Was not it B. ?

:E Give that man a prize!

Linktrained, marketing isn't the only powerhouse that can lead to changes like new software. Legal can also be most persuasive when they remark casually that not fixing these here bugs will cost the company billions....

But it's a REALLY hard sell for an engineer to get something changed on aftermarket equipment unless she can engage Legal.

Not really. All they needed to do was to:
1. Take a deep breath and watch what the aircraft did under the pitch and power that was set before the AP dropped out.
2. Make gentle corrections to ensure that altitude and heading were maintained within some reasonable limits given the turbulence encountered.

Instead, they panicked, pulled back on the stick, and, with the help of autotrim, pulled the airplane up into a stall.

I'm the wrong person to ask about that. Someone who is familiar with the actual implementation of the flight control system should answer that. Dozy?

There isn't a figure published but it's very easy to move it at say 10 times the auto trim rate.
The biggest advantage is that I now know where it is and if and when I moved it.

If thats true Ed, then any stall recovery proceedure for an Airbus must include immediate check of Trim status. Otherwise the gadget designed to save has also been inadvertently designed to quietly kill.

The sufficiency of the autotrim's response to any emergency pitch down commands would have to be immediate and adequate, otherwise the system "as a system " couldnt be safe /recoverable in an attitude upset.

It WAS... Wait for it... B !

Engineers can wave reports like BEA's in front of their favourite bean counters and ask how much has been budgeted for any Legal costs should we be mentioned, even in passing. Our Publicity Department may have ideas for "correcting" publicity requirements for years ahead. Ticket Sales... etc.

Safety is no Accident

"Ears popping"
 

Until quite late in the descent (if I remember an early post) the cabin was normally pressurised (8-9000 feet ?), so in spite of the climbing and descending (or most of the latter) there would be little or no chage in the pressure on the pilots' eardrums, until the pressurisation controller announced that it was unable to keep up with the rate of descent.
Until that moment, the crew would probably do their own version of the "Valsalva manoeuvre" - i.e just "click" their eardrums without doing the full "hold nose and click" business. At least that's what I remember doing almost automatically in the 1950s designed military aircraft I was privileged to operate, and those aircraft were pressurised to a much lower degree than passenger-carrying aircrtaft, so your eardrums acquired a fair degree of flexibility.

In any case, with all the other things their eyes and brains had to contend with, eardrums would be, to say the least, Low Priority.

@ A33Zab

I have a question on AOA data. The FDR records their actual values, which we get in the report. But wasn't these values fed to the ADRs, which in turn were rejected? Where did the PRIMs get the data from and how did they elect the data as correct to trigger a stall warning? (Shouldn't AOA be correlated -at least to a certain degree of incidence- with airspeed, which was unreliable?)

Hi Smilin Ed,

I generally share your views about autotrim but we have to acknowledge that many pilots love it and find the bus safe, easy and pleasant to fly.

When I was teaching pilots converting from Airbus to the 777, they had to adjust during the first sim sessions about having to trim and a few lesser handling problems. To those who complained I said from now on they would have to earn their living honestly :). Whilst I am sure manual trim is right with a classical control column and wheel, I cannot say it’s the same with a sidestick controller a la Airbus.

Choosing “C*” law implied autotrim: you cannot rely only on elevator to compensate for speed, configuration, or thrust changes while maintaining a constant path stick free. Besides, streamlining THS and elevator improves aerodynamic performance. The 777 uses a “C*-u” law in normal mode and does silently autotrim when compensating for configuration or thrust changes but you have to “trim to speed” to compensate for speed changes. In another design, Gums had to trim the Viper to “gee”. None of these designs preclude implementation of powerful protections at both ends of the flight envelope.

For Airbus, longitudinal static stability (LSS) is restored by software in normal law when approaching the limits. I believe this was one of the Special Conditions of certification and was called “proof of concept” at the time. The problem which I think was disregarded is the lack of LSS in ALT2 law. Imagine the THS could not move further up as soon as the stall warning activates: most probably, even with consistent full back stick, the AoA would not increase enough to the point where the stall alarm is silenced.

It is Dozy Wannabe’s opinion that the THS rightfully moved to 13 deg up in AF447 because it’s what the PF requested. It is apparently not Dozy’s opinion however that the THS should continue to trim up in normal law when high AoA protection is active and the pilot pulls on the stick, since the software prevents it. Where is the logic?

I don't have any numbers, but the question is not so much the speed of actuation (which would be the same), but the speed with which that actuation is demanded. You can crank the manual trim wheel round to neutral in a couple of seconds, and the actuators will catch up, but to do it via autotrim will take sustained nose-down inputs over several seconds, which appears to be something that many pilots aren't comfortable with doing.

Autotrim is designed to take the load off normal control inputs, not recover from extreme THS angles. I think to correct the THS angle with autotrim in time to effect a positive recovery would have required recognition of the situation and corrective action at a considerably higher altitude than with manual trim operation.

Concorde of course had that, even though it wasn't a 'cartoon', but a separate electro-mechanical instrument (the "Icovol"). It showed the position of all eight control surfaces, and the control mode for each (blue or green electric signalling - aka FBW - or mechanical).
And yes, Concorde also had an AoA indicator.....

Airbus will also display that on the ECAM FLT CTL page. (but you all knew that, right?... Is the suggestion for a separate screen always displaying this?

Pitch control and longitudinal stability
 

Hats off the Rudderrat for pointing out one of the unusual characterisctics of the FBW concepts that emphasize Gee versus AoA for pitch laws.

The stability that Smilin'Ed describes can be implemented quite easily in the FBW systems, and the plane tries to achieve the TRIMMED AoA versus the TRIMMED gee if you relax pressure on the stick. The system can also make the "move" aggressive or rather wimpy.

Both the 'bus and the Viper engineers went with the "trim to gee" approach, which has some problems, as the Rat describes. The 'bus actually compensates for pitch in Normal Law by adjusting the trimmed gee when not in level flight (not so the Viper, as it goes to whatever trimmed gee the pilot sets - minus 2 or so, up to plus 3.5 gees) . So with the 'bus in a turn of 30 deg back we get a trimmed gee of about 1.1gee for neutral stick. Kinda neat. In a climb of 30 degrees, we would get a trimmed gee of 0.87 or so. but here's the biggie.....

If the system loses the pitch/roll correction, then it looks like it seeks one gee. So RudderRat's observation is very germane. You can wind up with a plane trying to increase pitch even with a neutral stick input. In fact, it's an aspect of the Viper logic that "helped" us achieve the deep stall none thot possible. If you zoom at sufficient pitch, you can "fly" past the "protections" on inertia and have a plane incapable of introducing a useful nose down pitch moment, especially if the c.g. is far enough aft.

My personal opinion is most pilots would prefer the "trim to AoA" approach. The plane would closely resemble, if exactly "feel" and act like older planes. It would "unload" the gee if you were holding back pressure commanding an AoA above the trimmed AoA, and exploit the "natural" aero characteristics of the plane - really great statically stable or less stable as with the Viper.

I guess the engineers decided that the "gee" command approach could be implemented more reliably/easier using internal sensors of the FBW system itself. But even so, the Viper used the AoA probes for limiting our AoA, and it worked very well.

Test Pilots
 

Let's face the facts that led to this tragic accident:

1) A captain, knowing that in the next hour the plane would be in a more dangerous portion of the flight, ITZ transit, decided to take a rest break. He suggested the PF take the break, but changed his mind when pilot flying says he isn't tired. Then, instead of giving command to the fresh (and more experienced) pilot, he gave it to the 'baby pilot'...even having to ask if he had the correct license to be the Pilot-in-Command! Still find it interesting the first BEA report says he didn't have the right license but the third does...sure that was just a clerical error. :confused: POOR COMMAND DECISION

2) When the autopilot/autothrust kicked off, the plane had a nose-down (or at best level) attitude with reduced engine thrust because of the compensation the autopilot was making in that moment for turbulence. Pitch was not correct, power was not correct so doing nothing would not have been acceptable. The pilot needed to make two immediate corrections, one to correct the roll and the other to correct the slight nose down (or flat) attitude. If he followed standard pitch/power memory items where he would have leveled the wings, put 4-5 degrees of NU and added the bit of power the autos reduced seconds before, he would have been fine. His continued nose-up inputs, even as he is saying he is going back down, shows a very poor understanding and control over the primary flight control, the side-stick. POOR TRAINING

3) Problem came with over-control in BOTH axis. His over-corrections were noticed (in both axis) by the PNF. The PF did not correct the situation and being 'commander' refused to release control to the PNF who seemed to have a better grip on the situation. POOR TRAINING

4) Just before the (real) stall, the airplane was fully recoverable with proper procedures. Again, despite the stall warning and advice from PNF, the training of the PF was not up to the task of simple stall recovery i.e. GET THE NOSE DOWN. Of course, he might have reverted to his 'Normal Law' stall training, holding stick back to get max performance. Too bad he was in 'Alt2 Law' which doesn't give those protections/help. POOR TRAINING

5) After that point, the pilots/aircraft are into the 'Test Pilot' regime. NO ONE could have known the performance of the aircraft in this flight envelope because no one is stupid enough to put this aircraft into that position to get that data. The whole point to training and hiring competent pilots is so they NEVER get into this situation to start and if they do, have been trained with basic flying skills and upset recovery to have a chance to save the plane/passengers and their own lives. NONE of the three pilots on the flight deck at this time had a clue what to do. VERY POOR TRAINING/HIRING/PROMOTING

My take on this accident is the pilots, especially the PF, gave us one of the worst "FAILS" in aviation history. The plane can be blamed for the initial problem, but many crews have walked away from identical incidents. In this case, very poor training on flying at altitude, no teamwork between the two pilots, clueless about the true nature of the emergency etc... If it were me, I'd put 10% on Airbus for the pitots and the way the stall alarm functioned. The other 90% has to go on the pilots which really means Air France and their training/hiring/promotions...

After reading the entire three reports, no way you are getting me on an Air France flight be it an Airbus or Boeing craft... I can accept one pilot messing up under high stress, but for ALL THREE to remain so clueless indicates a BIG problem within Air France IMHO.

question for Mimpe?

Can you expand on that "quietly kill"? The way I see "Trim"'s function is a smoothing of flight in longer time frames than when needing to Manoeuverr? In other words, say a Pilot is in a situation that requires "maneuvering" more than cruising, wouldn't the Trim work against him?

As in, favoring his inputs NOSE up? Damping his nose down elevators? With a large trim nose up, his Pitch responses would be out of his comfortable rate and trend experiences? And also be variable as the THS was in transit? just a thought

Wouldn't Trim be far down his list of "controls"? in an emergency.? Why would he want any additional inputs at the tail other than his own, even if he was aware they were happening.?

The simple fact that the Trimming Horizontal Stabiliser is manually controlled by a central wheel that requires total focus and many repetitive (read, time consuming) and large hand excursions seems to be an awkward step especially in a confusing and emergent setting?

ALTERNATE LAW: "STAB TRIM" NOSE UP 7

OR TRIM : UP 7 or 6, or 10, or 13

OR TRIM :DOWN 3

Again, would a pilot even want turtle trim in his quiver in UPSET RECOVERY?

The main difference is that in Normal Law, all the inputs to the FCU and FMC can be trusted - therefore it is likely that the computers will know more about the plethora of aircraft parameters at any given time than the pilots, who are relying on an instrument scan and therefore only a subset of available data at any given time.

Alternate 2 is only triggered when one or more of the flight surface, sensory or computation systems have failed. At that point the system is designed to acknowledge that the computers' ability to assess the situation is impaired, and give complete control to the pilot. This presumes that the pilot knows his aircraft and how to handle it (which makes the assumption that Airbus designed their aircraft systems around the idea that pilots are largely numpties not only wrong, but pretty ironic in hindsight).

In essence, when you get ADR DISAGREE and Alt 2 clicks in, the computers are saying to the pilot "I can't trust these readings and I don't have your problem-solving skills. You can not only see the instruments but can also see outside. I'll help you keep the shiny side up based on what I know, but if I'm wrong you can override me". Unfortunately in this case the pilot made control inputs way in excess of what was appropriate, overrode the "soft protections" in doing so and stalled the aircraft, apparently because he was not appropriately trained.

Hi DozyWannabe,

You made an interesting observation about pilot instrument scan and I agree with you:
but I disagree about the help the automatics will give:
So the pilot only has a subset of available data with UAS, he does not have the benefit Longitudinal speed stability in ALT2 Law, he has to keep it shiny side up himself with Roll Direct (which he appears to have found "very twitchy"), and the auto trim is silently trimming the load from the up elevator well beyond alpha max.
What a magnificent design!

He has pitch and power, which should be all that is necessary.

I'm still not sure about the argument there - the flipside is that in the case of a small-intermediate upset he has complete control of elevators and trim in one hand and can manage throttle in the other.

Twitchy my foot. Control at altitude is about very small, corrective movements until you get a feel for how the aircraft is responding (even this "know-all", "office-chair" non-pilot knows that much - and no, that is not directed at you... :)) - he was mashing the stick halfway to the stops from the get-go, so much so that the PNF repeatedly chastised him for it.

I don't know what JD-EE intended.
The 'Icovol' was hardly a primary flight instrument, but then of course Concorde may have been FBW, but it was not a glass cockpit, so a space for it was found on the central panel.

I've refrained from commenting on the autotrim issue, because the only autotrim I know was the Concorde one.... and that was a separate computer, and an electro-mechanical actuator, that moved the neutral point (in pitch) of the stick. And autotrim also moved the manual trim wheel, with the obvious 'ping', 'ping' of the bicycle bell.

As an ancient, I'm still somewhat astonished about the lack of "feedback" from the aircraft to the pilot, which seems to have disappeared slowly....

No 'stick force', other than a side-stick spring (crikey, even MS Flight Simulator and other games now come with force-feedback joysticks).

Concorde had FBW, so no direct force feedback from rods and cables either, but it had a quite sophisticated 'artificial feel' system and I've never heard complaints about it.

No 'throttle position feedback' - on Concorde the autothrottle computer controlled an autothrottle actuator, which moved ... yes, the throttle levers themselves.

No 'bicycle bell' on the pitch trim.

Concorde of course had André Turcat..... who stuck his nose into everything, including a lot of aspects of the "pilot-aircraft interface".
For those of you who read French, read his book "Essais et Batailles" (sadly it's never been translated).

I tend to think the FBW family (A320, etc.) never had the benefit of 'ancient' pilots like him.... but I may be wrong.

@ChristiaanJ - the A320 did in fact have Gordon Corps.

Also, MS Flight Sim provides a rough-and-ready simulation of everything from single-engined Cessnas (where feedback comes directly from the surfaces and is therefore appropriate) to the A380 (where it arguably isn't such a big deal). When the force-feedback on a computer joystick or yoke malfunctions or dies, you can hit Alt-F4, get up from your chair, go to the store and get a replacement. If force-feedback were to go wrong at 35,000ft your options are more limited.

@Config Full:
 

- The stall warning is processed in FWCs not FCPCs
- It's the ADR itself which sets AOA to 0° (SSM=NCD) if CAS is below 60 Kts.

(A. did already changed that with 'BUSS' option, AOA is always available [only SSM will be NCD if CAS <60Kts])

- In ALT/DIR LAW the AOAsw triggervalue is 10,8°, this value is modified by a value which depends on MACH (and in other flight phases by S/F Config and Speedbrake setting)
The minium AOAsw triggervalue is 3,8° @ or above M 0.82.

Sorry, I know the name, but I thought he was Servanty's British counterpart. I'll do some Googling.

I get your point.... but as I said, I can't remember any serious issues with the mechanics of the Concorde artificial feel, which was a lot more complicated than the force feedback on an MS joystick.
My point being, really, that implementing reliable force feedback on an aircraft sidestick is no more difficult than what we did over 40 years ago.
And if one 'dies', there's always the other side.....

@DW:
 

Hi DW,

It depends what is understood by LLS, indeed you loose the protections but as long Nz and Nx are available there is speed stability due to thrust changes.

In fact you can see it in the 'simulated' graph in the 3rd report.

@ 00:09:14 THS 3.0° @02:10:09 THS 2.8°
@ 02:10:09 THS 2.8° thrust went down & THS moved to 3.1°
@ 02:10:30 Thrust is going up again and THS retrimmed to 2.9°

Hi,

IMHO .. it's not a way to argument for justify the non presence of an artificial feedback
If the brakes go wrong when landing your options are more limited
This can be an argument for no brakes at all ?
So .. as usual .. more we are .. more fun

About Planes and "flying Systems" Design
 

We could imagine his leadership and role, to the (technical) success of the fantastic bird.

When the crew in some machines (by design) are considered almost an accessory* to it's operation why they should invest in interfaces and feedback? :E

An "accessory" also convenient after "multiple design faults" allowing the publishing of "operator error" (citing lack of training, etc).

(*) Just exaggerating to express how i feel after reading AF447 published "circonstances" reports.

THS speed rate:
 

I’ve got these figures of THS rate.

The THS itself (the ballscrew) is driven by 2 hydraulic (B & Y) motors.

Max. Operating Load and Max. Speed (2 motors)

18940 daN (42578.81 lbf) ----- 0.4°/s
16950 daN (38105.11 lbf) ----- 1.0°/s
6770 daN (15219.56 lbf) ---- 1.2°/s

Limit load (both compression as tension) is 32500 daN (73062.9 lbf).
I assume this limit load is mentioned as being the limit for the Hydr.
motors to drive the THS.

Half speed for 1 motor (1 hydr. B or Y system failure) operation.

The rate for the BEA mentioned 1 minute to go from 3 to 13 ANU was 10°/60 = 0.16°/s.


For going back to the 3° ANU position:
The hinges are on the trailing edge and therefore the aerodynamic
load on the drive spindle is in the AND direction it would take 12 sec. to travel from 13° back to 3° ANU.

The manual trim wheel THS displacement is ~0,65° a stroke, if one need
1s to complete the stroke and 1s to re-grab the wheel, the rate will be 0,325°/s.

All of this didn't need to be known, just a hand on the moving handwheels would have been sufficient to cancel the autotrim orders.

active artificial feel.
 

I understand that the mixed mode (PITCH NZ / ROLL DIRECT ) will not be easy to handle but wasn't the idea behind Nz to use the SS more or less as a toggleswitch?
toggle NU > release > toggle ND to cancel the order?

So no need for an active artificial feel for a toggle switch.

From own experience it requires considerable force to move and keep SS to full aft and fwd stop.

http://i474.photobucket.com/albums/r...99/GeeA330.jpg

A33Zab

Could the limit load be structural? It is suspiciously similar to K/2, a common critical failure constant?

Could you clarify "toggle"? There are many types. A bump toggle would satisfy your meaning here (above)?

A33Zab,

Let's not confuse the issue.

IIRC, the AB SS provides a 'proportional' signal, from either pots or resolvers, and NOT a 'toggle' signal like the ancient Atari joysticks (LOL).

So your graph and post may be somewhat misleading to the uninitiated....

CJ

Thanks, that was close call. I came very close to almost understanding what he wass saying.

@DozyWannabe.
DW, I understand your point that in ALT2 the computers are missing some handles but to lose s/w would require triple AoA sensor failure or triple ADR failure or double FWC failure (and btw the UAS check list asks to respect stall warnings) so I still cannot see in what way preventing up-trim based on s/w AoA could be harmful.

Test pilots
 

PuraVida post # 1726

Perfect analysis! Totally agree.

@ChristiaanJ:
 

Trimming Is Not Hard
 

Dozy: No, I have not flown an airbus. My problem with autotrim when hand flying any aircraft is that I may not want it to follow my control inputs. Unfortunately, this crew seems not to have known anything about trim. Maybe having the trim announce to the crew that it is moving would have been a wake-up call and someone would have said, "oops we don't want that," but I doubt it. It certainly would have got my attention. Sadly, the men seated in the cockpit were completely out of their element.

Pilot incapacitation is not much of an issue in a multi-seat cockpit. Trimming is not difficult and, for a pilot worthy of being called a pilot, it is a natural activity which is so routine that it is done almost unconsciously. I have flown aircraft with the pitch trim on the yoke, on the stick, and on the console. I never found it a burden to activate any of them. I prefer it on the stick or yoke since that does free up the other hand, but the difference is really not material. What I don't want is having the system changing the trim when I want to remain trimmed for a particular speed or AoA. In the case of AF447, the system gave up and turned it over to the people seated in the cockpit, except they didn't turn over everything and I believe that situation makes it more difficult to hand-fly. Autotrim has its place, but not in my cockpit when I am trying to hand-fly.

airtren, you can lose lift without stalling.

Try it sometime.

Fly along with your flaps down. Then, raise the flaps. You'll lose lift, but you don't necessarily stall. You'll probably also descend, if you were flying level previous to the flap raising, unless you take the time to also add power ...

You only lose lift if you maintain attitude. If you pitch up with flap retraction you maintain the same lift until you stall.

Did the artificial feel include airspeed as input ? - intuitively I feel it must do (and I believe B777 does), but deltas are very different so maybe it wasn't needed ?

If so, what did the system use in the event of uAS ?

On a vaguely related note, did Concorde ever suffer from pitot icing / similar uAS or at FL60+ was it simply not a problem ?

S/W can be spurious, particularly in uAS, but also, I believe, with one AOA broken (it would be outvoted by the other two for the PRIMS, but would still trigger SW - think Perpignan but other way round).

Stall warning is a warning to the pilots to asses the situtation and take action. It is not a certainty based on which the plane itself should take action or restrict the pilots action - the FBW Bus has all that in normal law, but if you get as far as s/w then this has already failed. Or in other words, if you get a s/w on a bus, the plane is already uncertain of its data.

The logic running through is that automation doesn't act (or restrict pilot actions) on known-bad data - but it may still warn pilots. Sadly in this case, I think the pilots believed the stall warning to be spurious due to uAS, and possibly believed they were overspeed and acted to correct that instead.