Reduced fatigue life due to end fastener loads

[koopas]

Good morning all,

I was reading the FAA paper by T. Swift entitled "Repairs to Damage Tolerant Aircraft"

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

I've got a few questions regarding the highest rivet loads at the first row (end row) of fasteners (pages 3, 4, figs. 1, 3, & 5 of the paper).

I can grasp the rationale behind how load is being incidentally transferred from the skin into the external doubler starting at the first row. Let me paraphrase: the load sigma_hoop is being partially reacted by the first fastener with the force F. The load that's not reacted is being bypassed to the next fastener in the basic skin as shown by sigma_bp. Now, the force F at the first skin fastener is being reacted in the doubler hole by and
opposite load F.

First question: Is force F acting on the skin and doubler holes opposite and equal in magnitude?

Second question: Refer to Fig. 1c, regarding the reacted force F in the DOUBLER skin hole, shouldn't force F's direction be in the same direction as force F of the skin hole? Isn't the doubler "helping out" with countering the hoop stress sigma_hoop? I am not physically seeing it. Perhaps someone could slowly walk me through what's happening in Fig. 1c.

Third question: Ref. fig 1c, is sigma_d acting on the doubler related to force F on the doubler hole in the following way: F = sigma_d*(0.05*width of doubler) ?

Fourth question: On page 4, the first row rivets are calculated to experience 187.2 lbs each. Is that value equal to force F? I.e. is the force F equal to the fastener load?

Fifth question: The bearing stress in the basic skin is then computed at 187.2/(0.04 x .19) ~ 24 ksi, where 0.04 is the basic skin thickness and .19 is the rivet diameter. Would the bearing stress in the doubler be computed as 187.3/(0.05 x .19) ~ 20 ksi, where 0.05 is the doubler thickness. Does this explain why cracks in external doubler repairs occur at the outer row, on the thinner basic skin?

Last question: I don't understand why the end fastener load is a function of doubler thickess. I would think the end fastener load would be constant, only being dependent upon the basic skin's characteristics. As previously discussed, the skin hole reacts some of the hoop stress with force F, and that load is reacted into the doubler. The doubler simply seems to passively receive the load. How can increasing the doubler thickness increase the fastener load? In fact, due to the larger doubler cross sectional area, the stress in the doubler should be reduced. I am confused.

Thanks for shedding some light,
Alex

 

[edbgtr]

Hi Alex

In order for you to understand the rationale behind fastener load distributions in joints, I recommend that you read some definitive work on the subject, which will help you to understand what Swift is trying to tell you. It is all a matter of stiffness of the joint, at it's ends, as to how much the end fastener loads up. The same effect is seen in bonded joints and the analysis is similar except for the discrete nature of the fastener spacing. The fracture mechanics analysis can be found in any good text on the subject.

Please read Prof Michael Chung-Yung Niu's book "Airframe Structural Design", Section 7.7 et seq. and also "Airframe Stress Analysis and Sizing", Section 9.7 "Bonded Joints" et seq, by the same author.

For a detailed summary of the various analysis techniques developed for this problem, get a copy of "Analytical Design Methods for Aircraft Structural Joints" by McCombs, McQueen and Perry, Technical Report AFFDL-TR-67-184. This report, with additional handwritten notes on fastener spring constants, can be purchased from, W (Bill) F McCombs, at 2106 Siesta Drive, Dallas, Texas, TX 75224. Phone him for the latest pricing on content and shipping on (214) 337 5506.

Regards,

Ed.

 

[i278]

koopas,

Nice link, thanks for that. I'll give my offhand answers to your questions, but keep in mind--they're worth what you paid for them:

1. Short answer: yes. Just as you figured, I'm sure.

2. I think you're putting the cart before the horse here The doubler isn't being loaded by the hoop stress, it's being loaded by the rivets. The skin is loaded in tension, obviously, and the doubler is restraining the axial strain of the skin. So F in the skin loads the skin in compression, and F in the doubler loads the doubler in tension.

3. Close. The area term would be (0.05*width of shaded strip) as only the idealised strip is being considered.

4. Yes. F is the local rivet shear load.

5. Yes, that would be right for the doubler bearing load. And yes, it goes a long way to explain subsequent failure of the skin at those locations. Hence the effort to transfer load gradually into and out of doublers and patches (fingering, tapering, stepping the doubler, etc.)

Which leads into your last question. Increasing the doubler thickness (all else equal) increases its stiffness, and hence increases the stiffness difference between the skin and doubler. This will then portion an increased load to the first rows of rivets.

[bad humour]That's why I always make the edges of my doublers as thick as possible. I concentrate as much load in the first row as I can, and then cut that row off before installing it on the plane. Reduces loads astonishingly! hur hur[/bad humour]

I hope that makes sense; that's the way I learned it, anyway. Corrections welcome.

Regards

 

[Sparweb]

Watch out on the bearing areas, too. When Swift starts calculating bearing stresses, he's using a generous bearing area based on a universal-head rivet. Try the same thing with countersunk rivet joints and you're asking for trouble. There's a tradition of using 1/6 the bearing area for CS rivet heads (I'm trying to track down a reference to this rule-of-thumb). Now look THAT up on Figure 12 and see what kind of joint life you get (after you've extrapolated the chart beyond all reason)!


STF

 

[Wairau]

Can somebody please advise how I can purchase the paper by T Swift.

Graham Murphy

 

[edbgtr]

Graham

You can download it directly from koopas' link above.

Ed

 

[koopas]

Howdy i278, and fellow comrades,

Thanks for taking the time to answer my questions.

Lately, I've become quite interested in load transfer across metallic joints (it's been a bit slow at work

Armed with a combination of Niu's books, some Boeing and Douglas training manuals and repair design documents, and the gracious help of this forum, I've come up with the following conclusions:

1. At operational load levels, end fasteners load up the greatest, in proportion to the doubler or splice element stiffness. It's ironic that everybody mentions the word "stiffness" but it's rarely formally defined. From what I can recall, the stiffness of a bar in axial loading is "E x A" whereas a beam stiffness is "E x I". I wonder how stiffness is defined for a skin doubler that is essentially a thin sheet in plane stress.

To me, you increase stiffness by

a/ upping the doubler gage
b/ upping Young's Modulus
c/ using stiffer fasteners

Thus, increasing any of the three characteristics above will stiffen the doubler, draw more load to it, and SOMEHOW load up the end fasteners more. I write "SOMEHOW" because I've yet in my life done a FEA to obtain those loads (my next challenge). After one obtains the end fastener loads, you can determine the bearing stresses in each respective sheet by dividing the fastener load by the bearing area. Obviously, the thinner sheet with the smaller bearing area will experience the higher stress, as well as a correspondinly lower fatigue life.

Is it typical to back out the bearing stresses at fastener hole locations by first determining the load transferred by that fastener, then dividing by the bearing area?


2. In order to reduce the peaking effect at the end fasteners, you taper the doubler to decrease the stiffness by decreasing the doubler thickness. You can also use two doublers, the thinner one termed a fatigue doubler that lays under the thicker doubler. I presume that the tapering effect comes from the thinner doubler "sticking out" a few rows of fasteners beyond the end of its thicker counterpart.

Why is the fatigue doubler placed closest to the original member or factory skin? You could still obtain a taper effect, albeit not so pronounced, by first placing the thicker doubler next to the original skin, followed by the thinner doubler. Folks on this forum, as well as the SRM's, always mandate the thinner doubler be closest to the damaged skin.

Last, as I've written in another post, I will now spend some time trying to obtain those fastener loads with some FEA. Ahh...the joys of learning.

Thanks to the invaluable help everybody on this forum has provided me; I wish I could return the favor more often!


Alex

PS: for some reason, a coworker has given me a subtle hint that I should find another "line of work" after he laughed at my proposition to purchase a finite element program at the airline's expense.

 

[Sparweb]

Since wiser engineers than I have come to your aid on these topics, I won't wade into these waters again, but referring to your last comment about FEA, I've encountered the same attitude from those "above me". There are several sources to these attitudes:
a) "It's just a piece of sheet metal; how complicated does this have to be"
b) "The cost of the software doesn't justify the benefit"
c) "What makes you so sure a FEA model can give you a better result?"

I haven't encountered a) personally, but I expect it exists.
Our company's policy is b) If we passed the costs of a FEA software package, and the time to learn it, on to our customers, they wouldn't be our customers any more.
My personal opinion is much like c). I can see that using a FEA model can give you a reasonable picture of how the stresses might change due to a repair, but it's not going to give you the actual stresses, unless you are willing to painstakingly develop the loads that are applied to the sheet in question, be it fuselage, wing, tail skin or some bulkhead inside.

Taking a stab at the loads in the sheet is something you have to do in any event, whether you're using a simplified analysis, or the full-blown FEA package. Once you've done that, the nominal stresses are relatively easy to find, and the stress concentration factors can be looked up (Swift's paper, for example). Since the Boeing's are held to Damage Tolerance requirements, you are also looking for crack propagation rates and inspection times. This does not require FEA. Call the Chicago Certification Office, talk to Bob Easton, and he can set you up with the gospel on damage-tolerant repairs to skin.


STF

 

[koopas]

Sparweb,

Points well taken. Do you know of many airlines that utilize finite element analysis in their repairs/mods?

I am going to start sending my resume out to aircraft manufacturers out there...design is calling me. Wait! The grass is always greener on the other side, and a year from now, I'll be whining about being stuck in a cube all day doing FEA on the lav door handle.

Decisions, decisions,

Alex