Why I chose Subaru

I started building my Cozy MkIV in 1995. Immediately I started to research engine choices. As you can imagine, at that time, it was extremely difficult to obtain any facts related to risk. There was no place you could obtain Lycoming failure information. (Even now, a dozen years later, my website is the only place this is available Risk measurement).




I like facts. I do everything I can to avoid feeling based decisions. I also avoid decisions based on theory. I canít emphasize this enoughÖ.there is a huge difference between facts, theory, and feelings. Now Iíve noticed over the years that a large percentage of the population, Iím guessing something like 60-70%, donít make any distinction between these methods. At least not when it comes to arriving at conclusions. So all of their decisions are a blend of these components. This leads to emotional based conclusions. Leads to cognitive errors.


Rotary Engine:

One of the first engines I researched was the rotary engine. I worked with an engineer who was a rotary nut. Raced them in his free time. Itís tough to get the complete story from someone thatís so emotionally attached. But he was one of those special guys that is quite analytical. For example, when I asked about what commonly fails, he was able to rattle off all of the common risk items. ďIf you were going to put one in a plane, what would you do?Ē So he talks about replacing this component, another, on and on. Replacing the heavy cast iron components, some side cover component that develops leaks. Compression seals. Then we talked about the heat rejection issues with the engine. A large portion of cooling is accomplished using less efficient oil. Hot exhaust, obviously higher fuel use.

We also discussed the positives. The relatively forgiving failure modes. Many failures were gradual degradations, like the coolant leaks. Handled high torque well. Few moving components. Compact package likely to fit envelope. Etc.


This was enough that it was pretty clear I should NOT choose this engine. It fits some familiar failure patterns. The more things you change, the greater your risk. This is because every time you dip your hands into a system, you suddenly have added 30 potential risks. Installing something wrong, dinging it and not getting a good seal, foreign material getting in it. On and on. On top of that, having to design and construct your own components is extremely risky. Itís fun, just very risky. Also, from the descriptions, it was clear this is not a robust engine. Robust has special meaning. Iíve heard people describe the rotary as low risk because it has fewer moving parts. This shows they are clueless about what risk is. It doesnít matter how many moving parts you have. It matters how often the system fails. It matters the effect of the failure. See Risk measurement


So here it is, a dozen years later. We now have sources for fact based decisions. Iíve since done some preliminary research on actual rotary failure rates in aircraft. Itís atrocious. A very poor choice for aircraft. I estimate the actual failure rate is 4 to 8 times higher than the Lycoming.



What fails?

Lets look at the significant risk failures.



These engines are electron dependant. No mags. So Electrical related failures are very common. Typically loose wires, poor connections with no consideration of what happens if the circuit fails. Where the Lycoming would continue running after loss of electrons, the rotary goes silent. Many of the users resort to using custom ECUís. Low volume units that are hugely deficient in their programming and design. Two forced landings this year, almost a third.



Loss of compression periodically occurs. Three or four this year alone. Any foreign material in the combustion chamber wipes out the seal. The seal is sensitive to installation, wear of mating components.

Talk to any user and ask them how many times theyíve rebuilt the engine. Youíll hear two, or three times!



Exhaust failures are very common. Iíd estimate 90% experience failures. Custom exhaust manifolds, pipes, mufflers that fail due to the extreme heat and vibration. Substantial fire risk and risk to nearby components. Particularly risky to canards which have prop behind exhaust. Fire and odors not likely to be noticed soon enough. I donít recall any of these actually causing a crash yet, but itís risky.



Cooling related failures are fairly common. Loss of coolant internally through cover seal. The engine is reportedly very sensitive to coolant temp. Where reciprocating engines operate just fine at temps as high as 225 F, the rotary life is reportedly reduced if temps get over 185 F. So these guys have huge radiators, extra oil coolers, extra connections that see high energy pulses. One forced landing last year related to cooling.


Loss of power on takeoff:

If you are inclined to use aviation fuel regularly, then you will periodically experience power loss on departure. Some peculiar trait of the engine that shows up every 40 hours or so.


Voice of REASON:

I detail all of these failures because the rotary is the only engine that is propagandized. Grossly distorted promotions of this engine like no other. So here we have this engine that fails 4 to 8 times more frequent than any other, and itís promoted as the safer alternative. Lamar uses emotional argumentsÖ. like referring to reciprocating engines as ďhand grenadesĒ waiting to blow. More people will die as a result. Itís really sad. So I offer this information as the alternate view point.


Make your own decision:

You donít have to take my word for anything. There are now facts available just waiting for you to make a safe decision. All of the engine options have enough hours collected by pilots that you can determine your actual risk. Use the same method I used to arrive at 1815 mean time between failures (MTBF) for the Lycoming. Just apply that method to your engine of interest. You can measure it just by adding up all of the incidents reported in newsgroup archives. The facts are sitting there waiting for you.


But please, use the thinking part of your brain. DONíT allow yourself to make decisions based on some ďexpertĒ opinion, popular theory, or emotional need.


Subaru 2.5 engine:


Modern design methods:

My career has been focused on methods to reduce failures. Iím really good at it. There are patterns to success and failure.

Auto engine designs use new methods to increase reliability. They are very effective. Components are tested in environments far more severe than the engine will normally see. Computer modeling is accurately done to increase quality margin. Iíve had my hands in all of this. That helps me appreciate the significance of new auto engine advances. Iím told that Subaru, in one year, makes more engines than Lycoming has their entire history. That makes for millions of dollars available for product development that Lycoming can just dream of. More importantly, auto engines are routinely improved, modified. If Lycoming makes any improvement, that opens them to lawsuits for all previous designs. So they stagnate. Look how long it took them to implement fuel injection. Clearly, long term, auto engines are the way to go.


What fails?

I had detailed information on the failure rate of the 2.2 liter engine. See Risk measurement for details. The 2.5 is higher hp and itís design is essentially the same as the 2.2. So it was pretty easy to extrapolate which components were likely to fail on the aircraft. My copilot buddy worked on Subaruís for a living. A very analytical fellow, Brian was a wealth of information.

Basically all the facts suggested no deficiencies. During my ground test phase, I uncovered a problem with the 2.5 liter engine. Itís prone to head gasket leaks. More details on my 2.5 install page.



It looked pretty clear I could adapt the engine to the plane with few modifications, so that factor alone greatly reduces my risk. If you analyze root causes for conversion failures, mods are the number one cause.



This is an all aluminum engine. It doesnít matter what type of engine you look at, the weight follows hp. So I knew that it was in the ball park. As it turned out, my plane is the lightest of all the Cozy IVís to date. You may not believe it, amusingly enough, I never did weigh the engine!


The emotional factor:

Have to admit, part of my reason for using this engine was because I like the concept of contributing to the auto conversion cause. Iím also motivated to ďbe differentĒ. I enjoy the challenge of using my QA and testing background. I did my best to subtract these motivations from the decision, but easier to say than do.



I ended up with a very safe conversion. I had better cooling characteristics than most. I could deliberately† inflate the engine temps during taxi, then depart full power. Full power to 12k ft (no pilot O2), and all that while the temp dropped to itís normal 199F. Incidentally, I did this test twice when the local temps reached 100F (rare in NW Oregon).

I had only one incident during the 240 hours, 4 years of flying this engine. I experienced loss of around 300 rpm while cruising at 12k ft. This was caused by partially clogged fuel injector. I knew better, but I opened up my fuel system a few hours earlier. Obviously introduced a contaminant post filter.

I did have one component that I replaced. My TPS (throttle position sensor) developed a dead spot. This gradually caused the engine to run rough when first started up. Never a risk item.


So now I pursue the new 3.0R engine installÖ...