In this groundbreaking article, first published in the July 1965 edition of Air Facts, Richard Collins raised the question–heretical at the time–of whether twin engine airplanes really were any safer than singles. His cogent, well-researched argument started a debate that rages to this day. The safety issues remain unchanged, nearly 50 years later, although piston twins have certainly faded in popularity. –Ed.
Some practice makes trouble.
Why do pilots buy twins? A reader wrote in here a few months ago that he felt more secure flying his twin in bad weather and at night as well as when flying over forbidding terrain. Other people say they feel more secure all the time. Some undoubtedly buy twins for prestige purposes, for they certainly have that—in the eyes of laymen as well as other airmen. Some are probably sold for real nuts and bolts reasons: a fellow who had a generator failure while IFR in a single-engine might buy one as much for two generators as for two engines. Whatever an individual’s reason is for selecting a twin, the fact that the airplane should always land on an airport if properly operated is probably the strongest attraction. Security.
Security is a nebulous thing, though. In the light twin it is theoretically there in every good measure, but in tabulating a group of accident reports we found the somewhat startling fact that three types of popular light twins seem to have a higher fatal accident frequency than the single-engine fleet’s average. One is almost four times higher. Also, the twin fleet as a whole doesn’t come out as well as the single-engine fleet. Better comparisons might be drawn between the single-engine retractables, and the light twins. All but one of the retractables seems to have a better record than the twin fleet, and the retractable with the best record comes out twice as good as the twin fleet’s average. The retractable with the worst record is more than twice as good as the twin with the worst record. All this is on the basis of what percentage of the active fleet is in a fatal accident on a yearly basis.
Figures are figures, and hours flown and passenger miles and airplane miles and all that are not considered in the preceding approximate comparisons, and they would undoubtedly temper things considerably and make the twins look better. It is still interesting, though, that it comes out like it does on an individual airplane vs. air basis.
This is not an indictment of the light twin. In the group of accidents tabulated there were 81 serious light twin mishaps. The casual factors were largely the same as found in single-engine accidents but in bigger doses: non-instrument pilots in bad weather; instrument pilots descending below obstructing terrain; alcohol in the pilot instead of on the props; no fuel, or, no fuel in the tank selected; stalls at low altitude; and so on down the familiar line.
It’s not an indictment of the light twin because you can look at all the accidents and in most every case you wonder why in the name of peace the pilot did what he did. It’s Monday morning quarterbacking, but most of the causal factors are so simple.
The fact that the twin accident frequency is higher than the single-engine’s makes us wonder if that sense of security mentioned a moment ago might not at times cause a pilot to push a little harder in his twin than he would have in his old single-engine airplane. It’s more common to think of twins as “all-weather” and “night” machines, when really they are no more so than the single-engine airplanes unless the engine on the single stops—which it almost never does if properly cared for and fed.
Also the thought of twin-engine safety might well lead some pilots into an overly complacent state of mind. You do feel secure in one of them. They are stable, and most are very quiet and smooth. The drone of the engines themselves inspires confidence. All this, however, doesn’t change the rules of the game. This is probably the most important one thing a person stepping up to a twin needs to understand.
The main thing in any study of light twin accidents, though, is to ferret out those which were in some way connected to an asymmetrical thrust condition—for they are the ones peculiar to light twins (Skymaster excepted), and it is here and here alone that the pilot/airplane/air relationship is different than it is in the single engine airplanes.
In the group of 81 twin accidents, 15 were twin-connected. Eight of the 15, more than half, occurred while single-engine procedures were being taught or practiced.
There is an interesting comparison we feel compelled to draw here. We took a random sample of 81 fatal single-engine accidents, and 2 of them occurred during forced landings after an actual engine failure, one occurred during a forced landing after the aircraft had run out of gas, and none seemed to occur during practice forced landings. Maybe nobody practices them. FAA certainly has put less stress on forced landings lately, and for good reason. They awoke to the fact some years back that practice forced landings were causing many more accidents than were actual forced landings.
In the twins, the thought all this brings to mind is that we have airplanes so potentially wonderful, but also airplanes which we have obviously not learned to use properly. Something must be out of place in our training system, or in what people are required to do in order for the FAA to give them a multi-engine rating. More than half the twin-connected accidents and ten percent of all twin accidents are too much, proportionately, for the training process. It’s almost as if they award multi-engine ratings to the survivors of the curriculum.
FAA says that altitude is worth more than speed in a twin and that excessive speed doesn’t contribute as much to the chances of success in case of engine failure as does altitude. They can, for a fact, demonstrate this. Take off, climb on both engines at the best single-engine rate of climb speed, maintain this speed through an engine failure and the subsequent feathering operations, and then continue to maintain it while climbing. That is the absolutely most efficient operation (unless you have an angle-of-attack indicator and use it).
Furthermore the FAA can demonstrate how it is possible to lose an engine below the minimum single-engine control speed, and still fly away without losing control of the airplane. On a check ride they can pull a mixture on one engine while the airplane is below minimum control speed and expect the applicant to retard the throttles to reduce the asymmetrical thrust condition to the point where the airplane is controllable. The applicant is also expected to lower the nose so he won’t stall the airplane.
They can do all these things, and they do them every day. To our knowledge there has only been one fatal accident while an FAA inspector was testing an applicant on such items, so they do fairly well at administering the test. It’s beside the point, but interesting, that the government compensated the families of the applicant as well as the families of the observers who were along on this flight.
It is our feeling that FAA puts too much emphasis on doing what is right in theory. The loss of control in training/practice accident shouldn’t happen, but it does, and many of the accidents apparently occur when the instructor pilot, to make the trainee jump through the FAA hoop, presents the man with a difficult situation, and the man responds in a completely backward manner. For instance, if the instructor pulled the right mixture back at 75 mph and the trainee feathered the left prop things could get interesting.
A couple of typical “loss of control during training” accidents follow.
On one, an instructor with about 13,000 hours was checking out another instructor who had about 7,000 hours. It was indicated before the flight that single-engine procedures would be practiced. It was reportedly the policy of the check pilot to set up single-engine flight by moving the mixture for the selected engine to idle cut-off. Later, witnesses observed the aircraft at about 3,500 feet flying level with abrupt engine sounds. One thought a propeller was stopped. Then, according to the witnesses, the airplane fell into an uncontrolled tumbling descent from which it entered a normal spin which sharply transitioned to a flat spin after five or six rotations. The aircraft contacted the ground while descending in a near-vertical path relative to the surface. The wings were in a near level attitude at impact.
In another, an ATR [ATP] pilot with almost 8,000 hours was giving some dual to a low-time Commercial pilot in a light twin. The aircraft was observed at about 3,000 feet in apparently normal flight. The sound of increased power followed after which witnesses saw the aircraft enter a relatively flat spin. As the airplane descended, it nosed down more steeply and the rate of rotation increased. One prop was feathered and the gear and flaps were extended. We quote from the last sentence in the accident report: “. . . it was concluded that a practice single-engine maneuver, with the aircraft in a landing configuration, was being made in which insufficient airspeed resulted in a loss of control and inadvertent spin.”
We used those two accidents for examples because they are typical. Good experienced pilots in each case, and what would generally be considered adequate altitude for practicing — yet the aircraft spun in.
In a case of a person out practicing single-engine work in a twin alone, the following example:
The pilot was working on his multi-engine rating and was seen making several practice take-offs and landings. Then the aircraft flew over the airport at about 500 feet with the right prop feathered and the gear retracted. As it flew by, witnesses saw the right prop turn a few times. Then the gear was extended. The plane was next seen circling a nearby town at about 300 feet—the right prop still feathered but still turning a few times intermittently. Presumably the pilot was attempting to restart the engine. Next it was seen three miles from the town and at this point the airplane entered a right spin and flat spun to the ground. There was no evidence of any malfunction which would have led the pilot to shut the right engine off for reasons other than practice, and there was no reason the right engine wouldn’t start except that its fuel selector was in the “off” position.
The loss of control in that accident wasn’t caused by a suddenly induced asymmetrical thrust condition, but it is significant that when the pilot finally did lose control the airplane came down in a near level attitude with the angle of descent almost vertical.
There is obviously a high element of risk involved in exploring the more critical low-speed characteristics of a light twin under asymmetrical thrust conditions. We would just rather take somebody’s word for the very most one of them will do, and work out operating practices that will give us the most possible value from the two engines on the Twin Comanche which we fly, and which will give us maximum protection from asymmetrical thrust conditions at difficult airspeeds.
The important speeds on the Twin Comanche are 80, minimum single-engine control speed; and, 105, the best single-engine rate of climb speed.
First, a look at the minimum single-engine control speed. This was arrived at, on the Twin Comanche, with the right engine wide open, gear up, flaps at take-off position, the left engine windmilling, and not more than five degrees of bank toward the live engine.
On the one we fly we had the left throttle all the way back, the right engine wide open, and the gear and flaps up and when the airspeed showed about 85 or 90 the airplane didn’t feel well nor did it want to go very straight. So, we decided 90 would be our minimum control speed.
We aren’t saying that the airplane can’t be controlled at 80, for we are sure it has been properly demonstrated and that it can be done. These are demonstrations of the very most though, or the very least in this case. They are not unlike the barrier take-off demonstrations. They say, for instance, that a new Cessna 180 will clear a 50 foot barrier in 1205 feet at gross weight in standard air at sea level. It will, for a fact, do this—but, if the 50 foot barrier were a brick wall instead of a piece of string we don’t imagine you would find many volunteers to demonstrate the airplane’s true ability.
With 90 as our minimum single-engine control speed, we will turn everything off and land and try again another day if an engine sputters below that speed.
If we did have to abort a take-off because of a sudden loss of power on one side below 90 we would pull the throttles and the prop controls all the way back. There are two reasons for the props. First, the decelerative effect of the props in flat pitch is considerable when the throttles are retarded. If the airplane was off the ground with not much airspeed and a fairly high angle of attack this sudden decelerative effect would probably lead to a stall before too much could be done. They feather fast on a Twin Comanche so if there was what seemed to be more veer than rudder would take care of, it should go away before things get out of hand if everything is completely neutralized.
Second, if a decision is made to abort because of a power loss or failure it should be an irrevocable decision and feathering the propellers would keep us from changing our mind and trying to make it anyway.
On or Off?
One other item on the minimum control speed. It’s often said that this speed should be attained with the airplane still on the ground. Maybe so on some airplanes, but others get awfully light on the back wheels before the speed is reached. On these it seems you would be just as well off in the air as on the ground—maybe better off, because an asymmetrical thrust condition on the ground with not much weight on the back wheels might be harder to handle than it would be at 25 feet.
Back to the take-off. If there is still plenty of runway ahead after reaching 90 we think it is best to leave the wheels down until the best single-engine climb speed is reached, 105 on the Twin Comanche. If anything happened during that period then it would still be possible to shut it down and land without the embarrassment of remembering the wheels had been put up but not put back down prior to a hastily arranged landing.
At 105, everything still going well, we put the wheels up and accelerate on to 120 to have a little extra, climbing some as we accelerate. When we get to 120 we keep that and climb with full power until an altitude is reached which we could remain at and handily fly around and land in case one engine stopped. Then cruising climb can be established and procedures forgotten.
That is not the most efficient way to fly the airplane, but we think it is the safest way for us to fly it. To fly a light twin with complete efficiency you have to base things on what we feel are minimum items, and you have to stick your neck out and practice things which we don’t think are safe to practice.
In a twin, once you have decided that you would go instead of quit in case of trouble, the two main things in case of engine failure become not to hit anything and not to lose control of the airplane. The not hitting anything is served best by climbing, and the not losing control is best served by going fast enough to maintain complete control even if you are a little slow on the draw and your reaction times are a little below par. It seems that too much stress is being put on not hitting things and not quite enough on maintaining control, for in every one of the 15 twin-connected accidents studied the pilot lost control of the aircraft due to inadequate airspeed. In not one did he hit something because he was going too fast.
We took the Twin Comanche out and feathered the left prop and opened the right engine up. It would indicate almost 150 miles per hour in level flight at 3,000 feet. It flies good like that, too, and feels like a lively and responsive bird. At 105, the best single-engine rate of climb speed, it would climb about a couple of hundred feet a minute, and it flew like a dog compared to the way it flew at 150. That’s true of any light twin. The logic of the matter is that nobody loses control of an airplane that will go 150 in level flight and can be controlled down to 80 unless they aggravate it unnecessarily, or unless they put too much stock in trying to make it go up and not enough in making it go.
If anything does happen to one of the engines in a twin, there’s no need to be overly disturbed—most of us fly on one engine all the time. The record being what it is, the main thing seems to be not to try too hard to make the airplane climb, and not to get in any hurry to get around the field and back down if it happens during take-off. After all, if your trouble starts at 200 feet, and if you jump through the hoop just like FAA says, and your bird climbs 200 fpm on one, you are going to be a couple of miles away before you get to 400 feet. We’d swap a hundred of that, or even all of it, for extra speed if we felt like we needed or wanted it—learning from the experience of others as set out in accident reports.
We are not against practicing single-engine procedures in twins. In fact we are very much for it, and we practice a lot, but we do leave a big margin between the speed at which we fly and the minimums. We also like to practice single-engine landings for in the event an engine fails this is something you have to do every time—and you wouldn’t want to do it the first time with your family or boss aboard. We practice them carefully, though — starting at about 3,000 feet by feathering a prop, and then planning the approach and landing carefully on the longest piece of asphalt in the area—so there won’t be any go-arounds. As a matter of fact we’d never go around on one engine in a twin. It would be better to land on the last few hundred feet and slide gracefully off the end than to be exposed to the far more hazardous risks involved in a go-around from an engine out, flaps and gear down, approach.
Think back to those comparative figures: Twins, 15 of 81 peculiar to the twin with 8 of these in practice. Singles, 3 of 81 peculiar to the single in the sense that the only engine quit running for one reason or another, with no accidents in practice forced landings. We think it is a very pertinent comparison and it guides us when we practice in the twin we fly.
Everybody talks about engine failures on take-off in twins, and there have probably been a million words written on the subject. FAA’s attention to this matter—their requirement that pilots be exposed to abrupt power failures at low airspeeds — seems to account for 10% of the fatal accidents in twins. Yet, with all this background, the percentage of time where one of these airplanes is vulnerable is minuscule.
Still there is a hazard. In the 81 twin accidents and 15 twin connected accidents there were two cases of an actual engine failure on take-off getting the best of the pilot. The 81 accidents were almost a two year supply, so you might say that about once a year something malfunctioning inside an engine during this critical period causes an accident. In both of the cases related to an actual failure the pilot flew too slowly and lost control. Both took off from big airports and both aircraft were well below gross weight. In either case the theoretical possibilities of a successful flight were as good as you could ask for but they didn’t make it. It’s easy to say, but we think we could have made it using our procedures in either case, and we think you could have too.
In light twins, then, there are really only two burrs under the saddle.
The fact that pilots push a little harder and feel more complacent in twins when they really shouldn’t is one thing. That is something for people who fly twins to ponder and make peace with themselves about.
Then our keeper, the FAA, needs to look in the mirror the next time it points a finger in the name of safety, and quit making people practice protecting themselves against something which isn’t as great a hazard as the practice itself.