Learning from Mistakes

The media never gives you the full story. In the tragedy of the Colgan Continental Flight 3407 the press has lost interest and moved on. But the NTSB (National Transportation Safety Board) is diligently working and has uncovered key facts about what caused the crash.

I've taken the following from a posting by James Fallows, an editor of The Atlantic, who is a pilot and has been following this story on his blog:

... according to NTSB evidence, the effect of icing was minimal. Instead, the flight crew had deliberately or inadvertently slowed the plane themselves, by pulling the throttle back to nearly the "flight idle" position -- and leaving it there. Reduced power is normal when descending or deliberately slowing for an approach, but apparently the power was left too low for too long as the plane's speed decayed to a dangerously low level.

The extraordinary NTSB animation of the flight's last 2 minutes and 39 seconds dramatizes how it happened. At time 1:40, the plane begins slowing from its cruise speed of about 185 knots. By 2:04 -- with the autopilot holding a constant altitude and the power setting still low -- it had slowed all the way down to 140 knots. That is where the power should have come back in, because the plane had reached its proper approach speed and shouldn't safely go much slower. But the crew left the power at idle, and within four seconds the plane was slowing below 130 knots - at which point the "stick shaker" gave its warning and, tragically, the pilot reacted in exactly the wrong way. The animation shows how quickly this all could happen, and what it looks like when a plane goes into aerodynamic "stall."

The effect of the pilot's wrong reaction to the stall warning has been frequently discussed in the wake of these hearings. The inattention to approach speed is in a way more puzzling, since it was not an instantaneous, instinctive thing.

Managing the power on a descent so that you slow down -- but don't get too slow -- is one of the fundamental tasks on an instrument approach. (The others are monitoring the navigation indicators, to make sure you're on course; and monitoring the altitude, to make sure you don't get closer to the ground than you're supposed to at each stage of the descent. Instrument training largely consists of an instructor constantly reminding you of whichever part of the "scan" of flight instruments you are neglecting.) That a two-person team would not notice this first-order aspect of flight safety is very hard to explain.

Talks about aviation safety often refer to an "accident chain" -- that is, a sequence of events leading to disaster, which could have been broken (avoiding a crash) at any point. Here there seems to be a two-link chain: If the crew had noticed and corrected the decaying air speed, they would never have had to worry about the "stick shaker" alert. Or, if they had responded properly to the stick-shaker, they (probably) could have recovered from the effects of letting the air speed decline.

None of this makes anything about the situation better. It may clarify some of the training changes that need to occur.