Service History Based Approach to Damage Tolerance 1

[SAITAETGrad]

Does anyone know a good reference for developing an inspection interval based on found cracks? I had one suggested to me recently with a lot of obvious problems. It's not an approach I'm familiar with as I have minimal experience with correcting service difficulty it DT structures. If possible, I'd like to see some sort of example of how this can be properly bounded.

 

[rb1957]

Look into DND's C130s maintenance ... they fly with known cracks.

Analytically, without OEM data, I think it's almost impossibly. You "can" monitor crack growth (that sounds dangerous in the commercial world) but how do you know the critical crack length (maybe you can assume a conservative limit load, and define the geometry corrections then you can estimate it). If you have some data about how the crack grows, then you can tune a crack growth analysis to match this data.

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

 

[SAITAETGrad]

Hi RB...thanks for the response. In this case it's more about adjusting vs. a fleet than one aircraft. Cracks appearing earlier than expected due to unaccounted loads.

Some obvious challenges are the population statistics, relative assumed initial crack length, etc (actual from service vs. potential from rogue flaw), etc.

I hope my English is passable. I'm from Saskatchewan...so it's tough...

 

[rb1957]

so one plane has rogue flaws or loading ? not sure we can talk sensibly about a population of one ?

Not sure I get where you're coming from (problem-wise, I figure where you are from your handle).

If cracks are appearing unexpectedly then something has occurred that wasn't in the DTA (obviously) !
Has the plane been operated differently ? Are the assumed loadpaths correct ? Has something happened in the build (eg clamping up a clevis or a step) ? but this is heading down a path quite different to your OP.

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

 

[LiftDivergence]

SAITAETGrad,

Do you mean trying to characterize a stress level based on service history and found crack lengths, etc. Or do you mean more of a statistical approach. What is the reference this other person recommended?

Keep em' Flying
//Fight Corrosion!

 

[SAITAETGrad]

Okay, it appears I've confused a lot of people. I apologize.

What I'm looking for is any methodology to work back from a history of service difficulties on a structure detail reported over a number of airframes over time.

The objective would be to determine a bounded inspection process for the structural detail.

The initial analysis wasn't just out in terms of magnitude - a severe parasitic load regime was not anticipated.

I'm seeing a lot of potential difficult questions to answer with such an approach - like what's the residual strength required, how to deal with variance in crack development time to detectable, etc.

That's not to say clever people haven't done this before.

If so, I'd really like to know about it.

 

[GregLocock]

Back on topic. I know that some people have successfully monitored growing cracks in aircraft as a way of predicting/avoiding in service failure. But two things come to mind. The first is Liberty ships. The consequences of a crack accelerating are not good. The second is one of the lessons Feynman drew from the O ring disaster. The time to start investigating is when the system is not behaving like it was designed to, not waiting until a failure occurs.



Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[SAITAETGrad]

Sorry Greg. Not about monitoring known cracks.

The parts affected cracked out & were replaced.

About finding cracks in same parts in same usage over fleet population. I just want to speak about what a rationalized effort working backwards would actually look like.

Belief presented by one authority is you can look at one case then divide life by 4...dvi...repeat 40 times to dsg or next replacement. I'm just trying to inject some reaaon from external source.

 

[Sparweb]

"Life divided by 4..."
That is "Safe Life" design practice. Has nothing to do with DTA. Somebody's barking up the wrong tree.

I doubt this will help, but a few thoughts. Well, just more questions, sorry:

Have you been able to examine the failed part, or given it a detailed inspection to determine the cause and pattern of failure? Metallurgical test to show its alloy and temper actually are as designed? Knowing if it had flaws, modifications, or injuries at some point in its life would be valuable. Evidence of these events could justify the conclusion that the life and inspection currently used on the fleet is OK.

Known unknowns, vs. unknown unknowns: How certain will it be, if a new load spectrum is attempted? Say one is selected, given the known age of the specific aircraft, a scale factor, and perhaps even tweaking the stress ratio R and a Goodman diagram, whatever it takes to get a model that predicts the failure at the time that it actually happened. On this one aircraft, will that be applicable to the fleet in general?



STF

 

[SAITAETGrad]

Sparweb - thank you. The physical problem is understood. Just need to break through the noise. A design change and corrective action needed. A spectrum can be made from test data. Some unfortunately have head in sand or are looking for a quick out or simply dont have the knowledge and rely on the guy established as the specialist. People and money problem hardest part.

I don't need to work back from damage in this case except maybe as validation. Just want to contrast between proposed and a third party approach...and maybe pick up some tools for next time. I am getting an earful that working from service history is the best possible DT data to determine inspections...but have no idea how you would make that work in a bounded way. Looking for something well thought out if it exists. I simply can't blow off the attempted approach of the specialist who others trust.

Not exactly divide by 4...more correctly according to specialist inspect at 1/2 and every quarter after forever. Assurance cracks will be easily found by dvi so no life limit...just keeps going strong until you find one...

 

[rb1957]

If you have crack growth data (crack length and cycles) then you can tune the DTA to this data.

DTA is (apologies if this is olde news) cyclic loading, crack geometry, material props.

Maybe MSG-3 logic is applicable if not doing analytical DTA, though I don't see how pre-mod experience helps for post-mod substantiation ... if that's what you're after ?

you guys out west speakwrite funny ...

maybe you're trying to substantiate the improved design based on pre-mod service experience ? (us guys out east have similar problems ... expressing ourselves !?) Maybe showing that the post-mod design lowers the stress in the crack location ? 1/2 the stress equates to 16x the life (for typical Al structure).

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

 

[Sparweb]

...more correctly according to specialist inspect at 1/2 and every quarter after forever. Assurance cracks will be easily found by dvi so no life limit...just keeps going strong until you find one...

Okay that's in Eastin's playbook. Some of us western cowpokes ain't got much more book-learnin' than that.

STF

 

[rb1957]

basing an inspection interval on a fatigue analysis is not DTA. Fatigue doesn't tell you much about crack growth and nothing about residual strength; although safelife is a reasonable substitute for threshold inspection. If your safe-life is much longer than service life (more than 2 times) you have an argument for "slow crack growth, non-inspectable" which is a USAF term (and not really applicable to civilian operations).

Sure maybe some people will buy a repeat inspection based on fatigue, particularly if you can support this with a story about how (in)significant cracking is to the safety of the airplane (and how features in the structure will limit crack growth, maybe to discrete members and maybe you can show fail-safe); not sure I would without a very warm fuzzy !?

The issue is going to become how detectable is the cracking ? for example external doublers on the pressure vessel skin require NDI.

An option, maybe enough for today and make this ultimately tomorrow's problem (for someone else) would be to strip down the structure at the safe-life (= threshold, cut-off at 1/2 service life) and effectively zero time the installation and rebuild it. Not saying this is a good option !

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

 

[WKTaylor]

Damage on one aircraft should be repaired/eliminated [by replacement] when no other similar-to examples exist.

EXCEPTIONs that I developed, working as a aerospace field service engineer in the 1990s, thus...

I worked with various fighters [squadrons, multiple acft] and was faced with this dilemma on a regular basis.

These particular acft had a mandatory redundancy in primary structures and systems elements and a slow/stable crack-growth requirement. These 'components' [almost] always had very clear guidance in the tech data for cracking, corrosion, wear, etc damage.

However secondary structure and systems components often had poorly defined cracking, corrosion, wear, etc damage limits: Inspect when it is required by tech data; IF you see IT then you must repair or replace IT.

Yeah Right.

A small fleet wide sampling was used to determine if a one-off problem was really a one-off problem... or if the sampling revealed more acft with similar issues... IF SO... then a fleet-wide inspection was accomplished.

During these inspections, the size of each detectable defect was determined/recorded for each location on each acft. See below for applicable examples. This data was tabulated and the damage [by acft/location on the acft] was charted on a spreadsheet... then arranged in worst-to-best case order [worst-defect to barely detectable defect order].

Here’s where the ‘logic’ became specific to multiple highly stressed fighters: most flights [training, actual combat, etc] were very similar [for the fleet], and all/most aircraft had similar flying hours and operational exposure. I ass-u-med that the worst case damage detected occurred on the last flight before it would fail [next flight]... which lead to the obvious plan to minimize effect on the fleet, thus...

Based on this analysis [and back-up simplified stress analysis] I set a max damage limit [below the worst case] for 'unserviceable' = too-close to failure and must be grounded and maintenance action taken.

Damage exceeding my cutoff limit required immediate repair or replacement to attain a fully serviceable condition. This 'usually' affected only a small population of acft, immediately.

Damage less than my cutoff limit was allowed to remain in service with recurring inspections that I set based on relative ratio of damage size to my-damage-limit size... and usually resulted in 0-5, 5-10, 10-20, 20-40, or up-to-100 etc, flying hour limits before mandatory re-inspections for damage growth.

Damage that grew to the cutoff limit required immediate repair or replacement [noted] to attain a fully serviceable condition.

Meanwhile repairs or parts replacements began immediately on a worst-to-best case recorded damage sequence. These repair/replacement actions were accelerated to ensure that 'the problem went-away quickly'... and allowed breathing room to still fly Full Mission Capable [FMC].

NOTE. I rarely worked in a vacuum [very few exceptions]. I always contacted my depot counterparts and laid-out my plan... and sent a few damage specimens [removed from service for repair/replacement] to them for failure analysis... and strongly suggested that the remaining fighter fleet [worldwide] be sample inspected for the same defect. Curiously, we often discovered that wildly unpredicted damaged was limited to an acft production-run [block] or to a particular operating base/environment [etc].

NOTE. Commanders were reluctant to allow unscheduled fleet sampling or fleet-wide inspections, since it ‘wasn’t in the book’ and manpower was scarce and FMC jet-count could drop for scheduled missions... temporarily or critically ‘depending on findings’.

NOTE. For military commanders, I learned that a clearly presented path/goals was vital... and reminded these ‘command-pilots’ [mostly] that these inspections to mechanics are similar in nature to combat reconnaissance for warfighters... had to be done for mission success/safety.

Examples where these judgments came in handy.

CRES bleed-air band-clamps that cracked thru spot welds.

Belly and bay Panels that chattered/wore-out-edges/holes by chaffing/fretting [especially during deployment in a desert environment due to trapped fine-sand].

Plain bearings/extruded hinges that wore-out prematurely.

Roller bearings that wore-out prematurely.

Wiring that was chaffing/abrading against a titanium bleed air duct.

Circuit breakers that tended to corrode the wipe-contacts ‘closed’ in a seacoast environment.

Titanium hydraulic tubing that abraded the wall [to 3/4 thickness] due to Teflon lined clamps in a desert deployment environment due to trapped fine-sand.

Titanium skin panel cracking damage located at the base of the leading edges of both vertical stabilizers.

Pulled/broken-off fastener heads in certain lightly loaded skin fairing panels.

Regards, Wil Taylor

o Trust - But Verify!
o We believe to be true what we prefer to be true. [Unknown]
o For those who believe, no proof is required; for those who cannot believe, no proof is possible. [variation,Stuart Chase]
o Unfortunately, in science what You 'believe' is irrelevant. ["Orion", Homebuiltairplanes.com forum]