While icing was initially high on the list of potential causes of the loss of control of a Bombardier Q400 on approach to Buffalo Niagara International airport on 12 February, evidence emerging from on-board recorders and other aircraft in the vicinity at the time is hinting at a more complex chain of events.

Colgan Air flight 3407, operating a Continental Connection service, crashed into a house while on an instrument landing system approach to Runway 23 about 8km (4nm) from the airport at 22:15 that night, killing all 49 on board and one person on the ground.

Information from the cockpit voice recorder and flight data recorder as well as air traffic control tapes paint a picture of a relatively normal winter flight for the north-east USA with ice and light snow. Problems emerged as the crew began configuring the aircraft for landing at 2,300ft (700m) over the final approach fix for the Runway 23 ILS.

Soon after leaving New York Newark, the crew turned on the de-icing system, which uses pneumatic boots on the wing, fin and tailplane leading edges to clear ice. The system automatically and sequentially inflates the boots on the outer and inner wings followed by the tail section, and then repeats the cycle. The pilots also selected a switch that adds 20kt (37km/h) airspeed to stick shaker and stick pusher stall warning system trip levels, a safety factor that takes into account the fact that wing ice might increase the aircraft's stalling speed.

Colgan-Air-q400-445 
 © PA Photos

The aircraft was flying on autopilot until roughly 30s before the crash. Although the US National Transportation Safety Board has recommended that pilots hand-fly aircraft in icing conditions so they can feel it if ice accumulations change control loadings and effectiveness, this is not compulsory. The Q400 flight manual calls for manual flying only in severe icing conditions, and Flight 3407 appears to have been exposed only to light or moderate icing. Trim motor movements commanded by the autopilot did not appear to be unusual according to the flight data recorder.

Reference airspeed for the approach, says the NTSB, was 119kt and the crew selected the landing gear down at 134kt and flap to 5° passing the ILS outer marker. The NTSB has said the autopilot was engaged, but gives no detail about autothrottle. Some 34s later, following selection of the flaps to 15°, the stick shaker activated followed by the stick pusher, which pushes the control column forward to prevent airspeed from decaying toward an aerodynamic stall.

Just after stickshaker/pusher activation, the Q400 pitched 31° nose-up and the autopilot disengaged. The crew selected gear and flap up, and the aircraft entered high-rate left and right roll excursions - respectively 46° and 110° - and pitched 45° nose-down, symptoms typical of a deep aerodynamic stall. The crew selected full power, but the aircraft hit the ground at a descent rate of nearly 10,000ft/min (50.8m/s) in a 30° nose-down position at 100kt, below the stalling speed of the Q400.

The cockpit voice recorder shows the pilots talking about "significant" ice build-up on the windscreen and wing leading edges, but there is no mention of the yellow caution lights that would have lit up if the pneumatic de-icing system had failed. The engines also appear to have been operating properly.

The investigation into why the nose pitched up is likely to include questions as to whether this was the crew's reaction to the stick pusher, or if it was an issue connected to the location of the aircraft's centre of gravity if the centre of lift had shifted forward, and whether trim changes made by the autopilot might have contributed. Although there appear to have been no anti-icing/de-icing system warnings, the NTSB will be examining the potential effects of a failure or partial failure of the systems.

If Icing didn't bring flight 3407 down, what did?

If icing now looks likely not to have been the only cause of the Colgan Air Bombardier Dash 8 Q400 accident, the investigators have to consider what other factors were involved.

The possibilities include the fact that because the crew were aware they were in icing conditions they raised their workload and may have been hyper-sensitive to non-standard aircraft behaviour - like the operation of the stick shaker/pusher. In addition, all the problems happened at about the time the aircraft was intercepting the ILS localiser and, a little later, the glideslope. This high workload process can distract the crew's attention from critical performance issues like airspeed monitoring. Even small amounts of ice contamination on flying surfaces such as the wings and tailplane can alter their stalling angles of attack and also move the centre of lift, usually forward.

Source: Flight International

Topics