Twin and Turbine’s editor Dianne White posed an intriguing series of questions to me:
“When working on our multi-engine rating, we spend most our time practicing engine-out technique, which is critical. But is engine failure what kills pilots and their passengers? Or are they more the result of flying into stuff – weather and terrain? When we do recurrent training we once again go through single-engine work, but what other scenario-based training could we be doing based on what is really killing people? If you were designing a recurrent training session for multi-engine pilots, what would it include?”
Take a moment and think about your response to those questions. Then read on, and using real data I’ll answer her questions from my viewpoint as a multiengine instructor for nearly 30 years, in simulators and in the airplane itself.
The Multi-Engine Record
Most authoritative studies of general aviation accident causes rightly quote AOPA Air Safety Institute’s annual Nall Report. This report, available on the AOPA website, relates and comments upon trends in light plane crashes by careful review of the most recent NTSB Probable Cause findings. The Nall Report, however, does not break accident causation down by airplane class. It does have some statistical breakdowns of single- and multi-engine crash scenarios, but it does not provide separate studies for single-engine and multi-engine airplanes. As one researcher put it, AOPA “chose[s] to focus on accident types rather than aircraft types.”
To answer the first of our questions – is engine failure what kills pilots and their passengers? Or are more the result of flying into stuff such as weather and terrain? – we need to go deeper into the data. Further analysis of 10 years of NTSB data (2005-2014) reveals there were 365 fatal multi-engine piston and light turboprop crashes during that period (averaging 36.5 per year or one every 10 days for a decade). The top five cause scenarios are listed in rank order below.
These top five causes account for 49.5 percent of the fatal events, meaning that fatal accidents are distributed across a small number each of a wide variety of individual scenarios.
Looking at these numbers, it appears that if anything we need to be focusing more on engine-failure scenarios in recurrent multi-engine training. The trouble is, it is not safe to accurately practice the most critical engine failure in an airplane, and it is not accurate to safely practice them. The most critical event is an engine failure immediately after takeoff. Move an unseen hand to the fuel selector value and shut off the gas right after the pilot retracts the landing gear – the most accurate way to surprise the pilot with a power loss at a low airspeed/high angle of attack/close proximity to the earth scenario – and you’ve done several things that put the flight in serious risk. It may be realistic, but it’s extremely risky.
Follow accepted technique to simulate engine failure with a throttle reduction done at least 3,000 feet above ground from a speed somewhat higher than initial climb speed is far safer, but the level of pilot-perceived stress is much lower (collision with the earth does not appear imminent), the throttle reduction is a clear marker of the “failed” engine, and in non-turbocharged airplanes the reduced maximum available power on the “good” engine means the airplane is not as quick to diverge from desired yaw, roll and pitch targets as it would be in a real-world engine failure at a lower, more powerful altitude.
We have no choice but to present engine failures extremely conservatively. It’s safer, but it’s also much less realistic.
I’m a big proponent of simulation at least annually to give multi-engine pilots more accurate engine-failure scenarios. But flight training devices and true simulators have a “realism gap” as well. Simply put, the first time you have an engine failure right after takeoff is the first time you will have seen and felt what it’s like. That said, we need to continue to focus on this deadliest multi-engine scenario as often as we get to training.
The remainder of the Top Five fatal scenarios can be grouped into three areas:
- Basic aircraft control – visual (avoiding loss of control during/immediately after takeoff and stall during approach/traffic pattern).
- Basic aircraft control – instrument (avoiding loss of control/controlled flight into terrain: IFR in IMC)
- Flight planning and decision-making (avoiding attempted visual flight in IMC and fuel mismanagement)
Many readers know that I spend most of my time in the piston Beechcraft world. As part of that focus, I publish what I call the Beech Weekly Accident Update, a compendium of known accidents in Beech piston airplanes. It’s free at www.mastery-flight-training.com if you’re interested. Looking back through five years of my reports (2012-2016) we can take a look at the comparative data on two groups of airplanes that are almost identical in systemic design, numbers in the fleet and typical use, and which differ only by the number of engines: the Model 33 and 36 (“straight tail”) Bonanzas, and Beechcraft Barons. I do not intend to address the “single versus twin” argument here. What I’m looking for, instead, is whether the mishap record supports treating twin-engine recurrent pilot training substantially differently from that of single-engine pilots.
Note these are all reported mishaps, not just fatal crashes. In many cases the information is preliminary or does not meet NTSB reporting threshold and will not be investigated beyond the initial report. But given that those assumptions are the same for both single- and twin-engine airplanes, listed in order of frequency among Barons, we can see some trends.
There were several more scenarios affecting only the single-engine Beechcraft. It’s interesting to note the situations where a specific cause more frequently occurred in the twins, and in which there was more than one occurrence:
- Landing gear collapse on the runway
- Loss of directional control on the runway
- Landed long/runway overrun
- Landing gear mechanical failure
- Hard landing
- Attempted visual flight in IMC
- Loss of control/stall during go-around
- Damage due to thunderstorm/turbulence encounter
It is of course telling the types of mishaps that occurred more frequently in singles than in the twins, but that’s outside the scope of Twin and Turbine. Expectedly, the Bonanzas were involved in far more reported crashes due to engine failure than the Barons. But the otherwise virtually identical twin-engine fleet is not immune to loss of thrust during takeoff, and engine failures that begin in cruise flight only to end badly as the pilot maneuvers to land on one engine.
Multi-engine Recurrent Training
The combination of these two data-dives – NTSB reports and my more informal Bonanza/Baron comparison – suggests that for multi-engine recurrent training:
- A strong focus on engine failure procedures and techniques is indeed warranted. We’re not benefiting from the capability to fly on one engine in most cases like we should. The benign way we must present practice engine failures during takeoff and in flight (to keep from killing more pilots training than we save in real-world failures) does not accurately portray the rate of departure from controlled flight, “surprise factor” and natural fear resulting from all this happening so close to the ground. Ideally multi-engine pilots should use a combination of simulation and in-airplane training to keep their engine-out skills sharp and train to handle as many of the variable as possible.
- Basic airplane handling and maneuvering is vital to accident avoidance. We need to be fluent with the avionics and use autopilots to reduce workload. But eventually you must turn off the autopilot – or it turns itself off – and when that happens we need to be just as able to fly the airplane by hand in the conditions we’ve chosen to fly. This includes flight at low airspeed and high angle of attack, stall recognition and recovery.
- Basic attitude flying and maneuvering by reference to instruments. Hand-flying in addition to, not in place of, avionics and autopilot fluency; position and altitude awareness to avoid controlled flight into terrain.
- Takeoff and landing practice, including stabilized approaches, insistence on touching down in the runway’s landing zone, directional control during the takeoff and landing rolls, avoiding inputs during landing that may result in inadvertent gear retraction, and go-arounds from short final in the full landing configuration, are equally important to avoiding the most common accident scenarios.
- Flight planning and decision-making needs to be part of the ground instruction part of flight reviews and other recurrent training, including fuel management and weather strategies.
Very frequently pilots of multi-engine airplanes get regular Instrument Proficiency Checks (IPCs), and have their training provider endorse the IPC flight as a Flight Review as well. Although a multiengine IPC requires maneuvering and an instrument approach in simulated single-engine flight (on the Rating Task Table in the Instrument Practical Test Standards), an IPC does not include most of the tasks and skills that result in the historic causes of most multiengine crashes. It’s legal to count properly endorsed IPCs as Flight Reviews. But it does not help you retain and build upon the skills that are most likely to keep you out of the NTSB record.
Keep getting those regular IPCs. Don’t stop, the record shows. It’s working. But add an additional instructional session covering visual maneuvering, more simulated engine failures, accuracy takeoffs and landings, go-arounds, and a flight planning and decision-making quality control check at least annually. If possible, make every other one of those a trip to a multi-engine simulator. When I help pilots design a personal multi-engine training regimen (and I do), that’s the recurrent reality check I recommend.