I translate the original as follows:-

"The position of the rudder limiter of the rudder (RTLU) is frozen at 2 h 10 min 04.5 and the unavailability of the rudder travel limiting function is recorded between 2 h 10 min 17.5 and 2 h 10 min 18.5."

Do you believe that the wording used in the French version effectively means that, "The RTLU is locked at the allowable deflection for the last valid Mach speed"? If so, that statement matches the operational description of the device.

In respect of the earlier variations; would seem that there were some short-term Mach fluctuations prior to the defining one.

Hi mm43;

Thanks for sharing my curiosity. I think your translation is perfect and, yes, I believe it means what you say (i.e. the 'freeze & latch' corresponds to the CAS & Mach at 02:10:04,5). The "short-term Mach fluctuations prior to the defining one" must have been pretty large, and must have come from a different source than the recorded CAS & Mach. Then there is the value itself, which doesn't correspond with IR #2:

Hi HN39;

Thanks for the reply. I will answer the PM in due course.

Looking at the Yaw Damper trace, I believe that the maximum rudder angle allowed by the Yaw Damper (+/-4°) was not exceeded, but whether the damper was acting correctly is another matter.

The DA displayed by the TLU is rather puzzling. Perhaps the TLU angle doesn't represent the actual rudder angle, but there doesn't seem to be a valid reason for that to happen. As the AMM doesn't shed any light on the situation, we may just have to wait for the BEA to explain exactly what was happening.

As you have already pointed out, if the deviations shown in the TLU traces are to believed, where did this Mach data come from?

Having searched the various definitive sources and found nothing, I reverted to the Cathay Pacific notes originally compiled by Andy Tracey and came up with this for Yaw and another for ALT-2.

EDIT : Basic control schematic
http://oi40.tinypic.com/2hncsad.jpg

@Lyman
"His Pull was additive to an input made by AutoPilot"

Forget !
Impossible . :ok: You know : BZ said : pilots are only housekeepers, cow-boys or fools ! It would be a real algorithm bug in AB philosophy to add a computer and a human input !:=

BZ , himself , a cow boy , cutting telepheric lines in Alps with a Mirage 3 .

Short memory Bernie ! :ugh:

For the uninitiated, is that post supposed to mean anything in particular? :ugh:

In the 60's Bernard Ziegler's F84 collided with the line of a cable car in the Alps, killing 6 people. It's routinely dragged out to serve as proof that anything Airbus is inherently evil. Link

@Zorin:
Ah! The light dawns. Thanks!

Bon Temps Rouler....

ne plus plus. At handoff, the airframe was rotating up, as the ascent diminished, (The PITCH on PFD showed 0, four degrees 'low'). My surmise is that the PF's initial NU input created an unwanted PITCH moment (Normal Law, protected), and that caused the STALLSTALL, (the first WRN, a 'transient'). Still sufficient for the highest g accel, though.

Lyman, for what it's worth ...

IF you are a pilot flying on instruments,
AND
you (or your friend HAL) make a pitch input that causes the nose to go up ...
AND IF the nose goes higher than you intended it to to go when you or HAL made that input
THEN
you use your hand on the flight controls to make an immediate counter correction that is usually accomplished in two parts (particularly if your task is to fly smoothly in order to keep your passengers as comfortable as you can)

a) first you stop the pitch rate (and you watch all of this on your attitude indicator/AH) and
b) make a second correction to return the nose to the pitch you actually desire for the flight conditions or performance you wish to complete your mission, or the segment of the mission you are on.

As an adjunct:
IF during this process you find your aircraft to be rolling beyond the AOB you desire (which I think would be "wings level" or "0" degrees AOB)
THEN you use that same hand to stop roll, and work the wings back to level via small corrections, or one larger correction. Given what A330 experienced pilots who have posted here have shared, at that altitude the small corrections might be the better technique, as they adivse us that the A330 is sensitive to input in that flight regime when hand flying.

ASIDE: Any pilot who cannot correct for pitch and roll at the same time by referencing his attitude indicator cannot be considered to be qualified to pilot a passenger jet. I am sure both of the pilots in that cockpit had demonstrated time and again that they could make such an input in two axes successfully. I seriously doubt AF would have hired them had they not been able to. The answer to why neither did in this instance, particularly the pilot who handled the controls at the handoff from HAL to Human, remains incompletely answered to my eye, other than to consider comfort with the aforementioned sensitivity to input. What did he see?)

So, even if the plane at hand off was in a non stable state -- which is part of your point -- a change in nose pitch, or a continuation of a nose pitch change in one direction or another is not necessarily a fait accompli if the pilot whose hand is on the stick flies in accordance with standard instrument flying principles, principles he had previously shown he could apply consistently. (For this and other reasons the "pitch and power" chorus has been singing in three part harmony on these forums since about 01 June 2009 .)

Your analysis of what the pitch result was, in and around that point in time, must to be complete (IMO) include a consideration of what flying performance was desired at the time.

At the altitudes in this event, 11 degrees nose up borders on an unusual attitude, for the flight condition and peformance desired on that leg of the mission: straight and level.

"The aircraft did not immediately respond..." BEA

This plays Hell with your primer, Lonewolf. Notwithstanding the Pitch was four plus degrees low on the pilot's AI. The STALL WRN suggests an AoA of extreme value, v/v cruise, and the a/c was maneuvering out of phase with its attitude, eg, Nose Low, and ascending.

At the very least, we suspect the airframe was not consistent with a ho hum cruise, and the a/p (imo) quit due controls/response out of limits. To include airspeeds discrepant: either disagree or simultaneous migration past a/p limit value. (30 knots < second and continuous). See the Mach variations as described above.

The only aspect consistent with a STALL Warning, at that speed, is AoA/Mach disagree, which is consistent with the accepted wisdom, here, as I see it, but not enough acceptance/credence is lent to the conditions extant at a/p loss/handoff/HAL malfunction.

It is troubling that without the conditions leading up to handoff, so many here have arrived at conclusions. You suggest a lack of skill is responsible.
The DFDR is reporting Inertial data, not what PF or even you would be seeing at the beginning (or 'continuation' of UPSET).

It is vital that the conclusion be exculpatory of the Bus. If due Weather, then the a/p v/v limits in stink is suspect. If a/p exceeded, the Bus is painted as poorly designed. How handy for a baby pilot to put them into the wet.

My starting point is that the crew inherited an upset airframe. The autopilot's limits are well past what is technically accepted as "Upset" in the regs. So naturally, ICE and inexperience are the uniform of the day.

The story is here, in the start, not at the (actual) STALL.

What blows my mind is your willingness to set aside the fact of (at least) 37 previous incidents of ice-caused UAS using a known-to-be-defective pitot. Instead you become an intellectual contortionist, sir.

There are plenty of confusing factors to this accident without all this stuff you're coming up with.:sad:

Lyman, taking a plane that is not stable, if HAL did something odd, is roughly the same as taking the plane from another pilot who is flying badly or has vertigo. If you fly in multi-crew aircraft, I run with the assumption that the above is one of the tasks that you are required to do, and trained for.

If you are not trained to do that, one wonders "Why not?"

I refer you again to my old refrain: what did he(PF) see? Though the betting money is on all of the AH's working, the PNF did switch sources to a different inertial gyro for one reason or another.

What was that reason?

More than one possible answer, and since they didn't verbalize why that was done, the why is left open to some question. Those who have flown the A330 may be able to offer better estimates that I, so I won't attempt to.

Nothing further.

Out.

The PNF was apparently looking at all three displays between 02:10:25 and 02:10:36:
Subsequently he made following selections:
02:10:40 F/O on ADR3
02:10:44 F/O on IR3
02:12:15 CAPT on ADR3
02:12:19 CAPT on IR3
02:13:25 CAPT on ADR1

Logically, PNF wanted to be sure that PF was flying with information that looked good to PNF and so that he, PNF, could be sure of what was appearing on PF's panel. (Since monitoring the PF's PFD was a bit of a stretch for PNF.)

Obviously PNF trusted the ADR3/IR3 data and likely didn't understand why PF insisted on flying around with his nose in the air.

Then later, to ensure that he could also conveniently read the ADR3/IR3 data on his PFD, he switched his displays to the same data sources. (Despite the loss of redundancy.)

That is my take on these actions. Sorry about all the acronym soup.:confused:

It seems to me he primarily switched the Air Data source, just taking the Inertial Reference along without any particular reason.

Thank you both for your thoughtful estimates. :ok:

Popular Mechanics: What Really Happened Aboard AF447
 

Popular Mechanics has a new article on its Web site today discussing "What Really Happened Aboard AF447." The article includes the final conversations of the crew, along with analysis. According to the PM version, the captain does finally realize the plane is in a stall, and in the last seconds tries to point the nose down, but they run out of time.

Print - What Really Happened Aboard Air France 447 - Popular Mechanics

I disagree. Having read the book, I commented it here.

The "picture" in Otelli's book indeed adds some points to the data contained in the BEA's last interim report, but on the other hand it lacks of data/information on some other points.
=> Describing it as "fuller" is in my opinion a bit of an overstretch, and carries implicitely the impression that some things are "hidden" by "the authorities" (BEA), which constitute a polemic, not a fact.

---------

The quoted article lacks some nuances (which can be of importance, when really wanting to understand). Let's try to correct some of that:

"It's better than nothing." feels pessimist. The exact translation for that exists and is used in french: "C'est mieux que rien". That's not what the pilot said.
"C'est toujours ça de pris." is optimistic. They turned on the anti-ice, as a precaution against potential icing conditions. They didn't turn on the anti-ice as a mean to try to limit an already bad known problem.


In french, "éventuellement" means eventually in the sense of "possibly". Not in the sens of "in the end".


The BEA report differs :
2 h 10 min 33: (PNF) Selon les trois tu montes donc tu redescends
=> According to the three you’re going up so go back down
"Les trois" (the three) is important, it means the PNF refers to 3 indications (either the 3 ADIs, or the ADI and the V/S and the altimeter)


...

I leave it there, just wanting to try and convince you:
1/ to read the article (it's interesting in quoting extensively Otelli's book, only published in french AFAIK)
2/ but to exercice caution. The advertised "full transcription" (and implicitely "full explanation") is marketing to sell paper... ;)

Cheers
AZR

At the crucial moment of a/p loss, the airframe was climbing, and the Nose was Pitched down. Added to that, the bank angle was about 8 degrees Right wing low (turbulence?). The a/c had it right, they were in a robust upwelling, an updraft, and the Nose was commanded down, though the a/c still climbed. We do not know for how long the a/p had been maneuvering in this fashion, though it can be said that the pilots themselves, in discussing temps and calling the back for a HU on turb tells us it was not new.

It is this possibility that has been actively ignored. I believe the PF made a correction in Pitch that was perhaps the opposite of what was needed. Climbing in the airmass, a NU creates more climb, in g and in VSI.

Much is made of their presence in a cell, though the possibility of a robust UD is dismissed?

The column may have been 120 knots. If so, this plays havoc with airspeeds, AoA vanes, and 'feel'. The airspeed in descent could have been twice that reported by the DFDR. "I think we have crazy speed..."

As with any epic disaster, those whose fingerprints are on the investigation call the shots. Though BEA have not done so, the focus is on the 95% of what is utterly irrelevant to the disaster.

What caused the climb to Stall is virtually ignored, and when addressed, data is ignored, and the audience is expected to sign on.....

I think PF did fly attitude, more's the pity. And Power was not a player, not at first.

Lyman claims that the aircraft was climbing, nose down. I have two questions:
1.Is that supported by the reported data?
2. How is that even possible?

No, the V/S was going through zero, decreasing
It is possible, if the airmass is rising fast enough.

Wings were level, rolling right, 8 deg 2.5 sec later.

Hazelnuts39 Pls. take note of the VS and the Attitude, two seconds prior to a/p loss.

Note that in two seconds the actual VS had changed from 500 fpm up to a like number down. That is remarkable, and the a/c was at what weight?

The Nose was in transit as the VS reversed, and they were not in tune.

I think the a/p was attempting to rate the UPDRAFT with elevators, and it eventually required too much ND (v/v AoA) to keep up (level). The Screen reported ND at handoff. Not Level.

Have you a reference for the deflection of Vanes due airmass? That geometry is beyond my current take. Something was causing the Vanes to read crazy high, and it was not 'cruise Nose'.

The change corresponds to a normal acceleration of 0.85 g, well within the ICAO “light” turbulence criterion of 0.5 g peak to peak. (3.5 seconds of normal acceleration of 0.85 g changes V/S downwards by 1000 fpm.)

The vanes are shown on the same page 42 as AoA IRS1,2,3.

Relevant graphics from the BEA's IR#3 that I posted on page #27 of this thread are available as a "cross-hair version". Enlarged with additional scaling and timing marks, the chances of getting the actual up/down rates etc. will be improved.

Really nice one, mm43,
Unfortunately, can't save it in an off-line format, and the cross-hair pointer doesn't show any data.

Give us a clue on how to use it?

CJ

@Christiaanj;

With the enlarging of the graphic, additional scale marks have been added, along with an overlaid time scale in each section. The adding of the cross-hairs enables a closer examination of any points of interest with the x:y co-ordinates now easily readable.

To save for offline use, right click outside the graphic (pale blue background) and Save As (webpage complete). The javascript file that generates the cross-hairs will also be saved.

Hmm - if work lets up I might have a play with that JavaScript, see if I can't tart it up a little. :)

So - based on mm43's web graphic we have:

• The pitch attitude *bang on zero* at disconnect
• Importantly however, trending nose-up
• Followed by the last A/P command bringing the pitch attitude to about +1.5 degrees NU
• A wobble in V/S around the time of disconnect, but *well within the tolerances for "MODERATE" turbulence*
• A very clear difference in elevator response from 02:10:08 onwards, possibly indicative of manual control
• A/P elevator control has a max NU of 0.6 (correcting for the bump beginning at 02:10:00) and a max ND of 0.5 - compare this with a max NU of 2.1 and a max ND of 0.6 under manual handling

I still see an aircraft that was relatively stable for the turbulence they were in until the PF starts overcontrolling.

How stable is this Airbus? What would happen if the PF wouldn't touch the controls for a while until all the instruments would be functioning and a/p could be switched on again?

In the two seconds between A/P disconnect and the PF's first control input, the FCS commanded elevator nose-up which returned pitch to almost normal and normal acceleration to approx. 1 g.

@ mm43
Excellent work, i hope we see more of it, especially the times when the stall warnings happened.

I tend to disagree.

At the very moment, when the autopiloit disconnect occurred, we have the following values:

-vertical speed zero with tendency to decrease
-normal acceleration .85 g`s (@mm43, i think your labeling is off by one, the line you labeled 0 should actually be 1 g)
-pitch attitude 0°
-AOA 2.25°

The moment the AP drops out and stick is in neutral , NZ-Law will maintain flightpath / 1 g with reference to the flightpath, the aircraft was in at AP disconnect. (Whatever position intended by the AP before does not matter any more).

That would be a descent under the above recorded parameters.

franzl

RetiredF4 but if you extrapolate the rollrate of 4 deg/sec to the right between 2:10:05 and 2 :10:07 you will only a short time descent peacefully before you dive.....

RetiredF4;

I simply read the traces. After Clandestino in post #393 (p.20) I'm not so sure about Nz law maintaining flightpath / '1g' in turbulence.

The point I am making is this: Whilst the airframe accelerated upward, the Nose Pitched Down. It is at this exact moment that the a/p quit.

The Nose is "trending" UP, the VS is trending DOWN, and that signifies...

"MANEUVERING"

For five seconds, the airframe is not stable v/v level, speed, or acceleration. Well and good to pontificate re: what PF must do, but this massive airliner is spunky as a little Robin, and that ain't right.

"I have the controls...." Whether OR NOT the PF knows the genesis of disconnect, he is well within his rights to assume the a/p has left because of control demands, not ICE. At this point, he does not know, and after two plus years on, neither do we.

I sense HN has some misgivings about a/p in the stink. So did Smilin Ed, and so do I. The A/P has quit, the a/c is wanting management, and the Bus has reverted to "Other than NORMAL LAW". Free to STALL, you're on your own. The BUS has left the building.

Make NO mistake, what the pilot does now is the precursor, and (I think) the procuring cause of manually induced UPSET and LOC. Is there a 'sit on your hands' command for the AUTOPILOT? Thought not.

Wanting to establish a 'normal' flight path is a natural for any pilot flying an airliner. Perhaps there is a flaw in the training concerning this situation.
NOT "I have the Controls", but instead "monitor the Stick, No Touchee".

Wait, what good is there in 'monitoring' something that does not move?

I do the same, but then lets see what Clandestino wrote:

There is one word missing at the end, it is 0 g change. As the basic and overall topic of Nz-Law is to maintain a flightpath ( = flightpath stable = 1 g flight ) the sidestick in neutral commands no deviation from this 1 g flight, which means no loadfactor change at high speed or no pitch rate change in low speed.

He is wrong there. What he might be is referring to s the fact, that a dampening input is added to the normal LAw command.

To sum it up, if the SS is in neutral (= 1 g flight) and turbulence is disturbing the flightpath away from this one g flight, an opposing command is added to the existing 1 g flight command or like Clandestino likes to put it to the 0 g- change command).

If not hands off, but with a SS command of 1.2 g an updraft and thus increase of commanded g change by .4 g (which would give us 1.6 g) would stimulate the system to counter this increase by adding up to the commanded 1.2 g, thus maintaining the ordered 1.2 g of the SS as close as possible.

Any deviation from the flightpath (1 g with hands off, ordered g change with SS or AP inputs) will be opposed by the dampers by an respective counter action command within the capabilities of the system.

Let´s face the fact, anything else would be not helpful at all.

That the SS command adds to the already existing g, thus the SS command not giving a specific amount of flightpath change (load factor change at high speed, pitch rate change at low speed) is against any basic principle the C* law is funded on. But anybody feel free to point me to some reference here. Imho such an aircraft would not be controllable.

In the discussion there were also comments concerning load factor demand in high speed and pitch rate in low speed. Some here seem to believe, that in low speed the aircraft keeps the last pitch, if the stick is in neutral.
The aircraft will maintain a given flightpath, if we intend to change that flightpath with a SS input the gain will be load factor change at high speed and pitch rate change in low speed, but it still will maintain flightpath and not pitch in low speed with SS neutral (assumed that no protections are involved).

Again feel free to correct my research here.

When discussing the crew actions, their motivation of those actions and the resulting aircraft behaviour, which again constitutes an feedback channel to the crew, it is of immense importance to know how the systems work.

franzl

Some basics
 

A319/320 description Nz Law

Manoeuvre demand law as basic flight mode
- neutral speed stability with full flight envelope protection
-Vertical load factor control proportional to stick deflection : C* law
- independent of speed, weight, center of gravity ;

stick displacement : Nz = n Nz = n + 1g
stick neutral : Nz = O Nz = 1g


C* = q + K_C* x delta_Nz
q = pitch rate
delta_Nz = incremental load factor
K_C* = C* crossover gain - chosen to balance q and delta_Nz contributions

Flight path stability instead of speed stability
- control inputs are made to alter the flight path, not to hold it.

Medium-term flight path stability :
- maintenance of parallel trajectory 1g in pitch even after atmosphere disturbance.


C* (pronounced "C Star") is the popular name for a control law in which Nz (g) and pitch-rate feedback are blended. (In the late 60s and early 70s, Nz feedback was called the C law. NASA space shuttle approach studies added pitch-rate feedback, which was called C*.) At low speed in a C* airplane, pitch rate is primary; at higher speeds, g is primary. The changeover is transparent and occurs at about 210 knots in the A320 ("Fly-By-Wire for Commercial Aircraft: The Airbus Experience," C. Favre, 1991).


4.3.4. Longitudinal static stability. The A320's C* pitch control law is a manoeuvre demand law: the pilot's control inputs are interpreted as a demand for a given level of manoeuvre rate and the control system provides the surface deflection needed to generate this rate. Releasing the side-stick commands flight path stability.

. C* (pronounced ‘‘C-Star”) is a term that is used to describe the blending of the airplane pitch rate and the load factor (the amount of acceleration felt by an occupant of the airplane during a maneuver). At low airspeeds, the pitch rate is the controlling factor. That is, a specific push or pull of the column by the pilot will result in some given pitch rate of the airplane. The harder the pilot pushes or pulls on the column, the faster the airplane will pitch nose up or nose down. At high airspeeds, the load factor dominates. This means that, at high airspeeds, a specific push or pull of the column by the pilot will result in some given load factor.


And who likes to dig deeper and compare it to the 777, can read here.

Going back to my armchair again:O

franzl

Thanks for pointing that out. The graphic label now reads 1.0 (1,0)

Evidence. Three seconds prior to a/p disconnect, the airframe was subject to a robust Updraft (from DFDR). The FPC ordered Nose Down, to arrest the ensuing ascent (climb). This is of record.

After this event, the a/p quit, and PF took control.

Still looking for direct evidence of ICE, not unsubstantiated opinion.

SS neutral at low speed = zero pitch rate demand

What you see at lower speeds (ALT2) with speed bleeding off and with a 'hands off' neutral SS is essentially pitch attitude maintenance while the FPV (if selected) gradually drifts lower (not FP maintenance) as AOA increases.

(No turbulence damping in ALT2.)

:)