Accident Statistics: Certified vs. Non-Certified Engines

"RU ok" wrote in message
...


Barnyard BOb --- a really 'regular' guy


You "Otto" be regular!

Rich S. :^O

Ron Wanttaja wrote in message . ..
A while back, I posted the result of a cursory look at the accident
statistics for aircraft powered by auto engine conversions. It *was* an
awful shallow pass, and at the time I promised to look into the issue
deeper.

I have since obtained the NTSB accident databases for the years 1998, 1999,
and 2000, and am ready to provide more exact figures.

The nice thing about the accident databases is that they usually describe
the type of engine that powers the aircraft. Unfortunately, the FAA
registration database is a lot more vague. A lot homebuilts are merely
described as having experimental engines; a number don't even have an
entry. So we can't do the classic "x% of auto engined-airplanes have
accidents every year vs. y% Lycont-powered planes."

Does this take into account things like modified Lycomings? We have
an O-360 with an Ellison TBI. It isn't a certified engine, but it's
not an auto conversion or rotax or something either.

"Morgans" wrote in
:


"Ron Wanttaja" wrote

Evahbuddy's a critic. :-)

To make it clear, there were three engines that fit that particular
category:

Jabiru
Rotorway
Rotax/Bombardier 9XX series

Ron Wanttaja

WWeeeeeelllllll,
actually, the Rotax 912 has been certified, also, right?

Otherwise, good work. g

Uhuh; but you can still get the uncertified version also.

On 11 Jan 2004 07:47:52 -0800, (Bob Martin) wrote:

The nice thing about the accident databases is that they usually describe
the type of engine that powers the aircraft. Unfortunately, the FAA
registration database is a lot more vague. A lot homebuilts are merely
described as having experimental engines; a number don't even have an
entry. So we can't do the classic "x% of auto engined-airplanes have
accidents every year vs. y% Lycont-powered planes."

Does this take into account things like modified Lycomings? We have
an O-360 with an Ellison TBI. It isn't a certified engine, but it's
not an auto conversion or rotax or something either.

The database engine description column does not provide indication of
accessories, but the narratives mention non-standard components if they
have anything to do with the accidents. I suspect the NTSB investigator
probably lists "Lycoming" if the engine is *mostly* Lycoming...like in the
case of one Seawind crash, where the plane had one standard mag and one
non-certified electronic ignition.

I went through the 1998-2000 accident reports and did my own assessment as
to causes. Allow me to summarize the causes for those cases involving a
loss of power due to mechanical problems. Here's an explanation of the
categories:

"Undetermined": Loss of engine power for unknown causes
"Engine Internal": Failures in the interior of the engine, such as a valve
or connecting rod
"Fuel - System": Problems in the fuel system from the tank caps to the
gascolator, inclusive
"Fuel - Engine": Problems "forward" of the gascolator (or "aft" if it's a
pusher).
"Ignition": Problems with the ignition, wires, or plugs
"Drive system": PSRU or drive shaft
"Oil System": Oil pumps, oil lines, etc.
"Carb Mech": Mechanical problems at the carburetor

I've lumped the Lycoming and Continental numbers together. The percentages
are the ratio between the cases with the particular cause, and all the
non-fuel-exhaustion loss-of-power incidents involving that engine.

Be advised that the results below do NOT necessarily reflect the NTSB
"Probable Cause" for the accidents. In loss of power accidents, the NTSB
investigators often list the cause of the engine failure as a secondary
cause, and fault the pilot for not being able to land the airplane
separately. I read the reports (yes, over 800 of them) and made my own
determination. The data below includes both the primary and secondary
causes.

------------------------------------------------------------------------
Results:

(As ever, this displays better if you've got your reader set to a
fixed-spaced font)
Lyconts Auto
Undetermined 31.5% 25.9%
Engine Internal 20.4% 29.6%
Fuel - Engine 3.7% 0.0%
Fuel - System 25.9% 14.8%
Ignition 1.9% 14.8%
Drive system 1.9% 3.7%
Oil System 7.4% 0.0%
Carb Mech 7.4% 7.4%
Carb Ice 9.3% 0.0%
Cooling System 0.0% 3.7%

Sample Size 54 27

Note that the sample size is low enough that one anomalous event runs the
percentages higher.

Ron Wanttaja

Ron Wanttaja wrote in
news
A while back, I posted the result of a cursory look at the accident
statistics for aircraft powered by auto engine conversions. It *was*
an awful shallow pass, and at the time I promised to look into the
issue deeper.

I have since obtained the NTSB accident databases for the years 1998,
1999, and 2000, and am ready to provide more exact figures.

The nice thing about the accident databases is that they usually
describe the type of engine that powers the aircraft. Unfortunately,
the FAA registration database is a lot more vague. A lot homebuilts
are merely described as having experimental engines; a number don't
even have an entry. So we can't do the classic "x% of auto
engined-airplanes have accidents every year vs. y% Lycont-powered
planes."

Instead, we can take another tack: We can catalog the number of each
type of engine in accident aircraft, then take a look at how often a
loss of engine power was a factor in the accident. The figures don't
include cases where the cause was traced to carburetor ice or the
pilot running out of fuel.

Presentation of data:

The "ENGINE" column describes the general category of the engine, one
of four types:

"Certified" Engines include Continentals, Lycomings, Franklins,
Pratt
and Whitney, Jacobs, Vendeyev, LOM, and Walter.

"Auto" Engines include those identified as Subarus, Suzukis,
Fords,
Volkswagens, Revmaster, Chevrolet, GM, Mazda, Honda, Stratus, or NSI.

"Non-C/4" are four-cylinder, non-certified, non-auto conversion
engines. They include the Rotax 912 series, the Jabiru, and the
Rotorway.

"Two-Stokes" include Rotax 4* and 5* series, Yamahas, KFMs, Hirth,
2SI,
and Cuyuna.

The next column is "ACC". This is the number of accidents in the
1998-2000 timeframe that involved each category of engine

"PCT" is the percent of the total accidents where that category of
engine was installed.

"LOP" are the number of accidents where loss of engine power was
involved.

"LOP%" is the percentage of cases where accidents involving aircraft
mounting that category of engine suffered an engine-related loss of
power.

---------------------------------------------------------
The Results:

ENGINE ACC PCT LOP LOP%
------ --- --- --- ----
Certified 332 51% 57 17%
Auto 95 15% 27 28%
Non-C/4 70 11% 13 19%
Two-Strokes 134 21% 46 34%

Of primary interest here, I think is the percentage of accidents where
a loss of engine power occured...17% for certified-engine-powered
planes, vs. 28% for auto-engine conversions. Two-strokes were even
higher; almost a third of their accidents involved a power failure.

It's interesting to note the non-certified four strokes are doing
practically as well as the certified engines. The Rotax 912/914
series alone does even better... a LOP% value of 13%.

An interesting side note: Lycomings outnumbered Continentals by
nearly four to one....

Ron Wanttaja





I wonder what these figures really tell us.... I believe type of
accidents as a percent of total accidents really can't tell us anything.

What if accidents of other types (the rest of the %) where actually
lower in homebuilts with auto engines?

What if things like icing, or VFR into IMC was more prevalent in
certified aircraft, skewing the engine related numbers into a lower
percent...

It would also be my guess (and only a guess it is) that a greater
percentage of homebuilts with non-certified engines are fair-weather
daytime fliers, and therefore more likely to not be as subject to night
perception problems, flying into mountains.. etc., therefore reducing
the total number of "other" type accidents and making the engine-out %
higher.

I also wonder if homebuilts on average experience less other mechanical
type failures such as control linkage, etc., since the builder is
intimately familiar with the workings of the aircraft.

IMHO the ONLY valid percentages would be total registered auto engine
powered aircraft vs engine-out accidents compared to the same in
certified... now I do realize that those figures would be near
impossible to compile short of contacting every single person with a
registered aircraft that the engine type is not listed in the FAA
database... and even more accurate would involve total hours… ok, that's
highly improbable…..


Please note the above is based on no research whatsoever...I am likely
completely off base since I'm new to aviation in general.. ;-)


ET

Future student pilot and future Sonex builder.

On Sun, 11 Jan 2004 19:39:57 GMT, ET wrote:

I wonder what these figures really tell us.... I believe type of
accidents as a percent of total accidents really can't tell us anything.

What if accidents of other types (the rest of the %) where actually
lower in homebuilts with auto engines?

That was basically the start of this exercise, several months ago. The FAA
registration database lists the type of engine, but for about 3,600
homebuilts, it is either blank or merely lists "AMA/EXPR" for an engine.
That's about 13% of all homebuilts. Many of these are undoubtedly auto
conversions, but most are probably stock Lycomings or Continentals.

My January 6 2004 database shows 113 homebuilts with Ford engines. But how
many of those 3,600 "AMA/EXPR" engines are Fords? If even a twentieth of
them are Ford conversions, that would double the "known" quantity.

In the almost 700 homebuilt accidents in the 1998-2000 time period, the
NTSB records show only one "AMA/EXPR" engine installed on an accident
aircraft. Either those 3,600-odd airplanes aren't flying at all, or the
NTSB investigators are looking at what was *actually* installed under the
cowling.

What if things like icing, or VFR into IMC was more prevalent in
certified aircraft, skewing the engine related numbers into a lower
percent...

It would also be my guess (and only a guess it is) that a greater
percentage of homebuilts with non-certified engines are fair-weather
daytime fliers, and therefore more likely to not be as subject to night
perception problems, flying into mountains.. etc., therefore reducing
the total number of "other" type accidents and making the engine-out %
higher.

Heh, heh...you don't think I downloaded 21 megs of accident reports (at 56
kbaud!), read a thousand or so individual reports, and generated about a
gigabyte of databases and Excel spreadsheets for just a RAH posting on auto
engine conversions, do you?

I've just completed an in-depth analysis of homebuilt accident statistics.
In it, I not only determine the causes of homebuilt accidents during the
1998-2000 time period (35 basic accident categories, plus some
sub-categories), but I compare the accident rates for homebuilts to a
representative sample of certified light aircraft (Cessna 172s and 210s,
leaving out accidents occurring during primary training). And determine
the accident rates for first flights and in the test period, as well.

I completed the three-part article just this week. Hopefully, it'll run
sometime in the summer. Since I don't address the engine issues in the
series (I figured three articles was enough) I could post them here without
"stealing my own thunder."

I also wonder if homebuilts on average experience less other mechanical
type failures such as control linkage, etc., since the builder is
intimately familiar with the workings of the aircraft.

Sadly, no.

Ron Wanttaja

Ron Wanttaja wrote in message . ..

I completed the three-part article just this week. Hopefully, it'll run
sometime in the summer.

In which publication?

On 13 Jan 2004 14:35:24 -0800, (Corrie) wrote:

Ron Wanttaja wrote in message . ..

I completed the three-part article just this week. Hopefully, it'll run
sometime in the summer.

In which publication?

Whichever one accepts it. :-)

I'm a freelancer, so I never can count on an article getting printed. I
submitted the accident series to KITPLANES, but haven't heard whether
they're going to take it, yet.

Ron Wanttaja

On Wed, 14 Jan 2004 02:28:02 GMT, Ron Wanttaja
wrote:

On 13 Jan 2004 14:35:24 -0800, (Corrie) wrote:

Ron Wanttaja wrote in message . ..

I completed the three-part article just this week. Hopefully, it'll run
sometime in the summer.

In which publication?

Whichever one accepts it. :-)

I'm a freelancer, so I never can count on an article getting printed. I
submitted the accident series to KITPLANES, but haven't heard whether
they're going to take it, yet.

Ron Wanttaja

I curious, Ron. How long after submission do you wait for
acceptance/rejection before submitting to another publication?
Is there an industry standard?
Thanks

On Wed, 14 Jan 2004 05:05:24 GMT, andy asberry
wrote:

I'm a freelancer, so I never can count on an article getting printed. I
submitted the accident series to KITPLANES, but haven't heard whether
they're going to take it, yet.

I curious, Ron. How long after submission do you wait for
acceptance/rejection before submitting to another publication?
Is there an industry standard?

In the pre-email age, it used to take three months to find out whether an
editor was going to take an article or not. Actually, it generally took
three months or more to find that the editor was NOT going to take
it...they'd usually get hold of you fairly fast if they decided to go for
it.

For individual authors, it's considered bad form to send a piece to more
than one publication at a time. For some weird reason, it's perfectly OK
for an author's *agent* to do that. So I'd never send an article to a
second magazine until I got a definite "no" from the first one. In the
past, I've waited three months before politely "pinging" the editor.

It's best to be sure before sending the article on to another editor...if
the editor of the first magazine thought he'd told you and put the article
in the production cycle, the SECOND editor will be very miffed if he or she
sees the article they bought appear somewhere else, first.

Now that email is so prevalent, notification comes a lot quicker. The
accident series is the first non-solicited article I've sent the new editor
at KITPLANES. While I've done several articles for him on assignment, one
doesn't really need confirmation of acceptance on those, since one already
knew the editor wanted an article on a particular subject. But in this
case, it came "out of the blue" for him...and since the editorial office
just moved to the opposite coast, things are still shaking down.

If he decides it isn't right for him, there are other outlets (Like CBS or
ABC's "20-20" :-). If all else fails, I just post it.

The wait for notification could be worse. My wife (a romance novelist) has
had editors take ten months or more to get back to her about novels she's
submitted.

Ron "It's fun helping with research" Wanttaja