Air France 447 - On topic only!

[ca787546]

What I find strange is not the fact they haven't released the whole transcript, but why they selected to release just those few particular quotes uttered by the pilots out of the whole critical 4 minutes. For example whatever instructions the senior captain gave after he returned to the cockpit would be quite important; why was nothing he said mentioned?

And although Air France's response isn't all unexpected, it's a little off-putting that they would circle the wagons so soon, before all the information is made public. Are they worried about something that might come out when all the truth is known?

BTW here's a pretty good timeline of the events:

http://www.spiegel.de/international/world/bild-765764-220365.html

According to this, the senior captain entered the cabin just after the stall occurred and according to the article, he gave instructions to push the nose down. However, the co-pilot didn't do so until 2:12:17, almost 30 secs later. What happened in that half minute between 2:11:40 and 2:12:17? And why was nothing in the CVR transcript released from 2:12:40 to 2:13:32?

I'm sorry, I missed the part where the captain gave instructions to push the nose down. What Spiegel has published here is basically the exact same thing that the BEA note says, minus several points of information and in a graphical form (based on the link you put in your post). Nothing new, au contraire, they dumb it down a bit and in the process miss some important information already released by the BEA. I think that link expresses very clearly why the BEA wouldn't release all the information prematurely but on the contrary be careful with what to release before all investigations have been concluded. If a serious and respectful magazine such as Spiegel can miss details when creating a quick picture of the accident, imagine what other less reputable parts of the media industry would do. It doesn't seem at all strange to me. Rarely do we see a full transcript of the CVR released as soon as it is available, and even more rare is the release of all the FDR recorder as soon as it is available.
I believe once they have completed the full analysis of all the CVR and FDR data, a detailed accident report, or a preliminary report will be released. Once the full accident report has been released, then the relevant information from the CVR and DFR will be released to manufacturers and operators, and eventually down to the people on the media.

[EconomyClass]

A stall warning occurs before the actual stall. The system can prevent the pilot from pulling up further and stalling the plane, and it can add thrust at that point as well, so it doesn't need to push the nose down. This is how the Airbus works in Normal Law. It is still possible to briefly stall it with aggressive pitch and/or rapidly falling airspeed, but it will promply return to an AoA beneath stall. The problem is that systems like this need airspeed data to safely make these decisions. Without airspeed data, you need a human to decide, even in a panic situation.

You lost me. It happens before a stall, the precise time when action is necessary to keep the stall from happening. Which is when something has to change thrust and angle of attack. So there's no way to force those things to happen? I just read a story in a German publication that said the Air France manual said to increase thrust. But supposedly when Marc Dubois entered the cockpit, he said to decrease thrust. So their written procedures must be wrong since they did what the instructions said and crashed the plane.

There's a more general issue here. I was thinking about past fatal incidents where pilots attempted inputs to planes' controls and the planes did the opposite. At least two 737 crashes and an MD80 crash in the Pacific by an Alaskan Airline plane. Does the data recorder record the input, or does it document both the input and the behavior of the component to which the input is directed. Because the article I read said the horizontal stabilizier was at 13 degrees, which it said would keep the nose up no matter what the pilots did. It raised the possibility that the system which accepts the input and directs the mechanism had a bug in it. Not only that,it says someone noted before the crash the same anomaly in a simulator.

[ca787546]

You lost me. It happens before a stall, the precise time when action is necessary to keep the stall from happening. Which is when something has to change thrust and angle of attack. So there's no way to force those things to happen? I just read a story in a German publication that said the Air France manual said to increase thrust. But supposedly when Marc Dubois entered the cockpit, he said to decrease thrust. So their written procedures must be wrong since they did what the instructions said and crashed the plane.

There's a more general issue here. I was thinking about past fatal incidents where pilots attempted inputs to planes' controls and the planes did the opposite. At least two 737 crashes and an MD80 crash in the Pacific by an Alaskan Airline plane. Does the data recorder record the input, or does it document both the input and the behavior of the component to which the input is directed. Because the article I read said the horizontal stabilizier was at 13 degrees, which it said would keep the nose up no matter what the pilots did. It raised the possibility that the system which accepts the input and directs the mechanism had a bug in it. Not only that,it says someone noted before the crash the same anomaly in a simulator.

Ok, let's talk about some basic aerodynamics. There are different ways one can change the pitch of an aircraft in flight. Depending on the location of the engines, and increase or decrease of power will likely change the pitch. In the A330, an increase in thrust would generate a pitch-up, and a decrease would generate a pitch-down. Then we have the flaps and slats. Flaps specially can create a pitch down when extended, and a pitch up when retracted. Then we have the center of gravity. Moving the center of gravity backwards (say, transfer weight, such as fuel) can create a pitch up, and moving it forward a pitch down. Then we have the spoilers on top of the wings, usually extending them will create a pitch down, and retracting them a pitch up. Then we have the THS (trimmable horizontal stabilizator-the one that ended up at 13 degrees), obviously deflecting this surface up would create a pitch up, and deflecting it down, would create a pitch down. And finally, we have the most known elevator. The elevator is what you most commonly use to control the pitch of the aircraft, and that's what is directly controlled by the sidestick inputs. I don't know how many of you have flown an aircraft, but the way you usually fly and aircraft and trim it for cruise is to level the aircraft using the elevators, and after the aircraft is level you "trim out" the pressure on the controls. How do you do that, well, on the 330, you would move the THS until the there is no need to maintain any pressure on the control to maintain the pitch, in a way that if you were to let go of the controls, the aircraft wouldn't pitch up or down. In the FBW airbus, the THS is normally controlled by the flight computers, by this relieving the pilot from having to constantly trim out the aircraft. What you get is the following results; in a normal aircraft, if I point the nose up to 10 degrees, I would have to apply backward pressure on the controls, then use the trim to take away the pressure, and then the aircraft would maintain 10 degrees nose up, same if I was to put 10 degrees nose down; in the FBW aircraft, the pilot pulls back on the sidestick to bring the nose up to 10 degrees, and let go of the controls, and the computers automatically trim out the aircraft, so when you let go of the controls the aircraft remains at 10 degrees nose up. This autotrim feature is only available in Normal and Alternate control laws, when you degrade to direct law, or mechanical backup, trim is at the control of the pilot. The aircraft under UAS is likely in alternate law, therefore the autotrim function is active.

[Fear_of_Flying]

I'm sorry, I missed the part where the captain gave instructions to push the nose down. What Spiegel has published here is basically the exact same thing that the BEA note says, minus several points of information and in a graphical form (based on the link you put in your post). Nothing new, au contraire, they dumb it down a bit and in the process miss some important information already released by the BEA. I think that link expresses very clearly why the BEA wouldn't release all the information prematurely but on the contrary be careful with what to release before all investigations have been concluded. If a serious and respectful magazine such as Spiegel can miss details when creating a quick picture of the accident, imagine what other less reputable parts of the media industry would do. It doesn't seem at all strange to me. Rarely do we see a full transcript of the CVR released as soon as it is available, and even more rare is the release of all the FDR recorder as soon as it is available.
I believe once they have completed the full analysis of all the CVR and FDR data, a detailed accident report, or a preliminary report will be released. Once the full accident report has been released, then the relevant information from the CVR and DFR will be released to manufacturers and operators, and eventually down to the people on the media.

Hi, click "back to article" at the top of the page opa-opa linked - I think it's actually pretty interesting.

[ca787546]

Hi, click "back to article" at the top of the page opa-opa linked - I think it's actually pretty interesting.

Thanks FearOfFlying, I read the article. It's a little embellished, but I guess it has to be since it isn't aimed to be technical. It does raise some interesting points. I wonder who are these sources close to the investigation that are mentioned though. It is very likely the captain indeed notice the stall situation once he came in. You have to understand that the people at the BEA (along with colleagues from the NTSB, BFU, CENIPA, and AAIB) likely have a movie-like animation of the accident along with the cockpit voice recordings playing in sync, and are analyzing this information very carefully, along with any wreckage recovered that may give further hints to the causes and factors of this accident. We are basically (more or less like the rest of the media) grabbing the bits of information generally available and drawing conclusions that may or may not be right.

[tone]

For convenience here are the links.

Part #1 - http://www.spiegel.de/international/...765764,00.html

Part #2 - http://www.spiegel.de/international/...5764-2,00.html

Very interesting analysis on Part #2.

Could the horizontal stabilizer being kept at 13º throughout the events eventually being the final factor that did not allow the recovery on the final moments, just after reaching FL100, after the captain entered the cockpit?

Is this a possible software bug or just the result of a forgotten detail under an extremely stressful, near panic situation?

[ca787546]

At 10000' with such a high angle of attack, such low airspeed, such high rate of descent, and at such a high nose-up attitude, I think it was very difficult to have recovered. I think if you were to recreate such conditions in the sim, most pilots would find it very hard to recover in time, and if they did, they would come very very close to the ground. You would have to push the nose hard until the Angle of Attack decreased, and start applying thrust as needed to regain energy, but the loss of altitude required to recover may very well exceed the 10000 feet. You're talking of over 180 tons of mass moving downwards, with not much airflow over the elevators or stabilizator making them quite noneffective.

[Evan]

By the way, the FCOM and FCTM do state CLEARLY that if the safety of the flight is impacted one should follow the memory items for UAS... if the aircraft is in level flight and clear of terrain I would not follow them, as the safety of the flight is not impacted.

Man, I have to to tell you, if this is still a common perception among Airbus drivers, I am very troubled to hear it.

WHile I understand the directive "if the safety of the flight is impacted one should follow the memory items for UAS", The FCOM and FCTM I have say nothing about terrain and gives no examples of when it is better not to use the memory items. For this reason it is ambiguous and therefore flawed and should be listed as a contributing factor in this crash.

If I was heavy and flying at 41000 feet in the A330, and my airspeed drops to 60 knots, I am going to try to not touch anything unless required if the aircraft is stable.

Agreed, but then the question is what the hell are you doing at FL410 at 205t? They were nowhere near that altitude.

When the Autothrust is disconnected without the pilot's command, it goes into thrust-lock mode. The thrust levers remain in their position, but looking at the EPR or N1 indications on the Engine/Warning display will quickly tell you the thrust they are locked at and you may move the Thrust Levers to match the TLA indication to that of the thrust being produced.

True, but irrelevant because in this type of 'panic' situation, usually accompanied by turbulence and always accompanied by aural and ECAM distractions, can we expect the crew to read the LCD engine data, identify an N1 reduction trend or register the tiny TLA indications? I think not. Therefore, with no instant, obvious reference to TLA, you can't expect the crew to verify the actual thrust setting in the first vital moments of the incident.

Imagine the case stated above, where the aircraft is heavy and close to its maximum ceiling, with protections lost due to a loss of airspeed indications. I pull gently on the sidestick but end up increasing the nose slightly more than I needed, as I try to correct this, I increase the thrust to climb power, which creates a further pitch-up momentum. Now my nose is at 7.5 or 10 degrees nose up.

Allow me to modify the scenario: You are in autothrust, THR CLB mode, cruising at .82M. N1 is around 90%. Entering turbulence, your speed target is commanded back to .80M. A/THR reduces N1 to around 70% to achieve the target speed. Then —BAM— ADR disagree. A/THR disconnects and thrust lock freezes N1 at 70%. A/P disconnects but auto pitch trim holds the level FPA, as the a/c continues to slow beneath .80M, by raising the pitch above your current 3°. This is not a static situation. There are two trends in effect. The speed is slowing and the pitch is increasing. If you do nothing what will happen?

Now what if you use the memory items? You get the thrust levers out of THR LK and back to CL. You plant the pitch at 5°. You have increased N1 from about 70% to about 95%. At FL350, how much thrust effect are you really going to get in pitch? How hard is it to very quickly correct that back to 5°? Now you are flying at a thrust/drag ratio that will neither stall nor overspeed for at least the time it will take to go to the QRH and set more exacting commands. I don't see how you can lose.

On the other hand, in opting for stasis (leaving things as they are), I see very clearly how you can get into a high AoA situation without touching a thing.

[guamainiac]

Evan watch that statement. They were almost at that altitude and in warm air. Warmer air than expected.

Factor for the two. Remember ... it is relative.

Dial in what would be density altitude.

[EconomyClass]

I'm curious about one thing. In simulation, how do they simulate the behavior of plane components as they would be in the specific meteorological situation this plane encountered? There isn't any real air passing over real wings. It isn't minus 50 degrees or whatever it is at 35000 feet. I'm remembering that the 737 crashes were because of what happened to a tail control at very cold temperatures. They finally simulated that in an accident lab with liquid nitrogen (as I recall). That is how they finally discovered that Boeing had built a servo that did the opposite of what it commanded. And the overwhelming majority of those planes did not have that malfunction. It took years for them to finally come to a clear diagnosis. So how is it people here, so shortly after any real evidence appears, can conclude the Airbus did what the pilots told them to do? I'll bet Boeing said the 737 was safe back then, too, right?

[Evan]

Because the article I read said the horizontal stabilizier was at 13 degrees, which it said would keep the nose up no matter what the pilots did. It raised the possibility that the system which accepts the input and directs the mechanism had a bug in it.

The trimmable horizontal stabilizer (THS) can ALWAYS be controlled mechanically (even during total power loss) by a wheel on either side of the throttle quadrant. This wheel has priority over any computer inputs and, when moved, cancels the computer trim for the remainder of the flight. The horizontal stabilizer was therefore controllable at all times (unless there was a dual hydraulic servo failure, which is extremely unlikely, and there is no indication of that).

I don't know how many of you have flown an aircraft, but the way you usually fly and aircraft and trim it for cruise is to level the aircraft using the elevators, and after the aircraft is level you "trim out" the pressure on the controls. How do you do that, well, on the 330, you would move the THS until the there is no need to maintain any pressure on the control to maintain the pitch, in a way that if you were to let go of the controls, the aircraft wouldn't pitch up or down.

It is important to note that at no time in normal flight do you move the THS to trim out elevator control force on the A330. The only way to do so is to move the THS trim wheel, which defeats autotrim for the remainder of the flight.

In the FBW airbus, the THS is normally controlled by the flight computers, by this relieving the pilot from having to constantly trim out the aircraft. What you get is the following results; in a normal aircraft, if I point the nose up to 10 degrees, I would have to apply backward pressure on the controls, then use the trim to take away the pressure, and then the aircraft would maintain 10 degrees nose up, same if I was to put 10 degrees nose down; in the FBW aircraft, the pilot pulls back on the sidestick to bring the nose up to 10 degrees, and let go of the controls, and the computers automatically trim out the aircraft, so when you let go of the controls the aircraft remains at 10 degrees nose up. This autotrim feature is only available in Normal and Alternate control laws, when you degrade to direct law, or mechanical backup, trim is at the control of the pilot. The aircraft under UAS is likely in alternate law, therefore the autotrim function is active.

Duly corrected, but also very relevant to this incident is the fact that the auto pitch trim will also be frozen during 'abnormal attitude' excursions.

[guamainiac]

Is there any significance regarding what appears to be a change in direction during the final moments?

It appears they ... I am reluctant to use the word but I will ... spiral and they were past 180 degrees from the original heading and in a turn to the right.

[Evan]

Huttig pointed out that Airbus published a detailed explanation of the correct behavior in the event of a stall in the January issue of its internal safety magazine. "And there, all of a sudden, they mention manually trimming the stabilizers," he says.

I have never seen an Airbus A330 memory item or checklist procedure for stall, but electrical pitch trim is frozen at 30° AoA and/or <60kts, so if manual trim is not clearly mentioned, we have some clear Airbus liability.

This is the one I have for the A320:

As soon as any stall indication (could be aural warning, buffet...) is recognized, apply the immediate actions :
- NOSE DOWN PITCH CONTROL . . . . . . . . . . . . . . . . . . . . APPLY
This will reduce angle of attack
Note: In case of lack of pitch down authority, reducing thrust may be necessary
- BANK . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . WINGS LEVEL

• When out of stall (no longer stall indications) :
- THRUST . . . . . . . . . . . INCREASE SMOOTHLY AS NEEDED
Note: In case of one engine inoperative, progressively compensate the thrust asymmetry with rudder
- SPEEDBRAKES . . . . . . . . . . . . . . . . . . . CHECK RETRACTED
- FLIGHT PATH . . . . . . . . . . . . . . . . . . . RECOVER SMOOTHLY
• If in clean configuration and below 20 000 feet :
- FLAP 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . SELECT
Note: If a risk of ground contact exists, once clearly out of stall (no longer stall indications), establish smoothly a positive climb gradient.

No mention of THS pilot actions.

[ca787546]

Evan, have you flown airbus in the FBW family as well, or have flown FBW aircraft? I have about 5000+ hours on the A319, A320, and A330-200/-300 combined. I am aware that the THS is not controlled by the pilot in normal law, I was simply explaining it layman terms since some of the people here have zero flying experience while others have several thousand hours flight time, or aviation engineering experience of some kind. I have to say, I have encountered Unreliable Indication situations in the A330 in climb, and I have accounts from several colleagues that have encountered it in cruise at heavy weights around thunderstorms. The common denominators in all these UAS cases was that although they were short-lived (from a few seconds to less than a minute), taking the time to examine the situation (wait a minute, we know we're NOT doing 60 knots on a climb, or in cruise all of a sudden!), evaluate the factors (altitude, speed, thrust condition, weather conditions, attitude) and not simply react like a robot (power to climb, 5 degrees nose up! right away!) ended up in a successful outcome.
Also, you must be aware of the flying qualities of modern jets when near their maximum performance ceiling, I didn't ever say I was at 41000 feet with 205T, you added that weight out of your own imagination. I have been at 41000+ feet in the A330, and the aircraft at such altitudes has to be handled very carefully, just like any other aircraft near it's performance ceiling. We know the Critical AOA decreases as mach increases, and therefore at such high altitudes handling must be performed very careful.
I recommend you read the following information, go to the stall section. I just finished reading it a few minutes ago, it's good reading.

http://www.ukfsc.co.uk/files/Safety%20Briefings%20_%20Presentations/Airbus%20Safety%20First%20Mag%20-January%202011.pdf

[Deadstick]

ca787546,

Thank you so much for your comments. It's enlightening to hear from someone that actually rides in the front, and the same to MCM and the other professionals that contribute.