Standard Stall Recovery - Going Against the Script?

On 1st June 2009, Air France Flight 447 was flying from Rio de Janeiro to Paris. During the flight, the Airbus 330-200 flew through some weather and ice crystals began forming in the three pitot tubes. These measure the differential between static air pressure and the dynamic, to provide airspeed readings to the cockpit. As these became blocked with ice build-up, there was a disparity between measurements from the three tubes, which disconnected the aircraft’s autopilot and autothrottle. The ice across the airframe also induced additional drag, so when the autopilot disconnected, the aircraft began to descend. Sat here in the comfort of my room, it’s easy to say ‘oh, you just needed to apply some more power’, however in the moment the crew did not acknowledge that the autothrottle was also disengaged.

The handling pilot pitched back on the controls to prevent height loss, which in turn reduced the airspeed. With erroneous readings from the blocked pitots, the only indication that the aircraft had stalled was the audible stall warner, which sounded for 54 seconds throughout the incident. The next part of the incident is the one that took my interest in this report as an aspect of human factors to explore and understand further.

There is a design within the system of the A330-200 that below a certain speed, when the angle of attack is ‘invalid’ (which I read as unusually high), the stall warner stops sounding. I can only speak for UK military and civilian flying training, having experience of both systems, however the procedures (I expect are similar across the globe). When you experience signs or symptoms of an approaching stall, you conduct the Standard Stall Recovery (SSR). While this can differ slightly for different airframes, the principles are similar, and this is something that is taught in the earliest stages of training. The first action is to lower the nose and simultaneously apply full power. Accepting that you will encounter some height loss, this action reduces the angle of attack, increases the airflow over the wings, maintains control of the aircraft and prevents going into an actual stall.

If we pause on the incident there and dive into the human information processing system, to understand how we embed processes such as the SSR into our long-term memory and strive to make them ‘automatic’.

During training, when the SSR is taught, the stages make their way into our Working Memory (sometimes called ‘short-term memory’) as we receive the information. Its capacity is finite (some quote 7 ± 2, though this is quite hotly debated – Doumont, 2002) and time limited to around 30 seconds. To that end, when learning the SSR, trainee pilots must rehearse it and practice it regularly to encode it to the Long-Term Memory (LTM). Once there, the information can be retained indefinitely, providing its retrieval in practiced frequently. We practice stalling at least once every three months, in various configurations, which rehearses this ‘script’ within our LTM. The cue for actioning the SSR script is usually light buffet of the aircraft or hearing the stall warner, at which point the actions follow. Rehearsing this routinely works to make the actions somewhat instinctive and automatic.

Back to Air France Flight 447. If the pilots have learned the SSR (or a variation of such) what happened for them to not reach for the script and switch into automatic reaction? While the pilots tried to make sense of their situation, they were most likely oversaturated. They went from a relatively benign state of arousal in the cruise, to autopilot disconnecting and the displays showing erroneous readings. This sharp shift in arousal state can degrade performance (see my article from last week on how much pressure is enough). The crew were busy trying to build their situational awareness that they hadn’t noticed the audible stall warner for 54 seconds.

Now this is the interesting point. The stall warner stopped sounding when the angle of attack was invalid, and the speed was so low. The first action of the SSR is to lower the nose, however as the handling pilot did this, the stall warner began to sound. This completely contravened their LTM’s understanding of the SSR. Usuallywhen the nose is pitched forward, the stall warner stops. Power is also applied, paying attention to not allow the nose to rise as this could re-enter the stall. If the stall warner sounds as you pitch forward, then it disrupts the SSR procedure as you are expecting the inverse to happen.

Many of the processes we operate in life are based on cues we receive from the world around us. When we’re driving, if we see the brake lights of the car ahead of us, that’s our cue to also brake. When our alarm sounds in the morning, that’s our cue to wake up. If the car in front of you has their brake lights are out, you may end up driving into the back of them. The cue wasn’t there for you to action the appropriate response. Or you were driving too fast…

Air France Flight 447 brings up several learning points, from both a procedural and technical perspective. Other A330-200s had similar problems with pitots icing and the automatics disconnecting, which cause Airbus to look into improving the electrical heating of the tubes. Other crews handled the situation differently, which speaks to the human performance aspect of this case study that led to the unfortunate crash. When something goes against the script, it requires activation of our cognition to process the discrepancy and understand why the system is doing what it is. That’s why a sound understanding of aircraft technical knowledge is crucial for aircrew, as it will aid that decision-making in abnormal handling situations.