Febuary 2011 Issue
Safety Is No Accident
Are common assumptions about auto-engine conversions supported by an analysis of the data?
Back when I was a young ’un, I had a marvelous idea for my dad: “Let’s bolt a propeller on the Pontiac’s engine and put it on an airplane!” Modestly speaking, though, my 8-year-old self couldn’t claim to have invented the use of auto-engine conversions on aircraft. I was at least 40 years too late.
Like several other companies, Maxwell Propulsion Systems produced engine conversions for aircraft in addition to race-car engines.
Those looking for the lowest-cost route to flight have always been drawn to auto engines. The economies of scale meant that Henry Ford could churn out millions of Model A engines for a far lower per-unit cost than Continental could produce thousands of A-65s.
It’s not just folks looking for cheaper engines. Many also are dismayed by the lack of technical advancement in the aviation-engine world. They hope to take advantage of the gains in material and computer technology that are featured in modern car engines. Or perhaps they’re building replicas of airplanes that had in-line or liquid-cooled engines, and can’t fit an opposed four- or six-cylinder aircraft engine.
In these cases, builders’ gazes often slide to the engines that provide trouble-free power for their ground transportation. But auto-engine conversions in aircraft have a spotty safety reputation. Is it deserved? Let’s take a look at the NTSB accident statistics.
Auto-conversion developers often use production aircraft as test beds to simplify comparison to traditional engines. Northwest Aero Products re-licensed this Cessna 172 in the Experimental Research and Development category to test its Chevrolet V-6 conversion for homebuilts.
I’ve gone through more than 2000 accidents in my 10-year database (1998 through 2007) and assigned them to one of seven engine categories:
• Traditional (production reciprocating engines such as Lycomings, Continentals, Franklins, etc.)
• Auto Conversion (Ford, Chevrolet, GM, Honda, Subaru, VW, etc.)
• Two-Stroke (Rotax 503, 582, Cuyuna, etc.)
• Non-Certified Four-Stroke (Rotax 912, Jabiru, RotorWay, etc.)
• Aftermarket Traditional (Experimental versions of traditional certified engines such as the Superior XP-360)
• Foreign (LOMs, M-14Ps, and other foreign reciprocating engines) and
• Turbine (all jets and turboprops)
The number of accidents involving aircraft with the last three categories is low enough to be statistically insignificant.
Figure 1: Engines in Accident Aircraft.
Figure 1 illustrates the kinds of auto engines that were installed in fixed-wing homebuilts in the accident database. Volkswagens and their derivatives lead, with Subaru bringing up a distant second.
A similar graphic for rotary wing homebuilts would have been pretty boring: Subarus constituted more than 80% of the auto-engine conversions, mostly on gyros.
Accidents Caused by Engine Failures
For the first pass through the data, let’s examine the number of accidents that were caused by a problem with the engine. The bars in Figure 2 represent the percentage of the total number of accidents that were due to engine-specific problems such as internal failures, ignition-system failures, cooling problems, etc. This includes fuel problems on the engine side of the firewall, but not problems with fuel tanks, airframe fuel lines or potential pilot-induced issues such as carburetor icing.
About 13% of the accidents involving fixed-wing homebuilts with traditional powerplants were due to engine failures, versus over 35% of the accidents occurring to aircraft with auto engines. If an accident occurs, it’s almost three times as likely that it was due to the engine when an auto conversion was installed.
Figure 2: Accidents Due to Engine Failure.
Note the if. The results do not mean that such aircraft have an accident rate three times higher. About 99% of homebuilt aircraft don’t crash in a given year. We’ll take a stab at comparing the actual accident rates later.
It’s sometimes difficult to adapt horizontally opposed, air-cooled aircraft engines to replica aircraft that used liquid-cooled engines. Auto engines are often suitable.
One interesting result in Figure 2 is the difference between fixed-wing and rotary-wing auto-engine statistics. Sixty-five rotorcraft with auto-engine conversions had accidents, but the engine was the cause in only four. This is much better than the fixed-wing rate.
It’s a curious result. The vast majority of the auto-engine rotorcraft accidents involved gyros. Perhaps because the gyro is operating in autorotation full time, an engine failure in a gyro is less likely to result in a reportable accident. Perhaps the operating conditions for gyros are more favorable than for fixed-wing aircraft. Or maybe the gyro community has figured out how to make a Subaru conversion reliable.
Engine Failure Causes
What are the causes of the engine failures? Figure 3 shows a comparison of auto engines and traditional aircraft engines. Note that auto conversions have a higher rate right where you would expect it—in engine cooling and in the systems that convert engine power to a slower speed for a propeller or helicopter rotor.
Figure 3: Causes of Engine Failures.
Traditional air-cooled aircraft engines certainly can overheat, but the failure modes are usually gradual. Many auto conversions are liquid-cooled, which produces the opportunity for a failure, resulting in a catastrophic loss of coolant. Builders also are often on their own when it comes to designing coolant systems. Baffles for traditional air-cooled engines can produce their own problems, even though good examples are as close as the nearest Cessna.
Similarly, most traditional aircraft engines are direct drive, and do not require a propeller speed reduction unit (PSRU). No matter how reliable a mechanical device is, not having one completely eliminates the potential for failure.
Two other failure categories stand out when comparing auto conversions to traditional engines: ignition-system failure and electrical failure.
The AeroVee engine used in the Sonex is a Volkswagen conversion.
Magnetos, cursed 19th-century relics that they are, don’t require an external power source. No one trusts them, so usually there are two installed for redundancy. When one quits, the other usually keeps going, and smart owners don’t fly until the bad one’s working again.
Modern electronic ignitions are more reliable—but sometimes only a single system is installed, or two systems have a single, common failure point. The Ignition category covers only those instances where the ignition system itself fails. Most of the Electrical Failure cases are unrelated problems that result in cutting off power to the ignition. When you combine the two, you find that about one in six auto-engine failures is due to cessation of the ignition system.
Notice that the Auto-Engine category scores significantly better in the Carb Ice category. This is one item that can be both hardware- and pilot-induced. But many auto conversions install the carb atop the engine in a relatively warm place, and they are probably not as prone to icing.
Figure 4: Specific Engine Types.
When replacing traditional engines with conversions such as this V-8, careful attention to CG issues is necessary.
The totals for individual engine makes are fairly low, but Figure 4 compares the results for three of the most common on the NTSB accident listings. The GM listing includes both liquid-cooled engines and the air-cooled Corvair.
Figure 5 compares the percentage of the total accidents versus the number of hours on the aircraft. About 7% of all traditional-engine accidents happen in the first 10 hours versus more than 20% of the auto-engine conversions. Nearly a third of all auto-engine homebuilt accidents occur in the first 40 hours, versus 12% of the traditional engines.
To me, the message is pretty obvious: The problem with auto-engine conversions isn’t a lack of fundamental reliability, but in getting one to work properly in the first place. The test period is intended to drive out “teething problems,” but an engine failure is almost always a critical situation.
Figure 5: Hours on Accident Aircraft.
The Subaru’s horizontally opposed layout reduces the length of the engine package, making installation easier.
Accident Rate Comparison
OK, admit it: The main reason you’ve slogged through this far is to find out the difference in the accident rate between traditional engines and auto-engine conversions. Easy to compute, right? Just divide the total number of accidents for auto-engine conversions and traditional engines by the total number of each type on the FAA registry.
Not that easy, I’m afraid. The FAA aircraft registry isn’t too descriptive when it comes to describing the engines on homebuilt aircraft. As Figure 6 shows, about a quarter of homebuilt registrations (more than 7500 aircraft) merely state “AMA/EXPR” in the Engine Make/Model columns. It’s a generic category that basically states that a non-type-certificated engine is in use.
Figure 6: Engines Identified for Homebuilts in FAA Registration Database.
These could be auto engines; they could also be traditional engines where the builder decided not to maintain certification status. Builders should install their own data plates, but some don’t.
Getting insight into these registrations is important if we want to compare accident rates among engines. We know many of them are auto conversions, but how many? There are four times as many “AMA/EXPR” engines in the registry than official auto-engine conversions! We need to estimate how many of those AMA/EXPR are traditional engines and how many are auto conversions.
To begin, I cross-referenced the N-numbers with AMA/EXPR engines with my NTSB accident database. The NTSB accident reports are pretty specific as to engine make and model, which meant that for more than 100 AMA/EXPR-engine aircraft, I now knew what engine they actually carried.
Only 10 airplanes are listed in the FAA aircraft registry as having Mazda rotary engines, but many more might be registered as having AMA/EXPR engines.
It stands to reason that there should be the same proportion of traditional and auto engines among those aircraft as with the overall AMA/EXPR fleet. The percentages were 29% and 31%, respectively. So we can add 29% of the 7500 AMA/EXPR to the Traditional Engine column, and 31% of 7500 to the Auto-Engine Conversion category. The impact to the Traditional column was minor, but the number of auto-engine conversions more than doubled.
Using this process, the accident rate for auto-engine homebuilts came out about 23% higher than for homebuilts with traditional engines.
Is it possible to beat the higher accident rate? Certainly, there are numerous examples of successful conversions. But converting an auto engine to fly is not as simple as some might lead you to believe. Remember, the goal isn’t just getting the homebuilt to the airport for the first flight. You’re looking for thousands of hours of safe and reliable operation. As the above discussion illustrates, it’s going to take more work, and more caution, to achieve that with an auto-engine conversion.
This SE-5A mounts a Ford V-6 inside its narrow cowling.
Keep in mind that all homebuilts have one major thing in common: the word “Experimental” on their airworthiness certificates. Just as a builder has to understand there’s more risk flying an Experimental/Amateur-Built aircraft than a Standard category one, there’s going to be more risk using an unproven engine to power the aircraft.
The ideal aircraft powerplant has a good power-to-weight ratio, and is reliable and inexpensive. In reality, you’ll achieve only two out of the three. Make sure reliable is one of them.
Ron Wanttaja is a systems engineer, engaged in satellite orbit/constellation design and analysis, launch vehicle and onboard propulsion system trades, and operations concepts for space systems. He worked on the early design studies for the International Space Station.
IMPORTANT CUSTOMER SERVICE NOTICE TO KITPLANES SUBSCRIBERS:
All mailed correspondence, including subscription invoices, renewals, and gift notices, will bear our address:
PO BOX 8535
Big Sandy, TX 75755
Third parties claiming to be selling KITPLANES subscriptions are not legitimate.