On January 9, 2006, a Cirrus SR20 crashed at Lancaster, California, during a training flight. The airplane had apparently stalled during an attempt to turn back to the runway after a simulated power loss. The helicopter-rated private pilot and his instructor both died on impact.
The Cirrus had arrived half an hour earlier from Van Nuys, 30 miles south, and had made several touch-and-go landings on Runway 6. The wind was 060 at 10 knots, gusting to 17. Either the pilot or the instructor -- the National Transportation Safety Board's report is ambiguous -- then asked the tower for permission to do a low approach and simulate an engine failure during climb-out. This was to be followed by a teardrop turn and a landing on Runway 24. There was no other traffic, and the tower approved the maneuver after calling the pilots' attention to the gusty wind. The student (or the instructor -- there is actually no way to know who was handling the controls) executed the turn successfully, though with the loss of "a significant amount of altitude," according to the tower controller.
The pilot then requested permission to repeat the maneuver, which the tower granted. The execution was similar the second time, with a slight turn to the right at the departure end of the runway, followed by a "sweeping" turn to the left. The tower controller did not see the Cirrus depart from controlled flight and hit the ground.
The turnback after power loss is not a training requirement. In fact, as far as the FAA is concerned it is not even a valid maneuver. As every pilot who has been to ground school knows, the standard advice regarding total power loss during the initial climb is to crash straight ahead. Why? Because experience has shown, on countless occasions, that an airplane attempting to return to the runway after such a power loss is very likely to stall and spin, just as this Cirrus did.
The inflexible and draconian nature of this advice has given rise to a lot of debate. Mostly it is the failed turnbacks that get reported, so how do we know what the odds of successfully completing one really are? Obviously, there must be some altitude at which a turnback is possible. What is it? And what is the best way to make the turn?
Conscientious instructors take an interest in questions like this. For what it's worth, a consensus of sorts has arisen to the effect that the turn will consume about 800 feet of altitude, and that the angle of bank for the least loss of altitude (this can be demonstrated mathematically as well as practically) is 45 degrees. If there's a crosswind, you should turn into it so as not to get carried farther away from the runway. And so on.
There are plenty of unanswered questions. What about flaps? Full flaps would reduce the radius of the turn and add stall protection, but increase the glide angle. How does the strength of the wind down the runway affect the chances of success? Because different airplanes perform differently, pilots are advised to try out the turnback maneuver at altitude -- meaning, I would think, at least 3,000 feet above the ground -- to see what it feels like and how much altitude they lose. Of course, it's going to feel different at low altitude, and different wind conditions may have unforeseen effects. One requirement is to keep the nose down, and that's a good deal harder to do at 200 feet than at 2,000.
As for the Lancaster accident, the NTSB framed the probable cause this way:
"The student pilot's failure to maintain an adequate airspeed while maneuvering, and, the flight instructor's inadequate supervision of the flight. A factor in the accident was the strong tailwind encountered as the airplane turned from an upwind to a downwind during the teardrop maneuver."
The reference to the wind betrays a misunderstanding of the effects of wind in turns. A steady wind has no effect on a turn, as is evident when we maneuver at altitude and cannot notice any influence of the wind, whatever its velocity. What can affect an airplane is not the wind itself, but any rapid change in the wind. Such a change can come from gusts or from a change of altitude.