Flight 447: Air France on the Hot Seat

  • E-Mail this Article
  • View Printable Article
  • Text size:

    • A
    • A
    • A

The French BEA (Bureau d'Enquetes et d'Analyses) is livid, over the leaking of the cockpit voice recorder transcript from Air France's Flight 447, which crashed in the South Atlantic in June of 2009. After reading the transcript, which was revealed in unknown entirety in a book called Erreurs de Pilotage, by Jean-Pierre Otelli, it's easy to see why.

Not to put too fine a point on it, the transcript reveals a degree of cluelessness and abrogation of command that you don't often see in the professional airline world.

But it's not as if what the transcript reveals is a shocking revelation, although the degree of confusion is disturbing. In preliminary findings revealed last summer, The New York Times reported that the pilots in the two command seats were, improbably, never trained in hand flying a jet transport at high altitude. They were evidently intended to be system monitors and radio minders while the captain was temporarily out of the cockpit on a break. They got into trouble when the autopilot and autothrust dropped offline because of faulty airspeed data from iced-over pitot tubes.

But BEA also said the bogus airspeed data lasted for no more than a minute of the Airbus's four-minute descent. In other words, once the Captain returned to the cockpit, they had nearly two minutes to recover from the persistent stall they appeared to be holding the aircraft in. Why the Captain didn't forthrightly either issue decisive commands or take control is one of the mysteries BEA will have to sort out.

Wherever you come out on the ethics of Otelli's premature publication of the transcript, he does raise a difficult question the French will have to answer. In his words, who's going to be held responsible for this mess? Although BEA's report isn't due to be published until early next year, I wonder if his book is going to force action before that. I don't think it's exactly expecting too much for the pilots of a transatlantic airliner to be able to fly on raw data well enough to keep the wings level and avoid any stall, much less one lasting three minutes.

Have we become so arrogant in the age of automation that, institutionally, we think it unnecessary to train people to actually hand fly on instruments? Are all Air France pilots so poorly trained, or was this crew simply an anomaly? And how about other airlines? It strains credulity to think that three ATPs in a glass cockpit can't put their heads together to recover a stall. And what of Airbus? Are the displays and alarms badly designed enough to have contributed? Could this have happened to any pilots, no matter how well trained? We await BEA's illumination.

Ahead of that happening, I suspect Air France better have some answers and solutions long before the BEA makes its recommendations.

Here are two links with both a summary of the BEA interim report and the complete document.

Comments (130)

"Have we become so arrogant in the age of automation that, institutionally, we think it unnecessary to train people to actually hand fly on instruments?"

Sadly, the answer is likely, "Yes."

Posted by: Gary Dikkers | October 23, 2011 10:36 PM    Report this comment

I haven't read the transcript, but if the Captain was on the flight deck, why wasn't he flying the aircraft instead of the copilots?
What concerns me more is the comment that the stall warning was going off and they didn't seem to know what to do! (Kind of like the Coolgan crash).
I think back to my basic flight training: when the stall warning goes off - lower the nose!

Posted by: Richard Norris | October 24, 2011 6:07 AM    Report this comment

The three ATP pilots knew how to recover from a stall which is so basic in learning to fly any aircraft, but it is apperant that something about the automated system made it impossible to do. Certification should be suspended on this aircraft system until positive pilot control ia all circumstances is proven and demonstrated to all pilots type rated in these aircraft. With four minites to respond it had to be beyond the crew which leaves only an improperly design control system regardless of it's sofistication!

Posted by: Chris Hayne | October 24, 2011 8:21 AM    Report this comment

Was this another Airbus "feature" intended on not violating any aircraft limits, or simply that ALL the pilots on board goofed and kept the aircraft in a stall until it was WAY too late?

Posted by: R. Doe | October 24, 2011 8:34 AM    Report this comment

The problems associated with cockpit automation is nothing new, and has been with the industry since the first automated systems were introduced. Recall that when the first automated Airbus was introduced, it crashed into trees due to an improperly selected autopilot mode. It eventually lead to a redesign of the MCP to better indicate the current mode to the pilots. There's no denying that this automation has ultimately made flying safer, but I think flight training is still trying to catch up to how to properly teach how to most effectively use the automation.

There have also been cases of airliner crashes due to the computers providing conflicting information in the wake of a faulty sensor. It's easy to second-guess the pilots while reading the accident reports, but when you're in the dark or instrument conditions and you have all of the alarms and contradictory warnings going off at the same time, it can be quite difficult to figure out what is going wrong.

This ultimately sounds like an accident caused by poor CRM and pilot training. The cockpit automation may be a contributing factor, but ultimately the captain should have acted as the PIC he was and take control. Also, the fact that two apparently-unqualified pilots were even at the controls at all is appalling.

Posted by: Gary Baluha | October 24, 2011 10:00 AM    Report this comment

What was yhr life ot the cockpit crew during the 48 hours preceeding the fatal flight ?
Is this used as element of the inquiry by BEA ?
Ph.D Paris

Posted by: philippe ducrocq | October 24, 2011 10:55 AM    Report this comment

Is a complete stall recoverable at 38.000 feet with a 330 center of balance chosen aft for sparing fuel?

Posted by: philippe ducrocq | October 24, 2011 11:05 AM    Report this comment

Is a complete stall recoverable at 38.000 feet with a 330 center of balance chosen aft for sparing fuel?

Posted by: philippe ducrocq | October 24, 2011 11:05 AM    Report this comment

Why should we think the flight's PIC should know any better how to sort through the conflicting information than the two at the controls? He was brought up through the same training system as they were.

Posted by: Andrew Cleveland | October 24, 2011 2:22 PM    Report this comment

"In preliminary findings revealed last summer, The New York Times reported that the pilots in the two command seats were, improbably, never trained in hand flying a jet transport at high altitude."

If that's the case, please don't call them pilots. A person in the cockpit who can't hand fly the jet is not a "pilot." They may as well have had a steward or stewardess siting up there.

Pilot: A person who operates the flying controls of an aircraft.

Posted by: Gary Dikkers | October 24, 2011 2:49 PM    Report this comment

Unfortunately, kinda reminds me of the old joke about the airliner equipped with an autopilot and a dog- the autopilot is there to fly the airplane, the dog is there to bite the pilot if he tries to touch anything!

Posted by: Anthony Johnstone | October 24, 2011 3:09 PM    Report this comment

... and the pilot is there only to feed the dog.

Posted by: Glenn Killinger | October 25, 2011 9:17 AM    Report this comment

With turbulence and windshear, airbus says just keep the stick back. How can the pilot know when this is not the good action? Airbus do not say anything about that because this means they still need the pilot.
Airbuses needs the big push botton "Pilot has control" to be installed. And the the pilot will assume its responsability. Be sure...

Posted by: ANGEL ROS | October 26, 2011 2:31 AM    Report this comment

Guess you can't pass links through, so Google "a330 flight control system matrix" - the first hit should be a link to a Airbus flight control laws from the Airbus Driver web site.

Posted by: Keith Ellis | October 26, 2011 4:54 AM    Report this comment

Why should Air France alone be on the hot seat? Airbus should be roasted as well. This discussion is focusing on "pilot error" more than in the historical Airbus "fly by wire" systems and structural design errors. Computerized control systems and structural failures have been the cause of several EADS AIRBUS AIRCRAFT accidents leaving over 600 innocent passengers and crews dead. Is it possible that Air France, Lufthansa, New Zealand and American Airlines pilots are all incompetent? Will the likes of AA587 and AF447 systemic failures happen again? Do the “fly by wire” computer systems leave the aircraft unstabilized to where hand flying is not possible or safe? Is the Vertical Stabilizer structure a little or is it a big problem? I suspect that The Erreurs de Pilotage, by Jean-Pierre Otelli is another plot by Airbus to defocus from their responsibility. The degree of confusion is disturbing, I agree Paul.

Posted by: Rafael Sierra | October 26, 2011 6:21 AM    Report this comment

It apears the vertical stabliser, found floating, may have seperate prior to impact? If so a trim system failure may have made a recovery imposible. I hate to hear that three ATP pilots did not know how to fly an airplane. My conclusion, don't ride on an airbus until all this mess is sorted out by other than the manufacturer.

Posted by: Chris Hayne | October 26, 2011 6:31 AM    Report this comment

I don't know if this is the result of the investigation but here is a news clip on an A330 AD that was issued this year: "The European Aviation Safety Agency (EASA) has updated its airworthiness guidance for Airbus A330s and A340s related to a concern regarding flight control laws on the widebodies.

Last year, EASA issued an airworthiness directive (AD) highlighting a flaw in the existing flight control system. In certain circumstances, a pilot could reengage the aircraft’s autopilot after it has disconnected due to unreliable speed information when such a move should not be possible. In some instances, the reengaged autopilot, using erroneous speed information, could result in an abrupt “pitch up” command.

The new AD notes that Airbus has now issued new flight control computer software standards “that will inhibit autopilot engagement under unreliable airspeed conditions.”

EASA says that it considers the upgrade of the three flight control computers the terminating action for the AD. Carriers have 10 months from the issuance of the AD on Aug. 2 to upgrade or modify the flight control computer software according to the prescribed standards."

Posted by: Mark Hancock | October 26, 2011 7:09 AM    Report this comment

By the way, that news item is from Aviation Week, 2 Aug 11 (need to give them the credit they deserve).

Posted by: Mark Hancock | October 26, 2011 7:11 AM    Report this comment

I noticed that Raphael mentioned AA 587 with regard to a similar "systemic" failure. If memory serves, 587 was a non-fly by wire A300. Perhaps automation would have prevented the departure of the tail fin due to a possible structural weakness in the airframe.

Posted by: Steve Tobias | October 26, 2011 8:27 AM    Report this comment

Tail fins are part of the structural system. The AA587 was an A300, the New Zealand aircraft that went down in 2008 in the south coast of France was an A320 and the AF447 was an A330. All lost the Vertical Stabilizers, in my opinion, the structural system is not strong enough for an aircraft 200 foot long as even with "fly by wire" inputs the rudder can be applied rapidly left and right eventually exceeding established designed loads limits. FAR 25 Section 351 structural design parameters here are insufficient.

Posted by: Rafael Sierra | October 26, 2011 8:44 AM    Report this comment

"Please don't be so quick to claim that you could have recovered - this was not a Cessna 150!"

I have not read any comments here that suggests anyone thinks they could have recovered this situation when these three pilots could not. But the fact that they could not indicates a profound failure in either skill, training, aircraft systems or a combination of all three.

All transport category airplanes are conceived as a transportation system, of which the pilots are a critical part. The Airbus philosophy has been to diminish the importance of the pilots. What is unknown--and perhaps by be illuminated by this accident investigation--is which elements of this system failed, causing this accident. Has Airbus gotten it wrong? Do Air France and other airlines have the training wrong?

My guess is all of the above. At this stage, I would neither hold the pilots responsible nor blameless. But the confusion in the cockpit is inarguable and it is, at the least, perplexing as to why the Captain didn't exercise more authority than he appeared to.

Posted by: Paul Bertorelli | October 26, 2011 8:47 AM    Report this comment

Can't you just hear the Airbus/Air France instructors (in French) "Work with the systems, let the airplane work for you, it wants to fly, this is a new generation airplane, forget everything you knew...etc etc. etc.

After all the bigger they are the easier they are to fly right?

Many of the system designers and instructors have forgotten that ultimately, Dr. Bernoulli rules, regardless of the black boxes.

Posted by: MIKE HAND | October 26, 2011 8:54 AM    Report this comment

There is a suggestion that one of the pilots was holding the aircraft into the stall and that the rest if the crew did not realise that was so.

As an old pilot who has been nowhere near fly-by-wire, I wonder if the airbus dashboard should have (if it does not already), as a primary instrument, a coarse display of the positions of each flying control as commanded by each pilot and by the autopilot. The display would highlight which of the three controllers actually had control.

Posted by: R L S Butler | October 26, 2011 8:57 AM    Report this comment

To be fair, all modern transport jet aircraft use computerized fly-by-wire systems, and are susceptible to producing contradictory warning messages to the pilots. However, not all of these systems behave the same when receiving faulty data. It seems at least part of the problem in this case is that the displays don't clearly indicate when certain failure modes exist, or get buried in a list of other faults.

I personally prefer the way the Boeing flight control system works over the Airbus model, but it's important to realize that all of these aircraft can suffer from similar problems. When the aircraft loses valid speed data, how is it supposed to warn the pilots? This is not a new problem (I recall at least one incident involving a B757 with invalid pitot-static data), but it seems the answer has been to throw both stall and overspeed warnings. I think more research needs to be done in determining what the best human-computer interface should consist of.

Posted by: Gary Baluha | October 26, 2011 8:58 AM    Report this comment

Get of your high horse Paul. You might well have done NO better in that particular episode. The pilots' training was obviously deficient.... but DAH.... nothing new there. I have 32 years with a major US carrier plus Corporate, and military experience. Times that high altitude stall recovery or upset training has been conducted in the simulator....ZERO, Zilch, Nada. No regulatory requirement.

There are plenty to share the blame in this mishap. Airbus design & certification, the Pitot system manufacturer, AF Management & Pilot Training, oversight by regulatory authorities, and a general lack of imagination by all interested parties. Kind of like 911?!

Sure, the pilots are culpable. But they are dead, and everybody else is still above room temp engaged in obfuscation......with plenty of help from you.

In the US, we briefly had an upset training progam for Airline pilots conducted by an FAA subcontractor in a Lear, a response to the United COS and US Air PIT upset mishaps. I heard it was great training, but it didn't last, and very few had the benefit. No bucks. No Regulatory requirement. Hence, a needless expense.

In the end, AF is just another sad study in Economics of the Air, olde boy.

Posted by: John Langenheim | October 26, 2011 9:02 AM    Report this comment

R L S:
If one pilot was holding control inputs contrary to another pilot's inputs, that would indicate a breakdown of proper CRM techniques. There is only supposed to be one pilot handling the control inputs at one time. I'm also pretty sure Airbus' response would be that the pilots need not concern themselves with the actual aircraft control surfaces, and the pilots should be working with the computer, not against it.

I have no doubt that the pilots were trying to recover all the way to the point of impact, but it also seems clear that this is another case of poor CRM. If nothing else, it would seem to indicate that the training the crew received was inadequate.

Posted by: Gary Baluha | October 26, 2011 9:14 AM    Report this comment

Thanks Gary. I just wonder if the captain, standing behind the pilot seats, could actually see who was doing what.

Posted by: R L S Butler | October 26, 2011 9:21 AM    Report this comment

Sure, the pilots are culpable. But they are dead, and everybody else is still above room temp engaged in obfuscation......with plenty of help from you.

Thanks for that all knowing observation. I'll climb off my horse here and see if I can dust it off to make any sense of it, because on its face, it does not.

I am not obfuscating anything because there's not enough data to make a full determination here. I merely raised the questions that will have to be answered and, one hopes, dealt with. Despite the confusion in the cockpit, the crew deserves some benefit out doubt.

In your fashionable cynicism, you may spin that in any way you like.

Posted by: Paul Bertorelli | October 26, 2011 10:10 AM    Report this comment

So what is the moral of the story? Never trust your life to a bean counter.

Posted by: Andre Abreu | October 26, 2011 10:14 AM    Report this comment

I'm not quite sure what your point is. If none of us (or at least, those of us qualified to fly transport-category aircraft, of which I'm not) would not have been able to recover from the situation AF447 found themselves in, that would seem to be a failure of Airbus' design of the aircraft. Everything seems to indicate that mechanically, it was a perfectly flyable aircraft, and they had at least some reliable attitude instrumentation. I don't mean to minimize the difficulty they would have had, but it sounds like this was essentially a partial-panel-type of situation.

If the Airbus flight control system discourages raw-data hand flying, and training doesn't include this, then we have a pretty serious problem in my opinion. UA232 proves that a well-trained and coordinated crew is always going to triumph over automation any day (granted, the DC-10 isn't as fully automated as an A330, but it still proves the point).

Posted by: Gary Baluha | October 26, 2011 10:20 AM    Report this comment

This further strengthens my aversion to all things "Air Bus." When the tail came off of the Air Bus (I forget the model) in New York that was it for me. When I fly on the airlines one of my first questions is about the equipment...if it is Air Bus or ATR I just get another flight with better equipment. As for the qualifications of the 3 pilots...I am at a loss...I like to think that the dudes wiggling the controls are as interested in arriving alive as I am...now I have to second guess that too. Sad.

Posted by: william laatsch | October 26, 2011 10:47 AM    Report this comment

I would love to read this transcript. I wonder if "push the nose down" was ever mentioned. Is it really possible that this is not possible in any plane (not in a flat spin)?

Posted by: ARTHUR THOMPSON | October 26, 2011 2:21 PM    Report this comment

If I recall correctly (and backed up by wikipedia, so I'm not putting 100% faith in that), if a swept-wing aircraft enters a Mach tuck condition, it may develop into an unrecoverable dive. At the upper flight levels where AF447 was, the speed difference between a stall and over-speed is very small. Since they were without reliable airspeed indications, it's possible they wouldn't have known if they were in a stall or over-speed condition.

Of course, I'm not an expert in aerodynamics so I'm just guessing about possible scenarios. However, I do know that there are certain aerodynamic conditions a modern jet transport aircraft can get in where recovery is difficult or impossible. This may ultimately be a case where a delayed proper response is what led to the accident (which still goes back to improper training).

Posted by: Gary Baluha | October 26, 2011 3:05 PM    Report this comment

I expect they were trained that the aircraft will not stall. In normal flight control law, the FBW system is supposed to prevent a stall, regardless of how hard the pilot pulls back on the stick. Apparently, the FBW system had reverted to one of it's failure modes - like alternate law or abnormal alternate law. I am unaware of the indications (if any) that the crew receives when this happens - but the aircraft can stall in these modes. The issue goes back to how much training the crew receives in flying the airplane in anything but normal law and if they are aware that the airplane can stall in alternate law.

Posted by: Keith Ellis | October 26, 2011 3:25 PM    Report this comment

I would find it difficult to believe their training hadn't covered stalls in non-normal law. If us armchair transport jet pilots (excluding those of us who actually do fly these aircraft for a living) know about this, it would be incredible if they didn't.

I think this has more to do with training. If pilots are taught that you normally can't stall the aircraft regardless of what control inputs you apply, the law of primacy means the pilots are going to have a hard time applying the correct inputs when that normal safety barrier is removed. Perhaps Airbus should rethink the side-stick concept and have it provide a more direct mechanical feedback system so even in normal law, the pilots will be well aware that they are at the outer edge of the flight envelope and would be entering a stall if not for the computer preventing it.

Posted by: Gary Baluha | October 26, 2011 3:46 PM    Report this comment

The FRENCH version of RUSSIAN ROULETTE has now entered the Air-Line Business....
Vive la Difference OUI OUI

Posted by: Lance van Merlin | October 26, 2011 5:06 PM    Report this comment

I agree with Gary Baluha, regular training is needed to keep pilots skills and knowledge current and proficient especially under unusual situations. However, assuming that there are structural and ergonomic design problems that training will not fix, this assumption is based on the last 11 year Airbus accident/incident history. Here is a thought then; Ralph Nader did it with the Convair. American Airlines, Continental, Delta, FEDEX. United Airlines and other carriers maintain an operational Airbus fleet of about 1100 aircraft, a mix from A300s through A340s. US carriers need to openly join admitting that there are in-built design threats and have an Airbus recall and/or have Congress force the FAA ground all US registered Airbus aircraft. Let me reassure you Airbus lovers that should this be a Boeing or any other aircraft manufacturer’s problem I would be just as interested in promoting a fix. In the meantime I don’t fly Airbus, unless I have to, but that is another story.

Posted by: Rafael Sierra | October 26, 2011 8:03 PM    Report this comment

I agree with Gary Baluha, regular training is needed to keep pilots skills and knowledge current and proficient especially under unusual situations. However, assuming that there are structural and ergonomic design problems that training will not fix, this assumption is based on the last 11 year Airbus accident/incident history. Here is a thought then; Ralph Nader did it with the Convair. American Airlines, Continental, Delta, FEDEX. United Airlines and other carriers maintain an operational Airbus fleet of about 1100 aircraft, a mix from A300s through A340s. US carriers need to openly join admitting that there are in-built design threats and have an Airbus recall and/or have Congress force the FAA ground all US registered Airbus aircraft. Let me reassure you Airbus lovers that should this be a Boeing or any other aircraft manufacturer’s problem I would be just as interested in promoting a fix. In the meantime I don’t fly Airbus, unless I have to, but that is another story.

Posted by: Rafael Sierra | October 26, 2011 8:04 PM    Report this comment

I agree with Gary Baluha, regular training is needed to keep pilots skills and knowledge current and proficient especially under unusual situations. However, assuming that there are structural and ergonomic design problems that training will not fix, this assumption is based on the last 11 year Airbus accident/incident history. Here is a thought then; Ralph Nader did it with the Convair. American Airlines, Continental, Delta, FEDEX. United Airlines and other carriers maintain an operational Airbus fleet of about 1100 aircraft, a mix from A300s through A340s. US carriers need to openly join admitting that there are in-built design threats and have an Airbus recall and/or have Congress force the FAA ground all US registered Airbus aircraft. Let me reassure you Airbus lovers that should this be a Boeing or any other aircraft manufacturer’s problem I would be just as interested in promoting a fix. In the meantime I don’t fly Airbus, unless I have to, but that is another story.

Posted by: Rafael Sierra | October 26, 2011 8:06 PM    Report this comment

I have been retired for ll8 years. But, when I was on the A330 the company's position was that the autopilot could do a much better job of fllying the airplane than the human pilot. We were encouraged to engage the autopilot at 100 feet after T/O and disengage it on rollout. There was very little "hand flying" done in training.

Posted by: Unknown | October 26, 2011 8:27 PM    Report this comment

Well, the autopilot is in fact better at flying the airplane in the most efficient and effective manner than the human pilots. Even in general aviation aircraft like what I fly, the autopilot is more consistently precise than I am, and can be a life-saver for non-instrument rated (or rated-but-disoriented) pilots. However, no general aviation pilot would be caught not being able to effectively hand-fly the aircraft.

Admittedly, flying a transport-category jet isn't the same as flying a general aviation aircraft, and the former definitely gain a huge advantage from the efficiency afforded by fully-automated FBW systems. However, I think AF447 proves that the airlines still need to keep their pilots' skills sharp in hand-flying. If I recall correctly, some certain US airlines forbid, or at least discourage from, using certain autopilot modes in certain phases of flight, to force the pilots to hand-fly and keep up their skills.

IMO, the more automated the aircraft is, the more training the pilots should have in learning how to cope with automation failure modes.

Posted by: Gary Baluha | October 26, 2011 9:01 PM    Report this comment

Hi Gary: The point of my ramble was this. In the airline industry it is economics (bucks & euros) that are the primary determinant in how the system functions. Obviously maintaining a safe operation is a key consideration since crashing and killing the passengers is bad for business. It is expensive, wasteful, and will diminish future revenues, perhaps even threatening the viability of the enterprise. But in day to day operations, the dollar is the master served by an airline's management, and hence all the airline's suppliers, including the airframe manufacturers that supply our magnificent machines. Everything has its price.

Training is just one of the many functions that must have a cost accounting. It must meet regulatory requirements...but do so economically. And if the industry feels that a given regulation is too costly, they will fight it all the way to the White House.

The scope of training received by an airline pilot is almost totally determined by regulatory requirements. If it is not mandated, don't expect to receive it. That would be "guilding the lily" in the eyes of any bean counter worth his salt. Training is an evolutionary process, in that it is constantly being adjusted in response to the last mishap (or near mishap). I hope and expect to see "Unreliable Airspeed", high altitude upset, and deep stall recover at my next training event. Have I received that training in a simulator before? Not that I can recall.

Posted by: John Langenheim | October 26, 2011 10:56 PM    Report this comment

More.. Hi Gary. I guess Paul B. is the only guy that gets to be windy. I couldn't post my full ramble.

I would suspect that pilot training at Air France is not much different (better or worse) than at any US airline. It meets regulatory requirements, but often times that "bar" is not appropriately set to insure that a pilot has the skills that are needed vice those that are mandated.

The AF crew had a lot of things going against them. In the dark, in the weather, with unreliable instrumentation in a very complex automated airplane, the junior relief pilot unexpectedly handflying close to maximum altitude, with an inadvertant entry into a deep stall. And the Captain just woken up from his nap. Anybody train for that one on their last PC?

Finally, I disagree with you a bit that they had a "fully flyable aircraft". If the autopilot is functional but by design not capable of flying the aircraft in its degraded condition with compromised control laws, the aircraft is somewhat less than fully flyable.

The Accident report will be interesting to read.

Regards, John

Posted by: John Langenheim | October 26, 2011 11:00 PM    Report this comment

Earl Kramer said, "We were encouraged to engage the autopilot at 100 feet after T/O and disengage it on rollout. There was very little "hand flying" done in training."


I don't get it. For your airline, apparently the only reason for having real pilots on board was for appearances and to take over on those rare occasions when the sierra hit the fan. So it seems even more important that your company's real pilots should have received periodic training so they could actually fly the jet when needed.

Aren't you telling us the pilots in your airline were ill prepared to do their most important job?

Posted by: Gary Dikkers | October 26, 2011 11:19 PM    Report this comment

This wasn't the first time a jet transport was stalled at altitude and the crew couldn't recover by simply relaxing back pressure to achieve below-critical AOA. (See NTSB/AAR97-05 for the ABX DC-8-63F accident on 12/22/96.) So, whatever was going on here - where a cockpit full of supposedly professional aircrew were demonstrably incapable of doing what I teach primary students to do in a light airplane every day, viz., recover from a stall by adding power AND RELAXING BACK PRESSURE - is nothing new. I will say that with the Airbus sidestick controller, it's impossible to see the full or mostly-full aft stick pitch input. However, the ABX Stretch Eight crew would have had obvious clues with the conventional yoke arrangement. Then of course there's the Colgan Q400 quandary. I will just say that the proposed IACO multi-crew licensing concept regimen isn't likely to resolve this appalling lack of basic airmanship without some robust training modules to teach how to, and how NOT TO, recover from stalls!

Posted by: WILLIAM VAUGHN | October 26, 2011 11:31 PM    Report this comment

I recall reading in the transcript that, upon returning to the flight deck, the PIC told the copilot that he was holding the nose up. The response was, "I am?" Yet nothing changed. That's a bit disconcerting.

Still, I refuse to blame the pilots when their airplane is not working properly, especially if it is also distracting them with an array of useless and demanding warnings/alarms.

Many years ago, I had a passenger on board a cross-country Cessna 206 flight who went nuts as dusk made the ground below start to fade into darkness. I showed him how I could see cities ahead of us and pointed them out on the sectional chart, and tried to get him to look for the other towns and roads shown on the map, hoping this would calm him.

Nothing calmed him. He was beyond frantic. At one point he opened the door to jump out. With pandemonium in the cockpit, it is almost impossible to think, let alone fly. If my airplane was also broken, I may not have survived the flight.

The mental overload on that AF flight over the dark Atlantic had to have been worse. My take on this event, like others have mentioned, is to avoid Airbus airplanes. For an automated flight control system, this was an inexcusable flaw. Another flaw to go along with tails that snap off instead of bending a little.

Posted by: S. Lanchester | October 27, 2011 1:13 AM    Report this comment

AA587 was mentioned in passing, with reference to lost empennage. Pilot error was claimed.

The Associated Retired Aviation Professionals (ARAP) refers to this on their website, under the heading “AA587 News.” ARAP had demonstrated TWA800 was sabotaged. A cover-up and motives were exposed independently as a terrorist act in several books by Investigative Reporter Peter Lance.

With AA587, the claimed cause was rudder excursions by the copilot who had the takeoff. Supposedly the wake of the preceding 747 t/o—even given the 5-mile heavy separation--migrated into AA587 path and the driver sought to correct. Training error or something else?

Additional data—officially ignored—included witness claims AA587 shed debris (in the water and on a building roof) before the claimed wake turbulence event. Most telling are views from surveillance cameras of smoke/fire erupting from the AA587 center section.

Seems consistent with clothing bombers or other devices near a/c center sections? AA587 co-pilot was officially blamed when rudder travel was recorded by the FDR, but the bump recorded prior was not investigated as severed control lines.

Google “twa800.com/aa587news” and/or “USRead AA587.”

Posted by: Wash Phillips | October 27, 2011 3:20 AM    Report this comment

The following comment is from an aircraft engineer choosing to remain anonymous. "Pilots don’t really fly Airbuses. They ask the computer to do something. If the computer approves, the airplane does it. If the computer doesn’t approve, then the pilot has no way to make the airplane do it. Approach the stall speed, and the airplane will simply refuse to lower engine power or to raise the nose higher. Bank steeply, and the airplane will simply refuse to roll any further in that direction. Pull up at 2.5g, and the airplane will refuse to pull up any tighter (despite the fact that it is aerodynamically and structurally capable of doing so) even if it would make the difference between hitting a mountain or avoiding it."

Posted by: Rafael Sierra | October 27, 2011 3:49 AM    Report this comment

The guy who was flying it into the sea was one of those ab-initio fellows. Never really got proficient on a HP single or twin, Right Seat Direct after a LOT of theoretical (and simulator) training. This is the flaw of EASA and many other non-FAA training schemes. And caused by lack of GA that would allow pilots to gain their experience with only a few bags of expired checks at risk. Reminds me of the Egypt Air flight where none of the forward view passengers touched the controls and were only frantically trying to reengage the A/P which had disengaged-because..... they were upside down.

Posted by: ROBERT ZIEGLER | October 27, 2011 6:25 AM    Report this comment

John L.:
Thanks for the clarification, I understand your point now.

However, I still feel the pilots had a fully flyable aircraft. Their job certainly wouldn't have been easy, but the aerodynamic controls were still all fully functional, and their engines were still turning. If the mere fact that the autopilot disconnected itself due to unreliable instrumentation means the aircraft is now in serious danger, I don't think i'd want anyone I care about flying on these Airbus aircraft. But the US Air landing on the Hudson shows these planes should still be safely flyable outside of normal law. AF447 had instrument conditions to contend with, but i've flown planes partial panel with the issues they had (sim or with a safety pilot, of course), and it is possible. Three ATP-level pilots should have been able to figure it out (though I can't fault them if poor training was to blame).

Posted by: Gary Baluha | October 27, 2011 7:41 AM    Report this comment

Just for clarification the airbus control system has 3 modes that the pilot can select, one of them is pure hand flying without any envelope protection. So don't blame it on the plane, it was intact, all controls working, and the engines developing full power. At least at lower altitudes nobody should have been afraid of overspeed and just push the nose down when the stall alarm is blaring!

Posted by: ROBERT ZIEGLER | October 27, 2011 8:10 AM    Report this comment

Most of the comments here are from the point of view of a PIC in flight during an emergency. But, this is not your father's Boeing. If any of you could go to maintenance training on this aircraft, and learn how the systems work and what the design philosophy was... you'd never get aboard one. Airbus believes that pilots thinking too much is what causes accidents. They think their airplane knows better than you. It's rather scary.
And, sorry, but even in direct law the side stick control inputs go through the A/P system.

Posted by: David Kyle | October 27, 2011 8:32 AM    Report this comment

The US Air landing was great, but the flight control system was in normal law, Sully had envelope protection and could pull the stick fully back and hold it and the airplane would be at minimum speed with no danger of stalling. The Air France airplane was not intact, all three pitot static systems had iced over (granted they had thawed at lower altitude) but the FBW system had gone to one of it's non-normal laws, did not revert back to normal once the pitot static systems had recovered and that same system ignores the AOA vanes even though they were fully functional. An Airbus pilot NEVER has direct control of the airplane - sidestick inputs are always through the FBW computers even in direct mode. If all the computers fail, the pilot is forced to fly pitch through power and stabilizer trim and roll using asymmetric thrust and the manually connected rudders.

Posted by: Keith Ellis | October 27, 2011 8:42 AM    Report this comment

Hi Gary - I have never been an Airbus driver so I don't know if the Hudson 320 would have been operating in something other than "Normal Law" flight controls.

Once the AF crew departed controlled flight, their problems were seriously compounded. Given the degree of confusion, instrument limitations / ambiguities and startle factor / denial, they were in uncharted territory as far as Airline training goes. The use of continued inappropriate "pro-stall" inputs may well have induced a full nose up trim condition that would have further delayed or prevented recovery without aggressive and prolonged correct "stall recovery" inputs and the time necessary for these to take effect.

Upset training is part of the curriculum, but as practised in the simulator is done at intermediate to low altitudes and usually with some degree of visual reference and full instrumentation. The pilot flying at the time is the pilot recovering, and inadvertantly stalling the airplane during the recovery attempt so the other guy has to take over is not part of the scenario. You usually have some inkling before the episode starts that it is going to happen. And with Boeing, there are readily apparent tactile and visual indications of control yoke position so the non-flying pilot can see what his counterpart is attempting to do.

As you suggest, training must have been an issue. We never go out to practice a few stalls in the "old airliner". Perhaps we will start doing so soon in the sim.

Posted by: John Langenheim | October 27, 2011 8:50 AM    Report this comment

But the stall warning was not continually blaring. It was designed to cease alarming at indicated speeds below 100kts because the airplane was assumed to be on the ground. So every time the crew put the nose down, the horn would sound. If they raised the nose, the warning would cease. Just another of many Airbus gotchas.

Posted by: Bob Merritt | October 27, 2011 8:55 AM    Report this comment

I thought I recall it being mentioned that the aircraft went to (or perhaps one of them manually selected) alternate law once both engines had failed.

Separately, even the modern Boeing aircraft are FBW systems, with all control inputs going through a computer first. That's the price we pay for designing purposely-unstable swept-wing aircraft. However, the Boeing philosophy is that the pilot always has the last word, and if he/she wants to go beyond the normal flight envelope, it won't prevent that. I don't know how the new 787 FBW control system works, but I seem to recall the 777 FBW system specifically designed to allow pilot override. Does the Airbus "direct law" mode not also allow this?

Posted by: Gary Baluha | October 27, 2011 8:58 AM    Report this comment

Robert, before you bash ab-initio training, it will be coming to the U.S. before too long. Sorry this is off topic, but - everyone wants Sully and Skiles flying their brand new Dreamliner, with Christina Ricci as your personal flight attendant serving you Chateaubriand and champagne in your free upgrade first class seat, with your 3 carry on bags and steamer trunk checked for free while you fly coast-to-coast for $99. Not reasonable. Given the proposed 1500 hour rule from Congress (700 from the FAA) for Part 121 pilots, where do you think they are going to come from? Who can afford that much training/flight time to then go to work for a regional making $17,000 per year. Not going to happen. The military is training many less pilots, there is a 10 year commitment, are making over $100K when they are done, have an pension and medical benefits in retirement and many are just flying drones remotely from a computer. They are not going to leave the military to fly for the airlines. I'm not defending the Air France pilots, but the traveling public has demanded rock bottom fares, the airlines have delivered, but you are not going to get pilots like Sully and Skiles doing this job anymore nor are airlines going to train like they used to (I've been doing this for 28 years now) unless people start paying for it.

Posted by: Keith Ellis | October 27, 2011 9:07 AM    Report this comment

Keith - Bingo! You are right on target. Well done!

Posted by: John Langenheim | October 27, 2011 9:11 AM    Report this comment

From an publication on Airbus control laws-
DIRECT LAW Direct law is the lowest level of computer flight control and occurs with certain multiple failures. Pilot control inputs are transmitted unmodified to the control surfaces, providing a direct relationship between sidestick and control surface.
Control sensitivity depends on airspeed and NO autotrimming is available.An amber message USE MAN PITCH TRIM appears on the PFD. If the flight controls degrade to Alternate Law, Direct Law automatically becomes active when the landing gear is extended if no autopilots are engaged. If an autopilot is engaged, the airplane will remain in Alternate Law until the autopilot is disconnected. There are no protections provided in Direct Law, however overspeed and stall aural warnings are provided. The PFD airspeed scale remains the same as in Alternate Law.

Posted by: Keith Ellis | October 27, 2011 9:12 AM    Report this comment

I have spent the last 6 years as a simulator instructor for the Global Express Bizjet. We have relegated modern pilots to "Aircraft System Supervisors" (ASS's) who can't wait to click on the autopilot and play with the FMS's and iPads. When the automation goes "Tango Uniform", so does their flying. Our training programs need to take a hard look at how we let this happen and how best to return to training pilots and not ASS's. I saw Sullenberger on the news commenting on this accident and he mentioned that the automation geeks have taken out of the cockpit an indicator that could have made a difference - Angle of Attack! My airplane has it but the geeks decided to hide it from the pilots! We really need to learn from this accident and get back to basics!

Posted by: George Findlay | October 27, 2011 9:30 AM    Report this comment

One or two comments:
The right hand pilot, took control when the pitot tube iced and knocked out the autopilot.
He pulled back hard -- no -one knows why and from later bits of the transcript may not have even been aware that he was.
By the time the speed indicators came back, the plan was stalled and flying slower than the computer engineers writing the stall warning program even imagined. The stall warning was designed to cut out below something like 40 knots.
So while falling stalled there was no stall warning. The one time the stick was pushed forward, the speed went over the 40 knot mark and the stall warning sounded again.
The pilots had not had uniform training, the youngest in the right seat was pure Air France cadet onwards, the left seat was a "late entry" Air France training, and the Captain had been absorbed into Air France during one of the acquisitions.
This might have added to the confusion and the captain being macho with the weather, as the transcripts now show us.
A recent TV program has highlighted safety-culture concerns with Air France -- Concorde, Toronto and 447 make for a bad 10 years. Things may be changing.

Posted by: Brian McCulloch | October 27, 2011 9:41 AM    Report this comment

My experience, thousands of hours flying Boeings, 777 FBW and conventional control. Thousands of hours flying F-16 FBW. Currently Airbus 320 Captain. These are all outstanding airplanes by any measure. They fly millions of passengers millions of miles or kill bad guys with an efficiency and level of safety we could have only dreamed of 20 years ago. The Boeing hull loss data from 1959 - 2010 proves it; all modern airliners less than 1 hull loss in one million departures, the A330 .46 the 767 .38. The one exception the MD-11 at 3.69. Now there is a real point of discussion. How many will not ride in a MD-11? The FBW systems on the 777 and Air Bus are very similar, but there are some differences mostly minor. There is one significant difference, feedback provided to the pilot. The Boeing yoke and throttles move to reflect flight control and engine changes while the autopilot is on, in the AB they do not. This takes some getting use to. I wish the throttles moved on the AB, but I understand the increased complexity/weight required to do that. Different design philosophy. I am disturbed by the AF crash, there is a level of complacency and atrophied basic flying skill that can creep on to the flight deck of modern airliners because the systems are so good and reliable. I try to keep these things in perspective and it is best that a knowledgable group of aviators as ourselves does not jump to conclusions based mostly upon emotion instead of the cold hard facts.

Posted by: Tommy Williams | October 27, 2011 10:06 AM    Report this comment

George Findly,

Great point about AOA. Had the AF jet had an AOA indicator visible somewhere in the cockpit, the pilots would have known exactly what was needed to recover.

Posted by: Gary Dikkers | October 27, 2011 10:10 AM    Report this comment

I was told by a line pilot that the FBW accepts the average of the two side-stick positions. If one is full forward and one full back, the net result is no change.

Can anyone confirm this? We were enjoying Mexican food and beer at the time.

Another point I think important, is that line pilots have never flown into a stall condition with these aircraft. And that even when done by test pilots during certification, a deep stall condition is not allowed to develope. So, we have three pilots in a flight condition no one has experienced before.

Posted by: Edd Weninger | October 27, 2011 2:46 PM    Report this comment

"So, we have three pilots in a flight condition no one has experienced before."

All the more reason their jet should have had an AOA indicator.

Posted by: Gary Dikkers | October 27, 2011 4:04 PM    Report this comment

It is the average of the two side stick positions, although you cannot double roll or pitch rate by applying both in the same direction. BTW, learning to fly a sidestick takes about 10 minutes, very intuitive. Also, much more comfortable cockpit than a yoke.

Posted by: Tommy Williams | October 27, 2011 4:10 PM    Report this comment

Looking at a cockpit photo of the A330-200 on airliners*net, there is a mechanical attitude indicator just to the right of the pilot. Doesn't that constitute an angle of attack indicator?

Posted by: THOMAS M CONNOR | October 27, 2011 4:44 PM    Report this comment


Posted by: THOMAS M CONNOR | October 27, 2011 5:12 PM    Report this comment

Sorry, the devil's making me do this. I think the crew was afraid of the airplane and did what the French do best when afraid. They surrendered.

Posted by: Ronnie Gravitt | October 27, 2011 5:42 PM    Report this comment

I have been following this accident because I avoided flying an Airbus due to what I considered to be a poor design philosophy favouring automation over manual control. It seems to have been vindicated quite a few times. The basic principle is KISS (Keep it Simple,St----, and automation may have surface gloss of simplicity, but as soon as something doesn't work properly it turns ugly very quickly. As most computer owners know, probably sooner rather than later something won't work properly.

Some comments have mentioned it but I don't see any that came right out and said it. I haven't seen the CVR transcript, but I believe the aircraft was flyable, the aircraft did react properly to control inputs. I suspect the pilots had no idea that the airplane was stalled, due to the lack of physical cues and the confusion resulting from cockpit displays and controls design. As one pilot stated, this accident would not have happened in an earlier generation aircraft.

Posted by: Hal Marsden | October 27, 2011 7:15 PM    Report this comment

Tom, the mechanical standby attitude indicator is not an AOA indicator.

Posted by: Tommy Williams | October 27, 2011 8:13 PM    Report this comment

Hal, I agree with you, things can get very complicated with multiple system failures in a highly automated cockpit. Of course they can get pretty complictaed in a less automated cockpit too. My experience with the 777 and Air Bus both FBW is that there is very little difference. Click the autopilot and auto thrust off and they fly very similar to a conventional airplane. Degrade the flight control systems to Direct Law/Mode and they are essentially 727s with no protections. Hand flying either of these airplanes requires no special skills beyond what a 727 driver has it's just we highly automated airplane drivers do not do it so much. Both the Boeing and Air Bus are great airplanes that are a joy to fly, nothing sinister in either. The 727 driver is forced to keep his skills up, he has no choice. The FBW pilot can also maintain his skills, but he must disconnect some automation, auto pilot and auto thrust, to do it.

Posted by: Tommy Williams | October 27, 2011 8:25 PM    Report this comment

In response to Keith Ellis: I respectfully contend that it's bogus to say the public demands crummier pilot experience by choosing lower airfares. I've been flying the airlines for 50 years, and never have I seen a checkbox that says "Pay $30 more, get more experienced pilots." No, you pay twice as much as the guy in the seat next to you if you pick the wrong time to book your flight.

Posted by: John Schubert | October 27, 2011 9:03 PM    Report this comment

Hal, this is not the right analysis. I read the entire cockpit conversation in the original book in French yesterday. In a nutshell, there has been a black hole simultaneously in the mind of 3 pilots. The flying pilot made mistakes like a first hour student. There was intermittent pitot icing removing airspeed input at the higher altitudes, that went away as the airplane descended. There were not "multiple system failures", just pitot icing. They kept the plane in a permanent stall, at an angle of attack often exceeding 40 degrees (and the angle of attack indicator never failed), and even put in full deflection aileron input without the plane departing in a spin. Talk about excellent aerodynamic engineering at airbus. They had all stall indications including severe buffeting. It was pure and simple Darwin-Award type of ignorance that not only eliminated 2 definite and one potential award winners, but also more than 200 innocents. That is why I think that ALL pilots should earn their wings alone, at night, with ice, with a few bags of expired checks as only potential victims. And do NOT increase allowable simulator time for qualifications.
And regarding a higher up comment: Nobody was playing "macho" with the weather, they had simply the gain of their radar incorrectly adjusted, so did not see the weather. They made a lot of simple omissions/errors, so their minds were elsewhere: None were in a state of mind to successfully land a conventional gear airplane.

Posted by: ROBERT ZIEGLER | October 28, 2011 2:57 AM    Report this comment

Robert, I don't read French and haven't seen the book in English. Does the book seem credibly researched and presented?

Posted by: Paul Bertorelli | October 28, 2011 6:37 AM    Report this comment

The sad truth is that the pilots only had to maintain the power at the previous level and maintain the pitch attitude somewhere near what it was when the pitot tubes failed or iced over.

Unfortunately, the pilot flying pulled the sidestick (all the way?) back and held it. That caused the trim to pitch the aircraft up. When they topped out at 38,000+ feet, they tried to use "approach to stall" recovery techniques. Such training is/was the only stall recovery training you get in jet transports.

The stall was recoverable, and the lower they got, the more recoverable it would become, but the nose has to come down far enough to stop the stall buffet and recover. They evidently did not understand that point. Colgan 3407 and the Airborne Express DC-8 Accidents were similar failures on the part of the pilots and their training and mind set.

Posted by: THOMAS OLSEN | October 28, 2011 6:45 AM    Report this comment

The most glaring failure here is that their training failed to teach this set of pilots the most important aviation rule. That is: FLY THE AIRPLANE FIRST!

Whatever their warnings were telling them, it is always, always necessary to maintain control of the aircraft.

There is a lot more "autopilot dependence" in every phase of aviation. Private pilots who cannot make crosswind landings but who can use a Garmin to find their girlfriend’s garage to the air carrier pilot who sets the autopilot in climb and then picks up a magazine and ignores the autopilot until ATC asks him where he is going. The reliability of the autopilot and the way they are used in accordance with policy diminishes basic piloting skills and even basic thinking. This is a fact has been clear for at least 25 years that I am aware. There are lots of examples fatal accidents caused by over reliance on autopilots. This accident is just the one where the most lives were lost because the industry failed to act. Manufacturer’s, regulatory agencies and airlines keep buying and flying airplanes that discourage pilots from doing what was necessary to maintain basic skills.

Posted by: THOMAS OLSEN | October 28, 2011 7:06 AM    Report this comment

These highly automated systems are very good and in the aggregate are a positive in regards to flying safety. What this AF accident brings to light is that there are some "gotchas" that we have not adequately addressed. Airline travel is as safe as it has ever been and we must strive to make it even safer. What we are now working on is bringing hull loss rates to less than .10/million departures. What a wonderful problem to have. We are doing a lot right and we must insure our "fixes" do not create more problems than they solve.
The challenge, most air carrier operations are performed more efficiently and safer with auto pilot and auto thrust engaged, but the pilots must be fully prepared to hand fly if needed. Probably in an other than normal situation. The skill set required is perishable, it must be exercised on a regular basis. Where is the best place to exercise that skill? As an A-320 Captain I wrestle with that decision on every trip. I can keep my SA at a higher level by using automation when in busy environments. Should paying passengers be exposed to my "practice" as I click off automation to hand fly? They certainly are at times. Is it the safest way to operate? Probably not.
I have an RV-8, so I do a lot of hand flying in it, but does it translate to A-320 hand flying proficiency? Maybe, but I think not a lot. Obviously, I do not have all the answers, maybe one of you guys do.

Posted by: Tommy Williams | October 28, 2011 7:52 AM    Report this comment

For anyone who hasn't read the most recent interim report, I would suggest reading it as it does answer a lot of questions being asked here. I'm only halfway through reading the 114 page report, but it seems to indicate dysfunctional crew coordination at the very least. Also interesting is how the Captain's relief pilot (and PNF) kept telling the PF (in the right seat) to lower the nose.

We can talk about the erosion of hand-flying skills due to automation and how automated systems may be a hindrance in certain failure conditions all day. However, I think ultimately this comes down to improper crew coordination. How could the other two pilots allow the PF to continually apply the incorrect control inputs without taking some action?

Posted by: Gary Baluha | October 28, 2011 8:02 AM    Report this comment

For those who don't know, the Angle of Attack Indicator measures the angular difference between the chord of the wing and the relative wind. It is independent of attitude and speed. For each wing configuration (slats, flaps, etc) there is a "critical angle of attack" at which the wing will stall. In my USAF days flying the C-5, we had to know the angle of attack for various flight conditions (cruise, holding, and approaches with different flap settings) just in case that was all we had. Perhaps they were ahead of their time in providing and teaching us to use these valuable tools!

Posted by: George Findlay | October 28, 2011 9:47 AM    Report this comment

"Should paying passengers be exposed to my "practice" as I click off automation to hand fly? They certainly are at times. Is it the safest way to operate? Probably not."

Which raises this dilemma: Perhaps, systemically, we are at the practical and economic safety zenith. In other words, having pilots maintain flying skills by hand flying erodes the inherent safety built into having the automation do everything reliably and consistently.

So maybe the price you pay is occasionally having a low-skill pilot or crew crash an airplane because having the pilots actually fly causes more crashes. It could be the numbers will show this.

In that case, AF 447 is the cost of doing business. Hard sell, that, but it could be true.

Posted by: Paul Bertorelli | October 28, 2011 10:33 AM    Report this comment

And by the way, at the end of the blog text, I have placed links to both a summary of the BEA interim report and the full document.

Posted by: Paul Bertorelli | October 28, 2011 10:55 AM    Report this comment

Paul, I can't help but think the next logical step will be to go to a fully automated flown by computer with no pilot - since they need 'hand-flying' time to stay current and that is obviously less efficient (and safe) then the computer flying.

If that happens, it will be the time for me to drive - which is getting to be a more economical choice for anything under 400 miles, but that is another discussion.

Posted by: Richard Norris | October 28, 2011 11:00 AM    Report this comment

Except that I think AF447 shows the importance of having qualified pilots up front for when the automation does fail. It just happens in this case, apparently they weren't fully qualified.

Posted by: Gary Baluha | October 28, 2011 11:06 AM    Report this comment

Paul, well put. Remember the days of spin training, actual V1 cuts practiced in an airliner, not the sim. Actual approach to stall training in a swept wing airliner. It was all good stuff, made for well qualified pilots, if you survived training. Accident rates were astronomical and would not be tolerated today, yet very rarely we call on our brethren to exercise the skill set that we had back in the day, but they are not proficient or have never been exposed to that level of frequent highly dynamic flight environment. We see it now in individuals who have never really max performed an actual airplane who are professional pilots who consistently make landings equal take offs, so they must be doing something right. The airline industry has benefited from a group of aviators who have been trained in the hands on method and although they may not be very proficient have the experience to fall back upon when things go way wrong. That experience base is changing and we as a profession must come to grips with it and acknowledge it and smartly figure out ways to mitigate that lack of experience.

Posted by: Tommy Williams | October 28, 2011 11:24 AM    Report this comment

If the Colgan-type first pilot will be required to have 1750 hours (or a number like it) before he/she can earn professionally, there will be lots of time for advanced flying training of the sort that we military pilots got "free".

Posted by: R L S Butler | October 28, 2011 11:31 AM    Report this comment

Tommy Williams asked: "Should paying passengers be exposed to my "practice" as I click off automation to hand fly?"


Not unless your hand-flying is rough, dangerous, or somehow inconvenience the pax.

If you're practicing stalls and falls, chandelles, whifferdills, and steep turns they might not care for it.

But if you hand-fly through takeoff, climb out, and level-off, they shouldn't even realize what's happening. The same is true for pulling back the power by hand, and flying the penetration and approach. The pax should not be able to recognize what a smooth pilot is doing -- even when hand flying.

Posted by: Gary Dikkers | October 28, 2011 1:18 PM    Report this comment

In airline flying duties are divided between the pilot flying (PF) and pilot monitoring(PM). While automation is on PF selects airspeed, altitude, heading and at some airlines FMS changes. PM runs the radios. If automation is disconnected the only thing PF does is fly all other duties fall to the PM. Nice day, low traffic cooperative ATC, hand flying is fun, but in and out of major airports and wx things get very busy and we do this a lot. Lot's of exposure, some would argue not worth the risk.

Posted by: Tommy Williams | October 28, 2011 4:39 PM    Report this comment

In many cases the distraction created by re-programming a revised clearance is much greater than simply hand-flying, or at least reducing the automation level until later. The culture of the company I worked for did not encourage manual flight to say the least. More disturbing to me was that many of the younger pilots would never even consider hand-flying, at least in part because other captains would not let them, but also because they enjoyed the challenge of 'computer games' more than flying! They would never manually adjust power (Boeing) and I doubt they had any idea what power settings were normal for a given flight condition, which is after all only about half of flying an airplane!

Posted by: Hal Marsden | October 28, 2011 6:31 PM    Report this comment

Paul, I read the "Erreur de Pilotage" transcript two days ago. It is well written and documented. As for the author’s points of view they are as subjective as all the previous comments on this very page.

Being based in Paris, my PPL flight instructors are Air France pilots. Many AF pilots fly privately or are benevolent instructors in French "aeroclubs". Therefore I can attest that all the upper comments regarding lack of hand flying skills are simply infatuate nonsense. These pilots instruct students like me stall recoveries day in day out.

Following the publication of the preliminary report, all the AF pilots have already undertaken training for high altitude stall recovery.

The only comment I heard on this topic from one of the pilots (747 copilot) was that it is difficult to recover an airplane at high altitude. The yoke/joystick has to be handled extremely gently in order to avoid the airplane stalling again.

Taking into account the skill regression due to stress at night, in IMC, overflying a CB, I can imagine that even a well trained and current pilot would have less than 100% probability of doing it right when startled by the unforeseen situation.
Being only a low time PPL I refrain to comment on the crew interaction.
I felt compelled to write this comment as some of the comments above assume that AF pilot training is somehow subpar to US standards.

Fly safe!

Posted by: Radu Damianov | October 28, 2011 7:01 PM    Report this comment


All my jet time is in fighters or high-performance jet trainers -- no pilot flying with another pilot monitoring -- just me. I've flown in and out of major airports, in and out of weather so bad it made me wish I'd become a professional banjo play instead of a pilot, and have done other things where it even -- at times -- got "very busy."

Occasionally I would turn on the "altitude hold" feature when in crappy weather and I needed to root around in the map case to pull out an approach plate, but that's the limit of my autopilot or FMS experience. Other than those brief forays using "altitude hold," all my flying has been hand-flying.

If you were to turn off all the automation on your A-320 and just fly the jet, you might even enjoy it. If you could forget about a couple of hundred pax sitting behind you, it wouldn't even be that much different than your RV-8 -- just you and the flight controls winging through the sky, talking to ATC, en route to your next destination. The PM could still help you with the radios and navigation, and if occasionally you bounced one on, could also sooth your ego with a snappy, "It's a bit gusty today isn't it Sir?" before offering to buy the first round in the hotel bar.

Posted by: Gary Dikkers | October 28, 2011 11:36 PM    Report this comment


I've spent 26 years 3700 hours flying single seat fighters Vipers and Eagles mostly. I too rarely used the alt hold or heading select. It's different in an airliner. The aircraft is much less responsive and the crew concept, although a great backup, can be cumbersome at times. I do hand fly the A-320 and almost always land it, but not nearly as much as we did in fighters. IMO, more hand flying of airliners is really not the answer. The answer is a greater experience base before pilots climb into any seat in a 121 operation.

Posted by: Tommy Williams | October 29, 2011 7:30 AM    Report this comment

It's great to say 121 pilots should have more experience, but where do you propose they get that experience? With a lot of airline pilots due to retire in the coming years due to age limitations, there aren't enough military pilots to fill the gap. That leaves the rest of the pilots with airline aspirations to fund their own training experience while earning barely-livable wages flight instructing or flying for the regionals. This just leads to building hours, not necessarily experience.

Also, I think it is worth pointing out that general aviation and pilot certification is vastly different in the US than it is elsewhere in the world. Two pilots with the same hours in the same aircraft will have much different experiences if one is a US pilot and the other is a european pilot (though I'm not going to say one or the other's experience is "better", to avoid that off-topic discussion).

Posted by: Gary Baluha | October 29, 2011 9:19 AM    Report this comment

It begs us to consider why Airbus was unable to make a simple thing like pitot heat with reliability. That isn't a complicated system at all, one of most basic electrical devices on the aircraft and they couldn't get that right? What other things did they screw up also?

Posted by: Richard May | October 29, 2011 12:33 PM    Report this comment

The soon to be requirement for pilot to hold an ATP to fly as a Part 121 pilot, Captain or F/O, is a good thing. When the supply of pilots shrinks because of this new requirement the equilibrium point in the supply and demand curve will shift requireing employers to make their airline pilot jobs attractive enough to get the military pilots to leave active duty, motivate those interested to put up with the not so good jobs for little salary. This was the way it was when I was hired. Airline jobs were very good. Good enough to leave the military, good enough to work long hours for little pay. The experience will be gained in the military, flying corporate, flight instructing, flying single pilot night cargo. Most of them very arduous methods to achieve the ATP prereqs. This is no hit on ab initio pilots many I have worked with are outstanding people who have a real passion for flying, but to develop that "sixth sense" it is experience that counts. Quality time in the seat.

Posted by: Tommy Williams | October 29, 2011 4:17 PM    Report this comment

"IMO, more hand flying of airliners is really not the answer."


I agree. What the airlines need are aggressive training in partial panel recoveries, unusual attitude recoveries, upset training, and manual recoveries with disabled or barely functioning automatic systems.

Of course the question is how an airline would pay for it? My guess is that in the arcane and mysterious science of bean-counting and risk analysis, it is more cost effective to suck it up and absorb the cost of the admittedly rare catastrophe than it is to bring all their flight crews up-to-speed.

You and I know that when flying jets in the military, almost all of our time was spent training and preparing for all manner of possible emergencies and scenarios. The airlines simply can't afford to pay for the amount of training we received.

Check SIx!

Posted by: Gary Dikkers | October 29, 2011 9:02 PM    Report this comment

Re: the school of thought that the stabilizer on AF447 separated in flight, preventing stall recovery. If the stabilizer departs the aircraft, it will pitch over nose-down VERY abruptly. It will NOT revert to a nose-high attitude. Since the FDR did not show that any such pitch-over (nose down) occurred, then its obvious that the stabilizer did not depart the aircraft prior to impact. Back when a stabilizer departed from Capital Airlines Viscount in a severe thunderstorm over Baltimore, the plane pitched nose-down abruptly. This snapped off one wing; snapped the fuselage into three pieces; and all five pieces fell to earth independantly. When a Capital Airlines Viscount was inbound to Baltimore and ran into a whistling swan, the impact tore off one stabilizer. The plane pitched over abruptly nose down and crashed. When the stabilizer iced up on a Viscount landing at Saginaw one winter day, the stabilizer stalled; the nose pitched down abruptly; and the Viscount nosed over and crashed inverted (half way through an outside loop) from about 200 feet and hit just short of the runway. In short, the stabilizer holds the tail down. If it stalls or departs, the plane will pitch nose down - - NOT nose up.

Posted by: Carl Jordan | October 30, 2011 8:11 AM    Report this comment

Carl , it seems that nose down happens after loss (or stall) of horizontal stabilizer . What was found floating after 447 crash looks like vertical stabilizer ?

Posted by: philippe ducrocq | October 30, 2011 9:59 AM    Report this comment

Carl , it seems that nose down happens after loss (or stall) of horizontal stabilizer . What was found floating after 447 crash looks like vertical stabilizer ?

Posted by: philippe ducrocq | October 30, 2011 9:59 AM    Report this comment

The Airbus tail snapping threat goes on, it makes no difference whether it is a non fly-by-wire airplane (A300s) or fly-by-wire airplanes (A320s, A330s, A340s). All aircraft can be made to exceed Vertical Stabilizer lateral load limits either by rapid successive un-commanded rudder movements or by pilot input, however, not all aircraft have had Vertical Stabilizers snap off. China Airlines Flight 006 from Taipei to Los Angeles in 1985 is a good example.
Seven years after AA587 (A300), a New Zealand aircraft (A320), crashed during a stall recovery caused by pilot induced oscillations much like AA587, triggering the Vertical Stabilizer to snap off.
In the case of Air France AF447 during the recorded rapid descent, passing through 4,000 feet, the Captain warned the PF to ease off the rudders. It is at this moment where the Vertical Stab is suspected to have snapped-off eliminating any chance of recovery.

Posted by: Rafael Sierra | October 30, 2011 11:48 PM    Report this comment

(Cont.) Not much is being said on this as Airbus has encouraged press releases to depict a flat impact on the ocean’s surface prompting a separation of the Rudder. This is wrong, the Rudder and Vertical Stabilizer were found floating attached in one piece, thirty kilometers away from the point of impact. The AF447 tail floating over open water leads to the same conclusion, a frail Vertical Stabilizer structural design one where even Airbus recommends not using rudder during stall recovery. There are just too many Airbus aircraft flying under the same risk, therefore, Airbus is targeting the crews and training procedures obscuring the flimsy aircraft problems. Airbus aircraft should be designed to fly safely and not made to disintegrate in flight, pilot competence is definitely just as important.

Posted by: Rafael Sierra | October 30, 2011 11:54 PM    Report this comment

the French book is written by Jean Pierre Otelli, who is an airshow pilot with some 14,000 hours and also aerobatic instructor. He is a prolific aviation writer, and the book is his 6th or 7th describing the prelude to accidents. He simply prints the cockpit voce recorder conversation, with an interlude of well formulated technical explanation what is going on. My view is that he is pretty much not biased, and sticks to the facts. The most glaring issue is that in spite of many dozen stall alarms, the stall or stall recovery never forms the subject of any communication on the flight deck. He draws no conclusion the way the NTSB would, but offers a combination of human factor issues which may be at the root of the accident.

Posted by: ROBERT ZIEGLER | October 31, 2011 8:02 AM    Report this comment

I think these speculations on mechanical integrity problems are no longer justified after the CVR and Flight Data Recorder are available. This was a perfectly functioning aircraft with engines putting out takeoff power (that had a frozen pitot temporarily at altitude), and was CFIT by holding it in constant stall from FL380 to ZERO feet after the A/P disconnected.

Posted by: ROBERT ZIEGLER | October 31, 2011 8:08 AM    Report this comment

And now Robert's Conclusions:
Expect that these type of accidents will INCREASE dramatically in the near future. Many have already happened. An industry that does not nurture, support and protect ... and VALUE the environment young pilots can learn in (General Aviation) will eventually fall on its nose. Boys flying checks at night in Minnesota winters in a so called "known ice" Aztec will be good airline pilots, especially if they also have a commercial glider rating and conventional gear time. Ab-inito trainees will not, even if they have been flying a Bonanza like an airliner for 100 hours. DIXIT.

Posted by: ROBERT ZIEGLER | October 31, 2011 8:17 AM    Report this comment

"The most glaring issue is that in spite of many dozen stall alarms, the stall or stall recovery never forms the subject of any communication on the flight deck."

Robert, the stall alarm is programmed to go OFF when below 100kias and ON again when over 100kias adding to the confusion. Many messages were presented and it seems like the crew could not "compute" the data torrent fast enough to recover.

Posted by: Rafael Sierra | October 31, 2011 11:35 AM    Report this comment

Rafael! Never heard before of thirty kilometers between floating rudder-and-vertcal-stab and point of impact This supposes that the rest of the plane was still flying after separation or the stab or both ! Interesting !

Posted by: philippe ducrocq | October 31, 2011 12:27 PM    Report this comment

"Many messages were presented and it seems like the crew could not "compute" the data torrent fast enough to recover."


Many messages presented perhaps, but missing was the single piece of information that would have allowed them to quickly recover their airplane -- angle-of-attack from a straight-forward, easy-to-read AOA indicator on the glare shield.

Posted by: Gary Dikkers | October 31, 2011 1:17 PM    Report this comment

Read about Airborne Express DC8 crash over Virgina, especially the analysis and CVR transcript. These guys went up to do stalls specifically on a dark night above an overcast. They stalled the aircraft as intended but could not recover.

Posted by: Stephen Ericksons | October 31, 2011 1:48 PM    Report this comment

Are you sure you want your airline pilots hand flying the aircraft? It makes me sick to my stomach thinking about it.
Upset recovery is great training in a simulator. Stall training in a simulator is worthless because you can't exceed the envelope as sim data is obtained only from flight tests inside the envelope and wind tunnel test data. A stall in the sim, particularly at high altitude, is not anywhere near what happens in the airplane. Approach to stall is controlled flight. Stall is uncontrolled.

Posted by: Stephen Ericksons | October 31, 2011 1:54 PM    Report this comment

"They stalled the aircraft as intended but could not recover."


And I bet they didn't have an AOA indicator either, right?

Posted by: Gary Dikkers | October 31, 2011 1:54 PM    Report this comment

Gary Dikkers, you are correct. AoA is Key info easy to "compute".

Posted by: Rafael Sierra | October 31, 2011 2:24 PM    Report this comment

Philippe ducrocq, the Brazilian Navy reported finding the Vertical Stabilizer approximately 30 kmtrs from the now estimated point of impact. I am speculating on scenarios in search of clues defining a mid-air break-up. Also, Post mortems carried out on 16 of the first 50 bodies found floating in the sea reportedly have traces of hypoxia supporting the theory of mid-air break-up. The report mentioned that the victim’s clothing appear to have been stripped off, apparently caused by the free fall from 35,000ft.

Posted by: Rafael Sierra | October 31, 2011 3:23 PM    Report this comment


Not sure how people were being stripped naked from 35,000' when CVR data is obtained from 4,000' AGL, unless they were doing something completly unrelated to the accident ;)

Posted by: Tommy Williams | October 31, 2011 6:17 PM    Report this comment

The Missing Link - I guess you can't put a link in a post for some reason.

The Airborne Express Stall Mishap NTSB Report is a worthy read.

Once more try....


If no link or luck.....google Airborne Express DC-8 NTSB.

Posted by: John Langenheim | October 31, 2011 7:07 PM    Report this comment

Tommy Williams, this is the report I'm referencing;

By Henry Samuel in Paris
1:00PM BST 15 Jun 2009
The post mortem reports from Brazil came as Air France announced that it had upgraded all speed probes on its long-haul A330 and A340 aircraft in the wake of the June 1 crash.

An investigation into the loss of AF 447 from Rio de Janeiro to Paris, in which an Airbus A330 jet plunged into the Atlantic with 228 people on board, has focused on the "inconsistent" readings from its airspeed sensors. Ice-clogged probes could have confused the plane's flight computers and led the pilots to fly too fast or too slowly into a storm.

Air France last week promised to replace its sensors within days. On Monday the main SNPL pilots' union said the carrier completed the upgrade of its 34-craft fleet over the weekend.

Post mortems carried out on 16 of the first 50 bodies found floating in the sea reportedly have no trace of burn marks or smoke, supporting the theory that the accident was not the result of a blast.

Their clothes had been stripped off, presumably in the rush of air as the plane fell from as high as 35,000ft.

Posted by: Rafael Sierra | October 31, 2011 7:18 PM    Report this comment

"Traces of hypoxia". I had not realised that low oxygenation was detectable. If however it can be detected, the symptoms of hyperventilation would be the same. It seems impossible that the full transcript from the flight deck should not have indicated cabin loss of pressure, without which noone is leaving the aircraft at 35,000ft.

Tommy's remark is a bit tasteless. We here are naturally concerned about what was happening on the flight deck, but their last four minutes must have been hideously scary for the passengers and crew in the back.

Posted by: R L S Butler | October 31, 2011 7:21 PM    Report this comment

"Their clothes had been stripped off, presumably in the rush of air as the plane fell from as high as 35,000ft."

There's a lot of presuming going on here, like. I like Tommy's presumption better.

Posted by: THOMAS M CONNOR | October 31, 2011 7:31 PM    Report this comment

What "rush of air" would the passengers be experiencing while inside the aircraft cabin? I didn't read anything about rapid decompression in the BEA's interim report (nor any post-mortems on the passengers recovered, though it did mention that was on-going). Clothes being stripped off due to a rush of water, though, might make some sense.

Posted by: Gary Baluha | October 31, 2011 7:42 PM    Report this comment

More thought provoking media reports;

"Typically, if you see intact bodies and multiple fractures — arm, leg, hip fractures — it's a good indicator of a midflight break up," said Frank Ciacco, a former forensic expert at the U.S. National Transportation Safety Board. "Especially if you're seeing large pieces of aircraft as well."
The large number of fractures observed during autopsies performed so far, and the relatively large pieces of wreckage recovered, are consistent with a mid air breakup of Flight 447 Airbus 330.

Posted by: Rafael Sierra | October 31, 2011 8:16 PM    Report this comment

I apologize for my comment, yes a bit tasteless. It's that aviation gallows humor, but still no excuse.

Posted by: Tommy Williams | October 31, 2011 8:30 PM    Report this comment

I do not think there is any argument that there is CVR recording all the way to impact. Media reports of inflight breakup are interesting and probably what we would still be assuming had we not found the CVR and DFDR. It was my initial assumption too. I was wrong and yet I believed I was so right.

Posted by: Tommy Williams | October 31, 2011 8:37 PM    Report this comment

I recognize that the BEA’s report based on data from the voice and flight recorders has helped understand events prior to AF447’s impact and that it has changed speculative thinking in our aviation circles and with the non-flying public. Although the BEA’s report is not complete, there appears to be confusion and disorganization in the cockpit, perhaps due to rapid flow of misunderstood system failures and unexpected extreme weather conditions. Air France crew training is in question and the manufacturer is disclaiming responsibility. Airbus and Air France’s reputation and legal liability is at stake, but so are the lives of more passengers and crews if this investigation is allowed to be influenced by the manufacturer’s interests allowing for a reoccurrence. Between the initial factors and the cause of the accident, say the pitot tube problem, the computer malfunction, bad ergonomics, the crew behavior and performance, there is also a threatening structural design deficiency. Thanks for letting me share.

Posted by: Rafael Sierra | November 1, 2011 6:01 AM    Report this comment

There seems to be a consensus here that these pilots were all buffoons and under-trained. I disagree.

Every airline pilot I've met during 50 years of flying (none for airlines) has had, what I judge to be, a higher than normal level of intelligence.

As for being under-trained, they were not trained as test pilots. I'll wager the Airbus had never been flown into that condition, deep stall at 38,000', even in clear skies with all instrumentation functional. Their conditions were way beyond the envelope.

AF may have been some training in a simulator, but the sim algorithms were not based on real data, only simulated projections of what the aircraft might do.

Posted by: Edd Weninger | November 1, 2011 11:25 AM    Report this comment

It will beinteresting to see the time-lines of the CVR and FDR aligned.

Posted by: Edd Weninger | November 1, 2011 11:27 AM    Report this comment

Read the Interim report no 3. It has just that at the very end of it.

Posted by: Gary Baluha | November 1, 2011 11:34 AM    Report this comment


They may not have been under-trained, but if their Airbus didn't have an AOA indicator they could use, they were certainly under-prepared.

Posted by: Gary Dikkers | November 1, 2011 11:50 AM    Report this comment

Edd, Google the following and you'll get to a pdf file with all you want except rudder input data fo some reason.

AF447 F-GZCP CVR/FDR/Narrative Timeline Combination based on BEA July 2011 Report


Posted by: Rafael Sierra | November 1, 2011 5:02 PM    Report this comment

This spreadsheet, AF447JulyReportVer13.PDF, gathers a partial history of the last minutes of AF447. I am not sure who arranged it. Comments by others in the internet mention that although it is a good analysis it is an incomplete form of presenting the real picture as it still lacks other needed forms of communication as the intonation of the pilots and visual clues like body movements. All combined would enhance the understanding of what was really going on then. A group of experts will be organized to study all human and technical interfacing factors involved in the accident.

Three BEA investigators specializing in human factors;
* A psychiatrist specializing in risk analysis;
* A human factors aviation consultant;
* A type-rated A330 pilot;
* An A330 test pilot

Posted by: Rafael Sierra | November 2, 2011 3:35 AM    Report this comment

Forget about Air France, nobody here can influence what happens with major airline ops. However I think the Cirrus is GA's Airbus. The growing number of technology assisted loss of control incidents and accidents and loss of control during landing and takeoff (ie when the pilot is hand flying) should be a wake up call for future flight training. SAFE has started the dialogue but there is a LOT of work and hard reflection required....

Posted by: DAVID GAGLIARDI | November 6, 2011 7:13 PM    Report this comment

This topic has generated a good discussion world wide and has enhanced a better understanding of the AF447 accident factors. Many have contributed techcnically and others, like I, have speculated based on historical information. My interest started after AA587 and continued fueled by similar accidents. There is a trail before AF447 that brings to mind the de Havilland DH 106 Comet accidents. I hope more true information is revealed in the next BEA report.

Posted by: Rafael Sierra | November 13, 2011 10:43 AM    Report this comment

Add your comments

Log In

You must be logged in to comment

Forgot password?


Enter your information below to begin your FREE registration