Analyzing Pilot Performance

The twin-engine turboprop was en route from Newark, N.J., to Buffalo Niagara International Airport. At 10:17 p.m., it crashed into a house about 5 nm from the airport, killing both pilots, both flight attendants, all 45 passengers and one person in the house. The airplane was in visual meteorological conditions at the time of the accident.

The Safety Board determined that the probable cause of the accident was the captain's inappropriate response when the airplane's stall-warning stick shaker activated. There were four contributing factors: (1) the flight crew's failure to monitor airspeed and recognize instrument cues that airspeed was decaying; (2) the flight crew's failure to adhere to sterile cockpit procedures; (3) the captain's failure to effectively manage the flight; (4) Colgan Air's inadequate procedures for airspeed selection and management during approaches in icing conditions. Among the NTSB's 46 separate findings was that there wasn't enough ice accumulation to seriously affect the crew's ability to fly and control the airplane. The Safety Board said that the captain and first officer weren't adequately trained to deal with an unexpected stall, and found that the airline should have identified the captain as a candidate for remedial training based on his previous performance.

The captain, age 47, held an ATP certificate and a first-class medical. He had received his type rating in the DHC-8 three months before the accident. He had 3,379 hours, with 3,051 in turbine-powered airplanes, according to company records. The first officer, age 24, held a commercial certificate and a first-class medical. Her total time was 2,244 hours, with 774 in turbine-powered airplanes.

NTSB Chairman Deborah A.P. Hersman felt strongly that an additional factor in the probable cause should have been pilot fatigue, but investigators couldn't determine the extent to which fatigue might have contributed to performance deficiencies. The night before the accident, the captain slept in the company's crew room, and his sleep was interrupted several times. At the time of the accident, he had been awake for 15 hours. He commuted to work from Tampa, Fla. The night before the accident, the first officer commuted from her home in Seattle, Wash., to Newark. She changed planes in Memphis shortly after midnight, and arrived in Newark at 6:30 a.m. She rested in the crew room for five hours before her work schedule began. She had slept a total of about 8.5 hours: 3.5 hours in a cockpit jump seat while deadheading from Seattle, and the rest in the crew room. Some statements on the cockpit voice recorder (CVR) suggest that the first officer was experiencing symptoms associated with congestion or the onset of a head cold. Also, the CVR recorded the first officer sniffling and sneezing multiple times during the flight.

The cruise portion of flight was uneventful. The captain and first officer chatted a lot. At 9:49:18 p.m., the CVR picked up the captain making a yawning sound. At 9:53:40, the first officer referred to her stuffiness, saying that it "might be easier on my ears if we start going down sooner." Shortly thereafter, Cleveland Center cleared the flight to descend to 11,000 feet. At 10:03:38, the Cleveland controller told the flight to contact Buffalo Approach Control; the first officer called Buffalo, reporting that they were descending from 12,000 to 11,000 feet with the ATIS. The approach controller said to plan for the ILS approach to runway 23.

At 10:04:16, the captain began the approach briefing. Less than a minute later, the approach controller cleared them down to 6,000 feet. Thirty seconds later, the captain continued the approach briefing. At 10:06:37, the airplane descended through 10,000 feet. From that point on, the pilots should have observed the sterile cockpit rule.

At 10:07:14, the CVR recorded the first officer making a yawning sound. The flight was cleared down to 4,000 feet. The captain asked the first officer about her ears, and she indicated that they were stuffy and popping. They discussed ice accumulating on the windshield, and the captain commented that he was seeing ice on the leading edges. Ten seconds later, the captain and first officer began a conversation unrelated to their flying duties, including the first officer saying that other company first officers were "complaining" about not yet having been upgraded to captain.

At 10:12:18, the approach controller cleared the flight down to 2,300 feet. The captain and first officer performed flight-related tasks, but also continued a conversation that was unrelated to their flying duties. At 10:12:44, the approach controller cleared them to a heading of 330 degrees. The captain called for the descent and approach checklists, and the approach controller told them to turn left to 310 degrees. The autopilot's altitude hold mode became active one second later, as the airplane was approaching the preselected altitude of 2,300 feet. The airspeed was 180 knots at the time. The captain called for five degrees of flaps, and the CVR recorded a sound similar to flap handle movement. Afterward, the approach controller cleared the flight to turn left to 260 degrees and maintain 2,300 feet until established on the localizer for the ILS approach to runway 23. The first officer acknowledged the clearance. The captain began to slow the airplane less than three miles from the outer marker to establish the appropriate airspeed before landing. According to the flight data recorder (FDR), the engine power levers were pulled back, and both engines reached minimum thrust at 10:16:02. The approach controller then instructed the flight to contact the tower, and the first officer acknowledged. This was the last communication between the flight and ATC. The CVR recorded sounds similar to landing-gear handle deployment and landing-gear movement, and the FDR showed that the propeller condition levers had been moved forward to maximum rpm and that nose-up pitch trim had been applied by the autopilot.

The first officer said the gear was down. At that time, the airspeed was about 145 knots. The autopilot added additional nose-up pitch trim, and an "ice detected" message appeared on a cockpit display. At the same time, the captain called for 15 degrees of flaps and for the before-landing checklist. The CVR then recorded a sound similar to flap handle movement, and FDR data showed that the flaps had been selected to 10 degrees. FDR data also showed that the airspeed at the time was 135 knots. The CVR recorded the stall-warning stick shaker and a sound similar to the autopilot disconnect horn, which repeated until the end of the recording. When the autopilot disengaged, the airplane was at 131 knots. FDR data also showed that the control columns moved aft, and the engine power levers were advanced one second later.

While engine power was increasing, the plane pitched up, rolled to the left and reached 45 degrees left wing down, then rolled to the right. As the airplane rolled to the right through wings level, the stick pusher activated. The stick pusher applies a nose-down control column input to decrease the wing's angle of attack after an aerodynamic stall. At 10:16:37, the first officer told the captain that she had put the flaps up. FDR data confirmed that the flaps had begun to retract by 10:16:38; at that time, the airplane's airspeed was about 100 knots. FDR data also showed that the roll angle reached 105 degrees right wing down before the airplane began to roll back to the left, and the stick pusher activated a second time. Despite the stick pusher trying to lower the nose, the airplane's nose was only one degree down, not low enough for stall recovery.

At 10:16:42, the CVR recorded the captain making a grunting sound. FDR data showed that the roll angle had reached about 35 degrees left wing down before the airplane began to roll again to the right. The first officer asked whether she should put the landing gear up, and the captain said, "gear up," followed by an expletive. The airplane's pitch and roll angles had reached about 25 degrees nose down and 100 degrees right wing down when the airplane entered a steep descent. The stick pusher activated a third time. FDR data showed that the flaps were fully retracted at 10:16:52. About the same time, the CVR recorded the captain stating, "we're down," along with the sound of a thump.

The investigation revealed that during training, the captain was shown a video produced by NASA, which dealt with tailplane icing and showed that if an airplane's tail stalls due to ice, the recovery procedure involves pulling back on the yoke rather than pushing it forward. However, investigators concluded that it's unlikely the captain was attempting a tailplane stall recovery, and that there was no evidence the DHC-8-400 was susceptible to tailplane icing. The Safety Board found that the captain's improper control inputs for stall recovery were consistent with being startled and confused.

Peter Katz is editor and publisher of NTSB Reporter, an independent monthly update on aircraft accident investigations and other NTSB news. To subscribe, write to: NTSB Reporter, Subscription Dept., P.O. Box 831, White Plains, NY 10602-0831.