@BOAC:
A nice variation on the old gag about Unusual Attitudes recovery.
Something about Shave the Bat rings a bell. This must be included in aviation vernacular, but I'm not sure what practice it ought to be applied to.

"Captain Pteropine sure shaved the bat on that one!"
(Made in reference to a hand flown landing very near stall brought on by a sudden windshear attack in the flare ... )

more automation?
 

Some nice points, Garage.

Some do not realize that one of my original jets was the VooDoo, and it had the dreaded "pitch-up". Couldn't deep stall it due to the aero and c.g., but it had two limiters when not in autopilot control stick steering - a basic AoA limiter that we could overcome with 65 pounds of back stick, and a 'pusher" that snatched the stick outta your hands. This was back in early 60's, folks. So it has been tried and flown a long time ago. FBW systems don't act like that! You can pull back at 100 pounds and the command will be "x" gees, or "x" AoA, and if more than the design then that's all you'll get until airspeed is too slow. In other words, the jet will give you all that the designers figure you'll need.

One of my problems with the 'bus is that it seems to ignore the AoA when other air data is outta the loop. I have no problem with the basic "limits" implemented for the 'bus. They seem very reasonable for the mission and the aircraft's aero capabilities. Although I have my personal ideas of implentation, I'll stick with what we have to work with today except ignoring AoA.

So we don't need bitchin' betty telling you to lower the nose! And how is the 'bus gonna do this if it has ignored the AoA? Huh? just use the AoA limiter function and the sucker will gradually lower the nose to reach the 1 gee baseline ( corrected for pitch, so it will be lower than 1 gee in a climb or dive). The problem is if you are climbing at an extreme attitude, then you run outta air molecules over the THS too quickly to get the nose down!!

I also caution folks about the "direct law". Sounds like the jet would fly like a "normal" jet, but it ain't so. There are still control surface deflection rates plus pitch rates computed independently of air data and such in the way. Further, the 'bus ain't no F-22, and getting 3 or 4 gees within a half a second might be a bit disconcerting, ya think? Only way around this, way I see it, is to employ control stick feedback as we now see in several PC flight sims. In other words, you get a "feel" of how much you are asking the beast to do. Only problem is that kinda system need some gee and "q" inputs. Gee ain't a problem, but when the design ignores speed under certain circumstances, then all bets are off. And such is the 'bus control logic at present.

respectfully,

thank you for a very imformitive post, i cannot thank all of you enough for your posts and links again i take my hat off to every professional person to post on this forum. thanks again from a slf:ok:

Machinbird;
The text I have is not the easiest to read/interpret. One could gain the impression by reading some of the statements, that once the THS is moved manually through the trim wheels, that it is in "manual, mechanical mode", period. However, the relevant statements which make me believe otherwise, are:




...........5 The mechanism is now set in the electrical control position.

I think that makes it pretty clear, that once manual input stops, autotrim is reinstated. It also makes sense from the POV of the safety of flight - you cannot have a 1g AFS function safely with manual trim requirement.

BOAC;
The THS would move, doing just as we would do in a non-AFS airplane...trim control column forces out to keep "elevator forces neutral", (bearing in mind that the forces aren't real, being masked by hydraulic servos and are artificially introduced by springs, ;).

The question that the engineers had to come to terms with however is, how do you do that in a FBW system?

For the above reasons suggested by Machinbird, you can't just rely on human input to do the trimming in a FBW automated system.

Maintaining "1g" flight is one solution. Using the elevator, (short term pitch control) as speed changes means that according to Nz Law the THS, (long term pitch control) would change position to maintain full elevator authority.

Maintaining 1g is an indirect response to increases and decreases in airspeed. A decrease in speed causes slightly < 1g as the aircraft pitch reduces, an increase, the opposite. Even in manual flight, the THS responds, both to stick and Nz laws.

Clearly, there are circumstances which you would either not want autotrim or autotrim would not be available in system or sensor failures.

Our human feedback loops are completely transparent to us but are extremely sophisticated and are the "loops" in a non-fbw airplane that we "tune" when we learn how to fly.

But those "loops" must be made visible (from their "automatic transparency!), then designed and built into a FBW system independent of humans for such a system to even function in a rudimentary way, and decisions must be made about such loops about the information signalled back to the servocontrols, the AFS and the crew.

Otherwise, any autoflight system would quickly become unmanageable, highly fatiguing and fundamentally unsafe.

PJ2
Thank you very much for the analysis with references. That is crucial information in understanding a large range of issues regarding the Manual Trim and Automatic Trim systems.:ok: The FCOM references are indeed skimpy and confusing.

This is to express my gratitude to A33Zab (AF447 Part 3 #1818 11th june) for digging up and posting the complete Stall Warning schedule of the A330. That is information I've been searching for in vain since s/w was first discussed early in the AF447 threads. So I've promptly incorporated that schedule in a graph that I posted earlier: StallWng2.

THS Answers.
 

Hmmm, doesn't the A/C makes inputs of its own in ALT law, trimming the THS ? Sure, this was due to strange PF inputs but was he really aware of the move ?
[/quote]

The actual logic inside the FCPC is not open to the public, so we have to rely on what is stated in the released documents.

"Pitch:

The side sticks and in some cases the pitch trim control wheels control the aircraft in pitch.
They act on the elevators and on the THS depending on the different laws.

1. Nz law.
This FCPC law is the normal pitch law engaged during the flight phase.
The pilot commands a load factor via a pitch action on the side stick.
The Nz law executes this command, depending on the aircraft feedbacks, so that:
- the short-term orders are executed by the elevator servo controls.
- the long-term orders are executed by the THS actuator (autotrim function).
The gains depend on Vc, on the flap and slat position and on the CG position."

From study material and answer to Machinebird's question:

"An override mechanism, which is installed in the PTA (Pitch Trim Actuator),
makes sure that the mechanical control through the trim wheels cancels the electrical control.
When a manual command is made with the trim wheels, the override
mechanism gives priority over the electrical command from the FCPCs.
It mechanically disconnects the PTAoutput from the mechanical input(via
electro-magnetic clutch) and also operates the overriding detection
switches which in turn signal the FCPC's to stop any electrical command
from the FCPC's."

It's not stated but it seems logical that when a manual pitch trim wheel
command is made or an autotrim command is cancelled by holding the
manual trim wheels (THS runaway) the side stick should be released to neutral.
In that case there is no SS input anymore and when manual input on Trim wheel stops also,
the electro-magnetic clutch is released again and THS system is ready to
take new SS orders.


Cryptic schematic but could not find any better:

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



In ALT 2: The HI AOA Protection is lost, in case of a dual ADR failure (or ADR DISAGREE).

Agreed, except that when IAS dropped below 60, then 30 kts, I think that was due to changes in the fuselage pressure distribution near the static ports (over-reading static pressure) due to high AoA, as much as or even more than due to the pitots under-reading total pressure. For what it's worth, CAS never drops below 140 kt in my simulation. That 140 kt is quite capable of pointing the AoA vanes in the right direction.

And don't forget Otto Lilienthal (recently celebrated on another PPRuNe thread), and who still knows the Minimoa?

Questions & Comments
 

PJ2, in answer to your post #258 :

More accurately, it did not remain in Normal, it would have returned to it at the end of the 10 seconds monitoring process. This ADRs monitoring process by PRIMs specifically allow return to Normal law in some cases. During the process itself, it is in Alternate 2.

No I do not. I expect it would have acted upon its sole authority following Normal law and quite possibly its protections, interfering badly with what PRIM1 was doing according to PF orders. Being in Normal while Master was in Alternate 2, it would have deemed its Normal law better than what was asked by Master PRIM (PRIM1), thus resisting it.

Not quite. Especially on elevator control, the need to activate all servos simultaneously under certain conditions make it necessary to cater for dual PRIM outputs onto parallel servos. Such thing is therefore not positively excluded from the design.

Alternate 2 law was latched by at least Master PRIM1, and likely PRIM3, thus triggering the Alternate law ECAM and PFD effects.

Not quite. Their wording was extremely careful (BEA text in italics):

From 2h10min05, the A/P then A/THR disengaged and the PF said "I have the controls". The airplane began to roll to the right and the PF made a left nose-up input.

The monitoring process takes place in Alternate 2. 10 seconds elapsed, then :

At 2h10min16, the PNF said "so, we've lost the speeds" then "alternate law [...]".
The airplane pitch attitude increased progressively beyond 10 degrees and the plane started to climb. The PF made nose-down control inputs and alternately left and right roll inputs.

The climb is not correlated with the initial left nose-up input. 10 seconds between the two. Time for PRIM2 to return to Normal. I posit these nose-down inputs were an unsuccessful reaction to the zoom-climb/pitch-up (manual THS should have been used then, easy to say now of course), and not the initiators of the following :

The vertical speed, which had reached 7000 ft/min, dropped to 700 ft/min and the roll varied between 12 degrees right and 10 degrees left.

I posit these are only the result of depleting kinetic energy. No updraft here, speed lost for height gained, no added energy. Roll variations might be the result of conflicting actions of PRIM1 and PRIM2 in different laws over ailerons and roll spoilers which the PF would therefore have had enormous difficulty controlling.
I agree that we should be questioning this action. After flight controls have cleared the investigation filter ;)

It was actually "to the nose-up and left stops".

If flight controls were compromised, as the theory I offer seriously suggests (this is no fancy), then PF actions cannot be understood without the full traces of computers states, surfaces actuation/position, and so on, simply because his actions and the aircraft reactions would make no sense to him. This could lead him to trying things which would make no sense to us.

What would cause PRIM2 in Normal to command the zoom-climb pitch-up?

same thing as that 2000 A340 AIRPROX?

PJ, thanks for that graphic, A33Zab, much thanks for your diagram.

In walking through PJ's proposed event sequence, it's the same place I get stuck as the first read through I did of the second interim report.

I still don't understand ...
Was the pilot flying (or think he was flying)
a g command,?
a nose attitude?
an AoA (with no AoA reference in the cockpit)
another profile and expecting something else to happen next?

For that matter, what response was he expecting from the aircraft?

A guess here: he expected to escape the stall alert. Apparently, we go down the timeline and the stall alert went away. Did the crew deem the maneuver successful? :confused: Based on pre-event conversation, there was still probably a concern with altitude, since they were worried about their scheduled climb and the temps at altitude not developing as forecast.

This leads to a question I still have: why no deliberate act to descend with stall warning no longer active, as they'd been concerned with altitude just a few minutes previous to all this going down. (Guess: deeply concerned with airspeed, or its lack of indication)
==
If PJ's sequence is about right, the Captain re-enters and has to figure out what's going on while they are still in a diagnostic mode. He has something between two and three minutes to assess and call for the correct remedial actions( X, Y, Z and beyond) until recovery.

It appears that he may have suggested, or ordered, engines to idle, nose down commands ... and the stall alert goes off again.

With no disrespect to the dead intended, did the next CVR recorded voice transmission come out as "what's it doing now" in three part harmony? :confused:

Whilst there's a slight break in the Tech action...

LW 50:

I think I flew with Harry Pteropine when he was an FO.

Last I heard, he retired from a fly-by-night cargo operation and became a safety consultant.

He was noted for being able to hang upside down and view all the facts from a different angle.

:ok:

"Managing and handling" advanced Systems (Human machine interfacing issue)
 

The System "output": Data (displays, aural info, etc.) and inputs to a/c flying surfaces (e.g. to THS, etc.) must be easily (and very fast) understood and understandable by the crew. It´s absolutely necessary an interface designed to present in real time a clear and reliable "picture" of "what´s going on" (and changing) in the plane.

And also even potentially creating "extra CRM issues".

My comments are not just on AF447 case. Other incidents/accidents can be related to these issues. And are not for just A (or B) but for "advanced planes", an important, "powerful" (and evolving :})trend.

Smilin Ed

I think you misunderstand - you seem to be describing the machine/human interface. Human factors does not refer to aircraft and how they operate but to people and how they respond to the situations they find themselves in. An awful lot of the discussion in these threads has focused on the possible trouble with the systems and how this has impacted on the pilots involved (a lot of this clearly ill-informed) but very little has focused the pilots themselves. Some have certainly asked questions but answers there have been few. The key to this disaster is the flight deck crew not the systems - which no doubt can be improved, systems always can. Information so far released contains hints of complacency and casualness (for example there does not seem to have been any sense of urgency in entering an area of turbulent weather). More examination of possible mental states would I believe be more fruitful. I realise that some of us don't really want to go in this direction (I am a pilot too but not on big jets) but to understand this accident I would suggest it is necessary to leave no stone unturned no matter how painful it is. Some have already shown this aversion to commenting on the pilots and I have to regretfully submit that this is not the right approach.

- I'm sure the majority here see the 'pilots' actions as a major factor in this accident and thus a subject to be examined thoroughly, but the question is 'why did they do xxx' - if indeed they did.

Could you answer #302 please?

In normal law (AoA law) no stall warning, no stall: AoA remains at alphaprot.

From the BEA Note:

So, what does it tell us ?

So we know pretty safely AoA was 4°, speed was >215 kts.
If we now assume a pitch of >10° (see first quote),
this gives a FPA of >6° at ~400kts TAS.
This equals a V/S of >4000 fpm.

Taking this together it does not seem that the depleted energy was the reason for the reduction of the climb rate.
There is only one way how with the data points we have the V/S can have gone down to 700 fpm if speed was >215 kts an AoA was 4°. And that is that the Pitch must have reduced after the initial Pitch Up.

Therefore I conclude that the Nose Down input of the pilots have not been unsuccessful. They might have yielded slower response than expected or resulted in confusing somatogravic results but going purely by the facts I tend to disagree with your scenario that the control system did not follow the Nose down commands.


Edit:
The somatogravic point is the one which is the most intreaguing for me.
If they accidentally overcontrolled the aircraft leading to the initial climb, (e.g. due to mxing up with Roll correction input) then started to deplete energy due to the steep climb, a sudden and significant lowering of the nose (reducng V/S from 7000 to 700 fpm would fall into this category) would give a 'falling sensation'. Being relatively close to the stall speed, a subsequent pull up reaction to any pitch would lead to a reasonable 1g result, because there is no more lift available.
This 1g fits well into the normal range of things and would feel like normal attitude again. in complete darkness with no external references I'm afraid these sensations can play nasty tricks on you. Once your personal G force reference system is shifted (Falling vs. stable 1g), its difficult to go back to the original one. Re- aligning this personal reference system with the instruments in a situation where you are not sure which instruments are working properly might not be easy.
During and post stall any NU input will yield </=1g and thus feel benign.

.

V/S can have been reducedby other factors as well like bank angle / roll to left and right, decaying airspeed, drag by deflected ailerons / rudder.

Concerning AOA: The relationship between AOA and pitch is not factual, theoretically a pitch of 90° with an AOA of 0° is achievable (not in an transport ac, but in a fighter ac).

I did observe lots of misunderstandings concerning unloading (stick forward) in nose high situations after more than 1 g was applied. 0,5 g already feels like a lot of unloading, stops the increase in pitch rate and reduces the AOA somewhat, but doesn´t change the (upward) flightpath immidiately and drastically. To really get the nose travel downward fast (and that is necessary in a nose high low energy state) unloading to at least zero g and below zero AOA is necessary, a vey uncomfortable thing to do if you are not used to it. And it takes some patience to wait till the nose is definitively below the horizon until further actions (like rolling wings level and applying power) have enough positive influence.

So i agree, nose down input had some effect, but not enough (Unload not high enough and not long enough).

Applying TOGA at that point was IMHO the worst thing which could be done by the crew. As long as the nose was above the horizon this TOGA produced a pitch up from the underslung engines and some noise, but no positve effect for recovery. That coupled with nose up stick input kicked the ship straight in to the fully developped stall.

That is no direct blame to the crew at the moment, as it affords training and expierience to act accordingly in a situation like this. A fighter pilot gets lots of training and application of those procedures in daily flying, the stall aproches in landing configuration or in medium altitude for air transport pilots in the simulator falls way short to that demand.

Concerning the somatogravic illiusions, spot on!

retiredf4,
No disagreement here. The only thing which I did was to derive the Flight Path Angle:
Flight Path Angle = Pitch - AoA
And consequently VS = Sin(Flight Patch Angle) * TAS
All I wanted to show is that 700 fpm at 400kts TAS means a Pitch attitude of significantly less than 10°..
In fact it would mean a Flight Path Angle of ~1° and thus a pitch of ~5°.

Regarding Roll it was stated that it varied between 10 - 12° left/right, so you would have to multiply the V/S with the Cosine of 12°, which would drecrease the Vs by less than 5%. Doesn't change the whole picture though.
Any speed reduction is already included as i assumed the 215kts IAS to be correct at that moment.