Proof Load to Ultimate Design Loads on Bonded Composite Structures 5

[SAITAETGrad]

We are looking at a small design and production cycle (3 articles) of interior monuments fabricated from fiberglass honeycomb sandwich panel materials.

For our substantiation, we would like to proof load the structures to the ultimate design loads.

We would proof test all three to account for variability in manufacture. We are expecting no permanent deformation to the structure (i.e. the ultimate design loads are considerably less than the allowable limit loads for the structure). After post test general visual inspect and tap test, we would like to install the articles on the aircraft. Any detectable permanent deformation or damage would be cause for rejection.

Our local approval authority has disallowed this approach in the past. The concern is that the proof test would cause considerable hidden damage that would be undetectable even with a battery of NDT methods.

But all composite structures have flaws, voids, etc. from limitations in the manufacturing process (regardless of NDI applied). Our proof test would likely involve some low level audible creaks. But, we would hold the ultimate design load for three seconds without additional strain under load or permanent deformation with the load removed. This should be indicative of sufficient residual strength to repeat the ultimate design load in service.

What am I missing here?

 

[Jetgirl8]

I think part of the problem could be your terminology. As repeatedly discussed in these forums, limit loads are the maximum loads expected in service and ultimate loads are limit times a safety factor (typically 1.5, but check your governing FARs). Using the terms as just defined, I would expect an ultimate test to have potential for large damage.

When the future's architectured
By a carnival of idiots on show
You'd better lie low

 

[SWComposites]

"post test general visual inspect and tap test" are essentially worthless for finding damage (crack, delaminations, etc) in the structure. Perhaps if you proposed a full ultrasonic / thermography type inspection you would get a different response from the authorities.

Alternatively, you could take one article, cycle it to your ultimate load a number or times (say 30; would need to get the authority to agree) to agree), then disassemble the part, do a full NDI, then cut it up in critical areas and do a detailed inspection including cross section photos. This might be enough to convince them that you can take another ultimate load case in service.

I going to assume that you are attempting to certify by test, and that you do not have material allowables and analytical predictions of margins, correct?

SW

 

[rb1957]

ultimate load tests (1.5*limit0 don't have to result in structural failures. interior monuments are typically designed for stiffness and abuse loads and will typically have high margins for static strength.

it is a pretty common rule not to allow ultimate load test articles to be delivered as serviceable parts. this rule sometimes gives exceptions to metallic structures where the static MS is large, the test article shows no sign of "distress", etc. the test article can be hiding some localised plasticity but if you can't detect it is it there ? and if it is there and so hard to detect is it critical ?

composites are still new to the business and react differently so they're dealt with with a lot more caution. can you show a detail of the design is "clearly" critical and test that (to destruction ?) to help your authority agree that there's nothing wrong with the design ? joints come to mind ? have you done a full analysis of the monument.

 

[SAITAETGrad]

Hello Jetgirl,

Please understand that there is the potential for damage and that we are testing to the ultimate design loads (i.e. 1.5xlimit). Since they are interior components, they also must handle the emergency landing loads which are ultimate design loads.

The component is over-designed, however, to handle the ultimate design loads without permanent deformation (i.e. below the material allowable limit load). This is not unusual for interior components. So, catastrophic damage could of course occur in the test but it would not be expected by the design.

SW, agreed that GVI/tap are limited and as you say it needs to take one more ultimate load. If the stress/strain of the component is linear, should one not expect the residual strength to remain? Whatever damage that was done to the component taking it to the ultimate design load was carried before it was released.

I have seen test reports from large OEMs with the same test procedure. What was their justification? Perhaps they were in error in their thinking. It does follow that on subsequent loadings the composite material will have additional micro-damage.

Yes, SW, we don't have all the test data. But, we will certainly go that way if required. I'm not looking for a way to circumvent the requirements but just a better understanding of them.

The statistical procedure is a means to provide certainty for designs where repeated full scale testing is not cost effective and weight penalties are of concern (wings,etc.). Here we are at the opposite end of the spectrum where we are not concerned with weight and full scale testing of each manufactured component is less expensive than the building block approach and weight is a tertiary issue.

 

[SWComposites]

"I have seen test reports from large OEMs with the same test procedure. What was their justification?" > they certify by analysis supported by test, and have a whole boat load of test data at the coupon thru sub-component test levels to validate the analytical predictions.

"If the stress/strain of the component is linear, should one not expect the residual strength to remain? " > probably, but there can be repeated load effects, though those (usually) show up only after a larger number of cycles. Also, do you have a single test load case (9g forward?) or do you apply a series of ultimate load cases to the components? if the latter, then one has to consider the load sequences for subsequent loadings.

 

[berkshire]

Can you afford to test one article to complete destruction and note the loads?
If these loads are substantialy higher than the required proof loads. Then that may get the Feds off your back.
If they are not then your inspector may have a right to be suspicious.
B.E.

The good engineer does not need to memorize every formula; he just needs to know where he can find them when he needs them. Old professor

 

[rb1957]

OEMs and small companies address things differently. sometimes the OEM can get a break 'cause they have alot of supporting data/experience/process control/etc that a small company doesn't have (and so they can deliever a tested item) ... of course the flip side of this is "do they really need to save that small change ?". but small companies also cop a break 'cause they generally don't have teams of specialists that can consider every possible problem ans interaction; small companies cop the break of being able to say "it's insignificant" or "it's standard of type".

in your case, like SW, i think thye'd have alot of supporting data, full FEA, component tests, experience with the design, etc. I dodn't think subject the test article to a 2nd ultimate load would demonstrate anything. if you've already spoken to your local authority, and they've led you in this direction, i don't see much choice. Unless analysis is a new component to the discussion, or component testing ... i doubt you'll get much milage with "well, other people do it", for a multitude of reasons (some valid !?). You'll need to bring something to the table.

 

[SAITAETGrad]

SW: " 'I have seen test reports from large OEMs with the same test procedure. What was their justification?' > they certify by analysis supported by test, and have a whole boat load of test data at the coupon thru sub-component test levels to validate the analytical predictions."

Generally yes. But not in this case no. I have the report and it was a proof load on a very similar item with no further substantiation. The article was put into service with a GVI/tap test. I should say it wasn't something done at Boeing...but an OEM just the same.

But, as RB1957 pointed out just because they did it doesn't mean I can get any mileage on that alone. I would have to agree with all the comments made about the unknowns.

Berkshire, for a few reasons, your alternative is not in the cards this time. In any case, there is the question of variability in process which can't be done with one test.

Thank you all for your very helpful posts. They are greatly appreciated.

 

[Taz99]

I assume that you know of the GAMA publication "ACCEPTABLE PRACTICES DOCUMENT, CABIN INTERIOR MONUMENT STRUCTURAL SUBSTANTIATION METHODS"

From Chapter 2.1
If the test article, after testing, is to be used as a
production unit for installation in a production airplane, the procedures for ensuring that the delivered unit is
airworthy, that includes quality control requirements, refurbishment and overhaul, and completion, shall be
defined as part of the certification plan.

Clearly this indicates that it is not unheard of use the test article as a production installation.

I have only some minor experience directly on interior parts but for a pax to freighter conversion program that I worked on I did the stress analysis of a number of interior items made from composite panels. We then tested the 1st example of these items and installed them on the 1st aircraft.

As seems to be the case with you we had some limited material properties and allowables for the inserts etc. We designed the part so that the critical 9g crash load times a further factor of 1.15 was treated as a limit load and then we tested the 9g crash load. The rational was that the structure had been designed to be repeatedly loaded to the limit load, so doing it once in the test should not weaken it for future service. This was accepted by the local airworthiness authority and subsequently accepted by the FAA.

The main problem that I see is that you have to have at least some confidence in your limited design allowables in order to convince the FAA that when you say our stress analysis shows XXX lb is the limit load and we'll prove it in a test, that your analysis was in the ballpark to begin with, because the risk from the FAA point of view is you will apply a test load that turns out to be just below the parts ultimate strength and introduces some non visible damage and the structure is permanently weakened.

 

[rb1957]

good reference !

good idea to consider 9g as limit load, but this means that you're got analysis to show the structure good for 13.5g. you could "talk" your way there by using 9g as your ultimate load but having an required minimum MS = 0.50. either way there's analysis to support the test.

 

[rerig]

ALL,
This is exactly why I am a member of the Eng-Tips Forum.
Excellent responses to a well stated question!

Rerig

 

[Sparweb]

I've been through a couple of these certification programs, now, and interestingly I've taken different approaches on each of them.

On one of them, I was actually in charge of the lay-up and curing process, and the certification tests, too, leading to rather little difficulty with the selection of loads and passing the tests. In that case we tested an article to failure, showing it didn't fail until a load greater than ultimate was applied, which was satisfactory to the regulator. The test article had flaws in it, too, which made the test conservative, and the regulator acknowledged that.

I attempted the same approach on a subsequent project, except I didn't have much control over the fabrication process at all. The parts didn't come close to their target allowable strength properties, and the test failed well below ultimate. The supplier wouldn't discuss a "fix" to their layup processes and the budget didn't have room to educate a new shop. Fizzle.

A third project brought in a very professional facility to do the composites work, and allowables data were available, too. In that case, the regulator was satisfied with a proof test of a production article, and provided that it did not show signs of permanent deformation, nor make cracking sounds during the load test, it would be acceptable for installation on the aircraft. We had to use a significant factor for environmental effects, which was about 1.4x IIRC. Our parts were external, hence directly exposed to the environment.

I like the idea of considering the ultimate design condition as the "Limit" load case for the sake of these designs. Since the definition of Limit load is "once in the life of the aircraft", we can extend that to include a moment of testing during the life of this cabin monument.


STF

 

[rb1957]

the "ultimate load as limit" is something like semantics. the FARs defne the ultimate load (9g crash) and even if the calcs show a healthy margin the authorities are going to be cautious about allowing a part tested to ultimate to be delivered. however, they are quite prepared to accept delivering a part that has been subjected to limit load (without obvious deterioration, which of course there shouldn't be ..). so if you do your calcs comparing 9g crash loads against yield strength, and 13.5g loads against ulitmate and have healthy margins, there sholdn't be much objection to using the tested part (tested to 9g).