Getting your Trinity Audio player ready...
7 min read

Air France 447

The final report on the Airbus A330 Rio to Paris Air France 447 accident is not out yet but preliminary information provides a lot of food for thought. It is a safe bet that many thousands of words will be written about this. They will come from all points of view and represent a multitude of opinions. Here is mine.

The accident sequence started when the airspeed sensing system on the airplane went haywire, likely because of pitot system icing. When the autopilot was fed erroneous data, it disconnected. The captain had just gone on a rest break in the cabin so one of the two first officers in the cockpit took manual control of the airplane. To control what was an apparent increase in indicated airspeed, caused by the pitot anomaly, the nose was pitched up by the pilot. The airplane climbed from Flight Level 350 to 38,000 feet where it stalled. The airspeed was low not high. Everything was backwards.

The captain was called to the cockpit but apparently did not retake control of the aircraft.

Wreckage of Air France 447 is recovered

The Air France crew never recovered.

The crew flying never recovered from the stall and the airplane descended to the surface of the ocean. It had gone from 38,000 feet to the water in under four minutes. All during the descent the angle of attack was deeply into the area of aerodynamic stall, reaching as much as 45-degrees.

The nose of the airplane was held or trimmed up for most of the descent. The wings rocked back and forth but lateral control was never fully lost. Despite the high angle of attack, the engines were apparently running and responsive up to the moment of impact. It is the opinion of most that the stall could have been broken and the airplane recovered. This, though, would have taken a lot of altitude. It might have also taken an almost instantaneous burst of brilliance in the first couple of minutes after the stall, when a recovery could have been possible without crashing into the water.

When the airplane reached the ocean surface, it was descending at a rate of 10,912 feet per minute. The pitch attitude was 16.2 degrees nose-up and the airplane was in a 5.3 degree left bank. The groundspeed was 107 knots. Airspeed indications varied widely during the descent, as reflected by the flight recorder.

Flight 447 was flying in or near the Intertropical Convergence Zone where there are usually numerous thunderstorms. The crew made a small heading adjustment right before the event started, presumably for weather. The airplane was flying in some turbulence and they had reduced airspeed from .82 to .80. No significant turbulence was reported or recorded and it is unlikely the Captain would have gone to the cabin on a rest break had the path ahead appeared menacing.

The A330 is a highly automated airplane as are all modern airliners and new-production general aviation airplanes. Operation of these airplanes requires specialized training and the mental dexterity to operate complex systems and to compensate when they malfunction. This is not the first time an automated airplane has crashed because of a problem with one or more of the systems and it won’t be the last.

The lesson is that, in a heartbeat, the automation can crap out and hand the pilot an airplane that has to actually be hand flown using basic knowledge and good piloting skills. The information supplied by the systems can be flawed, too, as it was in this case.

Nobody will ever know much if anything about the thought processes of the pilots as this event unfolded. I have no idea what the sounds and feelings would be in an A330 in such a completely developed stall but it could not have felt normal. One observer noted that the word “stall” was not used by the crew. Surely, though, they knew that something was badly wrong even though, almost as surely, they didn’t know what was wrong. If they had known, they would have tried to do something about it.

When confusion is induced by erroneous instrument readings, the basics of flying remain the same but the basics of pilot thinking do not. Quite obviously there was an indication of increasing airspeed which is what prompted the pilot to pull the nose of the airplane up which resulted in the stall. At this point the pilots might have become completely confused because of conflicting information.

There was a similar accident years ago, in a Boeing 727. The airplane was climbing, the pitot heat was not turned on, and when it passed through the freezing level, in clouds, the pitot iced over. As the airplane climbed, this resulted in an erroneous and continuing increase in indicated airspeed. The pilot flying pulled the nose up to try to control the airspeed and stalled the airplane. As with the A330, the stall warning activated which was apparently thought to be in error because the pilots were trying to deal with increasing airspeed, not a stall. The 727 descended from 24,000 feet to the ground (elevation 1,090) in 83 seconds.

I can relate to such confusion because, to use that hackneyed old saying, I have been there and done that.

It was on May 28, 1975. I was flying a brand-new Cessna T210, en route from Cessna Field in Wichita to a fuel stop in Colorado Springs.

There was a front to traverse and it got wet and bumpy. The rain was so hard that the airspeed was jumping around a bit. To help with the bumpy part I asked for a higher altitude. To help with the fluctuating airspeed I selected the alternate static source.

The BEA's map of AF 447's final moments

The BEA's map of AF447's final moments (click for full size view).

As I pitched the nose up and started to climb, the airspeed started increasing. To that point in the flight there had been no indication of convection but I wondered if I had suddenly flown into an updraft.

Then I thought I should take stock. The pitch attitude was normal, the engine power was set for climb, and there was no increase in air noise as would be found with increasing airspeed. My plan was to maintain the correct pitch attitude and think for a minute.

As is always a good practice I thought back to anything I had changed. The static source was the culprit. The alternate static system had been incorrectly hooked up when the airplane was assembled and when I selected it, it completely closed the static system instead of letting it vent to the cockpit as it was supposed to do.

By going back to the primary static source, I calmed things back to normal.

I’m going to digress for a moment for another lesson on how brand-new airplanes need to be flown with caution and suspicion.

We were taking two new T210s to Monument Valley for some air-to-air photography. After stopping at Colorado Springs we both vaulted up into the Flight Levels in honor of the Rocky Mountains. The other T210, flown by Chuck Hinson of Cessna, was ahead and 2,000 feet above us at FL220.

Chuck called the controller and told him he had a power failure and was declaring an emergency. I called the controller and told him I was responding to the emergency by circling where I was because I knew the other airplane would be descending through my altitude and we were flying in clouds.

As Chuck descended, he realized he still had some power. The turbo hose on the airplane had been incorrectly installed when the airplane was assembled and it had come loose, making it a normally aspirated airplane at 22,000 feet, or, one with not much power. So, two brand-new airplanes with manufacturing screw-ups, all in one day.

Pilots do have to always be ready for glitches. Anything can and will eventually happen. In an automated airliner you would think the flight control system would have an autopilot reversionary mode that would set cruise power, maintain a level- flight pitch attitude and keep the wings level without regard to other inputs. From reading the preliminary findings on the A330 crash it is certainly obvious that they need to go back to the drawing board and evaluate all the warning and flight control system things that happened and overwhelmed the crew, all because of a relatively simple loss of accurate airspeed indication.

In the meanwhile, pilots can always handle what automation doesn’t comprehend when something goes wrong. By hand, we can set cruise power, maintain a level-flight pitch attitude and keep the wings level. The next step would be to troubleshoot whatever is going on. That might be called basic airmanship, something that needs to be stressed while preparing pilots to operate automated airplanes. Bells and whistles are fine when they work correctly but in the end the system that leads to a successful outcome is called “pilot-in-command.”

Richard Collins
15 replies
    • Jim
      Jim says:

      I am intrigued by Kenneth’s analogy in respect to technology. The technology exists for systems to be monitored and for that data to be compared to data taken from a fault free normal flight. When an anomaly is detected instead of reaching for a manual to counter the effects system exist that can tell pilots what to do and when to do it. Satnav’s have been given the data for speed limits, known camera locations and a host of other information sources. We have simulators that can be controlled to produce the “normal” flight data. Simulators can also be made to reproduce malfunctions or spurious data. How easy/difficult would it be – it’s got to come – the sooner the better when you have 4 minutes to regain controlled flight.

  1. Capt Bob Lawrence
    Capt Bob Lawrence says:

    Great opinion piece, Richard. I had the same thoughts. Having taught the B-757/767 for a major airline, one of our concerns was to be sure the pilots knew when and how to use the automation, but equally important when NOT to use it. Basic flying skills are still the best safety factor. The new “fly by wire” generation of aircraft give very little, if any feedback to control inputs, which makes it even more important.
    Best regards….Bob

  2. Kenneth Nolde
    Kenneth Nolde says:

    I am a GA pilot (2200 hours plus)and a retired USAF Nav (B-52, RF-4C) with 5000 hrs. Does it occur to Mr. Collins and others who are professional pilots flying aircraft like the AirBus, that the 447 was downed by the very technology meant to make flying the aircraft easier and safer. From what I have read, the pilots cold not precisly determine the A/C status. Moreover, the system appears not able/equipped to cope with incident by warning the pilots of the situation.

    I have questions such as: could the system be disconnected so hand flying was an option? In a stall was there a pedal “shaker?” Was this a “grey” area of the software?

  3. Shahryar Saigol
    Shahryar Saigol says:

    Hand flying a giant airliner at 38,000 ft is just not done in the airlines. I seriously doubt any pilots do it deliberately. With a single stick controller which doesn’t actually move those pilots found themselves in unfamiliar circumstances once the autopilot tripped out. In a GA plane any change in wind noise or control feel is immediately noticed. The IFR pilot is trained to ignore those feelings and trust his instruments. The overspeed alarm was the scariest sound those pilots heard. How many of us would have said “wait a minute, let me analyze this” with that alarm blaring, telling us we have gone supersonic and have only seconds left to keep this plane from tearing its wings off.

  4. Captain Sean
    Captain Sean says:

    It incredible how many geniuses there is in Aviation,there’s always one who knows every-thing and can fly a situation he has not entered in, in his opinion to text book form in the memory of all Pilots who have tried and fail to recover, shame on you to think that in the same situation you could have done better,you who are a legend in your own mind.Remember this, the day you think you know it all,you are now officially a danger to yourself.

  5. Vasconcelos
    Vasconcelos says:

    Let’s stop to look for guilty people!

    It was given a chance to learn with others accident, so … Let’s start to learn and increase our knowledge.

  6. Mark S.
    Mark S. says:

    Great piece. I’m afraid that automation has gone too far. Yes, hand flying well up in the FLs is unusual, but it’s not impossible, nor should it be. At the end of the day, it is the job of a pilot, any pilot, to fly the airplane and if that can’t be accomplished when automation fails then there is a serious problem.

    I think of the NW 727 which had been hand flown from FL32 to FL40, then once established at the new altitude plummeted earthward. They dropped the gear and regained control in the nick of time landing successfully and saving all onboard. After a media-fueled fairy tail involving disabling the slats and putting in 10 degrees of flaps for a speed advantage at higher FLs it was determined that the PCU on the lower rudder had gone into a hard-over. Great stick and rudder work saved lives that day.

    As to the second example; what is it about COS that brings out flaws in brand new aircraft? :) Dateline, 1979: I had flown a week old Beech Sierra to COS to visit family. With my sister, niece and brother-in-law riding along, I arranged a twilight tour of the skies over C-Springs. After an hour or so, we were on long final for 35L when the prop governor decided to go into full decrease. By pushing it foward manually I could keep the engine running so I delegated that task to my burly brother-in-law, whom needed plenty of his strength to keep the prop control in full increase. I flew the approach using the throttle alone for thrust control and we arrived safely. (I won’t glorify it by using the term landing.)

    The next morning, a run up revealed an unflyable aircraft and we taxied to the Beech dealer on field, requiring several restarts along the way. As it turned out, a bracket securing the control cable for the prop governor was improperly tightened and was allowing the cable sheath to flex. A couple of quick twists with a pair of end wrenches and we were on our way, safely. But the story doesn’t end there.

    When I got back to the FBO from whom I rented the airplane I told the A&P on staff about it and he had known there was a problem all along. Now I am an A&P myself and know that opening the cowl and fixing this could take all of three minutes on a slow day, but this guy let the plane go on a long (800 NM x 2) cross country in this condition. The approach path to 35 L at COS is above some dodgy terrain including rock outcroppings and other hazards. The chance of making an off-field landing safely in that area was slim, at best. At the time I was young and easygoing, and simply walked away when the A&P told me that. Were this to happen today, I would have gone after his ticket.

  7. Kim Hunter
    Kim Hunter says:

    My belated comment is in support of postings by S.S., C.S. and V., above.

    When a critical flight system suddenly provides erroneous information; when it alerts the crew that they are about to depart controlled flight; when visual ques are absent, how many of us really believe we’d do better?

  8. Zeek
    Zeek says:

    “ One observer noted that the word ‘stall’ was not used by the crew.”

    “Hey, you guys—anybody know how to stop this damn stick from shaking?”

Trackbacks & Pingbacks

  1. […] are other fairly recent airline accidents related to this. One was the Airbus A330 that was stalled at 38,000 feet after an apparent ice-induced failure in the pitot static system and resulting erroneous airspeed […]

  2. […] the wake of the Air France 447 crash, many have suggested that basic stick and rudder skills have declined among airline pilots to the […]

  3. […] airline crews can lose the ability to think. The Air France A330 that crashed in the Atlantic did so after the crew stalled it at 38,000 feet and never recognized […]

  4. […] the wake of the Air France 447 crash, many have suggested that basic stick and rudder skills have declined among airline pilots to the […]

Comments are closed.