Page 1 of 2 12 LastLast
Results 1 to 30 of 34

Thread: Air France 447 - pilots and computers in a storm

  1. #1
    Join Date
    Mar 2007
    Location
    Falls Church, VA (near Washington, DC)
    Posts
    8,654

    Air France 447 - pilots and computers in a storm

    I found the final report on the Air France Flight 447 disaster, in which the plane went in the drink between Brazil and Africa en route to Paris.

    https://reports.aviation-safety.net/...332_F-GZCP.pdf

    A comparison with the 737 MAX cases is interesting. Unlike with the MCAS fault, the A330 computer appeared to do everything right in terms of flying the aircraft. When the airspeed indication went flaky as a result of the icing of the Pitot probes, it disengaged the autopilot and autothrottle functions, leaving it to the pilots to manually control the flight path and speed until the heaters could clear the probes. As a matter of physics and basic airmanship this makes perfect sense, because if the thrust and attitude are held constant the air speed is not going to change, at least not in a few minutes. As I think I understood it from the report, there are procedures in place for safely flying and landing without airspeed indication. What this case appears to have in common with the MAX cases is crew confusion in response to being startled by multiple fault messages, including the stick shaker which was going off even though the plane was not in immediate danger of a stall. In AF 447 the situation appears to have been aggravated by poor coordination of the actions and responsibilities of the pilots. The captain was out of the cockpit on a scheduled rest break for this long flight, and a less experienced relief pilot was in the left seat. A very junior copilot was flying the plane in the right seat, and he had very little experience or training in manually flying in rough weather at cruise altitude. According to the report he did a poor job of damping the roll that was being caused by gusty crosswinds, jerking the ailerons around instead of having a soft touch, and in the midst of it he somehow thought he needed to keep pulling back on the stick. He apparently did not know what to do in the seldom-used alternative mode in which the computer regulation that would have prevented the jerky action was turned off. The confusion was such that the relief pilot was clueless about what his partner was doing with the right stick, and the design of the Airbus fly-by-wire system is such that the left stick did not move in unison with it. They were near the "coffin corner", which is normal in cruising at that altitude, and in their confusion they did not know whether they were going too slow, too fast or somewhere in between. The plane stalled and fell out of the sky.

    In most respects it is a good thing to cruise almost exclusively on autopilot, because the present-day systems can fly the plane better than any pilot can do manually, especially if they are getting knocked around by turbulence. Perhaps the flip side is that when something goes wrong the pilots could be rusty on manual flying in turbulence when they need the skill and experience the most. I would say it is a must to have them drill on such conditions in the simulators on a regular basis. I got the impression from the report that these pilots had not had much time in such drill, and that the investigators were recommending changes.

    I was reading the report while doing my warmup routine and technical drill on my horn, something that takes about 1% of my musical thought process. I may well have missed some key technical details. As usual, don't take my word as gospel.

  2. #2
    Join Date
    Apr 2011
    Location
    Norfolk UK and some of me is in Northern France
    Posts
    8,547
    thank you for finding that, it is both interesting and alarming. I add the note that it was night flying and pilots are drilled to believe their instruments rather than their gut. The report includes that pilots in general fear high speed stall (from wings going supersonic) even though this aircraft had good characteristics. The alarming part is that even with the captain returning to the cabin and knowing the aircraft was not super stalled, i.e. recoverable, they descended all the way. The confusing signals that arrived when the pitot tube recovered and was lost again seems to have been enough to confuse all three pilots. I guess in the dark and descending through cloud they really did not know their orientation with instruments cutting in and out. So in my view, not just pilot error here, the loss of speed indication got amplified by confusing signals. Scary.
    sicut vis videre esto
    When we realize that patterns don't exist in the universe, they are a template that we hold to the universe to make sense of it, it all makes a lot more sense.
    Originally Posted by Ken G

  3. #3
    Join Date
    Apr 2011
    Location
    Norfolk UK and some of me is in Northern France
    Posts
    8,547
    more: having ploughed through there is a near the end paragraph 4.2.2 noting the there was no AoA visual indicator for the pilots who did not realise they we stalled despite the stall warning and never therefore tried the correct manual recovery.
    The icing of all three heated pitots was in that case unexpected and the presence of high altitude ice crystals was a new discovery that we now know about. The confusing messages directly resulting added to the basically panicked pilot who seemed to think he was overspeed and the whole area of training for manual control at high speed was poor. Training concentrated on low altitude low speed manual control. The plane crashed with average 40 degree AoA in zero visibility. The instrumentation let them down, they were used to ignoring the stall warning! Gosh. No AoA indicator.
    sicut vis videre esto
    When we realize that patterns don't exist in the universe, they are a template that we hold to the universe to make sense of it, it all makes a lot more sense.
    Originally Posted by Ken G

  4. #4
    Join Date
    Oct 2005
    Posts
    26,636
    It seems to me a recurring theme is "make sure the information is good regardless of what system is flying the plane." Instead, the attitude seems to be, use computers to fly the plane based on the information they get, and if something goes wrong with the information, dump the whole thing in the pilots lap to figure out-- but without good information. It's as though they think it is only the software that needs good data-- pilots are "flying by the seat of their pants", and if they screw up, it's "pilot error." Just a suggestion: increase the ways you get information to whatever system is flying the plane, with built-in redundancies and cross-checks, so no single type of instrument failure, be it an active system or a passive information diagnostic, is ever difficult enough to overcome that the plane can crash as a result. Now how hard is that?

  5. #5
    Join Date
    Apr 2011
    Location
    Norfolk UK and some of me is in Northern France
    Posts
    8,547
    I am tempted to think that it should be possible to design a pitot or indeed other airspeed indicator and an AoA sensor that does not have moving parts or holes that can fill with insects or ice crystals. The obstacle to my doing that is the huge and justifiable hill to climb to get acceptance. But reading that report about AF447 and the available on the MAX crashes leaves me itching to do something better. But in the meantime the ergonomic interface needs an external review in my opinion, specifically to cope with these rare events. Planes have to fly blind so the instruments must be triplicated and multisource.
    sicut vis videre esto
    When we realize that patterns don't exist in the universe, they are a template that we hold to the universe to make sense of it, it all makes a lot more sense.
    Originally Posted by Ken G

  6. #6
    Join Date
    Mar 2007
    Location
    Falls Church, VA (near Washington, DC)
    Posts
    8,654
    My educated guess is that anything we can think of for more reliable airspeed probes or AoA sensors have already been tried by aeronautical engineers. If those devices are the only things that fail, there are still means of getting good indirect estimates of airspeed and AoA from the attitude, engine thrust settings, altimeter, rate of climb/descent indicator, etc. The investigators addressed findings of ergonomic deficiencies in the multiple fault warnings that were bombarding the crew when they needed to focus on immediately maintaining the attitude (level flight) and thrust that prevailed going into the onset of trouble. Had the copilot kept the thrust constant, held the pitch steady and done a smooth job with the ailerons to stop the roll, as an exercise in physics I think it would have taken the mother of all wind shear to get them into any immediate airspeed trouble. Of course no one can fly by the seat of their pants without seeing the horizon, but unless I missed something they had the artificial horizon.

    Don't get me wrong. I am not implying that I could have done it right with no more training than buzzing around in a light plane in daylight and clear weather. But I can put two and two together in hindsight and form sound opinions about what could have been done better in training and in composition and presentation of the fault warnings.

  7. #7
    Join Date
    Feb 2005
    Posts
    11,505
    Storms around the ITCZ can reach a bit higher

  8. #8
    Join Date
    Apr 2011
    Location
    Norfolk UK and some of me is in Northern France
    Posts
    8,547
    Quote Originally Posted by publiusr View Post
    Storms around the ITCZ can reach a bit higher
    I believe the hypothesis of ice crystals so high up was investigated and found to be the case, and that no doubt will exercise the designers of pitot tube heaters. Insects or plain dust are another worry, and the AoA sensors were probably victims of bird strike. I am not so sanquine as Hornblower that everything possible has been tried. I have been involved in different sensors and the adverse inertia that any proposed design faces is daunting. We do have technologies now such as pressure sensors embedded in elastomers that can resist a lot of those environmental factors and they could be configured to measure AoA (using shear) as well as dynamic and static pressures, but the path to acceptance is another thing. It is interesting that these kinds of sensors mimic the way we sense pressure and flow through our skin and are also used in robotics.
    sicut vis videre esto
    When we realize that patterns don't exist in the universe, they are a template that we hold to the universe to make sense of it, it all makes a lot more sense.
    Originally Posted by Ken G

  9. #9
    Join Date
    Mar 2007
    Location
    Falls Church, VA (near Washington, DC)
    Posts
    8,654
    Quote Originally Posted by profloater View Post
    I believe the hypothesis of ice crystals so high up was investigated and found to be the case, and that no doubt will exercise the designers of pitot tube heaters. Insects or plain dust are another worry, and the AoA sensors were probably victims of bird strike. I am not so sanquine as Hornblower that everything possible has been tried. I have been involved in different sensors and the adverse inertia that any proposed design faces is daunting. We do have technologies now such as pressure sensors embedded in elastomers that can resist a lot of those environmental factors and they could be configured to measure AoA (using shear) as well as dynamic and static pressures, but the path to acceptance is another thing. It is interesting that these kinds of sensors mimic the way we sense pressure and flow through our skin and are also used in robotics.
    I stand corrected, and your input is most welcome. This could be analogous to the 1950s, when most of our airlines were skeptical about jets and felt comfortable operating and maintaining piston engines. Juan Trippe at Pan Am was a fabulous exception.

    If I am not mistaken, those ITCZ thunderheads can reach 50,000 feet, far higher than AF 447 was cruising. They tried to go around one and did not succeed.

  10. #10
    Join Date
    Jun 2005
    Posts
    13,619
    One thing that occurs to me is that you might be able to use GPS type information to see how fast a plane is going and whether it is climbing or diving. In an emergency I wonder if pilots could use such data.


    Sent from my iPhone using Tapatalk
    As above, so below

  11. #11
    Join Date
    Aug 2005
    Location
    NEOTP Atlanta, GA
    Posts
    2,776
    GPS is widely used in aviation but I think the lag time is too slow for critical decision-making. As a backup, perhaps.

  12. #12
    Join Date
    Jun 2005
    Posts
    13,619
    Quote Originally Posted by schlaugh View Post
    GPS is widely used in aviation but I think the lag time is too slow for critical decision-making. As a backup, perhaps.
    Yeah, and I thought of another flaw too, which is that it would be hard for GPS to understand wind speed, so it might only grasp the speed relative to terrain and not relative to the air.
    As above, so below

  13. #13
    Join Date
    Feb 2005
    Posts
    11,505

  14. #14
    Join Date
    Apr 2011
    Location
    Norfolk UK and some of me is in Northern France
    Posts
    8,547
    Quote Originally Posted by Hornblower View Post
    I stand corrected, and your input is most welcome. This could be analogous to the 1950s, when most of our airlines were skeptical about jets and felt comfortable operating and maintaining piston engines. Juan Trippe at Pan Am was a fabulous exception.

    If I am not mistaken, those ITCZ thunderheads can reach 50,000 feet, far higher than AF 447 was cruising. They tried to go around one and did not succeed.
    The cold is extreme; I worked on railway pressure sensors exposed to the outside and embedded units have proved very tough for many years. I would imagine the elastomers would have to be laced with carbon or metal dust to conduct so they could be heated and maybe skinned with hard particles like tungsten carbide for long life but it would be a worthwhile challenge At first they would be used alongside conventional pitot for field trials, there is no alternative as wind tunnels cannot simulate all those rare hazards.
    sicut vis videre esto
    When we realize that patterns don't exist in the universe, they are a template that we hold to the universe to make sense of it, it all makes a lot more sense.
    Originally Posted by Ken G

  15. #15
    Join Date
    Jun 2015
    Location
    Houston
    Posts
    1,151
    Quote Originally Posted by profloater View Post
    The cold is extreme; I worked on railway pressure sensors exposed to the outside and embedded units have proved very tough for many years. I would imagine the elastomers would have to be laced with carbon or metal dust to conduct so they could be heated and maybe skinned with hard particles like tungsten carbide for long life but it would be a worthwhile challenge At first they would be used alongside conventional pitot for field trials, there is no alternative as wind tunnels cannot simulate all those rare hazards.
    That being said, and I haven't read the report and points noticed, It seems like the Captains whatever the experience level should look at the radar, and I have been in helos. with what was current radar equipment and I could see the intensity and size of the storms, should make a better attempt at evading the storm.

  16. #16
    Join Date
    Aug 2005
    Location
    NEOTP Atlanta, GA
    Posts
    2,776
    IIRC the storms that gave the most trouble were being “masked” in the radar by storms closer to the aircraft. The radar saw the closer storms but not the further ones until they flew through.

  17. #17
    Join Date
    Jun 2015
    Location
    Houston
    Posts
    1,151
    Quote Originally Posted by schlaugh View Post
    IIRC the storms that gave the most trouble were being “masked” in the radar by storms closer to the aircraft. The radar saw the closer storms but not the further ones until they flew through.
    I have seen those also, but if one is prudent to fly around the masking ones, the bigger ones will be shown. For helos. the pilots always dodged the red storms, so there was never a masking effect. I 'm not a airline pilot and I don't know the training and directives they receive, but it seems like it might be prudent for the passengers to dodge all the red storms should be able to keep all from colliding.

  18. #18
    Join Date
    Mar 2007
    Location
    Falls Church, VA (near Washington, DC)
    Posts
    8,654
    Quote Originally Posted by profloater View Post
    I believe the hypothesis of ice crystals so high up was investigated and found to be the case, and that no doubt will exercise the designers of pitot tube heaters. Insects or plain dust are another worry, and the AoA sensors were probably victims of bird strike. I am not so sanquine as Hornblower that everything possible has been tried. I have been involved in different sensors and the adverse inertia that any proposed design faces is daunting. We do have technologies now such as pressure sensors embedded in elastomers that can resist a lot of those environmental factors and they could be configured to measure AoA (using shear) as well as dynamic and static pressures, but the path to acceptance is another thing. It is interesting that these kinds of sensors mimic the way we sense pressure and flow through our skin and are also used in robotics.
    My bold. According to p. 44 of the report the AoA value is not directly displayed to the pilots in the A330. It is used by the computer to generate a stall warning. I have reached p. 92 in re-reading the report and so far I see no indication of AoA indication failure.

  19. #19
    Join Date
    Mar 2007
    Location
    Falls Church, VA (near Washington, DC)
    Posts
    8,654
    I ran this one by my engineer friend, who told me today that he had worked on a flight control software package for NASA. His opinion was that test pilots often are less easily startled or confused than relatively inexperienced commercial pilots. He said that if cost was no object, he would wring out a new system with test pilots and then beta test it with commercial newbies, with a test pilot in the cockpit to intervene if necessary.

  20. #20
    Join Date
    Mar 2007
    Location
    Falls Church, VA (near Washington, DC)
    Posts
    8,654
    Quote Originally Posted by Hornblower View Post
    I ran this one by my engineer friend, who told me today that he had worked on a flight control software package for NASA. His opinion was that test pilots often are less easily startled or confused than relatively inexperienced commercial pilots. He said that if cost was no object, he would wring out a new system with test pilots and then beta test it with commercial newbies, with a test pilot in the cockpit to intervene if necessary.
    Addendum: Please don't take this as meaning my friend and I think the junior commercial pilots are bad pilots. The point is that perhaps the actions of the test pilots are not a stringent enough test of the system.

  21. #21
    Join Date
    May 2007
    Location
    Earth
    Posts
    10,252
    Quote Originally Posted by Jens View Post
    One thing that occurs to me is that you might be able to use GPS type information to see how fast a plane is going and whether it is climbing or diving. In an emergency I wonder if pilots could use such data.


    Sent from my iPhone using Tapatalk
    One thing to remember is that GPS will give ground speed; pilots need air speed as aircraft move relative to the air, which dictates their behavior.
    Information about American English usage here and here. Floating point issues? Please read this before posting.

    How do things fly? This explains it all.

    Actually they can't: "Heavier-than-air flying machines are impossible." - Lord Kelvin, president, Royal Society, 1895.



  22. #22
    Join Date
    Mar 2007
    Location
    Falls Church, VA (near Washington, DC)
    Posts
    8,654
    In smooth air there will be no immediate danger as long as the pitch attitude and thrust are held constant. The air speed will not change appreciably in the few minutes needed to resolve the icing problem. In a thought exercise we could envision severe wind shear that could suddenly push the air speed outside the envelope. My educated guess is that such situations have been evaluated in hurricane hunter missions. I don't know how strong the gusts were in the AF 447 case. They were enough to cause roll, but beyond that I have no useful knowledge.

  23. #23
    Join Date
    Jun 2005
    Posts
    13,619
    Quote Originally Posted by swampyankee View Post
    One thing to remember is that GPS will give ground speed; pilots need air speed as aircraft move relative to the air, which dictates their behavior.
    I’m glad I made post #12.


    Sent from my iPhone using Tapatalk
    As above, so below

  24. #24
    Join Date
    Apr 2011
    Location
    Norfolk UK and some of me is in Northern France
    Posts
    8,547
    But in zero visibility there is only the gyro generated artificial horizon. Pilots cannot detect roll otherwise and can be flying upside down within a minute. I guess that is why the pilot was so anxious to correct roll, but for me his instinct to pull back on the stick when he was at high speed and altitude seems to be a panicked reaction.
    sicut vis videre esto
    When we realize that patterns don't exist in the universe, they are a template that we hold to the universe to make sense of it, it all makes a lot more sense.
    Originally Posted by Ken G

  25. #25
    Join Date
    May 2007
    Location
    Earth
    Posts
    10,252
    Getting a good static pressure or temperature measurement is not easy, especially in the complex flow field around an aircraft. One could consider flush-mount transducers, but they're also subject to ice and are delicate.

    Airspeed requires both total and static pressure. Aircraft generally don't use combined total/static pressure probes, possibly because static pressure is needed for more than airspeed measurement.
    Information about American English usage here and here. Floating point issues? Please read this before posting.

    How do things fly? This explains it all.

    Actually they can't: "Heavier-than-air flying machines are impossible." - Lord Kelvin, president, Royal Society, 1895.



  26. #26
    Join Date
    Dec 2011
    Location
    Very near, yet so far away
    Posts
    233
    Quote Originally Posted by profloater View Post
    But in zero visibility there is only the gyro generated artificial horizon. Pilots cannot detect roll otherwise and can be flying upside down within a minute. I guess that is why the pilot was so anxious to correct roll, but for me his instinct to pull back on the stick when he was at high speed and altitude seems to be a panicked reaction.
    You have to pull back on the stick when banking or coming out of a roll. You lose lift/airspeed during these manoeuvres hence the pull back.

  27. #27
    Join Date
    Jun 2005
    Posts
    13,619
    Quote Originally Posted by headrush View Post
    You have to pull back on the stick when banking or coming out of a roll. You lose lift/airspeed during these manoeuvres hence the pull back.
    I might be misunderstanding, but pulling back on the stick raises the pitch and thus lowers airspeed, right? So why would you want to reduce airspeed if you’ve already lost some?


    Sent from my iPhone using Tapatalk
    Last edited by Jens; 2019-Jul-31 at 12:07 AM.
    As above, so below

  28. #28
    Join Date
    Mar 2007
    Location
    Falls Church, VA (near Washington, DC)
    Posts
    8,654
    Quote Originally Posted by Jens View Post
    I might be misunderstanding, but pulling back on the stick raises the pitch and thus lowers airspeed, right? So why would you want to reduce airspeed if you’ve already lost some?


    Sent from my iPhone using Tapatalk
    It appears that in his confusion the copilot flying the plane at that point thought he was overspeeding. According to the report the buffeting is distressingly similar in overspeeding and in the onset of a classic stall when the airspeed is too low.

  29. #29
    Join Date
    Jun 2005
    Posts
    13,619
    Quote Originally Posted by Hornblower View Post
    It appears that in his confusion the copilot flying the plane at that point thought he was overspeeding. According to the report the buffeting is distressingly similar in overspeeding and in the onset of a classic stall when the airspeed is too low.
    I was asking about the comment in post 26 that:

    You have to pull back on the stick when banking or coming out of a roll. You lose lift/airspeed during these manoeuvres hence the pull back.
    Is that true?
    As above, so below

  30. #30
    Join Date
    Dec 2011
    Location
    Very near, yet so far away
    Posts
    233
    Quote Originally Posted by Jens View Post
    I was asking about the comment in post 26 that:



    Is that true?
    I may be misremembering, it may just be the lift. This is from my experience in simulators a fair while ago. What you say about airspeed sounds logical. I might have meant altitude not airspeed
    Last edited by headrush; 2019-Jul-31 at 03:36 AM. Reason: Add altitude

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •