Fatigue Life of FAR 23 Aircraft Repair 2

[Sparweb]

For years I have been using the same convoluted method of showing that the fatigue life of repairs I design for FAR 23 aircraft is satisfactory for regulatory approval (I don't get into the damage-tolerant stuff of FAR 25). The method was worked out by the boss, so I used it, but it relied on data with no source, and assumptions that didn't make sense to me. I recently came up with a new method that uses nothing but Mil-Hdbk-5 charts and Bruhn. I'd like to run the method's assumptions past the experts and see what they think:

[ul][li]The Kt=2.0 chart in Mil-Hdbk-5 for 2024-T3 aluminum represents the existing skin in the aircraft. Manufacturers design the skin around having holes for antennas in it. Round holes typically have a Kt=2.3, so using a Kt=2.0 exaggerates the life of the existing structure. I assume that the minimum stress is zero, so the mean stress is half of the maximum stress.[/li]
[li]The sheet of skin in question can only experience as much load as the fasteners at its edges can apply. Ie, a single row of AD4's spaced at 1" pitch can only apply 388 lb/inch of load to the sheet. For a 0.032" skin, that's 12 ksi.[/li]
[li]The stress concentration of my repair I could calculate from Bruhn, C13, but I prefer to use an old RAE Report 65004 that is specifically developed for the pressure vessel case, where fx=2*fy (ie. hoop stress is twice the longitudinal stress). This usually gives a Kt=4.0 to 5.0 for many repairs.[/li]
[li]The stress in the repair is the load per inch divided by the thickness of the skin+doubler (388 lb/in)/(0.032"+0.032&quot=6ksi [/li]
[li]I then use the Mil-Hdbk-5 chart on the next page, which is for Kt=4.0 to find the life of the repair. Same min/max stress assumption, and I get the life of the repair.[/li][/ul]
Would you accept that as the fatigue analysis if you were approving it?

I'm sure there are as many methods of analysis as stress engineers out there, but this method keeps the math down to a bare minimum, which I find to be the biggest barrier to understanding other people's analyses. I've tried other methods, including the FAA RAPID program and Swift's damage tolerance method. I found the heavy analysis required for a damage tolerance analysis seems most effective only at making another engineer's eyes glaze over, while no matter what repair I run thru RAPID, even bad ones, it will still say it's good for 300,000 cycles.



STF

 

[Sparweb]

Uh-oh, I just realized the charts are based on net section! I also have to account for the size of the cut-out and the size of the doubler. This cuts the life of the repair in half, and unfortunately for the case in front of me, is less than what I got for the basic airframe. Now I'm in big trouble!


STF

 

[KSshopper]

I like the last paragraph in your first post!

However, if I'm having trouble following your methodology. I understand you are trying to substantiate a repair for fatigue, but what in the repair are you looking at? The damage which was repaired or the first row of fasteners?

In your first paragraph you states the Kt for the aluminum skin is 2.0. What kind of cutout does this represent? An open hole has a Kt of 3.0. Also, the assumption on your stresses would be difficult to support.

A couple of things to think about when doing a fatigue analysis:

1. Do you really need a fatigue analysis? If the part is not a pricipal structural element, or it is compression dominated structure, you may be able to convince the DER a fatigue analysis is unwarranted.

2. Fatigue is an "elastic" phenomenon. That is, local yielding, or load redistribution that is normally assumed for a static strength analysis are inappropriate assumptions for a fatigue analysis. Therefore, you will find that straps, fasteners and doublers will not develop their full capability.

3. Fatigue life is proportional to stress by a factor of 4. For example if I increase my stresses by 20%, my life decreases in half. Therefore, the goal is to reduce your stress. If you can show analytically that your doubler reduced the local stress of your detail to less than or equal to the local stress of another original detail (i.e a lap joint or cutout) on the component, you should have no problem convincing the DER.

4. A finite element model is useful in determining the repair elastic behavior. Franc2D/L is free on the web, is easy to learn and has rivet and adhesive elements to attach doublers. You can get rivet stiffness from Swift's paper.

 

[Sparweb]

Given that this is the external skin on top of a pressurized aircraft, I would say, yes, this is fatigue sensitive. Besides, FAR 23.365 and 627 together make it mandatory.
Funny, I've always been using 2.3 for open holes in infinite sheets. I can see it getting higher if you consider a narrow strip of sheet, say between two frames, but in the whole skin, DIAM<<PANEL.
I'm fairly confident about the fasteners themselves. I don't know how true or conservative the method is, but if the rows of repair fasteners can transfer more load per inch than the rows at the edge of the sheet, then the thing is quickly approved by the DAR. (The official letterhead I use has a maple leaf on it - our system is similar to yours).
Actually, the DAR has already signed off on this project, but I'm personally not comfortable with the results. I designed the repair, and I got the static strength right, I'm sure, and it's now in the aircraft. The DAR has a few points he considers in a fatigue analysis to be satisfactory, but they don't satisfy me.

I'm concerned that I've not reduced the stress concentration in the skin around the cut-out edges far enough that the fatigue life isn't affected.

To use your proportional stress rule of thumb, are you using net section stress X Kt? In my case, the doubler reduces the net section stress by 31%. The cut-out is a 5&quot;x3&quot; rectangle aligned longitudinally. I get a Kt of 4.0 from RAE Report 65004 (a very obscure source, I know. If you have a better one, that does rectangles, I'd like to see). If the stress concentration factor rises by more than 31%, them I'm stuck.

I made the assumption that the skin has a normal Kt of about 2 because the top skin of this aircraft (Beech King Air) has &quot;staggered&quot; joints (following around a frame, then a stringer, then another frame) with about 3/8&quot; radii in the corners.



STF

 

[Michael996]

Stress concentration factors are for fatigue analysis. Not, stress analysis.

 

[Sparweb]

Hi Michael996,

I did the static stress a while ago already, and it's only the fatigue analysis I'm working on now. Part of my problem is indeed that I may have used the wrong stress concentration factor when calculating the life of the undamaged skin. But if I've mistyped something above, or worse, blatantly misunderstood something about fatigue analysis, let me know. I'm still on the steep part of the learning curve.


STF

 

[Nigel]

It sounds like you have done a skin repair. Was the damage beyond the limits of the SRM?

Drop me line so we can chat about this.

Nigel Waterhouse & Associates
Aeronautical Consulting Engineers

Transport Canada and F.A.A approval & certification of fixed and rotor wing aircraft alterations: Structures, Systems, Powerplants and electrical. FAA PMA, TC PDA.
n_a_waterhouse@hotmail.com

 

[marios]

Hi Sparweb,

I try to give a small contribution to this discussion.

- a value of 2.3 for an open holes might correspond to a Kf (fatigue concentraion factors)
- the curves present in the MIL-HDBK-5 cannot directly be used in design analysis, they are referred to specimens (mirror polished, small compared to the real structural part etc., axially loaded or bending loaded, surface finish and surface treatment different from the real part) and they show a medium life: probability of
50% to fail ...
- if there is more than one row try to consider the joint efficiency for static checks and for computing the transfer load
- the SF present in Niu's book give a good method to estimate the Kt (fatigue quality index)
- be careful with CSK rivets because crack usually starts in these holes when the thickness ratio is poor.

PS. I'm interested in Swift's publications, do you know where I can find them in the net?

marios

 

[Michael996]

Get a copy of Petterson's Stress Concentration Handbook.

The Kt of 2 is most likely a safety factor.

 

[Sparweb]

I've dug up a couple of papers by T. Swift, but only hard copies, I'm afraid. I also have a paper published by the Chicago FAA ACO on certification of Damage Tolerant repairs, also, only in paper form. I know somebody I could ask about electronic copies, though. I'll try to get back to you soon, marios.

The last nail is in the coffin, my original proposal of a &quot;streamlined&quot; fatigue analysis has been completely shot down. Back to the &quot;old&quot; way which, as I said at the beginning, is rather convoluted, but gets the job done. I should admit that the analysis has already been done the other way, modeling stress in the skin, finding SCF's for the cut-out, applying some rules on joint efficiency, particularly at the edge of the overall skin panel, and coming up with a margin of safety that way. I was hoping to taunt a more experienced engineer into revealing the guarded secret of &quot;How I do it&quot;. No such luck.

Thanks to all...


STF

 

[astroclone]

Check this bad boy out.

http://www.agingaircraft2001.com/pdfpapers.html

 

[Nigel]

Below are links to all the info you will ever need to get you started in fatigue and DTA.

If anyone needs FAA or TC approvals drop me line.

http://www.tc.faa.gov/its/worldpac/techrpt/ct93-69-1.pdf

http://www.tc.faa.gov/its/worldpac/techrpt/ct93-69-2.pdf

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

http://www1.faa.gov/certification/aircraft/FAA-AIR-90-01.pdf

Nigel Waterhouse & Associates
Aeronautical Consulting Engineers

Transport Canada and F.A.A approval & certification of fixed and rotor wing aircraft alterations: Structures, Systems, Powerplants and electrical. FAA PMA, TC PDA.
n_a_waterhouse@hotmail.com

 

[jetmaker]

Michael996,

FYI: Stress concentration factors DO need to be considered in Ultimate Static analysis. Originally, I thought not, but a colleague showed me when and why they need to be applied.

Case 1) Brittle materials. Brittle materials will reach fracture strain at the location of a stress concentration well before the basic material.

Case 2) Sections in which the stress concentration influences the majority of the net section. For example, a hole in a finite width plate or near an edge. As the net area drops compared to the gross area, the region of plastic deformation will extend completely across. This will result in a significantly higher net section stress than if one just assumed P/Anet.

Thought this might be useful information for some out there.

Regards,
jetmaker