How much process validation is needed? 5

[KirbyWan]

Howdy all,

I work for an FAA repair station. We do, among other things, bonded metal panel repairs. These are aluminum panels that have new skins PAA anodized using a process based on BAC5555. Where we can't PAA anodize, we use a non-tank etching process such as Boegel. Details are bond primed and assembled in a clean room using tooling to maintain part contour and cured in an autoclave. This is just to give an understanding of what kind of processing we do.

Currently we run a process validation coupon, usually a lap shear, sometimes with a climbing drum peel, flatwise tensile or floating roller peel, with every autoclave run, and I think this is excessive and would like to create a specification to detail when process validation coupons are required. I'd like to get a sense from the community of what standard industry practice is for process validation. Are we over testing? Is this right where we should be? I know for manufacturing usually one test is done per batch of adhesive film used, which is acceptable because of the higher production rate and high quality of tooling used.

My plan was to have the process spec say something like: "run a process validation coupon with every autoclave run unless Engineering has determined for a repair process that alternate testing requirements are acceptable." We would have a section for specific part numbers of panels that have had a consistent history of acceptable results and use less frequent testing in those instances. Does that sound reasonable?

Thanks for your comments.

-Kirby

Kirby Wilkerson

Remember, first define the problem, then solve it.

 

[rb1957]

sounds like a prolongation specimen that's included in a forging ... to verify the process.

I think you are right to be concerned about monitoring (and recording) your process, as "cooking" composites and/or metal bonding is so dependent on a good process.

another day in paradise, or is paradise one day closer ?

 

[MintJulep]

DOT/FAA/AR-TN06/57 Best Practice in Adhesive-Bonded Structures and Repairs may offer some insight.

 

[KirbyWan]

Thanks MintJulep.

I have seen that document before, but forgotten about it. Time for another read through.

I did want to get a sense of what was actually done by other bond shops too. How much testing is normal. If we bond an acoustic structure with both supported adhesive for the non-perforated skins and unsupported adhesive that has been reticulated for the perforated skins are lap shear coupons needed for both? If we have gotten consistently good results could I reduce the testing to just the supported adhesive? If we use the same adhesive system on a panel through multiple bond runs can I use one test for all of the bond runs and should it be the first or last bond run?

I have in the many years of working here seen only a handful of test coupon failures and feel in general we have a solid process, particularly when performing repairs on our most popular product lines where the experience is deeper.

rb1957,

We do composite work as well and often run an aluminum lap shear coupon set to verify proper cure of film adhesives used during composite bonding. As I say above I feel confident in our overall process and want to reduce the testing to minimize cost while still having a good check to make sure process failures don't get through.

Thanks everyone for their input.

-Kirby

Kirby Wilkerson

Remember, first define the problem, then solve it.

 

[blakmax]

As the author of DOT/FAA/AR-TN06/57 Best Practice in Adhesive-Bonded Structures and Repairs I strongly urge you to add a wedge test ASTM D3762, but note the acceptance criteria in that document because the ASTM 3762 stated criteria in para 10 is totally inadequate and the FAA has a program running with UoU to establish valid criteria.

It is important to understand that the strength of an adhesive bond depends directly on the chemical bonds formed between the surface you have prepared and the primer and/or adhesive. It is comparatively easy to generate short-term strength. Indeed, a simple scuff-sand and solvent clean will often generate adequate lap-shear strength if tested soon after manufacture, but if left for an extended period the results would be totally unacceptable. The mechanism of loss of strength is that the oxide layer formed during preparation hydrates (Al2O3 -> Al2O3.2H2O) and in the process the chemical bonds formed during cure will dissociate to enable the hydration to occur. The important fact is that the lap-shear test for QA will NOT interrogate the resistance of the oxide layer to hydration. The wedge test is by far the most effective test for long-term bond environmental durability.

Lap-shear will provide an indication of inadequate adhesive cure, a condition where gross contamination has occurred or where materials or processes are deficient, but only the wedge test will tell you if your processes for preparation have been effectively implemented.

Now as an extension of this discussion, one of my concerns has been the use of coupon tests at all. It must be realised that coupon tests only IMPLY quality; the do not guarantee it especially for bonded repairs. In this case you are using an autoclave and that gives you a fighting chance because the specimens and the part have a high probability of seeing the same cure conditions. The same can not be said for repairs performed using localised heat and pressure methods such as heater blankets and vacuum bags where temperatures and pressure may vary significantly and the problem is far worse when people attempt to use a single heater blanket and a small number of temperature sensors. The results from your coupon test will now depend strongly on where the coupon is located within the repair zone because it may see a totally different cure cycle to the repair itself.

I have for some time been an advocate of Quality Management, rather than Quality Assurance. In Quality Assurance, the repair is said to be good if a number of predetermined conditions are met (typically lap-shear climbing drum, T-peel NDI etc.) But that process relies very strongly on the validity of the duplication of the exact processes and cure cycles, and as stated for in-situ repairs that duplication becomes tenuous. Further, I have seen a technician bust his buns to get the QA specimen correct and take short-cuts for the actual repair because he knew that if he was to go to the coast for the weekend, that specimen had to pass.

In contrast, Quality Management is based on knowing that the processes have been properly validated before you use then and that your management of the cure cycle is undertaken in a sufficiently rigorous manner that you have a very high degree of confidence in the result. It relies heavily on the use of quality materials and technicians who are guided by written processes that have been validated prior to use and the understanding of the technician of the consequences of failure to follow the approved procedures, and written documentation to record critical process details. If you know the processes are valid, you know the materials are within spec, you know the technicians are proficient and the equipment is capable of providing sufficient data to enable a reasoned assessment of the cure cycle, then if the processes are done correctly then every coupon prepared would always pass.

In the organisation where I previously worked, we introduced Quality Management, and the sole test we undertook was the water break test immediately after solvent degreasing. We did not use any coupons cured with the repair. In 15 years with about 4500 bonded repairs, we only had three cases where the bond failed at the interface and a review of the quality records soon identified the short-cuts taken by the technicians.

I'd like to see some document (maybe and Advisory Circular?) which outlined the minimum conditions for repair processes. I am aware that CACRC have been developing such a document for autoclave based repairs, but what is really needed is a similar document for non-autoclave repairs. That document should be used as the basis for development of aircraft repair manuals. Maybe then we will get away from the rediculous approved repairs that are out there. (Scuff-sand and solvent clean, single heater blanket on complex structure, one thermocouple, no moisture evacuation prior to heating composites and sandwich structure to 100C (212F).

Regards

Blakmax

 

[SWComposites]

Agreed. Process "control" specimens are really not and are just a waste of time and money. Much more important and cost effective to have detailed process specs, in-process controls, and data collection and analysis (SPC).

 

[KirbyWan]

That's great input Blakmax.

How general can the Quality Management process be? Am I qualifying an aluminum bonding process using our PAA prep, a particular bond prime and adhesive system and a specific cure cycle, or does it need to be referenced to a specific part? One of the issues is I can spend some time to qualify a repair process to our most common product lines, but we get a large variety of other structures on an infrequent basis. If I can qualify the process and eliminate unneeded testing that would be great.

I've never been a fan of the wedge crack test, but my experience was with the process defined in BAC5555. Putting the crack test coupons in a hot, wet (120° ±5° F 95-100% rh) environment for 60 to 75 minutes seemed too short of a period to cause an effect and I had never had a test fail. I kind of want a test to fail once in a while to let me know that what I'm using to validate my process will catch a bad process and I never got that sense from this process. Perhaps we just have a good process, or perhaps my perspective is wrong and I need to revisit this.

Thanks for your post,

-Kirby

Kirby Wilkerson

Remember, first define the problem, then solve it.

 

[blakmax]

The ASTM STD D3762 and BAC 5555 both have the same problem in that the exposure period and acceptance criteria are inappropriate. If I had results that were 0.5 in. growth in 1 hr of exposure, I would consider that a disaster. As stated in the DOT/FAA document, reliable results require a maximum growth of 0.20 inches in 24 hrs exposure and 0.25 in. in 48 hrs exposure. As well, there should be less than 10% interfacial failure. If these criteria are applied as part of the process qualification program then only valid processes will be used and experience shows that such processes do not disbond in service. If you use these criteria you will soon show that scuff-sand and solvent clean, Pasajel, HF etch and alodine are not appropriate processes if you want the repair to last.

The urban myth I heard is that BAC 5555 criteria were set based on the fact that BAC could "never" produce a bad bond and therefore they set the standards to suit.

Incidentally our process used a grit-blast and silane coupling agent, and we never used a primer.

Regards

Blakmax

 

[tbuelna]

Further, I have seen a technician bust his buns to get the QA specimen correct and take short-cuts for the actual repair because he knew that if he was to go to the coast for the weekend, that specimen had to pass....

That's the problem in a nutshell. Unless the coupon is prepared exactly like the related component, then any results from testing of the coupon may not be representative of what exists in the finished component.

These co-cured coupons don't validate the process itself. The "process" is written set of instructions for performing a specific work task. The coupon only serves as a final QA check to validate that the process was complied with. The reason a coupon is used is because it is not practical to perform a destructive inspection on the component itself. What can be more important is thorough documentation/verification that each step in the process has been performed correctly. This requires the technicians and inspectors to be meticulous in their work.

 

[blakmax]

Some time ago, the ISO standard ISO 2001 on quality processes once described a "special process" as one which was such that the final outcome could not be verified, and adhesive bonding was one such process. The only way of verifying the process is to destroy the bond. The ISO standard directed that special processes should be managed by certification by persons assessed as competent in the processes that they have complied with validated process specifications using materials and processes of known and validated quality. (This is a paraphrase and readers should verify the exact text. I think it was the 2002 version.)

Hence, if the process has been appropriately and accurately validated, the technicians have been appropriately trained in the process and certified as competent, the materials are certified to be of appropriate standards, and the technicians competently follow validated process specifications which have critical process steps verified by rigorous documentation, then the product should be reliable.

Regrettably later amendments to ISO 2001 removed the relevant para (4.9.2 from memory). BUMMER!

In essence, set up your materials acceptance, handling, storage and exposure procedures, validate your processes BEFORE YOU EVEN START CERTIFICATION testing, make sure your technicians are well trained and understand the importance and consequences of process variations and have a quality management approach to verification that all critical process steps are compliant, then the process will be valid and every "QA" test will pass.

In contrast, QA says that if you can manipulate the processes such that your coupon tests pass, it is possible to meet QA requirements yet produce a bad product.

I am reminded of a very good friend of mine from WPAFB Materials Labs who was asked to investigate persistent bond failures from a certain US ALC. By absolute chance as part of the investigation he checked the water-break test aspect of the process QA, and found that every single component passed that test. When the technician was interviewed it was discovered that the technician had on his own initiative actually added an amount of detergent to the water used for the test because that gave him a good result every time. He honestly thought that his role was to get every part to pass, not to reject poorly processed parts!

Regards

Blakmax

 

[KirbyWan]

On a side note. There is a line in BAC5555 that states "Prepare two wedge crack assemblies containing 5 specimens each from 2024-T3 clad and from 7075-T6 bare in accordance with ..." I had one person insist that this meant you should have two wedge crack test panels, each with a 2024-T3 clad plate bonded to a 7075-T6 bare plate. I think I tilted my head sideways like my dog does when i try and explain philosophy to him. I think it's possible to read that meaning from the words on the page, but it seems like such a tortured reading that I wonder how he got there while bypassing the obvious meaning.

-Kirby

Kirby Wilkerson

Remember, first define the problem, then solve it.

 

[KirbyWan]

One other issue I have is with environmental chambers. We have one, it's a huge, noisy and finicky machine that I would like to get rid of. Are there suggestions for a minimal environmental chamber that would meet the needs of processing wedge crack coupons? Wouldn't placing the wedge crack tests in warm water produce an equivalent environment? getting a pot of water to 120° and keeping it there would be much simpler than tracking the heat and humidity in an environmental chamber.

Thanks,

-Kirby

Kirby Wilkerson

Remember, first define the problem, then solve it.

 

[blakmax]

Oh NO! Don't immerse the specimens in water! Our results showed that free water on the test zone led to a dramatic increase in failure rates. Not sure why but I think it may be something to do with the rate of absorption of moisture by the adhesive. The odd molecule grabbed from the atmosphere compared to an abundance in the liquid form.

We expose our specimens either on their sides so water can run off or with the wedge down. I have seen tests undertaken quite effectively using a water bath (with a closed lid) at the required temperature with the specimens in a glass beaker on a rack above the water. A bit more agricultural than an environmental chamber, but still quite effective.

Regards

Blakmax