Dozy, read what he says:

"I have no control of the plane at all" He wouldn't say that if his inputs were having effect.

Not to burst your balloon, But I don't care what you think, HE thinks his controls have no effect. And, erm....he's flying.

But the traces show that his inputs *were* having an effect, and that the aircraft was too aerodynamically unstable for the flight surfaces to have the effect he was expecting.

It's been pointed out more than once that the PF was a sailplane pilot, and as such should have been more likely to understand the basics of aerodynamics than many.

The question he should have asked himself (and his colleague) was "*Why* do I have no control of the aircraft?". The clues were there, and at that point there was still plenty of troubleshooting time.

[The desire to absolve the pilots I can understand (and to some degree is already backed up by their apparent lack of training), but why the continual grasping at straws to conclude the aircraft must have either forced them into making the mistake or suffered a failure above and beyond that we already know?]

first of all...

@ rouli The Viper limiters were inviolable - very hard. The only way to manually control the horizontal tails was with the pitch override switch, and that thing did not work unless AoA was above 30 degrees In other words, we could use it to "rock" outta the "deep stall". I had some excerpts from the Code One article on that in a very early thread here when we were discussing the possibility that the AF jet had an aft cee gee and the same problem as the Viper at extreme AoA's.

@ EMIT TNX for the nice words. I knew about the Belgian, as it was a few months previous. The thing about getting slow or even close to a normal AoA was you lost roll authority. I commented on this during my approach, and 170 knots was prolly too slow, and I should have stayed at 180 or so. The engineers told me that I had approximately one pound to play with out of the 17 pounds required for max roll command. Another factor was that the yaw trim reversed as you slowed down below "x" AoA. So it was a good thing I simply drove the sucker onto the concrete with no flare or attempt to soften the touchdown. That technique became the procedure after another incident or two.

@ Doze Yeah I know the three modes, but feel that the "alternate" should be more basic and eliminate the THS auto-trim feature. old ground, I know and we had a good discussion on the pro's and con's two years ago.

As we discussed two years ago, I believe the autotrim shouldn't be the issue because it is one of the things that Alternate Law has in common with Normal Law and as such would feel more natural to a FBW Airbus pilot who has spent 99% of their time flying in that mode.

The only reason autotrim became a factor here was because the PF commanded inputs which were way in excess of what should have been expected in Alternate reversion. The setup in the case of dropping out of Alternate Law is to give the pilots full authority, which makes perfect sense as long as the pilots understand the situation and make their inputs accordingly.

Interestingly (if the sim session was anything to go by) the A320 does have hard limits on autotrim in Alternate Law - which required manual intervention from the TRE to match the trim state of AF447 during our session. Why this was not carried forward onto the widebodies is a question for Airbus themselves, but I wouldn't be surprised it if was a reaction to the perception that the A320 was excessively limiting - if this was the case then the lesson should be "be careful what you wish for".

The point that I still stand by is that a drop from Normal Law means that you cannot rely on the protections and more care must be taken when making control inputs. Alternate Law does make sense, as does the requirement for at least two data sources for protections to be active - I hope you can at least follow my reasoning there even if you don't necessarily agree. The role autotrim played in moving the THS made their lives more difficult - that's a given. But the inputs that caused the autotrim to command that deflection should never have been made in the first place.

To auto trim or to not.....
 

I agree with you for the most part, Doze.

Our "autotrim" was referenced to the gee we had trimmed for. So If I pulled hard and relaxed, then the thing tried to get back to the "trim" setting I had. thought we went over this before.

As I mentioned, some folks trimmed to zero gee when entering a fight so to "extend" and gain energy they just had to relax on the stick and HAL got to that wonderful zero gee condition. My understanding is the 'birds on the wing and slot use a tad of trim for something like 0.9 or so. Makes the close up work a bit easier.

[The desire to absolve the pilots I can understand (and to some degree is already backed up by their apparent lack of training), but why the continual grasping at straws to conclude the aircraft must have either forced them into making the mistake or suffered a failure above and beyond that we already know?]

1. There are mistakes and failures far in excess of "what we know". Just a stabin the dark.

2. No desire to absolve the pilots. I am asking questions in areas that have been foreclosed by a rush to conclude.

3. Questions I ask are studiedly biased at times to elicit a different look, or pov.

4. I have a place in my heart for "patsies". The "fall guy", the pilot, iow.

Much of my passion here has to do with how absolutely crazy it is that this happened. "Airworthy aircraft, meet certificated pilot crew, see you in Paris."
From the brief prior to launch through "I have no control".... It makes me angry.Most of my anger is reserved for AF. Except the pilot group, who showed stones in demanding a r/r pitots.

The flip side is there is so much to learn, to change, to re-assess, and then re-introduce ourselves to a more honest approach to Air Travel. The driving force morphed from an innocent joy in having breakfast in Paris, and dinner in San Francisco into a numbers game. It would have been alright, but for the greed and sloppy handling of things that cannot be compromised, and must always be held close. We're in this together, and a mild jump in camaraderie and integrity would be nice. There is no relax, in the shop, on deck, or the ramp. Let the heathens choose colors for the FA's uniforms, but keep them out of the equipment, and all its supporting iterations.

This crew is not as derelict as some would suggest. Unfortunately they were a bit shy of what must be a new standard for all of us.

Fair enough, but the system you're describing suits a fighter, not an airliner. I hate to repeat myself, but the very fact that in the F-16 the human pilots had the option of ejecting in extremis makes the design considerations of the system as a whole very different.

The investigation seems to be pretty thorough so far and has in part led to the first collaboration between Boeing and Airbus of its type regarding response to stall.

Not to seem rude, but you've enthusiastically backed every theory from vertical stab separation to jammed stabilisers to inappropriate FD commands to try to point the finger at the aircraft.

I've got no problem at all with that, but the outright fiction (e.g "no UAS procedure existed") goes way beyond that, to the point that it can sometimes appear as deliberate attenpts to misinform.

Again, fair enough - but you only seem interested in pursuing a single line of inquiry.

There are no "patsies" in this case (and much as it may drive CONF to distraction, neither were there any at Habsheim). The aircraft was mishandled - the evidence to back that assertion up is overwhelming.

I am relatively certain that there is no drive to "protect" Airbus within the French aviation safety community. I'm equally certain that there never was. Like all other manufacturers, I have no doubt that their legal department is poised to argue the toss if necessary, but thus far there's been no evidence to suggest that there is. The BEA gains nothing from it - so why does this suspicion remain?

I'm hoping you mean Rio De Janeiro or Sao Paulo - otherwise we're talking about a completely different flight.

No-one (not even the DGAC) is suggesting that the crew were "derelict" - merely that the evidence suggests that they mishandled the emergency. There's a big difference.

Sticking my neck out, but this is why I think that when he said that he was almost entirely correct.

1. With the wing completely stalled, the ailerons would have been useless - no roll control, but maybe some adverse yaw.
2. At that sort of AoA the rudder power would be severely limited because (a) it was working at something like 70 deg effective sweep and (b) it was probably sat in the low energy wake of the stalled wing.
3. The elevators and THS were working.
4. When you roll an aircraft around the fuselage axis at high AoA you get an increase in sideslip and a reduction in AoA just from the geometry.
5. Changes in AoA produce changes in pitching moment and that in turn changes the pitch attitude. So there is a pitch/roll coupling.
6. At high AoA, as I have said before, the Dutch roll changes to a lightly damped or constant amplitude roll/sideslip oscillation. This will be accompanied by a corresponding pitch oscillation. [Check out the traces - the roll and pitch oscillations have similar frequency and much the same relative phasing throughout once the aircraft is stalled]
7. He would not be able to control the roll oscillation - with zero aileron effectiveness no amount of thrashing of the sidestick is going to make any difference.
8. He would not be able to suppress the pitch oscillation by using elevators because the motion would have been perpetually regenerated by the roll motion.
9. No wonder he thought he had lost control!
10. The only vestige of control left to him would have been a steady application of down elevator to reduce AoA, after which the other problems would disappear. Tragically this was the one option he did not try.

In the period from 02:10:57 through 02:11:38 they may well do, but the results certainly don't. As I surmised in a previous post the CL was hanging on as demonstrated by the synchronized roll/pitch oscillation. So when the PF made the statement, "I've lost control", he had realized that his inputs were having no effect on the outcome.

I see that Owain Glyndwr has made comments complimentary to the above.

Hi DozyWannabe,
http://www.blackholes.org.uk/PP/AF elevator2.jpg

The elevator basically follows the sidestick inputs up to 02:10:45 with deviations mostly around zero - which is what a sailplane pilot would expect. Thereafter, the FBW computers are attempting to maintain the pitch attitude requested by pf, despite the speed fall off and well beyond Alpha Max, by applying stab trim and lots of up elevator.

At 02:12:15 down ss has no effect - full up elevator is still applied.
After 02:12:30, despite nose down inputs, the elevator never reduces beyond -15 degs and for the most of the time, both elevator and stab trim are fully saturated at full nose up.

Since the crew were never aware of their control surfaces' positions (unlike a sailplane) - I'm not surprised that pf thought he had no control - because (fortunately) he could not pull back any harder even when he tried.

The PF didn't know that at 02:11:32. Until then the plane was still responding in pitch.

That said, you're right about the rôle of the control law. The response of the elevator to sidestick demands depends on the pitch rate and normal acceleration that the airplane already has.

When was Flight Idle selected ( for a few (10 ?) seconds only )? NOW this is said to be a part of the new joint Boeing/AB Stall recovery for aircraft with their engines mounted under the centre line or wing.

IIRC the pitch of AF447 was reduced during this short period, before TOGA was restored and the pitch went back up and remained up.

( For these brief moments AF447 WAS indeed a " stalled glider", with whatever advantages this might or could mean. Think of AirTrans, which was not stalled as I recall, but did fly as a glider to a safe landing.)

As a stalled glider of course the nose should be lowered, to get the wings working properly as usual. And then power restored to make it an aeroplane.

As has been discussed "STALL" appears not to have been recognised.

mm43, rudderrudderrat - cheers.

I know that in stalled air that the ailerons would not have had the desired effect, I was just trying to squelch the notion that the surfaces didn't respond.

I'm trying to remember whether the elevator traces are absolute or relative to THS angle, because you can see clearly the elevators beginning to come down when nose down is applied for a few seconds, first by the PF and then by the PNF - unfortunately too late in the sequence.

@gums, #803 thread 8

:\ Shame to me ! for mismatching the Viper limitations ! Sure it is hard-limited and very well protected inside its enveloppe in normal law.

Connection with AF447 :
@ Dozy
To be master or slave concerns precisely who decides when reaching the limit and the protection at the limit and if you can override it.

Those limits weren't plucked out of the air though, they were defined by the test pilots (a group of the best and most experienced in Europe) who went up in the prototype A320 and found them. So the "master" of the protections is still a pilot - just not the pilot in the seat at the time.

Apparently both.
Up to that time the HDG was constant meaning the PF was able to maintain the vertical FD bar in the center, but at 02 11 30 he applies full left stick (even the PNF is 'helping') but the HDG is increasing, the vertical bar cannot be maintained in the center ...

(…) je n’ai plus le contrôle de l’avion là
J’ai plus du tout le contrôle de l’avion

No control in roll
No control in pitch

I'm not able any more to keep the FD bars in the center.

What is terrible about that is that just before the FD bars reappear the PF had finally managed to stabilize the situation.
Sir, you've been criticized so much up to now, you was doing well, well enough to arrive in Paris, I just wish you took a big breath before following those reappearing commands ... I just wish those commands never came back.




mm43, I know I'm asking a lot, could you add to your graph the HDG, lateral inputs, ... in fact all the parameters.
Your graph is terrific, only a video animation would be better.

@CONF iture - you sound awfully certain of this. Do you have proof that he was trying to follow the FD or do you just wish that were the case?

If they'd followed UAS procedure the FDs would have been turned off and would not have come back.

@gums, #803 thread 8

:\ Shame to me ! for mismatching the Viper limitations ! Sure it is hard-limited and very well protected inside its enveloppe in normal law.

Connection with AF447 :
@ Dozy
To be master or slave concerns precisely who decides when reaching the limit and the protection at the limit and if you can override it.

Given that the limit is set pretty close to the point at which structural damage or loss of control is possible, why would overriding them be a good idea?

It is nevertheless regrettable that the BEA is not an NGO
His ultimate boss .. is the state
If for some reason .. a survey of BEA threaten the state ... the survey could be classified as top secret .. or state secrets
A perfect example is a Caravelle accident very long ago ...
Another example is the disappearance of a fishing vessel where a submarine is suspected
I don't think the BEA had free hands in those inquiries ... and indeed .. he gains nothing ..

From Owain,

2. At that sort of AoA the rudder power would be severely limited because (a) it was working at something like 70 deg effective sweep and (b) it was probably sat in the low energy wake of the stalled wing.

Pursuant to your observation, I would like to add that the Rudder/Vstab may have had little if any directional control as a system, and that the Rudder may have had an opposite effect to control deflection.

Would you rule out that Rudder may have reversed its effect, if not wholly, then intermittently?

The Rudder, at 17 degrees sweep from vertical, plus AoA, supplies just the sort of tangential bias from airflow from under the HS and even behind it, at this point? That the airflow past Rudder in this direction works opposite its intended force? Keeping in mind that the elevators, fully UP, do not shield this surface from under tailplane flow?

The Directional system is also in a transiently Stalled wake, that of the HS/elevators?

Perhaps off the wall, but the corner of the Rudder that exhibited damage was in this very unusual flow, and in effect, played leading edge to V/S system, collecting its drag from beneath and behind the HS/elevators? As such, it would be in vulnerable and undesigned for airflow, perhaps in some sort of leading edge flutter that corrupted the structure as seen in early photos?

structural damage are calculation marge , then first permanent deformation, then rupture. Why to override ? to avoid the ground for instance...

I think the limit load is set for the computers at a value that merely meets certifications for aircraft in general.

If I'm definitely nose down for the ground, I'll risk 2.75, or 3.0

I still remember reading Pierre Baud beginning 1988 explaining that the system accepted both SS inputs to be added...

Dozy, you open the difficult problem of the specifications. Has the flight engineer to discuss them ? It is from his responsability to ask and ask and ask : "Do you really want that ? Do you realise that means also that ? aso".

An example was Ariane5 V501 crash : not only the rocket crashed due to an single wrong carry, but the engineer said later that after the both (wrong) failures of the two inertial systems the computation of the path had to be stopped ! It was the spec ! He obeid without discussion... (8 billions FF 1996).:E

I don't understand how exceeding 67 degrees of bank or pulling into a stall will help you avoid the ground.

That is supposed to be an abnormal operation though. Normal procedure only has one pilot in active control at any one time, and if necessary this can be enforced through the override switch.

I have it on good authority that the test pilots and engineers (aeronautical, electrical and software) worked together to an almost unprecedented level, and nothing made it into the spec unless all groups had reviewed and signed off on it.

Now that *was* rocket science. ;) Very different disciplines involved with a very different design brief.

So simple that you get it wrong : Beside the load factor protection you might have other protections as well depending of the type of malfunction, and 2 levels of alternate law too ...
Also between ALT1 and 2 a confusing mixture of pitch anf roll control.

What is needed is to keep things simple :
Everything works fine - go for the normal law with all the grigri.
Something is in doubt - go straight to direct law to get rid of all those grigri

Your graceful degraduation is not wanted - keep things simple.

Not my airplane, but the following represents my thoughts on how the AF447 crew got so far out of sync with the aircraft.

Normal Law-makes sense
Alt 1 Law-graceful degradation, but the risk is that a heavily stressed pilot may not recognize his reduced protections. (NA to AF447.)
Alt 2 Law-graceful degradation-NOT.
It is an awkward stepchild. It is fairly clear from the narrative that Bonin never integrated the loss of protections into his flying. It would have been better if the aircraft had dropped into full Direct law. It is likely then that the stick response would have been sufficiently different that failure to recognize Direct law would be impossible. Direct law has the added benefit of the HS trim staying put at last setting unless changed by the crew.

It is fairly clear to me that the Crew of AF447 lost confidence in the flight control system's proper operation, and that doubt kept them from recognizing that they were stalled. The doubt began with the initial roll oscillation and escalated as the aircraft stalled. Flying in a unfamiliar part of the envelope was a contributing factor. (ALT2 at altitude)

When you put a guy in an unfamiliar situation, you are flipping coins as to how he will respond. Every so often you will come up tails. (i.e. the wrong answer).

Lyman

Well they were at least 250 kts too slow to get any sort of aeroelastic reversal and the sidelip/control deflections never got into the non-linear range let alone stall, and even stall is not the same as reversal, so I guess I would rule out rudder reversal.

Frankly, I haven't a clue what you mean by these remarks!

Don't be daft - take a look at a side elevation of an A330 at say 35 deg AoA. The HS sits well behind the fin LE and even if the HS were stalled there is no way that its wake would impinge on more than the very bottom of the rudder. And we know that the HS was operating with no more than about 10 deg AoA so it wasn't stalled and the elevators remained effective, so there wasn't any HS wake anyway.

As for the directional system being in a transiently stalled wake of the elevators, that is even more ludicrous since the elevator hinge lies at the same longitudinal station as the rudder TE at the fin root. No way that any elevator wake (even if it existed) could wash over the rudder.

You don't give up do you? That rudder damage was collected when the aircraft struck the water tail first and the APU was thrown through the top of the fuselage striking the rudder root TE. There is, and never has been, any evidence to support your wild theory of rudder failure in flight

@CONF iture:
 

Then keep it simple:

Normal, Alternate(PROT LOST) & Direct(PROT LOST)

Forget about the 'grigri' you still have in whatever Alternate....it states PROT LOST....that is what is the key in the ALt/Direct messages.

The 2 LAW/MODE you suggest is that referred to B. FBW? then you forgot to mention SECONDARY mode as degradation (i.a. no envelope protection) of the Normal Mode.
---

There is an simple and effective autotrim cancel 'feature'.... just hold the manual wheel.

I've been wondering what the Flight Director would command in pitch if (for whatever reason) it ignores the fact that airspeed is below VLS. I would think the FD would then command the change in pitch attitude that changes the current vertical speed to the selected value, i.e.:

ΔPitch=ΔFPA=ΔVS/TAS (1° =~ 700 fpm @ 400kTAS)

That ΔPitch, i.e. the angle between the aircraft symbol at the center of the attitude display and the FD pitch bar, was within ±1.5° in the 23 seconds after the FD became available. Would that explain the PF's sidestick inputs?

Exactly.

Assuming VLS not a factor, the FD commands are a function of the VS, not pitch attitude, (or FPA if that was the selection) in existence (which becomes the 'selected' and FMA displayed VS) when the bars returned.

This will remain so as long as there is no FCU selected change of the VS value, or vertical mode change thru pilot action, or the computations for the FD's again become unreliable.

As the airspeed decays at a given rate, the FD pitch bar will command a pitch rate commensurate with the pitch attitude required to maintain that VS (FPA) at successively lower airspeeds.

Eventually this becomes a losing proposition...

@HN39:
 

It remains Nz law, so the delta between NzCmd and NzAct = FD bar position. e.g. if you are on commanded trajectory the FD pitch bar will be neutral. (NZCmd = NzAct)

AMM:

Quote:

---

The pitch FD bar command is computed by using the measured vertical acceleration (NZ) and the NZFD command (pitch outer loop).
The gain and the limitation of the bar command are different according to whether HDG-V/S or TRK-FPA has been selected on the FCU.
With HDG-V/S selected, FD pitch control command is limited to plus or minus 22°5, and with TRKFPA selected,
FD pitch control command is limited to plus or minus 50°.

---

You need to calculate NzCmd from 'smoothened' SS input...

Max SS = +/- 16 degrees.

http://i474.photobucket.com/albums/r...99/SS_vs_G.jpg

Am i correct, that there is no such procedure, no such advice, no such recomendation and no such training for it?

Only in direct law it says "use manual trim only". But there, because the trim does the only correct thing, it stops working in automode.

OK465 and A33Zab,

Thanks for your replies. The question I have: In the time fragment shown, is the PF following the FD commands or is he chasing a pitch attitude?

http://i.imgur.com/jvWWZ.gif?1

@RetiredF4:
 

I know, the only refrence made is in

FCOM 1.27.00 Description:

"Mechanical control from the pitch trim wheel has priority over electrical control."

and

Flight controls normal law.

Automatic pitch trim is frozen in the following cases:

– Manual trim order
- ......

I've read that during training crew is prohibited to touch the wheel,
IMO that's a shame, they should know the dangers and benefits of such action.

To be exact:

In Direct LAW it is PFD amber message "USE MAN PITCH TRIM"
in backup (Elevator not available) the message is red "MAN PITCH TRIM ONLY"

A33Zab

in backup (Elevator not available) the message is red "MAN PITCH TRIM ONLY"

Where is the colon?

MAN PITCH : TRIM WHEEL ONLY

Nit pick?

HN 39:

If I interpret your graph correctly, your FD pitch bar plot represents the angular difference between the pitch attitude symbol (theta) and the FD bar, i.e. mostly within + or - a degree and one-half until around 72s.

If this is all referenced around your upper plot of a possibly active VS of +1500 displayed on the FMA, then you graph clearly shows that when aircraft vertical speed was below +1500 (48s-55s), the FD bar would be above the aircraft pitch symbol (theta), and vice versa when the aircraft VS was above +1500 (56s-67s), the pitch bar would be below the aircraft pitch symbol.

At 67s-72s the bottom began to fall out.

As to your actual question, since the trend for pitch is a generally consistent increase (not a specific attitude) with some minor variation (as would be expected control-wise with the speed decay) until the aircraft VS starts to go deeply negative, it sure looks like an attempt is being made to follow the FD until it can't be. I think as a practical matter the variations in SS input could be a result of the 'mushy' pitch response to the inputs as the stall is entered, as they are generally back and forth around neutral until the FD's are once again removed later. At which time the input is definitely only NU indicating lack of cues requiring pitch 'fine tuning'.

I hope I stated this clearly, and it is of course just an opinion...there may be other explanations which fit.

However, I've observed a number of pilots making their first attempts at PRM (precision radar monitor) breakouts where the FD's are initially turned off, but can as a matter of course inadvertently return or be reselected in other than the desired mode, and the results can be spectacular both vertically and horizontally, and they're not even stalled.

(Question: Where did your FD pitch bar info come from? Surmised?)

Calculated as (VSsel-VS)/TAS in consistent units, radians converted to degrees. Maybe that is too simplistic?

OK465

IYO, is this the first time in the last two minutes PF has chased his tail? Or is it possible it is a close reprise of what got him to 38k in the first place?

lyman

HN 39:

Other than the possible oscillatory FD effects of a variable rate of decay of TAS due to pitch inputs, it looks to me very representative of the precision associated with someone attempting to fly a flight director, to the extent of making the hairs stand up on the back of my neck.

You're the physics dude, I defer to you on the details. :)

(Lyman: You need to check IR #3 for when the FD's were actually available. My opinions only refer to the insightful graph. :))