CASA Proposes To Restrict Jabiru Engines

jnmeade

Cleared for Takeoff
Joined
Sep 25, 2005
Messages
1,216
Location
Eastern Iowa
Display Name

Display name:
Jim Meade
In a rather strong proposal, the Australian aviation agency CASA proposes to restrict the use of Jabiru engines due to what they consider an excessive number of failures. (They show some numbers comparing them to Rotax.) Students would not be allowed to fly them solo. Pilots couldn't carry passengers.

http://www.australianflying.com.au/news/casa-moves-to-restrict-jabiru-engine-operations

"Draft Consultation Document 1425 effectively bans the use of Jabiru-engine aeroplanes for carrying passengers and for solo student training.

The proposed restrictions also include a constraints on flying over populous areas and the need to placard the aircraft with operating limitations."

"Jabiru founder Rod Stiff was quick to condemn the regulator for proposing the action.

"CASA has based this on spurious statistics and the commercial gripes of a few," he said last Friday. "I have tried to build a strong and safe aircraft and I reluctantly boast that in 23 years of Jabiru flight no one has been killed or seriously injured as a result of my doing.

"Jabiru is the safest light sport aircraft in Australia."

RA-Aus has responded by admitting that Jabiru engines have a failure rate higher than that of Rotax, but believes owners and pilots should be left to make their own decisions about safety. A statement issued comments:"

There's more. It's not a long read and is interesting so don't rely only on the excerpts, which are just there to tease you. :)
 
I didn't see any distinction between the 2200 and 3300 Jabiru engines in the cited report. I know in the St. Louis area of three 2200 engine failures -- two with massive damage. But not the 3300s.
 
The Jabiru has had a reputation for good while now. This does not surprise me, but it is sad that the Australian government has to step in to "protect" it's citizens. You would think that the consumers would just stop buying them. I guess they must be loads cheaper than a Rotax. I have heard enough that I would never buy one and would do my homework about their failure modes before I ever flew behind one.
 
I used my 15-year database of US homebuilt accidents to examine engine-related events.

We have no way to tell how many of a particular type of engine is installed in US Experimental aircraft. The FAA registration database includes an engine column, but thousands of homebuilts don't specify an aircraft type.

However, we can determine what percentage of accidents affecting aircraft with a given type of engine were caused by problems with the engine. The larger the percentage, the higher the likelihood that the engine is having difficulties...either due to design issues or the difficulty in properly installing or operating the engine.

Here are the percentages:

Continental Engines: 18%
Lycoming Engines: 16%
Auto-Engine Conversions (including VW): 38%
Jabiru: 22%
Rotax (all): 28%
Rotax 912: 12%

This does indicate that Jabiru has a higher incidence of engine problems than the Rotax 912.

Ron Wanttaja
 
Now if this is a CASA decree, I wonder if it only applies to VH tail planes? Most of the planes flying behind a Jabiru in Australia are going to be RAA registered, and I'm not sure how much say CASA has in the matter.:dunno:
 
Some very interesting numbers right there...and a counterpoint to the "I would never fly behind a Rotax" crowd.

:stirpot:

I'd like to see the percentages on how many engines installed. If Lycoming and Continental make up 34% of failures, but over 50% of installs, that's pretty good. If they only make up 20% of installs that's bad. Bare failure numbers without context only paint part of the picture.
 
I'd like to see the percentages on how many engines installed. If Lycoming and Continental make up 34% of failures, but over 50% of installs, that's pretty good. If they only make up 20% of installs that's bad. Bare failure numbers without context only paint part of the picture.

These are failure rates per accident, number of installed engines does not matter here. This is the percentage of accidents/incidents reported to the FAA where the engine was determined to have been the cause, for each engine type.
 
Some very interesting numbers right there...and a counterpoint to the "I would never fly behind a Rotax" crowd.

:stirpot:

Was this based on number of flight hours or time?

Lots more lycomings and contis flying lots more hours.
 
Those percentages don't come out to 100...
 
These are failure rates per accident, number of installed engines does not matter here. This is the percentage of accidents/incidents reported to the FAA where the engine was determined to have been the cause, for each engine type.

Exactly.

It's near-impossible to determine the total number of installs. The FAA lists thousands of homebuilts with "AMAT/EXP" engines. Others give the make without a model (you don't know whether the "BOMBARDIER" engine is a four-stroke, or two stroke).

But the NTSB accidents almost always list the make, and usually the model, engine. Say there are 100 total accidents that occur with Brand X engines installed, and 100 total accidents that occur to Brand Y engines. If 25 of the accidents with Brand X engines happened due to issues with the engine, and 10 accidents with Brand Y were due to the engine, it's a logical assumption that Brand X engines are more likely to have problems.

I don't have my database handy, but if I remember correctly, there were 59 Jabiru-engined homebuilt accidents in the span of my 15-year database, and about 67 Rotax 912-engined homebuilt accidents in the same period. Thirteen Jabiru engines were the cause of the accident, as opposed to eight Rotax 912s.

By "caused by the engine" I mean specific problems with the engine...not just the fact that the engine quit running. Fuel Management, Carb Icing, and other issues are tallied separately. If builder or maintainer error led to the engine problem, this is also tallied separately.

It is a relatively small sample size... my minimum is 50, and both are just above that. But one would assume these engines are more common in Australia, and the Australian authorities can run the same sorts of analyses.

Ron Wanttaja
 
Here are the percentages:

Continental Engines: 18%
Lycoming Engines: 16%
Auto-Engine Conversions (including VW): 38%
Jabiru: 22%
Rotax (all): 28%

122%

More engine failures than can be fit into conventional math! Time to go back to flying gliders....

Edit: EdFred beat me to it. Slow typist.
 
Those percentages don't come out to 100...
The list I posted doesn't include Franklin engines, Allison engines, Walther engines, Rolls-Royce engines, Lycoming clones listed under another name, Rotax 914s, any two-stroke engines, etc.

Ron Wanttaja
 
122%

More engine failures than can be fit into conventional math! Time to go back to flying gliders....

Edit: EdFred beat me to it. Slow typist.
Ah, I see the issue.

The percentages I posted are the percentage of engine-related failures FOR EACH ENGINE TYPE. Adding the percentages between types is meaningless. I could list five different engines, each with 100% of the accidents caused by the engine. The fact that that would add up to 500% is immaterial.

Note also, that this is NOT THE PERCENTAGE OF ENGINES OF THAT TYPE THAT CAUSE ACCIDENTS. Some people see that 38% for auto-engine conversions and say, "Wow, 38% of auto-engine aircraft crash!" That's not the case at all.

The percentage is of airplanes that *did* crash, not the overall fleet.

Ron Wanttaja
 
The percentages I posted are the percentage of engine-related failures FOR EACH ENGINE TYPE. Adding the percentages between types is meaningless. I could list five different engines, each with 100% of the accidents caused by the engine. The fact that that would add up to 500% is immaterial.

Note also, that this is NOT THE PERCENTAGE OF ENGINES OF THAT TYPE THAT CAUSE ACCIDENTS. Some people see that 38% for auto-engine conversions and say, "Wow, 38% of auto-engine aircraft crash!" That's not the case at all.

The percentage is of airplanes that *did* crash, not the overall fleet.

Let me know if I got your procedure correct or not:

(1) You start with a database of records describing crashes.
(2) For each engine type E:
(2.1) You extract only those records where E is mentioned. You find you have extracted D records.
(2.2) You examine the D records and count A records where a failure of engine type E was listed as a cause of the accident.
(2.3) If A is less than 50, you feel the sample size is too low to go to the final step. [Or maybe you meant if D is less than 50, don't process further.]
(2.4) You generate percentage P which you then post, where P = 100*A/D.
 
Ah, I see the issue.

The percentages I posted are the percentage of engine-related failures FOR EACH ENGINE TYPE. Adding the percentages between types is meaningless. I could list five different engines, each with 100% of the accidents caused by the engine. The fact that that would add up to 500% is immaterial.

Note also, that this is NOT THE PERCENTAGE OF ENGINES OF THAT TYPE THAT CAUSE ACCIDENTS. Some people see that 38% for auto-engine conversions and say, "Wow, 38% of auto-engine aircraft crash!" That's not the case at all.

The percentage is of airplanes that *did* crash, not the overall fleet.

Ron Wanttaja


Okay, let me see if I understand this correctly:

For example, for all the airplanes with continental engines that crashed, 18% of the crashes were engine related.


Sent from my iPhone using Tapatalk
 
Let me know if I got your procedure correct or not:

(1) You start with a database of records describing crashes.
(2) For each engine type E:
(2.1) You extract only those records where E is mentioned. You find you have extracted D records.
(2.2) You examine the D records and count A records where a failure of engine type E was listed as a cause of the accident.
(2.3) If A is less than 50, you feel the sample size is too low to go to the final step. [Or maybe you meant if D is less than 50, don't process further.]
(2.4) You generate percentage P which you then post, where P = 100*A/D.
Yes, that's right. I do use "D" (total number of accidents with that engine, not the total number of accidents DUE to that engine) for my 50-accident threshold. I'll compute it anyway, but just won't usually talk about it publically unless I've got 50.

This stems from when I was computing accident rates for different homebuilt types.

Ron Wanttaja
 
Okay, let me see if I understand this correctly:

For example, for all the airplanes with continental engines that crashed, 18% of the crashes were engine related.

Yep!

Ron Wanttaja
 
Those stats are crap.

What Id expect to see is engine failures per hour flown, and percentages totaling 100% over all the makes that had a failure over a X year period.

The stats should also exclude any fuel contamination or running out of fuel situation, minus those caused my an engine driven pump or ther engine based system.
 
Last edited:
Ah, I see the issue.

The percentages I posted are the percentage of engine-related failures FOR EACH ENGINE TYPE. Adding the percentages between types is meaningless. I could list five different engines, each with 100% of the accidents caused by the engine. The fact that that would add up to 500% is immaterial.

Note also, that this is NOT THE PERCENTAGE OF ENGINES OF THAT TYPE THAT CAUSE ACCIDENTS. Some people see that 38% for auto-engine conversions and say, "Wow, 38% of auto-engine aircraft crash!" That's not the case at all.

The percentage is of airplanes that *did* crash, not the overall fleet.

Ron Wanttaja

So the percentage number represents the total number of planes that used that engine that were involved in a crash, that percentage is the number of those crashes where it was the engine that was the causal factor. Do I have that correct?

As in "Of all crashes in a plane with a Rotax 912, 18(or whatever it was)% of those were due to engine failure."
 
Those stats are crap.

What Id expect to see is engine failures per hour flown, and percentages totaling 100% over all the makes that had a failure over a X year period.

The stats should also exclude any fuel contamination or running out of fuel situation, minus those caused my an engine driven pump or ther engine based system.

I thought it was a very clever way to make the information available somewhat representative of trends.
It would be nice to have something about failures per hours flown but as far as I know that information is not readily available.
It would be fun to know the cause of the failure but again as far as I know that information is not readily available.
It appears to me that many engine failures in experimental amateur built aircraft are due to installation errors rather than a weakness in the engine.
The percentages are not intended to add up to 100% but that has already been discussed.
 
Those stats are crap.

What Id expect to see is engine failures per hour flown, and percentages totaling 100% over all the makes that had a failure over a X year period.

The stats should also exclude any fuel contamination or running out of fuel situation, minus those caused my an engine driven pump or ther engine based system.

That's just a different measure, it doesn't mean these stats are crap.

What if you had 100,000 total hours logged for engine type 'E', and there were 10 engine failure accidents in that 100,000 hours, or one every 10,000hrs...reliable engine, right?

What if I then tell you that the total sample is 20 engines. And that of the ten that failed, all of them failed in the first ten flight hours, and the other ten made the rest of the hours without incident, around 10,000hrs per engine?

Kind of changes how things look, right? Now it's not really a reliable engine, it's one that either goes 10,000hrs or fails in ten, 50/50 chance either way. No statistic tells the full story without context.
 
Those stats are crap.

What Id expect to see is engine failures per hour flown, and percentages totaling 100% over all the makes that had a failure over a X year period.

The stats should also exclude any fuel contamination or running out of fuel situation, minus those caused my an engine driven pump or ther engine based system.
Fuel exhaustion/starvation/contamination, carb ice, and maintenance/construction-related issues are not included in my engine category. I track them separately.

The trouble is, any sort of "Failures per hours flown" parameter needs an "hours flown" input. That's not available for most GA aircraft, especially homebuilts. There are FAA estimates, yes... but then you're at the mercy of whoever is making those estimates. The FAA says that an RV flies 29 hours a year, and a Cessna 172 flies 200 hours per year. Personally, I think the pilot who owned a 172 but now flies an RV flies about the same amount, but it's not my opinion that counts.

My own opinion is that if two different types of engine can be installed in the same homebuilt, the hours per year will be about the same. I suspect the Jabiru and Rotax 912 are installed in about the same types of airplane.

Ron Wanttaja
 
My own opinion is that if two different types of engine can be installed in the same homebuilt, the hours per year will be about the same. I suspect the Jabiru and Rotax 912 are installed in about the same types of airplane.

Same kinds, yes, but not the same. For example, there almost no Sonexes with 912. I would estimate perhaps 5 in whole country, if that. Most of them fly with factory supported engines (which coincidentially are much cheaper). On the other hand, you have to be disconnected from the reality in order to install a Jabiru on an RV-12. The only popular type that flies on both that I know is Zenith 701 and Zenith 601. It's a policy of Zenith to allow customers install whatever they like.

Now I halfway expect you find data that blows my preconceptions out of the water :)
 
Same kinds, yes, but not the same. For example, there almost no Sonexes with 912. I would estimate perhaps 5 in whole country, if that. Most of them fly with factory supported engines (which coincidentially are much cheaper). On the other hand, you have to be disconnected from the reality in order to install a Jabiru on an RV-12. The only popular type that flies on both that I know is Zenith 701 and Zenith 601. It's a policy of Zenith to allow customers install whatever they like.

Perhaps I should have said, "The same kinds of missions," vs. "the same kinds of homebuilts." Basically, I thinking that the Jabiru and the Rotex 912 are installed in homebuilts that appeal to the same kinds of owners, and who would fly similar amounts no matter which plane they selected. An owner with a Zenith 601 with a Jabiru would probably fly the same annual hours if he had an RV-12 with a Rotax.

Now I halfway expect you find data that blows my preconceptions out of the water :)
I'll try to keep your preconceptions' hulls wet, but you *do* make me curious. I'll dig through the FAA registration database tonight for some quick summaries of engines by types.

Ron Wanttaja
 
A bit on engine installations on several small homebuilts, using my 1 January 2014 copy of the FAA Registration Database (e.g., NOT accident database).

I looked at RV-12s, Zenair 601s, Zenair 701s, and Sonex. *Only* aircraft listed as Experimental Amateur-Built were included. Sonex included Waiex, but not Xenos.

Rotax:
RV-12: 81%
Zenair 601: 8%
Zenair 701: 39%
Sonex: 0%

Jabiru:
RV-12: 6%
Zenair 601: 28%
Zenair 701: 11%
Sonex: 29%

Not Specified:
RV-12: 6%
Zenair 601: 35%
Zenair 701: 21%
Sonex: 63%

Ron Wanttaja
 
The FAA says that an RV flies 29 hours a year, and a Cessna 172 flies 200 hours per year. Personally, I think the pilot who owned a 172 but now flies an RV flies about the same amount, but it's not my opinion that counts.

Keep in mind that the 172 is used A LOT for training and the RV... pretty much zero. My logbook shows that when I was training for my IR, I flew a 172 about double the hours in that year than I fly my Mooney annually now.
 
Here are the percentages:

Continental Engines: 18%
Lycoming Engines: 16%
Auto-Engine Conversions (including VW): 38%
Jabiru: 22%
Rotax (all): 28%
Rotax 912: 12%

I was looking at my database again in regards to some of the additional questions, and found I'd messed up the extraction process. Should have known to use a brute force method rather than try some multiple cross-references.

The numbers for all are lower. The difference between Jabiru and Rotax 912 is less, but still has definite gap.

Continental 13.4%
Lycoming 11.9%
Auto Conversions 28.8%
Jabiru 16.9%
Rotax (all) 20.4%
Rotax 912 12.1%

My apologies for the error.

Ron Wanttaja
 
I was looking at my database again in regards to some of the additional questions, and found I'd messed up the extraction process. Should have known to use a brute force method rather than try some multiple cross-references.

The numbers for all are lower. The difference between Jabiru and Rotax 912 is less, but still has definite gap.

Continental 13.4%
Lycoming 11.9%
Auto Conversions 28.8%
Jabiru 16.9%
Rotax (all) 20.4%
Rotax 912 12.1%

My apologies for the error.

Ron Wanttaja
i suspect that's a little skewed, calling an aerovee or rotorway an "auto conversion" is about as accurate as calling a rotax a "snowmobile conversion". I'd like to see data on the aerovee in its own right, I'll bet it is pretty darn reliable
 
Keep in mind that the 172 is used A LOT for training and the RV... pretty much zero. My logbook shows that when I was training for my IR, I flew a 172 about double the hours in that year than I fly my Mooney annually now.
Oh, definitely. I used an RV and a C-172 for an example. The FAA doesn't specify an aircraft type...those are the estimates used for homebuilt and production-type GA aircraft, not just those types. 29 hours/year for homebuilts, 200 hours/year for production-type aircraft.

The problem is, we have no insight into *how* those numbers are generated. When I talked to the FAA and Nall Report folks a few years back, it was a combination of survey results and estimates (ha!) on how many homebuilts are inactive.

Now that roughly 1/4th the homebuilt fleet has been removed from the registry via the re-registration process, do you want to bet that the assumptions didn't change?

The Nall Report (which is based on FAA hour estimates) says that homebuilts have an accident rate about seven times higher than production-type GA aircraft. When the computation is made, the estimate of annual hours flown by production-type aircraft is also nearly seven times higher.

Just seems like one heck of a coincidence.

My argument is that for *comparable use* (e.g., privately owned personal-use aircraft) an owner will fly a homebuilt or a production aircraft about the same number of hours per year.

When I compute relative accident rates using that assumption, I find that homebuilts have a ~45% higher accident rate...not good, but beats the multi-hundred-percent higher values some folks come up with.

When I leave out accidents that occur in the test phase, homebuilts have a ~16% higher accident rate. That actually seems pretty realistic to me, when you consider the planes are amateur-built, amateur-maintained, and occasionally amateur-designed.

Ron Wanttaja
 
i suspect that's a little skewed, calling an aerovee or rotorway an "auto conversion" is about as accurate as calling a rotax a "snowmobile conversion". I'd like to see data on the aerovee in its own right, I'll bet it is pretty darn reliable
The problem is, between 1998 and 2012, there were only eight accidents of aircraft where the NTSB said it had an Aerovee engine (or Aero-Vee). That's really too small a a sample size to make an estimate.

And for that matter, my January FAA database says there is only one Sonex registered with an Aerovee engine.

But if you look at one of my earlier posts, you'll see that 72% of Sonexes either have a blank for the engine entry in the FAA registry, or merely state it's an AMAT/EXP engine. A lot of these are probably Aerovees. But how does one tell?

Look at the accident report for ANC11LA060. What type of engine did it have? "Volkswagen Aerovee 2.1". Is it a Volkswagen...or an entirely different engine?

And one correction: I have never referred to a "Rotorway" as an auto engine conversion. In my database it's flagged as a "non-certified four stroke," same as the Jabiru and Rotax 912.

Ron Wanttaja
 
I was looking at my database again in regards to some of the additional questions, and found I'd messed up the extraction process. Should have known to use a brute force method rather than try some multiple cross-references.

The numbers for all are lower. The difference between Jabiru and Rotax 912 is less, but still has definite gap.

Continental 13.4%
Lycoming 11.9%
Auto Conversions 28.8%
Jabiru 16.9%
Rotax (all) 20.4%
Rotax 912 12.1%

One could conclude from the above that pilots who fly auto conversions make so few mistakes and have so few fuel exhaustion or weather problems that engine failures are the dominant cause of their accidents. :D

Obviously I'm little surprised to see engine failures accounting for such relatively large (in my opinion) percentages of accidents.
 
Obviously I'm little surprised to see engine failures accounting for such relatively large (in my opinion) percentages of accidents.
About one-third of all homebuilt accidents start with an engine failure. It may be a mechanical issue, it may be fuel management-related, but it's even more vital for homebuilt pilots to be ready for an engine-out.

For my combined Cessna 172/210 database, about 24% of the accidents involve engine failure.

Keep in mind that the NTSB will blame many of these accidents on the pilot, even if the cause of the engine failure is mechanical. If they feel a successful forced landing should have been possible, they'll assign pilot error as the Probable Cause.

Ron Wanttaja
 
Those stats are crap.

What Id expect to see is engine failures per hour flown, and percentages totaling 100% over all the makes that had a failure over a X year period.

The stats should also exclude any fuel contamination or running out of fuel situation, minus those caused my an engine driven pump or ther engine based system.

Well then. Have a go at it and report back. No rush. Take a couple days if you need to.
 
Back
Top