The conditions where NDI may not prevent failure of adhesive bonded structures 5

[blakmax]

I'd like to bring attention to a paper I recently published which brings into question the frequent reliance on adhesive bond NDI to assure structural integrity. Such faith is based on damage tolerance analysis to determine tolerable defect sizes, combined with in-service NDI to detect defects larger than the estimated defect size. The tolerable defect size is usually determined by either testing using artificial defects such as teflon insert, or by FEA where elements are disconnected to represent the disbond. Such approaches imply that the adhesive adjacent to the defect is pristine and exhibits full strength.

In real bonds, failure may be due to interfacial degradation or porosity, both of which will result in a low bond strength adjacent to the defect, so the assumption of the bond being in pristine condition adjacent to the defect is unsound.

The paper shows that for bonds where the interface is susceptible to degradation, and the overlap length is short, there is a real chance that the bond strength can degrade to an extent such that failure may even occur without any detectable disbond being present.

I'd welcome discussions on this paper.

Regards

Blakmax

 

[btrueblood]

Interesting paper, and somewhat scary in its conclusions. How soon do you see updates to FAA reg's. occurring? Hopefully real soon, given the rapid adoption of composite materials in commercial aircraft.

 

[Sparweb]

I don't think the FAA would act on something that wasn't more directly linked to the cause of an accident or incident. Interpretation of a failed joint coming from the NTSB would get peoples' attention. Which brings me to my question - how did the subject of the paper evolve, if the disbond is not the cause of the rotor blade failure? (it was fatigue cracking at a root attachment bolt hole). The blades of the R22 were designed under the older "Safe-Life" philosophy, not "Damage Tolerance". Subsequent attempts to impose DT driven inspections on R22 rotor blades has proven to be less than satisfactory. This seems to be because the inspections of the bonded elements of the blade are only looking at the secondary load path, not the primary one. Tap testing a blade with a cracked bolt hole will not detect the crack in the bolt hole, it will only catch the skin delamination caused by the shifting members around the bolt crack.

Sorry that turned into a diatribe about Robby, and not about your paper. :[

The point I was really trying to make is that the cited example doesn't illustrate your point as well as you want it to. In a strange way, if the skin bonds of the R22 blade had been poorer and weaker, and suffered from adhesion failure sooner, then the tap tests could have detected the problem, and inspectors might have snagged the bad blade before it failed catastrophically. That tortured logic doesn't really help the case you are trying to make. The blade simply wasn't designed for damage tolerance.


STF

 

[blakmax]

Sorry Sparweb but you are confusing my paper's content with a totally different crash. This aircraft had absolutely no bolt hole failures.

You also miss the point that the paper indicates that for short overlap joints where interfacial degradation is occurring, failure may occur in the total absence of disbonds which could be detectable by tap testing. So suggesting that if the " if the skin bonds of the R22 blade had been poorer and weaker, and suffered from adhesion failure sooner, then the tap tests could have detected the problem" is simply not true. The paper also shows that the particular blade had been inspected TWICE by tap testing and had also been visually inspected within 80 hours of the blade failure.

You also state: "I don't think the FAA would act on something that wasn't more directly linked to the cause of an accident or incident. Interpretation of a failed joint coming from the NTSB would get peoples' attention." I suggest you google A08 25-29 dated June 9, 2008.

The FAA is paying attention. As a direct result of this investigation, FAA Advisory Circular AC 20=107 was updated to address adhesion failures, and at a meeting in Salt Lake City in July 2014 the FAA announced it was proposing a new AC to specifically address adhesive bond issues , in particular on how to prevent adhesion and mixed-mode failures.

 

[Sparweb]

This one?

7...Romeyn, A., Engineered System Failure Analysis Report: Main Rotor Blade Fracture Robinson R22, VH-OHA, OCCURRENCE 200302820, 2 Nov. 2005, Appendix B, Items 48 and 49.

It is from the references of your paper.

Google gave me this link:

http://www.atsb.gov.au/publications/investigation_reports/2003/aair/aair200302820.aspx

If I was looking at the wrong accident report, then I am sorry.
There have, unfortunately, been MANY R22 blade failure accidents. Far too many.


STF

 

[blakmax]

The information referred to is in the table on page 102. It is interesting to note that in the report it was concluded that the disbonds in the root fitting did not play a part in the failure. In reality, the disbond would have caused a significant change in load distribution at the very fastener which cracked, so I don't agree that the failure was unrelated to the disbond.

I also note the previous comment "The blade simply wasn't designed for damage tolerance." Then why does the manufacturer give acceptable defect sizes for the tap test? I know the blade was certified on a safe-life basis, but safe-life fatigue testing will never interrogate the resistance of the interface to in-service environmental degradation.

Thanks for the feedback

Blakmax

 

[WKTaylor]

Blakmax... A discussion about Robinson helo main rotor blades was held ~2004 Thread

http://www.eng-tips.com/viewthread.cfm?qid=106740

Main rotor blade failures. At that time fractures appeared to be the main concern.

A few off-the wall comments, this thread.

I've been accomplishing adhesive bonded repair for years. My primary resources for repairs were DAC PABST [Primary Adhesive Bonded Structures Technology] testing/design documents. Are You familiar with the series of documents related to the PABST program? They have MANY rules for design/fabrication/repair/bondline-protection You have not mentioned.

They developed/enforced rules for bondline design; cleanlieness; PAA surface prep [aluminum]; adhesive bond primer application [exceptionally thin]; film adhesive tack-application on sheet metal with minimal voids; installation of similarly prepared skins/doublers/fittings/honeycomb; proper cure parameters [heat, vacuum debulk pressure application, etc... based on adhesive variables, etc]; Post bond physical, dimensional and non-destructive tests... and hot-water immersion tests [for leaks]; over-coating EVERY exposed bondline with epoxy primer and sealant [improved environmental durability], etc. NOTE. Many bond design elements included 50-to-100xT(min) over-laps and highly tapered material thicknesses or scalloped [wavy] edges to reduce edge-peel and improve stiffness parallel to the lap/joint... and increasing edge length.

Bond quality must be rigorously enforced. One miss-step in "the process" and WHAM!!!... failure prevails. One time I was tasked to re-qualify out-of-date very expensive film adhesive. The first time around the lap-shear failures were all-over the map and usually mixed-mode +/15% or more: so I shot down that batch of adhesive. A couple of weeks later the shop chief begged me to try requal that same adhesive again... he was desparate no adhesive stock was available. The second time, the lap-shear failures were within +/-2% and consistent cohesion failure modes... but still fell fell ~5% below minimum allowed. At that time I noticed something exceptionally important: The second technician was very precise and careful in his handlling and approach to the adhesive bonding processes [even for simple lap-shear coupons]. I realized that the first technician was NO WHERE NEARLY as thorough in is approach to the process as the second tech. At that point, I still had to DQ the film adhesive; however I certified the second Technician as 'the only one' to-do [or to lead] all critical follow-on adhesive bond work... and DQ'ed the first tech from ever touching the bond-process without adult supervision.

When I worked with a foreign entity doing USAF Depot maingenance, I paid attention to their commercial and their govt military parts fabrication. They made many sets of rotor blades for OH-6 local manufacture helos. Blade disintegration during heavy-weight lift-offs on hot days destroyed 3-acft before they were all grounded. Found-out that the blades were being assembled with 180F service-temp adhesives, in-lieu-of 250F+ service temp adhesives mandated by Hughes. The blades were painted all-black [top/bottom] for low Vis... hence blade skin temperatures ran close to 170F just sitting in the sun on the ramp on a summers-day... at high ambient humidity. Stresses/vibration during lift-off combined with heat-reduced strength of the adhesives caused the blades to fail in a rapid creep mode during run-up and initial pitch angle lift change. IF I recall correctly they disbonded along the trailing edge which caused a massive assymetrical drag and vibration load as one or more blades peeled/rolled... tumbling the helo.

Aluminum Honeycomb is beatifully strong/light... when assembled correctly. However the least contamination of foil edges will result in no attached adhesive fillet... but thats how it will look on X-ray/US images due to the non-wetting-thickening of the fluid adhesive along the foil edgs!!!! Weird how bad adhesion can look like good adhesion edge-on. DESTRUCTIVE TESTING of random samples/parts/coupons during fabrication is a MUST!

I noted limited adhesion-failure detection success using the Fokker Ultrasonic Adhesive Bond-Test methods per MIL-STD-860... probably why it is no-longer used. Metal-metal seemed OK, metal honeycomb was never satisfactory.

Technicians performing tap-test of homeycomb-aluminum structure have to be specifically trained and equipped to hear the fine difference between a solid hit... and a tinny semi-solid hit. In the 1980s and 1990s my hearing was up-to the challange... not so now since I have significant hearing-range loss. Also, a great tap-test should also include a firm touch-feel test for softness or skin loosening.



Regards, Wil Taylor

o Trust - But Verify!
o We believe to be true what we prefer to be true.
o For those who believe, no proof is required; for those who cannot believe, no proof is possible.
o Unfortunately, in science what You 'believe' is irrelevant. ["Orion"]
o Learn the rules like a pro, so you can break them like an artist. [Picasso]

 

[WKTaylor]

NOTE.

If I recall correctly, MOST of the PABST documents are available on line at DTIC

http://www.dtic.mil/dtic/

search: 'PABST' or 'primary adhesive bondted structures technology'

Regards, Wil Taylor

o Trust - But Verify!
o We believe to be true what we prefer to be true.
o For those who believe, no proof is required; for those who cannot believe, no proof is possible.
o Unfortunately, in science what You 'believe' is irrelevant. ["Orion"]
o Learn the rules like a pro, so you can break them like an artist. [Picasso]

 

[tbuelna]

Reliance on the ability of NDI processes to detect a certain minimum size/shape of flaw, both during manufacture of the component and during its service life, in order to ensure structural integrity of flight critical structures, is not specific to composite parts. It has long been common practice with metal parts subject to fracture analysis and fracture control planning. The fracture analysis of metal parts considers what the minimum detectable defect size/shape is based on the particular NDI processes used to inspect the part.

In this case it does not seem that the capability of the NDI process is the problem. Rather it seems more likely that the issue is how carefully the NDI process was used. We should not criticize the aircraft industry's reliance on controlled and validated processes to ensure quality and safety. This approach has been shown to be very effective when applied properly. If it can be established that the specific QA process and analysis used for qualification of the part in question is not adequate, then take another look at how that can be improved.

 

[blakmax]

Thanks for the comments tbuelna but the issue is not how well an existing technology is applied. It is more about the adequacy of the process itself.

Current FARs require demonstration of static strength, fatigue resistance, damage tolerance and resistance to environmental effects. The real key to this issue is that all of the regulatory framework is based on the assumption that the bond will maintain sufficient strength to always fail by cohesion, where the adhesive retains a high level of strength and the failure is by fracture of the bulk adhesive layer. Hence, to demonstrate compliance with the regulations, manufacturers would demonstrate static strength by following the "building block" approach detailed in AC 20-107. This would be backed up by a full scale test. Next, they would demonstrate durability by undertaking fatigue tests. The damage tolerance tests would demonstrate that the structure could sustain limit load with known defects and usually these were based on artificial defects as I have previously discussed, and I also pointed out that this approach is deficient for failures other than cohesion failures. I have also pointed out that almost every manufacturer evaluates environmental effects by moisture conditioning specimens, and while this adequately represents the effects of moisture on the bulk adhesive properties, such tests do not interrogate the environmental resistance of the interface between the adhesive and the substrate to which it is being bonded.

I stress that the current regulatory framework is specifically focused on one failure mode: cohesion fracture of the adhesive and this is appropriate for post-production evaluation of bonds because it can detect defects and assess their size against the tolerable defects established by damage tolerance analysis. Where this framework breaks down is that by far the most frequent failure modes in service are by interfacial degradation which causes mixed-mode and adhesion failures at much lower bond load capacity than exhibited by cohesion fracture of the adhesive. So the regulations assume a high strength failure mode, but practical reality clearly shows that failures can occur at loads well below limit load, especially if the bond exhibits porosity or if the overlap length is insufficient.

The whole purpose of the paper is to show that current NDI methods are absolutely suited to finding disbonds after production where there is an air gap. The reality of service inspection is that almost all of the defects found in service are due to interfacial degradation, not cohesion fracture so the current regulatory framework is not capable of assuring continuing airworthiness because by the time an interfacial (weak adhesion) disbond occurs, the adjacent adhesive interface is already degraded well below the strength that would be exhibited by a cohesion fracture failure. By the time NDI can find a disbond, the adjacent bond is already well below strength. Now if these conditions occur in a bond with a very short overlap length, my paper shows that the bond load capacity can be significantly compromised well before any disbond actually exists. Therefroe bond failure at a low load can occur even without any detectable disbond being present.

So the question is not how well an inspection process is implemented. Even the most competent NDI technician will not find any defect under conditions where the bond may actually fail in service.

Will Taylor, I am very much aware of the PABST program, because I think that this program was fundamental to developing at least the basics of recognising and dealing with adhesion failures. Their guidance on PAA anodising was really ground breaking, but even this work was based on evaluation of bond strength tests. Later work referenced in my paper at Reference 14 has identified that testing based on ASTM D3762 wedge tests (but with modified acceptance criteria) provides a valid short-term test which identifies processes which provide adequate resistance to bond degradation in service. RAAF and USAF experience quoted in this reference shows that where processes validated using this test with modified acceptance criteria there were almost no bond failures for on-aircraft repairs. I continue to be astounded that even last year I found a repair procedure from a (very) large European manufacturer for a helicopter structure where they: 1. used scuff-sand and solvent clean for metallic structure (I thought that even Noah didn't use this on the Ark!!!!) 2. they cure adhesive bonded repairs at 250F (120C) on structure (irrespective of sub-structural heat-sinks) using a single heater and a single temperature sensor. How do they know if the adhesive has seen a full cure cycle, or if the adjacent laminated (and not dried) composite has not been heated to a temperature which causes delamination????

My paper specifically addresses bonded joints rather than sandwich structure which has it's own unique failure issues. see

http://www.adhesionassociates.com/p...P Nations, Aging Aircraft Conference 1998.pdf

I'd be happy to start/participate in a thread on repair of sandwich structure if you like? I continue to be gob-smacked (Australian term for totally surprised... gob=mouth...) by what passes for "approved" repair methods for bonded repairs on aircraft. I seriously think that there needs to be an FAA AC that outlines the minimum standards for aircraft repair methods.

Regards

Blakmax

 

[Andries]

Hi Blakmax and other Gents,

Thanks for a very informative discussion on an important topic.

Regards,

Andries