The following may help your search for evidence:-
http://oi39.tinypic.com/juxp8y.jpg
Then review the CAS trace in BEA's IR#3 - second from top on page 112. I'm sure you will note the increasing CAS during the climb

447 entered an updraft, it is on record. after exerting its influence on her, she started to climb, and reached a VS of 500fpm. What was the velocity of the updraft? 500fpm? Hardly. let's be conservative and call it 60 knots, vertical. Before the airframe rises, the pitots are sampling the airflow. If 60 knots on the nose, the IAS is 60 knots increased. If on the tail, 60 knots reduced. How about straight up? 30 knots less, and sampled simultaneously by all three probes. This is enough to cause the a/c to reject NORMAL LAW, and degrade to ALTERNATE. It also explains some Mach deviations reported in the DFDR. After sufficient time to outwait a possible anomalous loss of AS, the A/P is commanded to quit, after the RTLU locks.

Now this does not solve the problem of what happened, but it does suggest that perhaps an obsession with ICE may want a look. It may also call to question the reliance on this autoflight system in a cell.

Air France 447 Flight-Data Recorder Transcript - What Really Happened Aboard Air France 447 - Popular Mechanics

The best so far and probably a synthesis of what the final BEA report will say. So helplessly sad. :ugh:

The comments reading is interesting
I notice that many are pointing to the dual input system and again yoke versus side stick and also the lack of CRM

Updraft/downdraft velocities
 

The vertical wind speeds in this graph have been derived from the recorded AoA, pitch attitude and the aircraft horizontal and vertical speed components. The calculation is very sensitive to small errors in AoA and pitch attitude, and should therefore perhaps not be taken too seriously.
http://imgur.com/XvEwhhttp://i.imgur.com/XvEwh.jpg

HazelNuts39

I take your point re: sensitivity. Such vulnerability to errors of small magnitude is out of context (sic) for any graph of windspeed. One is tempted to say that the nomenclature is reversed, and that the airmass is the line "Aircraft". However, to claim the airmass line as the a/c would involve a belief that the v/s of this airframe is ridiculously responsive to airmass, let alone controls.

So the graph is essentially useless, imo. A response of airframe of even 500fpm to an updraft of 2500fpm is to stretch, especially when considering the a/c is in sensitive airspace, and in autoflight. Also consider the airframe, per graph, is responding far too quickly to light forces, and reversing out of all sense of Physics.

Do you take my point? The airframe is climbing at a rate completely inconsistent with the data. No updraft of 2500fpm will cause this aircraft to leave its cruise level with a functional (sic) autopilot latched. Nor will it cause said autopilot to command 4.5 degrees nose down (-1 degree PITCH) from cruise (+3.5 degrees) at Mach .80.

All this well before disconnect, no PF touching anything.

mm43

I am not saying there was no ICE. I am proposing that the loss of autopilot and the degrade to AL was caused by loss of AS, actual, as a result of Updraft. Once in the climb, in warm wet air, the airframe, including Pitot Probes and Statics, may well have ICED. I believe that it was likely, even. I also think that it was not consequential, since Airspeeds returned quickly. At that point, however, the Airspeed was remarkably lower, due Updraft, AoA, and climb gobbling up energy.

Rates versus gee versus "q" and Ps
 

Excellent post, Retired. And coming from an old stick-and-rudder-and-AoA dude, heh heh.

- As with the 'bus, our Viper control inputs resulted in close to "absolute" rates and gee, and were not "additive". The 'bus has some attitude and roll values added that we didn't employ. e.g. Pitch gee is biased for pitch attitude up to "x" degrees of pitch in Normal law. Gee is also adjusted for bank angles up to "x", so making a level turn and keeping the jet level is easier than what we had in our little jet. This aspect of the Nz laws helped us enter the dreaded deep stall at extreme pitch attitudes, as relaxing stick pressure resulted in a continuous one gee pitch command and the jet didn't go over the top in a ballistic trajectory at zero AoA or gee.

- The discussion about blending pitch rates with the gee command is a good one. Our laws reflected that, especially in landing configuration when rate dominated gee command. We also biased the AoA curves to provide us the "feel" of a conventional jet where you trimmed/commanded for an AoA. In other words, the neutral speed stability had to be overcome.

- Both of the above made the Viper feel like a much larger jet. Going thru turbulence and thermals at low altitude/high speed was very smooth. Was like an old Cadillac with the "soft" suspension versus the teeth-splintering up and down ride.

- The dynamic and static pressure inputs influence body rates due to the "gains" that are used for control surface deflections in magnitude and rate. As I have pointed out before, the Viper used fixed values for the gains when the air data was deemed unreliable. Seems the 'bus gives up and tells the pilot - "you have the controls". Heh heh.

MM43's input on handling the frozen sensors hits home for this old dinosaur. Had the static ports freeze up once in the SLUF and was descending for the approach in weather. Hmmmmm... I am not going down and speed is building up. Since we had an inertially-derived vertical velocity in the HUD, it was obvious I was going down and I checked groundspeed ( also inertially-derived), Clue light comes on and I realize I had frozen static pressure. BFD and waited for the radar altimiter to kick in at 5,000 feet and then the ice on the ports finally melted, so things were back to normal.

gums

And certainly without PAR, eh?

I'm really relieved to read that.

It is apparent that HazelNuts39's graph is based on sound method and best available data.

To dismiss it because it does not support a theory, seemingly based on numbers that have been pulled out of a hat at random, is foolish.

@ Lyman

No PAR, just ol' Gums and the ILS radio stuff and some skill and cunning. Oh yeah, no PNF to worry about.

OTOH, worst WX I ever landed in was a PAR, and a crusty Navy Chief talked me down in 100 and a quarter official, but more like 50 and an eighth. Call it WOXOF, if you will.

I might be making a totally stupid remark, but the "wind chart", even though it's not reliable, made my mind think "physics of flight".

Don't know the A330, neither how silent/loud it is and how loud changing airplane velocity is. These "speed of airflow" changes are not always created by changes in airplane speed, but also by violent up- or downdrafts. I wonder if, even when the airplane reduces its speed in reality, the sound of increasing violent updrafts could've created a false sense of speed.

On the other hand you have the physical characteristics of low speed stall (low speed because the "stall" protection is here primarily a low speed protection), which in my brain has always been a "silent" maneuver because of low speed. Yes buffet will be there, but it can also be a characteristic of higher speed flight.

If that is the case the "initial pull" seems logical because the physical environment gives you the impression of being in an airplane that increases speed. Add a "stall" warning on top, and confusion can be total. Sense and computers give totally opposite impressions to the pilot and it is only the "wind sound level". We all know the problem of visual illusions, however they can also be physical.

What is FCPC1 and which aeroplane has it installed?

Because you had G-trim, very useful feature for tactical fighter, absolutely unnecessary for aeroplane that spends 90% of its time in straight and level.

You ask for others or you are testing?


GENERAL DESCRIPTION A330/340 Flight Controls

The surfaces are controlled by three types of computers, depending on their
functions:
- the Flight Control Primary Computers ( FCPCs ) ( 3 per A/C ),
- the Flight Control Secondary Computers ( FCSCs ) ( 2 per A/C ),
- the Slat and Flap Control Computers ( SFCCs ) ( 2 per A/C ).
The FCPCs and FCSCs enable to control the aircraft in the roll, yaw and pitch
axes.

and so on....

As said before, it is useless to discuss matters without having basic knowledge of the systems.

@Lyman

I know you have "pulled" a post where you accepted that ICE was a factor, though stated again the inability of the A/P to maintain altitude in what appears to have been a fairly steady updraft (+/- turb) for about 25 seconds prior to A/P disconnect.

The A/P was pitching the aircraft ND in an attempt to maintain assigned ALT, and clearly from the graphics, the RVSM parameters were being exceeded. Perhaps the A/P disconnect was for that cause, and the UAS just happened along with the rising "warm/wet" air. The RTLU "latch" could only have been caused by the "speeds", and coincidentally the RVSM and UAS issues collided - time wise.

After reading the "talk in the cockpit" I thought why are the hand control sticks not linked together mechanicly:rolleyes:-So each pilot can feel & see the input of BOTH controls, Not, One pulling to climb-One pushing for Air speed. With HAL 5000 on the fritz with frozzen petots :=,Looks like a major design screw up in anyones book.:ugh:

That's our Lyman! (ba-dum-tish!)

In all seriousness though, what Lyman is not taking into account is the time factor of the calcuations for those readings and traces. One of the things I noticed on my A320 sim trip was that when an altitude capture is selected, the V/S display on the FCU module will initially display a surprisingly high value before it settles - I believe this is because the value is computed in real time and relates to what the V/S *will be* if the current pitch and thrust commands are maintained - of course, when the autopilot stabilises pitch after the initial pitch-up or -down, this value becomes something more realistic.

My theory is that the values on the trace are also calculated in real-time and those large bumps just before disconnect are a result of what the V/S *will be* if the current trajectory is continued, the pitch is retained and/or corrections are not made. The autopilot was engaged long enough to ride out the bump but did not have enough time to return the pitch attitude to 4 degrees nose-up, eventually coming to rest around 1.5 degrees nose-up. The aircraft would have regained cruise level on it's own, but slightly more slowly. Everything that happens after then (including the elevator movements that command the zoom climb) are a result of the PF's overcontrol of the sidestick.

I suspect that what may have spooked the PF was not something he saw that we don't know about, but something we do know about - namely this "bump" just before disconnect, which briefly brought the pitch attitude below 0 degrees. The residual autopilot command was enough to bring the nose back above zero, but I think he mentally programmed himself to correct it manually and unfortunately overcooked it. From then on they were in the zoom climb and things quickly stopped making sense for him.

@hillberg - read the thread from the beginning, it's always going to divide people but it has already been covered to death here and elsewhere.

TSB reports over threads ,A !!!! load of "Computers" & No "pilots" and control "Inputs" that have no feed back from PIC & SIC, A flawed approch to innovation in flying.

A330 AP/FD & A/THR Conditions
 

Assembled from separate logical snippets.......(could be not complete!)

http://i474.photobucket.com/albums/r...DATHR_COND.jpg

Based on the close correlation of normal acceleration, V/S and altitude (each being either the time-derivative or the integral function of one of the others), I don't think the traces support that theory.

According to FCOM 1.31.40 p.15 description of indications on PFD:

We'd need someone more qualified to confirm, but moderate turbulence by any definition should not be enough to take out autopilot/autothrust and I don't think it did in this case.

The climb was caused by the human and not the computer, depite Lyman's ever more bizarre protestations.

@hillberg - we're not going into the force-feedback loop again if we can help it - suffice to say there are as many valid reasons for dropping it as there are for keeping it, so it becomes a matter of personal preference.

DW
I do not think whether it's a matter of preference by anyone
This is something that can affect (in some cases) the (safe) conduct of flight
In fact .. there are currently two systems (the yoke and the sidestick) with or without feedback
There are questions about these two systems concerning the conduct of the flight and one of the two will prove better than the other
It's to weigh the advantages and disadvantages of each and make the best choice for flight safety

Hi HazelNuts39,

How come there is a trace about that in the FDR, the choice being:
V/S from AirData
or
V/S from Inertial

(i.e. no choice "V/S from both sources").

Thanks :)

Vertical speed is derived from the IRs with the barometric input providing a damping function. If the IR becomes invalid the V/S indication changes to pure barometric. This is indicated by an amber box around the digital V/S display.

I seem to remember the intertial V/S being considered invalid above a certain value. 10 000 ft/min is what springs to mind.

@AZR
 

Inertial V/S data actually seems to be a hybrid signal (IR & Barometric),
as posted earlier by Takata in august 2011:

http://www.pprune.org/tech-log/45687...ml#post6624383

Dozy Wannabe

1. Interesting concept: "Anticipatory Instrumentation". Extrapolation is of limited value, and not to be relied upon.

2. The Updraft could easily have been well in excess of 10k fpm. I used 5000 as a low, and conservative, value.

3. VS not reliable via IR. Airspeed not reliable via IR.

4. In airdata compromised situation, you believe the Baro VS? Let alone a smoothed combination of the two? Especially in the presence of rapidly moving temps and pressures?

5. There is a good deal more evidence to support UPSET caused loss of speeds, than ICE.

6. For that matter, there is plenty of evidence to suggest the upset was caused by this environment, as well as UAS.

7. At Loss of Auto, the airframe was losing altitude, and vertical acceleration. Combined with Nose Low on the screen, what would you do? Wait, right, ask your advisors/instructors?

Challenge. Describe the accelerations on the Pilots' bodies as a result of this maneuvering.

The V/S trace seems to confirm what you remember. Would you have a reference for it?

Hi DozyWannabe,
It's not so much the force-feedback - it's the control surface displacement feed back loop that is missing.

Please see page 7 of http://www.easa.eu.int/rulemaking/do...%202011-09.pdf

“c) The use of side stick controllers together with electronic flight control systems which provide control augmentation and control deflection limiting systems could affect piloting awareness that the aircraft is approaching a control limited flight condition. It may be that return to normal flight condition and/or continuing of safe flight needs a specific crew action. In these circumstances a suitable flight control position annunciation is required to be provided to the crew, unless other existing indications are found adequate or sufficient to prompt that action.
(2) CS 25.777 Cockpit controls
Side stick controller force-deflection characteristics in pitch and roll together with displacement sensitivity and gains need to be evaluated. The intention is to show that normal inputs on one control axis will not cause significant unintentional inputs on the other. Consequently a new paragraph CS 25.777(i) is proposed requiring a suitable assessment.”

Dunno if it was posted before ...

Crude Google translation ......
http://www.bea.aero/docspa/2010/f-zb...74fa196764388a

Hazelnuts39

The topic came up in a discussion with a tech rep some time ago. The discussion had veered off the main topic at that point so I didn't pursue it any further.

Thinking about it a bit more, 10 000 ft/min seems a bit low as a cut-off value. I would expect the IRs to be able to provide valid data at higher vertical speeds.

With regards to the PF and the V/S display, there are two slighly different layouts of the display. One where the only visible part of the pointer is the one covering the V/S scale, and one where the full lenght of the pointer is visible.

I do not know which version AF has, but if it is the first one the position of the pointer is less obvious when pegged at one end of the scale.

http://i42.tinypic.com/27wymtx.jpg

how sensible is your calculation for smal changes in AoA? the elevator flys about 0.1 sec behind the wing, if you fly with CAS 250 into this sort of fast increased updraft the elevator had ca.160 feet/min lower updraft than the wing-----> the result in AoA is about 0.2 deg aditional-diference between elevator and wing

the bird pitch up, as if you pull 0.2 deg, if he fly into this strong updraft

(and visavers pitch down 0.2 deg if you fly outta)

is your calculation so sensible that this effect plays a role???

grity

along those lines, which surface has the greater moment arm to affect Attitude? And how is this accomodated?

Up Elevator. Per PF

Grity,

The sensitivity of the calculated updraft velocity to errors in AoA or pitch is evident from the formula given in the header of the graph. For example, for GS + HW of approx. 500 kt, the updraft velocity changes by 88 fpm for 0.1 deg error in the recorded AoA.

Quite another matter is the pitch response of the aircraft when traversing an updraft zone of increasing or decreasing vertical velocity. That effect exists and has been discussed on an earlier AF447 thread, but is smaller than you suggest: The updraft increases about 2000 fpm in 5 seconds, and 40 fpm / 500 kt is about 0.05 degrees. The actual effect is even smaller due to the downwash behind the wing and other damping effects.

@RRR
 

The F/CTL page - showing the control surfaces position - was automatically callled and in view on the System Display @ triggering of the F/CTL RUD TRV LIM FAULT (02:10:19).

Which could be in compliance with:

Next,

Not very demanding IMO, SS is equipped with dampers, springs (Roll) & spring rods (Pitch) in all axis.

For me, the supposition that aft stick at a/p release was inadvertent was put to bed long ago. I have to give a minimum amount of credit to PF, and I do believe still the a/c required aft stick as well as Roll left at drop.

An aircraft in an updraft will drop her nose, and in a downdraft, the reverse. By herself, it is basic to the geometry of the architecture. I could not tell you whether or how much attention was paid that fact by the software, (DOZY?). But in this a/c, if in an UP with the kind of energy I believe was in play, her Nose wanted down on the natch. The a/p may have "known" this, but it would not have abated at a/p quit. This may have been a major player in the PF's insistence on Nose UP, when in fact, Nose Down would be needed. Nose down is also contrary to g demand in PITCH, and the PF may have had help in Pitching this mama UP. NO?

Lyman, your constant restating of your theory is somewhat annoying to me. Saying the same thing many times does not make it correct. Flying at night, in turbulence, should not be overly challenging. If the aircraft needed a bit of nose up at AP disconnect, then 1/2 to 3/4 seconds of aft stick would have been sufficient. What PF was doing must have been quite different.

From the control position graphs of the first 10 seconds, PF was badly over-controlling the aircraft in roll. The energy and force required to make these lateral control inputs were likely substantial.

The Airbus stick is angled forward in its neutral position (to discourage cross channel inputs I understand). I suspect that with the drag from the roll channel viscous damper, PF could not get adequate leverage on the stick to keep up with his perceived need for fast lateral input. Once the stick is canted back towards vertical, there is higher mechanical advantage available. Please note, I am only discussing the control inputs up to the stall. Afterwards, there were different reasons for his aft stick input.

If someone (perhaps A33Zab) could come up with the force/velocity characteristic for the lateral channel viscous damper, then we could look at the energy and force required to make those lateral inputs. It might be very educational when looked at in terms of human energy expenditure.

I must confess a certain scepticism about such a suggestion. It suggests, to my mind, an unfamiliarity with the controls worthy of a tyro which whatever his reactions I don't think we can claim the PF to be. One shouldn't get too wrapped up in trying to find a mechanical explanation (red herring fillets). Rather there should be more examination of the psychological elements of this accident.

I mesured in your calculation at 2:10:13 increases of updraft 1000 fpm in 0.7 seconds, that was my base for delta AoA ca. 0,2 deg

but you are right this is not so much......

AF447's initial altitude drift went virtually unchallenged

:(

AF447 Roll Oscillation & PF's control inputs.
 

OC, So what are your thoughts about the following definition?
Do only tyros experience PIO? Was the aircraft's sustained roll oscillation before the stall a PIO? Had PF ever flown in Alt 2 Law at altitude?
Why would a somewhat experienced pilot be unable to fly smoothly?