude of major cycles and hysteresis loops. But you can still use the sequence of those cycles in a stress-life analysis based on S-N data or strain-life analysis to compute cyclic damage.

LEFM is obviously beneficial but has its own limits, generally being applied to macroscopic cracks with plastic zone size limits. Yes, there are exact analytical solutions for stress fields around crack tips, SIF, etc. which analytically tell us the crack tip extension to expect. But in the end, we still correlate to test data and tested da-dN/deltaK data. There is still an element of scatter, curve fitting, and correlation the same as with S-N data.

Both are data used to address the same problem... accumulation of damage due to cyclic or repeated loading. The S-N data is generally derived from tests on pristine specimen and the reported life is from "pristine" all the way to fracture. The da/dN data usually starts with a saw-cut specimen (intentionally induced macroflaw), and this data supports a different mathematical model.

That doesn't mean stress-life or strain-life approaches are outdated or low-value. As a matter of fact, for establishing ICAW for commercial aircraft the FAA requires checking results from both methods to support inspection threshold determination. If you only use macroscopic crack life to define an inspection threshold it may be unconservative.

I feel it is prudent to use both methodologies as appropriate. But I agree that the actual duty cycle and load spectrum should always be used whenever possible.

Keep em' Flying
//Fight Corrosion!

 

[Ng2020]

Crack growth analysis most definitely does not supersede fatigue analysis by miner's rule.
These analyses provide different pieces of information, cycles to crack nucleation vs cycles to a specific crack length, so they are not necessarily interchangeable. They complement each other.

 

[jpaero]

I was wondering if there are other
7075-XXXX extrusion S-N curves in MMPDS and if there's any appreciable difference amongst them. Just to get a feel if different forms have any impact.

 

[WKTaylor]

You might [unsure] find more SN curves in the following...

Battelle Aerospace Metals Handbook [ASMH] Vol 3207 - 7075 [bare]

ALCOA white papers.

ASM Specialty Handbook: Aluminum and Aluminum Alloys [1993] ISBN: 978-0-87170-496-2

Regards, Wil Taylor
o Trust - But Verify!
o We believe to be true what we prefer to be true. [Unknown]
o For those who believe, no proof is required; for those who cannot believe, no proof is possible. [variation,Stuart Chase]
o Unfortunately, in science what You 'believe' is irrelevant. ["Orion", Homebuiltairplanes.com forum]

 

[crackman]

I believe you will find the curves you are looking for in "Fatigue Data Book: Light Structural Alloys ", 1995 by ASM

If you are looking at a pure theoretical case for a notch in a bar, then I think the reference should give you what you need. However, if you are trying to perform a fatigue analysis of an actual aircraft joint, then it would be totally inadequate. For a fatigue analysis, whether it is strain life or safe life method, you must use joint SN or joint strain life data. The SSF method in Niu which by the way was invented by Lars Jarfall (see ICAF 1967 "Optimum Design of Joints: the Stress Severity Factor Concept") is based entire on joint SN data. I was fortunate enough to meet him over 20 years ago at ICAF and to get copies of his work before he passed away. Jarfall based his method on fatigue tests of riveted and bolted joints, not un-notched or notched specimens. Notched and un-notched lab specimens are very different than real world structure and typically are highly polished hence a large if not enormous scatter in life. The strain life method, in its currently updated approach, is better as the test data is based on the time to grow to a specified detectible (SEM) crack length and also is now based on joint data. Early on it too used unnotched specimens and results were not very good as you might expect. Just from a first hand experience level, when I worked at an OEM over 20 years ago and conducted hundreds of joint fatigue tests, we plotted them up against the MMPDS type curves and it was completely obvious as to how inadequate the MMPDS curves were.

Anyways, with respect to crack growth replacing fatigue, there is quite a bit of validity to this for the commercial world. The latest requirements for 25.571 mandate that all single load path and multiple load path structure not detectible below complete load path failure must use crack growth. Also, if you read the WFD requirements, MSD and MED susceptible structure must be analyzed using crack growth. The only area which still permits a fatigue evaluation is multiple load path structure that is inspectible for complete load path failure. This means a completely failed spar and skin or frame and skin or stringer and skin and so on. Obviously much of this analysis is usually done by the OEMs and of course they use joint fatigue data and they have a full scale fatigue test to back this up. That is not to say fatigue is not a valuable tool for doing preliminary design reviews and checking details. However, think of fatigue as an analytical method as a micro analysis versus versus crack growth being a macro type analysis. With a micro analysis, all variables come into play: hole fit, hole condition, squeeze force on rivet, clampup, friction, surface finish, and so on and so on. This makes for a very difficult prediction without testing actual joints with all parameters covered.

Best of luck on your project.

 

[LiftDivergence]

Agreed with regard to 25.571(a)(3)(i) and (ii) specifying macroscopic crack growth use for determining inspection criteria when safe-life criteria are not being used.

However, I will note that this language has been implemented "recently" with Amendment 25-96. The aging aircraft safety rule prevents showing compliance with anything lower than 25-54, regardless of TCDS basis, but the point remains.

I would also note that as I mentioned above, if you were to examine fatigue life of a structural detail based on stress-life (S-N) data and then the same detail using the assumption of a macroscopic flaw and LEFM, it is not always a sure thing that an inspection threshold based on the rogue flaw analysis would be critical. By the time you apply appropriate scatter factors to the S-N analysis, your inspection threshold might be earlier using that method. Like I said, although the language added to the later amendments of the CFR is specific, it is not always conservative.

The FAA actually has published guidelines for establishing inspection threshold and repeat criteria which are more detailed than those given in 25.571. These specify that the analyst must consider both statistical fatigue life evaluation, and rogue flaw LEFM evaluation when setting a threshold, and use the more critical of the two:

Note that proceeding guidance (Safarian etc.) has altered the factor of 4 to a bona-fide scatter factor accounting for many different variables.

This is why I feel I would continue to recommend both approaches. Use your peak stresses/spectrum/modified goodman/Palmgren-Miner/scatter factors to check a threshold for inspection based on S-N data. For the same detail, also use your stresses/spectrum/macroscopic rogue flaw/factors to check a threshold (typically done with NASFLA). Obviously only the LEFM analysis will be useful for establishing a repeat interval. I would almost never recommend using a safe-life evaluation per 25.571(c).

Keep em' Flying
//Fight Corrosion!