Aftermath: Off to a Bad Start

Two fatal accidents show that takeoff is not always as simple as it looks.

Aftermath

Aftermath

(February 2012) Increasingly, flight is automatic. Navigation no longer requires an ounce of brains. Auto-pilots climb and descend, hold altitude, track waypoints and execute entire flight plans while idled crews ponder football scores. Autoland systems bring airplanes safely to earth, and even to the decks of aircraft carriers, in zero-zero weather. Only one phase of flight remains uninvaded by automation: the takeoff.

Too difficult? You wouldn’t think so. Certainly the basic act of taking off — throttles forward, hold centerline, rotate — could easily be automated. If there’s any complication, it’s in the element of judgment that arises — very seldom — when an engine fails, a tire blows, a deer strolls onto a darkened runway or some odd and inexplicable thing happens to befuddle and paralyze the pilot. It’s not that a computer could not deal with these eventualities as well as a human can — not always very well, to tell the truth — but rather that the possible vicissitudes of takeoff are so various, and the proper reactions to them so dependent on circumstances, that no program could foresee them all.

Here are two takeoffs that went bad. The conditions and the equipment involved were completely different; the outcomes were the same.

The density altitude was over 6,400 feet when a Beech A36 with three aboard, including an instrument student and his instructor, began its takeoff roll. There was an 11-knot quartering headwind, gusting to 24. The runway length, 5,100 feet, should have been ample: Handbook performance, which the flight school required that pilots record in writing prior to each takeoff, predicted a ground roll of 1,900 feet.

The engine sound drew the attention of several witnesses. One, his memory perhaps retrospectively influenced by what was soon to happen, described the engine as “clanking and banging ... like a thresher machine,” but several others said it was steady and smooth. One of those, a CFI and mechanic with 30 years’ experience in general aviation, first saw the airplane when it was a third of the way down the runway. He immediately felt that something was wrong. The engine sounded “like it was turning about 2,300 rpm.” He kept saying to himself, “Shut down. Shut down.” But the airplane continued, not accelerating appreciably, and rotated when three-quarters or more of the runway was behind it. The nose came up, but the airplane rolled another 200 yards on its mains before breaking ground.

It climbed only a few feet, snagged a seven-foot-high boundary fence with its left main gear, crossed a canyon and struck a wooded slope a thousand feet from the airport boundary, killing all three occupants.

Examination of the engine and what was left of the airframe — the cabin area was consumed by fire — yielded no evidence of a mechanical malfunction. The only noteworthy fact was that the light gray color of the spark plugs suggested lean operation, which became “excessively lean” in the summary version of the accident report.

The school had a second A36. The two airplanes were similar in most respects, but one — the accident airplane — had an altitude-compensating fuel pump and the other did not. In principle, the one with the altitude-compensating pump would take off at a full rich mixture setting, regardless of density altitude. The NTSB suggested that the accident airplane might have been leaned anyway, robbing it of power. But instructors who used both airplanes said that they leaned them both in the same way, setting the EGT at 40 degrees F rich of peak (“best power mixture”). This was logical, regardless of the type of fuel pump. An instructor who had flown with the instructor on the accident flight said that her technique was the same, and that she was “very thorough and diligently followed company rules regarding takeoff and performance data on the company form.” Their standard takeoff procedure, incidentally, included a 60-knot call-out and response.

In order to ascertain whether the compensating fuel pump could have played a role in the accident, an NTSB pilot flew a number of takeoffs under conditions similar to those of the accident and in a similarly equipped A36. The effects of leaning were negligible. In fact, the only way the pilot found to seriously compromise the airplane’s takeoff performance was to reduce power to 2,300 rpm and 23 in Hg. At that setting, the airplane would not accelerate to flying speed with full flaps; but with takeoff flaps it used only 500 feet of extra runway.

Notwithstanding its own tests, the NTSB decided that the probable cause was “a partial loss of engine power due to the certified flight instructor’s failure to comply with the pilot operating handbook requirements for the mixture setting during takeoff. Also causal was the instructor’s inadequate supervision of the flight, failure to monitor the airplane’s performance, and failure to initiate an aborted takeoff in a timely way.”

It’s hard to imagine that something did not feel strange to the instructor at the very least, even supposing that the 174-hour student was oblivious. If one were programming a computer to handle the takeoff, it would monitor speed and distance covered and would abort the takeoff if rotation speed were not reached in the expected distance. It would do this more precisely than a human being could, but even a mere human can make a mental note of the location of landmarks along the runway and judge whether the takeoff is progressing as expected.

Similarly perplexing was the takeoff of a Pilatus PC-12 during a January snowstorm in Colorado. The pilot and a single passenger perished when the airplane, after taking off and beginning a shallow right turn, flew inverted into the ground a mile abeam the approach end of the runway.

The reported weather conditions were 1,200 and three-quarters, with moderate snowfall; but witnesses described the snowfall as heavy. The NTSB attributed the accident to the pilot’s decision not to deice the wings before departure. Witnesses said there was an inch of wet, slushy snow on the wings as the airplane taxied out. The FBO had offered deicing, but the pilot had declined. Perhaps he simply didn’t want to take the time — he had an appointment at his destination — but it is equally likely that he guessed that, given the rate of snowfall, the wings would not remain free of snow for long anyway. He touched the snow on the wings during his preflight; he may have concluded that it was not adhering — the airplane had been hangared until a few minutes earlier, and may have felt warm — and would blow off as he accelerated.

The danger of contamination is that it lowers the stalling angle of attack of the wing. Since rotation and initial climb are times when the airplane is operating at a comparatively high angle of attack, there is a risk that the airplane will stall when the pilot believes he is executing a normal departure. Wing contamination can lead to a stall-spin soon after takeoff, and the typical radar track in that case ends rather abruptly with a curlicue.

A gradual off-course turn lasting more than a minute could be seen as more suggestive of a disoriented pilot or a malfunctioning autopilot than of a contamination-induced stall. It is not unusual for pilots of turbine aircraft to turn on the autopilot shortly after takeoff, and the accident docket (CEN09FA126) mentions that three days before the accident the pilot had reported an unspecified “autopilot problem” and that a third party had spoken of “repeated autopilot disengagements.”

On the other hand, the air temperature was 19 degrees F, and it is possible that in the 22 minutes that elapsed between the airplane’s emergence from the heated hangar and the takeoff, parts of its wing skins cooled sufficiently to cause the wet snow to adhere to them. Partial stalling on sections of the wing might cause unusual vibrations or control responses and quite a bit of extra drag, distracting the pilot to the point that he failed to keep the airplane on heading and upright.

It’s often the case that an airplane accident leaves you bewildered, even after the investigators have closed the book on it. The snow was conspicuous, but was it actually the cause of the accident? What could have kept a flight instructor, experienced in an airplane, from noticing that it was not accelerating properly, when even bystanders did? These questions don’t have easy answers. But they do tell us that the takeoff is not always as simple as it looks.

This article is based on the NTSB’s reports of the accidents and is intended to bring the issues raised to our readers’ attention. It is not intended to judge or to reach any definitive conclusions about the ability or capacity of any person, living or dead, or any aircraft or accessory.