Plane Answers: "Chit-chat" did NOT doom Colgan flight 3407

"Gear down." The captain called.

The co-pilot responded by lowering the gear and pushing two knobs called condition levers forward. Just two seconds later, the approach controller told her to contact the tower. The co-pilot immediately looked down to change to the tower frequency, while acknowledging the controller. After she had spun some dials to enter the tower frequency in the VHF control panel, she looked at the gear handle to call out that it had extended completely-that it was now down and locked.

Two seconds later, the captain called for the flaps to be lowered to 15 degrees. Before even having a chance to look up and check on the flight's progress she needed to move the flap handle from 5 to 15 degrees.

In the 22 seconds that it took for the co-pilot to put the gear down, push the condition levers forward, change the frequency, verify the landing gear position and select flaps 15, the airplane had slowed from 180 knots to 133 knots and the stall warning system activated.

She was relying on the captain to fly the airplane or, in this case, monitor the autopilot, while she performed her non-flying pilot duties. Every pilot has been in this situation before, where rapid-fire actions can take the non-flying pilot's attention away. But usually being out of the loop for twenty seconds isn't enough to cause a problem. Up to this point, she had done everything right.

Now let's think about what the captain may have been dealing with:

He was in level flight at 2,300 feet with the flaps set to five degrees. He may have been tired, and so he likely felt like letting the autopilot take care of intercepting the final approach course. The autopilot was holding the altitude and heading and since the Dash 8 Q400 doesn't have any autothrottles, he was manually setting the power to the proper setting to maintain a speed of about 180 knots.

At one point, the speed picked up to 186 knots. He pulled the power back slightly to let it settle at 180 knots which took about 6 seconds.

A few seconds later he called for the gear to come down. The co-pilot brought the gear down and pushed the condition levers forward. The condition levers essentially control the pitch of the propellors. Pushing them forward drives the prop blades to a finer pitch, resulting in a higher prop RPM, but also more drag. These levers are procedurally moved forward so maximum thrust is available in the case of a missed approach. So putting the gear down and the condition levers are two actions that will result in a significant amount of drag.

But somehow, the captain was distracted. He had just pulled the power back prior to calling for the gear to come down. He didn't touch the throttles for the next 27 seconds, which means there was no way he had glanced at his airspeed for that half-minute. He could have been checking to see if there was any more ice accumulating or glancing at his approach plate.

The point is, he had become distracted and the co-pilot was out of the loop while she accomplished her required duties. The motion of the gear coming down and the condition levers coming forward meant that there was little time to react with the throttles.

This wasn't the first time a pilot failed to notice a loss of airspeed while on approach. In fact, less than two weeks later another accident occurred while flying an approach on autopilot. Turkish flight 1951 which crashed short of the runway in Amsterdam was equipped with an autothrottle system, but it had failed at 1950 feet, when it reduced the power to idle slowly without the crew noticing.

In an age when the flying public seeks comfort by thinking airplanes just land themselves, it appears that a reliance on automation may have led to two separate accidents in the month of February alone. Autopilot use is generally encouraged by many airlines as a way to reduce a pilot's workload.

But I'm certain that if the autopilot had been off in either accident, the pilots would have found it difficult to maintain altitude as the airplane slowed, which would have made it immediately obvious that more power was needed.

In both of these cases the autopilot masked this, making it easier to become distracted.

The Stall

When the stick shaker activated on Colgan flight 3407, the autopilot turned off automatically. Somehow the captain let the nose of the airplane reach nearly 30 degrees, and even though he correctly responded with full power, it wasn't going to prevent the continued loss of airspeed as long as he had the nose pointed up between 20 and 30 degrees.

The co-pilot had been thinking about ice for the last half of the trip because of the build-up she had seen earlier, and this might have been going through her mind as she heard the stick shaker activate at the exact moment she was moving the flap handle from 5 degrees to 15. She very well may have associated her flap selection with the stick shaker, and if a movement in the flight controls results in something going wrong, I could see most pilots tempted to move the flap handle back where it was before the problem began (in this case, back up).

This is exactly what she did, which made the recovery much more difficult for the captain, since an extra 20 or 25 knots would be needed to fly at the reduced flap settings. Bringing the flaps up is also a recovery technique in high-wing turboprops that encounter enough ice to stall the tail. So this may be further proof that she was convinced tail-icing was their problem.

The captain may have also thought tail icing was his problem and the reason the nose wanted to drop, completely misreading what the 'stick pusher' was trying to tell him. Reports have indicated that the captain had watched the NASA video on tailplane icing recoveries during training just a few months earlier. This is a video which will definitely leave a lasting impression on any pilot.

Considering the lack of sleep both pilots had, it's easy to come up with a scenario where a misdiagnosis of the problem–deciding between a tailplane stall or traditional stall–led to the accident.

The Aftermath

Non-essential chatter wasn't a factor in this accident since the pilots had been quiet for more than two minutes prior to the airplane slowing. The NTSB will likely look at the training these pilots had received and how fatigue may have played a role in the accident.

It's been verified that a lack of sleep can be equivalent to drinking while on the job, so the NTSB will likely factor this into their final report. And perhaps some attention will be given to the audible alerts pilots receive with specific attention given to how accurately they're interpreted and how long the reaction time is with various warnings.

The airplane manufacturers may determine that a warning prior to the stick shaker is warranted. A "caution, too slow!" warning may be all that's needed.

But first, training and procedures need to be considered to avoid this scenario. A great deal of time is spent during recurrent training on FAA mandated scenarios and emergencies that become repetitive. A program that introduces an even wider range of failures and scenarios in the simulator might be a better way to prepare pilots.

The NTSB will also likely criticize the turnover that has resulted from commuter airlines that see themselves as a stepping stone to the majors. An airline that decides $16,000 a year is an acceptable salary for a pilot might have to rethink their strategy as the flying public recognizes the need to continue to attract the best pilots possible.

This accident could become a catalyst for a number of changes that have been needed for a while. Proper crew rest, adequate training, and upgraded safety warnings could be around the corner. Let's hope so.



Do you have a question about something related to the pointy end of an airplane? Ask Kent and maybe he'll use it for next Monday's Plane Answers. Check out his other blog, Cockpit Chronicles and travel along with him at work.