Thirty Seconds of Silence

By now you've likely heard the news accounts of Colgan 3407's last minutes as captured by the Cockpit Voice Recorder (CVR). The NTSB released the CVR transcript today along with hundreds of pages of analysis, much of which is actually more informative than the CVR. Of course, the contents of the CVR transcript are easiest for the general public to grasp - and appeals most to voyeuristic impulses - so that is what was mostly reported.

Most news accounts that I saw focused on the two most sensational elements of the transcript: a statement by the First Officer regarding her inexperience in icing several minutes before the crash, and the crew's last expletives and cries in the moments before impact. Both tidbits are completely irrelevant to the investigation in and of themselves, yet do indirectly offer some clues. The First Officer's statement - that she'd never been in icing before Colgan and was glad to get a winter under her belt before upgrading - was noteworthy mainly in that it was part of a rambling, unceasing conversation that began in cruise and didn't abate until a few miles outside the marker. Put bluntly, sterile cockpit was not observed. The crew will be harshly and universally condemned for this, from the NTSB to the web boards to the lawyers. It is important to note that one cannot point to this conversation as the cause of the accident because most everything was normal while it was ongoing, and the oversight that resulted in the upset took place several minutes after the conversation had ceased. The superfluous chatter may, however, be taken as a symptom of a crew that was not 100% focused on the task at hand.

The last thirty seconds of the CVR, meanwhile, are notable for just how little gets said. Beyond the few exclamatory comments relayed by the press, there are only a few fragments and half-sentences exchanged from the time that it becomes clear that all is not well until impact: "uhhh," "I put the flaps up," "should the gear up?" and "gear up oh #!" It's an indication of how thoroughly surprised and quickly overwhelmed this crew was. There were no standard recovery procedures used, no callouts or coordination, only furious reaction. If the CVR was the only tool at the investigators' disposal, the one and only clue it would provide is a constant noise in the background through the end of the recording: the staccato chatter of a stick shaker.

The Flight Data Recorder (FDR), on the other hand, tells the grim story with an accuracy and clarity that the unfortunate crew never knew. It records in exquisite detail the status and output of every single major sensor and system on the aircraft, 121 parameters in all. The NTSB painstakingly graphed out every pertinent piece of data against various timelines, from the entire flight to the last desperate seconds. Simply thumbing quickly through the graphs gives the very distinct impression that things went south very quickly: the lines go from straight and orderly to severe and chaotic in a manner of seconds. Much closer study of the data contained within, taken alongside the CVR transcript, gives one a very good idea of just what happened on that snowy February night.

The best place to start is at 22:16:00 Eastern Standard Time, a mere 53.9 seconds before impact. Up to this point, everything is utterly normal so far as aircraft control goes. The Q400 is level at 2400 feet and 180 knots, intercepting the localizer for Runway 23 at Buffalo; the flight has been cleared for the approach. The autopilot is engaged, and the flaps are extended to five degrees (Flaps 5). The Captain is the Pilot Flying and the First Officer is the Pilot Monitoring. The aircraft has a moderate amount of ice on it, but not enough to affect performance significantly. The crew has correctly switched the "Ref Speeds" switch on the ice protection panel to "Increase," which decreases the angle of attack at which the stall protection system activates. During their preparation for the approach, the crew set a V-app speed of 118, which was correct for their weight and configuration but did not include the mandatory 20 knot icing additive. It's a small but important detail that got overlooked during their conversation.

At 22:16:00, the Captain reduced the power levers to flight idle to slow down for landing. A few seconds later he called "gear down, loc's alive" and the FO extended the landing gear and moved the condition levers to their maximum position, 1020 RPM. Both landing gear and flat-pitch propellers on the Q400 are quite draggy, and the rate of deceleration increased. In the next ten seconds, the airspeed decreased from 170 knots to 149, already a fairly slow speed for Flaps 5. At 22:16:21 the FO noted "gear's down," and the Captain immediately called "Flaps 15, before landing checklist."

Perhaps the FO noticed the low speed cue rising menacingly on the Primary Flight Display, because she paused a few seconds before moving the flap handle, only put it to 10 degrees, and began "uhhh...." One second later, the stick shaker began clattering away, almost immediately accompanied by the autopilot disconnect horn. It would continue to blare through the rest of the recording, although it could've been silenced by pushing the autopilot disconnect on the control wheel. At the moment the stick shaker activated, 22:16:27.4, the flaps were moving through 6.7 degrees and the airspeed was 126 knots.

Neither the Captain nor the FO said anything about the stick shaker. In fact, neither pilot said anything at all for a full ten seconds. The Captain did react, however, immediately and decisively. Within half a second, he hauled back on the yoke with almost 30 pounds of force. It was enough to send vertical acceleration from 1G to 1.44G. He subsequently relaxed the back force somewhat, but pitch continued to increase to a maximum of 30 degrees by 22:16:33. Meanwhile, the Captain advanced the power levers to around 70 degrees and left them there; neither he nor the FO seem to have noticed that they weren't at the rating detent, much less the mechanical limits. The Q400's engines can produce up to 130% torque in an emergency, but in this case they never went above 80%.

The Q400 is notorious for its left turning tendencies when power is advanced, and they were magnified by the aircraft's slow speed. As the aircraft pitched up, it also began banking left despite increasing right rudder. By the time the aircraft reached its apex of 2600' feet, it was in a 45 degree bank. The Captain responded with significant right control wheel deflection, which deployed the roll spoilers on the right wing.

We don't know exactly when the stick pusher fired because the FDR only samples it once every four seconds. We know it was active by 22:16:34, and that the Captain responded with renewed stick force of around 40 pounds. The airspeed was now under 100 knots; angle of attack reached 23 degrees. At 22:16:34, the aircraft simultaneously pitched down rapidly and snapped to the right, rolling straight through level and reaching 110 degrees of right bank within seconds. It was at this point that the Captain muttered "Jesus Christ," the first word he spoke since calling for Flaps 15 and the before landing checklist. At the same moment, the First Officer put the flap handle to zero without prompting, and then announced "I put the flaps up." The flaps move slowly on the Q400; they only reached Flaps 0 a few seconds before impact. Its questionable whether the situation was recoverable at this point, but retracting the flaps certainly didn't help matters.

Using full deflection left aileron and rudder, the Captain was actually able to recover from the past-vertical right bank to a shallow left bank. The pitch was bobbling near zero, and the angle of attack fluctuated between 15 and 25 degrees. The stick pusher momentarily stopped firing, and suddenly the control wheel back pressure increased to over sixty pounds. The pusher fired again, and the Captain overrode it for the last 12 seconds with 80 to 120 pounds of force. The plane snapped right past 90 degrees bank again; this time the nose dropped well below the horizon, to about -45 degrees pitch. "Should the gear up?" the FO asked; the CA responded, "gear up oh #."

The crew fought to the end, getting the bank back to around 30 degrees and the pitch to -25 degrees. The airspeed momentarily increased to 160 knots. The angle of attack, though, was still well in excess of 20 degrees and the rate of descent was around 10,000 feet per minute. The crew's last words make it clear that they knew the gig was up a few seconds before the recording ends.

It was 27 seconds from the time the stick shaker went off until impact. When the shaker went off, the airplane was still flying just fine; in fact, the angle of attack was only about 8 degrees. At this point the recovery procedure was fairly straightforward; the Captain practiced it several times quite recently in the simulator. Why he instead reacted by yanking back on the yoke is going to be the cause of a lot of speculation throughout the aviation world, but will never be truly known. Even after the pitch up and subsequent stall, recovery was not out of the question with the right stall recovery technique. Most airlines, however, do not teach stall recovery, and even if Colgan had it is impossible to say whether this crew would have recalled and executed it in time to save the airplane.

This is the easy part of an accident investigation: finding out what happened. Investigating why it happened can be much, much more difficult. This is what the NTSB is focused on now; they'll have plenty to say on the subject, and I'll have a few things to say myself. In the meantime though, this accident provides a reminder of how quickly things can turn bad in our profession. A read through Colgan 3407's CVR transcript should provide more than enough motivation to focus on the matter at hand and leave the distractions at cruise altitude.