Adding 1.15 to margin of safety for repairs.

[KirbyWan]

Howdy all,

I dug around trying to find an old discussion about this topic and could not find it, but was able find the FAR (25.625) that mentions it.

From my understanding, when I design a repair I add a fitting factor of 1.15 so that my repair has to be at least 15% stronger then what broke. I do not need this added fitting factor if there are multiple load paths, i.e. more then 5 fastners, but I can not find a reference for this.

Could anyone point me to a discussion of substantiating repairs and the use of fitting factors? Could engineers experienced in the repair of aircraft talk about how they design repairs? I would like to avoid a discussion of damage tolerance to simplify the discussion and because I usually work on secondary structure.

Thanks,

-Kirby

Kirby Wilkerson

Remember, first define the problem, then solve it.

 

[ESPcomposites]

Fitting factor - Good question. From what I have seen it is program specific and I have seen both 1.0 and 1.15 used and sometimes selectively applied to certain structure. I believe it comes down to how the test data is supported and if it is absorbed in there or not. I have also seen the rules change when weight savings comes into play. It would be good to hear if there is a more defined rule though.

Repairs - Perhaps you can be more specific?
- What original structure material, what repair material?
- Bolted joints only or are you interested in bonded joints as well?
- Repair to primary or secondary structure?

Repair General - That being said, I can give you some rough guidelines.
- Generally stiffness match and restore strength. This is vague though its the first "golden rule". For stiffness, you want the structure to have about the same overall stiffness so the repair does not draw extra load into the damaged region. This would be bad for the original structure, especially since you covering potential cracks with repair that may not show any sign of damage. Obviously too soft is not acceptable. Generally, if you are the same material, consider the same gage or the next gage up for a bolted joint. At a basic level, you can Et match the repair to the structure, or alternatively EA match a cross section over approximately the size of the repair. Consider the bending stiffness (EI) as well. For strength, you are supposed to "restore" strength, but it may not always be 100% achievable depending on what is sizing the original structure. In compression, the repair can achieve 100% of strength generally, but it may not able in tension due to the extra holes for fasteners.

- For bolted joints, you want at least two fastener rows for load transfer, preferably three. Otherwise the bearing loads may be too high and your DaDT and/or strength will suffer. It may also not couple the structure well with a single fastener row and the overturning kick due to eccentricity will be bad. I have not seen a repair with just a single row for load transfer.

- For bonded repairs, you may need to consider a failsafe scenario (i.e. may want to consider a failed repair and make sure you are good for limit load).

- For references, consider Niu's books. There is some basic info there. Also, most of the general structural analysis techniques are relevant to the repair, so you can use other references for that aspect. I have a few other books with some general design rules and will list when I see them next (not with me at the moment).





Brian

http://www.espcomposites.com

 

[rb1957]

applying fitting factors is somewhat arbitary (IMHO) in the 1st place. you could apply it to a repair, but i don't think it's required (unless it's single loadpath, ie failure = a smoking hole in the ground). as for 5 rivets or what not, maybe AC43.13

 

[YoungTurk]

I'm assuming you've already reviewed 25.625 & 25.619?

http://ecfr.gpoaccess.gov/cgi/t/tex...=14:1.0.1.3.11&idno=14#14:1.0.1.3.11.4.172.11

Also, I suggest reading this previous thread if you have not already:

http://www.eng-tips.com/viewthread.cfm?qid=264393

I don't see repairs having any specific criteria (at least not per the regs). I also don't see any verbiage speaking to single vs. redundant load paths with respect to fitting factors, those considerations apply to FDT, DTA & establishing allowables.

My general interpretation is that any "uncertain" fitting requires use of the factor. See my posts in the previous thread for a more detailed opinion regarding what is uncertain. It comes down to what you use to define the strength of your structure (be it repair or original).

Of course, aside from regulations, determining when to use special factors can be / is often dictated by either company procedures or approver judgement. If either of those parties dictate using a 1.15 "fitting factor" for your repair work, then you've got your marching orders!

 

[SWComposites]

Typical sheet metal repairs are not considered 'fittings', so the 1.15 factor is not applicable. Besides, most repairs to aluminum structure are fatigue critical, not static strength critical. And it is very difficult to restore the original structure pristine strength (rather than only restoring ultimate load capability) and it will be even more difficult (impossible) to restore original strength * 1.15.

 

[ESPcomposites]

But I think the definition of "fitting" may vary by company and is not specific to just "fittings". It may apply to joints such as this in the repair. Boeing appears to have a different definition than Lockheed. Even at Lockheed, the approach was program specific, and not universal within the company. As the good thread by Youngturk mentioned, this may be partly explained with how each company develops allowables. If there is significant test data that has been statistically digested, then there may not be an additional need for a factor. But if the "raw" material allowables, without much consideration for the effect on the joint, then a fitting factor may apply. But that is only part of the story and it may simply come down to the program/DER requirement.

If the original structure was designed with a fitting factor already accounted for (i.e. repairability or similar detail in near proximity), then you may consider it for the repair for consistency. In that sense, you would not be over penalizing yourself or "double dipping". Provided you are consistent within the program philosophy, there is not real difference.

Brian

http://www.espcomposites.com

 

[Sparweb]

When I first saw your questions, my first thought was to point you to the Advisory Circulars on the topic. Alas, they are not as helpful as I'd expected. Believe it or not, AC 29-2 is more helpful than AC 23-19 or 25-21.

Our practice is to use the fitting factor (23/25/27/29.625) only in cases where certification is carried out by analysis alone, and only in joints or structures with one or very few load paths, and only when there is enough uncertainty in the strength of said joint. Manufacturing repeatability is a main concern here, but wear-and-tear plays its part. Your question circles around repairs, and as we all know, work done in the field is not expected to be as consistent as work done in the factory.

The proper use of fitting factors gets drilled into us here because because our preferred "proof of structure" is by test. If we were to load-test an assembly to 1.15x the ultimate load, and it failed, then we would have failed not because the structure isn't strong enough, but because we were being too conservative. That is an expensive mistake to make on an engine truss that goes pretzel-shaped.

While thinking about trusses: legs of seats are as often in tension as compression. Column buckling is really hard to calculate but easy to test, so you would think that a test would save you the uncertainty AND the fitting factor. Beware, you are not relieved of using the "1.33" fitting factor for seats, because 2*.785 always applies.

Despite what I've said about not risking too much conservatism, test loading an extra 15% can afford you some freedom to change the design in the future, or change to cheaper/lighter/thinner materials, or make other changes that come up when confronting the reality of turning your design into a real part or repair. When you find yourself working on a design that will clearly have a large safety margin, you can choose to "use it up" with factors such as this, or just declare the MS as you find it.

It could be interesting to compare the meaning and purpose of the 1.15x fitting factor from *.625 with the "environmental" factor used in composites to account for moisture, temperature, and production variability. Roughly the same concept, just adapted to the different material. Maybe getting side-tracked on that, though.

Steven Fahey, CET

 

[der8110]

Repairs have to meet the same regulations as an original design. A recitation of the regulation is in order:

For each fitting (a part or terminal used to join one structural member (so secondary structure could be excluded) to another), the following apply:
(a) For each fitting whose strength is not proven by limit and ultimate load tests in which actual stress conditions are simulated in the fitting and surrounding structures, a fitting factor of at least 1.15 er (no option for 1.0) must be applied to each part of—
(1) The fitting;
(2) The means of attachment; and
(3) The bearing on the joined members.
(b) No fitting factor need be used—
(1) For joints made under approved practices and based on comprehensive test data (such as continuous joints in metal plating, welded joints, and scarf joints in wood) er (MMPDS or 26.613 allowables); or
(2) With respect to any bearing surface for which a larger special factor is used.
(c) For each integral fitting, the part must be treated as a fitting up to the point at which the section properties become typical of the member.


Also 5 rivets, or even 5 rows of rivets, might not qualify as multiple load path.

 

[ESPcomposites]

Der8110, that looks reasonable and might clear up some of the confusion (for me at least). Can you list the source as well or perhaps I missed it?

Brian

http://www.espcomposites.com

 

[YoungTurk]

der8110,

Your version of 25.625 has a few notes that mine doesn't. Can I get the annotated source as well?

In general I agree with your notes; with the possible exception of excluding secondary structure from the definition of structural. I'm not sure why we'd call it secondary structure and then say regs referring to structural members dont' apply.

ESP,

My previous link to 25.625 on eCFR appears down, here is RGL:

http://www.airweb.faa.gov/Regulator...8AA80AC9AAC8F8838525667200503969?OpenDocument

You can also navigate to 23.625 at either site, but the verbiage is the same.

 

[der8110]

my apologies, the text in red was my editorial comment. This consisted of the statements:
(so secondary structure could be excluded)
(no option for 1.0)
(MMPDS or 26.613 allowables)

This goes back to Bulletin 7a, fittings are defined as attaching primary to primary. Since then, CAM 4b, preambles to Part 25 amendments, AC25-21, and the policy memo on Factors are all rather useless.

 

[737eng]

Per my view and what I require for a repair that I review and approve as a Static Strength DER-
Per 25.625 if you are using tested allowables to design your joint than you don't need to utilize the 1.15 factor. If you are using computed allowables (i.e. Fbru*d*t) then you have to include the 1.15 factor.
You have to design your joint to the minimum failure mode; fastener shear, bearing, tear-out, net section failure, transition failure(example -head bending). The only way to account for transition failure is by testing, so if you are not utilizing a tested number, add the 1.15.

The 1.15 factor is not designing to be 15% stronger than what broke and it is not a saftey factor per say, it is a way to account for the unknown without adequate test data.

 

[irq]

I have 3 questions to the current discussion.

Q1:
I am also interested in reference, which directs the avoidance of fitting factors for multiply load path. Looking through your discussion I still can not find the answer. Could you quote here an appropriate text from AC or other regulatory documents and give the source?

Q2:
I have seen many times the FF 1.33 applied for the attachments/fittings, which connect interior parts to the frame. IMHO to conservative, since the components were also tested, but as this is program specific requirement I have to accept it. Let's assume the fitting assembled from two parts. They are bolted together on 2 elongated holes for some adjustability and the load is transferred via friction existing between parts. How do you treat such slippage connections during limit load to avoid moving between parts? (I assume that in case of ultimate load the movement between parts is allowed). Would be FF 1.15 enough or do I need to apply 1.33 again? Are there any special regulations for this type of connections?

Q3:
Do we need to apply any FF for the structural attachment if the substantiation is done by a detail FE-Analysis?

 

[KirbyWan]

Thanks for everyone's comments. I think I have hit on one of the many places in aircraft repair where there is confusion about what the proper methods to design and validate a repair are. I'm going to go through the many references different people have included and try and come up with a standard method for designing and validating repairs with the hope of sparking more discussion and through that discussion reaching a consensus on what consititutes a good methodology.

Youngturk: That was the thread I remembered and searched for but could not find. Thanks for the link.

-Kirby

Kirby Wilkerson

Remember, first define the problem, then solve it.

 

[IRstuff]

This is not my field, but the concept of arbitrary factors seem, well, arbitrary. The concept of making the repaired part stronger makes sense conceptually, but given that you don't necessarily know the factors that went into how a given part was broken, or how much stress the part experienced, would a factor of 1.15 even make consequential sense?

If it breaks again, does the next guy know to add 1.15 to your repair, or is he going to add 1.15 to the original?

TTFN

FAQ731-376

 

[Sparweb]

DER8110's quotes and references from Aero Bulletin 7A reminded me that many different fitting factors and other general "rules of thumb" have been in use for a long long time. It seems their usage has never been consistently enforced as much as the special factors from 619 (casting/bearing/fitting).

Niles & Newell sum up the point about uncertainty: "It is well to provide a margin of safety of 20 per cent at least on all parts of fittings when the loads are 'accurately' known, and a 50 per cent margin is not unreasonable when they are indeterminate."

Not an iron-clad rule they give - more an appeal to the analyst's judgement.

Other souces I've found are much the same, offering "guidance" instead of rules, and even EF Bruhn offers only "good practice" in D1.10.

I haven't opened AB7A (or AB26) in a long time, and what a different perspective that gives. It does not read like the airworthiness reg's of today. What I find is similar to the advice in Niles & Newell, so it looks like 20% was the standard practice at the time (1930's to 1940's).

Other observations made in these books suggest that wear on a part reduces its strength below design values if it is subject to motion, and this should be catered to in the design allowables and/or margin of safety that is accepted. Also, pre-stress on parts once they are installed can be a significant fraction of the working load in service, in mis-aligned holes for instance. Zero applied load doesn't always mean there is zero stress on the part. Once it's explained it seems pretty obvious, but what factor to use when that condition is found is clearly a matter of long-standing debate.


IRQ:
I recommend you look up the current FAR's on the FAA website and read through the paragraphs from FAR 25 619 through 625, and then skip to FAR 25.785. That's the difference between a 1.15 factor on 1.33. Use 1.33 for seats, belts, litters... everything that directly attaches a person to the airframe.

Kirby,
Thanks again for bring up this topic. As you say, you've "defined the problem", and now can work on solving it.

Can I expect the book to be published in, say... November?


Steven Fahey, CET

 

[KirbyWan]

SparWeb,

Thanks for the input. I'm better at doing the reading then the writing, so I don't know if I can hit a november deadline, but maybe I need a deadline to actually get something done. Maybe I'll just keep this tread alive by posting my thoughts as I read the AC's that you listed.

-Kirby

Kirby Wilkerson

Remember, first define the problem, then solve it.

 

[sigmatau]

No confusion, if you're using calculated allowables, the 1.15 factor is to be included, otherwise you're using test data. Unless a larger special factor is being used (25.619). Specical factors are in 25.621-25.625.

Jim Brown, DER Structures

http://www.faader.us