The takeoff seemed to go well, right up until the left engine failed. Autofeather quickly had the propeller blades of the failed engine turned to minimize drag. I raised the landing gear; I did not need to worry about flaps because I never take off with flaps in this airplane. The multi-engine pilot’s rosary—power up, clean up, identify, verify, feather—was complete.
So why wouldn’t the airplane climb?
With the hangars approaching quickly, there was no time to look around and diagnose much. I decided to fly at the “balked landing” airspeed, the climb speed for a late go-around, like when a deer crosses the runway while you are in the flare. It worked!
The simulator instructor said that he had never seen anyone successfully handle that situation before. This was not a source of pride. The situation was caused by my error: the pilot who never uses flaps for takeoff had tried to take off with full flaps.
Multi-Engine Training
Nobody needs to tell a new multi-engine pilot about the amazing performance and the advantages at night and in IFR conditions. That’s not what a new twin pilot needs to learn.
First there are the systems, which are often comparable to high-performance singles like a Cirrus or Malibu. It’s interesting when the systems are not comparable to a sophisticated single; more on that later.
Second, and more important, are engine problems. That’s where the safety issue is most important, because engine problems often happen when the airplane doesn’t have any energy to spare: near the ground at low airspeed. The pilot needs to act decisively and correctly.
It doesn’t matter how sharp you are or how recently you trained… there will always be some reaction time and some hesitation about what to do next. The best way I know to reduce this time is to make all of the decisions in advance, which is how things are done in Transport category airplanes. Decide whether to continue the takeoff when the airspeed is V1, decision speed. Lift the nose when the airspeed is VR, which is rotation speed. Climb out at V2.
The next steps depends on the airplane, which is why studying the POH/AFM is a crucial step. Regardless, the most important thing is to get the airplane climbing.
What Should Happen
A multi-engine pilot must understand how the airplane’s controls function. When the nose swings to one side, as it will when an engine fails, it’s yawing about the airplane’s vertical axis. The way we control yaw is with rudder. In fact, aileron input makes the situation worse in most cases, because of adverse yaw. The student has to be ready to use a lot of rudder. There are some subtleties to this, especially at lower power settings, but hefty rudder input is generally the correct response.
Actually feathering and securing a failed engine is an important part of learning to fly a multi-engine airplane the pilot has not flown before. This typically comes when the student has almost completed the curriculum. I have students do this at 5000 feet agl near the airport, just in case. I’d really rather do it in a simulator, but that can be hard to come by.
After a few rehearsals, I turn one engine’s fuel off. The now well-prepared student goes through the drill. Then we fly around on one engine for a bit—everyone wants pictures for social media—and practice gentle maneuvering at or above blue line speed—VYSE, best single-engine rate of climb—and then we carefully go through the restart procedure. (See the sidebar on restarting.)
Once the failed engine is feathered, the securing procedure should be slow and deliberate. In a piston airplane, for example, the magnetos should be shut down one at a time. Shut down alternators or generators with care, too. You may also want to configure the fuel selectors to crossfeed the “good” engine. Don’t forget to close the cowl flaps, if any. They add drag. But be sure to open them when restarting. There should be an engine failure/securing engine checklist. Once your memory items are complete, run the checklist to make sure you covered everything.
What Happens Too Often
Nobody is perfect, and people make mistakes in training. That includes instructors.
Once when checking out a new hire, who was already a rated multi-engine commercial pilot, I simulated an engine failure by reducing the power on one engine using the throttle. What should have been a routine exercise almost turned into a nightmare when the pilot used full aileron to try to counteract the yaw. That just made the yaw worse, increasing the risk of a VMC rollover, which is too often fatal. (A VMC rollover happens when there is not enough airflow over the rudder to overcome the yaw from the failed engine. The subscript “mc” means “minimum control.”)
The only way out of this situation was to take control of the airplane and reduce power on the “good” engine. Once the ailerons were neutral, we could add power to both engines and climb away. I mean I could climb away—that pilot wasn’t going to touch the controls again until after a lot more ground training.
While I like to be high above an airport when practicing a genuine engine failure, I had one instructor who, during recurrent training, decided to test my ability by shutting an engine down with the mixture control as soon as I raised the landing gear. In many standard operating procedures, raising the gear is the cue that the takeoff is going to continue as an airborne emergency. So there I was at 50 feet agl with a failed engine, going flying. Identify—verify—feather kept us flying. I was not happy. The extra “realism” is not worth the risk.
The worst situation I’ve seen was a high-altitude failure with a new hire who was already multi-engine rated. Instead of being slow and deliberate in securing the failed engine, he reached over and shut down the magnetos to the good engine, leaving us in a very low-performance glider. I was able to reach past him and turn the magnetos back on before the engine actually quit. He, too, saw the inside of a ground school classroom again.

Many twins, including most of those used as trainers, have some kind of feathering accumulator. This means that some kind of energy is stored—it varies from type to type—to help push the propeller blades out of the feather position into a position that will produce thrust.
But not all airplanes have this feature. While it is unlikely that someone will want to restart an engine that has actually failed in flight, it’s needed for training. This is another case where the instructor (MEI, in this case) often faces a situation which hardly ever happens in actual operations, including emergencies.
When training a new pilot, especially an airplane owner, it is especially important that the instructor understand this system’s capabilities and use before flying. It could be a chance to do a single-engine landing outside of a simulator, which I have never done. I aim to keep it that way.
Older Airplanes
I am getting ready to start flying a classic Piper Apache. I have flown Apaches before but it has been quite some time, so I am studying the Owner’s Handbook. Owner’s Handbook? This airplane was built in 1960, so there is no POH/AFM.
The Apache is not a modern aircraft. Its fuel system is complicated, and the description does not go into the detail found in a POH for a sophisticated single. That means spending a lot of time running scenarios like “right engine failed but left fuel pump inoperative.”
The landing gear and flaps are hydraulic, but the stock airplane’s single hydraulic pump is on the left engine; if that engine fails, then the pilot needs a Plan B. Similarly, the stock airplane has a single vacuum pump, on the right engine. If that engine fails, the pilot has to fly partial panel.
This scenario describes the industry’s early definition of which engine on a twin was the “critical” one. In many earlier twins, it was the one with either the hydraulic pump or the generator/alternator, depending on the power source when raising landing gear, and/or flaps. As the systems on light twins matured with time, generators/alternators were installed on both engines, and the industry largely moved away from hydraulic systems, making an engine-driven pump unnecessary. The critical engine became the one offering the worst performance when it fails.
There isn’t much performance data in the Apache’s owner’s manual: landing distance, takeoff distance, all-engines rate of climb and single-engine rate of climb. The Beech Super King Air 200’s paperwork has 12 charts. For the Apache, performance data like accelerate/stop distance (something you’ll study when you get your multi-engine rating) is only guesswork.
The Instructor’s Role
The multi-engine instructor has much more responsibility than a CFI instructing in a single. During private pilot training, there are very few situations that leave no room for recovery; for example, stall practice occurs at or above 3000 feet agl. But things can go wrong in a twin really quickly. Don’t take unnecessary risks, and be ready for low-level student mistakes.
Remember, too, that an applicant for a multi-engine rating, under stress, may fall back on bad habits learned in airplanes with relatively benign low-speed and stall behavior. They also may not fully understand the position and operation of some switches and controls. Everyone has to start somewhere.

— J.B.