I can barely read a DFDR trace, and as such I'm certainly not going to attempt to second-guess the PF's thought process, but whatever it was, the overriding theme seems to have been to keep the nose up at all costs. The reason I mention the "rocketship approach" is because I wonder if he'd got it into his head that as long as he had the nose above the horizon and TOGA set, he was safe.

While the sim was being set up, I used the time to try out various different grip techniques on the sidestick. No matter how hard I gripped or whether I was moving with the elbow or wrist I could not get more than a fractional deviation in the pitch channel when moving it stop-to-stop in the bank channel without trying to add pitch input deliberately. This is not to say it didn't happen to the PF, but I don't think it's as easy as Machinbird thinks it is.

If we speculate on the PFs mental state (always a dangerous thing) I think we have to assume that he had only one thing in mind. All other information which might have contradicted this view was ignored or excluded. Sadly, the planes behaviour was not a factor in his actions. Evidence in similar situations tends to indicate fixation on one thing not a variety.
A study of cultural issues here would also yield dividends. Korean Airlines experience in the 80's and 90's is highly relevant with supposedly better trained military pilots messing up big time. It is also interesting to note the lack of CRM prevalent during that period. It is not where the pilot is from that is important but the culture and environment he/she operates in.
AS TTex600 notes past accident history does not indicate that things were necessarily better. But as accidents have gone down with better technology and more reliable aircraft, human factors have become more visible. Whilst a case can definitely be made that training is being neglected the airline culture has a more significant role.

Could you elaborate a little? I know you're talking about people you're actually meeting on the line here, but as far as this thread is concerned I don't think anyone has suggested that "Cause is pilot error, and we're done". Pilot error itself has myriad contributing factors including rest time, training, airline culture and human psychology.

As far as the airframe is concerned, it appears that recovery is relatively straightforward at the apex of the zoom climb, and becomes progressively more difficult as the aircraft sinks further into the stall. From the point they pass about 30,000ft on the way down it's impossible to know whether any airframe could have been recovered other than by using techniques which your average line pilot would not know, and in any case, from the moment they actually stalled they were in test pilot territory to start with.

As I posted previously and reenforced by TTex600 there seems to a problem with training and the overall approach. First as I noted and then again by TTex600, the simulator training syllabus for most carriers (all?) seems to have frozen in time, something like 25-30 years back - there is a lot of engine out/engine fire/engine fell-off type stuff, but little related to high altitude/UAS/near-upset. Does the current training regime match the real world as it stands now? My opinion is engines are very much more reliable than might have been experienced 30 years back, so are all these engine out situations worth the sim time? You tell me - I'm not a pilot, but I build sims, and model all kind of engine malfunctions, but nothing at all related to stall buffet cues... (at least in the sound department, may be the vibration boys do?).

Secondly, I know of several simulators that can be rapidly reconfigured to alternate aircraft - the cockpit representation is generic, but can be set-up as a twin jet, twin turboprop, light-prop, well in fact pretty much whatever. Further though, the flight modeling includes extended envelopes, up to and beyond stall - since some of the types modeled can be flown into and recovered from such an upset. So my question for the gallery is: "Would such a simulator be seen as an important adjunct to the existing FFS Level D devices?". You might say, "why bother? Put the pilot in a real plane and do the same thing." But I can very quickly see the finance department jumping up and down at the cost - the sim can run near 24/7 (ok, 20/7 with maintenance downtime), but doesn't eat fuel or cost much to run (versus any aircraft) and doesn't run the risk of falling out the sky.... The cost of that sim, well depends on a lot of details, but it should be on the right side of $5M, maybe quite a bit below that if smart choices are made.

A quick costs calc: ($5M/(10 years*365*20))=$68/hr purchase cost... (assumes sim lasts 10 years, runs 365 days/year/20 hours a day)

- GY

You've already acknowledged the source of my statement. I was talking about the way some other pilots regard the cause of this accident. The lack of airline pilot participation in this very forum is a good indicator of the overall lack of interest I see in my fellow pilots. Most airline pilots I speak to about AF447 have only a passing familiarity with the details, they don't have the interest to go deeper than "the pilot stalled it and didn't know how to recover". I'm not a psychologist, so any conclusion I draw from their attitude is pure speculation, but I think that most Airbus drivers I talk to don't want to accept that the airplane is partly at fault because that would mean that it could happen to them. Very few human beings want to believe that some event could overwhelm them, I believe it is called "denial".

Well, they can't say the aircraft was entirely blameless because of the issue with the pitot tubes for starters, but are you saying you believe there's another issue at play?

I think you'll find a lot more pilot participation in the non-public areas of the forums, by the way.

pilot interest
 

Hi there,
We as pilots are very interested in this forum despite all the comments.
I personally fly an A 330 last model and i feel confident with this bird.
I flew Boeing planes too and i believe is all the same job.
What we need is to fly more manually and train people to do it.
My Navy time told me that the checklists are written with the blood of others like this case
Best Regards
Alcalde

TTex600
Once again it comes down to human factors - know your machine and how to deal with issues. This was lacking. No aircraft is perfect and despite the current machines being far superior to previous machines they can still go wrong and have their foilbes. Pilots need to understand this and to be able to react correctly when it happens. Those that don't crash. You mention denial and this is certainly a factor but the very fact of denial is rooted in the mind of the denier. You are right that pilot error is far too 'glib' an explanation for the accident but it was a recoverable problem if the pilots had been able to react in the correct way.

OC, Human Factors?

How well trained were the AF447 crew to handle UAS at altitude. Judging from the horrible example we have been puzzling over the last 2+ years, they were not adequately trained for the successive situations that they encountered. Whatever the emphasis of the training program, it did not prepare them for what they encountered.

I'd like you to think (way?) back to when you were learning to fly. Did you ever get in way over your head? Do you remember that feeling? Did the instructor steady things up for you and allow you to resume control? Were you lucky (if solo) and puzzle things out for yourself?

These guys got in way over their heads. There was no one there to punch the freeze button and explain what went wrong. To their credit, they kept on trying to the end. Unfortunately they did not understand that which is perfectly clear to us using our retrospectiscope.

The purpose of training is to keep us from getting in over our heads. As you fly, you will periodically find little things you didn't know or did not properly understand. Some of these are of minimal import. A few can be life threatening.

When training departments decide what to emphasize, there is a risk assessment of some type used to allocate training effort. Most accidents seem to occur in the takeoff and landing phase. Very few in the cruise phase. This accident unfortunately occurred in a less emphasized phase of flight. Furthermore, the system failure, UAS, had only recently begun to be visible in the safety radar scope, so training in that subject had to displace other training to be included in the time-limited training syllabus.

To sum up, the crew of AF447, particularly the two co-pilots were not ready for the UAS emergency that they encountered. They handled that emergency about as well as you would handle a night landing aboard an aircraft carrier.

Machinbird
I totally agree with what you say and I hate to think what a mess I would make of a night landing. However, training or lack of it is part of the human factor equation. As you mention getting in over one's head is part of learning to fly. But we should learn from that and emerge wiser and more aware. At first it seems so easy and then suddenly it isn't but after the experience we have more respect for the environment and the aircraft. I would hope that pilots flying the big jets would be aware of how to troubleshoot and deal with the issues they face. However, some are not and the whys and wherefores are what make up the human factors. Reading the accident report on Korean Airlines Flight 6316 is instructive. One crew dealt with the instrument problem successfully, the next crew did not and crashed the aircraft. Once again there were all sorts of human factors involved such as CRM (or lack of it) and lack of communication and training. Certainly training for UAS needs to be improved and it needs to be taught as a recogniseable issue. But the fact remains that there was a procedure availaible and a checklist to follow at the time. Maybe not clear but it did exist. It wasn't used and there seems to have been no attempt to troubleshoot the issue professionally. This is cultural as well as a gap in training.

Old Carthusian, one can't "know his machine" when the machine is Dr. Jekyll or Mr. Hyde.

From what we presently think we know, the pilots dealt poorly with the UAS which lead to an upset. Once upset, they were faced with control harmony different than normal. How is the crew to "know the machine" when the machine isn't consistent?

it's consistent in that it does what the pilot commands it to do. It's the pilot who is not consistent in knowing what commands to apply

To offload pressure on the primary flight surfaces, exactly the same as autotrim does. Autotrim is a way of solving the problem of not having the surfaces transmit the pressure through the primary flight controls - nothing more, nothing less.

I'm sure that other airframes also "wander" to some degree, and always have, because even calm air is never 100% calm - it may not have been as apparent in the steam-gauge days due to the analogue nature of the instruments.

That's one way of describing it - another equally valid way is "gives you exactly what you ask for in a manner that is appropriate to the conditions". In terms of flight path stability (which is after all, what the manual combination of thrust, PFC input and trim were trying to achieve in conventional jets) it is relatively unmatched.

Except that is not a fair description in that it isn't "random". The whole point of the "graceful degradation" aspect of the design was that it gives you as close to Normal Law handling as possible despite the fact that some of the systems required to give you that handling are not functioning correctly. A fairer analogy would be a power steering setup that has a partial failure mode making the wheel half as heavy as it would be with power sterring out completely.

How is it possible to quote someone on 22nd Nov 2011, 21:40 who doesn't post the quoted statements until 23rd Nov 2011 05:31??

Now that's spooky. :eek:

(HAL is at it again)

TTex600
I think that Jekyll and Hyde is a bit too much to describe an A330. I'm certain that it's not that bad. Iomapaseo sums the situation up excellently. Any airliner would have ended up in the same situation as AF447 if the pilots had performed the same actions. The accident wasn't platform related but crew related.

@chrisN - vBulletin uses server time as it's timestamp, otherwise threads would be unreadable. My guess is that the server was restarted at some point or there's been a database hiccup somewhere.

@Lyman - The sidestick is spring-centred, the PF does not have to "reindex it to neutral".

DW, your post now numbered 457 has jumped in 2 above my 459, to which yours is a reply. You clearly did not send it at 09.32 UK time, so I wonder if your computer has defaulted to a non-UK time zone? The MS default is western USA – or something – might be worth you checking (it happened to me once).

Regards – Chris

Lyman; perhaps I have missed someting, so please explain how the PF was without attitude display.

My post in reply to #465 has goe into 459!!! So here goes again;

Quote:
Without attitude display, PF was blissfully thinking he had only the ROLL to mitigate?
Lyman; perhaps I have missed someting, so please explain how the PF was without attitude display.

Several posters have commented on the system's apparently putting posts out of order.

I can't claim to have much knowledge on how the system works in the background but my understanding is that posts are ordered by DTG.

Having said that we did have a period some time ago when the system went a tad strange and was logging incorrect DTG. Perhaps we are revisiting that situation again.

Either way, if you can tolerate the problem for the short term please and I'll refer it to the experts for attention.

Indeed, this post has been logged as having been made about 8 hours before it was made and posted in that incorrect DTG order - how quaint.

Some of us have arrived at a conclusion, some of us have not.

That about sums it up.

"Dead men tell no tales"

I get the feeling that some of you would defend Jack the Ripper if he happened to use your favorite brand of butcher knife.

If the airplane consistently does what it is commanded to do, how do you explain "auto trim"? When does the pilot command trim?

Why does it wander around the selected speed? (for you Boeing , Embraer, Canadair, Douglas drivers - the bus will vary by ten or more knots below 10Kft when fully automated. It trims for flight path, not for speed)

For the open minded readers, I don't condem these actions as defective or bad; they just are what they are. Contrary to the spin (partisan positioning) you read here, the Airbus takes what you input, processes your input through its brain and outputs whatever it decides appropriate. The aircraft responds to the SS one way in normal law, and another in abnormal, and another in direct. Imagine, if you will, the steering wheel on your automobile randomly varying tire steering angle for a given steering wheel angle. Fun, huh?

OK, I don’t know how for certain, but post 456 suddenly became 457. I think DW’s was inserted above. If not his, another was.

I suspect it is to do with our different computers (and maybe time zones) having different “real” times of posting, and the PPRuNe server sorts them into an order it recognises.

But what do I know?

Chris N.

Machinbird
Your post touches the issue of CRM as well as training very nicely. A proper division of responsibilities and teamwork come into play. These don't seem to have been present in this cockpit at all. The PF acted in a way that suggests he had no idea of what was happening and the PNF failed to take any action to correct this situation or to redefine the approach to the problem beyond saying "go down". At first the situation wasn't one where an intensive response was needed - sitting there, following the procedures would probably have solved it with AS returning after maybe a couple of minutes. Several threads back PJ2 suggested this very approach.

From grity, PF input NURL at 2:10:08. The stick started at neutral. The a/c did not. It was NDRR. Did PF not re-index his stick to neutral? Did he carry on with NU 1/3 back as his neutral? It looks like it, the PITCH (stick) stays reasonably consistent whilst the roll was active.

Was he trying to maintain PITCH with a 1/3 back bias? That would explain the climb with no ND to arrest it. Did he climb, thinking he was 'maintaining'?

Hi,

DW
"When does the pilot command trim ?"
You don't answer the question .. you just explain what is trim
With autotrim the pilot not command directly the trim
The position of the trim is a result of a combination of the elevators position and other factors
He have no power over the trim movements unless he switch to manual trim

My one reservation regarding the above is that knowing procedures and having checklists is not sufficient to do the job in some situations, e.g. the "night carrier landing equivalent type situation" or other similar hands on intensive task. Only recent training buildup is sufficient to allow these tasks to be accomplished..

In this case, the crew probably knew their (current) UAS procedure, but they never got that far. They were likely hung up by the PF's problems in roll over-control.

For those currently flying the 'Bus, would you action the ECAM procedures while the PF is still having difficulty with the actual flying part of his duties?

Establishing PITCH is an immediate concern, it cannot wait for ECAM, or conference with PNF.

TT. Do you get my question? Neither Pilot appeared to understand how in the weeds they actually were, from the very git. I believe the PF actually thought he had a reasonable NOSE, not one that was increasing its AoA. His lack of awareness not only seemed blase, it didn't get an instant "Hey!" from PM. Without attitude display, PF was blissfully thinking he had only the ROLL to mitigate? imo.

Again, at the outset, the a/c attitude was quite different from stick position.

Is this an awareness issue on the BUS?

Machinbird, If one goes "by the book", the PF orders "ECAM Actions" or "Stop ECAM Actions" as necessary. In other words, the PF should be in command of the situation. I would not expect the PM to proceed with ECAM actions outside of my command. At the very least, the PM should query the PF for guidance before beginning ECAM actions.

In reality, I suspect that some PM's might begin the actions if they perceive the PF to be overloaded and they believe the ECAM actions beneficial to resolving the issue. If I believed the PF, in my case that would be the First Officer, to be in difficulty flying the airplane, I would take over. I believe that my First Officers would likely do the same if they believed I was having the same difficulties. They would likely wait slightly longer before taking over from the Captain, but they would take over.

Lyman, I get what you are saying, No I don't think this is an issue in the Bus. At least not any more so than any other EFIS/"tape for speed and altitude" aircraft. The initial upset could have happened in any airplane that uses computer generated flight instruments. If I remember correctly ( I don't live and breath this accident and don't have the time to look back at the BEA rpts), the A/P gave up and gave the airplane back to the pilots while the aircraft was still stable. I have followed your thoughts and don't have any idea if you are correct. Hopefully we will learn all the facts when the final reports are issued.

Dozy. You misunderstand. PF never allowed the Stick to visit its neutral point after taking control. It was aft the entire time of grity's pictorial. My assumption is that he 'established a neutral point', in his mind only, and hence disregarded the aft stick he consistently input.

See how the PITCH trace stays aft of neutral? PF may have been 1/3 of the way back at odds with the stick. (Actually at odds with the attitiude,:=).

NR. I can't imagine that he had an attitude display. How could he have, and keep asking for more NU?

Please note that posts can be deleted, in which case, I believe the post count and posts numbers past the deleted one(s) get updated.

Post number is not critical and can alter but post order should be appropriate to DTG. Clearly something is not working quite right at the moment.

I've passed the information back up the totem pole so we can presume that the computer folk will have a looksee at the problem for us over the next few days.

(Following on from my previous post, this one is now logged with the correct DTG and is posted in the correct order - definitely quaint).

From the 'Las Vegas Slot Machine Technique & Aviation & Other Technology' weekly for today:

:)

What did PNF have when he referred repeatedly to "go back down" and "According to the three you’re going up so go back down" ? From the phrasing (and maybe need to go back to the original French) would it be attitude, vs or something other ?

BEA state "Reading the three instruments (the two PFD’s and the ISIS), the PNF noticed that the airplane was climbing and asked the PF several times to descend." - how the heck they know which instruments PNF was reading I'm not sure, but maybe the assumption is that there was no other possible source to infer "climbing". If so, then logical conclusion is that no source for that information was unavailable...


And yet, like you, I've always wondered if attitude info was in some way lost. Post after post on here has said no way would you pull the nose over 10deg up in cruise, so it seems the only way to explain PF actions... but then not PNF comments. The HF report is going to be very interesting I think - there is much in that domain that just seems to make no sense.

infrequent flyer 789.

The a/c was NOSE DOWN at handoff. It was also climbing at 1000fpm. So, yeah, VS. Robust UPDRAFT. I don't pretend to know what each pilot saw, but it seems that one was using ATT the other VS. At least at first. Then, perhaps each one, feeling 'burned' by one or the other of the instrument readings, switched horses.

Never the twain to meet. Frankly, in the longer term, the PNF had the better guess. It is clear from the absolute gitgo that the pilots were not flying the aircraft as most experts on this thread would have. But they were not up to the standard evident here, alas.

The ship was climbing 1000fpm. What does that say about the Zipper? With the turbulence ("fortes"), I still am not convinced ICE was the culprit re: speeds.

Hi,

You can maybe never know .. as it's no alarm for pitot tube problems (like clogged by ice or other substance) on the A330
So it can be a real pitot tube problem or " turbulence ("fortes") "

I suspect the probable outcome will be described as "super-cooled" icing.

It appears from an article in news.com.au published a few hours ago that the lowest temperature that pure H2O will remain as a liquid has been extended from -41/-42°C to -48°C.

This would cover the AF447 situation where the nominal OAT was -43°C and an updraft containing super-cooled droplets could have created an "ice-block" on the pitots and possibly other surfaces within a second.

The initial pitch down and roll to right at A/P disconnect may have had as its origins, the same cause.

[Car analogies - the staple diet of internet discussion boards, how did we manage so long without one ?]:E

Not so much random as predicatable change in feel. Not disimilar to power-steering failure... which happens. Most of us will never see it and never (these days) drive without power steering or remember what it was like.

However, in this case there was no change in control law in pitch axis - which was where it all went wrong - but a change in roll law may have pre-occupied the pilot and lead to failure to manage pitch. More like failing to steer the car right while worrying about accelerator or brake failure. Those do happen, and probably more often than steering fail. Many modern cars can go into a "limp home" mode for all sorts of failures (MAF sensor, DPF...) which is exactly the same principle as the degraded control laws. Fun ? Probably not, but not considered dangerous enough that drivers are given any training (or sometimes any information) in advance of it.

In the aircraft case, at least there is the sim to train and prepare for it. Except someone apparently decided it wasn't necessary to train the AF crew...


While we're on cars, it maybe worth comparing this case with the Prius accelerator issue. Initially blamed on the design/manufacture (type) and the computer control (fbw) going wrong. Turns out that in the end it's down to an age-old human-factors issue of not figuring out that something on the floor is stuck under the pedal. Yes, an element of bad design - floor-mat - but mostly HF issue that's happened before on other types. Just that the new computer control (and maybe the foreign mfr.) is a convenient initial scapegoat - one that some will still believe in even after all the investigation and reports are done and point a different way.

Also contributing is automation dependency - changes to reduce driver workload. Where a whole nation of drivers can't co-ordinate more than one foot, or take a hand off the wheel to change gear (because it's got a phone in it) - resulting in the removal of the mechanically connected clutch & pedal. Which, on those rare occaisions where your car decides to accelerate away uncommanded, is actually a really really important safety backup. [feel free to draw different conclusions on the a/c equivalent for that one].

Lyman,

Look on page 42 (IR#3, en). At 02:10:05 the vertical speed was zero, the next 5 seconds it was descending 400 fpm max, and at 02:10:10 the V/S became positive, i.e. it started to climb.

Moderators: This was posted around 19:00 GMT, in reply to Lyman's #483, posted 24 Nov 2011, 18:42 GMT.

The important thing to note is that in the videos the stick is quickly "blipped" in a given direction and allowed to return to neutral. What we see in the AF447 traces are inputs that are large, sustained and held for noticably longer.