Any interest in the Statistics of engine failures?

hindsight2020 said:
Nope, but since my perception is my reality, 6 bangers fail more than 4 bangers, Continentals fail more than Lycoming, and turbos fail more than NA. As such, I fly behind a Lyco four banger. Since nobody can quantify otherwise anyways, my religion is as good as my science. :D
Click to expand...
As someone who builds engines for a living, I'd say you nailed it for commercial service. The variable is private owners who let their Lyco 4 bangers sit and rust.
 
tyndall said:
As someone who builds engines for a living, I'd say you nailed it for commercial service. The variable is private owners who let their Lyco 4 bangers sit and rust.
Click to expand...


So, in your opinion, what would be the maximum time that you would allow the airplane to sit between actual flights, not short ground runs, which I don't do by the way....
 
44Gort said:
"Three engine failures where the engine ceased to drive the big fan due to faults in things attached to the engine."

Were these engine accessories; like oil pump or were they outside the engine compartment?
Click to expand...

Failed muffler Internals blocking the outlet, Mud dauber nest blocking the carb. throat intermittently after coming in via the carb heat check, fuel vapor lock after a short stop for fuel on a hot summer day.
 
Last edited:
Lycoming O-320H2AD - 70hrs SMOH, sucked an intake valve, apparently the A&P had used the wrong valve keeper.
 
Graueradler said:
Failed muffler Internals blocking the outlet, Mud dauber nest blocking the carb. throat intermittently after coming in via the carb heat check, fuel vapor lock after a short stop for fuel on a hot ssummer day.
Click to expand...


Thanks. :)

You are right. These were accessories failures and didnt need the engine torn down.

gort
 
SteveR said:
Lycoming O-320H2AD - 70hrs SMOH, sucked an intake valve, apparently the A&P had used the wrong valve keeper.
Click to expand...


Just curious, did you get any compensation/warranty money for this?


Sent from my iPhone using Tapatalk
 
Its been pointed out that "Cylinder" failures may not fall into this study.

While its true that many treat Cylinders as accessories, some cylinder failures cannot be predicted.

If a cylinder is monitored by inspection and compression testing, and it is determined that it should be removed/replaced or OH'd, then this shouldn't be considered an engine failure. It was maintenance.

OTOH, if a cylinder fails catastrophically, by losing a heat or sucking a valve, which causes further internal engine damage, then Yes, this could be considered a "Core" engine failure.

Why?, because sometimes the failure of a connecting rod results in the loss of the piston which then can cause the head to fail. Its difficult to determine which happened first. the Head or the Con-rod.



Engine failure:
Make
Model
TSMO/FRM

:D
 
what if a cylinder wasn't installed correctly....and a bearing spins? :eek:
 
SPAJC said:
So, in your opinion, what would be the maximum time that you would allow the airplane to sit between actual flights, not short ground runs, which I don't do by the way....
Click to expand...
Once every couple of weeks or so for an hour. Which works out, coincidentally, to 100 hours/year.;)
 
Maintenance Induced Failures may apply in some cases.

I'll have to parse the data as it comes in.
:)

Engine
Make
Model
Time in Service
 
A discussion about engine failures needs to be split into two groups. One would be pilots/owners who've experienced engine failures and who can provide their perspective as to cause. The second will be guys reciting second, third, to more likely 50th hand accounts and don't really have a grasp on the context or cause of the failure. Both may have merit but I'm far more interested in the former.

I had an engine break a rod in flight. No warning. It pounded holes through the case but the engine kept running. It made prodigious amounts of metal in doing so.
 
tyndall said:
Once every couple of weeks or so for an hour. Which works out, coincidentally, to 100 hours/year.;)
Click to expand...


I usually fly more than that anyways most of the time. So, I should be good. It's been a difficult month this month for free time though (two babies at home), so I was actually wondering what the generally accepted rule was. Thanks!


Sent from my iPhone using Tapatalk
 
I agree that there is substantial discussion of both types, but unless we develop a form which sterilizes the data, there's no way to avoid it.

I'd like to keep it as simple as possible.

"Just the Facts 'mam". --J. Friday

:)
 
44Gort said:
Thats sort of the data Im interested int.

No, there isnt an accessible database where I could pull out the failure data.

This would be a Poll of the POA members.

IF you had a failure, just post the make, model and the hours on the engine when it happened.

Ill collate the data an prepare a report.

thanks.
Click to expand...

OK that gives some kind of numerator but how would you get any kind of reasonable denominator? Or if you're going to exclude non-failed engines completely then it's meaningless.

In other words, say we know that, among engines that failed, they failed after 433 hours on average? What use is that stat without knowing what % of total engines fails at all?

We would have to have everyone report how many hours are on there engines AND whether and at what point they failed. Limiting the survey only to failed engines makes it meaningless to my thinking.
 
tyndall said:
Once every couple of weeks or so for an hour. Which works out, coincidentally, to 100 hours/year.;)
Click to expand...

It works out to 26 hours per year. I think you meant twice a week for an hour?
 
"say we know that, among engines that failed, they failed after 433 hours on average?"

Thats the point. We dont know that.

We do know that engines fail, and that sometimes they fail prior to being overhauled and thier TBO. Sometimes they dont fail until some time after a recommended TBO. that would be good too. But I would guess, that would be even more rare.

Id like to know that number

Im trying not to read more into it than that.

Like they say: K.I.S.S.
 
FYI....if you haven't already done so, I highly recommend studying 14CFR part 33 (Endurance Testing) and get an understanding of how TBO is determined. Like me, I'm sure you'll be surprised. IMHO....it's very subjective and in some cases arbitrary.

Based on how I've done forecasting and predictive analysis using test data for other systems....TBO is quite the opposite. Run an engine for a 150 hours....do a teardown....measure parts....and pick a TBO number. I'd rather the process be driven by data (even something close to statistically significant)....IMHO it ain't.
 
Last edited:
44Gort said:
"say we know that, among engines that failed, they failed after 433 hours on average?"

Thats the point. We dont know that.
Click to expand...

Turns out that we do (even though we didn't know that we did).

http://blog.aopa.org/opinionleaders/2014/03/13/do-tbos-make-sense/

Dr. Ulrich analyzed five years’ worth of NTSB accident data for the period 2001-2005 inclusive, examining all accidents involving small piston-powered airplanes (under 12,500 lbs. gross weight) for which the NTSB identified “engine failure” as either the probable cause or a contributing factor. From this population of accidents, Dr. Ulrich eliminated those involving air-race and agricultural-application aircraft. Then he analyzed the relationship between the frequency of engine-failure accidents and the number of hours on the engine since it was last built, rebuilt or overhauled.

ulrich_hours-300x202.png
Click to expand...
 
That graph is a total waste since it doesn't address total number of aircraft in any sample group, just number of failures. I've seen other reports that dispel the 100 hour engine failure risk as well as the first power reduction failure risk, etc. Guys can find support for whatever opinion they want supported. That doesn't necessarily validate it as important.
 
Like any good science or engineering, every experiment, every engineering report needs to be able to be reproduced. If its not reproducible, then it cannot hold as much value.

While we "know", intuitively, that infant mortality of engines is real, and while we 'know' that TBO is a somewhat arbitrary number recommended by manufacturers that may have a stake in having engines overhauled; These things we "KNOW" need to be proven.

Mr. Busch as published several articles and mention Dr. Ulrich's work in each of them. There has also been a USCG white paper discussing the cost/benefits of Reliability Centered Maintenance instead of TBO centered maintenance. Each article refers back to a previous article and the work told to Mr. Busch by Dr. Ulrich. (references on request)

None of these 'publications' cites the data or provides any data to support their assertions.

Those that have worked in Manufacturing, Maintenance, Repair and Overhaul (MRO) "know" that it is impossible to support a conclusion (such as TBO), if you dont have the data. Design Engineering cannot be done without supporting data that shows the design will work. Theoretical calculations alone are not enough, especially if real world data exists. (which is why they keep such data).

Those POA members that have actually done reliability engineering work, know that making assertions without the data is tantamount to having the sales department make engineering decisions.

The goal should be to collect data from as many sources as possible and to determine if the assertions made previously can be verified.

Manufacturers have managed to have the engine data stripped from NTSB and FAA databases. [http://app.ntsb.gov/avdata/]

For this reason alone, previous reports tend to massage the ACCIDENT data and infer that this reflects ENGINE data when may not.

The ONLY Best Data on engine reliability are the reports on engines that fail or not... not airframes, or aircraft accidents.
 
Interesting....I did USGC aviation RCM during the 90's while at TAMSCO.

Good luck finding data to support single digit percentile distributions....:goofy:
 
Last edited:
44Gort said:
Like any good science or engineering, every experiment, every engineering report needs to be able to be reproduced. If its not reproducible, then it cannot hold as much value.

While we "know", intuitively, that infant mortality of engines is real, and while we 'know' that TBO is a somewhat arbitrary number recommended by manufacturers that may have a stake in having engines overhauled; These things we "KNOW" need to be proven.

Mr. Busch as published several articles and mention Dr. Ulrich's work in each of them. There has also been a USCG white paper discussing the cost/benefits of Reliability Centered Maintenance instead of TBO centered maintenance. Each article refers back to a previous article and the work told to Mr. Busch by Dr. Ulrich. (references on request)

None of these 'publications' cites the data or provides any data to support their assertions.

Those that have worked in Manufacturing, Maintenance, Repair and Overhaul (MRO) "know" that it is impossible to support a conclusion (such as TBO), if you dont have the data. Design Engineering cannot be done without supporting data that shows the design will work. Theoretical calculations alone are not enough, especially if real world data exists. (which is why they keep such data).

Those POA members that have actually done reliability engineering work, know that making assertions without the data is tantamount to having the sales department make engineering decisions.

The goal should be to collect data from as many sources as possible and to determine if the assertions made previously can be verified.

Manufacturers have managed to have the engine data stripped from NTSB and FAA databases. [http://app.ntsb.gov/avdata/]

For this reason alone, previous reports tend to massage the ACCIDENT data and infer that this reflects ENGINE data when may not.

The ONLY Best Data on engine reliability are the reports on engines that fail or not... not airframes, or aircraft accidents.
Click to expand...
So, what's going to be different about your data compared to the data mined from the NTSB reports?
 
NTSB reports would be a very incomplete source of engine failure info. The majority of engine failures don't result in NTSB investigations. Most are addressed at the home field after a serious vibration led to finding metal in the filter, an RPM deviation led to the discovery of a broken crank, a funny smell led to the discovery of a separated cylinder, etc. In my earlier example of a broken rod that beat holes in the case? The NTSB had no reason to become involved. If you're interested in crashes attributed to power loss? That data may interest you. If you're interested in overall engine failures? I doubt there's any good way to compile a good data set to make any summary that's useful.
 
So far, there hasnt been ANY engine data mined from the NTSB reports. Only Accident data that says engines were a cause or contributing factor to the accident.

Im not looking to be different, I just want to SEE what the data looks like.

Several poster's have tried to jump ahead of the data to see what a study like this could mean. My goal is not to interpret the data, just to publish it.

I know it's frustrating, but we're not trying to Predict anything, we'd just like to confirm what everyone "Seems to Know" but haven't established with real data.

A freedom of information request to the NTSB is being prepared to release the Engine data for analysis and publication. Up till now, TCM / Lycoming, etc have kept this data out of the public view, citing it as proprietary information. Accident investigations are not proprietary.

All we want is to see the data reflected in the post accident investigations; including engine logbook entries; namely, TSMO or TSFRM at the time of the incident.
 
Stewartb said:
NTSB reports would be a very incomplete source of engine failure info. The majority of engine failures don't result in NTSB investigations. Most are addressed at the home field after a serious vibration led to finding metal in the filter, an RPM deviation led to the discovery of a broken crank, a funny smell led to the discovery of a separated cylinder, etc. In my earlier example of a broken rod that beat holes in the case? The NTSB had no reason to become involved. If you're interested in crashes attributed to power loss? That data may interest you. If you're interested in overall engine failures? I doubt there's any good way to compile a good data set to make any summary that's useful.
Click to expand...

Agreed.

OTOH, the NTSB is a HUGE database, that has managed to collect nearly 100,000 incidents in the period 1982 - 2014. A big data-set like that is easier to work with than surveying several, perhaps many smaller sets.

I understand your doubts, but until I hit a brick wall, Im going to see where this rabbit-hole leads.
 
I know first hand of three engine failures. No crashes, one off-airport landing. No accidents, no injuries, no NTSB reports. Not sure that data is captured anywhere?
 
Engine failure rate data will be strongly context sensitive. Maintenance vigilance and type of flying will have a profound impact on failure rates. A pencil whipped towplane is going to have a different outcome than a carefully maintained personal aircraft flown regularly. Composite statistics may have little meaning for any specific, individual case. The best intrinsic data might come from well maintained flight school aircraft fleets.
 
"The best intrinsic data might come from well maintained flight school aircraft fleets."

That will give us data on well maintained aircraft. Do you think it would be statistically different than just general aircraft?

Thanks.
 
If you're lucky someone would cough that up....but, it's doubtful.

A large school, like Riddle, would have the data....but, it isn't going to be representative of the average GA owner....who's plane sits for long periods and flys a couple hours here and there and the colder regions have different wear due cold starts.


But, if you should get flight school data....I'd expect it to be more optimistic than your average GA owner.
 
Last edited:
Henderson State University in Arkadelphia, AR has an aeronautical program with a fleet of Maules and a few others. They might be willing to share info if theyhave it since they are a public entity.

Contact
Department of Aviation
HSU Box 7611
870-230-5012 phone
870-230-5005 fax
 
Checkout_my_Six said:
If you're lucky someone would cough that up....but, it's doubtful.

A large school, like Riddle, would have the data....but, it isn't going to be representative of the average GA owner....who's plane sits for long periods and flys a couple hours here and there and the colder regions have different wear due cold starts.


But, if you should get flight school data....I'd expect it to be more optimistic than your average GA owner.
Click to expand...

Agreed.

We already know that large organizations, dedicated to flying lots of hours, and with MRO; that the aircraft are operated better and have fewer issues on an hourly basis.
 
UPDATE:

So far, Ive had no luck even getting permission to file an application for a Freedom of Information request to the NTSB. Thier website is designed not to allow entry.

I will have to file a request, by certified mail, to even get the ball rolling.

-------------------------------------------------------------------

What I have gleaned from their database is the usual statistics of how many planes had accidents or incidents that were maintenance related or even engine related.

There is NO data in the database to relate engine hours to accidents or to conclude any relationship between these at all.

Thanks for watching. :)
 
and even "if" you get hold of maintenance logs....you will have no idea why items were replaced....and the condition of the replaced item (failed or suspended).

I've done this as a contractor for maintenance organizations and had access to interview folks who did the maintenance entries. It's a mess I tell ya.

But....keep trying. :D

Again...you will get much more failure data and meaningful information from an engine shop....than a flight organization.
 
44Gort said:
"The best intrinsic data might come from well maintained flight school aircraft fleets."

That will give us data on well maintained aircraft. Do you think it would be statistically different than just general aircraft?

Thanks.
Click to expand...

Unless you own as "general aircraft" such a statistic would be meaningless. And I would expect a wide variation of engine failures rates for engines not maintained properly or not flown regularly. Failure rates will only be meaningful and useful for a cohort of engines of similar manufacture, maintenance, and flight usage. But if you aspire to maintain your engine to a certain standard and fly regularly, you can expect failure rates similar to the rest of the cohort. The problem is similar to trying to predict your chances of a heart attack by using rates for the general population: if your cholesterol is stratospheric and you are a smoker, those general population statistics won't be of much value to you personally. You have to assess your personal chances against a similar population, not the general one.
 
[The problem is similar to trying to predict your chances of a heart attack by using rates for the general population: if your cholesterol is stratospheric and you are a smoker, those general population statistics won't be of much value to you personally.]

Again, we're not trying to predict anything. Medical reports relating cholesterol with heart disease are not predicting; they are gathering evidence and making a correlation based on statistical controls.

AFTER you have the data, THEN you try to compare YOUR Symptoms to those of the failure group.

At this point, there isnt any data to support that any failures have any cause. They just happen.

Gather the data first.
 
You must log in or register to reply here.
Back
Top