Kenya Airlines Crash Quandary

Air Safety Week,  Feb 7, 2005  

• E-mail
• Print
• Link

Following the PNF's callout of "positive rate," the PF answered: "positive rate of climb, gear up." At that moment, the warning sounded, the PNF said "uhooo," which showed his surprise, and he did not retract the gear. The CVR transcript shows that 19 seconds later the PF asked for the audio warning to be turned off and that the PNF pushed the "EMER AUDIO CANCEL" button. No other action by the PNF was determined until 21h 09min 23s when he ordered "go up," while the radio altimeter had just called out 10 feet and the "overspeed (VFE)" CRC was set off. This "go up" order by the PNF shows that he was becoming aware of the proximity of the ground, albeit only one second before the impact. The radio altimeter "300," "200," "100" callouts could have led the captain to give this order earlier. He does not appear to have heard them (as in "they did not register").

The stall warning, inhibited while the airplane is on the ground, was initiated just after the airplane lifted off. The PF reacted rapidly by pushing the control column forward, which interrupted the initial climb and put the airplane into a descent. The time the crew had to interpret and cope with the warning before hitting the sea was less than 30 seconds.

Findings

The FCOM used by the airline states that whenever a stall warning is encountered at low altitude (stick-shaker activation), it should be considered an immediate threat to a safe flight path. It specifies that at the first sign of an imminent stall or upon stick-shaker activation, the following actions must be undertaken simultaneously: thrust levers in TOGA position, reduction of pitch attitude, wings level, check speed brakes are retracted. The investigation showed that the pilot flying reduced the pitch attitude but did not apply TOGA thrust. As in the Halifax 747 crash, reduced power takeoffs introduce another risk element.

The following elements contributed to the accident:

* The PF's rote action put the airplane into a descent without the crew realizing it, despite the radio altimeter callouts;

* The GPWS warnings that could have alerted the crew to an imminent contact with the sea were masked by the higher priority stall and overspeed warnings;

* The conditions for a takeoff performed toward the sea and at night provided no external visual references that would have allowed the crew to be aware of the proximity of the sea.

In the case of a false alarm, the taught procedure will not stop the stick-shaker. This situation is not limited to the A310. For all public transport aircraft there remains a real probability that a false stall warning may generate confusion during a critical phase of flight, at a low height.

Warnings inhibited until after lift-off can be an instant-opening Pandora's Box of surprises. But muted warnings would seem to make sense, because lowering the nose for a stall warning (albeit false) would conflict with raising the nose in response to an audible GPWS alert of imminent ground contact. Despite the GPWS being muted, the AC still counted off the descent heights but this muffled cadence was probably lost in the cacophony of alert sounds, stickshaker distractions and final overspeed alarm. There are similarities here to the Bahrein GulfAir A320 crash. However, like the A300 rudder system, you couldn't build a better rat-trap than this setup. Information overload is a legacy of the information age. As early as the BEA Trident (G-ARPI) "droopretraction stall and stick-shaker" crash in Staines UK in 1972, investigators have been perplexed as to why pilot inaction has become the norm in such cases (ones that can't be explained away by disorientation). No provision had been made for an A310 stick-shaker failure on takeoff scenario -- even though the report's annex logs a history of prior incidents, including for the A310.