Fatigue Test of Aircraft Wing

[lLouie]

Hi,

I want to conduct a fatigue test on an aircraft wing. I have performed a fatigue analysis with random loads applied to the wing. Additionally, the loads on the wing have a maximum of 1.2g in the positive direction and 0.4g in the negative direction. How should I determine the loads for the wing? Should I consider the maximum loads it experiences? or Should I use rainflow counting to determine how many times it has been subjected to different loads and then perform the test accordingly?

Sincerely,

 

[rb1957]

if you have performed a fatigue analysis then you know "how many times it has been subjected to different loads".

if you have performed a fatigue analysis then you should know "the loads for the wing". How do you do a fatigue analysis without applying loads ?

Now this question "perform the test accordingly" is an enormous effort, designing a load control system for a fatigue test is a significant task; well beyond the scope of a student.

Maybe you have done your fatigue analysis with a truly random load sequence, though I doubt it. One key aspect is developing a quasi-random load sequence, a block of fatigue loads where the loads are picked at random from the spectrum loads (so that a spectrum load level that occurs 1/1000 flights occurs once in a bock of 1000 flights) then repeated on the test.

"Should I use rainflow counting" ... no! you use rainflow in the analysis as a "fudge" to get closer to reality. But when you're testing you are in the real world.

This is a very complicated topic, well beyond a student forum on the internet. You may have tried a google search and probably not found much. There are courses online where fatigue testing is covered as part of fatigue analysis, but only at a very high level. But the people teaching these courses can get you the help you need.


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

 

[SWComposites]

Yes, you need to use a rainflow or similar analysis to determine an appropriate fatigue spectrum for an aircraft wing fatigue analysis or test. But you have to obtain actual flight data and mission profile usage for the aircraft in order to do that. Then for a test you need convert the complete fatigue spectrum into a series of fatigue load blocks (magnitudes and cycles), then randomize the blocks, and so on. Its very complicated.

What exactly do you mean by "conduct a fatigue test on an aircraft wing"? An actual test? For what aircraft? And the loads cases for the aircraft are very likely more than just a simple up and down bending cases; usually there are torsion cases, and control surface position cases, gust load cases, unsymmetric maneuver cases, takeoff and landing cases, and so on and so on.

 

[rb1957]

sorry I disagree. Rainflow is an analysis adjustment to get a better analytical answer (ie closer to test results) than simple Miner's rule. A test spectrum should the load sequence from the load inputs (the gust spectrum, landing loads, etc) without any adjustment.

At least that is the approach I've seen at 4 different OEMs on both sides of the pond in some 40 years of doing this. For example, in a test spectrum no one (well, no one I've worked with) would add a GAG cycle to a fatigue test spectrum, this is baked into the applied loads. But "everyone" would add a GAG cycle (or maybe the rainflow equivalent) when they do fatigue analysis.

I'm going to put this down to a misunderstanding. If you think it is correct to apply rainflow to determine the test spectrum loads, then I suggest you review any of the many YT videos on teaching math.


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

 

[lLouie]

First of all, thank you for taking the time to answer.

I just want to learn how can be done fatigue test for aircraft wing. I took a flight load from a company. It can change the load but I know that load is randomly. I did fatigue analysis using this load in nCode DesignLife for a metal part which is connected to wing. Actually, I want to do fatigue analysis of metal part which has 40000 cycles.

I know the fatigue test is very difficult for me, but I want to develop a method to do test. I thought a method, however I'm not sure exactly;

I will use flight spectrum which is almost 4 hours. in rainflow counting. As I said, this load changes between 3.2g and 0.4g (I wrote wrong before writing, sorry)
I will find the loads as cycle. For example,
- There are 1000 cyclse for 0.1g
- There are 3000 cycles for 1g
- There are 200 cycles for 2g... like that. Then, for testing, I will apply these loads according to the number of repetitions I have determined.

As you said, there are no any good information in the internet for fatigue test.

 

[SWComposites]

some reading on aircraft loads spectra:

https://www.tc.faa.gov/its/worldpac/techrpt/tc17-40.pdf

https://www.tc.faa.gov/its/worldpac/techrpt/afs120-73-2.pdf

https://smartutsa.wordpress.com/wp-...07/load-and-stress-spectrum-generation_ld.pdf

https://aeronauticausa.com/wp-content/uploads/2023/10/AFgrow-Spectrum-Presentation-2022-9-12-22.pdf

https://www.researchgate.net/publication/269128151_Flight_Loads_Spectra_of_Two_Business_Jets

https://apps.dtic.mil/sti/tr/pdf/ADA265055.pdf

this outlines the 5x5 (5 flight types and 5 loads levels) approach for developing an a/c test fatigue spectrum:

https://asmedigitalcollection.asme....Airframe-Fatigue-Test?redirectedFrom=fulltext

Are you planning to conduct an actual fatigue test? Or a simulation in a FEM?

 

[rb1957]

I'd also look into "MIT Opencourseware" ... an astonishing amount of learning is available.

I feel you have grabbed hold or some words, and some ideas, and maybe ESL "English as a Second Language" is creeping in too. I'd start with basic engineering texts on fatigue, there are many. Once you get your understanding on the concepts and the analytical approach there's another reference I use ... ESDU 679023 ... if you want real world spectra. but if you just want ideas then pick a spectrum ... for 1000 flights assume 1 cycle 1g+-2g, 9 cycles 1g+-1.5g, 90 cycles 1g+-1g, 900 cycles 1g+-0.5g. The load spectrum is typically plotted on log/linear graph (cycles are log axis). Use your loadcase for some nz and interpolate for 1g and factor for different levels. This is very simplistic and ignores alot of small factors we take into account.

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

 

[lLouie]

I have flight spectrum data like the flight load spectrum in the articles you share. It just is not separated into load conditions such as taxiing and maneuvering. All available in one flight profile. Thanks to nCode DesignLife, I did fatigue analysis in a FEM. I just want to do fatigue test, however I do not know the method. No one shares their experiences for test. I studied analytical approach of fatigue. I believe that, my theory is good, but not for test. Sorry, this may be a bit presumptuous, but I need definitive test information. I'm looking at the articles that you shared.

In addition, I know that the fatigue is a deep subject which has many factors to do analysis and test. I try to do my best.

Thank you so much.

 

[3DDave]

To properly do wings you need to know the g load and the lift and moment distribution, though the rigs I've seen seem to only do lift. They distribute the load across the wing with a large number of attachments to push and pull the wing. You need to be talking to the aerodynamics experts who designed the wing to give you the pressure distributions.

The typical loads are the wing-root loads that the wing applies to the fuselage, but that won't tell where on the wing those loads originate, so they won't tell what the bending distribution across the wing.

 

[rb1957]

ok, if you want to do a fatigue test, actually test a wing, you are in seriously deep waters and this student forum isn't the right place. indeed this site isn't the right place ! The amount of stuff you have to learn is enormous !! it takes people years with very different skill sets to pull this off. You need structures designers, electronic systems designers, load control designers, etc. If you look carefully OEMs write up their tests for journals and symposium, but they leave out the details !

If you have a spectrum of flight loads, ok good starting point. you don't need to know what is causing the load, just the frequency of the load. So graph the spectrum for a number of flights, this is called an exceedance curve. Now divide this into a number of discrete loads, like 1/flt, 10x/flt, 100x/flt. These will be different loads to be applied on test. 100x/flt is IMHO pretty excessive. then you need to consider what is the ground load for the wing. where is the landing gear ? zero is a reasonable starting point, but when you're running a test you don't like zero load, you like to keep a small load applied to keep the whiffle tree stable. "whiffle tree" ... yes, this is how load is distributed on the wing, google it.

This shows you the complexity of what you're asking ... I'm babbling through things, haven't touched hydraulic controls, how to control these with different loads, a million other things. If you're near a university with an Engineering department, they can help you.

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

 

[lLouie]

I have wing loads to do fatigue test in different g loads. If you think that it is suitable for starting, firstly I can use the flight spectrum to find how many repeats occured in different g loads in one flight. After I determined the repeats of load in one flight, I can do fatigue test.

For example,
I have 4 hours flight spectrum, I will calculate the g repeats.
For 1g 1000 repeats
For 2g 100 repeats
For 4g 4 repests and so on.. Then I will obtain the moment/force in 1g, 2g and 4g. After I calculated the force/moment, I will use the hydraulic system to do fatigue test.

https://www.ansoneng.com/sample-page/other-stuff-of-interest/howsat-hawk-outer-wing-static-test/

https://advances.asee.org/wp-content/uploads/vol01/issue04/papers/aee-vol01-issue04-p04.pdf

Actually, we think same but as you said the fatigue is very complicated.

 

[rb1957]

if you have multiple flight loads cases, then you probably have different load input ... like fwd or aft CG, flaps ? gust vs manoeuvre. you want to pick the one that is closest to typical 1g flight. From these loads you need to extract 1g steady flight and delta 1g (gust or manoeuvre) ... so you might get 1g and 2g. You don't really need 1g, 2g, 4g, etc ... 4g = 1g steady + 3*1g dynamic.

In 4 hours of flight you get 4 4g loads ... WOW ! that is very severe. You can hanve the test loads represent 4 hrs of flight (or 40 or 400) depending on the number of loads you want. If you want 1/400 hours then you need 400 hrs of loads. this is a "block" of loads, that the test will repeat. So if you have a 400 hr block, then you have this high load occurring once, and you may have 10, and 100, and 1000? other loads representing the loads of 400 hrs of flight. The less frequent loads are scattered though the 400 flight block using some random (or quasi-random) distribution

And this is all very simplistic.

You also need a ground load (could be zero, or small load). You may want to consider the effect of flaps ?

When you're setting up your test rig, you need the "tare out the specimen". ie the rig should support the weight of the test specimen, the test specimen should not be hanging on the rig. clear as mud ?? when you are loading the specimen you need to apply loads (airloads) and the inertia response. Ideally the loads balance one another, put the specimen will need to be grounded in the rig, and the loads being reacted at the ground should be recorded.

You understand "whiffle tree" ?

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

 

[rb1957]

but first, a basic question ... why are you doing this ? If this is for certification, then drop everything and talk to your certification office. At least they will put you on the path they think you should be on, most likely they'll also help you ... they won't answer these questions (how do I set up a fatigue test ?) but they will point you towards people who can and they will review your test set up to ensure it meets their requirements.

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

 

[lLouie]

I understand the ''whiffle tree''. I talked to test manager, but he does not know the fatigue test method. He said, if you learn how the fatigue test does, we can apply. Actually, this is my thesis topic. Although, I searched the fatigue test, I did not find useful information. It is very complicated for me. This is my last resort, because noone helps me in real life. They just say I need to do research. I swear, I searched many articles, but all articles do not give useful information.

 

[rb1957]

if this is a thesis topic, is it really a test, an actual physical test, or "just" a paper on doing a test ?

I've just googled "airplane fatigue testing" and thee look to be many relevant hits. "airplane fatigue testing controls" hit a Linkedin page which should help a lot (a lot more than I can).

You've done a fatigue analysis using ncode, so you have a fatigue spectrum ? I don't know ncode in detail ... it "could" work from a continuous gust spectrum, or it may need discrete load levels (as you'd need for a test) ?

If your test manager doesn't know how to do fatigue testing, then you have a difficult task ahead, as there is an enormous body of knowledge to understand to get this done. If your test manger understands static loads, then that is a good start. You can determine an acceptable minimum number of load levels with ncode (I suspect). The issue is how to control the loading of these different levels; maybe "MTS" (at MTS.com) can help.

Where in the world are you ? India ?? China ?? Africa ?? East Europe ??

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

 

[lLouie]

In East Europe. As you said, there are many articles which are related to fatigue test, however, useful information is not given in articles. How many loads are used? How did the test done? I could not find any information.

Yes, I have load spectrum acceleration(g)-time. Thanks to it, I did fatigue analysis but, I confused the fatigue test.

I will look at the MTS. Thank you

 

[rb1957]

How did you determine your fatigue spectrum ? Is it "just" a set of loads or does it represent something real ? (like a gust spectrum, like from ESDU 69023)

How does ncode work ? does it use a continuous spectrum (a curve) or does it use a set of load levels ? if the latter, then you can use the analysis to determine a set of loads equivalent to your spectrum. I would do it differently myself, but it'd take too long to explain.

I'm surprised there are no texts that describe how to derive a test spectrum from a load spectrum.

If this is a thesis, I'm surprised that you are not getting support for your faculty, I mean they should want you to succeed ?
Is this a Masters or Doctorate thesis ??

I don't know what you searched, but I easily found this document (from the above google) that seems to be very relevant (and the author is in the same part of the world) ...

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

 

[SWComposites]

ok, eastern Europe. Go talk to the folks at Warsaw Institute of Aviation, who tested a composite wing box component technology demonstrator for the ILX-34 aircraft.

https://www.scribd.com/document/663...rator-Case-Study-Warsaw-Institute-of-Aviation

https://bibliotekanauki.pl/articles/2105142.pdf

 

[SWComposites]

I'm surprised there are no texts that describe how to derive a test spectrum from a load spectrum. its not that surprising, its typically company proprietary. The best public reference today is probably the last reference link I posted above. When CMH-17 Volume 3 Rev H finally gets published in the next year or so, Chapter 12 on Damage Tolerance will have a section on test spectrum development.

 

[lLouie]

Hi,

I'm very much obliged to you. I have read the documents which are shared by you.I will create an initial test for fatigue. Although, I'm not expert, I will do my best thanks to you.