KayPam

1. Yes, I think BEA indicated a possible reason for wanting to climb : the degraded altitude calculation indicated 400ft under target altitude. One must reckon that at this time, its unlikely that any pilot would consider themselves in an emergency and therefore would feel free to descend without proper ATC authorisation
2.
3. You mean that the PF climbed against the PNF's recommendation ?
I read the original transcripts (not translated) and the main problem with them is that the oral tone is lacking, hence it is impossible to know whether some sentences were comments or order. "You are descending now" and "you must descend" are two equally likely translations of "tu descends, là". Impossible to decide between them without hearing it, and even when hearing it it's likely that one could not decide which it was.
We even can read things like "là tu montes, donc tu descends" (literally "here you climb so you descend" which makes no sense in English, again translation problem and tone required)

4. I can bet (and probably could verify if I had time) that all artificial horizons were correct. About the indicated speeds, one should check parameters IAS1, IAS2 and ISISIAS in the recordings, but I believe these were indeed all lost (?) temporarily

5. An aircraft not responding to control inputs would be classified as a catastrophic failure and should not occur more than once per billion flight hours. It was not the case this night.

6. I don't know that for sure, but it does seem a bit stupid to lose the ground indications (GS, attitude, position) of a perfectly working IRS just because the pitot has frozen. So my guess is that Airbus engineers aren't that stupid ?

7. I think nobody knows for sure what was displayed on the PFDs, but there should have been at least an attitude indication on the emergency horizon.
I don't think it would be wise to follow the FDs in an abnormal situation.. And they should switch off on their own if the APs switch off due to lack of information I don't see how FDs could compute orders if the APs cannot.

I'm not appointing blame on anyone here (someone was asking if the French jurisdiction was doing it, just on the preceding page), just pointing out a few basic theoretical facts that imho should be kept in mind in case of such an incident. (because it starts with a simple incident : pitot information loss)

Don't you think the PF could have decided between high speed stall and low speed stall by looking at his GS ? Even with wind, the difference between the two should be in all cases sufficient to decide between high speed and low speed.

We should learn from accidents and try to improve safety. I would suggest these
1. The creation of a color code for each displayed parameter :
- white - normal, confirmed by multiple sources
- other color - abnormal, at least one source disagreeing significantly
- other color - abnormal, no redundance
- red - unreliable or unavailable information
1bis. there could be a new status page indicating which sources are known to be unreliable (e.g computer sending error signals), which sources are feeding information where, and if this information is consistent with other sources
2. The improvement of the standby instruments systems towards a very high reliability even at the cost of a slightly lower precision. For example, maybe the precision of modern DGPS could allow to compute a reasonably accurate attitude indication (with 7 receivers on each extreme point of the aircraft).

My (debatable) idea is that GPS would help confirm values computed by the usual way.
e.g if you measure a -8000fpm GPS descent rate, a -10000fpm indicated VSPD would be confirmed but if 0 fpm was displayed on the PFD, it would trigger an alarm (and color coded indication) (as well as the contrary)

Uplinker

The idea of a computer Status page might be useful, but could risk adding further confusion when basic flying skills were all that were needed.

See post #1119 on page 56, which although I had the added (self induced) complication of a suspected V Alpha prot event, really showed me how AF447 could have occurred, and in particular that the "STALL STALL" was continually sounding, but I did not hear it.

I really think we should all be allowed to regularly practice far more basic flying and basic instrument failures in every one of our regular 6 monthly SIMs.

For example

• Partial panel work, e.g. loss of two ASI, or V/S, or ALT, or ATT, and flying by reference to what you have left.
• Unusual attitude recovery.
• Unreliable airspeed but not necessarily a madly varying speed tape, instead more subtle failures.
• High altitude handling, including stalls and TCAS RAs.
• Engine failures not only at V1, but in the intermediate climb, cruise, approach, or Go-around.
• Memory drills.
• More Go-arounds and windshear events.
• Flying by sole reference to pitch and power.

We are always briefed to use the automatics, and a large amount of SIM time is taken up with learning all the automatic modes and functions. But we should NOT allow ourselves to get so rusty that we don't actually dare take the automatics out and fly manually when they are not providing the solution.

Busy line flying often means that we don't have the chance or don't want to risk screwing up an approach by flying fully manually, (AP and A/THR off), but if the CAA mandated three fully manual approaches every six months - as they used to for autolands - it could provide just enough of a jump-start start for the collective pilot confidence and skills to return.

Concours77

"2. The aircraft did not climb immediately, it started climbing ten seconds after pilot input. (Also BEA). Could this have added some urgency to the NU input? Not unresponsive but "lackluster response."? Or could this be data input problem that caused a too extreme Pitch target? see below, g value?

"3. You mean that the PF climbed against the PNF's recommendation ?"
That's hard to say. I did not read (certainly did not hear) "Unable" from PF. I am concerned that the PNF is seemingly quick to abandon his alarm about the ascent; did he suss PF situation (as it was perhaps) the A/C would not Pitch down, at least not sufficiently to arrest or reverse the ascent?

"5. An aircraft not responding to control inputs would be classified as a catastrophic failure and should not occur more than once per billion flight hours. It was not the case this night."
Not my inference, only that the A/C was initially slow to respond to NU. This would be important as it may have set the stage for over control in Pitch, as we see with Roll?

Not discussed, but is it agreed that Automatic Trim might well be inhibited with the loss of A/P and UAS? It seems to me a straightforward and (existing) trim value should be sustained? The goal in manual flight is maneuverability? Not Trimmed cruise or a not necessary pitch bias? It turned out the trim ended up biased fully NU, eliminating any sense of a neutral point with the stick? It resulted (imo) in giving the PF a confusing input as to stick position, and authority? He was concerned with Overspeed, ("do you think we have some crazy speed?") not likely to push the stick forward to the stop and sustain it there?

That's pretty much what I have... The responses have been gracious, and illuminating.

Thank you to Winnerhofer, KayPam, and Uplinker

KayPam

I agree.

But I don't think the general public will agree to buy more expensive tickets, in order to allow for that supplementary training.

Concours77 : just regarding the overspeed issue : yes he did think he was in overspeed (at least it seems so). He was worried about high speed stall
But aircraft do go up to VD without high speed stalling..
Who even knows, beyond themselves, how far Airbus test pilots and engineers push the airplanes in the high speed range ? I read somewhere that the A380 went up to M0.99.
Given the sound barrier (ultra sharp increase of Cx at M1) it's unlikely to go beyond it, and so a high speed stall seems very unlikely with a A380.

Maybe pilots should be given more theoretical knowledge about the real limits of their aircraft ?

This would allow for a better "overview" of the system used, would not allow any voluntary exceedance of the parameters, but could prove useful in extreme situations.
This could help in situations like AF447 :
I am unsure whether I am stalling or high speed stalling. If I was high speed stalling, could I destroy my aircraft by pushing down to a -10° pitch ?
If a pilot knows for sure the answer to this question, he can make a more rational decision.

Winnerhofer

https://www.youtube.com/watch?v=VBNZ46zxY5E

Concours77

Is that audio an actual portrayal of the CVR? That contains a great deal more pilot discussion than I read from the report? Also the ambient noise adds a lot, if constructed from the actual data?

There was a time I could have translated that....that time is passed. Is there an English translation?

KayPam

You may or may not have noticed that much of the speech is told by the off-screen narrator. He should have a distinctive voice.
Regarding the text said by the pilot, I think it is accurate for the most part. But there are differences with the CVR transcript.

If you used to know French, you can probably still read the text in the BEA report and compare it with the audio.
https://www.bea.aero/docspa/2009/f-c.../annexe.01.pdf

BUT the tone is not available ! And that would be very useful to understand the meaning of many sentences.
I really don't understand why the CVR recordings are never released.
Anyone has an idea ?
Is it just the privacy of persons who have passed ? or is there any more obscure reason ?

Concours77

Three confused pilots in high anxiety, piloting a 100M jet, with 228 souls on board descending into the sea. Can't be good for business.

The interface is the bottom line...man/machine.

Contain, Control, Manage... Damage control...

Winnerhofer

The DCVR released by the BEA is hashed.
There's no way that when Dubois asked Bonin if he had licence, that the latter just answered "Yes".
There's more that came after that "Yes".
A source claims that a brief argument ensued as Robert entered.
The source claims that Bonin's spouse was on the flight deck all the time which explains the mysterious fourth voice.
Why was Dubois standing?
Why were Robert and Bonin not strapped properly?

KayPam

I do believe that the no good for business argument might be the most important one
Does Pprune know about the Habsheim crash ?
A 320 pilot, we're told, crashed his aircraft in the forest at the end of a short runway where he was making a low pass, with the utmost incompetence.
In practice, we can clearly see different markings on the DFDR photographed just after the crash, and the DFDR presented to the press when the judges seized it ten days later.
There are also gross unconsistencies in the BEA report including the DFDR data.
So in my humble opinion of 20-year late observer, there is at least something wierd with this case. If any one would like to discuss it further we could find a place to do it.

Back to AF447, there was an article in vanity fair which "revealed" supposedly secret conversations.
We learnt tremendously interesting details, like that a flight attendant was allegedly carrying brasilian meat in her hold luggage !

What does the D stand for in DCVR ? Digital like in DFDR ?

According to a French commentator, Dubois was standing because he did not have the **** "to grab one of his subordinates by the collar and throw him away from the controls". One will or will not subscribe to this analysis.

Why would you think that the copilots were not strapped properly ?

Regarding the "are you full ATPL" question, there are (...) just after it in the CVR transcript so that might be the following discussion about it.
Or, the discussion could have taken place in the full two minutes before the "yeah" and the next line transcripted.

Concours77

Why did Robert give up on trying to get Bonin to descend. What did he find out that satisfied him that Bonin could not descend? He made no effort to take control.

Why did Bonin reach for spoilers?

Did he really say "Eh, what are you doing?" That implies he assumed pilot error....then he stands in the middle, giving silent assent (as did Robert) that Bonin should continue to try to regain control.

The "sense" of CRM conveyed from the comments in the report is....bizarre.

A question about stick position? With full NU Trim, does the stick neutral point remain centered? If so and with full back stick, is the back limit (stop) sensed or displayed? Felt? How would a pilot trying to descend acclimate to a stick with Full NU bias? Does ND require more travel than when Trim ="0".? It would with full NU trim, no? Is authority a function of stick travel?

KayPam

An airbus in alternate law should keep its attitude stabilization. In direct law this will be lost (USE MAN PITCH TRIM)
But I'm not sure whether af447 was in alternate or direct law, but it's obvious that a full nose up THS will highly reduce your authority to pitch down, should you want it.

I do not understand all of your questions.
However you might be interested in this :
The airbus stick always feels the same, +/- 18° forward and backward, +/- 20° laterally.
Then the computers do all sorts of computations before transmitting them to the servocontrollers

Concours77

The a/c switched from Normal Law to Alternate Law 2 (b) @ 2:10:07.

That means Roll goes to Direct, Pitch remains g load. The Rudder switched to RTLU 2 degrees each way? Four degree sweep. The heading after loss of a/p drifted @ 1.5 degrees per second turning right....pretty sure that's correct.

Thanks.

alf5071h

There has been time to reflect on this accident and consider what has (has not) been learnt.
Although surprise, startle effect, was considered it was not rated particularly highly.
The effects of startle have been discussed more recently in http://www.pprune.org/rumours-news/5...ml#post9613307.
Most of the points in the link could apply to AF447; unable to comprehend, communicate, no CRM ..., and an instinctive reaction. Also that startle applies to both crew, each independently, attempting to understand the situation. Furthermore, the surprise condition could continue, refreshed or renewed as other warnings and alerts occurred, and particularly that the situation was not improving with actions which were believed to be correct.

The last point might be debatable, but similar inappropriate actions can be identified in other so-called LoC accidents, e.g. Colgan, CRJ. In all of these accidents including AF447, the crew manoeuvred the aircraft into an upset condition.
The industry should consider the effect of training for one scenario on other situations, the possibly of inappropriate instinctive response.
Colgan had recently trained for 'tail stall' where the recovery required a sharp pull-up and flap retraction; the aircraft encountered a conventional stall, but the crew's actions were for a tail stall.
The CRJ suffered an IRS failure which resulted in an erroneous nose-up ADI indication. The PF appears to have taken upset recovery actions with a quick nose down control response.
In AF447 the crews actions were consistent with the memory items for dynamic UAS, whereas the situation required the follow-up actions applicable for level flight.

Did the then recently required UAS refresher training consider these different conditions?
The checklist differentiated between memory items and follow-up drills by an unusual qualification of 'safe conduct of the flight', which was not defined. Was this difference explained during training?
Would the relief PF had any experience of handling the UAS conditions; do operators' SOPs require the Commander to take manual control in all emergencies?
Did the most recent training consider alternative drills for 'loss of airspeed', together with multiple and confusing distractions of the ADC problems, systems reversion, and the First Officers' experience limited to reading the memory items of the abnormal drill.
The crew did exactly what they were taught; they were incapable of detecting the error, or consider other actions, or communicate due to the effects of surprise. 'Last practiced, best remembered'; particularly if surprised.

If the industry continues to believe that there is no limit to training, that humans can always be improved, without considering the situations human have to manage, and the limitations of human performance, then we may continue too encounter these rare and surprising accidents.
Surprising to us, but are the regulators also too surprised to realise? Knee-jerk reactions for LoC recovery training - instant reactions and action, whereas nowadays there is more emphasis on awareness and avoidance, but what about situational surprise.

Concours77

The Stall recovery training (more specifically approach to Stall) for this aircraft involved "limit altitude loss". That has some obvious drawbacks once the airframe is actually stalled. The Captain, Melvin Renfro, was surprised by an initial, and "spurious" Stall warning. He had left wing anti ice ON, which elevated the onset of the aural Stall Warn by twenty knots. It was, essentially, a fatal response to a bug speed. Lack of SA, hair trigger neuronal, male ego, etc.?

His initial Pull may have been related to either Tail Stall training, or "limit altitude loss". In either case, the unnecessary pull was the proximate cause of the actual Stall. Once Stalled, the aircraft was lost. I remember the altitude on long final at the time as about 1000 feet. He continued to pull on the column, overcoming the pusher, which requires seventy pounds pull to be defeated.

Muscle memory? Training? Inexperience? Lack of sterile cockpit?

In some cases too fine a point is put on "causes" that are irrelevant?

In 447, we actually are aware of the PF's concerns: "What's that smell?..."
"Do you feel crazy speed?" "Maybe spoilers...." "Mais non...." No mention of Stall Warn (seventy actuations continuous in the descent). As far as we know, right?

We are missing the most important (CVR) data available, not even CAM airstream signature is released? No intonation, no rebukes, no questions, no aural record. That is a disgrace.....

PEI_3721

Concours77, you appear to believe that its is a given right to hear the CVR; for you to interpret as you wish, or judge crew thoughts based on your experiences and knowledge, often limited to national and cultural norms. Whatever you might conclude is unlikely to change aviation; self-satisfaction is not always the best way forward for safety.
The world-standard for accident investigation and publication of data has been discussed previously.
The accident investigators report their findings and recommendations, the industry judges and heeds their advise. Operators and individuals are free to make their own interpretation, discuss aspects from unique viewpoints, and perhaps improve their sector of aviation, but don't expect everyone to agree.

Concours77

I take your point. I do not believe the data should be published without regard to its audience.....I do however take note of who is first on scene, who decides a course of action, and who are the "invited parties."

Government agencies, airframe manufacturers, engine manufacturers, end of.

All of the above have varying degrees of conflict. From the outset of most accidents, conflicts are rampant. To say that the usual agencies have no bias, and are completely objective, is to invite a chuckle.

I don't speak for any other than myself, but it occurs to me the occasional "ombudsman" would be welcome.

Boeing/pencil whipped skin inspections. Boeing/confidential logs: lithium batteries
Rolls Royce/test bed explosions: Trent. Airbus/ g data Habsheim.

Incomplete data relative to any accident invites suspicion... Transparency and disclosure are missing; the customer is compelled to "Trust, don't verify".

My position is not a popular one...or not?

Winnerhofer

https://www.youtube.com/watch?v=ZFyrTUN5_as

Concours77

Many mistakes made in that video. Many wrong statements. The scapegoat is identified.

Timing of first Stall warn is quite wrong. Last words of crew was not Bonin asking what's wrong. Bonin is not "Captain". "Perfectly good airplane"? Aircraft did not do as described.

Captain's first comments are wrong, etc. Even from the limited CVR data released, that depiction is nonsense, though undoubtedly "sufficient" for public consumption....

Machinbird

At present, the most critical element in an aircraft is the man/woman in control. How he perceives, how he analyzes, what he does, is under control of a brain that has been hard-wired to handle fear since before we became men. Some of the things that are hard wired in stressful, fear-inducing conditions are contrary to common sense and survival while in control of an aircraft.

How do we train to understand the effects of fear?
How do we train to handle the flow of chemical signals that are awash in the body during fearful incidents? Isn't the simulator an imperfect tool for inducing fear?
Where is the human operating manual that documents what the effects of of extreme fear are and how to handle fear? Have you seen one?

As an old Navy jet carrier pilot, I have seen fearful incidents. They can last for fractions of a second, several seconds, or a large number of minutes. With experience, you learn to gain control of your mind and your body and to perform at a high level.

But how do we teach new aviators to handle fear without undergoing actual fearful experiences?

Should we just assume that a pilot who is trained to fly in the center of his aircraft's performance envelope is going to somehow be ready to handle fear in an unusual situation where things suddenly go pear shaped? Does the simulator adequately train pilots to handle fear?
I don't think so.

Fear is the factor that is not adequately addressed in AF447 and the other accident that alf5071h mentions.

I think we had better find ways to address the fear factor effectively or else we will find pilots removed from the cockpit by automation that cannot become fearful.