PA28/32 Wing Spar AD - Show me the data!

[Eric Pauley]

Again, not knowing statistics, I could be way off base, but perhaps part of the problem is that the "signal" is so weak. True results outstrip False results by an order of magnitude, making individual False results strongly influential compared to individual True results.

But the "problem" is that even though "total False" is small in comparison to the size of the tested fleet "total True", there have been dozens of spar failure events so maybe it's hard to just ignore/accept them without trying to understand why they're happening.

It is unfortunate that we don't have data from other aircraft designs to try to determine if PA28 spar issues are an outlier with respect to planes of similar design and usage.

I've made substantial changes to the original post based on your (and others') feedback. Let me know if this helps clarify things!

 

[Half Fast]

I've made substantial changes to the original post based on your (and others') feedback. Let me know if this helps clarify things!

MUCH better! Thank you!

It does make me wonder what the basis was for the original time factor in the AD. I think you've shown that factored service hours are a weak basis, at best, but then the question becomes whether all planes in the series should have ECI on some regular basis without regard to FSHs.

I suspect that the original AD was mostly a compromise attempt to ease panic. It seems likely that the FAA wanted many more, possibly all, planes to receive the inspection, and Piper made an argument to scale that back with the FSH criteria.

Do you plan to share your analysis with the FAA and with Piper?

 

[Eric Pauley]

MUCH better! Thank you!

It does make me wonder what the basis was for the original time factor in the AD. I think you've shown that factored service hours are a weak basis, at best, but then the question becomes whether all planes in the series should have ECI on some regular basis without regard to FSHs.

I suspect that the original AD was mostly a compromise attempt to ease panic. It seems likely that the FAA wanted many more, possibly all, planes to receive the inspection, and Piper made an argument to scale that back with the FSH criteria.

Do you plan to share your analysis with the FAA and with Piper?

Interestingly, Piper originally wanted a 5000h TIS requirement in the old SB, and the FAA created FSH for the AD. Piper has since proposed 2:1; who knows what the FAA will propose next.

I definitely plan to create an extensive write-up and send it to the FAA/Piper. If anyone is interested I would be happy to receive continued feedback, especially from those with engineering credentials that could lend credibility (at the expense of beating a dead horse, I am a statistician and not an engineer!!!).

 

[TCABM]

This is absolutely correct. To compute statistical power (i.e., how confident we can be in the negative result) we'd need to hypothesize some other distribution. Things get very tricky very quickly.

Your problem is you don’t know if the sample is representative of the whole. A little more rigor in bucketing the data might be useful.

We can’t infer x thousand hours is significant, nor can we infer the ECI is a valid predictor of actual failure.

I get the factor service hours tries to account for severity of use but it’s just another math-driven assumption.

What we know is the fleet isn’t breaking up in flight at the same rate as ECI no-gos were documented. That leaves us with either random dispersion or another factor isn’t accounted for. I vote the latter.

 

[hindsight2020]

That's the thing, nobody needs to "show you the data", you're captive audience. I mean this with due respect, that TLDR bowl of numbers-salad you proferred is little more than impotent rage. Or are you the guy who unironically thinks the NPRM cOmMent pEriOd democratizes your participation in this hobby? *plastic legs break under chair*

Look, the details of this nonsense were hashed out ad nauseam 4 years ago, including a regurgitation of the points you're attempting to make in this thread. Yes, it's largely nonsense on the technical merits, to say nothing of empirical reality of zero losses since, but it's settled for you nonetheless: You're gonna get a life limiting final to that AD or worse, a recurring intrusive inspection. What people missing the ball don't get is that, this is all, and always was about ERAU. A... syndicate (I'd use different words but I don't want to open myself to libel) that doesn't own a single Arrow anymore mind you. Alas, that overpriced dino engine did me a favor last year, and I don't look at gift horses in the mouth.

fwiw, in terms of actual day to day safety and ease of ownership/dispatch, I'd actually prefer a life limit than the recurring eddy inspection, if you understand what is involved in that NDT. Which was my objection to the interim in the first place, and the reason I refused to voluntarily participate (well, that and moral hazards). That assembly was never meant to be recurringly perturbed, but to each their own. Good luck going forward on -28/32 series ownership.

 

[Half Fast]

Without significantly more in-flight breakups (God forbid) it will be tough to show any meaningful correlation. We don’t even really know whether ECI failures would have eventually failed in flight, but the cost of finding out is too high.

OTOH, the upside is that I’m even happier with my baby Beech.....

 

[Klaus M]

Your statistics support the theory that Piper produced a number of inferior spars. The percentage of spars presumed cracked from the ECI is low. So, what's the odds of ERAU having a second PA-28R-201 built in the same year also exhibiting cracks? Yet, other spars in the same fleet didn't exhibit cracks.

This quote below is copied from http://www.kathrynsreport.com/2018/04/piper-pa-28r-201-arrow-iii-embry-riddle.html and shows evidence that not all PA-28s or PA-32s are having the cracking wing spar problem.

Examination of the left-wing main spar of another Piper PA-28R-201 owned and operated by the same operator as the accident airplane; manufactured in the same year with a serial number two previous to the accident airplane; and with similar hours, cycles, and operational exposure found a similar fatigue crack that initiated near the lower forward corner of hole LC-1.

Eric, Your 2002 was built during a time period that Piper was still trying to show profits and climb out of dept. I would suspect that your spars could have a similar construction as the two ERAU 2007 aircraft.

I've seen a number of PA-32/28 accidents where the wings broke off during a very hard landing or collision. The spars would bend and crack during the failure. None of the aircraft exhibited just a crack without bending and distortion also. The metal temper process is precise and a little sloppiness can cause too soft of metal that will bend easy and too hard will cause brittleness. It's a fine line getting it right.

 

[Eric Pauley]

Your statistics support the theory that Piper produced a number of inferior spars. The percentage of spars presumed cracked from the ECI is low. So, what's the odds of ERAU having a second PA-28R-201 built in the same year also exhibiting cracks? Yet, other spars in the same fleet didn't exhibit cracks.

This quote below is copied from http://www.kathrynsreport.com/2018/04/piper-pa-28r-201-arrow-iii-embry-riddle.html and shows evidence that not all PA-28s or PA-32s are having the cracking wing spar problem.


Eric, Your 2002 was built during a time period that Piper was still trying to show profits and climb out of dept. I would suspect that your spars could have a similar construction as the two ERAU 2007 aircraft.

I've seen a number of PA-32/28 accidents where the wings broke off during a very hard landing or collision. The spars would bend and crack during the failure. None of the aircraft exhibited just a crack without bending and distortion also. The metal temper process is precise and a little sloppiness can cause too soft of metal that will bend easy and too hard will cause brittleness. It's a fine line getting it right.

Let's look at what the data tell us. Here's failure rate with CIs by year (in 5-year buckets):



Note: this does not account for the fact that some year buckets may tend to have more hours:



Picking out a trend in the ECI failure rates would basically be speculation given the large error bars.

 

[Eric Pauley]

Let's look at what the data tell us. Here's failure rate with CIs by year (in 5-year buckets):

View attachment 131098

Note: this does not account for the fact that some year buckets may tend to have more hours:

View attachment 131099

Picking out a trend in the ECI failure rates would basically be speculation given the large error bars.

Interestingly, the only 2020 aircraft to be inspected had 3300 hours, all commercial, and the right front bolt hole failed.

 

[Klaus M]

Reviewing the documents from the link in your original post under "Source Files" and assuming, that's the only data you're working with. How can you tell if those are all different aircraft?

I was involved with dozens of commercial PA-32s that where first in line to be inspected. The first round of inspections had over 60 percent failure. A couple months later it was learned that the holes needed dressed before doing ECI. The second round of inspections had almost 90 percent passing. The third round of inspections 99 percent of the bolt holes passed the ECI. Each time the inspection was performed the NDT Specialist had to report each and every ECI (separate report per aircraft one result for each hole). Many aircraft had 4 ECI hole reported three times. Possibly, Twelve ECI hole reports for one aircraft.

Out of the 50 plus aircraft that I'm knowledgeable of three wings where replaced in total. Those aircraft are operating mostly in Alaska with quite a few in Washington State. All the Cherokees with +30,000 hours passed.

 

[Eric Pauley]

Reviewing the documents from the link in your original post under "Source Files" and assuming, that's the only data you're working with. How can you tell if those are all different aircraft?

I was involved with dozens of commercial PA-32s that where first in line to be inspected. The first round of inspections had over 60 percent failure. A couple months later it was learned that the holes needed dressed before doing ECI. The second round of inspections had almost 90 percent passing. The third round of inspections 99 percent of the bolt holes passed the ECI. Each time the inspection was performed the NDT Specialist had to report each and every ECI (separate report per aircraft one result for each hole). Many aircraft had 4 ECI hole reported three times. Possibly, Twelve ECI hole reports for one aircraft.

Out of the 50 plus aircraft that I'm knowledgeable of three wings where replaced in total. Those aircraft are operating mostly in Alaska with quite a few in Washington State. All the Cherokees with +30,000 hours passed.

Interesting point. Are you able to see evidence of duplicate report lines in the tables from reports that you're familiar with?

 

[MRC01]

Since statistical tests are a random process, it's difficult to say for sure whether there is "some evidence". All we can say is that, if there were no difference, we'd randomly see evidence at least this compelling 1 in 3 times. Such statements are patently disatisfying, but such is life with statistics...

Agreed. That's much weaker than the common 95% / P=5% standard, but it's not nothing. If you had to flip a coin, at 67%, it's more likely than not to be real, so if you guessed there was a non-random influence at play you'd be right more often than you'd be wrong.

 

[Tantalum]

This was cool.. Thanks

 

[Klaus M]

Interesting point. Are you able to see evidence of duplicate report lines in the tables from reports that you're familiar with?

Out of the 84 PA-32-300s listed none of the total times posted match any of the 4 aircraft reports in my records. The idea of a 50 year old Cherokee doing the ECI with less than 5000 hours doesn't make any sense to me. A number of aircraft listed only have approx. 3000 hours. Not many have over 15,000 hours. Most of the commercial aircraft that I know are +15,000 hours and many over 20,000.

If I owned a Piper affected by the A.D. I would contact owners of flight training aircraft. Ask them if they would share their ECI results. How many of them replaced a wing?

You can readily acquire the aircraft owner's information by searching https://registry.faa.gov/AircraftInquiry/Search/MakeModelInquiry . Another resource is https://www.airframecomponents.com/ . They did many of the spar replacements. The aircraft I know that had wing replacements acquired them from salvage yards like https://suttonaircraft.com/ & https://www.wentworthaircraft.com/ . If you decide to do a little calling around it would be great to hear what you find out.

 

[WDD]

It's always about story telling.

Another quote attributed to Einstein that rings true but probably wasn't really said by him:
"I've reached a conclusion, and if I had more time I would give you the simple version of the answer".

It takes as much time or more to simplify the answer / conclusion as it does to find it.

Nice work volunteering your time to dig into the data like this.

My career was in Market Research/ Competitive Intelligence. Most of the battle was in getting the client to define in one sentence what the real question was, what they really wanted to accomplish.

1. Are ECI failures consistent with fatigue (which is additive and would be consistent with life-limiting components), or with random occurences such as hard landings, which suggest recurrent inspections but are not consistent with life limits.
2. Are Factored Service Hours/Commercial Service Hours an accurate depiction of risk? What is the relative hazard of commercial and non-commercial service hours?

Give the direct answer "Yes or No", within no more than one sentence. Then maybe one chart to depict why.

THEN in the detail section you can have all the data / detail.

Stories - tell them what the story is about, the answer, a hard fought very short answer why, and then people will have a frame work to understand all of your details.

 

[Hang 4]

And are ECI failures consistent with a risk of the wing falling off? I go back to there have only been 3 incidents since ever. There was a lot of rumbling about the EARU Arrow being mistreated. The pipeline patrol one back in the 80's was ridden very hard. While the statistical analysis is interesting, the real question remains, is the ECI test indicative of an actual risk?

 

[Eric Pauley]

And are ECI failures consistent with a risk of the wing falling off? I go back to there have only been 3 incidents since ever. There was a lot of rumbling about the EARU Arrow being mistreated. The pipeline patrol one back in the 80's was ridden very hard. While the statistical analysis is interesting, the real question remains, is the ECI test indicative of an actual risk?

Unfortunately this isn't a question the data can necessarily answer. We can say that many more spars failed ECI than fell off.

 

[Half Fast]

Unfortunately this isn't a question the data can necessarily answer. We can say that many more spars failed ECI than fell off.

The worrisome thing will be when spars that passed ECI fall off.

 

[WDD]

The worrisome thing will be when spars that passed ECI fall off.

Begs to dig to the core questions:
1) Is there a problem?
2) Is "yes", does the proposed solution fix the problem?

 

[hindsight2020]

While the statistical analysis is interesting, the real question remains, is the ECI test indicative of an actual risk?

When everybody freaked out and ran to do these tests even on a voluntary basis (aka if the factor hours didn't call for one), there was a rash of panic when folks started to get false positives due to lack of cleaning of the hole area. Second tests after cleaning where the results were negative. ruh roh, mask off moment. Credibility shot.

Then there was another subset, where the destructive nature of the prep (the bolts require replacement) damaged the attachment, and at least anecdotally on the type cult site one person had to replace a wings on account of the attachment being damaged as a result of the perturbation, positive reading. CeRtiFied mX FTW.

The real irony is that as these attachments get perturbed on a reccurrent basis at a national fleet-wide basis by govt mandate, the very outlier crack initiation they were trying to find, will be induced by said maintenance imprudence in the first place. BL, maintenance induced failure is a big blind spot in this space.

 

[Lindberg]

It looks like the likelihood of failure increases with hours. So why doesn't that support a life limit?

 

[Eric Pauley]

It looks like the likelihood of failure increases with hours. So why doesn't that support a life limit?

Imagine that every hour you fly you roll a 500,000-sided die, and if it lands on 1 the spar fails. Would you (a) check the plane at some interval to see if you rolled one recently, or (b) quit flying after 12,000 rolls?

 

[TCABM]

Imagine that every hour you fly you roll a 500,000-sided die, and if it lands on 1 the spar fails. Would you (a) check the plane at some interval to see if you rolled one recently, or (b) quit flying after 12,000 rolls?

Depends. What if NDI’ing the wing results in more failures? The hypothesis hasn’t been proven, but there’s a pretty invasive procedure that provides no guarantees that’s been prescribed.

Besides, Piper didn’t design those structures to be disassembled, then re-assembled repeatedly, much less by your average A&P.

 

[Lindberg]

Imagine that every hour you fly you roll a 500,000-sided die, and if it lands on 1 the spar fails. Would you (a) check the plane at some interval to see if you rolled one recently, or (b) quit flying after 12,000 rolls?

If that were the case, wouldn't this graph be flat?

 

[hindsight2020]

Imagine that every hour you fly you roll a 500,000-sided die, and if it lands on 1 the spar fails. Would you (a) check the plane at some interval to see if you rolled one recently, or (b) quit flying after 12,000 rolls?

False dichotomy, because that's not how fatigue crack initiation and propagation works, at all.

Again, the irony is that the action of picking option-a negatively impacts the delta-K of the assembly in question in the first place. So even by your statistician-centering, engineering-ignorant approach to the positing of the question, defintionally option-b would yield a lower risk of dying by loss of wing.

Myopic owners just hate that narrative because opt-b gonad-punches their resale value, as the buyers of an "asset" that didn't have a life limit when they bought it (aka they end up overpaying). IOW, this pearl-clutching isn't about safety at all; people should just drop the pretense.

Lastly, don't mistake my position as simping for option-b either. I'm saying all of this as someone who rejected both -a and -b, by rejecting the need for the interim AD in the first place. One still waiting for ERAU to make me whole on the loss of resale value directly a result of their maintenance derelicition and suceeding moral hazard governmental imposition. I'm being facetious on the latter of course, I'm not actually waiting for a damned thing. What do I look like, a MOSAIC sucker? I digress.

 

[Half Fast]

False dichotomy, because that's not how fatigue crack initiation and propagation works, at all.

True, but I think the OP is trying to make the point that fatigue cracking is not yet supported by the evidence. The data so far could well be random failures, and if so, the dice comparison would be apt.

 

[Half Fast]

If that were the case, wouldn't this graph be flat?

No. The probability of rolling snake eyes at least once in N trials increases as N increases (which the graph shows), but the probability is exactly the same on each individual roll. Fatigue failures would have the probability increasing with each roll.

BTW - ever read the novel No Highway by Nevil Shute?

 

[Half Fast]

Hmmmm......

FAA Issues Airworthiness Directive For Certain Piper Aircraft Models

The Federal Aviation Administration (FAA) issued a new Airworthiness Directive (AD) for specific Piper aircraft models after an unscheduled inspection, prompted by a ground collision with an automobile, revealed a...

www.avweb.com

Do we know that the planes that had wing failures did not have this problem?

EDIT:
Oh, I see this new AD only applies to newer aircfraft, 2018 and later.

I guess Piper has a variety of ways to to have the wings land separately from the fuselage.

 

[Eric Pauley]

No. The probability of rolling snake eyes at least once in N trials increases as N increases (which the graph shows), but the probability is exactly the same on each individual roll. Fatigue failures would have the probability increasing with each roll.

BTW - ever read the novel No Highway by Nevil Shute?

 

[dmspilot]

No. The probability of rolling snake eyes at least once in N trials increases as N increases (which the graph shows), but the probability is exactly the same on each individual roll. Fatigue failures would have the probability increasing with each roll.

If the probability of failing inspection was random the graph would be flat. Not seeing the relation to the dice roll, because they aren't inspecting planes with more hours more times.

 

[Eric Pauley]

If the probability of failing inspection was random the graph would be flat. Not seeing the relation to the dice roll, because they aren't inspecting planes with more hours more times.

CALCULUS ALERT: The failures are cumulative in that each hour could cause cracking that fails the inspection later. The cracking (under the model) occurs with equal probability in any hour regardless of age. In this way the ECI failure probability increases, but the rate of increase decreases over time. This is contrary to fatigue, where the rate of increase in probability would go up as the part ages.

This is known as a Poisson process: https://en.m.wikipedia.org/wiki/Poisson_point_process

 

[dmspilot]

The failures are cumulative in that each hour could cause cracking that fails the inspection later. The cracking (under the model) occurs with equal probability in any hour regardless of age.

I think I see what you're saying. But, even if it is random per hour, doesn't that mean airplanes with more hours are more likely to have cracking, thus are more likely to have a failure? Isn't that the point of the inspection requirement?

 

[Eric Pauley]

I think I see what you're saying. But, even if it is random per hour, doesn't that mean airplanes with more hours are more likely to have cracking, thus are more likely to have a failure? Isn't that the point of the inspection requirement?

Absolutely. The data are certainly in support of inspections. What they do not support is (a) an initial period before inspection longer than the inter-inspection period, or (b) a life limit.

 

[dmspilot]

Absolutely. The data are certainly in support of inspections. What they do not support is (a) an initial period before inspection longer than the inter-inspection period, or (b) a life limit.

Agreed, with the same caveat I gave earlier about the quality of the data and statistic significance.

 

[Eric Pauley]

Agreed, with the same caveat I gave earlier about the quality of the data and statistic significance.

The bane of statistics! As noted it's impossible to determine statistical power without some alternative hypothesis, but generally the sample sizes used here ought to be sufficient that we'd identify a trend of the failure mode varied substantially. Caveat: if there is fatigue cracking but it only begins to appear well past the hour counts observed, it would not be visible. That becomes a question for the engineers.

 

[Lindberg]

No. The probability of rolling snake eyes at least once in N trials increases as N increases (which the graph shows), but the probability is exactly the same on each individual roll. Fatigue failures would have the probability increasing with each roll.

BTW - ever read the novel No Highway by Nevil Shute?

So the failures on the graph are cumulative?

And no.

 

[Half Fast]

So the failures on the graph are cumulative?

And no.

See post #71.

Shute’s book is a fictional tale about fatigue failure in a British airliner. It’s a decent yarn. It was made into a movie starring Jimmy Stewart, but the book is much better.