Boeing lap joints and bonding
 

I have no inside knowledge of the SWA failure but was heavily involved in corrective action after Aloha accident; see [/FONT]http://avstop.com/stories/aloha.html[FONT=Arial] for that story. So I'll make an informed guess of what happened.



Hopefully the problem is confined to Boeing 737s up to LN 288, because these used a bonding process that was found to be defective; the process was relied upon to reduce fatigue stresses across longitudinal lap joints in the fuselage. Aloha happened after the bond failed and skin cracked at the top row of rivets; which were critical because they were countersunk. Fix was to replace all those rivets with button heads. BTW you can spot an old, suspect aircraft by the prominent rivets. This did nothing to aleviate criticality of bottom row. Just postponed the evil day and my understanding is that it is the bottom row that failed on SWA.


For aircraft structures, bonding has special attraction of minimizing stress concentrations and associated fatigue failures that seem inevitable with rivets or welding. [/COLOR][COLOR=#000000]However adhesive bonding is never as easy as it seems. A major reason is difficulty in establishing, substantiating and maintaining process specifications that ensure bond reliability and durability. Not only is the adhesive prone to variability depending on temperature cycles, pressure etc. Perhaps more so, durability depends on surface preparation.

Boeing learnt these lessons the hard way, with a series of embarrassing failures. The worst was the Aloha B737 fuselage failure in 1988. In (Australian) Bureau of Air Safety Investigation Journal, June 1992 I wrote:

“(The fuselage) is made in panels which are typically about 4 metres long and 2 metres wide. These are joined together with rows of rivets. Obviously the skin is weaker where it is drilled for the rivets, so on early B737's Boeing engineers tried to reinforce the joints with epoxy adhesive. It was these joints which failed first and let the skins rip away from the aircraft.
The first Boeing 737 was delivered in 1967, the Aloha aircraft was delivered in May 1969 and by about then Boeing became aware of problems with the adhesive bonding process. The adhesive worked much like a two tube mix used by a home handyman except that the glue was premixed and held on a scrim tape. By keeping it refrigerated the adhesive reaction was suspended. At the right time in construction the adhesive tape was laid between the skins, these were riveted together and the glue then cured as it warmed up to room temperature. That was the theory. In practice the adhesive did not really bond to the aluminium skin, it only bonded to the very thin layer of oxide on the surface of the aluminium. Attachment of the oxide film to the metal underneath was dangerously variable. Also if the scrim was too cold when it was applied it attracted condensation which prevented proper adhesion. If the scrim got too warm it partially cured before it was in place and again adhesion failed. Whenever adhesion failed the rivets and surrounding skin were overloaded and the skin began to crack.

Boeing was not secretive about the problem. The bonding deficiencies and their rectification were discussed in many technical papers in the early 1970's. The whole U.S. industry was embarrassed because the Europeans had been successfully bonding aircraft for 30 years. In 1975 the U.S.A.F. stepped in with a large contract to catch up with the Europeans. It went to Boeing's arch rival, Douglas.

Boeing progressively improved the design of the skin joints and hoped that for aircraft already in service the problem could be controlled with enhanced inspections. From May 1970 onwards Boeing sent the airlines a series of bulletins recommending inspection and sealing of the joints. Cracking and corrosion still went on. Diligent airlines found it and fixed it. Less diligent ones stayed lucky.”Of the tech papers mentioned above specific mention needs be made of “Surface Preparation – the key to bond durability” by Boeing Chief Chemist Corey McMillan. It spells out the evolution of surface preparation and failures along the way.

Problem is that many aircraft are still flying which were built using surface preparations of dubious repute!

Apart from the faulty cold bond technique that led to Aloha, we knew at the time of Aloha that Boeings hot bond technique was better but not completely reliable. A year earlier in 1987 hot bond failures on B747 were addressed by SB 747-53A2279, FAA AD 87-16-13 and Australian AD/B747/59.

At the time, Australia tried but never really got Boeing or FAA to acknowledge the risk of hot bond failures on B737; that hazard still lurks!

Ironically the problem this time has been caused by a mod to the design designed to reduce the risk caused by cracking.

Distance between tear strips was increased from 10 to 20 inches to reduce possibility of cracking propogating from frame to frame as is believed to have happened in the Aloha incident.
This redesign is believed to have led to this problem, hence why only 579 aircraft manufactured between 1993/2000, of which 175 are believed to be above the 30,000 cycle trigger are affected.

Best laid plans of mice & men :rolleyes:

Mr @ Spotty M

I stand corrected. I always associate KPAE with Boeing. Forgot about ATS! Cheers:ok:

there is a list of all affected airlines and aircraft on jacdec.de

Ozaub, thank you for your illuminating and concise post.

Aloha bonding
 

Ozaub

Your assessment of Aloha 243 is close. The problem with the bonding was not the mixing process, because they did not mix the adhesive. They used a film adhesive, similar to double sided tape. The root of the problem was that the material they used was a room temperature curing system, so that they did not have to heat the structure (as the Europeans always did). That saved complex heating procedures for large structures and was much quicker.

The essence of the problem is the requirement to cure at room temperature. Because the material is a pre-mixed tape, it must be stored frozen, or the material will cure off prematurely. When the material was cut is was frozen and when it was applied to the structure it was frozen. As a consequence, atmospheric moisture condensed on the cold surface. Now adhesive bonds depend upon chemical bonds formed at the interface during the curing process. These chemical bonds give the adhesive strength and also durability. In the case of the early 737s and some 727s and 747s as well, the moisture inhibited adequate contact between the ahdesive and the metal, resulting in weak bonds which later failed in service. The disbonding led to higher than expected loads at the fasteners, which when combinded with knife-edge countersinks led to the fatigue cracking at multiple sites, and then to the failure once the small cracks linked up.

Adhesive bonded structures can be very reliable when the processes are correctly validated and implemented. In this case they were not.

If I may be permitted a daft SLF question - I only drive little airplanes with the small wheel at the back - I'm not sure really sure what correction Blakmax is making here. Ozaub described the bonding process in exactly the same way and listed condensation as one of the possible causes of the bonds failing. What'd I miss?

Regardless, I'm getting a great education here - thanks, folks.

Fact is despite the cycles and hrs, none of us can compile the data that resulted in this failure. Starting from the last rivits holding the skin, unknown "g" stress from hard/semi hard landings to turbulence (largest unmeasurable factor), the environment over 15 years "hot/cold/wet/dry". Just to speculate that this piece of skin of this aircraft had a poor run of luck based on the factors above is probably in atleast in one way the truth.

Cycles and hrs are not a factor given this type aircraft. Allthough the crown is least inspected and incinedence like this will likely prove for for more detailed inspection.

Human factors, pilots are often on the fence to report a (semi hard landing), this area of the aircraft is not scrutinized by maintenance and difficult to inspect at the gate. These are factors that will change.

Grounded:

Rubbish. Why do you think we maintain them at set time intervals and life a lot of components in terms of hours or cycles? Didn't you know that the Aloha aircraft was the cycle leader of the fleet?

Did I waste Six years of my life looking at cycles and hours flown then making judgements about component reliability?

readywhenreaching,

I am guessing that list is merely compiled from production line number, the affected batch were specified on the AD, but takes no account of cycles.

Of 579 aircraft in the batch, only 175 are thought to have reached the 30,000 cycle trigger point.

For martinprice. The point is that it was an active decision by the manufacturer to choose a specific form of adhesive which was highly susceptible to moisture problems. It had nothing to do with the mixing of the adhesive. It was therefore not a quality control problem, it was a design/material selection issue.

Hope that point is clearer now. :) Thanks for the interest.

Photos in Post #126
 

The second piece of skin in those pictures seems to have a "hump" or bulge to the middle section of skin between the longitudinal rows of rivets.

Is that likely just due to tension when the adjoining piece tore loose? Or from the disassembly procedure? Or something else?

Re: Photos
 

The fuselage is circular. The 'hump' Is the natural shape of the skin.

The fuselage is not circular, it is ovoid. :ugh:

There are bloody great beams running across each frame at at deck level that support the floor. They are in tension when the aircraft is pressurised and keep it in its shape.

This is the B707 profile it is exactly the same as the 727 and 737.

http://www.lx97.com/FUSELAGE339582.jpg

Sunfish
Yes I am aware of the true shape of a B737 fuselage. In much greater detail than I would care to.
I was sure when I posted my hastily written explanation of the 'bulge' in the cutout lap section that someone would correct me, and you have excelled with that. ;)
Thanks for the detailed post.

Sunfish
 

We use flight hrs and cycles for the bean counters. EG: 26 people show up to work and x amount of work is performed, 13 of them do 125% of the work the other 13 only actually complete 75%, this is a common factor in most workplaces, they all get paid the same and all the bean counters know is that it takes 26 people to accomplish the workload they have.

Don't get me wrong cycles and HRS are a solid rule of measure but not accurate, not absolute.

The answer to your second question is hard to answer. Did you enjoy your job, were you compensated well, were you appreciated for your skills? I am sure you count beans very well, not to belittle your task, just saying that the bottom line is you were payed to SPECULATE based on statistics.

@ grounded spanner and sunfish

Sure, I know the aircraft skin is curved. But to my eye (trained in photography, photo analysis and graphics arts), the curvature is different in that particular piece of metal in the area between the rows of rivets, than it is overall from edge to edge.

I.E., I see, in small scale, the same kind of change in curvature radius as at the "top of floor beam" in sunfish's X-section.

(And, yes, I know the piece of metal pictured does not come from the area of the floor beam :hmm: - which is why I find the change in what should be a smooth, constant-radius rooftop curve interesting.)

In the first picture, the reflection of the overhead lights makes a sharp change in shape as it crosses the lower rivet line. Which indicates a change in the curvature at that point - a distortion of the smooth curve.

In the third picture, end-on, there is also a change in curvature visible between the rivet lines.

I'd diagram what I'm seeing, but pprune doesn't allowed for direct uploading of images. :hmm:

pattern_is_full

Now that I look at the photos on something bigger than an iPhone, and with your detailed description, I do see what you are talking about.

The change in curvature in the first photo (something akin to a slight crease) is very normal with a removed skin section across a rivet line. It is sometimes called 'pillowing'. You can often see on an aircraft fuselage that the skin changes profile at a rivet line. This is because as the aircraft is pressurised, it stretches ever so slightly, and when unpressurised, the stringers behind (only slightly) return to a smaller diameter circle. Nothing to be concerned about as long as it is not pronounced and/or there are signs of distress (cracks, pulled rivet heads etc). That photo looks very normal (ignoring the torn section next to it!!).

Whether there is any extra curvature is impossible to tell. If there were any extra curvature in the skin section it would have to be material that had come from the lap joint. I cannot see any signs of paint distress at the lap joint, which means that the skin will not have moved by more than 1/16 to 1/8 of an inch. Any more than that and the lap would have let go anyway.

I would lay money that that unbroken section of skin contains cracks (significant ones) in the lower row of fasteners, just due to its proximity to the torn section, but that the cracks are not sufficiently large to continue to link up yet. The NTSB will be very interested in how much longer that section had before it would let go.

The stupidity is coming thick and fast now.

Grounded27:


I'm an engineer not a bean counter. The reason we use cycles and hours has nothing to do with bean counting. Those measures are surrogates for the strain history of the part in question and in metals that don't have a yield point (ie Everything except steel) the stress history determines when the component will fracture.

We make estimates based on experiment as to how long a component will last, then we apply safety factors to that estimate. When the aircraft is in service, we monitor all failures and continue testing to refine those estimates based on experience. In more than one case I dealt with, we actually reduced the number of failures by extending the time in service limits for an assembly, see if you can work out why.

In other words, try telling a turbine disk or a chunk of aluminium that its life in service is really infinite, it's just that bean counters make us change them.


Ozaub has already explained that Boeing had a problem with quality control in relation to the adhesive system it once used. That does not mean that the process was defective, it means that Boeing discovered after the fact that it could not precisely control the manufacturing conditions tightly enough to ensure a sufficiently reliable bonding system.


Before you yappers now scream for more of Boeing s blood, by "sufficiently", I do not mean 100% guarantee, I mean to an extent to where any defects are small enough to not result in stress concentration likely to cause severe crack growth during the expected service life of the aircraft which is god knows how many thousand hours and cycles (60,000 hrs? 60,000 cycles?).

The bloody aircraft has not done badly considering, and the safety measures worked as advertised. The only issue for Boeing is that the problem surfaced considerably earlier than expected.

The question is now to work out how extensive the problem is and determine what the best inspection methods are and what the repair schemes are.

I'm also getting fed up with what I term "The pprune effect" whereby every self proclaimed expert offers a solution from within their own expertise.

By that I mean; when there is an aircraft accident, Prune attracts the computer expert who posits the cause as a software problem. The chemist suggests the fuel was faulty. The teacher wonders about the pilot training and the lawyer blames it all on criminal negligence by the designer.

If you have never worked with aircraft, I wish some of you might be a little more tentative in coming to conclusions.

You work with statics, not based on experience but from actual, factual failures. Your experience is 2nd hand to these factors . Granted it's the best we will do from a cost v/s liability factor.

Really now, sounds like you are in vegas betting against the house, your odds are worse than someone skilled at counting cards, now that guy has experience, a mathimatical formula that leaves less liability other than getting busted up by a bunch of thugs.

Inspection processes will improve as will manufacturing processes, at the same time manufacturers will develope products that push limits, use new alloys that will fail as they push limits. Over time we develope new inspections and servicable on wing times.

Grounded Spanner - roger that.

I thought that perhaps when the main torn section cut loose, it applied a little extra tension ("tug") to its neighbor before breaking free, causing the hump. But normal pillowing makes equal sense. In any case, obviously a section of interest to the investigators, if only because it is the other half of a joint.

grounded27

You just might be confusing two different types of 'stress testing'.

See:
1.Stress testing - Wikipedia, the free encyclopedia
2.Stress Rupture & Creep Testing at WMT&R, Inc.

Very true sunfish.

However, if Boeing had implemented a proper safety case, they may well have avoided this mistep.

Aircraft structure is not subject to anykind of risk assesment.:ugh:

Bonding Conditions
 

Having built a couple musical instruments, I came to appreciate that gluing is a complex operation. It took hundreds of years to develop bonding technology for wood -- and today the proliferation of new adhesives is truly amazing.

I do love tape adhesives -- but they don't always work;)

Bottom line -- all adhesives have their particular quirks, some of which take decades to manifest themselves.

Yes, pulling out adhesive tape from the freezer and uncoiling invites condensation.

The amount of condensation depends on the ambient temperature and humidity of the day.

I doubt the Boeing factory floor is an environment where temperature and humidity are precisely controlled.

It may be useful to relate the discovered bond failures to the day they were performed and the temperature and humidity at the time. Are there environmental records of temperature and humidity on the factory floor? We may find ourselves limited to public weather records and then infer from building management procedures what the HVAC would have produced. My experience of building management is that HVAC is managed much more to the dollar sign than anything else:}

Are you kidding?

Everything that we do in designing and certifying aircraft has risk assesment built into it. That's why things break, including acts of God. We do a pretty good job of balancing risks, that's why it is safe to fly.

I'm 100% serious (and correct;)).

A probabalistic, risk based approach is only required during certification of systems and equipment. Structure and handling for example are exempt. If they wernet the 2001 AA A300 accident would have been prevented.

A possibility
 

Depending on type. During heavy inspection lower skins are common to replace, if there are enough skins that need replacement you have 2 options... Replace only a few at a time as not to allow for the hull to torque or often secondary procedure for additional shoring and the hull is shot with a transit allowing for many skins at the same time.

Now if something goes wrong and an aircraft comes off jacks after a multiple skin change, the skin seams could be subject to abnormal stresses. The whole frame being torqued.

That is needless and irrelavent speculation.

You really are serious! but not 100% correct :)

I really can't find fault in your words above, it was only your original statement that I challenged

The idea behind the cert standards is to require a specification of conformance within a specified range. Anything outside of that range is expected to be minimized to practical extent.

Thus a structure may be specified to a stress limit but therafter "managed" to a life limit.

OK, I'll cut you some slack;), but just tell me old boy what the probability of this piece of structure reaching design life target Boeing had to work to?

Can you answer that? Can you?;)

The other advantage of apply an SMS philosphy would be that the earlier failures would have resulted in earlier Boeing action.:ok:

Shell Management -- since you are the one swinging your ISO 31000 hammer at everything, why don't you contact Boeing and then report back to us the answers to your questions.

I didn't mention ISO 31000:2009 old bean.:= That a pretty poor Antipodean drafted standard that is compromised by its application beyond safety.:ugh:

If you were an aviator, you would know that it is ICAO that defines tsndards in aviation not ISO.:=

The affected aircraft gained winglets some time in 2009, is there a possibility that the extra weight of these winglets causes more wing flex which may stress the skin above the original design spec??

Shell Management

Given that they have conducted lots of fatigue tests on similar designs, they would probably statiscicaly expect cracking at about half their target life. Such cracks would be inspectable and a confirmation of their model. Now if the cracks are not easily inspectable then they better half the life expectancy even more and continually adjust their mainteance program based on findings.

Sounds to me like that's where they are.

This thread is deteriorating rather quickly.

Aircraft are designed and built under massive risk management programs that are driven by parameters set by the insurance industry starting with the hull loss probability. That then gets driven down into the design through Failure mode effects analysis (FMEA) right into the design MTBF's of the components - including the hull.

Those programs do not stop when the aircraft design is frozen, but continue for the life of the aircraft in service or until the company goes under, after which someone else has to be found to assume design authority if the aircraft is to remain in service.

We have one person here (grounded) that confuses bean counting with statistical analysis of failures and doesn't understand what a strain history is either.

To put it another way Grounded, why do you think an automobile manufacturer sets service intervals and component replacement schedules for the cars they make and how do they do it?

P.S. If you ever worked on aircraft replacing lower skins, then they must have been pretty old and crappy aircraft because Boeing started paying attention to bilge corrosion - sealing faying surfaces and filleting with PRC 1422G8 for example, around 1975.

Sunfish
 

Really, I was under the impression that it was as simple as keeping a clean driving record. Do you guys get a safe driver discount down under. Quantas has a wonderful record, shame they do not get a discount for it.



Sorry to have offended your profession. I have in depth knowledge of modern corossion preventative programs. CPCP goes way beyond using a variation of tank sealant in joints, I have had factory training by Dynatrol and yes have worked around some old aircraft "quick go wiki it and come back with more slander".

All your statistical analysis is great in general. BUT it is absolutely useless when something fails before you predict it.

To word it another way HRS/CYCLES/ANALYSIS are not worth a damb to the crew and passengers of an aircraft diving for breathable air.

Also events like this allow tools like yourself to change the shelf life of a product, initiate an inspection at an earlier date. I am sure it made you feel like you were doing the world a grand favor. Sir by mentioning you as a tool it is with the utmost respect, persons like yourself create job security for people like myself, we are truely a breed apart.

The only thing exciting in this business are things that can not be predicted, the unpredictable changes the industry.

winglets, sturcture life
 

Winglets: I would look more at the additional lift on the wing in critical conditions, which could be transmitted as bending load into the fuselage. That said, given the speed with which Boeing revised the inspection interval without any reference to modifications, I doubt they are relevant.

Structure life: I am intrigued by the suggestions that structures are not subject to risk assessment. This may have been the case decades ago but I thought all recent new aircraft designs had to be tested to show compliance, the so-called fatigue tests. Maybe the 737-300s, being derived from the 1960s -100/200 didn't need such a demonstration, or service experience was used.

Grounded:


Have you told the aircraft manufacturers, operators and the worlds regulators the wonderful news that all those extremely costly and carefully constructed aircraft maintenance programs they engage in can now be abandoned because failures "just happen"?

Why do you think we bring a particular aircraft in for you to maintain? Do you think we just pick its number out of a hat?


Shell Management, I was taught risk management by Exxon. The starting point for the airline, air traffic control and aircraft manufacturer risk management programs are the acceptable passenger death rates per million miles travelled as well as the calculated cost to the insurers of a fully loaded aircraft crashing in the middle of a block in central London or downtown Manhattan.

Leave the air-insurers consortium out of it := they don't have a hand in managing safety. They enter into it to gauge what the overall statistical risk is and what level of their money to set aside for the aftermath.

And nobody in regulated aviation goes around with managing a known problem against the number of dead bodies.

Risk management can address known problems that are understood well enough (all the potential causal chains in a link) such that the risk of all the links coming together within the time frame that corrective actions are implemented are far less than 1 single catastrophic event.

It's the unknowns that catch us by surprise and this thread subject is just another example of a surprise. But now that it's no longer an unknown a corrective action will be implemented (closing out the risk) in a period of time where only a surprise (unforeseen combination) will result in a catastrophe.

If critical combination in the future turn out to still be unknowns after an investigation then expect forced inspection and limitations forever. If this can't be reliably performed then expect the limitations in operations to be severe.

This is nothing more than managed risk at least far better than the average person freely lives their own life.

Risk Management is based on a theory of what should happen, it is nothing more than financial liability. Reality changes your factors and the variable factor of events we can not predict is reality. I do not understand the argument here? Our environment our reality is certainly unpredictable, thus my point. What should be is a big part of society, what is, is real.