News French plane lost over Atlantic

[Urwumpe]

All flight recorder and voice recorder data is confirmed to be readable and downloaded!

http://www.spiegel.de/panorama/0,1518,762799,00.html

 

[Cairan]

Other article, in english:

http://www.flightglobal.com/article...-data-and-cockpit-voice-recorder-data-is.html

 

[Turbinator]

16 May 2011 briefing

Following operations to open, extract, clean and dry the memory cards from the flight recorders, BEA Safety Investigators were able to download the data over the weekend.

These operations were filmed and recorded in their entirety. This was done in the presence of two German investigators from BFU, an American investigator from NTSB, two British investigators from AAIB and two Brazilian investigators from CENIPA, as well as an officer from the French judicial police and a court expert.

These downloads gathered all of the data from the Flight Data recorder (FDR), as well as the whole recording of the last two hours of the flight from the Cockpit Voice Recorder (CVR).

In the framework of the safety investigation directed by the BEA, all of this data will now be subjected to detailed in-depth analysis.

This work will take several weeks, after which a further interim report will be written and then published during the summer.​

Recovery of airplane parts:
http://www.bea.aero/fr/enquetes/vol.af.447/videos/remontee.des.pieces.mp4
CVR Recovery video:
http://www.bea.aero/fr/enquetes/vol.af.447/videos/remontee.cvr.mp4
FDR Recovery video:
http://www.bea.aero/fr/enquetes/vol.af.447/videos/remontee.fdr.mp4

 

[N_Molson]

So in the big lines, there was an instrumentation failure that disengaged the autopilot, the aircraft climbed to 38,000 ft, then stalled and fell into the sea in 3 minutes 30 seconds. The commander was back in the cockpit during the last 2 minutes.

 

[Keatah]

So would it be correct, to say the plane's autopilot disengaged due to an electrical fault.. and the craft drifted up into into the coffin corner?

 

[N_Molson]

No, the fault came from the 3 Pitot probes, or the system that process their data to the flight computer & backup instruments.

When the autopilot "realizes" that the data received is incoherent, it disengages itself. So the plane climbed to the point it stalled, and at 38,000 feet a stall is violent and very difficult to recover, since the air density is less than the third of what it is at sea level.

So the plane probably went into a violent spin. Add to this that it happened at night, that all the calculations provided by the flight computer using the speed parameter stopped to work, and that the commander was taking a short break (which is a normal procedure), and you got all the ingredients for an inevitable disaster. :shrug:

Edit : the engines and the flight control system worked nominally, though after the loss of the velocity informations, the fly-by-wire system went into a "degraded" mode.

 

[Evil_Onyx]

"Accident to the Airbus A330-203
flight AF 447 on 1st June 2009
Update on Investigation"

http://www.bea.aero/fr/enquetes/vol.af.447/point.enquete.af447.27mai2011.en.pdf

 

[N_Molson]

Here's an article with more precise information, I'll try to translate it :

Rio-Paris : l'ombre d'une erreur de pilotage

Quatre minutes, vingt-trois secondes. Le drame du Rio-Paris, qui a causé la mort de 228 personnes le 1er juin 2009, s'est joué très vite, selon la note publiée, vendredi 27 mai, par le Bureau enquêtes accident (BEA).

Ce document, qui se contente d'établir des éléments factuels relevés par les boîtes noires et ne commente pas les décisions de l'équipage, suggère pourtant que celui-ci, confronté à des indications de vitesse erronées, a commis une erreur de pilotage. Le pilote aurait définitivement perdu le contrôle de l'appareil en le cabrant, c'est-à-dire en relevant le nez de l'Airbus pour tenter de le récupérer.

Entre le brusque désengagement du pilote automatique provoqué par la défaillance des sondes Pitot – à cause du givre accumulé lors du passage à l'intérieur d'une masse nuageuse – et l'arrêt des enregistrements des boîtes noires, tout s'est déroulé très rapidement.

A 2H08, l'Airbus A330 entreprend de dévier légèrement sa route sur sa gauche pour éviter un amas de nuages. Malgré tout, les turbulences augmentent et l'équipage décide de réduire la vitesse. Deux minutes plus tard, le pilote automatique se désengage car les indications de vitesses sont devenues incohérentes. Le chiffre qui apparaît côté gauche – celui du pilote aux commandes – ne correspond pas à celui qui s'affiche au centre du cockpit. De toute évidence, une partie des sondes Pitot ne fonctionnent plus et ne donnent plus d'indications fiables. L'incohérence entre les diverses vitesses affichées durera un peu moins d'une minute. A 2H10, l'alarme de décrochage retentit dans l'habitacle deux fois de suite. "On a perdu les vitesses, alors" lance le pilote.

Comment réagit-il ? D'après le BEA, "l'assiette de l'avion augmente progressivement au-delà de 10 degrés et il prend une trajectoire ascendante". A 2H10 et 51 secondes, le pilote "maintient son ordre à cabrer" autrement dit, il lève le nez de l'appareil. Simultanément, il appelle à plusieurs reprises le commandant de bord qui, en repos, n'est pas présent dans le cockpit.

Selon un expert, cette décision "est inexplicable". Selon lui il aurait été préférable "d'incliner l'avion, de baisser le nez, afin de le remettre à plat et reprendre de la portance".

A 2H11 et 40 secondes, le commandant de bord entre dans le poste de pilotage . "Toutes les vitesses enregistrées deviennent invalides et l'alarme de décrochage s'arrête". A 35 000 pieds d'altitude (un peu plus de 10 00 mètres), l'équipage se trouve aux commandes d'un appareil incontrôlable qui va monter jusqu'à 38 000 pieds avant de perdre irrémédiablement de l'altitude. "On n'a aucune indication qui soit valable", constate le commandant de bord.

A 2h12, pourtant, les vitesses redeviennent valides et l'alarme de décrochage se réactive mais il est trop tard. A 2h14 et 28 secondes, après une descente qui aura duré 3 minutes 30, l'appareil s'abîme dans l'Atlantique avec une assiette de 16,2 degrés et une vitesse de 198 kmh.

Jean-Michel Normand, www.lemonde.fr

"Four minutes and twenty-three seconds. The disaster that caused the loss of the Rio-Paris flight, the 1st of Juny 2009, went very quickly, according to a note published by the BEA (Inquiry & Analysis Office) Friday the 27th of May.

That note, that only publish the facts revealed by the flight recorders without any comments on the crew decisions however suggests that the pilots, facing wrong airspeed data, may have commited a piloting error. It looks like that the pilot lost control while pitching up, which means he tried to recover control by raising the nose of the aircraft.

At 2H08, the A330 Airbus flight tries to slightly deviate its course on its left to avoid a cloud formation. However, the conditions are deteriorating and the crew decides to reduce the airspeed. Two minutes later, the autopilot disengages itself because the airspeed data became incoherent. The figure that displays on the left - the first pilot seat - doesn't match the figure displayed on the cockpit's center. Obviously, some of the Pitot's probes are not functioning and don't give reliable airspeed indications. The incoherence between the various airspeeds displayed will last less than one minute. At 2H10, the stall alarm buzzes in the cockpit, two times. "So we've lost the airspeeds ?!" says the pilot.

How does he reacts ? According to the BEA, "the aircraft pitch rises slowly but steadily over +10 degrees and the plane takes an ascending trajectory". At 2H10m51s, the pilot "maintains his pitch up order", wich means differently said that he rises the nose of the aircraft. At the same time, he calls repeateadly the flight commander that, taking his rest, isn't in the cockpit.

According to an expert, that decision "is impossible to explain". According to him, it would have been better to "pitch down the aircraft, by lowering the nose, in order to put it on flat level and allow it to regain lift".

At 2H11m40s, the flight commander enters the cockpit. "All the recorded airspeeds become invalid and the stall alarm stops". At 35,000 feets of altitude (roughly a little more than 10 kilometers), the crew is piloting an uncontrollable aircraft that will climb up to 38,000 feet before beginning to lose altitude in an irreversible manner. "We have no valid indications", says the flight commander.

However, at 2H12, the airspeeds become valid again and the stall alarm reactivates. But it is too late. At 2H14m28s, after a descent that lasted 3 minutes 30 seconds, the aicraft sinks into the Atlantic with a pitch of 16.2 degrees [note : don't says if it is + or - pitch] and an airspeed of 198 km/h."

Edit : oh and on TV an expert said that the cause that explains why the Pitot probes stopped working for a moment is very probably icing.

 

[Urwumpe]

Yes, most analysts now say that the pilots likely flew accidentally into an "ice pocket", while trying to avoid the worst thunderstorms like all airliners tried that day.

The reaction of the copilots is really weird, but when the captain arrived at the scene, the plane was already out of control. I would say this reaction by the copilots has to be investigated.

My tiny bet here: The cause of the accident is similar to the only fatal X-15 accident: Caused by inadequate instrumentation.

Even if Airbus won't admit it.

 

[Cosmic Penguin]

Yes, most analysts now say that the pilots likely flew accidentally into an "ice pocket", while trying to avoid the worst thunderstorms like all airliners tried that day.

The reaction of the copilots is really weird, but when the captain arrived at the scene, the plane was already out of control. I would say this reaction by the copilots has to be investigated.

My tiny bet here: The cause of the accident is similar to the only fatal X-15 accident: Caused by inadequate instrumentation.

Even if Airbus won't admit it.

If this is true, then this is yet another fatal plane crash caused by confusion by the crew when the pitot tubes failed and caused weird discrepancies on the instrumentation. Unfortunately, even with fatal examples before, the crew just could not maintain control on the aircraft.

For previous examples, see these two crashes:

[ame="http://en.wikipedia.org/wiki/Birgenair_Flight_301"]Birgenair Flight 301 (Feb. 1996, B757)[/ame]
Aeroperú Flight 603 (Oct. 1996, B757)

Both crashes happened at night conditions, and the pitot tubes/static ports failed for various reasons, causing confusion among the pilots and ultimately led to the crash when the crew misjudged the situation. Yet this happened again.

 

[N_Molson]

What is weird is that they seem to have ignored the HSI. In that kind of 0-visibility situations, I thought it was the first thing pilots had to look at... And even if the MFDs shut down because of invalid data, there is a backup mechanical one. :idk:

 

[Cosmic Penguin]

What is weird is that they seem to have ignored the HSI. In that kind of 0-visibility situations, I thought it was the first thing pilots had to look at... And even if the MFDs shut down because of invalid data, there is a backup mechanical one. :idk:

Well, there's [ame="http://en.wikipedia.org/wiki/China_Airlines_Flight_006"]this[/ame] accident, in which the pilots on a Boeing 747 failed to notice that the plane was rolling to the right after one engine flamed out, and the plane literally "back-flipped" and was plunging vertically downwards (the plane descended 30000 ft in under two and a half minutes!). Luckily the pilot finally realized the plunge at 11000 ft and leveled out at 9600 ft. Still, the plane was seriously damaged under 5g+ forces, though not affecting flight characteristics. And that happened in broad daylight...

 

[Urwumpe]

The HSI should have worked, but still they failed to realize the tail-heavy trim of the aircraft in that situation. This is already more than just strange. They didn't lower the nose until the captain arrived, and that was too late, the plane was already in a stall by then.

I would say (if we would do a forum investigation of the accident, we are of course just arm chair engineers here), the following points need closer research:

1. HSI indications in the situation, especially after downmoding to alternate law. What did change and how did it affect the interpretation of the crew?
2. The pitch trim system - what happened after the downmode? In normal law, trim is automatic.
3. The standard procedures - newspapers reported the copilots had been trained by old standards, which had been dangerously wrong for modern airliners for such a situation. How does it look like and how should it be done better?

The words of the captain make me feel strange... he said "We have no valid indication", which is likely referring to the velocity... but what if the indication of attitude was also wrong? For example: How much can you trust the inertial HSI indicators if also altimeter and variometer are unusable?

 

[N_Molson]

The A330 has leading edge slats. If those went random after the downmoding to alternate law, the plane may have become uncontrollable, especially at high altitude.

 

[Urwumpe]

The A330 has leading edge slats. If those went random after the downmoding to alternate law, the plane may have become uncontrollable, especially at high altitude.

Yes, that as well... the leading edge slats behavior should be investigated as well... I am still busy with finding my own points in the A330 manuals.

 

[Keatah]

Shouldn't the 'craft have maintained a safe speed by setting the pitch to 5.2 degrees and a throttle of 85%? I thought that is standard procedure for maintaining the specified speed?

Anyways, the big picture is that the plane was precariously balanced by the autopilot. Now, all of a sudden you ask people to immediately jump into the loop and maintain that balance? Ain't gonna happen! Not with the precision and exactness needed at the moment. By the time the pilots fully absorbed what was going on, the flight was over.

A machine can acclimatize itself to the situation at hand in second, a human cannot! Especially pilots that are not correctly trained. Especially pilots that are forced to go directly from a systems-monitoring "mode" to a flying mode. It ain't gonna happen!

 

[Urwumpe]

I would say the lack of automatic trim (if it happened at all) isn't the problem - it is the crew not noticing that they are stalling. The trim should be approximately right, when the normal law kicks out, but requires the crew to watch it from that alert on.

 

[Keatah]

After following the discussion here at airliners.net
http://www.airliners.net/aviation-forums/general_aviation/read.main/5219823/1/#1
I come to the simple conclusion that modern aircraft are becoming too complex. It would seem that in the name of efficiency and pushing performance to the absolute edge we've come to rely on systems that are needlessly complex

The problem, as I see it, is when these systems go out or have input failure from sensors - the whole autopilot gives up and cuts out. Immediately throwing the pilot back into the loop. By the time the pilot figures out what's going on the plane is 6ft under.

Can't Otto just say "hey dude we've got a problem. I'm going to keep doing what I was doing for a little while longer till you get back in the loop. But hey it's all good, if you want all control this instant just hit the red button!"

Instead Otto closes up shop leaving a pilot that's not 100% situational aware!

To me it seems that the complexity of this Airbus was a contributor as well as a stall warning that turns itself off if speed is less than 60 knots. But the craft is still in a full blown stall. Not only that, the plane tries to stabilize itself and prevent a wing-over which would greatly aid stall recovery and indicate to the pilot what's going on.



Now, back to my pitot tube rant from earlier on. Couldn't we have some sort of particulate counter? Something that spits out particles and tracks the airflow? Or something that ionizes the air and sees how long it takes to flow back a certain distance?

Why don't they do just like the glider pilots do and tie a string to the front and use that to help gauge air-flow direction?

Or why not a trap-door where if the airflow is too slow, the spring pushes it open.

What about SAR or Doppler radar aimed downward to see what the plane is doing? Or put 4 gps receivers in the wingtips and fore/aft positions. That way we can tell pitch and roll and yaw and other attitudes and trends. It might not be the most accurate. But surely we could get 2 or 3 degrees out of it. There'd need to be some slick software, but hey that's the $150,000/year jobs are for.

Or put a smoke generator (or chemical dye dispenser) in the nose and light it off, and with a high-speed strobe and digital camera you can take pictures of the smoke blasting past you. You put 8 of these in circumference around the fuselage or some other place (but close to the generator) and through image processing you can tell attitude, AOA, TAS, and probably the weather forecast for Hawaii through next month.

It's a like a bring-your-own-windtunnel-with-you. This would be a backup system, of course. And it's accuracy would be unmatched against the standard PFD instruments. You could put these EADU's anywhere once they are miniaturized enough. And all feeding into a central system that can graphically show the plane in relation the moving airflow, complete with speeds, G's, AOA, TAS, Everything except heading, you pull that from the GPS. This would most definitely appeal to the Nintendo-era of pilots we have today.

Or keep it simple and frakking derive a rough estimate of airspeed from the RAT.



I'm of the school where the autopilot should gradually drop out of the loop and while doing so, it continues to do what it was doing based entirely on IRU. If we make Otto smart enough he can follow the past 5 minutes trend or something and rely solely on IRU/INS for a few more moments. Long enough to let the pilots get back into the loop! Especially on Airbus, with all those confusing laws and stuff -- http://www.airbusdriver.net/airbus_fltlaws.htm -- can anybody make sense of that? And do it in 10 seconds while riding a crashing plane? No wonder why there are so many pilot errors on Airbus!

This whole "Setup" reminds me of WINDOWS-ME, a half-assed hybrid of dos and windows and drivers all over the place, with no genuine automatic maintenance and self-healing properties. Way too much technical info for the casual pilot.

They have this set up so the plane keeps the pilot in-line and behaving, all the while trying to second-guess what he's doing. Are pilots that incompetent today? All this electronic stuff seems to increase workload. Do pilots need all this because they are not capable of flying the craft?

What ever happened to the simple autopilot that holds FL350, and `277 ?? It either works or it doesn't! Good GOD!!!!

 

[Cairan]

I absolutely agree with you Keetah about the problematic behaviour of autopilots regarding auto-disconnect. Back in the day when "Otto" as you call it was not quite trusted and was monitored most of the time and limited to cruise duty, a disconnect was ok to deal with as the pilot-flying was monitoring it and ready to jump back, because he basically acted as supervisor to the autopilot. However, with the complexity of modern integrated autopilots, which can take over from takeoff rotation to landing roll, there should be a "graceful" disconnect where the autopilot doesn't just drop the stick without forewarning.

When I was flying with my instructor, it really got into my head to wait for his confirmation of having control of the plane when I wanted to hand it to him. Simple and effective:

Student: You got the airplane? Instructor: I got the airplane.

But as Keetah highlights, with the AP it's more like:

AP: Oh s--t! You got the plane!!! [autopilot disconnect horn] Pilot: [...] ehm , what?!!

It should be the same for airliners when sensors drop off:

AP: You got the airplane? [caution chime] Pilot: [hits AP disconnect switch] Yeah, I got the plane Otto.

 

[Keatah]

One other quickie: Why can't they have a like a hot-air compressor or something to blast out any obstructions or melt excessive ice. Sorta like blow'n your nose. This would be powerful enough to remove those remove-before-flight covers, AND it would most definitely blast out any bees'n'bugs! Cycling this would take 10 seconds. You close off the sensitive sensor with a valve (made in triplicate, with mechanical pilot hand-powered override). Then you either use bleed air from the compressor or set off a small gunpowder charge to clear the way. The explosive charge is small and in a cartridge and can be set up so the pilot has to manually drop it into the firehole. Then you re-open the valves and you now have a clear pitot tube ready to work! I'm sure there can be variations, like pumping hot oil through it or just hook it to the pilot's butthole and feed him a burrito with cream soup. That will not only clear out the stuffed tube, but all aircraft in the vicinity will benefit from the cleansing action!

Speaking of that method. You could go with another mechanical means; just have a rotor-rooter toilet cleaner setup type deal. A motor spins a wire that gets pushed through the tube. This hot wire would burn and drill it's way through and built-up crap.

Or make a rotating dual-pitot assembly. If it gets jammed up, then the pilot cranks it 180 and gets a fresh tube to play with.. Why not?

OR how about a mini-anemometer. Or hotwire (altitude and temperature compensated)..

And then there is the weight on a string, at zero speeds it hangs down. And at higher and higher speeds it swings more horizontal. Just affix this to the pilot's side window. Or along one of the supports. Make it deployable in emergency.

And finally you can make a waldo-glove to where the pilot can stick his hand into the airstream (at altitude) and feel how the wind is moving. Or a tiny hole where you and stick a stick through and gauge the speed from the resulting force needed to push the stick. Or you can just make a hose that sticks out the side and pump it up, the faster the plane flies, the more pressure is needed to overcome the sheering force and make it stand out straight.

What about a flexible pitot that can turn itself inside out? Something like sticking your fingers in those under-inflated animal balloons? Perhaps a pitot tube than can shed or dissolve its inner surface. Or spray itself with antifreeze compound..?

Or put fans on the wheels and lower the nosegear. The faster the plane is moving, the faster the wheel spins. This is so simple, it's beyond belief!! Modulate the brake pressure and measure it to see how much force is trying spin the tires. A spin-off (no pun intended), lower the engine speeds to idle and see how well they freewheel. I know *that* wouldn't be a safe procedure in times of emergency. But would it work?

And something novel - why not mount micro parachutes on strings, put on at the tail and one on each wingtip? Connect them to springs and compute the speed and angle of attack by how much these micro-chutes are pulling. And you can get precise direction by looking at how much off-center they pull. If they are 90` up then you are in a pancake fall. If they are centered and trailing exactly behind the plane then you're flying level. You measure their orientation by hooking them to a joystick-like mechanism. And it would be like playing a videogame with string on your joystick, you whip it around and pull it this way and that way. All indicating airflow force and direction.

As a bonus you could make them glow-in-the-dark to see them better or use blacklight, or a webcam type camera. You make these chutes tiny, like 3 inches across, or larger like a meter. It all depends.

Or better yet!!! Get a load of this. A frakking windsock! Just blow one out the tailcone or make a periscope type config where the pilot pushes it up through the roof. Make it configurable and pump it full of that liquid metal so it gets heavy and sags a little while at speed.

Another system, you could have an explosive tube pop out of the nose like an air-data boom, but just for airspeed and direction relative to the airstream. The Pinocchio Pitot! I'll get you the true airspeed!

Pilot disoriented? What about those cheap night vision goggles you get from fleabay? This would immediately solve up/down/bank orientation issues. The sky is black, the ground is white.. Why has this not been done? We don't need precision, we just need a rough confirmation. Are we going up or down? Banking?? Yah..


Or just simply have enough tubes in use to reduce the statistics and chances of failure to near zero. They use enough redundancy in the PFC and ACE computers, why not have the same level of redundancy in the sensors providing information to those boxes?

Yah, all these things can be done as backup, and we rely on a stupid straw with NO backup.
With all these ideas there is to be no excuse for stuffed-up pitot tubes and failed airspeed readings.