Should You Turnback?
or
The Possible `Impossible' Turn
by
David F. Rogers, PhD
Copyright (C) 1991 by David F. Rogers. All rights reserved.

Most of us fly single engine aircraft. If the engine quits on takeoff, should you attempt to turnback to land on the runway? The turnback problem is extremely complex. Like many complex problems, there is no single right answer. Each situation must be judged individually. Thus, the answer is a qualified maybe or the classical `it depends'. It depends on the Conditions, the Aircraft, the Altitude, the Proficiency of the pilot and on Planning. CAAPP for short.

Conditions covers a multitude of considerations. If there is a good landing area directly ahead, then you should follow the classical advice and land straight ahead. If there is a good landing area within ninety degrees left or right of the runway heading and none directly ahead and you have sufficient altitude, you should probably elect to turn and land in that area. If you are operating from a large airport and there is sufficient runway ahead, you should elect to land on the runway even if it means dribbling off the end at low speed. If the takeoff runway has disappeared under the wheels but a crossing runway is available, then a turn to land on the crossing runway might be appropriate. If the terrain ahead and to each side is hostile, e.g., a built-up area, rough or rising terrain or water on a cold winters day or a black hole after a night takeoff over unfamiliar terrain, then you might want to consider turning back for a downwind landing on the departure runway. If there is a strong cross-wind, then turning into the cross-wind keeps you close to the airport. The `Ifs' can be continued forever! Certainly any action taken is influenced by the aircraft you are flying. Your response when flying a typical low powered single engine trainer with a typical glide ratio of 8--10 will be quite different than for a sailplane with a typical glide ratio of more than 30. For one thing, in the trainer you'll need more initial altitude to execute a turnback to the departure runway. The bottom line is know your aircraft. Altitude at engine failure is critical. If, in a typical GA single engine aircraft, the engine quits at 50 feet agl, then you are} going to land straight ahead. If on the other hand the engine quits at or near pattern altitude (1000 feet agl) or is producing partial power, then you should fly a shortened close-in standard pattern and land on the take-off end of the departure or crossing runway. It's when the engine quits are 300--700 feet agl that the decisions become difficult.

As in any other aspect of flying, proficiency is critical. Part of that proficiency is based on knowledge. Do you know how to execute a proper turnback maneuver? The other part is based on practice. Have you practiced steep turns lately? Have you practiced an emergency engine out maneuver lately or was the last time several months or years ago when you got your last rating or took your annual/biennial flight review? How about stalls or maximum climb angle take-offs? In flying, knowledge and practice are the keys to safety and to decision making.

The final element of CAAPP is Planning. Before you pull onto the runway and apply takeoff power you should have planned what you are going to do if the engine quits. For example, say to yourself, ``If the engine quits at 50 feet, then ...... If the engine quits at 200 feet, then ...... If the engine quits at 500 feet, then ..... If the engine quits at 1000 feet, then .....'' Similarly say to yourself ``Today there is a left crosswind, the terrain is clear in that direction so, if I turn, I'll turn to the left.'' Surveying the surrounding terrain either when landing at the airport, by using a sectional chart or just by looking from the airport surface is helpful. Look for obvious obstructions--hills, buildings, built-up areas, bridges, etc. Also study the airport diagram for crossing runways, the location of terminal areas, fuel farms, etc.

Having said all that the question remains: If you elect to turnback for a downwind landing on the departure runway, what is the optimum maneuver? This question initially occurred to me one night more than a decade ago while flying a single engine aircraft off 33L at BWI (Baltimore-Washington International). At that time the end of 33L was a black hole. In the event of an engine failure, the only reasonable landing areas were straight ahead on the remaining runway or on the runway behind after turning back. Asking other pilots what they would do in this situation yielded as many answers as pilots. They varied from the classical `go straight ahead regardless' to `do a wing-over and land on 33L'. Really!

The problem intrigued me. Being a Professor of Aerospace Engineering and having a well equipped laboratory available including a modified GAT-IVS flight simulator, I set out to at least partially answer the question both experimentally and theoretically. The research study was conducted with then Midshipmen 1/c (now an Astronaut) Brent W. Jett.*

Fundamentally the turnback maneuver is an energy management problem. When the engine quits, the aircraft has a total energy equal to the sum of its kinetic energy (due to its velocity) and its potential energy (due to altitude). This energy must be continuously expended to overcome drag and maintain flying speed. The problem is to optimally expend the available energy while executing the turnback maneuver.

Using this approach a long time colleague Professor Bernard `Bud' Carson developed a theoretical solution for a decelerating descending turn that showed that a bank angle of 45 degrees at stall velocity is the optimal turnback maneuver. Why 45 degrees at stall velocity? Simplistically, 45 degrees, because here equal amounts of the lift are being used to support the aircraft against the pull of gravity and to turn the aircraft. At stall velocity, because the lower the velocity the smaller the turn radius and consequently the less time spent in the turn. Coincidentally, the smaller turn radius keeps you closer to the field and turns you around faster. After completion of the turn the aircraft is accelerated to best glide speed by lowering the nose.

The experimental research was conducted in a fully instrumented modified GAT-IVS simulator configured to correspond to a light (1600 lb) fixed gear, fixed pitch GA trainer. The simulator has motion but does not have a visual display. The only visual clues available to the pilot were a horizontal line and the letters N,S,E, and W for North, South, East, and West painted on the room walls. Thus, the flights were `essentially' performed on `instruments'. This also accounts for the somewhat arbitrary, but practical and reasonable, criteria for a safe landing. These were defined as: maximum decent rate less than 2500 fpm, rate of decent at touchdown less than 500 fpm, wings level +-5 degrees below 100 feet altitude, turn of at least 175 degrees completed above 100 feet altitude and maximum bank angle less than 55 degrees. The tests were performed in the no wind condition.

The experiments comprised seven different flights flown by each of 20 different pilots with experience ranging from a student pilot, several CFI's to military pilots with as much as 5000 hours. Only some of the flights are of current interest. In the first flight the pilot did not know that the engine was to fail at 500 feet. No instructions were given to the pilot. Eighty-five per cent of the pilots successfully landed straight ahead. Of the three pilots that turned back, all were private pilots and only one was successful. The other two crashed as a result of a classic stall/spin. These results indicate that standard training procedures are well ingrained.

The third flight is of interest. Here the pilots knew that the engine was to fail and at what altitude. They were told to turnback 180 degrees and land the aircraft. No instructions were given on how to fly the maneuver. Less than 45 per cent were successful. Successful flights correlated directly with experience level. More than 85 per cent of the unsuccessful flights were a result of the bank angle exceeding 55 degrees and a subsequent stall/spin.

The results of the fourth flight were illuminating. Here the pilots were specifically told how to fly the maneuver, i.e., using a 45 degree banked coordinated turn just above the stall velocity as indicated by a blaring stall warning horn. Upon completion of the turn, transition to the velocity for best glide speed was to be made. Seventy-five per cent of the pilots were successful. Again success rate correlated with experience level. All of the unsuccessful flights were the result of the pilot allowing the bank angle to become too steep. Unsuccessful pilots were given two additional attempts to successfully complete the maneuver. All but two were successful. Of those two, one was a student pilot and the other a private pilot with less than 100 hours. Thus, with practice, ninety per cent of the pilots successfully completed the maneuver.

The fifth flight was designed to examine the use of rudder in the turn. The instructions given to the pilot were the same as in the fourth flight with the additional requirement to use `full rudder' in the direction of the turn. The basic idea was to use the rudder to induce a significant yawing rate about the aircraft vertical axis and thus turn the nose of the aircraft back towards the runway faster. The maneuver was extremely difficult to perform since it required the aircraft to be cross-controlled with opposite aileron to prevent the bank angle from exceeding 45 degrees. Only 45 per cent of the pilots were successful. All the failures were a result of steep bank angles and the resulting stall/spin. The maneuver also corresponds to the case where the pilot attempts to `hurry' the turn by using excessive rudder.

Aerodynamically the high yaw rate induced by the rudder reduces the effective airspeed over the inboard wing, the inboard wing stalls but the outboard wing does not. As the results show, a classic stall/spin occurs. In fact, if there were sufficient time and/or altitude for the spin to develop fully, it would be inverted because of the increased rolling moment caused by the increased effective speed and hence lift on the outboard wing.

Finally, the last flight is of interest because of its high success rate in the simulator studies. The instructions to the pilot were the same as in the fourth flight except that a thirty degree bank angle was to be used. The thought here was to reduce the possibility of a stall/spin by reducing the bank angle. Ninety five per cent of the pilots successfully completed the turn back maneuver on the first attempt. The single unsuccessful pilot was successful on his second attempt. Lest we become too enthusiastic about these results note that there is a definite training factor at work here.

Based on the simulator study results, Jett properly concluded that a thirty degree coordinated banked turn at an airspeed just above stall velocity results in the best combination of performance and safety. However, the thirty degree banked turn leaves the aircraft at a greater lateral distance from the runway. As we shall see, this can have a significant effect on the ability to complete the maneuver at the runway end.

Before discussing some actual flight tests, several points need further clarification. What does `just above stall' mean. The theoretical analysis shows that the optimum velocity for the turn is stall velocity. Actually it shows that flying at maximum lift coefficient which corresponds to the stall angle-of-attack is optimal. The stall angle-of-attack is independent of velocity, weight, density altitude, etc. So, the objective is to keep the angle-of-attack just below the stall angle-of-attack.

An angle-of-attack indicator provides a much better measure of stall angle-of-attack than an airspeed indicator. Unfortunately, most GA aircraft are not equipped with an angle-of-attack indicator. However, consider that a stall warning device is in fact a crude angle-of-attack indicator and is thus (perhaps) a better control device for the maneuver under consideration than the airspeed indicator. Since most stall warning indicators are set to just below stall angle-of-attack (5--10 mph (kts) above stall), practically this means that with the stall warning horn (light) just on but not blaring (not at maximum intensity) indicates a near optimal angle-of-attack while maintaining an adequate safety margin. If a horn is fitted, then the aural input `somewhat' reduces the requirement to scan the airspeed indicator and thus reduces pilot work load.

Unfortunately, all these devices, including an angle-of-attack indicator, are basically designed to give correct indications only in coordinated flight. None of them work well in uncoordinated flight. For example, if the stall warning device is mounted on the left wing, then in an uncoordinated turn to the right using excessive right rudder the left wing may be operating below the stall angle-of-attack while the right wing is fully stalled. The stall warning device does not indicate a stall, the pilot increases angle-of-attack and a stall/spin results. Conversely, in an uncoordinated turn to the left with excessive left rudder the stall warning device indicates the incipient left wing stall, the pilot reduces angle-of-attack (back pressure) and the stall/spin is avoided. Practical angle-of-attack indicators have similar problems. Perhaps two stall warning devices are in order, one on each wing. You might also want to consider always turning toward the wing on which the stall warning device is mounted.

Note that during the actual turn maneuver the issue is not minimum sink rate nor is it maximum glide ratio. The real issue, as shown by the theoretical analysis, is energy management. In fact, the simulator study shows that the sink rate is quite high; nearly a 1000 fpm for the successful flights.

One might wonder what the effects of wing loading, weight, and drag (cleanliness) are on the results. These parameters have no effect on the optimal conditions, i.e., 45 degree bank angle at the maximum lift coefficient (stall angle-of-attack). However, as every pilot knows, they certainly affect the velocity at which maximum lift coefficient occurs in a 45 degree bank. They also affect the altitude loss during the turn. There are no surprises here. A heavy aircraft with a high wing loading, a low L/D ratio (glide ratio), in a dirty configuration requires a higher starting altitude to successfully complete the maneuver than a light, clean aircraft with a high L/D ratio (glide ratio). Each different aircraft (and each pilot) requires a different starting altitude to successfully complete the maneuver.

The results of the simulator study were subsequently verified by flight tests. These tests were not reported in the original paper. The flight tests were conducted using a Beech Bonanza F33A, a Piper Cherokee 140 and a Cessna 172. Tests in all three aircraft were conducted at an altitude of approximately 3000 feet. These flight tests verified the results of the simulator study for both 30 degree and 45 degree bank angles using a velocity just above stall for a 180+ degree turn.

A limited series of low level flight tests, each of which included an actual take-off and landing, were conducted with the Cessna 172. Lift off occurred at approximately 2000 feet along a 3000 foot runway. Winds were calm. Weight was approximately 2150 pounds. Engine failure was simulated at 500 feet agl by reducing power to idle with full carburetor heat applied. An immediate 30 degree banked turn at approximately 10 mph above stall velocity was initiated. The required heading change to return to the departure end of the runway was approximately 210 degrees. Upon completion of the turn the velocity for maximum L/D ratio (best glide ratio) was established. No flaps were used. In each flight test the aircraft would have landed between 200 and 300 feet short of the runway end. At approximately 100 feet agl power was applied and the aircraft landed.

Recall that based on the results of the simulator study, Jett properly concluded that ``Turning with 30 degree bank, coordinated rudder, at an airspeed slightly above stall, will yield the best combination of performance and safety.'' In light of the limited series of low level flight tests described above, this conclusion should be modified to more nearly reflect the theoretical optimum requirements, i.e., 45 degree bank angle at the maximum lift coefficient (stall angle-of-attack). The fundamental reason is that a 45 degree bank angle results in a significantly smaller radius of turn which decreases both the distance from the runway end extended and the distance to the side of the runway at the end of the turn. The simulator studies indicate that there is a negligible difference in altitude lost during the turn (2 ft. on average out of 340 feet) and only a minor difference in rate-of-sink (73 fpm on average) at completion of the turn between a 30 degree and 45 degree bank angle. The simulator study also shows that a 30 degree bank angle places the aircraft an estimated 225 feet further from the runway than with a 45 degree banked turn. Notice that this is about the distance that the aircraft was short of the runway end during the low level flight tests. Also notice that the low level flight tests were conducted in zero wind conditions as were the simulator studies. This is the worst case. Provided that the aircraft is turned into any crosswind component, a headwind or crosswind will help the aircraft reach the end of the runway.

The issue of overshooting the runway when taking off into a strong headwind is frequently raised when discussing the turnback maneuver. Let's roughly look at this possibility. To do so it is necessary to make some assumptions. Here, consider a typical light single engine GA aircraft departing from a 3000 foot runway. Lift off occurs at 2000 feet and a climb at an average speed of 68 mph is established with an average rate-of-climb of 600 fpm. With no head wind component, after 50 sec the aircraft is at 500 feet agl and 5000 feet from the take off point or 4000 feet from the departure end of the runway. Now assume that, when completed, the turn leaves the aircraft at the same distance from the departure end of the runway as it was upon initiation of the turn, the altitude is 200 feet agl, the airspeed at 85 mph (best glide speed) and a sink rate of 500 fpm. That's a lot of assumptions but they are reasonable and this is a rough calculation. In the 24 seconds it takes to lose the 200 feet of altitude, the aircraft travels approximately 3000 feet and lands 1000 feet short of the runway. This rough result qualitatively corresponds to the low altitude experimental flight test results obtained with the Cessna 172 for zero wind. Notice that even a modest headwind significantly contributes to the probability of reaching the departure end of the runway.

Now how much headwind is required to blow the aircraft beyond the take-off end of the runway? Allowing 30 seconds to complete the turn, the entire flight is only 104 seconds long. In 104 seconds a headwind component of 26 mph is required to carry the aircraft beyond the take-off end of the runway. That's a lot of wind for a light aircraft. Even if this occurs, the pilot can use flaps or slip the aircraft to increase the rate of decent, which also decreases the flight time and hence the time for the wind to act on the aircraft, and thus land the aircraft on the runway. Even if a long fast landing occurs, is it not better to dribble off the end of the runway or to deliberately ground loop the aircraft at low speed then to land into unknown or possibly hostile terrain?

What about safety? Can the average pilot successfully complete this maneuver. The simulator study indicates that training makes this maneuver reasonably safe. This should not be surprising. The turnback maneuver is well known to sailplane pilots. It is the standard maneuver when the tow rope breaks. For a sailplane, a typical starting altitude is 200 feet. Sailplane pilots are required to train for this maneuver and to perform the maneuver to successfully complete the check ride for the rating. Why not power pilots? Power pilots are required to train for and to demonstrate level stalls, departure stalls, approach stalls, forced landings, etc. to qualify for the private pilot rating. There's a lot of hostile terrain at the end of the runway, e.g., at island airports, mountain valley airports, city airports and almost any airport at night. If the engine quits on climb out, frequently, the best choice of terrain for a forced landing is behind you, on the runway. Let's train to be able to use it.

* AIAA paper number 82-0406 `` The Feasibility of Turnback from a Low Altitude Engine Failure During the Takeoff Climb-out Phase'' presented at AIAA 20th Aerospace Sciences Meeting, 11--14 January 1982, Orlando, Florida.

Copyright (C) 1991 David F. Rogers. All rights reserved.

A detailed technical paper on this subject appeared in the AIAA Journal of Aircraft as Rogers, David F., The Possible `Impossible' Turn, AIAA Journal of Aircraft, Vol. 32, pp. 392-397, 1995. It is available on this website as a pdf file.

David F. Rogers, PhD, ATP
Professor of Aerospace Engineering
1969 E33A Bonanza N2255A
dfr at nar-associates com (make at = @ and add the period please)

< Back to Technical Flying Index | Back to NAR-Associates Home Page >