'At 2 h 06 min 04, the PF called the cabin crew, telling them that"in two minutes we should enter an area where it’ll move about a bit more than at the moment, you should watch out". Do we assume that the seatbelt sign was on, and the passengers were told to take their seats? Was the going so rough that the FA's could not be seated? (Okay maybe some were sleeping). With about 5 minutes to go before v/s became negative, I would have thought that they could have been belted in. But with so many passengers found unseated, I wonder if they were able to due to the aircraft's movements. What a way to go, poor souls!

Uncloggable Airspeed Sensor
 

In my boyhood I was familiar with a number of biplanes where the airspeed sensor was a flat plate on a coil spring mounted against a scale on the interplane struts. However icing would change the drag coefficient and might even lock up the spring;)

Any blade antenna produces drag proportionate to airspeed that a force transducer could convert into knots. Whether the antenna people provide ice resistant coatings or proper heating elements is another question, but at least such a sensor would not be vulnerable to clogging. To be useful such a sensor would need to provide usable indications at altitude where the available speed range can be quite narrow.

promani, (and others) this foaming at the mouth to find somebody to blame, particularly a deceased somebody, is unseemly. It is also most definitely not BEA's charter. Their charter is to make aviation safer through impartial investigations of accidents.

It does not good to assign fault to say, the pilot, and wash your hands of the entire event leaving it to the court to finish stomping on a corpse to help some people obtain an emotional release (and financial killing.) The BEA must figure out WHY something failed and correct that situation.

Let's use some fanciful names and events to remove some of the emotion for a moment. Please bear with me here to get a notion of what I mean. We're not here to assign blame. We're here to make aviation safer.

Suppose we have a floobie. A floobie pingles widgets. This is a very complex floobie. It requires a rangant for it to function. At any given time it typically has hundreds of users per floobie. It has failed killing many users and the available data indicates the rangant "obviously" did something stupid.

Is this enough to make using floobies safer? It may satisfy some blood lust. "Let's take that rangant out, slap it around, and hang it from the yard arm even if it's already broken beyond repair."

It does not satisfy the need for safer use of floobies. We must figure out why the floobie's rangant twisted its left plooker button. Did the rangant get spurious inputs? Was the rangant's programming (training) incorrect or in some way inadequate? Was the rangant overloaded with conflicting data?

What actions can take place to make floobies safer to use? Some here seem to be suggesting ripping out all the telserboxes from the floobie and make the rangant do things the old way, fessling the grommetvents directly. There is a small point here if one neglects documented safety records that state usually floobies that use telserboxes all seem to pingle along more safely than when the rangant fessles grommetvents directly.

So it seems we must make sure the rangant does not mess up again under the particularly unusual conditions that contributed to its incorrectly twisting the left plooker button. What can be done to "clean up" the data input to this poor overloaded rangant when it is faced with a cascade of failures such as observed? Could better programming (training) of the rangant for real world conditions help? When their programming is tested they already know the test. "We start nearly feeblevetzed and you have to recover from a complete feeblevetz." What does the Rangant do when it's "surprised" (its programming had not prepared ot to recognize this) by feeblevetzing conditions?

It seems to me this latter objective is more important than the emotional release obtained by stomping that already dead defective rangant into itty bitty pieces in a fit of judicial temper.

And, yes, this has a deliberate resemblance to the case at hand here. Somebody determined enough could plot it out. I'm seeing shouts of "kill (blame) the pilot!" I don't see realistic searches for methods to repair the situation? I'm not even seeing many efforts to determine why a rangant might exhibit anomalous behavior when faced by the stimuli set we know, perhaps with the inclusion of some stimuli we don't know.

Machinbird

I had envisioned a tube similar to pitot, and in the airstream, but blunt, to forestall abrupt icing. I also think to alleviate unwanted angle sensing, the "bulb" could have "Cells" sensed through a computer that selects the group closest to stagnation point. Not a panacea, but perhaps a bit more ice resistant, and alleviating Drain hole aberrations, etc.

Sat in your old office once upon a time, now there's a foxhole!

JD-EE: 'Twas brillig .... excellent analogy.

JD-EE:

Great Post !!!

Flush Airdata Sensing (FADS) System
 

@Machinebird

You might find this of interest:
http://www.nasa.gov/centers/dryden/p...ain_H-2379.pdf

http://www.spaceagecontrol.com/nasa-tm-104316.pdf

kilomikedelta, henra - thanks guys. I was biting my fingernails over that one.

HN39, thanks for the data.
I took the liberty to change a bit the presentation just to help for the clarity.
Hope you don't mind.


http://i25.servimg.com/u/f25/11/75/17/84/af447_12.png

The PF commanded nose up in manual flight, alternate, law. The aircraft was just attempting to trim the aircraft to comply with pilot input. Quite normal.

No offense, but somehow making light of the inadequacies that may have caused so many to perish, I can think only of a possible reason why.

The Art is not the best, but it is the best, all things considered.

respect

Pitot tube
 

"Use of a gas behind the membrane would not work because of PV=RT."

Mariotte's law is not a problem: V, R are fixed values. T is measured, so P can be computed.

HN39

Thanks for the data set.

JD-EE

Well put. "If it ain't broke don't fix it" could be substituted for, "If it ain't fixed, it'll be broke". Your little rant was fair and to the point, but you are well aware that innovative and practical solutions are the bread and butter of this thread. Just a pity that at this juncture there is too much butter on the bread.;)

This post is to retract my earlier assessment that the initial response from PF may have been wrong. After spending a couple of days studying the report, I'm not sure anymore. It is unfortunate that BEA has released so little factual information. There is very little to go by. I'll just post some questions I scribbled in the margins of the report here before taking leave and waiting for the July report.

From 2 h 10 min 05 (4), the autopilot then auto-thrust disengaged and the PF said "I have the controls". The airplane began to roll to the right and the PF made a left nose-up input.

Why nose-up. Was this required to arrest the roll to the right?

The PF made nose-down control inputs and alternately left and right roll inputs. The vertical speed, which had reached 7,000 ft/min, dropped to 700 ft/min and the roll varied between 12 degrees right and 10 degrees left.

Propensity to roll right again?

The airplane was subject to roll oscillations that sometimes reached 40 degrees.

Were these oscillations symmetric? These oscillations correspond to a sharp turn to the right on the track shown by BEA in figure. Again reasonable to assume a/c wanted to roll right? Why?

Edit: remove nose-up comment.

The a/c seems to show a propensity to roll right from AP disconnect to end of flight, why?

My conclusion, there is more here than meets the eye. There is just not enough factual information to explain the oddness. There are enough doubts in my mind now to say the initial response of the PF was wrong.

Peace.

PF/PNF
 

1. The BEA report of 2009 describes the pilot qualifications as follows:

]1.5.1.1 Captain (Let’s call him Pilot A) TT: 10,988 flying hours, of which 6,258 as Captain Hours on type: 1,747, all as Captain

1.5.1.2 Co-pilot (Let’s call him Pilot B) TT: 6,547 flying hours
Hours on type: 4,479 flying hours Note: It goes on to say: “ …This pilot’s licences allowed him to perform the duties of replacement pilot in place of the captain …”


1.5.1.3 Co-pilot (Let’s call him Pilot C) TT: 2,936 Hours on type: 807 Note: There is no mention of replacement pilot qualifications as per Pilot B above.

2. The BEA report of 27 May 2011 makes the following statements:

The Captain was PNF, one of the co-pilots was PF At 1 h 55, the Captain woke the second co-pilot and said "[…] he’s going to take my place"

3. I had always assumed that it was Pilot C who replaced the Captain as PNF in the LHS but given that Pilot B was the only one of the two co-pilots authorized to the replace the Captain, I now conclude that Pilot B (high time on type) was actually the PNF in the LHS while Pilot C (very low time on type) was the PF in the RHS at the time of the accident.

JD-EE, good post, but...
I think the shouts are more in the line of "the pilot did it...", without ncessarily blaming him for not being able to cope.
Not quite true....
There are sugestions for improved instrumentation, improved (and less charged and confusing) displays, the addition of an AoA display, as well as better training. I agree with you there. Especially if the "stimulus set" is not static, but a complex "stimulus game play" that all plays out in a few minutes (which is a very short time...).

Unreliable airspeed drill or Stall recovery ?
 

AF 447 had stall warnings and unreliable airspeed indications simultaneously.

The A-330 unreliable airspeed drill in clean configuration at cruise flight level calls for the pitch to be set at 5 degrees nose up and for thrust to be set at CLB (which in effect puts the aircraft in a climb configuration)

The Stall recovery drill calls for nose down pitch control with thrust to be increased smoothly once out of the stall.

Because the auto thrust disconnected, if the pilots did not move the thrust levers, the thrusts would have gone to CLB on their own (A-330 thrust levers are set t CLB detent when in cruise on auto-thrust). The pilot's initial reaction was to bring the pitch up to 10 degrees nose up.

49 seconds after the start of the incident, after another stall warning, the pilot pushed the thrust levers from CLB to TOGA.

May 30 EDIT:

When the Autothrust disconnects due to a fault rather than by pilot action, the thrusts do not go to match the Thrust Level Symbols (blue doughnuts) as in the manual disconnect, but freeze in their present position.

CogSim, regarding the apparently silly roll commands I got to wondering if the PF was, indeed, the pilot who replaced the captain and whether he adjusted the stick properly when he sat down. If the stick was maladjusted I can conjure in my mind the roll being a spurious input. Of course, that does not necessarily apply for the times the command was all the way to the stops.

Another question comes to mind. We have the autopilot then auto-thrust disengage. Then the PF climbs with a roll. Is there a procedure that would call for a climb on AP/AT disengagement? (And did the pilot make his input before, simultaneously with, or in response to the stall warning? The wording seems to imply he acted before (perhaps barely) the stall warning. Was the pilot starting the unreliable airspeed procedure? (It was tens of seconds later that the thrust levers were shoved to TOGA.)

It might also pay to know precisely who was PF. Reactions honed in past experience may have led to an action inappropriate for an A300-200. That would indicate a cure is better training, with a lot more "unexpected events" during the training flight. (This event may end up as a bonanza for the people who make the simulators as more time may be needed for more unexpected events.)

Reading the various posts about UAS procedures, I have the impression they were all developed for low-level (shortly after T/O) UAS situations. In particular, I see the remarks/instructions about "preventing loss of altitude" (logical at low altitude).

Can anybody here tell this ancient (flight test engineer, not a pilot, and now retired, I should state), whether there are any formal high-altitude UAS procedures in existence (and trained), other than the basic "pitch and power"?
As implied earlier, the nose-up command "reflex" is understandable, if it results from an effort to "stop the altitude loss" in an UAS situation (inapropriate here).

Another possible difference between French and English versions
 

French:

"Le PF exerce une action sur le manche en butée à gauche et à cabrer, qui dure environ 30 secondes."

English:

"The PF made an input on the sidestick to the left and nose-up stops, which lasted about 30 seconds."

The way the French version is written is ambiguous and could also be translated:

"The PF made an input on the sidestick to the left stop and nose-up, which lasted about 30 seconds."

A small detail, but one that must be cleared up nevertheless

ChristianJ, I was more or less pleading for something to replace "the pilot did it" with "why did the pilot do it?" Who carries with it implicit blame for the person. Why at least "spreads the blame" to perhaps faulty stimuli, faulty training, he just had a bad day, and other possibilities.

And to be above board at the moment it looks like there is a training gap that may need to be filled and like there is a software design level inadequacy that can lead to stimuli that confuse even well trained pilots. (This latter comes from referencing past events as well as this crash. There is a sage old computing homily worth remembering, "Garbage in, garbage out.")

Agree, but what do you make of the stall warning ...
Let's say you want to practise a full stall exercice on a 757, would you still trim up past the initial stall warning ?
The question is why the automation is doing something you would not ?

That the low speed stability protection disappears in a scenario of unreliable airspeed is logic, but also auto trim should.

As mentioned earlier, the issue there is that there is very little data on the real aerodynamic behaviour of big jets 'outside the envelope'. For simulator purposes, most is just extrapolated from available data (see the NASA report on the subject).
Simulator manufacturers wouldn't mind, but they cannot afford the (highly instrumented and risky) flight testing needed to acquire such data....;

Not exactly. Its CLB and 5 degrees pitch.

TOGA is only at take-off and below 10,000 feet in the climb

I agree with your translation.
Basic problem with a "text" description, rather than the actual plots of the actual parameters, which we're unlikely to see before the July interim report, if even then.

OK, you put it better than I did.
"Who did it" is the first link in the chain, "why did he do it" is the second.

Minorite invisible

In French, is it not saying "during thirty seconds"?

In Normal Law, a pilot can input full left roll and full aft stick and hold it, without fear, "protections". N'est-ce pas?

There are several companies that specialize in aerodynamic modeling - I will leave them nameless for now, PM me for details - that CAN extrapolate meaningful modeling (supplemented by wind-tunnel testing to refine) that will provide valid flight model extensions. I have posted several times about the ICATEE RAeS - see RAeS Flight Simulation Group - Introduction to the ICATEE

Many of these Airbus "protections" are based on speed indication. In an unreliable airspeed situation as we had here, all "protections" and logic can be thrown out the window since the aircraft has no idea what speed its really at.

PF/PNF
 

TLB, I think Pilot B was PF in his usual RHS, and Pilot C took the LHS as the PNF.
Pilot B was at the controls with the airspeed indiction that was not recorded ...

Yes, it does say "during thirty seconds"

You are also correct about what would be possible under Normal Law, but the aircraft was in Alternate law, and the pilot knew it for he called it 11 seconds after the autopilot disengaged.

Which is backed up by the fact that Autopilot 2 was engaged as per the report.

GY & CJ, perhaps what is really needed is a reactions test for something like AP and AT disconnect during cruise. Does the person make the correct action when it happens? That takes care of the er "meatware" part of the equation. The second part of the equation is whether the "correct action" is properly defined.

The really unfortunate thing is spending an hour letting the people in the cockpit relax because sometimes nothing happens during this phase of the simulator check becomes "expensive" quickly. What a person does in real life is not necessarily what she or he does when knowingly facing a test situation.

(That's why I think looking into past experience and the reactions honed there may be well worth it. One tends to go back to "imprinted" reactions when in a crisis. Is extreme mobility between types a good idea or not? Most of the time it is OK, accident records prove this. But when fighting for that extra 9 in reliability speak you look at this type of issue. And this accident is really in the decimal places to the right of the 99. part of the number.)

Minorite invisible

"You are also correct about what would be possible under Normal Law, but the aircraft was in Alternate law, and the pilot knew it for he called it 11 seconds after the autopilot disengaged."

Yes, I am aware, but the PF made his correction in Normal Law, without fear (at the time) of any bad behaviour. The drop to ALT Law found him with a fist full of back stick and roll left? What does he do then? More importantly, what does a/c do? The THS?

thanks

milsabords. It is good that you are exploring these ideas. I don't wish to discourage you unduly. Please consider the below as constructive advice.

Let me challenge some of the assumptions above. Since the membrane is by definition flexible, is V really fixed? Is T realy uniform?

My personal philosophy in designing mechanisms and devices is KISS.
The fewer moving parts, (or variables) the better.

The law is the ideal gas law (of which the Boyle-Mariotte law is a special case.)

I do not think you would be building on a stable foundation if one used a trapped gas behind the membrane.

Minorite invisible, minor correction - it was PNF who made those statements. PF never, in the transcript, replied to the statements. (He may have replied and it was not noted in the partial transcript.)

NU input
 

A particularity of the sidestick, depending how you grab it, is that it can be easy to involuntary command a NU or ND input by implying a generous side input.
Was that NU input intentional ...

CONFiture

And "to the stops" infers either confidence or sloppiness. Or, over-reaction.

> It has been suggested on one or more of these threads that the corrective action in the event of airspeed failure above FL 100 is: TOGA and 5 degrees pitch.

> Not exactly. Its CLB and 5 degrees pitch.

> TOGA is only at take-off and below 10,000 feet in the climb

Minorite invisible,
Thanks for that. So my deduction remains: if the PF initially concluded (immediately following AP/AT disengagement) that his main problem (of many) at the time was that he was experiencing an unreliable airspeed situation, the fact that he increased power and raised the nose was exactly the correct response. Non ?
How the aircraft ended up zooming to FL 380 and high AoA is another story.