Pelican’s Perch #4: Engine Failure!

It ain't easy being contrary, but AVweb's John Deakin makes the effort worthwhile. Engine failure in a piston twin is no time to be messing with complicated procedures that some seem to favor. John lays out his straightforward ideas on how to react to this critical emergency - and explains why in detail. There's more to it than just "identify, verify, feather" or "dead foot, dead engine." Once again, the real world requirements that could save your life may not be well served by some of "the old ways."


The only other words that will more quickly capture the attention of the steely-eyed, granite-jawed airline captain are “Uh, oh, they forgot to load the crew meals,” or “Company’s talking about a cut in per diem.” The latter two don’t even bear thinking about, but we do cheerfully train for the engine failure.

But let’s omit airline pilots and modern jets from this discussion and talk only about piston-powered aircraft with more than one engine.

It is clearly impossible for a piston engine (“recip”) to work at all, especially in an airplane, much less run for hundreds, even thousands of hours. There are an incredible number of parts traveling in all different directions, temperatures that vary widely, from sub-freezing to nearly 2,000°F, and centrifugal forces that defy imagination, whether we’re talking about the little 65hp Continentals, or the giant, 500hp Wright R-3350, the all-time classic “Big Radial” from days gone by.

Impossible, I say, but like the bumblebee, somehow they do work, and fairly well at that. However, I think it’s fair to say that any pilot even moderately involved with flying any of them will probably experience an engine failure, which means we need to prepare and train for it.

These failures are a very big deal, for several reasons. The unbalanced power causes immense control problems and available performance is always critical, with rate of climb often measured in tens of feet per minute (sometimes up, sometimes down.)

Are Two Really Better?

Many jump much too readily to the conclusion that two engines are better than one, three are better than two, and so on and so forth. The grizzled old-timers will growl “When I say ‘feather four’, I want the Flight Engineer to say ‘Which four?'”. Or, “I want to look out my window, and see engines just as far as I can see.” Cute comments, but designers have long known that for any given power (or thrust), the less hardware (and human-ware) the aircraft has to carry around, the more economical the operation. So it’s a constant tug-of-war between the bean-counters who would like just one big engine with no pilots at all, and the pilots, who can never have enough of anything (“Faster horses, more women, better whiskey” – to which I would add “and more engines.”)

True to form for all pilots, the GA pilot will look longingly at the nifty Barons, Cessna 310s, and others, and think all the limitations of his single would just go away, if only he had two engines.

Well, maybe, maybe not.

The Critical Moment

Take any SINGLE on takeoff, gear just coming up, the far end of the runway just dropping out of sight-and the engine fails. If the pilot can find something soft to hit, or level ground, and keeps on flying the airplane until it stops, the odds are pretty good everyone will survive, and may even walk away from the crash. The aircraft should be doing around 60 knots or less at impact, and while that’s certainly no picnic, it can be, and often is survivable.

Now take the typical light twin, same situation. If the pilot is well-trained, current, alert, good, and fairly quick, and the airplane and the remaining engine are in decent shape, he will probably get the airplane around the pattern, provided the density altitude isn’t too great, and the airplane isn’t loaded too heavily for the conditions, or loaded too far aft. This assumes gross weight at sea level on a standard day. Gross at any higher density altitude may be legal, but it’s not very safe!

There are a fair number of caveats in that paragraph, and we could probably add some more, with a little thought. If those conditions are not well met, one of two things is going to happen. Either the pilot will lose control when the speed drops a little, and roll it up in a flaming ball (survival odds, zero), or he’ll pull some or all of the remaining power off and land it. It seems that too many try the former until it is too late, and unless you’ve seen a really, really valid demonstration of VMC, you have no idea how fast most airplanes can do a fair imitation of a snap roll, if mishandled.

In the unlikely event the pilot has the training and the presence of mind to reduce the power on the good engine and put the airplane down, the odds of survival are not going to be as good as they would be in the single, because the twin is much heavier, stalls at a much higher speed, and will be carrying a lot more fuel. That translates into much more energy to be dissipated, and unfortunately, some of that extra energy usually gets imparted to the occupants.

Frankly, I have little confidence that most GA pilots will successfully handle a full engine failure during that small window of time right after liftoff. Mercifully, actual complete engine failures during those few seconds are very rare.

Either we should train pretty hard for this to make pilots better able to handle it, or we should quit training for it at all, because we do kill some small number of people in training accidents!

On a brighter note, clearly, once the twin gains some speed and altitude, the safety does increase, because the failure is less critical from a skill point of view, and the extra engine will expand the options, and buy some time to think, to make better decisions, and to plan. Even if the airplane cannot maintain altitude on one, the very slow descent can be put to very good use. If an engine is going to fail at night, over the Rockies, IFR, we’d all rather have a spare engine, or three, but it is still no trivial matter, even for the expert.

How about VMC?

Most multi-engine pilots can usually rattle off a quick definition of “VMC,” if asked. What very few realize is that VMC is only a speed determined by test pilots, on which other useful speeds are based. That is the ONLY purpose for VMC. It is NOT a speed that is useful for pilots, in day-to-day operations. I’m not even convinced it’s useful to know it, although it’s certainly a common enough question on the various orals. Very, very few pilots are good enough, or quick enough, to maintain control with a sudden engine failure anywhere near the “book” VMC, even when expecting it, so a practical “real world” VMC is much higher, by some unknown number of knots.

In practical terms, unless the engines are supercharged, there’s no safe way to even demonstrate VMC accurately to a trainee, because the engines produce full power only at sea level, and only a suicidal maniac (or a test pilot) will attempt a VMC demo at low altitude. If we do the demo at, say, 3,000′, we’ve lost about three inches of manifold pressure, which is a fairly hefty loss in power. While that certainly will reduce the performance of the airplane on one engine or two, it makes the airplane much easier to control, because VMC will be much lower.

How to Handle the Engine Failure

Many GA pilots are confused over just how to handle an engine failure. What is most important? What to do first? What should the step-by-step procedure be? Every book written that I’ve seen is different, every CFI has her own variation, and when the unfortunate Applicant goes up for the Multi-Engine check ride, the Inspector/Examiner is very likely to say “No, no, no, that’s all wrong, here’s what I want to see,” and the poor Applicant learns yet another way to do it during the check ride. There are variations between instructors and check pilots within the same organizations, and very large differences between different companies, even when operating the same type of equipment.

Even highly-experienced pilots will get into heated arguments over this one. My two favorite ways to start a bar fight are to ask “What makes lift?” and “What are the best memory items for an engine failure?” Then I sit back, listen quietly, and leave when it gets bloody. Of course, pilots no longer hang out in bars, so this is much less fun these days.

I like the FAA’s Practical Test Standards (PTS) booklets. They are generally clear and reasonable, and serve the very useful purpose of somewhat limiting just what an Inspector/Examiner can do to the poor applicant. The PTS booklets force everyone to “sing from the same music.” The PTS remove much of the “surprise” from the Practical Test, and that’s generally good. Still, even the PTS and other FAA publications become a bit vague when talking about “immediate action” items for engine failure, and there are even differences between the FAA publications. The PTS also forces us to train pilots in ways that may not match reality, at least when it comes to engine failures.

I am always saddened when I hear someone say “This is what you gotta do for the FAA checkride, but if you ever have the real thing…” We should be teaching and checking for the real world, and the FAA should favor that.

Fly the Airplane!

I generally go ballistic when someone utters that tired old saw. OF COURSE you fly the airplane, that’s obvious, you ALWAYS fly the airplane (Duh!). However, for the engine-out case we need to say it again, and explain a little more.

Regardless of the procedure you choose to use, I strongly believe there is one simple, all-important general rule for all pilots flying propeller-driven twins, and while I don’t claim this as an original thought, I haven’t seen it discussed before, and I’ve rarely seen it done.


“Always run your airplane as if you are ALREADY on one engine.”


If you take nothing else away from reading this column, please learn and follow that simple rule. It may save your life. A few examples:

Are you getting ready to put the gear down in the pattern? WOULD you put the gear down at that point, if you ALREADY had one feathered? If not, don’t put it down while on two.

IF you put the gear down, would you want to pull it back up, if you subsequently lost one engine? If not, you’re extending the gear too early, immensely complicating the engine-out procedure. A suggestion for Instructors. I suggest/recommend you simulate engine failure right after your trainees put the gear down. The results might be educational for both of you.

Are you about to make a power reduction after takeoff? Could you safely do that, if you ALREADY had one engine feathered? Would you? If not, then maybe you shouldn’t reduce the power just yet, because you’d just have to remember to get it back up again, if you lose one.

What configuration do you normally use for an ILS approach? Gear down, and some sort of approach flaps? Would you use that configuration with an engine out? If not, I think you’re buying yourself a lot of trouble, and perhaps a crash if you do lose one. Do you think engines don’t quit on approaches, at low power? Don’t bet on it, I’ve had two do exactly that, one just last year in the C-46.

This generally implies full power after takeoff to 500 ~ 1,500 ft., depending on aircraft performance, your comfort level, and other conditions. Oddly enough, full power may also be easier on your engine than the so-called “climb power” many use, due to the “enrichment” feature on most of these engines. Of course, some of the larger engines may require a reduction to “METO” (Maximum Except Takeoff) power before reaching such altitudes.

On approach, it usually means you’re better off with just the gear down on an ILS glideslope, and if you’re doing a non-precision or circling approach, perhaps you might consider doing it “clean,” and wait to put the gear and flaps down until you begin the final descent for landing.

Gear for GO!

On takeoff, I suggest pilots think of AND USE the gear lever on a twin as the “go, no go” indicator. If you have not pulled the gear up, and an engine quits, pull the other engine back and land. Once you consciously decide to continue if an engine quits (and it should be a conscious decision, every time), THEN pull the gear up. That way, the gear handle is always in the “right” position for an engine failure. The moment you pull the gear, think like an astronaut “We have a GO!”

Speed Is Life

How about your speeds, in general? Think about the speeds you habitually use. Are you, at any time, at a speed that might not permit completion of the maneuver safely, if an engine quit? A prime area for consideration is the downwind and base, where many will too often let the speed drop too low in anticipation of the landing, while getting final flaps down. This will work just fine if the approach is steeper than a normal glide slope, but if you allow the airplane to get down to a “normal” three-degree glide slope while at minimum speed, you may be shocked to find that if an engine quits, you will not make it to the runway on one.

Many instructors teach a normal (two-engine) climb at fairly low speeds, but I’m very uncomfortable with this. In training, I see too many people lose far too much speed when I pull an engine, and these are situations where the pilots should be “spring-loaded” to expect the failure! The average pilot on a normal (not expecting the failure) flight is guaranteed to lose 10 to 20 knots before he even enters “the denial stage,” and as much as 20 to 30 knots if she’s not careful. With singles, altitude may be important, but with twins, “SPEED is LIFE.” Use any excuse you can to go for the speed, as early as possible, as soon as terrain and conditions permit. I’m talking about a speed with a good margin above the speed you would want with an engine already shut down.

There is one more area where “Fly the Airplane!” is most appropriate on multi-engine airplanes, and that is the case of “yaw control.” It is absolutely imperative that pilots remain alert to sudden yaw, and be prepared to stop ANY yaw with RUDDER. This is one of the very few cases in all of flying where genuine quick, unthinking, reflexive action is absolutely called for. It MUST be pure reflex, without conscious thought of any kind, whether the yaw is from a crosswind gust, the copilot slipping and kicking one rudder pedal by accident, or from an actual or simulated engine failure. If you cannot prevent a yaw of more than a very few degrees, you are not doing it by instinct. If, God forbid, you push the wrong pedal, it is an absolute indication that you are trying to think through the process, rather than performing a trained reaction. The yaw from an engine failure is far too fast to permit thinking about it! Once the yaw is controlled, the rest of the procedure is a careful, deliberate process, thinking about each step.

So remember, always fly a twin as if you were already on one, and control any yaw by pure reflex, with rudder. Those are the two major points I’d like you to carry away from this column.

Memory Items

Once the yaw is under control, there are a few items that need to be taken care of. They must be done carefully, while thinking about them, but there is not a lot of time to waste for a failure right after takeoff. There is simply no time, no room for a written checklist for these first few items, they MUST be done from memory, and the drill must be practiced often, even if only by the pilot drilling himself, without taking any action. At least once on EVERY flight, the multi-engine pilot should run through the drill, touching each item in sequence, and thinking about it.

If you adopt the procedure of always flying as if you already have one shut down, then I believe the actual engine-out procedures you use, or are forced to use, become slightly less important. If a new CFI, or a new company wants you to adopt a different set of memory items (sometimes called “immediate action items”), you’ll be able to integrate almost anything into your own flying, while wondering why your old memory items were so bad, and why the new ones are so good. Cheer up, your next CFI, or your next company, will change them again. If you operate in such an environment, then the rest of this column will be somewhat less useful to you. If you have the luxury of setting up your own procedures, read on.

If, AND ONLY IF, you always the fly the airplane as if one is already lost, then the only thing that really needs to be done when the engine quits is feather it. Many airlines adopted that lovely procedure before the jets came along. The first item on the memory checklist was “Feather Failed Engine,” usually followed by “Mixture Off,” and then “Checklist.” That’s it, leaving the pilot free to fly the airplane, then when time permits, run the written one, generally if someone else is available to read it.

Now We’ve Lost One

With an engine failure, stop the yaw WITH RUDDER ALONE. If you find yourself holding significant aileron with an engine out, you need RUDDER in that same direction, not the aileron! If you have left aileron applied, feed in left rudder (or less right rudder), and allow the aileron input to return to roughly neutral. Some books will suggest carrying the dead engine five degrees high, but this is invariably too much. Five degrees bank is the limit for VMC certification, and is intended to improve control, not performance. About two degrees, held with a very small amount of aileron input is more like it, for performance, and this small amount of bank is critical in all prop twins.

DON’T fool around with the rudder trim, until everything else is all done. Use that good strong leg to hold the nose straight, and it’ll be easy to identify the engine by thinking which leg is “dead.” This is the simplest, surest way I know to identify a failed engine, but if you’ve trimmed out the yaw, you’ve just made it somewhat harder to identify the failed engine.

To Feather or Not to Feather?

The only case where it is advisable to just feather the engine without further ado is when ALL power is lost, close to the ground, on takeoff, or maybe on the landing approach. At low altitude after takeoff, you simply can’t take the time to troubleshoot when one engine is windmilling, and dragging you down. On the landing approach, it is better to just feather and go ahead and land. In fact, if the engine fails on short final, it’s probably best to forget feathering or any other part of the procedure, and just fly the airplane to the runway. At all other times, at any decent altitude, unless the failure is obviously catastrophic, you’ll probably want to take some time to troubleshoot, try a different fuel tank, play with the engine controls, and see if you can restore some or all the power.

Some will add “POWER” to the memory items, either before or after the feathering. I don’t have a major beef with this, but again, if you’re on takeoff, the most critical case, you should already have full power, and all other times, you have the time to “fly the airplane,” which includes adding power in the normal manner, for desired performance, after feathering.

Others will add “gear” or “flaps,” and here my teeth begin to grate, because if you’re flying the airplane as if the engine had already failed, your gear and flaps will be precisely where they should be, already.

Assuming we’re flying the airplane as above, my procedure on takeoff becomes:

  • Identify
  • Feather
  • Mixture

That “Identify” confuses a lot of people. If the engine is jumping off the mounts, or fire is streaming out, you have certainly identified the engine! On the other hand, if yaw is the only indication, you should take the time to think “Dead leg, dead engine,” and then VERIFY that by pulling that throttle back, to see if anything changes.

Simple, easy to remember, easy to do, and gets the main drag items DONE and the engine is put to bed, leaving the pilot free to “fly the airplane.”

Time to Think

At any other time, my immediate action is “Hmm, what’s going on here, and what should I do about it?” One old saw is “First, wind the clock,” implying that NO action is the best thing. Unless you are in imminent danger of hitting something, or busting an MEA, there is NO need to do much of anything, except fly and think.

In my opinion, this is one of the few areas in the FAA PTS that lets us down in the real world. On essentially all PTS check rides the implication is that every engine failure requires an immediate shutdown, and because training is so expensive, we tend to teach that way, too. Say to the multi-engine trainee just before his check ride “Gee, I see a little oil coming out of the right eng…” and he’ll be into the engine failure case before you can finish the sentence with “…but I think it was there before takeoff.”

IF, after consideration, the decision is made to shut it down, then “Identify, Feather, Mixture” works well, too.

In some circles, more is added to the “Immediate Action” items. One typical checklist goes like this:


Power:Maximum (implies mixture, prop, throttles-3 items)
Throttle:Idle (failed engine)
Prop Lever:Low RPM (this does not feather the prop on most warbirds)
Mixture:Idle Cutoff
Feather Button:Push


When I am training and checking in those aircraft, I am forced to follow that procedure, and I do, but frankly, I don’t think it’s the best possible procedure. Does it work? Yes. Does it comply with the FAA PTS? Absolutely, in fact, done right, it may be better at doing that than “my” procedure. Does it work in the real world? No, I don’t think it does.

We instructors and check pilots who are constantly exposed to it can (usually) do it,and get it right, most of the time. Unfortunately, the average pilot, who gets a checkride and maybe a tiny bit of training only once a year, CANNOT reliably perform this procedure on that checkride, while flying the airplane at altitude, even knowing the simulated failure is coming. This is the single most-botched procedure I see.

Having seen so much trouble with this procedure at altitude, I estimate the odds of that same pilot getting it right on a real takeoff with an actual failure at ZERO. I hope I’m wrong. I think chances would be much improved with the “simple system” I advocate here, even if that system doesn’t exactly cover every possible case in the most complete manner possible.

So, fly the airplane as if the engine has already failed, control yaw with rudder alone, and carefully do a simple engine-failure procedure, and you’ll probably do OK when that engine fails for real.

Be careful up there!

For a complete list of references (and John Deakin’s comments), read his notes for this article.