Mostly it's the usual suspects: fuel, weather and runway prangs. But poor maintenance is a surprisingly big player, as Aviation Consumer reported.
|Multi-Engine Survivor's Guide
The accident record yields some useful advice for both the budding and inveterate pilot of a twin-engine aircraft (see charts at bottom of main article). Although we found some surprises in the data, we also didn't see any trends that convince us that the risk of an accident can't be mitigated.
Here's our take on how a twin owner can whittle down the risk, in order of importance and likelihood of accident:
Avoid short runways: Overruns, undershoots and excursions are common. Some occur because pilots attempt to operate from runways with marginal length or with obstructions. For most twins, 3000 feet of runway, with clear overruns, should be the practical minimum.
The longer the runway, the more options you have to accelerate and stop or handle an engine-out situation.
Be conservative about deferred maintenance: It's a fact of ownership that pilots fly broken airplanes and minor maintenance squawks aren't necessarily always risky, whether in a twin or a single.
But when it comes to engine squawks and landing-gear issues in twins, the owner with a highly evolved survival sense won't allow minor nits to slide by but will fix them as they occur. All too often a nickel-or-dime maintenance issue brings down an otherwise airworthy twin.
Know the systems: Twin systems aren't necessarily more complex than those found in singles but there's usually two of everything, thus diagnosing abnormals -- especially while in IMC or at night -- can be overwhelming. Owners should know fuel and electrical systems cold, including the operating theory behind each.
Keep it light: At gross weight, every GA twin made is supposed to be able to climb on a single engine. The reality is that some won't. To build in safety margin, keep the airplane as light as possible. If it's not necessary to tanker fuel, don't.
And if you've moved up from a single, don't assume constraints against carrying heavy baggage no longer exist. The lighter the takeoff weight, the better. Further, understand the implications of an aft CG on single-engine handling.
Train regularly: An engine-out situation in a twin is more demanding than in a single and far more likely to be fatal if mishandled. Engine-out and systems drills every few months are cheap insurance against screwing up during an emergency. And they don't have to be done with a CFI to be effective.
Give ice a wide berth: From reading the accident record, we suspect that many pilots of twins assume that boots and heated windshields -- not to mention extra power -- make them impervious to icing.
Yet a common icing accident is an ice-burdened twin smacking hard onto a runway, with serious damage to the aircraft. Known-ice packages are escape tools, not magic bullets.
Practice landings: As with singles, runway excursions and loss of control in crosswinds is the highest single explanation for why pilots crash twins, even if the miscellaneous category accounts for more wrecks.
Close the doors: It shouldn't happen at all, but cabin and baggage doors that pop open on takeoff are a measurable accident cause. Securing them properly will help, as will accepting that the airplane will fly perfectly well with a door ajar.
Have a pile of money: Given the spike in gas prices, the value of twins may again soften, making them appear to be attractive bargains. If you can't afford to maintain a twin, fly it regularly, train often and put enough gas in the thing, you're better off with a single.
This article appeared in the December 2001 edition of Aviation Consumer and is reprinted here by permission.
Like the speed of light and the freezing point of water, one of the constants in the universe is this imponderable question: Is a twin safer than a single? Does having two motors really add measurably to the safety margin or is it, like so much else in GA, undiluted marketing hype?
If we could answer this question, we could also turn lead to gold and all of our computer hardware would actually play when plugged. In the end, however, we think there are too many variables to say with assurance that an extra engine really does improve the safety margin for the average owner-pilot. We emphasize average as a means of noting that not all owners pursue the recurrent training that makes them competent in twins.
We think the best you can do is analyze the twin-engine accident record across a range of models, try to determine how and why twins crash, then decide for yourself what could have been done to prevent those crashes or to otherwise lower the accident risk.
And that's exactly what we've done here. Continuing our examination of accident records for individual models, we've reviewed the NTSB reports for six twins, ranging from the Piper PA-23 Apache/Aztec series to Cessna's popular 421 cabin-class cruiser. We looked at 545 accidents during the five years between 1994 and 1998.
And once again, the monthly dose of salt. Because of coding errors in NTSB reports, we're sure we didn't see all of the accidents. Moreover, NTSB reports sometimes don't give final cause or the final cause may be incorrect. In some cases, then, we've read the factual reports, taken them at face value, and assigned our own apparent final cause. Also, we don't have accurate data on hours flown for each model -- the best and only legitimate means of measuring one machine against another on an accidents-per-hours-flown basis. We've used estimates for each model provided by the Aircraft Bluebook Digest but these are based on reported hours at time of sale and may or may not be accurate. Absent better data, they're the best we can do.
Last, the other great unknown: How many twins land safely with one engine shut down but never make it into the NTSB's reports? Some, we're sure. Lots? We're not so sure. Some years ago, we examined the tower logs for nearby Bradley International Airport and learned that over the course of six months, an average of one twin a month landed with one engine caged. Surprisingly, half of these were turbine-powered airliners with a precautionary shutdown, a none-too-unusual development in the airline world. The rest were piston twins. That said, we still have no handle on how many twins land successfully with an engine out.
With these caveats in mind, we examined accident records for these aircraft: Piper's PA-23 Apache/Aztec, PA-31 Navajo and PA-34 Seneca, Cessna's 310 and 421 and the Beechcraft 55/58 Baron series. We realize there are performance differences between the Apache and Aztec and amongst the various Baron iterations. Nonetheless, for our analysis, we've lumped these types together.
Why Twins Crash
Right out of the blocks, our analysis revealed one difference between twins and singles. Our past accident reviews have indicated that the overwhelming reason that singles crash is because pilots lose control during takeoff and landing. They botch crosswind corrections, land too fast or too slow, careen into ditches or fences in the overrun or otherwise turn perfectly good pavement into an accident scene.
In general, twin accident causes are more diverse, however, which is why the "other" category in our charts resides at the top of the scale for each model (see charts at the bottom of this article). In reviewing the accident reports, we found so many oddball accident causes that we automatically assigned these to the miscellaneous category.
For example, how do you categorize a Cessna 310 pilot who experienced a night electrical failure in clear weather, had a flashlight, but still lost control of an airplane with two perfectly good engines? Or the pilot who neglected to remove the towbar after preflight, took off with it still attached to the nosegear then aggravated his problem by completely botching the subsequent emergency landing by shutting down the engines in an effort to save the props? That happened, too.
We do see some patterns in this diversity of accident causes, however, including a smattering of training-related wrecks -- although not as many as we expected -- lack of maintenance and quite a large number of controlled-flight-into-terrain accidents or CFITs. We see CFITs among singles, too, but not quite as frequently. We have a theory about this but more on that in a moment.
A Simple Failure
We were struck by the number of twin-engine accidents that began with a simple failure or even a minor abnormality that soon ballooned out of control, resulting in the loss of the airframe with occasionally fatal results.
For example, if you fly something like a Saratoga or a Cessna 210, it's hard to imagine how a baggage door coming open would cause you to lose control. Yet we found a half-dozen instances of this in the twin-engine accident record involving either cabin doors or baggage doors.
In one accident, the pilot of an Aztec spun the airplane into a hangar, killing all four aboard, after the nose baggage compartment door came open.
Investigators found no evidence of engine failure or any other anomaly other than the baggage door. We found several accidents that shared this basic scenario, some fatal, all of which resulted in damage to the aircraft.
With few exceptions, investigators note that the POHs for these aircraft advise the pilot that the airplane continues to be flyable with baggage and main cabin doors ajar. Even the simplest, most benign oversights can lead to expensive consequences. In one accident, the pilot of a Cessna 310 forgot to turn on the alternator switches after engine start, didn't catch his error and depleted his battery reserve. After lowering the gear manually, the pilot discovered that the left main failed to lock and collapsed on touchdown, damaging the airplane significantly because of a trivial mistake by the pilot. (This may argue for simply leaving the alternator/generator switches on, when practical.)
Some pilots kill themselves by simply not understanding the aircraft systems, chiefly the fuel plumbing. Mismanaging fuel crossfeed during single-engine drills can turn a ho-hum proficiency exercise into a genuine no-engines survival struggle. It seems obvious that if a pilot loses one good engine due to fuel mismanagement, he stands a good chance of losing the other when trying to right the situation.
When Engines Fail
|Common twin crash causes: fuel exhaustion, above, and runway loss of control, below. The former is often fatal, the latter rarely
Does having two engines increase the likelihood of an engine failure? We're not sure but we would guess it does, since the more engines you have, the more likely something will break. More to the point, if one does fail, would you have been better off with just one motor in the first place? In some cases yes, in other cases no.
Between 1972 and 1976, the NTSB investigated the outcome of twin-engine crashes and concluded that in the event of an engine failure that resulted in a crash, the likelihood of it being fatal was four times greater than a crash in a single. (Despite the paucity of current data, it's still a lead-pipe cinch that a single with a failed engine is coming down somewhere. It may or may not result in a crash.)
Our most recent review of twin accident data tends to support this finding in general, if not to the same decimal point. Typically, in crashes where an engine failed in a twin, the accident was fatal between 20 and 50 percent of the time; in singles, the fatal rate for powerplant failure is lower, on the order of 10 percent, somewhat variable with model.
Why this is so is not too difficult to understand. Bluntly, pilots sometimes screw up engine-out emergency procedures by reacting too late to the developing situation or simply losing control. In some cases, they expose themselves to certain disaster by expecting too much of the airplane's single-engine performance or pushing the weight and balance envelope to the breaking point.
In one accident, a Baron was loaded 800 pounds over gross weight and couldn't maintain altitude when one engine failed. No surprise there. It crashed, killing all six aboard. It didn't help that the pilot had a fresh multi-engine rating and little experience in the airplane.
Along the same lines, we often report that GA twins are poor performers on a single engine and even though the POH promises a 200 to 300 FPM climb on one motor at gross weight, the pilot has to do everything right to make that happen. We thus expected to see many accidents in which the airplane simply wouldn't climb. Happily, that wasn't the case. In fact, we saw only five over the five-year period we examined and of these, only a couple didn't involve extenuating or pilot-induced circumstances such as being overweight or botching the emergency drill.
One in which everything appeared to have been done just right involved the ditching of a Piper Navajo near Kennedy Airport in 1995. The Navajo, with six aboard, suffered a failure of the left engine while descending into Farmingdale, N.Y.
The pilot reported that the cowling was open -- probably from the departing cylinder that cratered the engine -- and the best he could do was a 300 FPM descent into the water. All six exited safely but one passenger died from cardiac arrest.
We expected to see more instances of the pilot feathering the wrong engine during the emergency drill but, again, only a handful of these were reported in the NTSB records. It's possible the frequency of such accidents is higher but simply isn't reflected or discovered in the NTSB investigation.
Similarly, given how poorly some twins do on a single engine, we thought we would find more examples of the multi-engine pilot's worst nightmare, the VMC rollover. Again, only seven of the 545 accidents -- about 1 percent -- we reviewed appeared to be genuine rollovers. That's a better record than we imagined, suggesting that VMC-roll avoidance training may work.
Let Me Demonstrate
Speaking of training, in the hoary days of the 1960s, when multi-engine training was booming, one school of thought allowed the mixture to be pulled to simulate an engine failure. Unfortunately, that often resulted in a real crash so by the mid-1970s, the instructional community had wised up, using zero-thrust rather than idle cutoff to simulate a failed engine.
Yet training accidents still happen. These are caused by across-the-board mistakes but the two most common seem to be runway loss of control and another big no-no some CFIs still haven't learned to avoid: the single-engine go around.
The runway biffs occur for the usual reasons, but are complicated by the fact that many landing drills are done with one engine at zero thrust. Obviously, this kind of training has to be done but it's riskier than landing with two engines, especially if a crosswind complicates the equation.
In the most benign form, the student lands hard and bounces and can't quite regain control before the airplane takes a tour through the edge lighting or into a ditch or suffers gear damage from the hard impact.
The more vicious variety is the single-engine approach in which a high sink rate develops and the pilot/CFI waits too long to arrest it or makes the training all too real by delaying use of the zero-thrusted engine. These tend to result in undershoots or wild excursions as the other engine is finally brought into the game.
Most baffling of all is the intentional single-engine go around. In a couple of cases, these involved airplanes that got too low on approach or a runway was blocked at the last minute by a vehicle or another aircraft. The CFIs let the student attempt the go around on one but the more conservative response is to bring the caged engine online and execute an orderly, two-engine go around right away. If that's not possible, landing is the better option, navigating around the obstacle on the runway as necessary.
These days, most CFI-MEs advise against the single-engine go around but a few hardcore throwbacks still teach it. We think they're nuts; it's something you wouldn't do in the real world -- at least in a piston twin -- so why train for it?
A related training issue is the notable number of "improper IFR" accidents we uncovered, which we generally ascribe to the CFIT or controlled flight into terrain category. We seem to see more of these in twins than in singles and our theory is that twins operate more often in bad weather, possibly because the pilots are more comfortable in weather with the extra engine. Yet without recurrent training -- especially instrument training -- lack of proficiency puts them into obstacles.
Things That Break
One of the most striking trends we noticed was how often a minor maintenance glitch turns into a bona fide emergency and all too often brings an airplane down. We recorded at least 25 accidents in which faulty maintenance was implicated. Our suspicion is that there are probably many more.
Landing-gear failures were common, due to broken or worn linkage or improperly installed or missing parts. So were engine anomalies of some kind, leading us to wonder if there's a twin-engine mindset that causes an owner to cut corners on engine maintenance because there's another one available if one fails.
Things like deferred exhaust maintenance, postponed magneto overhauls, fuel system foul-ups and prop squawks brought down more than one twin. The important lesson is that in terms of pure survivability, engine maintenance is more critical in a twin than in a single because if one quits in a twin, you've got your hands full and the consequences of botching the emergency response result in markedly higher chances of not living through the experience.
Owners who cut corners on both maintenance and training are asking for trouble, since they expose themselves to greater risk of actually needing the high-level survival skills they don't have because they haven't trained.
As noted, we can't argue that our accident rate data is anything other than a curiosity, since the hours flown by each model are so difficult to pin down accurately. Nonetheless, there are clearly model-specific accident patterns.
For example, we noted that the Beech Baron series has a dramatically higher number of gear-up and inadvertent gear retractions than do the other models. This is due in part, in our view, to the fact that in the early models, Beech put the gear switch to the far right and the flaps on the left, the reverse of the industry standard.
We also noted that the more complex twins -- the Cessna 421 and Cessna 310 -- seem to have a larger percentage of maintenance-related accidents than do other models. The Apache/Aztec series shared this dubious distinction, perhaps more due to aging airframes than complexity. Barons, on the other hand, had the fewest maintenance-related crashes, supporting owner claims that the airplanes are well-built and don't break much.
Two surprises: The dowdy Piper Seneca had the highest incident of in-flight breakup of any of the twins. Although not model specific, we were surprised at number of accidents apparently attributed to impairment, due to drugs, alcohol or illness. There were a dozen such accidents during the five-year period we examined.
Having reviewed the accident database, we conclude what most rational owners already know: A well-maintained twin with a current, well-trained pilot is probably safer than a single, provided it isn't flown over gross weight. The safety accrues, of course, from an additional engine with dual vacuum pumps and dual alternators.
If one engine fails, a competent pilot has real, not illusory, options by having meaningful backup systems to fill in until the airplane can be landed safely on one engine. The accident record seems to suggest that a pilot who's overtaxed by a system or engine failure is less likely to survive in a twin than in a single.
For that reason, in our view, on safety issues alone, an owner contemplating upgrading to a twin shouldn't consider the proposition unless there's enough money in the kitty to maintain the airplane to the highest reasonable standards and to both fly often and train often. To do less probably results in an overall reduction in safety.
|Twins Overall Accident Summary. Graph above shows overall apparent causes of crashes, by percentage of total. Graph below breaks down the miscellaneous/other category into specific causes, as listed in NTSB accident reports. Totals don't necessarily add up to 100 percent because some accidents are too weird to categorize.
|What "Other" Means