gallois wrote:
But the statistics now form part and are presented at an annual safety meeting for all clubs and their pilots.
Never heard of that. Especially no statistics with hours flown vs engine failures. If anyone has access to these stats, it would be very useful ! Even if instead of hours flown we have currently airworthy fleet size (the rest can be inferred in a .5-2x range, taking an estimated average hours flown per year per aircraft which cannot be much more wrong than a factor of two).
UdoR wrote:
Ok wow, nice discussion. So statistics are not available or show numbers between 1 per 1,000 and 1 per 100,000 flight hours. That is quite a range. 1 per 100,000 would be turbine reliability.
I think the study i posted is between 1,000 and 10,000 not 100,000
Study hardly describes it, since the study is about the analysis of components that fail in engines. The quoted failure rate per flight hour was taken from the FAA (for which I can’t find a direct source other than the ATSB study), who basically said “we don’t have hard data, we estimate the rate between 1,000 and 10,000”. We also don’t know exactly if this includes mismanagement or other factors. My point from the other thread was that a good portion of complete engine failures (the ones that are interesting, i.e. that lead to a forced landing without power) should be reported because they end up in a random field, from where (at least in France) even without damage you can’t legally takeoff without approval from the DGAC (more likely they send someone over to take the aircraft back to an aerodrome). Also mayday declaration. Also potential damage (for which an insurer will want some kind of official proof, i.e. an official accident report and not just an unofficial declaration from the pilot).
Under-reporting of in-flight shutdowns does not necessarily mean under-reporting of forced landings due to complete loss of power. Flame out and other temporary failures also exist, which do speak of the reliability of engines but are far less interesting when considering actual operation and risks.
Factoring in various mismanagement factors (again, we don’t know how the FAA estimated their seat of their pants number) you may well end up in 1/30k hours as opposed to 1/1,000 or 3,000 or 10,000. Which is a big deal.
Just curious – if you had extremely accurate stats on engine failures, wouldn’t the next question be “what failed on the engine?” and then “why did it fail?”
There are not going to be lots of answers to these questions, since there are only so many parts. Perhaps we can jump to that discussion – how to avoid engine failure, no matter how frequent or rare?
If we think of the engine as a system inside an aircraft, then the main failure mode is fuel mismanagement, including exhaustion. If you eliminate this risk, you’ve probably decreased your chance of an engine stoppage by more than 30%. Avoiding fuel contamination will probably get you another 10%.
Crank failures and piston connecting rod failures seem to be the major causes of “catastrophic failures” or “rapid unplanned disassembly” of an engine. These seem to be caused by either metallurgical issues, vibration fatigue, or ignoring a prop strike. Not much an individual pilot can do about these, but owners of course have a lot of power to avoid this failure mode.
Mike Busch makes a strong case that bad MX, like not taking care when removing a cylinder, is the cause of a lot of these catastrophic engine failures. Not sure he has any stats to back this up, but has shared several anecdotes.
Keep in mind that CFIT and LOC-I are killing more GA pilots than engine failures, so focusing on avoiding those will pay huge benefits.
Here’s a data point:
Our club has five airplanes (four until 2019) and has been around for over 60 years. In the years I have access to the data, which is about the last 10 years, utilization across the fleet was about 1300 hours p.a. I don’t think that to be significantly different in years past. To the best of my knowledge, there has never been an engine failure in any club airplane, although I obviously don’t know what happened way back in the 60s, 70s or 80s. As we’re part 91 we operate our engines on condition with an internally agreed limit of about 20% overtime. I say ‘about’ as this is somewhat contingent on parts and shop availability.
So there you have some hard data: at least 13,000 hours in a club environment w/o engine issues.
172driver wrote:
So there you have some hard data: at least 13,000 hours in a club environment w/o engine issues
What’s the probability that this will actually happen given a failure rate of 1/5000 hours? It means the MTBF or MTTF is 5000h. The probability of no failure:
R = e^(-t/MTBF) = e^(-13000/5000) = 0.074, or 7.4% Not very likely on other words. However, the 13000 h without in flight failures is more the norm IMO. Clearly the MTBF cannot be 5000 h.
Set MTBF to 50,000 h and the probability of surviving to 13000 h becomes 0.77 or 77%. I would say this is more in line with what we actually observe.
The TBO/TBR of a piston engine is around 2000h. There has to be some decent probability that it actually will run to TBO without an in flight sudden failure. If the MTBF was 5000 h, the probability of running to 2000h would be only 67%. Is this something we observe? I don’t think so. 50k h will give 96% This means that of 100 engines, 4 will fail in flight before 2000h. I think even this a lot, but not unreasonable like 67%. Lets say it’s 1000h as the FAA suggests. Then the probability of running to TBO will only be 13.5%
An engine may have a whole bunch more failures, but unless it happens in flight causing a dead stick landing (or worse), does it really matter?
happened to one of the most diligent pilots I know…
If the gear collapsed because of some technical problem then it happened. If he forgot to lower it down then it didn’t just happen but he wasn’t as good (diligent) as he (or others) thought.
172driver wrote:
So there you have some hard data: at least 13,000 hours in a club environment w/o engine issues.
As I mentioned above, my club has very similar stats. In the 10 years, we had exactly one actual engine failure. (Caused by a cracked cylinder.) The aircraft landed safely in a field and could be flown out after repairs.
In the same period, we had one engine stoppage caused by temporary unporting of a fuel tank with little fuel left and one caused by carb ice. In both cases the engine could be restarted.
(The actual engine failure was interesting in that all the other three cylinders had to be prematurely replaced at various times because of cracks discovered during scheduled maintenance. It looked like we got a bad batch at the previous engine overhaul.)
There is a big difference between a catastrophic engine failure and power loss.
Or landing at an alternate aerodrome and a landing in a field.
As I posted before. All these things are reported within the French aeroclub system and the FFAs SMILE.
Clubs affiliated to FFA now insist that all incidents are reported in the REX.
We had our club safety meeting a few weeks ago. As I posted in another thread there were very few accidents causing damage, injury or death due to engine failure or engine problems (and there were plenty of them due to loss of power in initial climb). Most accidents, by a long way, were down to bad management of approach and landing. Some of these may well have been initiated due to engine problems.
I don’t have the figures with me as I am just leaving Venice, by train I hasten to add.
LeSving wrote:
R = e^(-t/MTBF) = e^(-13000/5000) = 0.074, or 7.4% Not very likely on other words. However, the 13000 h without in flight failures is more the norm IMO. Clearly the MTBF cannot be 5000 h.
I think you need to take MTBF*1.5 in your calculation (because e^-1 does not equal 0.5, it’s more like e^-1.5).
Also, not taking not the probability of no failure but the probability of the first failure happening at a particular flight hour (taking 18,000 hours because we have 13,000 hours without and 13,000 with one):
So if you start with an “a priori” law of equal likelihood for the 4 different MTBF, you end up with a posteriori:
Not completely conclusive of course, but interesting. With just a little more data, things can get more precise.
