Sunday, May 29, 2011

AF 447 Likely Crashed Because of Pilot Error

The first thing I did Friday morning was to go to BEA's website and read the most recent note about AF 447. For the first time the investigators published excerpts from the voice recorder of the downed A330-200. After reading the note it was pretty clear to me that major errors by the pilots caused this tragic accident in June 2009.

As already known, the pitot probes failed during the Airbus's flight through the Intertropical Convergence Zone (ITCZ). Experts believe that super cold water hit the tubes and froze on contact. The tubes failed and the Airbus suddenly did not have any reliable speed data. Consequently, the auto pilot and auto thrust disengaged. The A330 switched into alternate mode, giving the pilots full control of the plane. According to the transcript, the two first officers in the cockpit tried to call the captain, who was resting at the time, but he did not respond. Even though at least one of the first officers was pretty experienced on the A330, they failed to correctly interpret the computer's error messages. They were too consumed by figuring out what all the error messages meant to just focus on flying the plane.

Finally, the captain was back in the cockpit, but the situation worsened. The first officer pulled the plane up and it climbed from 35,000 feet to 38,000 feet, which was too high for the current situation in the ITCZ. Climbing also slowed down the plane and signals went off indicating that the Airbus was about to stall. At an altitude of 35,000 feet, the speed of a plane has to be within a very narrow limit, otherwise it will stall if the angle of attack is not adjusted. A difference of only 20 knots can cause a plane to stall. However, the first officers did not react correctly. The plane was still too slow and turbulence was rocking the plane to the sides.

With the fly-by-wire's envelope system down the first officers pushed the sidesticks too hard and the plane began to roll. Normally, the A330's fly-by-wire system would prevent that, but since it was down, there was no automatic protection anymore. However, and contrary to earlier reports, the Airbus was completely controllable. No primary flight systems failed. The envelope system was down, but the fly-by-wire system was still intact. The pilots had 100% control over the plane.

The first officers were not able to stop the stall. With very high vertical speed, the A330 shot toward the Atlantic. Ironically, one first officer increased thrust to 100%, increasing the plane's speed even more as it was heading toward the water. Pretty soon, the first officer realized he was not able to regain control of the plane. At 10,000 feet, he handed the controls over to the second first officer in the cockpit, who only had a few months flying experience on the A330. They must have realized they were doomed. Ironically, the pitot tubes worked again after less than one minute. The last words in the cockpit were "go ahead, you have the controls," when the first officer handed the controls over to the second first officer. A few seconds later they hit the water and were dead. And with them all 225 people on board flight AF 447.

So what happened? First, the Airbus's systems did not fail. The only thing that did fail were the pitot tubes. This is not really cause for a crash if pilots are well trained in these situations. Pitot tubes have frozen over before and never did a plane crash because of it. In this case, a chain of unfortunate events caused the accident. The following is a list of all the errors and issues that led to this terrible accident:

1) Instead of focusing on flying the plane, the two first officers were too focused on the computer's error messages. Flying the plane at an angle of 5 degrees with thrust at 85% would have been enough to fly without working pitot probes.
2) The first officers made the plane climb to 38,000 feet instead of descending. This decreased speed and caused the stall.
3) They were not well trained to regain control of the stalling plane. This had not been part of their pilot training.
4) They were not trained to realize that the frozen pitot tubes caused these problems.
5) The most experienced captain was not at the controls when the problems started.
6) At 10,000 feet - the most crucial phase of the accident - the least experienced first officer was in control of the plane.

In my opinion, well-trained pilots should have been able to handle this situation. In a BBC documentary from 2010, pilots encountered the exact situation in a simulator and managed the situation very well. Of course a simulator is not a real plane and the pilots in the simulator knew what was happening, but I strongly believe that all pilots should have the training to handle such a situation and a stall.

I am convinced that the crash of AF 447 will result in lots of troubles for Air France and Thales, the manufacturer of the pitot tubes, as they will be charged with manslaughter. Air France will be blamed for not training its pilots well enough and Thales will be blamed for building faulty pitot probes, even though the company claimed that they are able to handle all real-world challenges.

For Airbus, this is a result they can live with. The A330's fly-by-wire system was working and the plane is not to blame for the crash. But that's the only good news...

8 comments:

  1. This is tragic. I feel pitot probe problem was a case study on Air Crash Investigation a while back.

    ReplyDelete
  2. This is a pretty stupid article. You are discounting the weather, defective equipment and inadequate plane design and blaming the pilots by painting a very warped picture that is psychedelic.
    Subsequent meteorological information reveal that the temperatures at that level were less than -80 Celsius (-172 F) and the storm reached a height of 18km (11 Miles).
    The pilots did and admirable with a plane that "had a stroke" as a Air France pilot described it. The problem is the airline, plane, weather - NOT the pilots. Respect their attempts to survive and save the passengers in impossible circumstances. After all, they made it possible for so much of the plane to be recovered. Blaming the pilot is the usual dirty tactic by airlines and aircraft manufacturers.

    ReplyDelete
  3. I'm not even a pilot, but I know enough from computer games to know if you have a possible stall the worst thing you can do is to pitch the plane up. Shouldn't the pilots first insinct to adjust the attitude and the thrusters to those predefined levels to avoid a stall? I mean they were over the ocean so there was nothing else they had to worry about.

    ReplyDelete
  4. There have been quite a few crashes related to Pitot tubes icing, including a resent A321, during a low level test...

    ReplyDelete
  5. this situation is really not that complicated.follow simple procedure thrust at 85% and angel at 5 degrees. the plane will continue to fly all the way to paris no matter what error messages are saying. they just had to follor procedure COMPLETELY PILOT ERROR.BUT,MAY THEY REST IN PEACE IM SURE THEY DID THE VERY BEST THEY COULD UNDER THE CIRCUMSTANCES

    ReplyDelete
  6. If all the Pitot tubes iced up evenly together and there was no immediate discrepancy between the 3 ADR's then the Flight controls may well stay in Normal Law and all protections would still work including Stall Warnings. Therefore if the speed is iced up slow you will get incorrect Stall warnings. Hence the warning in the QRH.

    If however if there is a discrepancy between the ADR's the Flight controls will revert to Alternate Law and then the stall warning will only be based on AOA directly from the vanes. Hence the warning in the QRH to respect Stall warnings while in Alternate or Direct Law.

    THE TRICK FOR THE PILOT IS TO KNOW WHICH LAW YOU ARE IN

    ReplyDelete
  7. Could Airbus program their flight computer to default to 5% and 85% thrust in this situation automatically?

    ReplyDelete
  8. Nathan -

    Yes, they could potentially do that. It's a bit tricky for the computers to realize the Pitot tubes failed. Not all inconsistencies in airspeed data are caused by Pitot tube failure. On the other hand, these numbers would change depending on the plane's altitude. But, overall, I guess it would be possible!

    ReplyDelete