Hi Gums,

Here is an article by Boeing where they discuss high altitude handling and stability in general terms.

High-Altitude Handling

It does describe the differences between the B-777 and the MD-11 in terms of longitudinal static stability and how it is achieved in both aircraft.

Given that Airbus is using C* based laws it would have tried to keep attitude. If the last commanded attitude was 10° NU, it would have tried to keep the nose there, if not corrected even into the stall.

This is confirmed in the report on page 187 (chapter 2.2.5.)

and
Simply levelling off and then letting the stick go should have helped until a speed of ~Mach 0,65 (Figure 65 in the report). After that even at level attitude the FCS would have started to trim into the stall once the propulsive limit had been exceeded. At that point sustained ND input would have been required.

Put still another way: If one removes the 'grass' or 'mayonnaise stirring' from the trace, the sidestick was at an almost constant 21% nose-up position between 02:10:50 and 02:11:30, i.e. the time the THS was moving.

Two stupidities were combined to put in water AF447
That of the pilot ( pull the stick as a mad with a continued stall warning )
That of the Airbus system (hold the trim in up position when knowing this was a continuous stall)
If Airbus was in full manual both stupidities to put the plane into the water should have been made ​​by the pilot
Pull the stick and put together manually the trim full up position
So I conclude that the auto trim really helped to put AF447 in the water because the system worked as per design

I'd like to see an experiment proving that before we make a definitive call - at this juncture we're into systems behaviour as yet untested, and unless we have someone from Airbus to confirm or refute that theory, your guess is as good as mine.

Which is, after all, the correct response to recover from the onset of stall and has been since aircraft were made out of wood, doped fabric and twine.

Ah, but the flight control computers are not programmed with the concept of stall and to the best of my knowledge never have been.

The stall warning parameters are encoded into the annunciator system, but the flight control computers are oblivious to stall in Alternate Law - and most would say rightly so, because for the computers to be able to override the human when in a degraded state would open the aircraft's safety to significant risk if the computers get it wrong (not to mention the fact that the backlash from the Airbus-sceptic brigade would be deafening if this were the case!).

The point is that outside of Normal Law, the systems are designed to defer to the pilot's inputs - no matter what those inputs are - for better or worse. This is based on the reasonable assumption that the human pilot will have a much better ability to adapt to circumstance than the computer ever could. In this case the crew were clearly overwhelmed and not only made mistakes, but repeated the same mistakes over and over again with tragic consequences.

You still have not fully understood the autotrim, THS, NzLaw without protections. Unfortunately BEA has not added normal acceleration in the graphs of the final report, therefore we have to deal with the bad graphs from Interim report 3.

From 02:10:25 until 02:10:50 the normal acceleration was below 1g. When PF after the stall warning 2 selected TOGA, the normal acceleration increased for a short time of about 10 seconds above 1 g. From that point on the normal acceleration was uninterrupted below 1 g until 02:12:00. When PF applied SS full NU at 02:11:40, the THS had already reached near full NU limit and the elevators moved now full NU. Despite those NU commands and despite the position of the THS the normal acceleration stayed below 1 g for another 20 seconds, until 02:12:00.

During all those times, where the aircraft was not able to maintain even 1 g, (and that started already before the apogee of the climb) the trim and the elevators would have moved on their own to achieve this 1 g with hands off SS neutral= 1g.

To stop the trim from further traveling, the SS would have to be moved and held at the present normal aceleration (which was below 1 g). To get the trim moving ND, the SS would have to be held to demand a g value well below the present normal aceleration.

Interesting, after PF started putting the SS to the NU stops and reduced the power at the same time to idle, the pitch of the aircraft decreased from +15° to -10° within 10 seconds. Power change and maybe the elevator deflection to full NU changed the forces of the airframe. What that did to the mental picture of the PF would be interesting to know. If PF would have gone to full ND at that point, the he might have had a chance to initiate recovery.

After the aircraft dropped outside the flightenvelope, which happened after the stall warning 2, any command except a definite ND command was bad, but the amount and the duration of the NU input or even a SS neutral input would not have saved the day and did not agrevate the NU travel of the trim. and the elevators. Because those were already maxed out by trying to maintain 1g.

NO, you are wrong there. It does maintain attitude, when autothrust is available. Without autothrust, when speed is decreasing, the aircraft trims to maintain the flightpath which is roughly 1 g. To maintan level flight when speed decreases, the attitude has to change, pitch has to increase, leading to a further decrease in speed and finally to stall. In your example the 10° pitch would have to increase.

You would be correct however, when low speed stability protection is available, which was not in AF447 case in Alt2b law.

Read it again please. BEA speaks from maintaining altitude. They talk about a level flight there. To maintain the altitude, 1 g flight is mandatory, the attitude has to change to maintain this 1 g for level flight.

Come on guys, we have been through this for years.

@franzl

I'm not disagreeing with your hypothesis in the slightest, I'm just saying we're so far outside of normal operational parameters that I'd like to see supporting experiments to attempt to prove it.

I suspect that the PF could have rescued the situation with full nose-down held as late as 02:12:20, but I can't prove it.

What you describe would be a pure C law. My understanding is that Airbus uses a C* law. A C* law will progressively switch from g command to pitch rate command when the speed decreases, i.e. at the stall it will be mainly pitch rate. That means if no SS inputs are made it will keep pitch rate 0 which means constant attitude.
If SS is left neutral it should keep constant attitude throughout. That is also what BEA considers neutrally speed stable.

Could you point me to any source that states that Airbus instead is using a pure C law?

Edit: Altitude is not a parameter in any FCS law. So FCS alone will probably not chase altitude. It may chase g or pitch or pitch rate or speed. That is what distinguishes C from C* and C*U (B777).
Altitude is only an A/P parameter.

You might be the best engineer or pilot, you will never be able to prove it. When BEA would have been able to prove it they would have done it. And they had more tools available than you or we will ever get our hands on.

We had this before.

As i understand it it only references the value to change, in low speed the reference will be pitch change, in other cases g command. But stick neutral it will be 1 g.

During all those threads i copied following sentence, source now unknown:

I thought of everything and so I had to raise troops for mount a counter attack :)
In normal law (autopilot engaged or not) the automatic trim is a must and a great help in flying the aircraft (thank's also to the protections)
In alternate law .. things are not all the same
With the autopilot off (the pilot fly manually) .. alternate law says there is no more pitch protection ..
But still .. auto trim remains active
piloting in manual (pitch matter) is using the stick and the trim (the pilot is using both hands)
Not with Airbus ...
The pilot does not control the trim .. and this one continues to act as in normal law .. but then it is no longer active protection ....
To me this is a bad design .. the pilot don't control (full manual control) entirely the pitch ...
In alternate law the auto trim is not a help ... instead it's like throw a lead buoy at somebody drowning

@franzl:

I know for a fact I'm neither, nor am I likely to be!

Incidentally, the "Normal Acceleration" parameter is very much present in Annex 3 of the BEA's final report - on pages 1-6.

I beg your pardon, Doze, but this is what my copy of the FCOM states:

http://www.sluf.org/misc_pages/a330_flight_mode.jpg

The attitude correction for the gee command ( load factor) provides an apparent "attitude" command, but it just ain't exactly the way HAL is working to "help" you. Additionally, this implementation also provides apparent "neutral speed stability". The basic aero of the jet still has positive speed stability, and positive longitudinal static stability, but HAL is trying to "help" you.

Only place I can find anything other than a gee command is in Direct Law.

show me how I am not getting my view of the control law from the text above from the FCOM.

As you, Doze, have pointed out many times, the primitive FBW system I flew back in the stone age would not be recommended for a commercial airliner EXCEPT FOR THE AoA LIMIT/PROTECTION!!! We also had a heavy AoA bias with the gear down to make the thing seem more like a "normal" jet.

Our gee command/AoA curve was designed for sustained turn rate at moderate gees - figure 5 or 6 gees. Think 25 degrees per second sustained rate with no loss of energy. And at 9 gees we got more than that for about two seconds until the AoA function kicked in - eat your heart out, heh heh.

For one more time, I iterate that I flew a no-kidding electric jet that did not have inherent static stability below 0.95 mach, yet had the apparent neutral speed stability that the 'bus has ( except when gear was down). We did not have the attitude gee correction for obvious reasons. We had no "direct law" except the manual pitch override deal when AoA was above 30 degrees and we were in a no-kidding deep stall, and that only gave us direct control of the horizontal tail. The 'bus has a lot more going for it and its pilots than the Viper, but ya gotta understand how the thing works and have some basic piloting skills.

Its the exchange with DW and his continued saying, that the THS was driven by the NU SS input of the crew. And that is not correct, imho it had no effect at all, because only a continued SS ND input would have prevented the THS from moving NU.

The exit of the flightpath was the failure of the crew, a following prolonged and prominent ND input could have corrected this failure in the early flight envelope extension.



It´s not in the section of the HF group, and i dont have the authority to change the wording of BEA. It would be nice toi know, what caused this marvalleous sentence.

In more traditional designs this is the de facto state of affairs, but this design is different.

The pilot controls the trim via the sidestick, just as in Normal Law - presence or lack of protections dont' come into it.

An opinion to which you are welcome, but looking at it from an engineering point of view, does it not make more sense to have the aircraft behave as close as it can to how it behaves when not in a degraded state?

Not at all - it behaves as it does in Normal Law, which makes sense - the loss of hard protections is a moot point. You can't blame the design for the fact that at least one member of the crew consistently applied inappropriate inputs.

That is the description of a G command law vs a pitchrate law.
It does not describe the switching from one to the other which is happening with a C*.
When switching to the other prevailing command law it will do so also for stick neutral, i.e. it will keep pitch rate constant if SS is neutral.
Otherwise it would still be a G command law. However I'm not aware that the laws switch depending on SS deflection.
My understanding is the following:
At higher speeds (lets say above 230kts) it will keep 1 g at stick neutral and when deflecting the g load will always be the same for a certain stick position, independent of the speeds (so at 280 kts the same deflection will yield the same g as at 230kts).
Below a certain threshold (which was still quite a bit above stall speed IIRC) the law blends over to a pitch rate law. I.e. at SS neutral pitch rate is 0, thus pitch constant.
A certain deflection will always yield the same rate of pitch change once this law is the dominant one.

Thanks, Retired, I think we are on the same page.

OTOH, I still believe the crew could have saved the day even after entering the stall. May have taken 15,000 feet, but a constant nose down command and manual trim should have done it.

The crew did not understand what was happening, and besides, "you can't stall the Airbus" mentality could have been at play.

I pray that many 'bus drivers are reading our excellent recap of the accident and all the ideas we have expressed to prevent a future repeat.

You are correct, but they are the same as in IR3. I should have said that i was looking for high resolution pictures.

The normal acceleration in the following graph would be helpful.

http://www.bea.aero/fr/enquetes/vol..../figure.28.jpg

I think it is both: The THS movement was driven by the continued NU SS input, and (after the stall) only a continued SS ND input would have stopped or reversed it.

Cheers HazelNuts39,

All - the point I'm trying to get across is that we're in the realms of theory here, and there are too many variables about which we are unsure to be able to make a definitive assertion one way or the other. For example, we know the speeds came back at about 02:10:35 (shortly into the zoom climb) - did this affect the C*/pitch angle transition at any point?

My personal assumption is that the THS may have continued to trim NU to maintain the commanded FPA or pitch angle, but I think it was the continued attpemts to pull up during this process that caused it to happen so quickly.

@ henra

See gums post, according to that it maintains 1 g with hands off, which would not be the case if it would maintain pitch.

If you would be right, BEA has it wrong:

This aircraft would under your definition of the NZ law maintain pitch attitude, but go into an descent to compensate for the loss of speed. It would be partial speed stable. It would not stall from level flight and it would not try to maintain altitude.

From FCTM Cathay Pacific:

Putting to one side the ineptitude of this crew, what worries me is that after 1100+posts (and several years) we still cannot really establish what does what and when in an Airbus control system. It kind of takes away the chances of 'Mr Average' getting a 'feel' for it, doesn't it?

Many years ago, I think after the very early Indian AB crash, someone told me that 23 different control modes were effective in pitch between cruise and touchdown - all 'un-announced', of course.

@BOAC

We're talking about how the system is likely to behave in extremis based on the information we have - to be certain we'd need to have an Airbus insider to spell it out.

But in the same breath, a pilot didn't have to understand the inner workings of the artificial feel system in older aircraft to be able to fly the things. As you've pointed out, all this technical back-and-forth is rendered somewhat moot by the fact that the initiating event was sustained and repeated control inputs which were thoroughly inappropriate for the conditions.

As to your other point, the fact that there are only 6 control laws in the A320 causes me to question your source.

So I ask the question:
Who is piloting a Airbus A330 .. the pilots or those who engineered the system ?

@jcj:

OK, so let's look at it this way. As a pilot, do you want your aircraft to behave as it does more than 99.9(rec)% of the time if something goes wrong, or to throw you into a control regime with which you've had little hands-on experience since converting to the type on top of the failure that caused it?

Dozy, BOAC said "modes" and you responded with LAWS.
I'd read more carefully.

I refer you to the table (matrix) of A330 flight laws that was quite popular, though it is vintage 2005, as discussions on this mishap began.

I'd invite you to look at that chart again.

There is an interaction between behaviors a laws and which protection does or doesn't turn on or off or is modified, see the notes.
These modes include all that ... with varied PROTECTIONS

PITCH ATTITUDE
LOAD FACTOR
AOA
HIGH SPEED
LOW ENERGY
LOW SPEED STABILITY
HIGH SPEED STABILITY
ANGLE OF BANK
MAN’UVER LOAD ALLEVIAT-ION
TURBUL-ENCE DAMPING
YAW DAMPING
TURN COORD
NORMAL LAW
ALTERNATE LAW 1
ALTERNATE LAW 2 (with subtle degrees ...)
DIRECT LAW
MECH BACKUP

If BOAC had meant "laws" I suspect that's what he'd have said.

The picture he was trying to paint was of the complexity of what a pilot faced in knowing what he can expect his plane to do, or to be able to do.

That's how I read it anyway.

As I understood it, the crash in India was a case of FMC-based mode confusion similar to that which caused the Air Inter crash in the Vosges - flight control laws and modes never really came into it in terms of the priamry factors behind the accident.

Response times of elevator and THS
 

Looking at the graph Dozy posted, noticed something I guess never computed before and apologize if it's been discussed to death and it just didn't make an impression. From 2:12.32 until about 2:12.49, pilot sitting in the right hand seat had predominately nose down inputs, with the first input going to the stop in the nose down direction. For this entire 17 seconds the elevators only moved to 15 degrees nose up from the bottom line of 30 degrees. The THS never moved.

Way too late, when the pilot in the left hand seat was finally given control, he input nose down almost to the stops from 2:13.45 to 2:13.52 so a total of perhaps 7 seconds until the dual inputs from the other guy evened out the inputs to around neutral. So full nose down for 7 seconds only moved the elevator to 15 degrees nose up and again, the THS never moved.

Just how long would a nose down input have to last to finally get the THS to full nose down? Took about a minute to roll to full nose up after predominately nose up side stick inputs from 2:10.47 to 2:11.45. Seems like an awful long time to have to wait in an emergency situation.

So can someone explain why it takes so long to get control surfaces to give the actions requested by the pilot. As I see it, the pilot wanted the nose down for 17 seconds and got nothing more than 15 degree nose up then the other wanted FULL nose down for 7 seconds and the computers once again only gave him nose up elevator. Does this have something to do with the cumulative inputs? Can a set of bad inputs cause the airplane to not respond quickly as was needed in this case??

No .. that I would not want that ..
The 99.9% is the behavior of the aircraft in flight which is operated by automatic systems , FBW and autopilot
The aircraft behave automatically .. that's is not of any help as experience when I must put my hand on the stick for fly it in another law than normal
When I have a problem (meaning that the flight management system is not more reliable at 100%) .. I'd rather manage myself (and with crew) and not have a hybrid system .. that can execute commands (eg trim auto) in my place
Hence the words many time hear when something go wrong:
It's my aircraft

PuraVidaTransportGo in direct law (full manual) .. and problem solved even with no elevators ( if you are enough quick to move manualy the trim :) )

PuraVidaTransport,
In both cases the ND inputs were not strong enough and sustained in the time to allow the elevators to go from 30 deg UP to a DOWN position.
For the second case, the ND input by the PNF was weakened by the NU input made by the PF.

So 1/5 only of the full stick deflection, pretty far from "the stick was held at least half-back for the majority of time the THS moved to the full-deflection position" stated earlier by DozyWannabe.

The THS movement only really started with Stall Warning 2.

WHAT if Direct Law ?
 

If Direct Law rules after AP disconnect, the worst case scenario is the PF is still full back stick when the CPT is back, but the THS is still at 3 degrees, so the stall is not that developed and the stall warning keeps warning ...

Captain, what do you think now ?
I think we're in a stall. Sure I could guess better if I could see you are pulling like mad on your controls ...

Just to keep the Oozlum Bird from disappearing "you know where", the following extract(s) from the Airbus A330 Instructor Support Manual provide the following:-As an aside, the PNF must have said, "Alternate law, protections lost."

- while your contributions are important, DW, please read posts carefully. I made no comment on the cause of the accident.

Response times of elevator and THS
 

That question was asked before, and i hoped the final report would answer this quetion. Bottom line is, that you have to ask the automation logic, because the SS only orders a change of g. How much change was ordered with full ND SS? I dont know, there are tables with g values corresponding to respective SS deflection, but i dont know wether there are speed gains involved or wether the load factor protection could have played a role in this.
Load factor protection was the only protection left, and again, i dont know wether it has some speed dependent gains as well.

Maybe an overlay of the graph you mentioned with actual normal acceleration values would shed some light.

Right, and that came within a second after the PF started to pull.

Hmmm, I don't see the disagreement with BEA in my post. If you put the aircraft in an attitude where the thrust is not sufficent to sustain the speed it will progressively continue into a stall. That is what BEA basically says.

Edit: Maybe just to clarify: When I say attitude I refer to pitch and not to AoA.

edit2: I assume the notion of Flight Path in BEA's description might be a traslation thing. Per my understanding the FCS does not care about Flight Path in the proper sense. That's the FMS which does that. FCS cares about behaviour relative to the air, not navigational.

Let's assume you place the nose 10° above the horizon. My understanding is it will try to keep the nose pointed at 10° no matter what the AoA. Given the fact that thrust will not be sufficient at that altitude to maintain speed in that attitude and given the progressively increasing drag with decreasing speed it will go into the stall and will resist any natural tendency to drop the nose with elevator and once its authority limit is reached it will call the trim for help.

Could you point my to where my assumption differs from BEA's?

The lift, and hence normal acceleration, is a function of AoA. If normal acceleration differs from 1g, pitch will change towards 1g with stick neutral.

HN, franzl, henra:

As this discussion of C* and its nuances progresses. I begin to get the impression, that attitude flying isn't an allowable mode in the A330, unless you transition to Direct Law. Or, in the case of pitch control, force the issue by use of Trim Wheels.

That does not make sense to me, and means that I misunderstand something essential. Also, based on the comments of those who fly Airbus aircraft, I infer that you can fly the A330 like any other aircraft.

As noted a few posts up, when PNF made a significant nose down input, and it was reduced in scope by PF coming back onto the controls, we have a non trivial CRM issue: two pilots fighting over the controls.
I have the controls
You have the controls

Not sure if PNF (LHS pilot) would have saved the day had his inputs not been interfered with, but I don't care what kind of aircraft you are flying in: If there are two pilots, and two sets of controls, and there are two people making inputs in opposition to one another, the odds of it all ending in tears goes up by orders of magnitude. :(