AF 447 Thread No. 6

[JenCluse]

via 'Layer 8', a few days ago:

"(United States) Air Force wants studies impact of human reliance on autonomous systems

at http://www.networkworld.com/community/blog/air-force-wants-understand-impact-automated-s"

I wonder how long the planning for the study has been in place?

Please resume normal conjecturing.

[worried SLF]

Hi all! I don't understand why everyone seems to ignore that even the conversation between two pilots ever since the start of the event didn't resemble a conversation between pofessional pilots? No call outs, no attempt to analyse the situation professionally, no checklists mentioned, no procedures. How badly were they trained exactly and how average were they? I find it frustrating that so many people seem to be reluctant to accept that what was found so far was more than enough to bring down the plane. Two FOs that couldn't fly complemented by a capitain that couldn't command put in one cockpit and as a result an event of frozen pitot probes that is not terminal by nature developed into a catastrophe. 228 dead. What is so surprising that other factors are needed? That modern planes still require pilots to know how to fly? After all, that's what happened. Pitot froze, obscure UAS procedures - this is on the surface, but if you scratch it, it's the case of A/P disconnect and the pilot's inability to fly the plane - as shocking as it sounds. Add unfamiliarity with plane's features. Multiply by two, add the capitain who failed to take responsibility, command and control when called to interfere (it's academic that it might have been too late by then). 228 dead. And insistance on an aviation forum that modifications to the plane are due to prevent this. I am a lay person, but I understand that a pilot without flying skills and solid knowledge base is a dead man walking, 100% at the mercy of the plane (automation) and his colleagues and a liability in some situations. And this is what happened. No addditional causes are needed.

Why there is insistance that the "plane did something" or that it caused the crew to do what they did? How can anyone take seriously AF's bull**** about "complexity of man-machine interface" after the converstaion in this cockpit was revealed and the latest report published? There is no and shouldn't be any interface in any sophisticated technology equipment that allows unproficiency. What's next then? Kids-friendly nuclear power stations? Interface didn't play any role in the event and for this crew any different interface could be just another thing not to know. To believe otherwise one would have to ignore all that was said and also what wasn't said in this cockpit. We have pilots that demonstrated that they knew and understood their plane - the one they were working on - badly. There is no reason to conclude logically that if plane was designed differently they would have known and understood it any better. They ignores stall alarm for almost a minute without saying a single word about it and then it was stopped by design, it's a pure conjecture to say that if design was different they wouldn't ignore it.


I disagree with recurring theme "something must have been wrong with the plane even if the investigation doesn't say it". In the light of what is published so far I don't understand where this conviction comes from. I saw failures on many levels but they are not technology-related. Does anyone blame technology when using it without reading the manual and breaking it? If you have a piece of equipment but only ever know one button there will be problems. And IMHO any debate about technology in the context of this crash will mean that these failures won't be properly addressed. It is clear that advances in technology no matter how big don't mean that professionals skills can be allowed to deteriorate especially when they are meant to complement and in times to compensate for this technology. Not a rocket science to understand. However I will agree that debate is due on why this is still allowed to happen. And please lets not try to make plane manufacturers responsible for pilots skills. Maybe instead airlines should be made to EARN their share of profits and prevented from attempting to take a free-ride on the back of manufacturers by manning new planes with untrained crews? Free-riding is unfair competition, and because as the result of poor training events like frozen pitot are turned into fatal, this is also a murder.

[Machinbird]

Mbird - what is more deep-rooted and more difficult to eradicate is the natural instinct to pull back hard when you see the ground/altitude zero rapidly approaching. Sometimes it works and sometimes.....................
BOAC,
Been there, avoided the temptation to pull right back. Still here to pester you.:}

It may not just be ground proximity that triggers the pull reflex. Think about some of the accidents you know where the guy flying pulled the wings off.

Were they all at low level? Doesn't "when bank angles exceed their comfort levels" factor in?

[jcjeant]

Hi,

Mbird - what is more deep-rooted and more difficult to eradicate is the natural instinct to pull back hard when you see the ground/altitude zero rapidly approaching. Sometimes it works and sometimes.....................

Methink .. the best training (for pilots) for self control will be a training for deep scuba diving
There .. 1 second of panic = immediate dead

[CONF iture]

Thanks a lot for the document mm43.

In paralele to the outer ailerons inaction, another point of interest :
in low speed, the auto pitch trim stops at Vc prot (below VLS) and natural longitudinal static stability is restored, with STALL WARNING at 1.03 VS 1g
This is in page 126 and relates to Alternate Law.
If I get it right, the THS should NOT have moved ... but it did !?
What was happening in that aircraft ?
This reminds me the WRG fault message ... only Svarin was interested to discuss.

[ChristiaanJ]

.....This reminds me the WRG fault message ... only Svarin was interested to discuss.I was interested as well, but the discussion petered out through lack of useable data.

[Hamburt Spinkleman]

If I get it right, the THS should NOT have movedYou get it wrong. In ALT2 with loss of 2 or more ADRs low speed stability is lost.

You should have continued readin thru page 127.

[DozyWannabe]

@Hamburt Spinkleman:

Correct - the important paragraph is on the following page:

In ALTN 2 the roll control is roll direct. In certain failure cases such as loss of VS 1g computation or loss of 2 ADR, the longitudinal static stability cannot be restored at low speed; in case of loss of 3 ADR it cannot be restored at high speed.

There's a reason that Alternate 2 is marked with a very clear "(NO PROT)" on the ECAM.

[CONF iture]

Hamburt,

My point of interest here is NOT the low speed stability or the roll control ... but the autotrim.
I think the autotrim activated as it did not have to.

Now, if it's normal by conception to autotrim under stall warning, it is aberration.

[mm43]

CONF iture; ChristiaanJ;

I raised the THS auto trim matter by PM with another poster a couple of days ago, and for the benefit of all, the excerpt from the A330 Instructor Support Manual follows:-
http://oi54.tinypic.com/k96jyd.jpg
The ALTN 2 provision clearly indicates that the longitudinal static stability cannot be "automatically" restored.

[DozyWannabe]

Now, if it's normal by conception to autotrim under stall warning, it is aberration.

This is a serious question, I promise I'm not being facetious here... Why do you consider it an aberration?

Would you not use manual trim in a conventional aircraft if elevator authority was not enough to help you escape the stall, or indeed if the elevators jammed? Autotrim is simply a way of providing trim function through the sidestick, and in the Alternate Laws the pilot has full trim authority through the sidestick. Alternate 2 removes some or all of the soft protections (soft meaning that sidestick input can override them, so in this case there would have been negligible difference had Alternate 1 latched, because the PF did not let go of the stick throughout the sequence).

The very reason for the existence of Alternate 2 is because with certain sets of failures, the FCU is programmed to recognise that it cannot manage the flight, and so full authority is given to the pilot. This is entirely in keeping with the Airbus engineering philosophy, which acknowledges that the last line of defence is the pilot, but could potentially cause problems in an operational sense if the airline concerned has used the existence of the FCU protections in approximately 99.9 (rec) of flight time to under-train pilots in manual handling.

In terms of systems architecture, Stall Warning is part of the notification subsystem and is not connected to the FCU protections in any way. The purpose of the notification subsystem is to tell the pilot, who in Alternate Law has full authority, if the aircraft is in a non-normal/dangerous state. The design relies on the pilot to heed that warning and take appropriate corrective action, as is the case in any other airliner. The notification and protection systems are complementary rather than co-operative or consequential - if the protection systems are functioning then the severe notification logic (e.g. overspeed warning, stall warning) should sound either briefly while the protections compensate, or not at all.

One of the fundamental misunderstood perceptions of the Airbus (and I suspect Boeing) FBW systems is that they are seen as a monolithic system in which everything is tied together - they are not. FMS/AP is distinct from FCU is distinct from reporting systems - they interoperate, but they do not form a cohesive whole from an engineering standpoint - nor should they lest they risk creating single points of failure. The protections are part of the FCU architecture, and the reporting systems are part of the avionics.

The design of the FCU states that absent specific parameters (in this case ADR data), then the pilot should have full authority - the safe limits will be notified via the reporting subsystem to which the pilot is expected to respond in a timely and correct manner, just as has always been the case before the days of FBW and the FCU system.

I don't see anything inherently wrong in retaining autotrim in Alternate Law, but the ramifications of retaining it must be trained for and understood.

[Lyman]

Doze,

THS: It was not retained, it resumed, having stopped at ALT LAW2 Latch.

You gonna leave your post as is?

[RetiredF4]

mm43

we had this discussion in thread 3 ages ago. I think it was TAKATA who explained it there.

The stopping of the THS trim is a function of the static longitudonal stability protection and is only available in Alt1, not available in ALT2.

In certain failures like 2 ADR which was the case with AF447, the longitudonal static stability cannot be restored, which means imho that THS trim remains available.

I would be cautious with the reference anyway, the instructor support manual is from 2001 and has been replaced by the FCTM some years ago. In the FCTM the wording is similar, but the part with the trim is missing. That is for the A320 as well as the A330 /340 family.

Also the Technical Training manual states, that in ALT2 only Load factor protection would be available.

[mm43]

RetiredF4;

I actually agree with you!

The ALT2 proviso "kills" further discussion on it.

[HazelNuts39]

This is what BEA #3 (p.41) says about it:
At the request of the BEA, Airbus conducted a simulation of the operation of the flight control computers, which involved recalculating the movements of the elevators and of the trimmable horizontal stabiliser (PHR) based on pilots’ inputs and compare the results against FDR parameters. This simulation could be continued up until the end of the flight. The recalculated deflection angles for the elevators and the PHR are consistent with the parameters recorded.

[Coagie]

worried SLF: Because it's unbelievible that none of the three pilots knew what they were doing. Admitting that these three were incompetent is difficult, because their incompetence tarnishes a historically heroic profession. Air France would have been better served by actors. At least the CVR would have sounded like professional pilots. Now Air France has no place to hide. Maybe some of the other carriers will learn, and if they don't hire competent pilots, they might, at least, try to train them to be competent pilots.

[DozyWannabe]

@Lyman - just this once I'll respond. You show me where it stops and resumes, and then I'll contemplate taking you seriously, otherwise I assure you that trying to engage me further will remain an exercise in futility.

[mm43]

@Dozy;

Think you'll find that Lyman will leave it alone.

Have been back and highlighted the ALT-2 proviso in the graphic a few posts back. As HN39 has carefully pointed out, the BEA have had Airbus prove the aircraft behaved as expected.

[Mr Optimistic]

Is Interim Report#3 still available at the link in the first post ? Tried to download it but no luck. Looking for extracts of the conversation to see the extent of cross communication of SS actions. May be one thing to not see what was happening, surely not deaf as well.

[DozyWannabe]

@Mr. Optimistic:

Both links (FR and EN) working fine for me. PM me if you have further issues and I'll send you my local copies.

[CONF iture]

This is a serious question, I promise I'm not being facetious here... Why do you consider it an aberration?
At this point, you would probably need to go up in the air and experiment, you on the controls, what a stall is.

The answer why you do not trim up approaching the stall and certainly not further up once stalled will get obvious.

Airbus put in place some restriction to not autotrim all the way when everything works fine, but decided autotrimming all the way was the way to go when data acquisition was known as deficient … it is aberration.

[CONF iture]

The stopping of the THS trim is a function of the static longitudonal stability protection and is only available in Alt1, not available in ALT2.
Can't remember such post ... and didn't manage to locate it either ?

I did not know the FCTM replaced the instructor support manual. For the little I've seen since yesterday, the instructor support manual seemed to propose interesting stuff that I have never seen in FCTM.

What about the Technical Training manual ... Any link ?

[Machinbird]

Quote:
Originally Posted by CONF iture http://images.ibsrv.net/ibsrv/res/src:www.pprune.org/get/images/buttons/viewpost.gif (http://www.pprune.org/tech-log/460625-af-447-thread-no-6-a-76.html#post6779975)
Now, if it's normal by conception to autotrim under stall warning, it is aberration.

Dozy
This is a serious question, I promise I'm not being facetious here... Why do you consider it an aberration?

I agree with CONFiture in this regard.
The Bus has little to restrict it from autotrimming nose up in Alt2, beyond the stall AOA, and I believe this is a potentially dangerous characteristic.

I recognize the Airbus is Certificated now under EASA CS-25.

If you look at comparable language in FAA AC 25-7B, Flight Test Guide for Certification of Transport Category Airplanes http://www.airweb.faa.gov/Regulatory_and_Guidance_Library/rgAdvisoryCircular.nsf/0/9f2b881724fefdf486257868003f38ee/$FILE/25-7B.pdf

(3) Procedures.
(a) The airplane should be trimmed for hands-off flight at a speed 13 percent to 30
percent above the anticipated VSR, with the engines at idle and the airplane in the configuration
for which the stall speed is being determined. Then, using only the primary longitudinal control
for speed reduction, maintain a constant deceleration (entry rate) until the airplane is stalled, as
defined in § 25.201(d) and paragraph 29c(1) of this AC. Following the stall, engine thrust may
be used as desired to expedite recovery.

Here is an aircraft being stalled for certification purposes while trimmed at a speed 13 to 30% above VSR. Now do you really want to give an automatic system the ability to trim to a higher trim value than the aircraft has been certified for?

A trim setting tells an aircraft to hold a certain AOA. To allow the trim to run up after stall warning says, "Lets increase the AOA some more." This reduces the aircraft's natural tendency to drop its nose in the stall.
If AOA for stall indication doesn't work in ALT2 because the Mach stall AOA correction cannot be calculated without a known good ADR, then they had better figure out another method of stopping the trim from autotrimming nose up after reaching VSW.

[BOAC]

I agree with 'aberration'

Hence my suggestion from PGF days of a requirement to 'push to over-ride' a trim warning when it tries to go past a 'sensible ' limit, presumably based on something like the trim for 1.x% of Vls at forward cg? Seems it might have helped the 737 at SPL as well as the Airbus family. What are the objections?

[DozyWannabe]

@Machinbird

The way you describe it, less knowledgeable people would assume that the aircraft autotrimmed nose up entirely on its own without any order or prompting to do so. Of all the UAS incidents on this type only one aircraft was trimmed in this manner and that was because the PF ordered it to do so. Autotrim is "automatic" only in the sense that it responds to pilot input when under manual control - it will not trim unless there is an input for which to compensate.

The Airbus may be capable of more than the certification standards in that regard, but importantly it needs to be recognised that the trim system is a little different to conventional airliners and this needs to be acknowledged in training and certification.

[Capn Bloggs]

The Airbus may be capable of more than the certification standards in that regard, but importantly it needs to be recognised that the trim system is a little different to conventional airliners and this needs to be acknowledged in training and certification.
And must be recalled in the worst possible scenario, even though I believe the trim, correct me if I am wrong, is NEVER used in normal operation?

[Zorin_75]

And must be recalled in the worst possible scenario, even though I believe the trim, correct me if I am wrong, is NEVER used in normal operation?
Once you're in that deep manual trim might speed things up somewhat (?), however first thing that would have to be recalled is to get the nose down, and the THS would follow suit just as it did on the way up (ideally you probably wouldn't even have pulled up until the trim went to the stops in the first place).

[jcjeant]

@Machinbird

The way you describe it, less knowledgeable people would assume that the aircraft autotrimmed nose up entirely on its own without any order or prompting to do so. Of all the UAS incidents on this type only one aircraft was trimmed in this manner and that was because the PF ordered it to do so. Autotrim is "automatic" only in the sense that it responds to pilot input when under manual control - it will not trim unless there is an input for which to compensate.

The Airbus may be capable of more than the certification standards in that regard, but importantly it needs to be recognised that the trim system is a little different to conventional airliners and this needs to be acknowledged in training and certification.

The point of Machinbird is not the one you described (or you have not understand his post)
The way you describe it, less knowledgeable people would assume that the aircraft autotrimmed nose up entirely on its own without any order or prompting to do so
Even a less knowledgeable people will understand the Machinbird point if they read carefully the entire post of Machinbird
Please read again carefully the entire post of Machinbird

[Machinbird]

The Airbus may be capable of more than the certification standards in that regard, but importantly it needs to be recognised that the trim system is a little different to conventional airliners and this needs to be acknowledged in training and certification. Dozy, I'd bet the test pilots never stalled the bus with full nose up trim. If so, that is not a certified condition. The only way the system would then be allowed to achieve that condition would be in some highly improbable one in a billion chance. But we definitely now know that the odds are much higher than one in a billion. I think this is an issue in the aircraft's certification that needs correction.
Now if EASA agrees, we can start fixing the problem.
Either test fly the aircraft with full nose up trim and declare it meets stall certification standards, or prevent the aircraft from autotrimming beyond the levels it was certified for.

[BOAC]

prevent the aircraft from trimming beyond the levels it was certified for. - NO NO NO! We do not need even MORE 'system' interference in how we fly. Crews MUST have full authority on all controls such as trim, but need to know when it is 'aberrating'??. Bring back pilots.:ugh:

[Machinbird]

Relax BOAC.
I changed 'trimming' to 'autotrimming' to make it clear the type of trim we are referring to.
Do you really want your trim system quietly doing things behind your back? (Actually out of the range of your normal scan).

[ChristiaanJ]

To me, "quietly" is the keyword, no?
I've repeatedly suggesting bringing back the 'pitch trim bicycle bell', but nobody has really picked up on that.
Yet, it's sufficiently different from the other 'bells and whistles' in the cockpit.

I still wonder why something judged to be useful on Concorde was deleted on the A330 (and presumably the other A3X0s as well ?)

[RetiredF4]

Machinbird

Do you really want your trim system quietly doing things behind your back? (Actually out of the range of your normal scan).

Although that question is not pointed at me, i like to comment on it.
The whole FBW-system as it is designed would not work without autotrim. Fundamentally nothing is wrong with it except the point when part of the FBW logic goes down the train. At that point the system should be at least as easy to understand and be as easy to fly as a normal conventional system.

In the case we discuss here the system is excellent and far better than a conventional design when everything is online and it gets a lot more difficult when part of the system goes offline.

Holding the stick back in a stuation where the aerodynamic performance of the aircraft does not allow further pitchup due to missing lifties would leave the trim in a conventional aircraft unattended , but caused an unnecessary full NU trim state in the case of AF447.

Whoever thinks, that this built in system function is a clever design should reconsider.

[DozyWannabe]

It's a perfectly rational design, because what it is doing is performing in exactly the same manner it does in Normal Law. A pilot whose commercial experience consists entirely of FBW Airbus time is going to find it easier to have autotrim on when things go awry than to suddenly have to cope with manually trimming the aircraft.

Let's be honest, how hard is it to understand that one does not make large and sustained inputs - which is what these were, make no mistake about it - without a bloody good reason, even more so at altitude?

[xcitation]

Dozy, I'd bet the test pilots never stalled the bus with full nose up trim.

Agreed. Test pilots of Airbus gently dance on edge of the flight envelope to ensure that the a/c can return quickly. Crudely departing with high energy is undesirable as the energy/time required to return is that much greater. That way the test pilot can stay ahead of the a/c.

[Machinbird]

Dozy
It's a perfectly rational design, because what it is doing is performing in exactly the same manner it does in Normal Law. Nope Dozy. In Normal Law it stops trimming at the V alpha protect if I remember correctly, but in ALT 2 it keeps on cranking away if airspeed is invalid until it hits the stops (or nearly so) as in the case of AF447.

But why make ordinary airline pilots into test pilots? Going boldly where no test pilot has gone before.

It is a certification issue. Gotta prove it meets the stall characteristic requirements with full nose up trim or stop it from autotrimming there. And for BOAC's benefit, no need to stop the pilot from doing what he needs with the trim. It is assumed that the pilots will act rationally.

[Organfreak]

DW,
It's a perfectly rational design, because what it is doing is performing in exactly the same manner it does in Normal Law.

How could it be "rational" when nothing else does (perform in the same manner as in NL)? This "rational design" may have been part of the reason they were all killed that night. Granted, the pilots may have not known a lot of things they should, but I am entirely on the side of, and in awe of, the estimable Gretchenfrage, on all of this stuff. I think it is relatively clear (as clear as anything in these thread can be) that the AB FBW is over-designed, by and for nerds.

I may not be a real pilot, but I do have a rawther large brain! :8

[Zorin_75]

It is assumed that the pilots will act rationally.
Isn't that assumption somewhat at odds with the condemnation of autotrim?
AF447 didn't enter the stall in full NU trim, from then on it took over half a minute of irrational control inputs to get there. Pilots acting rationally (even if only after the stall warning) would have never gotten to that trim state.
Not saying trimming up in a stall is not a bad feature, but how could it have been a factor here if there never was any useful attempt at recovery in the first place?

[A380 Jockey]

An autotrimmed FBW airbus which has been held in nose up condition will now sense the new nose up condition as the new neutral in a stalled condition ie a low airspeed condition. This further aggravates the stall and the rest then,as they say, is history.
Its as simple as that.
:)

[DozyWannabe]

How could it be "rational" when nothing else does (perform in the same manner as in NL)?

Look up the concept of "graceful degradation".

You're talking about handing a pilot who may have not flown under manual trim conditions since their last checkride (or possibly even earlier) the responsibility of handling the aircraft with manual trim (which he or she will not be used to) in the worst possible situation. How is that any better than a system which retains as much of the Normal Law handling characteristics (which the pilot would be most used to) as possible?

I think it is relatively clear (as clear as anything in these thread can be) that the AB FBW is over-designed, by and for nerds.

And with a comment like that I can't help but hope that one day, a member of the late Captain Gordon Corps' family bumps into you and gives you a well-deserved ding around the ear.

I may not be a real pilot, but I do have a rawther large brain! :8

With a head that big, I'm surprised you fit through an average-sized door.

[HazelNuts39]

If anyone is wondering (like me) to what extent AF447 might have been different if THS had not autotrimmed beyond stall warning, look up
Owain Glyndwr's post #450 (http://www.pprune.org/tech-log/460625-af-447-thread-no-6-a-23.html#post6663753), from which the following is quoted:
The nose was being held up by the application of elevator.

Of course, the THS setting made the elevator’s job easier, and if the THS had been (sensibly in my view) restricted to 3 deg the eventual AOA would have been lower, How much lower you can get from the first chart – with 3 deg THS and 30 deg elevator you could expect to arrive at 35 deg AOA – big deal! – you are still well stalled and although the descent would have been shallower the end would have been the same unless he had recognised early on that he was in a stall.

[RetiredF4]

Dozy,

there is no situation in cruise, where autotrim is improving a situation in ALT2 like that encountered by AF 447. If you can think of one, let me know. As we know, autotrim is inhibited in ALt1 at V-prot anyway and in direct law as well.

Then let´s look at it from the other way around.
How difficult would it be to expand the present feature from ALT1 to situations in ALT2 maybe with different and independent values? Is it rocket science?

Concerning your graceful degradation explain this degradation with autotrim of the THS:

In Normal LAW autotrim (a rally nice feature)
In ALT1 Law autotrim prohibit at VC-prot (looks sound to protect from entering unknown territory)
In ALT2 Law autotrim (that is your part to explain)
In Direct LAW autotrim off (looks sound to protect from entering unknown territory)

I would name that one non linear degradation (far off from gracefull).

Concerning your naming manual trim again, please do not push aside the inputs made to that issue from Clandestino. http://www.pprune.org/6770409-post394.html
Quote:
Originally Posted by RetiredF4
Manual THS trim (nowhere mentioned in the books except in direct law, and never trained)
Clandestino:
Rest of youth post is spot-on, I just have to comment on this. It is not trained because it doesn't work. You can forcibly move the wheel to position of your choice but the FBW will return it to the position it deems to be necessary once you loose your grip and all the while will try to combat the trim with elevator to satisfy G demand.

There's no use and no need for manual trim while auto-trim is working..

That should straighten up the manual trim option once and for all. It´s even worse, because the functioning autotrim in ALT 2 prevents manual trim in its original sense of implementation.

There might be pilots insane enough to pull on the yoke or sidestick in conventional and FBW aircraft until the aircraft stalls, but no one would trim while pulling.

To understand that above statement let me explain normal trim behaviour. With the intention to climb the pilot uses the yoke to change the pitch, once that change is reached he uses the trim to get rid of the pressure on the yoke. The trim comes into play when the change is achieved and not in the timeframe, where the change takes place.

In case of AF447 without autotrim the pilot would never ever have tried to achieve the desired flight path change with manual trim, because that is not the way to do it.

[Organfreak]

Mr. Wannabe said,
Look up the concept of "graceful degradation".

I'm already familiar with it, thanks for the advice anyway. I left your period where you put it, out of courtesy.

With a head that big, I'm surprised you fit through an average-sized door.

Assuming facts not in evidence, sir. I never said that I did. In fact, my cranium is so large I've been confined to the barn for years.

[Hamburt Spinkleman]

It is not ALT2 in itself that causes the loss of low speed stability and the uninhibited travel of the THS. It is only the loss of 2 or more ADRs that causes this specific condition. It is possible to be in ALT2 with a THS that behaves as in ALT1.

To say that no pilot would trim nose-up while pulling is a bit presumptious i think. I do not find it inconceivable at all that the PF would have trimmed nose-up, perhaps even to the same extent the autotrim did, if he had the means to do so. He was after all back to mechanical stop on the stick for a full 30 seconds.

Not much a what he did was the way to do it, but he did it nonetheless.

[DozyWannabe]

Franzl,

there is no situation in cruise, where autotrim is improving a situation in ALT2 like that encountered by AF 447.

The FCU has no concept of "cruise", as far as I am aware. It is a real-time processing system that does the job it was designed to do very well, but it is intentionally quite a simple beast in terms of design (because the simpler a system is, the less things there are to go wrong).


If you can think of one, let me know. As we know, autotrim is inhibited in ALt1 at V-prot anyway and in direct law as well.

You have to look at this from a systems perspective to understand. Autotrim as a system is never "inhibited" in any law other than Direct and Manual Trim Only - it is the *protections* that prohibit the aircraft from leaving the flight envelope by preventing any commands - either manual or automatic - from doing so, and if necessary providing corrective commands to keep the aircraft within the flight envelope. The protections are a completely discrete subsystem that is loosely-coupled to the others that make up the FCU system.


Concerning your graceful degradation explain this degradation with autotrim of the THS:

In Normal LAW autotrim (a rally nice feature)
In ALT1 Law autotrim prohibit at VC-prot (looks sound to protect from entering unknown territory)
In ALT2 Law autotrim (that is your part to explain)
In Direct LAW autotrim off (looks sound to protect from entering unknown territory)

Again, that is not correct if you look at the architecture as a whole. I'll repeat for clarity - Autotrim as a system is not "prohibited", nor "turned off" in any other mode or Law than Direct or Manual Trim mode. If the protections are active, then the autotrim commands will be treated as any other command that takes the aircraft out of the flight envelope and corrected accordingly. Think of it as two separate processes running alongside each other rather than as an integrated whole.

Put even more simply, imagine two people on either side of a wall that has a two-handled saw poking through it. The person on one side (let's call him Otto Trim) is told to push the saw forward and the person on the other side (who is physically stronger and called Pete Tection) is told to not let the saw through past, say, two-thirds of it's length. Pete will always stop the saw at the limit and will try to return it to the prescribed position if it goes past, but he is not explicitly aware that Otto's on the other side trying to push it because the wall is in the way, and as such does not interfere or communicate with Otto directly - all Pete knows is that he mustn't let it through past a certain point.

That should straighten up the manual trim option once and for all. It´s even worse, because the functioning autotrim in ALT 2 prevents manual trim in its original sense of implementation.

What do you mean by "original sense of implementation"? Using manual trim to set the THS to a certain degree of pitch and letting go will cause the autotrim to try to hold that pitch, unless a demand is then made on the sidestick for which the autotrim will try to compensate. It doesn't "remember" what it was doing X number of seconds or minutes ago, it simply tries to hold the trim set by the autoflight or pilot commands.

In any case, using manual trim and then holding on to the wheel will prevent the autotrim from re-engaging.

There might be pilots insane enough to pull on the yoke or sidestick in conventional and FBW aircraft until the aircraft stalls, but no one would trim while pulling.

"No pilot would ever..." is an impossible statement to prove.

To understand that above statement let me explain normal trim behaviour. With the intention to climb the pilot uses the yoke to change the pitch, once that change is reached he uses the trim to get rid of the pressure on the yoke. The trim comes into play when the change is achieved and not in the timeframe, where the change takes place.

I know you're not intending to patronise me, but I am fully aware of how trim works in the conventional sense - I even did it once or twice in the Chippy. Here there is no tactile feedback, so "trimming to the pressure" is impossible, and performing the same thing visually using the ADI as reference would be physically exhausting on a day-in, day-out basis. This was part of the reason autotrim was developed because the flight control design was a completely new paradigm.

In case of AF447 without autotrim the pilot would never ever have tried to achieve the desired flight path change with manual trim, because that is not the way to do it.

The only way you could possibly know that is with a Ouija board. The last time he trimmed manually would have either been during type conversion training, recurrent training or even back when he was flying trainers - because the Airbus FBW flight control paradigm does not require it except in case of major failure.

If all goes to plan I'm going to be doing some exciting research this weekend and I'll be able to argue from a much surer footing. If it turns out I've been wrong about anything you guys will be the first to know.

@Organfreak - Those were not intended to be personal insults, but I must confess I was flabberghasted with the throwaway manner in which you described the Airbus system design and those who developed it. I emphatically request you find out who Gordon Corps was, what he accomplished and how he died - and if you have the time, have a browse through the Flight International archives relating to A320 development (in which he figures very prominently) before you put your foot in it again.

[Dav66id77]

takata re: SPIKES.

The Airbus has this elegant escape from CFIT? NON?

Max power, roll full, and pull back max.

A max effort, 'at the limits' safety manuever. During which the a/c nibbles at Stall? and wing drop? No sweat, the Bus knows SPIKES.

The Pilot doesn't. As above, for Alternate Law, the STALL WARNING needs some looking into.

[Lyman]

Hamburt Spinkleman. Howdy. Isn't STALL WARN inhibited in Protection?

G Prot? Vc Prot? Load Factor Prot? The pilots did not hear the STALL WARN at the top of climb, Right? It shows in the trace, but where is the SV Cricket inhibited? On which side of the FDR? Is it silenced at the panel?

Thanks, hope you can help.

[DozyWannabe]

The Stall Warning and "cricket" annunciation is on the d*mn CVR - if they were picked up on the CAM or the hot mikes, it was audible in the flight deck. Must try harder.

Radio silence resumed.

[Lyman]

They clearly did not hear the WARN, and I do believe that the WARN is inhibited in prot, but you would know that?

Did you hear it?

You have been given one of three things, the 'annunciations'.

You are entirely willing to accept two absurdities to believe the one.

1. They did not hear it.

2. They heard it, but ignored it, without comment.

This is acceptable to you, because you are wanting to stop at what makes your position correct.

You have no desire to hear or see further, your mind is made.

There is one mind that will foreclose all doubt, the one that will be correct at all costs, even prior to investigation.

[Capn Bloggs]

Here there is no tactile feedback, so "trimming to the pressure" is impossible, and performing the same thing visually using the ADI as reference would be physically exhausting on a day-in, day-out basis. This was part of the reason autotrim was developed because the flight control design was a completely new paradigm.
A new paradigm? Perhaps that is the real issue. As for pilots trimming into a stalled situation, of course that is ridiculous. If you don't discount that, then you'd have to say there is no place in the cockpit for pilots becuase they are too incompetent.

[DozyWannabe]

Au contraire, sir - I'd *love* to know why they didn't hear it - and if it didn't sound then I'd be the first to say that would be a major problem. But unless you're accusing the BEA of falsifying evidence, then the presence of the Stall Warning alarm on the CVR is a documented fact, which closes that line of enquiry and this conversation.

@Capn Bloggs - As with any work, on the line I'm sure you'll encounter super-competent pilots, pilots who shouldn't be allowed in a flight deck and pilots of every level inbetween, with most hovering around an average competency level (which in most cases requires more general competence than a lot of jobs). However, you'd think that "No pilot would...":


Take off without permission
Retract LE devices below a safe speed
Fail to deploy high lift devices at all prior to take-off
Fail to turn on engine anti-ice when taking off into a blizzard
Shut down a healthy engine and leave the damaged one running


and yet pilots have done all of those things, and some pilots have repeated the same mistakes made by their predecessors. This is not to say I think pilots are incompetent, far from it - because of the staggering number of flights that get to their destinations safely every day - but "No pilot would..." turns out to be a demonstrably false assertion. "No pilot *should*..." on the other hand, and you won't hear me arguing. :)

To go back to your question, the change in paradigm was based on a desire to start from a "clean sheet" when it comes to flight deck ergonomics based on the technological leaps made during the Space Age, not because of any desire to impugn pilots as a whole.

[BarbiesBoyfriend]

For pitys sake.

If a pilot is holding the yoke full aft, that tells a story.

If a pilot is holding his wrist half-cocked back ,in a pocket in the corner of the cockpit, at night. that's another story.

You can yarn it however you like , but that's how it is.

[DozyWannabe]

@BarbiesBoyfriend

The ending of the story (aircraft pitches up to an attitude that has no business being achieved in cruise) is the same in both cases however. The proper response to the story ("That's not right - are you really doing that? I have control.") is the same.

[CONF iture]

"That's not right - are you really doing that? I have control."
Why such additional layer of uncertainty that yokes would naturally prevent ?
Also, pilots do silly stuff, let them do those by themselves, automation is here to help not to put you deeper in the mud.
Let him trim by hand if that's really what he wants.

[DozyWannabe]

Why such additional layer of uncertainty that yokes would naturally prevent ?

Because someone above either of our paygrades on both the pilot and engineering side decided that the benefits outweighed the drawbacks.

Think about it logically - the difference between looking over at the yoke to confirm what you're seeing on the ADI or out the window (unless you're watching your colleague's inputs like a hawk, which I suspect isn't the case given the average PNF workload) and asking them directly is what - a second, maybe two?

Also, pilots do silly stuff, let them do those by themselves, automation is here to help not to put you deeper in the mud.

And the autotrim does help - specifically it helps a pilot who is in manual control maintain the attitude he is requesting, and if the protections are available they keep the attitude within the flight envelope. Lose the protections and the autotrim will do it's best to keep up with demand, even if that demand is inappropriate.

As part of the research I was talking about, it is possible that the A320's equivalent of Alt 2 (Alternate with no speed stability) may function differently when it comes to trim, I'll keep you updated.

Let him trim by hand if that's really what he wants.

Which he (or she) can do by holding on to the trim wheel if he or she does not like what the autotrim is doing (although I have yet to hear of autotrim behaving inappropriately due to a software or hardware failure).

[Capn Bloggs]

Because someone above either of our paygrades on both the pilot and engineering side decided that the benefits outweighed the drawbacks.

Easier to build, saves money. That's it! Oh, and SS give you unimpeded access to your dinner tray. :cool:

[Machinbird]

Those who think yokes should be the cure-all for what the other guy is doing should look at this accident originally posted by Netstruggler in the current Rumors & News AF447 thread.
http://libraryonline.erau.edu/online-full-text/ntsb/aircraft-accident-reports/AAR97-05.pdf
An interesting mix of Perpignan and AF447 on a DC-8 airframe. No loss of airspeed, but a stall warning failure combined with a PF who didn't let the nose down promptly at the stall and two staff pilots + a FE on a Post Maintenance Check Flight.
The guy in charge (PNF) didn't apparently realize that they were still stalled and allowed the PF to continue to hold the Yoke too far back to recover.
When the wings started rocking-it was like they had both forgotten all they should know.

Oh yes, PF had trimmed too far nose up. Do you see a possible pattern?:hmm:

[Zorin_75]

Oh yes, PF had trimmed too far nose up. Do you see a possible pattern?Though it should be noted that according to the NTSB this has not been a factor...
Some of the conclusions sound quite familiar:
The pilot flying applied inappropriate control column back
pressure during the stall recovery attempt in an inadequate
performance of the stall recovery procedure established in ABX’s
operations manual.Although the pilot flying trimmed the airplane below the
recommended minimum trim speed for the clean stall, this action
did not contribute to the accident.The pilot not flying, as the pilot-in-command, failed to recognize,
address and correct the pilot flying’s inappropriate control inputs.The inoperative stall warning system failed to reinforce to the
flightcrew the indications that the airplane was in a full stall during
the recovery attemptThis accident might have been prevented if the flightcrew had been
provided a clear, direct indication of the airplane’s angle of attack.

[GarageYears]

There seems to be a tendency to let the mass of opinion become fact...

- Autotrim did not stall the airplane, the PF did that quite successfully and then held the SS NU for a good long time, resulting in the NU trim eventually. The stall would have occurred with or without autotrim. My opinion is this is a distraction and something for the anti-Airbus brigade to fixate on. That the PNF did not notice the eventual trim demand seems to be a topic that has received scant attention - I believe those trim wheels would have been rotating for a good while. But no comment???

- The stall warning system worked and sounded at one point for, what was it, 53-54 seconds continuously, with no comment from PF or PNF? Eventually the damn thing gave up since the airspeed dropped below the 60kt cutoff, but come on!!!... From then on the vertical speed and pitch attitude should have told ANYONE with half an understanding of the basic laws of aerodynamics that the airplane was no longer flying in the true sense and was stalled.

- I am also surprised that the PNF never seemed to really comment on the initial zoom-climb? Why not? Nothing like "Hey, we're at FL380 what are you doing?"

There seems to a lot that the PNF didn't do? Question the climb, question the stall warning, the pitch/VS combination. Unfortunately, he seems to have been somewhat fixated on the necessity to call the Captain back to the cockpit, just at the time when his full attention should have been on what the aircraft was doing.

[CONF iture]

Machinbird Zorin_75

It is a fascinating report netstruggler gave us the opportunity to read.
It is a reminder that stalling an airliner, intentionally or not, is a very serious matter.

Those who think yokes should be the cure-all for what the other guy is doing ...
You are correct, it cannot be the miraculous solution.
But it is one of the great piece of equipment that a multicrew has in its toolbox.
It is free non verbal communication between all in the flight deck.
The yokes have the merit to make that information available.
The sidesticks have the default to supress it.

During the attempted stall recovery, there were several indications of the PF’s excessive aft control column inputs that should have suggested to the PNF that, as the PIC, he needed to correct the control inputs and recover from the stall. These included the position of the control column, which was, at times, being held in the full aft position by the PF; continued aerodynamic buffet; the extreme pitch down moments (stall breaks) accompanied by roll-off into steep bank attitudes; engine compressor surges; and the instrument indications of low airspeed and high rate of descent. The Safety Board evaluated why, despite these cues, the PNF did not take control of the airplane or otherwise intervene effectively as the PF held the airplane in a stalled condition all the way to impact.
In my book, an aft control column, is the best shot to wake me up.

[RetiredF4]

Hi DW, being evasive?

Originally Posted by RetiredF4
there is no situation in cruise, where autotrim is improving a situation in ALT2 like that encountered by AF 447.

Reply by DW
The FCU has no concept of "cruise", as far as I am aware. It is a real-time processing system that does the job it was designed to do very well, but it is intentionally quite a simple beast in terms of design (because the simpler a system is, the less things there are to go wrong).

I described a flight phase as opposite to another flight phase (like one being close to the ground) where autotrim could help in Alt2 even at the extremes of the flight envelope (i´m not saying it would). My above statement stays. Should have explained to you, looks like.


Originally Posted by RetiredF4
If you can think of one, let me know. As we know, autotrim is inhibited in ALt1 at V-prot anyway and in direct law as well.

Reply by DW
You have to look at this from a systems perspective to understand. Autotrim as a system is never "inhibited" in any law other than Direct and Manual Trim Only - it is the *protections* that prohibit the aircraft from leaving the flight envelope by preventing any commands - either manual or automatic - from doing so, and if necessary providing corrective commands to keep the aircraft within the flight envelope. The protections are a completely discrete subsystem that is loosely-coupled to the others that make up the FCU system. .

Is it of relevance to this discussion? You may accept, that I look at it from an operators point of view. If the system is able to prevent trimming into a stall in ALT1 and Direct Law, it should be able to design one, that can do the same in ALT2.

But you did not answer my question.


Originally Posted by RetiredF4
Concerning your graceful degradation explain this degradation with autotrim of the THS:
In Normal LAW autotrim (a rally nice feature)
In ALT1 Law autotrim prohibit at VC-prot (looks sound to protect from entering unknown territory)
In ALT2 Law autotrim (that is your part to explain)
In Direct LAW autotrim off (looks sound to protect from entering unknown territory)


Reply by DW
Again, that is not correct if you look at the architecture as a whole. I'll repeat for clarity - Autotrim as a system is not "prohibited", nor "turned off" in any other mode or Law than Direct or Manual Trim mode. If the protections are active, then the autotrim commands will be treated as any other command that takes the aircraft out of the flight envelope and corrected accordingly. Think of it as two separate processes running alongside each other rather than as an integrated whole.

Put even more simply, imagine two people on either side of a wall that has a two-handled saw poking through it. The person on one side (let's call him Otto Trim) is told to push the saw forward and the person on the other side (who is physically stronger and called Pete Tection) is told to not let the saw through past, say, two-thirds of it's length. Pete will always stop the saw at the limit and will try to return it to the prescribed position if it goes past, but he is not explicitly aware that Otto's on the other side trying to push it because the wall is in the way, and as such does not interfere or communicate with Otto directly - all Pete knows is that he mustn't let it through past a certain point.).

You don´t need to explain how it is done, the technical aspect does not matter to the operator. Graceful would be logical, but this degradation is not.

- all Pete knows is that he mustn't let it through past a certain point. ). is exactly my point. Let Pete know, that autotrimming into a stall is no good idea, like Pete knows in Alt1 at Vprot and in direct law. Implement it, however you do it.

Originally Posted by RetiredF4
That should straighten up the manual trim option once and for all. It´s even worse, because the functioning autotrim in ALT 2 prevents manual trim in its original sense of implementation.

Reply by DW
What do you mean by "original sense of implementation"?

In any case, using manual trim and then holding on to the wheel will prevent the autotrim from re-engaging.

The original sense of trim is to aleviate loads on the system after a flightpath change / change of loadfactor has been achieved (long term).

FCPC will always try to hold a load factor demand, by using elevators in short term and autotrim in longterm. How to disable the elevators during manual trim? With SS, I know, but if properly used manual trim would not have been necessary in the first place.
But if the system would have reverted to manual trim like in ALT2 at Vprot or like in Direct LAW, THS would not have been trimmed Full NU by the FCPC via SS inputs.

Originally Posted by RetiredF4
There might be pilots insane enough to pull on the yoke or sidestick in conventional and FBW aircraft until the aircraft stalls, but no one would trim while pulling.
"No pilot would ever..." is an impossible statement to prove.

To understand that above statement let me explain normal trim behaviour. With the intention to climb the pilot uses the yoke to change the pitch, once that change is reached he uses the trim to get rid of the pressure on the yoke. The trim comes into play when the change is achieved and not in the timeframe, where the change takes place.

Reply by DW
Here there is no tactile feedback, so "trimming to the pressure" is impossible, and performing the same thing visually using the ADI as reference would be physically exhausting on a day-in, day-out basis. This was part of the reason autotrim was developed because the flight control design was a completely new paradigm.

You know very well that i´m fully aware about the functioning and the necessity of autotrim and do not question it. I even find it a clever and well thought out system. But the situation developing in ALT2 with AF447 was not being expected somehow and needs to be addressed and changed. It works in Alt1 at Vprot, why not do the same in Alt2? Nobody seems to be concerned to hinder autotrim in Direct Law, but you explain it would be difficult for the crew if autotrim would be hindered in Alt2 when predesigned values (aoa, speed, Trim value, take whatever would suit yourself) are reached?

Originally Posted by RetiredF4
In case of AF447 without autotrim the pilot would never ever have tried to achieve the desired flight path change with manual trim, because that is not the way to do it.

Reply by DW
The only way you could possibly know that is with a Ouija board.
If all goes to plan I'm going to be doing some exciting research this weekend and I'll be able to argue from a much surer footing. If it turns out I've been wrong about anything you guys will be the first to know.

You are going flying? Pull on the stick and start trimming while pulling (hope you have your chute with you. No need to use a Ouija board.

Overall design has to follow function, not vice versa.

[infrequentflyer789]

Those who think yokes should be the cure-all for what the other guy is doing should look at this accident originally posted by Netstruggler in the current Rumors & News AF447 thread.
http://libraryonline.erau.edu/online-full-text/ntsb/aircraft-accident-reports/AAR97-05.pdf


Yep. Yokes may or may not be better, but they provably do not cure the problem.

That report was also referred to way back in this thread: http://www.pprune.org/tech-log/173167-can-airliners-recover-stall.html

Post #18 (from Old Smokey) on that thread is particularly precient regarding the stall training issue. Since then, Colgan, Perpignan, Schipol, 447 etc. and finally in 2011 maybe something is getting done about stall training.

Some other bits from the report:


Following a December 20, 1995, fatal accident involving an American Airlines
(AAL) B-757 near Buga, Colombia, the Safety Board recommended that the FAA:
A-96-94
Require that all transport-category aircraft present pilots with angle of attack30
information in a visual format...

the Safety Board concludes that this
accident might have been prevented if the flightcrew had been provided a clear, direct indication
of the airplane’s angle of attack. Therefore, the Safety Board reiterates Safety Recommendation
A-96-94.
So that was 1996, and 1997, and now in 2011:


The crew never formally identified the stall situation. Information on angle of attack is not
directly accessible to pilots. ... Only a direct readout of the angle of attack could enable crews to rapidly identify the aerodynamic situation of the airplane and take the actions that may be required.
Consequently, the BEA recommends:

that EASA and the FAA evaluate the relevance of requiring the presence of an
angle of attack indicator directly accessible to pilots on board airplanes.
Is the industry going to lear from history or repeat it...

[infrequentflyer789]

You know very well that i´m fully aware about the functioning and the necessity of autotrim and do not question it. I even find it a clever and well thought out system. But the situation developing in ALT2 with AF447 was not being expected somehow and needs to be addressed and changed. It works in Alt1 at Vprot, why not do the same in Alt2? Nobody seems to be concerned to hinder autotrim in Direct Law, but you explain it would be difficult for the crew if autotrim would be hindered in Alt2 when predesigned values (aoa, speed, Trim value, take whatever would suit yourself) are reached?


Dropping autotrim in direct law already possibly killed several at perpignan, and similar change (although different system) probably contributed at schipol. I wouldn't say there is no concern about it.

The big problem with what you suggest is that the plane is in Alt22 because it doesn't have trustworthy values for speed, aoa and possibly other airdata. So what should it stop trimming based on ?

The whole reason the protections drop out is because it has been held to be more dangerous to "protect" based on invalid (or not trusted) data, than to hand full unprotected control to the pilot. I suspect this is a certification issue. Boeing FBW is exactly the same (not sure about the newer bizjets).

That whole assumption may need to be challenged now - possibly we're reaching the point where safety will be improved by systems overriding / protecting pilots even based on known-bad airdata. Be careful what you wish for...

[infrequentflyer789]

They clearly did not hear the WARN, and I do believe that the WARN is inhibited in prot, but you would know that?


Wrong - in fact opposite.

Stall warning shouldn't happen if everything is working, but in fact can and does happen in normal law. If it does, at least in some (possibly all) cases it is a signal to the systems that something is wrong (maybe airdata), and normal law will then drop out.

So, SW more or less inhibits normal law, not the other way round.

[Machinbird]

When the wings started rocking-it was like they had both forgotten all they should know. I'm disappointed that no one commented on this comment! Goes back to the earlier link posted relative to the pull reflex.
When the wings rock, you have to reduce AOA if you want to be in charge of a flying machine. Otherwise, you are just a passenger.
I thought all real pilots knew that instinctively.:rolleyes:

[RetiredF4]

infrequentflyer789
The big problem with what you suggest is that the plane is in Alt22 because it doesn't have trustworthy values for speed, aoa and possibly other airdata. So what should it stop trimming based on ?

Exactly on that, not having thrustworthy values.
With thrustworthy values the protections would stop the trim reaching special values , without those values it will continue to trim also it can cause harm.

Where is the logic?

[Lyman]

Infrequentflyer789

Yes, SW inhibits Normal Law, then. Do you have a thought as to the relevance of degrade at the Initial SW just after handoff? Are you convinced the ADRdoubled out and that was the immediate cause of ALT2? The SW had no effect? Any chance that the SW chirped and was silenced as in Protection? Which? If any?

[infrequentflyer789]

Exactly on that, not having thrustworthy values.
With thrustworthy values the protections would stop the trim reaching special values , without those values it will continue to trim also it can cause harm.

Where is the logic?



Logic is that when computer is half-blind and/or confused, it will hand over the responsiblity for staying inside flight envelope to the usually superior analysis and decision making skills of the real pilots.

It cannot logically limit trim to "safe" values if it no longer knows what those values are.

Autotrim didn't down this a/c - it was stalled by elevator alone and elevators were held nose-up all the way down. There was no diagnosis of stall let alone recovery attempt.

In contrast, the cessation of autotrim when things start to go wrong is a least a contributory factor in some crashes, if not a killer.

[DozyWannabe]

@Franzl

Autotrim is a separate system from the protections and the autoflight, as such under manual control and without protections it does exactly what the pilot commands. It does exactly what it says on the tin and trims automatically in the same way that manual trim is applied on older aircraft - there have never been any hard limits on manual trim wheels, why should there be one on their modern automatic descendents?

In this case we have one incident where possibly as a result of panic or poor training the PF commanded the autotrim - *after* the aircraft had already stalled - into an inappropriate position. I can say with some confidence that had the PF realised what was happening and pushed forward for an equivalent time to the amount he had previously been pulling back, the trim would have righted and control would have been regained. One incident in however many tens or hundreds of thousands of flights that this type and it's sister ships have made safely and without incident every day.

As such, it's far more realistic to train pilots to understand what the autotrim does as part of their type conversion (over and above never making and holding large control inputs for more than a second or two) than it is to redesign a system that works as well, if not better that it's manual predecessors and end up sacrificing pilot authority in the process.

@infrequentflyer789 - While you are correct in asserting that one should not hear the Stall Warning in Normal Law, and as such hearing it is an indication that something has gone wrong - as I understand it, the warning and annunciator system is not directly connected to the flight control logic - so it is never switched off or suppressed by anything directly. The FCU is a network of loosely-coupled, real-time processes with a degree of redundancy built in - not a monolithic, closely-coupled design.

[rudderrudderrat]

Hi DozyWannabe,
the warning and annunciator system is not directly connected to the flight control logic - so it is never switched off or suppressed by anything directly.
FCOM: DSC Aircraft Systems, 27, 20, Alternate Law:
"In addition, audio stall warnings (crickets + “STALL” synthetic voice message) is activated at an appropriate margin from the stall condition."

This seems to suggest that the "Stall" warning is only activated when Normal Law is no longer active.

[Hamburt Spinkleman]

Stall warning is also active in normal law. The threshold for activation is at 23 degrees AoA, which is far beyond what should ever be experienced in normal law.

[DozyWannabe]

I always read "activate" as "will sound" in that sentence as opposed to "becomes available". Would be nice to get some confirmation.

[RetiredF4]

@Franzl

Autotrim is a separate system from the protections and the autoflight, as such under manual control and without protections it does exactly what the pilot commands. It does exactly what it says on the tin and trims automatically in the same way that manual trim is applied on older aircraft - there have never been any hard limits on manual trim wheels, why should there be one on their modern automatic descendents?

Autotrim is commanded by the FCPC´s, not by the pilot. The pilot is ordering a loadfctor and the FCPC commands the elevators (short term) and THS trim (long term) to achieve and to maintain this loadfactor.
That is quite a difference to your saying and to older conventional aircraft. The FCPC will trim the THS all the way nose up with the stick untouched, if in 1 g flight (like in a climb) the speed decays. We discussed that before more then once.

And it is an operating difference as well. Trimming by turning a wheel is a deliberate act to trim, whereas automatic trim by the FCPC is a programmed syytem behaviour. Wether that is a seperate system or an incorporated one is of no relevance at all. Therefore there can be measures incorporated to prevent the trim reaching values, which are insane in FL350. There are enough data available after a pitot failure (altitude, attitude, AOA, GPS data, you name some more) to compute a normal trim zone (with trimming by the FCPC) and one where the pilot should decide wether he really wants the trim that far up by turning the wheel.

@ infrequentflyer
In contrast, the cessation of autotrim when things start to go wrong is a least a contributory factor in some crashes, if not a killer.


I remeber only those, where the trim beforehand wound all the way NU and ceased in the full NU position. If you refer to those, then it would have been better beforehand that the trim stayed in normal trim region.

If ceasing of trimming action in ALT1 is not a bad thing, i cannot follow the argument, why it would be bad in ALT 2.

What trim values are reached in normal operation except takeoff and landing?
Is Full NU a player above FL100 in any kind of normal maneuvering? In which one?

[infrequentflyer789]

@infrequentflyer789 - While you are correct in asserting that one should not hear the Stall Warning in Normal Law, and as such hearing it is an indication that something has gone wrong - as I understand it, the warning and annunciator system is not directly connected to the flight control logic - so it is never switched off or suppressed by anything directly. The FCU is a network of loosely-coupled, real-time processes with a degree of redundancy built in - not a monolithic, closely-coupled design.

Yeah, I know. I simplified a bit. SW doesn't directly inhibit anything, but once stalled it is likely that airdata discrepancies will start to show up due to the stall, and trip the plane out of normal law. That's assuming the SW is valid.

It's probably possible to take a broken bus and fly to SW in normal law and then recover without leaving normal law. Not sure I'd want to be in the back when someone tried it though. The guys at Perpignan were in direct law within 10secs of SW - see p112 (english version) of final report.

@rudderrudderrat:
See the traces on page 112 of the Perpignan report - SW clearly in normal law.

[DozyWannabe]

@Franzl

You're comparing apples with oranges here, or to be more literal, comparing the conventional, or "way we used to do it" method with the method on the FBW Airbus series, which is intended to be used in a completely different way - unless something has gone so seriously wrong with the aircraft that a drop to Direct Law is necesssary.

The FCPCs may command the physical movements, but it is the pilot that commands the FCPCs in manual flight. As the speed decays under Normal Law, the elevator/trim effect is limited by the AoA protections. Under all load factor laws (i.e everything above Direct) - even without protections, as the speed decays the control mode gradually transfers from "G" command to pitch command, meaning that the autotrim will not trim to stall with stick neutral due to decaying speed alone.

In this case the autotrim was commanded to move as it did to support the demands on the elevators made by the PF, who was - possibly for reasons we will never know - holding the thing between 50-100% back for the majority of the accident sequence. It was for this reason as well that the speed decayed, so in effect both the pitch and airspeed factors of the extreme angle of attack leading to and during the stall sequence were because of the consistent back-pressure on the stick, and for no other reason that I can see.

[infrequentflyer789]

I remeber only those, where the trim beforehand wound all the way NU and ceased in the full NU position. If you refer to those, then it would have been better beforehand that the trim stayed in normal trim region.


Yes, but it didn't, because (in the cases that come to mind) something else was broken.

It would have been better if the broken thing wasn't broken, but it was. Thereafter the loss of autotrim, coupled with lack of crew re-trim, caused deaths. Having got to that situation, it would have been better if autotrim had stayed engaged so that the trim did not impede the recovery.


If ceasing of trimming action in ALT1 is not a bad thing, i cannot follow the argument, why it would be bad in ALT 2.

What trim values are reached in normal operation except takeoff and landing?
Is Full NU a player above FL100 in any kind of normal maneuvering? In which one?

In Alt1 there is data available (known-good) that is not available in Alt2. Acting on known-good data is not a bad thing. Acting on known-bad data is.

As to what trim values are normally used, I don't know - but I doubt the designers put the mechanical end-stops where they are without reason. If no more than, say, 9deg nose up was ever needed, then the stop would have been at 9deg not 13.

[TTex600]

Airbus went to great lengths to design their FBW system to fly like a normal airplane. But in reality, the FBW Bus's do NOT FLY LIKE NORMAL AIRCRAFT. Auto trim proves that point. The Bus trims for "G", and normal aircraft trim for speed.

Airbus, EASA, FAA, etc, could fix all our problems with flying these computer games with one simple directive; change the training syllabus to state the opposite of what it states at present. Change "it fly's like any other airplane", to " it fly's like NO other airplane"........and train accordingly. Every Airbus pilot should be exposed to degraded flight characteristics at every training event, in addition to all other required maneuvers.

[infrequentflyer789]

Under all laws - even without protections, as the speed decays the control mode gradually transfers from "G" command to pitch command, meaning that the autotrim will not trim to stall with stick neutral due to decaying speed alone.


Not sure your're right there. Perpignan traces show exactly what you say can't happen.

I think you'll find that as speed decays, more elevator / trim will be needed to maintain stable attitude, and autotrim will happily provide it - up to the stops.

[infrequentflyer789]

Infrequentflyer789

Yes, SW inhibits Normal Law, then. Do you have a thought as to the relevance of degrade at the Initial SW just after handoff? Are you convinced the ADRdoubled out and that was the immediate cause of ALT2? The SW had no effect? Any chance that the SW chirped and was silenced as in Protection? Which? If any?

Sequence I see in traces is:

UAS -> A/P drop & Alt law -> stick-back -> elevator up -> acceleration up -> AOA up -> SW

Causality is not in doubt.

[DozyWannabe]

@if789

Wasn't the pitot-static system effectively completely hosed in the Perpignan case though (the systems must have been very badly compromised)? I must confess I didn't have an opportunity to follow that one closely at the time.

[Machinbird]

Wasn't the pitot-static system effectively completely hosed in the Perpignan case though (the systems must have been very badly compromised)? I must confess I didn't have an opportunity to follow that one closely at the time.To summarize Perpignan:
Perpignan accident was enabled when two AOA sensors froze at a low angle of attack while at altitude due to water intrusion. Crew later attempted to demonstrate activation of Alpha protect as part of a functional check flight of the aircraft and went well past into a full stall, autotrimming nearly full nose up in the process. Aircraft dropped into Direct Law (no autotrim) and crew was too busy to notice. High power setting plus nearly full nose up trim caused the nose to rise uncontrollably until stall. Did not pull out from the resulting nose down attitude.

[gums]

Have been on the sideline too long watching defense of a poor control law reversion sequence versus a defense of a poor crew.

Gums' view is there is a lotta blame to spread about.

Take the auto trim, and I do not like that term.....

- When normally flying the jet, HAL tries to reduce required stick displacement from neutral in order to achieve 1 gee that is corrected for pitch. So at 30 degrees pitch, the thing tries to achieve 0.87 gee with hands-off the stick. If I don't read that part of the manual correctly, lemme know.

- If the pilot is commanding more than one gee ( corrected for pitch), then HAL trims the THS to reduce the required stick displacement, or am I wrong on that as well? I see no AoA command when looking at the manuals, but I see some "protections" and warnings that hinge upon AoA.

- If I am at a 15 degree pitch attitude and let go of the stick, then what does HAL do to maintain 0.96 gee as my speed decreases? Does HAL trim the THS for more nose up?

- Then there's AoA sensors that are ignored because the speed is below 60 knots. So how is the stall warning considered valid when the AoA was being ignored? What the hell are you going to use if airspeed indications are unreliable? Maybe replace all the stuff on the LCD screen with a crosshair and dot, then tell the crew to "center the do"t, huh? BEAM ME UP!


Deeply stalled, deep stall, just plain stalled..............

In a plane with a super wing design and great directional stability, then it is easy to be stalled and not spin or have severe wing rock. Buffet could be very low and/or masked by the local weather conditions.

I do not defend the AF crew for the continuous nose up commands, but I can understand the confusion once the jet was "deeply" stalled and various warnings were being presented.

The facts now present clear proof that one can stall the 'bus without going into a spin or having severe wing rock or buffet. Anyone disagree?

The facts do not show that a recovery from the flight condition I just mentioned is impossible, especially with a fully trimmed THS.

Show us the pitch moment versus AoA/CG graph.


I only joined this discussion because I had FBW experience way before the 'bus and was interested in the plane. I also thot I could add perspective from lessons-learned 15 years before the A320 flew.

If I am off base here, or considered a dinosaur that still insists upon mechanical feedback to my stick or wheel or yoke, then tell me to stay quiet.

[Lyman]

gums. fwiw, I think you are in the money. As one of a rare few who continue to focus on the cause, rather than the effect, I for one appreciate the lead up to STALL that you discuss. For good or ill, and regardless of 'Blame', this a/c STALLED first, then it died. I like your basics approach, and 'ancient' pov.

if789. At what AoA did the first STALL WARN activate? To me, it seems a bit notable that at cruise, the STALL WARN activates. Since it occurs in such proximity to the a/p loss, my question would be what prompted the autoflight to approach Vs? Was it legit? Had the speeds gone south and the WARN was bogus? Was the vane compromised at this point? Thanks.

[Zorin_75]

The facts now present clear proof that one can stall the 'bus without going into a spin or having severe wing rock or buffet. Anyone disagree?
Affirmative on the stability part. As for buffet, there might have been some:
This modification of the behaviour in the load factor at the centre of gravity results in the
appearance of a high frequency component of an amplitude increasing to until about 0.1 g
peak-to-peak, and with a signature that is very different from a turbulence signature of
meteorological origin. Moreover, there is a noise on track 1 of the CVR, at about 2 h 10 min
55, which might be the impact of the microphone striking a wall, heard at a stable frequency.
Note: According to the simulation of the aircraft movements, at this time the turbulence observed in the
first seconds of climbing had stopped.
Additional analyses were conducted with Airbus to determine if this phenomenon could
correspond to buffeting. The difficulty with identifying this phenomenon lies in the fact that, on
the one hand, the concept of buffeting is defined as accelerations at the pilots’ seats and not
at the centre of gravity and that, on the other hand, no flight test has been conducted under
conditions that correspond exactly to those of the event (particularly in terms of Mach).

Note: Examination of flight test data revealed, based on the frequency and amplitude, that this
signature could in fact be that of buffeting. By drawing analogies with the flight tests, the amplitude of
0.1 g at the centre of gravity suggests that the amplitude of the buffeting at the pilot seat is high
(approximately 0.6 g peak to peak).

[RetiredF4]

DW

@Franzl

You're comparing apples with oranges here, or to be more literal, comparing the conventional, or "way we used to do it" method with the method on the FBW Airbus series, which is intended to be used in a completely different way - unless something has gone so seriously wrong with the aircraft that a drop to Direct Law is necesssary.


Read my posts, and read yours. I did not start comparing, that has been yourself.

Quote DW (bolding by me)
It does exactly what it says on the tin and trims automatically in the same way that manual trim is applied on older aircraft - there have never been any hard limits on manual trim wheels, why should there be one on their modern automatic descendents?

I might not know the design of the system, but i can get a grasp on how it functions looking from the pilots side. And from that point of view i find the necessity to stop trimming beyond a certain value, if protections are not available. As mentioned before, there are a lot of data available which could be used for such an safety feature.

Quote:
Originally Posted by DozyWannabe
Under all laws - even without protections, as the speed decays the control mode gradually transfers from "G" command to pitch command, meaning that the autotrim will not trim to stall with stick neutral due to decaying speed alone.

infrequentflyer
Not sure your're right there. Perpignan traces show exactly what you say can't happen.

I think you'll find that as speed decays, more elevator / trim will be needed to maintain stable attitude, and autotrim will happily provide it - up to the stops

That is exactly what happens, and i cannot understand, how anybody can feel good about it. We have been through this ordeal moons ago, but DW likes to forget about that.

[mm43]

Originally posted by gums ...

Then there's AoA sensors that are ignored because the speed is below 60 knots. So how is the stall warning considered valid when the AoA was being ignored?Surely these guys "got off the wheel" before it had a chance to turn full circle.

The Stall Warning is just that, and if it was to serve its designed for purpose, the PF/PNF would be expected to acknowledge it and take appropriate action. They didn't, and the wheel turned and got them in the a*se. So in that respect you can't blame the SW.

The PF not understanding that the pitch operates the same way in Normal Law and Alternate Law, but one "is protected" and the other is "not protected" is also a mystery. The PNF more worried about getting the CPT back to the FD than calling the ECAM and getting the QRH out is an even stranger mystery. This accident didn't happen in a vacuum - it happened to a flight that was under the control of two "apparently" competent pilots.

Granted there are issues that have validly been raised in respect of the SS visibility, but they are secondary to the human factor failures that we are currently aware of.
Pitch Characteristics - Normal and Alternate Laws

When acting on the stick the pilot commands a constant G load maneuver and the aircraft response is G load / Pitch rate. The pilot order is therefore consistent with the aircraft response "naturally" expected by the pilot, Pitch rate at low speed / Flight Path rate or G at high speedThe aircraft did as it was told - abeit the "wrong" thing, but it can't be "strung up" for that, with one proviso - there needs to be a means of preventing THS auto trim runaway once CAS/AoA/FPA have gone AWOL.

Lets face it, the pilots had control and the aircraft knew no better than to obey them.

[RetiredF4]

mm43
Lets face it, the pilots had control and the aircraft knew no better than to obey them.

To have control there has to be an understanding how things (trim, stall warning, LAW, aerodynamics) work.
Lets face the fact, that this knowledge probably was not present with the crew of AF447. And unfortunately the discussions in this and the former threads show, that it is not a single case problem.

[Clandestino]

That is exactly what happens, and i cannot understand, how anybody can feel good about it.

Lets face the fact, that this knowledge probably was not present with the crew of AF447There! You have effectively answered yourself; autotrim is good thing as long as one know how it works, applicable to any aeroplane system and one is required to know that by air law and the basic DNA-ingrained law of self preservation. With AF447 crew it was not just that they forgot about autotrim - they never verbalized what in the hell they thought was going on and what they thought they should do, indicating the rabbit-in-the-headlights defence posture. In the end, what CM2 did with the stick is what killed them. So far, investigators have published nothing that would indicate THS would not go nose down if only stick(s) were pushed forward.

I have to make very long&imprecise shot here: I suppose that in former life you were jealous of Bitburg based Ego drivers, with right hand gloves that didn't show wear at thumb.

Translation: F-15 has autotrim. How can anybody fell good about it? Very easily, I'd say.

Granted there are issues that have validly being raised in respect of the SS visiblitySo far, no official word has been provided on the matter. No certification authority concern. No accident investigation opinion. I do not consider mutual congratulations of anonymous posters, praising each other on understanding how Airbus is dangerous to be reliable.

[RetiredF4]

Hi Clandestino,

I have to make very long&imprecise shot here: I suppose that in former life you were jealous of Bitburg based Ego drivers, with right hand gloves that didn't show wear at thumb.

I have a chuckle on this one, like it actually. I still have some left hand gloves in good shape and missing the right ones with holes in the thumb.
I was flying in the backseat of an F15 double seater 1990 in a low level mission, and didn´t even recognize its autotrimming.

But to make my position clear: I stated several times and do it again here, i have nothing at all against a fine autotrim system. So its a bit unfair of you to put blame on me for not recognizing the advantages of FBW and its asociated systems.

There! You have effectively answered yourself; autotrim is good thing as long as one know how it works, applicable to any aeroplane system and one is required to know that by air law and the basic DNA-ingrained law of self preservation. With AF447 crew it was not just that they forgot about autotrim - they never verbalized what in the hell they thought was going on and what they thought they should do, indicating the rabbit-in-the-headlights defence posture. In the end, what CM2 did with the stick is what killed them. Investigation did not find a single reason why trim would go AND if only stick were pushed forward.

I agree with that, never did otherwise, and i knew somebody would jump on it. My concern is not about blame (neither to the pilots nor to the airframe), but about prevention. This trimsystem can be improved in the mentioned area without decreasing its overall performance, thus eliminating one hole in the different layers of cheese.

Why not talk about it and why not work on a change? Is it pride or neglicence, costs or arrogance which keeps us from recognizing, that there was a crew who was obviously not familiar with the behaviour of the aircraft despite their training, their hours of experience and their legal licences? Im pretty sure, there are others out there who learned a lot out of this accident and try to improve their knowledge base, but others will not and will never and might fall to the same situation.

Why not improve the autotrim system? There are accidents (not this one alone) where a correct functioning autotrim, not understood by the crew, contributed to the outcome of the happening. That should tell us not only that knowledge and training has to be improved, but that we also should improve the system to prevent that such misinterpretations proliferate into full accidents.

The statement, that they shouldn´t have got in this situation beforehand is correct, but it is not the way to look at the things in accident investigation and accident prevention and in view of flight safety. The task is not to close the hole in the first layer of cheese, but to take care of all known holes.

[HazelNuts39]

There are accidents (not this one alone) where a correct functioning autotrim, not understood by the crew, contributed to the outcome of the happening.Not expressing disagreement with your suggestion that this aspect merits revisiting, but could you please explain the implication that it contributed to 'this one'.

[RetiredF4]

I might try.

At 02:10:51 the stall warning sounded. AOA was 6°, pitch was 10°, elevators about 5° noseup, THS at -3° NU. At that point the stab trim starts taking over the load of the elevators and the THS trim starts to move uninterrupted to full NU position within the next minute. During this time period the SS was in the average NU, but never full NU (max NU was 13°, 5 times it was up to 8° ND. Note, that the elevators stayed at max 10° NU until 02:11:35: Then they moved further down and reached their 30° NU at 02:11:45 and about the same time the THS reached the full NU limit. (see the FDR in the BEA interim 3 report).

Would this THS trimming had been prevented upon the sounding of the stall warning, the elevators alone would have imho not been sufficient to keep the nose up in the stall over the following period. This THS is a mightiy thing and gave the crew the tool, to transfer their unfortunate SS commands into some kind of establishing the aircraft deep into the stall region.

The nose woud have dropped early in the stall, maybe the AOA would have been considerably lower and thus permanent stall warning available, the speed drop would have been less and all this could have finally helped in an recovery (which unfortunately was even not attempted).

I´m not saying, that this would have prevented the outcome at all, but it might in another situation on another day with a different crew.

[Lyman]

There is a bottom line in that conjecture, one I posted a very long time ago. It hasn't to do with competence of pilotage, but with common sense from the supposed brilliance of the design. "Just like a "Conventional" Aircraft", is buspeak for "Let's throw a bone." Arrogant.

Had the a/c STALLED earlier, without the authority of that monster slab, there would have been a great deal more energy retained, and the Nose would have fallen dramatically, imo.

Instead of MUSH, the pilots would have had at least one emphatic cue. Wait, let's make that two, with proper Buffet. Whether at that point the flight recovers is moot.

But it's worth an honest discussion.

[HazelNuts39]

RetiredF4;

Thanks for your elaborate explanation, with which I partially agree, but not quite. The airplane stalled a few seconds before 02:11:00, when the THS had hardly moved. The longitudinal control law imposes a certain pitch rate in response to the SS demands. If the THS had not moved, that pitch rate would have required a larger deflection of the elevator, but the pitch rate and hence AoA achieved would not have been significantly different, until the elevator reaches its stop. As to the THS being 'a mighty thing', that is true, but should not be exaggerated. According to Owain G's post, 10 deg of THS is equivalent to 15 deg of elevator. Therefore I agree that the effect of less THS as the airplane got deeper into the stall would have been somewhat lower AoA's, but that's about it.

[Clandestino]

Retired F4, I did not mean to infer that we shouldn't be discussing the Airbus flight controls in general or autotrim in particular; quite the opposite - it absolutely needs to be discussed to gain understanding. Given your experience and information you posses, your points are very valid and true. However, methinks that your perspective is somewhat lacking breadth. It is indeed true that having autotrim stop at stall warning would make AF447 behave differently - whether it would significantly affect the outcome is open to conjecture. Problem is we don't have infinite amount of cheese to patch up all holes. Patching up one, more often than not opens other holes. As DP Davies has beautifully demonstrated when analyzing pushers, every safety system is potential killer and risks an benefits have to be weighed carefully, especially cases of "operating when not needed" vs. "failing to operate when required". Before AF 447 there was nothing to suggest that autotrim concept as applied to FBW Airbi is flawed and I guess that even after final report is out it won't change.

The statement, that they shouldn´t have got in this situation beforehand is correct, but it is not the way to look at the things in accident investigation and accident prevention and in view of flight safety.It actually is but is not be-all and end-all. Obedient trim has good chances to be listed as one of fifity-something or more contributing factors but we shouldn't concentrate solely or it or ignore it completely. Balance has to be struck somewhere, not necessarily in the middle. As I see it, it was just a footnote in the history of certain crew that became (in)famous through spectacular error.

The biggest problem I see in discussions regarding the unwholesome fate of AF447 is that we are faced with horrible picture of three pilots who forgot how to fly in the midair so we are amusing ourselves with tiny technical details just to avoid to step back and face the sheer ugliness of the event.

Did I just blame the deceases pilots for the calamity? On purely technical grounds: no! I don't care a little bit about earthly theories of fault and blame but I do want that whatever daemon of the air got our unlucky colleagues gets caught and dissected and we won't be able to catch it if we put our traps in the places it doesn't visit.

It's not technical, it's in our heads.

[Machinbird]

Before AF 447 there was nothing to suggest that autotrim concept as applied to FBW Airbi is flawed and I guess that even after final report is out it won't change.Clandestino
I don't know much time you have in tactical jets, but you have enough chutzpah to have come to aviation that way. This is not a criticism by the way.

The performance of the autotrim system in this corner of the flight envelope likely does not meet certification requirements. Certifying this feature in Alt 2 at the stall was probably not envisioned in the initial requirements due to the high anticipated reliability of the airspeed system. It fell under the category of too improbable to worry about. Today we know that line pilots actually bumped into this corner of the envelope.

Despite Owain's assertions about relative power of the elevator versus HS (and I believe he is quite competent in this field) he was not working with the best data and his results may be slightly skewed.

Consider the Perpignan A320 and how throughly the THS overruled the elevator. Different aircraft, but designed by the same team with similar design philosophy.
Yes it is possible to stall the Bus in ALT2 using just elevator and thrust as we saw demonstrated by AF447, but who can say, had the trim had been limited, whether or not the crew would have seen the airspeed come off the peg a bit during one of their nose down trials and continued their nose down efforts.

THS running full nose up is a potential serious hazard and needs to be guarded against better. How to do that is one for the engineers (design type not wrench type) to puzzle out, but judging from what has recently transpired with the type, it may save two or more aircraft in the future. Expecting training to handle the problem completely is not realistic. If we see nothing in BEA's final accident report addressing this problem, then you should be concerned.

[RetiredF4]

When reviewing the FDR traces for my last contribution i stumbled on the the bottom lines on page 108 of BEA interim report N. 3.

At about 02:11:40 their is a spike indicating FO SS inoperative and a longer one shortly thereafter with Captain SS inoperative. It happens the same time, the THS and elevators reached Full Nose up. At the end the same indication Capt. SS inoperatie shows up again.

WHat would have caused this indication? Has it something to do with the priority button?

[Lonewolf_50]

Machinbird, in re training:

Consider the Perpignan A320 and how throughly the THS overruled the elevator. Different aircraft, but designed by the same team with similar design philosophy.

Yes it is possible to stall the Bus in ALT2 using just elevator and thrust as we saw demonstrated by AF447, but who can say, had the trim had been limited, whether or not the crew would have seen the airspeed come off the peg a bit during one of their nose down trials and continued their nose down efforts.

THS running full nose up is a potential serious hazard and needs to be guarded against better. How to do that is one for the engineers (design type not wrench type) to puzzle out, but judging from what has recently transpired with the type, it may save two or more aircraft in the future. Expecting training to handle the problem completely is not realistic. If we see nothing in BEA's final accident report addressing this problem, then you should be concerned.

The THS can be limited in movement by using the hand wheel in the cockpit. If it goes up too far it can be moved back down to where the pilots want it, since it appears that hands on the wheel overrides anything HAL tells it to do ... at least while the hand is on the wheel. ;)

Some threads ago, one of our Airbus 330 experienced pilots indicated that touching of the trim wheel in some sim training sessions was incentivized against. (Training note: This looks to fall into the realm of something called negative training, as in not correctly incentivizing a proper course of action, or incentiving against a given course of action).

Yet another Airbus 330 experienced poster here, Mikelour IIRC, described some unusual attitude training he encountered where using the trim wheel was one of the best ways to deal with it, though some crews had to be coaxed into using that resource, see above for possible reasons for that.

There seems to be some professional disagreement within the AB community, and the people who train the crews for various companies, on what is and isn't appropriate use of the trim wheels.

I don't think that changes the response to your concern:

if THS runs amok, you can manage it with your hand. What seems to add joy to this drill is the fact that when you aren't moving the wheel, depending on what HAL has in mind, HAL may move THS on his own cognizance once you reposition it. This point was also made by some of our Airbus veterans.

Good fun, and no sitting on your hands in an Airbus cockpit! :ok:

[Clandestino]

I don't know much time you have in tactical jets, but you have enough chutzpah to have come to aviation that way.Actually it is zero and I got into professional aviation through kind of integrated course. I'm an aviation buff that got very, very lucky to have both pictures taken by me and pictures of me posted on airliners.net. Don't concentrate on who is behind the nick but rather on what is written in the post.

The performance of the autotrim system in this corner of the flight envelope likely does not meet certification requirements.Given your previous post, I think you might be mistaking the description of flight test method with certification requirement.

THS running full nose up is a potential serious hazard and needs to be guarded against better.It did not run away, it was commanded to trim by FC computers.

How to do that is one for the engineers (design type not wrench type) to puzzle out, but judging from what has recently transpired with the type, it may save two or more aircraft in the future. Or kill two or three in the process if all implications of having such a system don't get well thought out and crew gets caught out by losing autotrim when they don't expect it.

During this time period the SS was in the average NU, but never full NUNeutral is 1G demand, full in clean is 2.5G demand. assuming linear sidestick response (I don't know whether it's precisely linear but during my time on bus it sure felt like it), half stick would be 1.75 G demand - impossible to meet below 1.38 Vs so with all air data rejected as unreliable and inertial reference available, both elevators and trim would try to meet it no matter what. That's why they ran to their limits.

The THS can be limited in movement by using the hand wheel in the cockpit.Except "don't do that", I can't find official reference what would happen if you try it. When we tried it out on the 320 sim, it simply ran to position demanded by FCCs when released.

There is another method of stopping the nose-up trim: push the stick forward.

Has it something to do with the priority button?Yes. FO SS inop means capt's priority button pressed and held an v.v. There's also latch out if priority button is held long enough but given the traces, it wasn't the case here.

[BOAC]

Quote:
THS running full nose up is a potential serious hazard and needs to be guarded against better.
It did not run away, it was commanded to trim by FC computers. .read what the guy (Machinbird) said - who said 'away'?? It is "running full nose up is a potential serious hazard" that is the problem.

[Clandestino]

By the time it got to stop, aeroplane was truly stalled and flight controls were trying to satisfy 2.5G demand which, as we all should know, is impossible to achieve below 1.58 Vs. Demand was given using RH sidestick. Do you dispute that or rather you believe that actual problem is dumb computer obeying supposedly intelligent human which got confused? Do you really want more intelligent computers recognizing when thinking in cockpit has stopped and taking over? IIRC one of the main complaints on PPRuNe was: Airbus computer doesn't let me to do what I want, sob, sob.

Well, on AF447 computers got confused by losing all three speed references so "decided" not to interfere with pilot's demands. That's what is meant by very technical expression "ALT2".

[Machinbird]

Given your previous post, I think you might be mistaking the description of flight test method with certification requirement.Clandestino, As a line pilot, you are not supposed to be taking your aircraft to any points in the envelope the test pilots have not already demonstrated. That is part of what certification is all about.

Automatic systems should not assist you in taking the aircraft outside the demonstrated envelope. If they do, they are improperly designed.

[DozyWannabe]

But autotrim under manual control is not "automation" in the classic sense of the term, it is entirely slaved to the demands of the pilot in control, which is why you have to be more careful with the protections out. Holding the stick back that far for that length of time is the antithesis of "careful".

I think we're going to hit something of a semantic argument here where those who are sympathetic to the Airbus FBW design will consider the fact that autotrim gives effectively complete trim control through the sidestick to redefine "primary flight controls", whereas those of a more traditional bent will insist there be a separation.

If you take the definition of "primary flight controls" to be anything you can do with the sidestick or rudder, then is is a simple matter of pushing forward on the stick to recover the trim position to normal while in the process of unstalling the aircraft.

[Clandestino]

As a line pilot, you are not supposed to be taking your aircraft to any points in the envelope the test pilots have not already demonstrated. That is part of what certification is all about.Exactly! Pulling further into the stall when faced with stall alarm makes a whole lot of the assumptions on which certification is based nil and void.

[Machinbird]

Holding the stick back that far for that length of time is the antithesis of "careful".Dozy,
Why would you be holding a control back for that length of time. It is because the aircraft was not performing the commanded action (It couldn't).
It may be a natural human instinct to re-select or hold a control when you do not get the expected action the first time, but as we see that this can be very dangerous. Things can be happening unseen (like the trim running nose up).

[OK465]

Automatic systems should not assist you in taking the aircraft outside the demonstrated envelope.

(my bold)

MACH: The elevator alone took the aircraft out of the demonstrated flight envelope. The THS behavior you find objectionable occurred as a result of being outside the envelope.

The flight test stalls performed in the aircraft were done at specific weights in specific configurations with specified CG's and at particular altitudes.

The three test data points for each stall were onset of the stall warning, onset of buffet and a termination point which was generally associated with a 'break' in the VSI occurring, a result of either 'pitchdown' or 'pitchup' depending on config. They are very structured and after that termination point you are 'out of the envelope'.

[RetiredF4]

Quote:
Has it something to do with the priority button?

Clandestino
Yes. FO SS inop means capt's priority button pressed and held an v.v. There's also latch out if priority button is held long enough but given the traces, it wasn't the case here.

Thank you for confirmation.

Another bit i found in the FDR traces and i like to point to:
Look at Page 111 of the BEA 3. interim report second trace from bottom. It is the normal acceleration graph.
02:10:28 - 02:10:50 below 1.0g, average g is 0.8, peak value is 0.6
02:10:50 - 02:11:10 above 1.0g, peak is 1.2
02:11:10 - 02:12:00 below 1.0g, average is about 0.9, peak is 0.7

In the timeframe from 02:10:28 up to 02:12:00, where the elevators and the trim moved to full NU (ordered by loadfactor demand from SS), only 20 seconds from the total of 92 seconds had been with a loadfactor of 1 g and more. All the other 72 seconds the actual loadfactor was below 1 g, giving the crew the feel of unloading (which actual was reality!) and doing the right thing. The SS NU inputs in this phase could be explained as controling the unloading.

By the way, you will be amazed, how much the body can feel an unload from 1 g to 0.8 g.

I´m fully aware that most of this loadfactor was created by the increasing vertical component of the flight. Did the crew know as well?

I´m open to other explanations.

franzl

[Machinbird]

OK465
The elevator alone took the aircraft out of the demonstrated flight envelope. 3rd BEA report
At 2 h 10 min 51, the stall warning triggered again. The thrust levers were positioned in the TO/GA detent and the PF maintained nose-up inputs. Not really disagreeing, but I believe the application of thrust at that point finished the job of taking the aircraft out of the envelope. They would likely have been able to do it with elevator alone as well, it would have just taken a little longer, particularly since PF was making inputs in both directions at that time. If they had not put in TOGA thrust, it may have kept struggling along until the Captain got back or the airspeeds returned.

Note: The THS had piled on at least one degree of trim by the time the aircraft left the envelope. If it had been stopped at stall warning, it would have been a bit harder to pull all the way into the stall.

The THS behavior you find objectionable occurred as a result of being outside the envelope.I guess I find it objectionable because it was unnecessary and had the effect of digging the hole the aircraft was in deeper. Not only do you have to rotate the aircraft nose down to recover, you have to run the trim back down to have a successful recovery and that would not be a quick process unless done manually. The fact that the aircraft was driven further into the stall by the THS again made things still worse.

The trim is essentially an AOA control if I understand the relationship between trim position and pitching moments properly. The more nose up trim you have, the more AOA you want the aircraft to fly at. The closer you trim to a stall AOA, the easier it is to pull into a stall.

How often does the Bus need to trim 12+ degrees nose up anyway? Aren't the flaps down when you do it and the cg is well forward as well? Would logical configuration related limits on trim position be appropriate?

[Lyman]

Notwithstanding the last bit of kick into STALL, the THS promised a long trim effort whilst Nose Down, subtracting prodigious amounts of altitude, adding breathtaking velocity at PULL OUT, with an aircraft that stubbornly limits G, (Probably a good thing, for that matter).

With that much NU, one expects less than docile 'stability' in the dive?

[HazelNuts39]

By the way, you will be amazed, how much the body can feel an unload from 1 g to 0.8 g. Did you take into account the variations of 'gee' due to buffet shown on page 43, together with the associated Note?
By drawing analogies with the flight tests, the amplitude of 0.1 g at the centre of gravity suggests that the amplitude of the buffeting at the pilot seat is high (approximately 0.6 g peak to peak).

[Lyman]

So they are actually identifying it as "buffeting"? Because if you read the addend closely, you will see how BEA works to make a "could be" an actual fact. So, I have been reading the BEA this way from the git; it is the creation of a separate reality before one's eyes.

"How did the crew disregard the STALLWARN, and the blatant BUFFET.....etc. etc. etc.........."

Numptification by proxy, and nary an harrumph save one. You fellas are too easy.

Grip/a/Get

[Clandestino]

The SS NU inputs in this phase could be explained as controling the unloading. If one take this phase out of context, this explanation is plausible. However, on pages 29-31 are some parameters traced more precisely than in appendix and they don't confirm that theory. First part, where load is slightly below 1G is where CM2 somewhat heeds CM1 advice to go down. Sidestick input is nose down, pitch goes from 12 to 6 ANU but aeroplane is still climbing and bleeding speed. Second phase, with G slightly above 1 is when stall warning goes of and CM2 reacts by pulling up again to 17.9° ANU - there are brief periods of pitch down command, quickly superseded by pulling again. G below 1 in third phase you mention is aeroplane stalling with oceanward acceleration accouting for Nz<1 till terminal velocity is attained. Stick is hovering around slight nose-up, to eventually move to full nose up.

IMHO, CM2 was not controlling the unloading by pulling, he was bent on pulling ever since he lost airspeeds and autopilot, for reasons not picked up by the CVR.

How often does the Bus need to trim 12+ degrees nose up anyway? Aren't the flaps down when you do it and the cg is well forward as well? Would logical configuration related limits on trim position be appropriate? Ouch. You don't trim for pitch, you trim for AoA. How many units of trim is needed for given AoA? Depends on speed, weight, config and CG position. A330 is long aeroplane with large speed range, so having powerful THS is necessary. I wouldn't suspect Airbus designers of being wasteful and putting too strong THS on their airframe. DP Davies has it all neatly explained.

[Machinbird]

Ouch. You don't trim for pitch, you trim for AoA. How many units of trim is needed for given AoA? Depends on speed, weight, config and CG position. A330 is long aeroplane with large speed range, so having powerful THS is necessary. I wouldn't suspect Airbus designers of being wasteful and putting too strong THS on their airframe. DP Davies has it all neatly explained.Clandestino, that wasn't a pitch attitude, it was a trim setting. Sorry if I wasn't clear.:O

I think my point is that the higher trim settings are rarely used unless the cg is way forward and the flaps are down. What is the highest trim setting you could encounter in a clean aircraft within the normal flight envelope?

[CONF iture]

As to the THS being 'a mighty thing', that is true, but should not be exaggerated. According to Owain G's post, 10 deg of THS is equivalent to 15 deg of elevator. Therefore I agree that the effect of less THS as the airplane got deeper into the stall would have been somewhat lower AoA's, but that's about it.
As we got little extra time in the sim today, we did experience a full stall from FL350.
Here is what I can report from the experience :

From the STALL warning we kept a light aft pressure on the sidestick
It was not long before we got a negative vertical speed of 15000ft/min
THS went to 12 deg UP under STALL warning
As we decided to exit the stall, full fwd pressure on the sidestick was applied
But we were unable to lower the nose
THS did not move
THS was then manually rolled fwd
Nose came down
Exit was then possible

I can't remember all the details, too much stuff to look at.
Thrust was kept at idle all the time.

Early fwd pressure on the sidestick at initial STALL warning should prevent a stall to develop.

[Capn Bloggs]

10 deg of THS is equivalent to 15 deg of elevator
I'm not an aerodynamics expert, but I can't believe that 10° of huge horizontal stab is equivalent to 15° of the relatively little elevator.

[Organfreak]

This seems to be breakthrough information in re the THS!
Stupid questions:
Why wasn't TOGA tried? Did somebody stop you?
Why has it been stated (someplace?) that the sims won't reproduce stall conditions?

[Machinbird]

Why wasn't TOGA tried? Did somebody stop you?
Why has it been stated (someplace?) that the sims won't reproduce stall conditions?

Organ guy, since I'm still awake, the answers are simple once you understand the principles:

TOGA is counterproductive in a stall. The low thrust line of the engines relative to the cg forces the nose even higher at high thrust levels. That is real bad for stall recovery.

The sims only accurately reproduce areas of the flight envelope that have had data collected by the manufacturer. Until AF447, no one had flown an A330 aircraft that far outside the allowed envelope. I expect the AF447 data is being used to make this terra incognita a bit more accurate, but it is not a complete data set as I understand.

[DozyWannabe]

@CONF

Was this an A320 or A330 sim session? Apologies, but I don't know which type you're actually on at present.

[Zorin_75]

Why has it been stated (someplace?) that the sims won't reproduce stall conditions?
You can determine the aerodynamic behaviour also through calculations and wind tunnel testing (which is BTW also all you've got up to the point of the first flight of a new aircraft, and they usually do pretty well out of the box nowadays). So the sim will of course do stall conditions (the other question being how much effort has at all been put into getting the model right for conditions that far out of the envelope), but to know how accurately your simulation reflects the real thing you need validation data from test flights and until now there hasn't been much data from free falling A330s to go on...

[RetiredF4]

Originally Posted by Retired F4
Tuote:he SS NU inputs in this phase could be explained as controling the unloading.

Clandestino
If one take this phase out of context, this explanation is plausible. However, on pages 29-31 are some parameters traced more precisely than in appendix and they don't confirm that theory.


I looked at those too, and i came to a differetn result and i´m not giving up yet. Those pages miss one vital information, the outcome of the actions in relation to airframe loadfactor. As the SS is a "loadfactor commanding device" (non technical expression) and the aircrew is feeling the loadfactor result, this value is important in judging the SS inputs the PF made.


Maybe someone capable poster could superimpose the graphs from P30-31 with the loadfactor and maybe the crew communication.

First part, where load is slightly below 1G is where CM2 somewhat heeds CM1 advice to go down. Sidestick input is nose down, pitch goes from 12 to 6 ANU but aeroplane is still climbing and bleeding speed.

agreed

Second phase, with G slightly above 1 is when stall warning goes of and CM2 reacts by pulling up again to 17.9° ANU - there are brief periods of pitch down command, quickly superseded by pulling again.

There we need to look at the TOGA power input as well, because that would have had a great deal in increasing pitch and increasing g-load. And as it looks like, an unexpected one.

G below 1 in third phase you mention is aeroplane stalling with oceanward acceleration accouting for Nz<1 till terminal velocity is attained. Stick is hovering around slight nose-up, to eventually move to full nose up.

Agreed, but again, the g load was less than 1 g, what would we expect a pilot doing, when the stall is not yet recognized and PF felt unloading ( and maybe even acceleration due to SI) for some time already? It was also the phase, where the altitude came down to the assigned FL350 again. The nose up limit stop could be the input to level off in FL 350.

At that point go back to Page 111, at 02:11:45 the THS and elevators and SS input all reach full nose up, and the pitch drops from 12° nose up to 12° nose down within 10 seconds and stays below the horizon until 02:12:15. I interpret that phase, that the THS was stalled and could not keep the nose up anymore. Look further to the loadfactor. The loadfactor increases despite the fact, that the nose drops (or because of it?) In that phase the THS got effective agin.

Must be a funny feeling, positive pitch unloading, negative pitch loading?

IMHO, CM2 was not controlling the unloading by pulling, he was bent on pulling ever since he lost airspeeds and autopilot, for reasons not picked up by the CVR.

My analysis is different.
After the initial unfortunate pullup and the following unloading hey did not recognize the stalled state, as it was against any training. They sure did not expect to be stalled with a loadfactor below 1 g. Their concern was the roll and not to unload too much, leading to the SS inputs. The application of TOGA - after the stall warning sounded- complicated their situation and led to an increase of pitch, loadfactor and altitude. Correction followed, again the loadfactor was being kept below 1 g to get the nose slightly down and recover the altitude FL350. But vertical descent rate picked up (unseen) and the pitch remained high. When approaching FL350 the level off attempt with full NU SS and THS and elevators also full NU the THS stallled and the nose dropped violently. The THS unstalled due to the pitchchange and grabbed air again, load factor got positive.

At that point i dont want to go any further at the moment.

I think it is not fair to say, they pulled all the way from the beginning. There where mistakes, big mistakes like the initial pull and like not recognizing the stalled situation, but the handling of the SS had different motivations than sensless pulling.

[HazelNuts39]

At that point go back to Page 111, at 02:11:45 the THS and elevators and SS input all reach full nose up, and the pitch drops from 12° nose up to 12° nose down within 10 seconds and stays below the horizon until 02:12:15. Could that be the result of the thrust levers moving from TOGA to IDLE?

[CONF iture]

It is a 330 simulator.
The main goals were to observe the autotrim behaviour and the response to the sidestick displacement as well as the THS movement.
Next opportunity, thrust could be applied and should applied and retarded just for further observation ...

Note : At no time the USE MAN PITCH TRIM PFD MSG was displayed.
What I figure, and that's only my own interpretation, the trim stopped at 12 deg NU when the AoA reached 30 deg and ABNORMAL ATTITUDE LAW took over.

[DozyWannabe]

@CONF

Wouldn't that law change have been notified in the ACARS message though? I'm pretty certain that some knowledgeable people in earlier threads were certain that ABNORMAL mode was never engaged.

[Clandestino]

Clandestino, that wasn't a pitch attitude, it was a trim setting. Sorry if I wasn't clear


My bad, apologies. Anyway I don't think that trim played significant part in the grand scheme of things and whether A330 THS is a) too powerful b) barely meeting certification requirements c) somewhere in between is for aerodynamicist to answer.

Those pages miss one vital information, the outcome of the actions in relation to airframe loadfactor.

There are acceleration graphs for all three axes in the appendix.

There we need to look at the TOGA power input as well, because that would have had a great deal in increasing pitch and increasing g-load. And as it looks like, an unexpected one.

Acceleration graphs don't confirm that, especially longitudinal accel, at the bottom of page 111, which comes as no surprise as at high altitude a) thrust is quite lower than low down b) there's not much difference between cruise and TOGA. N1 trace is on page 108. It hovers around 100% untill 2:10:45 when TLs are pulled back. Suddenly, stall warning fires and TOGA is selected at 2:10:52. There's dip in N1 with lowest being 80% at around 2:10:50. - just as stick is pulled.


After the initial unfortunate pullup and the following unloading hey did not recognize the stalled state, as it was against any training.

Or they were too distressed & distracted by "STALL STALL STALL STALL STALL STALL STALL STALL" to recognize they were about to stall. DozyWannabe made brilliant comment on forgetting how to recognize and deal with stall:"Like forgetting how to ride a bicycle".

Correction followed, again the loadfactor was being kept below 1 g to get the nose slightly down and recover the altitude FL350.

No. Sidestick traces are clear. Very short excursion into nose down were way too short to affect anything. Aeroplane stalled at her apogee and never recovered.

When approaching FL350 the level off attempt with full NU SS and THS and elevators also full NU the THS stallled and the nose dropped violently.

No. Have a look at TLA (thrust lever angles) and N1. Nose drop was due to power reduction. The aerodynamic stall of horizontal stabilizer, trimmable or otherwise, is way more violent than what is seen in pitch trace.

The THS unstalled due to the pitchchange and grabbed air again, load factor got positive.

No. TOGA was reselected.

I think it is not fair to say, they pulled all the way from the beginning.

Who said that? It was not they, it was him. He did not pull all the way, all the time but on average he pulled. He pulled when he shouldn't have. He died pulling. That's what RH sidestick trace shows. Is it non-PC to state in plain words what publicly available report has made clear through graph?


There where mistakes, big mistakes like the initial pull and like not recognizing the stalled situation, but the handling of the SS had different motivations than sensless pulling.

Maybe it indeed had. For the time being, I can't figure out what was the sense behind it.

[CONF iture]

I'm pretty certain that some knowledgeable people in earlier threads were certain that ABNORMAL mode was never engaged.
Report 3 itself mentions that the alternate law adopted was alternate 2B and it did not change again subsequently.
In the meantime, I observe that the Abnormal Attitude Law trace is not represented in the FDR data we've been given.

From my personal experience, I have to figure why the trim stopped around 12 deg UP and why it did not later on follow my full fwd request on the sidestick ... ?

[Zorin_75]

Stalled AF447 did not switch to abnormal attitude law (http://www.flightglobal.com/news/articles/stalled-af447-did-not-switch-to-abnormal-attitude-law-357394/)

[DozyWannabe]

From my personal experience, I have to figure why the trim stopped around 12 deg UP and why it did not later on follow my full fwd request on the sidestick ... ?

How long did you hold it and at roughly what deflection?

[RetiredF4]

All quotes Clandestino

Originally Posted by Retired F4
Those pages miss one vital information, the outcome of the actions in relation to airframe loadfactor.

Clandestino
There are acceleration graphs for all three axes in the appendix.

Exactly those i was refering, if you noted. But they are not incorporated in the pages 30 + 31, where they would directly show the outcome of the actions. Loadfactor finally is the vital indication what SS inputs and other factors accomplished to the aerodynamical behaviour of the aircraft.


Originally Posted by Retired F4
There we need to look at the TOGA power input as well, because that would have had a great deal in increasing pitch and increasing g-load. And as it looks like, an unexpected one.

Clandesstino
Acceleration graphs don't confirm that, especially longitudinal accel, at the bottom of page 111, which comes as no surprise as at high altitude a) thrust is quite lower than low down b) there's not much difference between cruise and TOGA. N1 trace is on page 108. It hovers around 100% untill 2:10:45 when TLs are pulled back. Suddenly, stall warning fires and TOGA is selected at 2:10:52. There's dip in N1 with lowest being 80% at around 2:10:50. - just as stick is pulled.

Think it over again. Some thrust vector is pointed downward and adding to lift factor instead of longitudonal acceleration. Also indicated by the VVi increasing again and altitude gain by further 500`feet. As we had a deceleration due to continuous loss of speed, you might see a slow down of longitudonal acceleration due to thrust, but no acceleration as you seem to expect (and PF did as well!). See BEA below

Quote BEA IR3 Page 91: (bolding by me)
02:10:54 The thrust levers are positioned in the CLB detent
02:10:56 The thrust levers are positioned on the TOGA detent.
The N1 increase progressively and reach 103% at 2 h 11 min 02.
 The copilot sidestick is positioned:
- between the half-travel position nose-down and ¾ of the stop
position nose-up with a nose-up position on average
- between 4/5 of the stop position to the left and 4/5 of the stop
position to the right.
 The pitch attitude fluctuates between 17.9° and 10.5° (Period
of 5 seconds).
 The THS varies from -3.8° to -8.3°.
 The roll angle fluctuates between 8.8° to the left and 4.9°
to the right (Period of 5 seconds).
 The angle of attack 1 increases from 7.4° to 18.3° while the
angles of attack 2 and 3 increase from 10.9° to 22.9°.
 The CAS decreases from 207 kt to 161 kt and the Mach
decreases from 0.66 to 0.51.
 The vertical speed changes from +2272 ft/min to
-3904 ft/min.
 The normal load factor decreases from 1.13 g to 0.75 g
(at 2 h 11 min 03) then goes up and stabilises at 0.85 g.

Quote:
Originally Posted by Retired F4
Correction followed, again the loadfactor was being kept below 1 g to get the nose slightly down and recover the altitude FL350.

Clandestino
No. Sidestick traces are clear. Very short excursion into nose down were way too short to affect anything. Aeroplane stalled at her apogee and never recovered.

Sidestick traces show a command to change loadfactor, and would be deflected whatever deemed necessary. And as the loadfactor was well below 1 g (although due to increasing sinkrate) and aircrew not aware that they stalled, there was obviously no reason to increase the unloading further. I give them that credit. We know now that they should have pushed further.... though.


Originally Posted by Retired F4
When approaching FL350 the level off attempt with full NU SS and THS and elevators also full NU the THS stallled and the nose dropped violently.

Clandestino
No. Have a look at TLA (thrust lever angles) and N1. Nose drop was due to power reduction. The aerodynamic stall of horizontal stabilizer, trimmable or otherwise, is way more violent than what is seen in pitch trace.


So you are saying, that adding power had no noticable effect, reduction had a big one? Generally you are right, i did´nt take the power change into my equation. Lastly it may have been a combination of both, as the engines with that pitch attitude produce a direct lift vector with part of the thrust.


Originally Posted by Retired F4
The THS unstalled due to the pitchchange and grabbed air again, load factor got positive.

Clandestino
No. TOGA was reselected.


I disagree. TOGA was selected 02:12:33. Recheck BEA IR3 page 113.

Between 02:11:45 and 02:12:30 THS and elevators had been full down, so SS input, whatever it was, had no aerodynamic effect as it didn´t change any flight control deflection. We can take those out of the equation for any pitch /AOA or loadfactor changes.

Power was idle until 02:12:10, when CLB was selected
Power was CLB until 02:12:33, when TOGA was selected.
Power change to CLB could only have an influence after 02:12:15, considering some conservative spoolup time from idle to CLB.

The G-load change however started already at 02:11:52 from 0.7 g to 1 g at 02:12:00 to 1.1 g at 02:12:10. The pitchdown had resulted in an decrease of AOA, thus wing and THS and elevators got more effective again.

Originally Posted by Retired F4
I think it is not fair to say, they pulled all the way from the beginning.

Clandestino
That's what RH sidestick trace shows. Is it non-PC to state in plain words what publicly available report has made clear through graph?

The SS graph does not indicate, what the reason for the overall pulling was. For that we have to look deeper and take the reaction of the airframe into account. Those are the position of the flightcontrols (as those are only indirectly positioned by the SS) and the loadfactor, as that one shows the final reaction to the inputs and is the only feedback (as there is no direct one) the crew had available (by instruments and by feel).


Originally Posted by Retired F4
There where mistakes, big mistakes like the initial pull and like not recognizing the stalled situation, but the handling of the SS had different motivations than sensless pulling.

Clandestino
Maybe it indeed had. For the time being, I can't figure out what was the sense behind it.

We have to open up our mind to grasp the unpossible and the unthinkable and have to put aside for a moment the obvious. Then we are prepared to dig deeper, to put ourselves in the Cockpit of AF447, in the LH or RH seat or in between / behind like the captain. As long as we make up our mind based on own trained behaviour, based on public oppinion and based on some own agenda (not saying that you do), we will not be open enough.

We also have to get rid of this permanent A vs B bashing or old vs new comparing. We have to think about every aspect of possibilities regardless who invented and designed it. It is hindering and distracting and it is without weight. Anytime i get dragged into that A vs B scheme by an answer to one of my posts, i feel uneasy with the response. Sometime i´ve got the feeling that by doing so some posters try to categorize the contributions and make them thus more or less trustworthy.

I for myself believe, that in the FDR´s is still a lot of truth hidden. The compilation of the vital stuff (Speed,Thrust, altitude, VS, AOA, G, SS, elevator, THS and cockpit communication into one graph and with better resolution will give us a better grasp on things. Unfortunately i´m too dumb to fiddle with those and make them myself.

[DozyWannabe]

@Franzl:

Allowing two F/Os, neither of whom had been trained in manual flying at altitude, to cross the ITCZ unsupervised with a known type-wide problem regarding UAS resulting in enforced manual control is in the realms of the unthinkable to start with - would you not agree? It's almost priming the system for an accident eventually!

Of course I'm talking with the benefit of 20/20 hindsight here, but even given that it has to be considered a spectacularly bad move. What are the "unthinkables" you are considering?

You make a good point regarding perception of acceleration, however if I've read the existing documents right then neither of these pilots had the fast jet experience you do that required this knowledge, and I'd venture to say that the only thing they'd be told to do regarding perception of acceleration and G would be to ignore it and concentrate on the instruments.

[RetiredF4]

Allowing two F/Os, neither of whom had been trained in manual flying at altitude, to cross the ITCZ unsupervised with a known type-wide problem regarding UAS resulting in enforced manual control is in the realms of the unthinkable to start with - would you not agree? It's almost priming the system for an accident eventually!

No, i would not agree. You want to pull my leg with such BS.

neither of these pilots had the fast jet experience you do that required this knowledge, and I'd venture to say that the only thing they'd be told to do regarding perception of acceleration and G would be to ignore it and concentrate on the instruments.

You need no fast jet expierience to feel the difference between 1.0 g and 0.8 g. My 75Kg would be 60Kg. And that reduction from natural 1g was not for a second like in turbulence (recurring short frequenced loading and unloading, feeling like bumps), but for much longer time .

In our jet we did the extension maneuver (unload and accelerate out of a dogfight) with 3.5 unit AOA equaling around 0.5 g. Most guys in training did stop the maneuver too early after felt 10 seconds, when only 5 had passed and at the beginning didn´t do the unload further then 0.7G

Concerning the "ignore it" passengers and CC´s would beat the hell out of you after .8 g´s for some seconds. Therefore it would be natural that they tried to keep it to the minimum in amount and time possible like they would do on any flight.

Your first comment shows some sarcastic talking, and the last one that you have very limited ideas concerning flying itself.

If that comment now offends you, so be it.

[DozyWannabe]

@Franzl:

I'm not being sarcastic at all!

I'm asking as simply as I can whether you think putting two people in the flight deck who are not trained to fly manually at altitude, when the type they are flying has a known issue that significantly increases the odds of them having to fly manually at altitude, could be considered a bad idea.

My personal opinion is that it's an incredibly bad idea with the benefit of 20/20 hindsight, and I was genuinely interested in yours.

As for fast jet experience, I wasn't saying that you'd need it to feel the difference, but that I'd expect that you'd need that experience to know instinctively what to do about it, and that the evidence so far suggests that neither of these guys, nor the Captain, had that experience.

Maybe they were instinctively trying to unload, but what little flying experience I have was underpinned with the proviso that your body can and will lie to you about your physical orientation and as such one should ignore it and trust the instruments in front of you.

I don't see how that could be misconstrued as anything other than an honest question and opinion. Just because someone has accused me of arrogance on another thread for having the gall to tell an actual pilot that the sidestick does not control the autopilot doesn't mean it's true!

[grity]

@RetiredF4,
you can think the rest of the day over the stall time after 2:11:12... but

again the mistake was one minute earlyer after the first tuch of the SS 2:10:07, a few sec after the end of the AP, mayby in a shock the first one ore two sec half pull.... the start of the zoom climb....

but not this pull was the problem, no, the problem was that the PF (and/ore the PNF) did not trust his correct instruments, he was not in the possition to decide if the instruments show him right ore wrong datas..... so he did not realise his climb, the falling speed etc

and for this decision if a instrument is realible ore not, it plays not a role if the values are analog or digital, no, you need a history ore the pathway of the values, and it can not be that it is impossible to show this devolution of datas (of the speed...of the altitude...) with pixels on a monitor in a manner that a pilot is in the position to decide correct. so in this case the PF was not in the position to decide correct.

the stall and the time behind was all later....

[RetiredF4]

@Franzl:

I'm not being sarcastic at all!

I'm asking as simply as I can whether you think putting two people in the flight deck who are not trained to fly manually at altitude, when the type they are flying has a known issue that significantly increases the odds of them having to fly manually at altitude, could be considered a bad idea. My personal opinion is that it's an incredibly bad idea with the benefit of 20/20 hindsight, and I was genuinely interested in yours.

We discussed the training or the lack of it at length, not only in this thread but in several others. To ride that old horse in comment of a serious issue from myself is sensless. You know my answer to these obvious issues long time ago. Your question is not honestly, it is there to distract.

As for fast jet experience, I wasn't saying that you'd need it to feel the difference, but that I'd expect that you'd need that experience to know instinctively what to do about it, and that the evidence so far suggests that neither of these guys, nor the Captain, had that experience.

My fast jet expierience isn´t even necessary in the discussion, its again a distraction you use. It is obvious, that any pilot flying with passengers unstrapped in the back tries to avoid g loads less then 1g, and when they are unavoidable when initiating a descent, to do it as sensible as possible. Therefore to shove the stick forward and thus to increase this unloading when already 0.8 (20% less than normal) are felt is expecting too much (also in the hindsight it would have been correct).

Maybe they were instinctively trying to unload, but what little flying experience I have was underpinned with the proviso that your body can and will lie to you about your physical orientation and as such one should ignore it and trust the instruments in front of you.

You dont get it? Read the loadfactor trace in the timeframe i´m talking about (my previous posts), they where unloaded, although due to other reasons and therefore didn´t increase the unload any further. Therefore the stick position (my assumption).

If you really want to bring your personal expierience into this, then look up the area of loadfactor protection in ALT2, what inputs (speed, aoa, or whatever) are being used, at what values this protections kick in and how it changes with different speeds, and how this protection would work out.

That could be usefull.

Otherwise i think we waste time with each other and others start getting bored. Therefore i will quit answering to such posts.

[CONF iture]

I for myself believe, that in the FDR´s is still a lot of truth hidden. The compilation of the vital stuff (Speed,Thrust, altitude, VS, AOA, G, SS, elevator, THS and cockpit communication into one graph and with better resolution will give us a better grasp on things. Unfortunately i´m too dumb to fiddle with those and make them myself.
Maybe mm43 would help at this ... He is pretty good (http://www.pprune.org/6628331-post1765.html).

[sensor_validation]

...

From my personal experience, I have to figure why the trim stopped around 12 deg UP and why it did not later on follow my full fwd request on the sidestick ... ?

I have seen this mentioned before, and I could design an auto-trim function that had this undesired behaviour - which shows same symptoms as classical "integral wind-up". If the controller doesn't know the output is limited it continues to ask for higher values, and then has further to fall before output comes back in valid range. Surely a correct implementation would include 'integral limiting' or 'desaturation' which would mean mean it would start trimming down soon after fwd input.

[OK465]

As we decided to exit the stall, full fwd pressure on the sidestick was applied
But we were unable to lower the nose
THS did not move

If the dynamics were unchanged, why would one expect the aircraft to require re-trimming?

[Clandestino]

Some thrust vector is pointed downward and adding to lift factor instead of longitudonal acceleration.Since it is: sine (aeroplane pitch + thrustline pitch) x thrust / mass, it is minor. It is also impossible to detect on Nz graph. All IRS accelerometers are strapped to aeroplane and therefore airframe referenced, not ground referenced.

So you are saying, that adding power had no noticable effect, reduction had a big one?Yes! I'll qualify that in a second.

The G-load change however started already at 02:11:52 from 0.7 g to 1 g at 02:12:00 to 1.1 g at 02:12:10. The pitchdown had resulted in an decrease of AOA, thus wing and THS and elevators got more effective again. Yes, but that pitch down was concurrent with thrust reduction! Next two were concurrent with elevators merely moving away from full nose-up! This aeroplane wanted to fly! Combined effort of engines, THS and elevator were needed to keep her stalled - as her attempts to pitch down into flying envelope have attested.

The SS graph does not indicate, what the reason for the overall pulling was.Agreed. They don't show why, just what.

For that we have to look deeper and take the reaction of the airframe into account. Penetrating as deep as it was possible with my limited means, I could find no fault in BEA's statement that aeroplane performed as designed & certified. Technical path has been well explored and not many pieces of puzzle found there.

It is obvious, that any pilot flying with passengers unstrapped in the back tries to avoid g loads less then 1g.Eeerm... not quite. It's impossible to avoid loads below 1G when leveling off or going into descent. I try to be as smooth as my George and he's limited to 0.7G

look up the area of loadfactor protection in ALT2, what inputs (speed, aoa, or whatever) are being used, at what values this protections kick in and how it changes with different speeds, and how this protection would work out. It uses no air data whatsoever. It's what it says on the box: load factor protection and you just need vertical accelerometer for it to work. Airframe referenced vertical, that is.

We have to open up our mind to grasp the unpossible and the unthinkable and have to put aside for a moment the obvious.Moderation is keyword. No use in opening mind so wide that the brains fall out.

We have to think about every aspect of possibilities regardless who invented and designed it....and discard impossible, implausible, improbable and just plane goofy ones.

What is so unbelievable about the picture of pilot pulling and pulling until the earth catches up with his stalled aeroplane and smites him? We have been losing aeroplanes to it for last century or so - trainers, transports, combat ones. Wolfgang Langewiesche has described the phenomenon accurately back in 1944. Heck, even your long time ago predecessor, Adolf Galland, managed to write off a training glider in such a manner. He survived spinning in because his glider had low wing loading and he kept the stick planted fully back until the impact, so he hit in fully blown spin, rather than in post spin recovery dive. Lady luck also lent her hand. Lucky for him, not so for those who would later stay for a split second too long in the reticle of his Me109.

First, I'm interested in why does it happen at all. When we solve that, then it's the question how did it manage to rise its ugly head in AF447's cockpit.

[bubbers44]

I know this is taking us away from the current rhetoric but wouldn't two competent pilots encountering ordinary weather conditions at night with an autopilot disconnect because of IAS loss not be able to keep the ac straight and level for a few minutes? The captain, who seems the only competent instrument pilot, was required by regulations to take his rest. Normally flying the long hauls from south america it is divided evenly so all pilots get equal rest. We had the extra pilot figure out the times usually depending on who was flying the trip on when each took their breaks. Our airline had competent FO's in the other two seats so didn't worry about the difficulty of that leg.

I guess airlines hiring low time pilots out of pilot mills can't do that now.

[infrequentflyer789]

Report 3 itself mentions that the alternate law adopted was alternate 2B and it did not change again subsequently.
In the meantime, I observe that the Abnormal Attitude Law trace is not represented in the FDR data we've been given.


Yes, and I don't see why it should not be there, even if it is a flat line. Perpignan report has it shown (well, the idividual pitch/roll law transitions are). That report also has better quality images of the traces.


From my personal experience, I have to figure why the trim stopped around 12 deg UP and why it did not later on follow my full fwd request on the sidestick ... ?

My bet would be on abnormal attitude law - entered based on AOA threshold. However, it would depend on how you got to the stall - presumably something was failed to drop you out of normal in the first place, can you tell us what was failed ?


Abnormal attitude law apparently doesn't give the "use man pitch trim" warning, but does inhibit autotrim, (and yes, that makes no **** sense to me either) which would fit with what you saw.

447 didn't trigger abnormal attitude because the ADRs, providing the AOA values, were already rejected by the FCPs due to the previous failures.

Perpignan did show the "use man pitch trim" warning (whether they saw it or not, who knows) because they failed all the way to direct law first. After that they then ended up back in abnormal attitude (which also doesn't make a lot of sense to me - pretty sure direct law is already full authority bar g-load protections, so why if already in direct would you need to switch laws to enable a recovery:confused:).

I guess after schipol, bournemouth, perpignan, etc., trimming needs to be part of stall / attitude recovery, whatever type you are on. Autotrim-stops-trimming (after trimming up into stall) is not a type specific problem or mfr or fbw specific (I don't think the 737 gives a warning on it either ?). I don't think it's going to be an easy one to solve (by engineering) either.


Also Conf, out of interest, presumably you flew alt-law in cruise in this sim - how did you find the pitch-normal / roll-direct combination ? As a non-pilot looking at the engineering of it, the control laws look mostly like sensible degradation, but that combination stands out as ugly - fly a different way in each axis! Is it really that bad hands-on, or is it ok ? [Clearly PF on 447 didn't find it ok....]

[RetiredF4]

Quote:
Originally Posted by Retired F4
So you are saying, that adding power had no noticable effect, reduction had a big one?

Clandestino
Yes! I'll qualify that in a second.

Im waiting.

Quote:
Originally Posted by Retired F4
The G-load change however started already at 02:11:52 from 0.7 g to 1 g at 02:12:00 to 1.1 g at 02:12:10. The pitchdown had resulted in an decrease of AOA, thus wing and THS and elevators got more effective again.

Clandestino
Yes, but that pitch down was concurrent with thrust reduction! Next two were concurrent with elevators merely moving away from full nose-up! This aeroplane wanted to fly! Combined effort of engines, THS and elevator were needed to keep her stalled - as her attempts to pitch down into flying envelope have attested.

Just read my post again. I was referring to an increase in g load beginning at 02:11:52 until 02:12:30 from 0.7G to 1.1 G, when power was in idle and elevators had been and stayed full down. No power change there, that started later. And yes, it wanted to fly, did i tell anything different? Did i tell it wanted to stall?

Quote:
Originally Posted by Retired F4
The SS graph does not indicate, what the reason for the overall pulling was.

Agreed. They don't show why, just what.


I´m interested in the "why", and as anybody can have a reason to do something or to don´t do something, the answer might be in the interactions of PF input and AC reaction as felt and understood by the crew.

Quote:
Originally Posted by Retired F4
For that we have to look deeper and take the reaction of the airframe into account.

Penetrating as deep as it was possible with my limited means, I could find no fault in BEA's statement that aeroplane performed as designed & certified. Technical path has been well explored and not many pieces of puzzle found there.

I didn´t tell anything about a faulty A/C, i try to get a grip on what happened, what the crew saw and felt, what the FDR schows in regard to SS input against elevator and THS reaction, and how the aircraft reacted to those. You see my motive in looking for something wrong with the aircraft, i can assure you this is not the case. I try to understand at the moment and i´m not content with your explanation that 3 pilots just f*ed up badly.

Quote:
Originally Posted by Retired F4
It is obvious, that any pilot flying with passengers unstrapped in the back tries to avoid g loads less then 1g.

Clandestino
Eeerm... not quite. It's impossible to avoid loads below 1G when leveling off or going into descent. I try to be as smooth as my George and he's limited to 0.7G

Now you are also selective quoting and answer to those selective quotes. Here is my complete sentence, in bold, what you left out.
It is obvious, that any pilot flying with passengers unstrapped in the back tries to avoid g loads less then 1g, and when they are unavoidable when initiating a descent, to do it as sensible as possible.

A bit unfair, isn´t it? Or was it because i didn´t mention the level off? Picky?

Quote:
Originally Posted by Retired F4
look up the area of loadfactor protection in ALT2, what inputs (speed, aoa, or whatever) are being used, at what values this protections kick in and how it changes with different speeds, and how this protection would work out.

Clandestino
It uses no air data whatsoever. It's what it says on the box: load factor protection and you just need vertical accelerometer for it to work. Airframe referenced vertical, that is.

On which box? Which values?

Quote:
Originally Posted by Retired F4
We have to open up our mind to grasp the unpossible and the unthinkable and have to put aside for a moment the obvious.

Clandestino
Moderation is keyword. No use in opening mind so wide that the brains fall out.


Getting personal now. Out of arguments? Or just tired? then dont answer.

Quote:
Originally Posted by Retired F4
We have to think about every aspect of possibilities regardless who invented and designed it.


Clandestino
...and discard impossible, implausible, improbable and just plane goofy ones.
What is so unbelievable about the picture of pilot pulling and pulling until the earth catches up with his stalled aeroplane and smites him?


It might be normal for you that such things happen, and they may actually happen. I lost friends who rejoined into the mirror picture of a lake in Labrador, everything might happen. Case closed.

First, I'm interested in why does it happen at all. When we solve that, then it's the question how did it manage to rise its ugly head in AF447's cockpit.

You wont find that answer, because you found the culprit already. And when you are not willing to look how it happened and thereby find out what caused the pilots to act like they did (like i and some others try to do), then you wont find out why it happend and you can can close the case.

[CONF iture]

I try to be as smooth as my George and he's limited to 0.7G
Nice admission for someone who fought so hard to state otherwise (http://www.pprune.org/tech-log/460625-af-447-thread-no-6-a-20.html#post6659626) …

[CONF iture]

IF789,

We did switch 2 ADRs off in order to degrade to Alternate Law.

As always, on the Bus, the less you touch the stick the better you fly.
We had no problem with the roll, but did the practice with no simulated turbulence.
Overcontrol and following oscillations can be easy depending of turbulence level.

The sidestick has that particularity, even in Normal Law, to easily, and well unvolontary, induce some roll when full back or fwd stick is applied. It is noticeable when conducting a GPWS procedure.

[Machinbird]

Clandestino
Anyway I don't think that trim played significant part in the grand scheme of things and whether A330 THS is a) too powerful b) barely meeting certification requirements c) somewhere in between is for aerodynamicist to answer. The control power of the THS is not what we are interested in so much as the effect of trimming all the way up to the stall. Test pilots avoid doing this for good reasons.
It is difficult to get the aircraft out of a stall when it is trimmed right up to the edge of the stall or beyond, particularly with a THS type aircraft.

Do you really think that just because the aircraft functioned just the way the designers set it to do in some obscure corner of the envelope, it is perfectly OK?

[Old Carthusian]

Autotrim - yet another red herring in search of a barrel to join its other compatriots. The evidence from this accident points to human factors, nothing more/nothing less. It isn't about the technical aspects or how the machine functions but what caused two supposedly rational beings to function the way they did. It's internal not external which seems to be missed by so many posters.
I have mentioned this before - KNOW YOUR MACHINE or professionalism. This was drummed into me by my flight trainers. No excuses - these pilots did not know what to do and this caused the accident. It seems to me that for some commentators the unthinkable is that the flight crew could have reacted as they did. Unfortunately that is the reality and grasping at straws doesn't help. There was nothing wrong with the aircraft but everything wrong with the human reaction. Try to focus on this ladies and gentlemen and the discussion will be much more fruitful.

[ChrisJ800]

"nothing wrong with the aircraft" hmmm I seem to recall there was a string of ACARS messages and errors including UAS, reversion to Alt 2 Law and other aircraft issues...

[Old Carthusian]

The aircraft performed as it was supposed to and the control system did the same. It responded to the UAS as it was designed - it was the crew that didn't respond appropriately. Remember, the aircraft stayed in level flight within the parameters of its designed flight envelope. It was the PF who put the aircraft into a stall.

[Old Carthusian]

The aircraft would have continued flying within the flight envelope - it was not in a dangerous state. You should go and reread PJ2's comments (on one of these myriad threads) on doing nothing. As a pilot one evaluates and then acts. One does not act and then wonder why things go wrong. One scans ones instruments, absorbs the information then in this situation pitch and power.

Now this is not about 'crucifying' the crew it is about understanding why the accident happened. The simple fact is that it was the crew who put the aircraft into the situation where it crashed - nothing else. The following points pertain; CRM, training, crew knowledge of SOPs, lack of knowledge of the machine, airline culture. All these combined to cause the accident in what was a survivable situation. This is the simple fact of the matter, if one assigns responsibility one has to look at the crew and the airline and its procedures.

[HazelNuts39]

What do you think the aircraft would have done, (and why) all by itself, second by second, from AP disconnect, if the crew had done absolutely nothing, and I mean absolutely nothing.Some time ago I wrote that the airplane's pitch and power at A/P disconnect was consistent with AP & A/THR maintaining altitude and speed while flying in an updraft of approx. 1000 fpm. It would have continued in those conditions as long as the updraft lasted. After leaving the updraft, it would have descended at 1000 fpm at the pitch and power and airspeed it had at A/P disconnect.

P.S. The above assumes that the FCPC maintains pitch after AP disconnect and leaving the updraft. If it maintains 1g, it would need to restore the still-air pitch of about 3 degrees nose-up, and the airplane would then decelerate at about 1 kt TAS per second at constant altitude.

[Flight Instructor]

I've just read a different thread on this accident and it threw up one big question for me. A lot of people seemed to kept mentioning deep stall (as this is also what the media called it) however was it deep stalled or just deeply stalled? Two very different things. Some people on the other thread seemed to have as little understanding as the many articles you can read about it.
Can an airbus even deep stall? Aren't ALL aeroplanes that can meant to be fitted with stick pushers? or is that something airbus could have got around?

Thanks :)

[sensor_validation]

@Flight Instructor
Stall (flight) - Wikipedia, the free encyclopedia (http://en.wikipedia.org/wiki/Stall_(flight)#Deep_stall)

[HazelNuts39]

was it deep stalled or just deeply stalled? Two very different things. For some people 'deep stall' is synonymous with 'locked-in stall' i.e. a condition that cannot be recovered by normal use of the flight controls. For others it just means deeply stalled.
"Can an airbus even deep stall?" Several 'experts' on this forum doubt it. The response of the airplane to slight relaxation of the pull on the SS seems to indicate that the airplane would pitch down with a determined nose-down input.
Aren't ALL aeroplanes that can meant to be fitted with stick pushers? or is that something airbus could have got around?The stalling characteristics requirements apply with all systems functioning as designed. In the case of the A330 that means Normal law, which does not allow the airplane to exceed the stall AoA, and can even effectively act as stickpusher as demonstrated in QF32. Alternate law is a failure condition that is judged on the basis of its probability of occurrence and the ability of the crew to handle it.

[chrisN]

RetiredF4 wrote (post 1633): “I´m interested in the "why", and as anybody can have a reason to do something or to don´t do something, the answer might be in the interactions of PF input and AC reaction as felt and understood by the crew.” (Re why did the PF pull back into and keeping the stall.)

Unless BEA release more CVR data or an analysis of that plus other data, I doubt if we will ever know for certain, but early on in one of these threads, there were at least two theories put forward by others:

1. Initial pull inadvertent, while “stirring mayonnaise” trying to control roll. PNF was telling PF to be more gentle with the SS, (but PF took little or no notice?).

2. Soon after, however, PF was evidently doing it on purpose. Possible reason – he thought they were overspeeding, and was trying to raise the nose to correct this. (If at the same time as TOGA, this seems to me inconsistent thinking on his part – but comments like “see the crazy speed”, later “I have been pulling up most of the time”, and reported high noise level which a witness is rumoured to have heard on the CVR playback suggested to some that he confused stall and downrush noise at high AoA with overspeed noise, probably never having heard either before. Didn’t he also try brakes, until PNF told him not to? Also, after PNF took over SS control briefly and stopped pull back, PF resumed (without the mantra “I have control”) and pulled back again, AIUI. That had to be deliberate, and not agreeing with PNF’s apparent belief that it was not overspeed.

(These were the tentative conclusions of ATPL’s – I am simply recalling them.)

These are the reason I mentioned that when a stressed pilot forms the wrong conclusion, he/she tends to stay with it regardless of ineffective attempts to correct the wrong problem, in my post 596 on the “final crew conversation” thread on R&N.

[gums]

As posted way back, at least one jet I flew had a true "deep stall" in which recovery using conventional control movements was not possible. I even posted the graph of pitch moments. The good news was unconventional control applications could "rock" the jet outta the deep stall.

I can not say if the AF447 jet was in a true "deep stall" without seeing the pitch moment charts for its c.g. and AoA. I would say the thing was "deeply" stalled and held there due to pilot inputs and the trimmed stabilizer that reduced nose down authority.

As another contributor posted here a day ago, a similar condition in the sim was overcome and he flew the plane out of the stall.

[Lyman]

I put great faith in Hazelnuts' analysis. The problem is in accepting it without further questions. The a/c's actual vertical speed was 1000fpm?

At the moment of a/p disconnect, does PF sit on hands? He has a Roll to the right to contend with also. Does he resist the urge to "Have the controls"? Because that means touching, and handling. Why touch and risk the displacement of the stick, if the book says 'don't touch'? Or, "Do not maneuver"? Sophie's choice? An after the fact pronouncement on an anonymous forum doesn't qualify as SOP.

Further, let us entertain that the a/c has not "settled" on 1000fpm, but was in some trend, and the PF felt, and understood the g forces to his satisfactory conclusion? Right or wrong, he is now 'seat of the pants', and that might be a bad thing. We know he is inconclusive about accel cues and audio feedback in the flight deck. In his own voice....

Was his flight path the result of a risky blend of poor memory/training/experience in hand flying with some 'cheek' input? TTex has posited armrest misplacement, and is it possible to believe that PF never got the chance to fasten his restraints after a conversation or other? The only clue we have of his re: stick work is his "I have been pulling for a while".

In the 5-10 seconds of loss of autopilot and manual acquisition, the die is cast. Following this 'potential' conclusion, we have examples of literally dozens of holes/cheese. My contention is that from this early point, the cheese was mostly hole, little cheese.

[CONF iture]

Some time ago I wrote that the airplane's pitch and power at A/P disconnect was consistent with AP & A/THR maintaining altitude and speed while flying in an updraft of approx. 1000 fpm.
I'm not sure where you see this ?

What I do see at AP disconnect time is a selected negative vertical speed of 5000ft/min and a vertical speed going in that direction (the quality of the given traces is somehow poor ...)
I also do see a pitch at zero (that's 3 degrees below the usual cruise pitch)

If what I do see is reliable (?) the initial action of the PF to pull is totally justified.

HN39, how would you justify the AP/FD vertical mode trace is not published ?

[Lyman]

At no time in two years plus has anyone questioned the NU input of PF at handoff. Only insofar as he should have left the a/c descending at 3 degrees. Have I missed something? This has been the position all along, the a/c needed handling.

The ND reflects A/P reaction to substantial UpDraft.

The politics of the initial BEA report reflect the wisdom of the PF's pull.
"The a/c did not initially respond..." It is their way of throwing PF and the a/c a bone...... The resultant climb, when it occurred, had the benefit of the airmass, and in AL2 should have trimmed NU. It did. Six Degrees? The 1.65 g was the result of these factors, not the pilot acting alone.

The jet was manouvering at a/p loss, and can be considered to be Upset, at this point. LOC came later. Twilight Zone, anyone? ffs.

An opinion can be put forth that this a/c was uncommanded for as long as 5 seconds. That qualifies as Upset.

How sinister is the creep of Urban Myth.

[HazelNuts39]

What I do see at AP disconnect time is a selected negative vertical speed of 5000ft/min and a vertical speed going in that direction (the quality of the given traces is somehow poor ...)
I also do see a pitch at zero (that's 3 degrees below the usual cruise pitch)If you are referring to the 'zipper' trace on page 111, I don't see any effect of that on the airplane's behavior. If you look at page 42 you'll see the AP maintaining altitude with very small variations of V/S due to the turbulence that is more explicit in the trace of normal acceleration. I explain the pitch of zero (that's 3 degrees below the cruise pitch in still air) by the autopilot pitching the airplane down to maintain altitude in a somewhat stronger upcurrent that commences at 2h10 (*). Pitching the airplane three degrees down in still air would change the FPA three degrees down, i.e. result in V/S=-1000 fpm. To prevent the airplane from accelerating on the downward flight path, the thrust must be reduced. The A/THR appears to do just that, although it doesn't quite reach the level required for that FPA (page 113).

If what I do see is reliable (?) the initial action of the PF to pull is totally justified.That may be so, I was only answering a question.

HN39, how would you justify the AP/FD vertical mode trace is not published ?I have no opinion on that, except that its omission in the interim report probably means that the trace does not provide significant information.

(*) That 'somewhat stronger' updraft may well be responsible also for the rolling motion and for the very particular type of ice particles (crystalline structure, particle size, density) that clogged the pitots of AF447.

[DozyWannabe]

If you are referring to the 'zipper' trace on page 111, I don't see any effect of that on the airplane's behavior.

Remember that could also just as easily be an "invalid value" rendered as the lowest possible value in range by the graphing software. I'm not arguing it's definitely one thing or the other, but given the aircraft's actual trajectory I'm inclined to think it's that. Given that it occurs prior to AP disconnect, I'm also inclined to think they'd have noticed and pointed out flickering values on the FMS display if they'd seen it as there wasn't much else going on in the flight deck at that time.

[Lyman]

Deep Stall? She came close a couple times to a smart recovery. Deep is a misnomer.

Clandestino: "The a/c wanted to fly." I agree, and there was airspeed available, just a too high AoA. The Stall entered unconventionally, and any recovery would have been the same. How many Stalls have been recovered in type? In these conditions?

[bubbers44]

Most pilots don't put it in a stall condition when they have a simple loss of airspeed and an autopilot disconnect. They have the experience to know what to do.

[grounded27]

I am surprised to see this thread go soo long. The simple answer is power and pitch, the varible being TC's.

[jcjeant]

Unless BEA release more CVR data or an analysis of that plus other data, I doubt if we will ever know for certainCan someone explain clearly why the BEA released only excerpts ?
Is there a law that prohibits the BEA to release the entire FDR ?
Is there a law that prohibits the BEA to release the entire CVR (transcript) ?
If there were leaks in the press and the publishing world (Otelli) BEA can only blame himself
If the BEA would provide comprehensive and clear informations (instead exerpts) .. journalists would not have to resort to some practices for gather informations (leaks)
Hide informations from the public .. how it's called ?

[Clandestino]

Retired F4, those remarks I made are not in any way personal and are absolutely meant to be general - i.e. applicable to anyone. I am sorry if you felt offended but here the offence is strictly in the eyes of beholder. Most of your questions have already been answered, too bad that besides having open mind, answers also require considerable foreknowledge of matters aeronautical to be understood. Somehow I don't think the answers given don't support the theory of "making sidestick input based on G feel".

Can we agree on that bending the facts to suit pet theories is not bound to be particularly productive if our aim is to truly understand what really happened?

The only thing I found in your post that has not been answered yet:

On which box? Which values?

Airbus FCOMs & AMMs, 2.5/-1G clean.

I have absolutely nowhere stated that I've found the reason for AF 447 catastrophe, claiming that I wrote so is severe misinterpretation. I've put forward theory I think currently fits with the facts as we know it. If anyone have a facts that are contrary with it, please post it, it's not about proving my or yours or anyone else's notion. It's about finding what is right, not who is right.

If this investigation gets wrapped up with "Pilot pulled for reasons undetermined" that would be a real failure. What made him pull must be brought to light.

It is difficult to get the aircraft out of a stall when it is trimmed right up to the edge of the stall or beyond, particularly with a THS type aircraft.

Do you really think that just because the aircraft functioned just the way the designers set it to do in some obscure corner of the envelope, it is perfectly OK? Who gives a part of a rat that the tail is attached to about what I think? This is the way our local universe, with its laws of aerodynamics, is structured. Yo want to go fast, high, far and heavy? It's a no-go without powerful THS. DP Davies book "Handling the big jets" has excellent treatise on why we need powerful stabs and how to handle them. Brief summary: In dealing with the consequences of having a variable incidence tailplane one basic fact must be kept in mind - it is very powerful.

(...)

The enormous power in variable incidence tailplane can be good servant when required but an impossible master when not required.

If you suddenly got an idea that the aforeqouted sentence somehow validates your notion that autotrim is lethal, a) perish the thought b) read the following sentence. Theorizing about particular features of any design without knowing and understanding the basic principles behind it is extremely unlikely to be meaningful. Not limited to THS. Not limited to aviation.

Nice admission for someone who fought so hard to state otherwise … My apologies, I keep forgetting that these are open fora so I have to keep discussion dumbed down to high school level for everyone to understand. You seem to understand that my statements "I do not fly by feel" and "I try to fly smoothly" contradict. Well, it is just not so. I do try to fly smoothly but not by the seat of the pants but by making small, measured inputs and crosschecking their effects against that big, bright attitude indicator in front of me. Leveling off without lifting the pax off their seats is best done by making small pitch change rate and that is done by looking primarily at the attitude indicator, crosschecking with altimeter and vario. Definitively not by using vestibular sense - it is last guard telling you you're doing something wrong but inflight it cries "wolf!" too often. Every instrument rated pilot must know the technique and use it, especially those who made it to widebodies, e.g. Airbus 330.

Hide informations from the public .. how it's called ? I would suggest reading Edgar Allan Poe's "The Purloined Letter" for an answer to that.

Not much is hidden, especially not the salient parts. The inability of some to recognize what is in plain view is quite ordinary but combined with their penchant for airing conspiracy theories gives spectacularly amusing results.

[jcjeant]

Hi,

Clandestino
Not much is hidden, especially not the salient parts. The inability of some to recognize what is in plain view is quite ordinary but combined with their penchant for airing conspiracy theories gives spectacularly amusing results. Salient parts are only parts of the truth .. not the entirely truth

Condorcet
The friends of truth are those who seek it, not those who boast of having found it

Paradoxally

Clandestino
I've put forward theory I think currently fits with the facts as we know it.
Are you sure that the facts you know are all the facts that know the BEA ?

[aerobat77]

"What made him pull must be brought to light."

i think it will never be. the pilot will not tell you why he pulled and any data recordings can only answer what he did, not why he did this.

[jcjeant]

Hi,

"What made him pull must be brought to light."

i think it will never be. the pilot will not tell you why he pulled and any data recordings can only answer what he did, not why he did this. The actions of a man (supposed normal mentally) are always the results of a thinkink
To make this thinkink that will lead to his action he must have facts to analyze
Recordings (voice and datas) can show what facts he had
If his actions are inconsistent with the facts available to it .. one might conclude that his thinking was not good, or as a last resort .. it does not have all his mental faculties (were altered) during the thinkink
So .. if we have all the datas who were available to this pilot .. we can conclude if is thinkink was good or bad
So .. yes we will know "why"
It's only three answers to "why"
Bad thinkink
Good thinkink relying on bad datas
Mentally disturbed

[jcjeant]

Hi,

Maybe it was already explained in many messages ( I don't remember anyway)

http://i.imgur.com/EEhhF.jpg

Why this difference of (displacement) movement of the elevators in reaction of pratically same movement (displacement) of the SS ?
Seem's that in the first 20 seconds .. the SS movements have not perceptible effects on the elevators ....

[DozyWannabe]

At a rough guess, I'd say as the airspeed goes down, the greater the potential deflection of the elevators to match the orders being issued from the sidestick...

[jcjeant]

Hi,

DW
At a rough guess, I'd say as the airspeed goes down, the greater the potential deflection of the elevators to match the orders being issued from the sidestick...

So the deflection shown by the elevators is not the direct result of the position of the SS but instead .. (a filtered order) a position commanded by computer orders in relation with the speed
That can be weird .. as we know that the problem was a unreliable speed due to pitot tube problem ....

[Lyman]

BEA have quantified that. They have not established a reason for it, however. I believe the THS was moving, though. "Initially the a/c did not respond"...

[Zorin_75]

(a filtered order) a position commanded by computer orders in relation with the speed That can be weird .. as we know that the problem was a unreliable speed due to pitot tube problem .... Uhm, no. There's a control loop feeding back the a/c response to the FCCs. No weirdness involved.

[DozyWannabe]

I suspect they're using different inputs to work out the aircraft trajectory/velocity than air data, otherwise there's no way that Alt 2 would work with ADR DISAGREE. In fact I suspect that the data they're using to work it out (probably inertial) formed the basis of the BUSS design.

[CONF iture]

I have no opinion on that, except that its omission in the interim report probably means that the trace does not provide significant information.
Looking at the selected vertical speed trace, one of the key thing I'd like to check is the AP/FD vertical mode trace - I personally do find it disturbing it is not published yet.

HN39, how would you justify the vertical acceleration trace in the Perpignan case has never been published ?

[A33Zab]

Hi,

Maybe it was already explained in many messages ( I don't remember anyway)

http://i.imgur.com/EEhhF.jpg

Why this difference of (displacement) movement of the elevators in reaction of pratically same movement (displacement) of the SS ?
Seem's that in the first 20 seconds .. the SS movements have not perceptible effects on the elevators ....



It was explained before.
During the first part the actual normal acceleration (Nz) was > 1G and during the second part the Nz was < 1G.

[CONF iture]

Old Carthusian,
Your mantra is KNOW YOUR MACHINE but in the meantime don’t want to hear about technical stuff … ?

The simple fact is that it was the crew who put the aircraft into the situation where it crashed - nothing else.
Equation is simple : If a type is repeatedly exposed to unreliable airspeed indication at cruise flight level, the manufacturer has to be aggressively proactive by publishing at least a safety bulletin addressed to all operators with a clear procedure to be followed if needed and to recommend all operators to train their crews for such eventuality.

When you have practiced once you are so much better equipped.

[bubbers44]

Every Boeing I flew had an unreliable AS chart for weight and altitude to hold a pitch setting and power setting. What these pilots did when they lost AS and autopilot was not anything remotely close to what the Boeing chart says. Everybody knows you can't pull up over 5 degrees at 35,000 ft in an airliner and still fly.

Unfortunately this is the new way of hiring pilots out of flight schools with no experience. We still have plenty of experienced pilots but they won't work for2,000 per month.

[bubbers44]

I know they were flying an Airbus 330 but the attitudes to maintain level flight are about the same. I went out of my way in my career to only fly Boeing AC. Had to fly MD80's for a brief period but that was just so I could marry my now wife of 21 years because of geographical problems flying the 727.

[Old Carthusian]

CONF iture

If you bothered to go back and read the previous threads (around 3 or 4 I believe) you would find a discussion of Airbus' procedure for UAS. It existed but wasn't used or even consulted in this case. The question is why? A stall warning was ignored for a very long time, again why?
This is not the fault of the manufacturer as much as you may wish it to be. This accident is a result of the actions or lack of them of the Flight Crew, nothing more and nothing less. The important thing to examine is what led the Flight Crew to act in the way they did and then to consider what can be done to avoid future occurences.
The suggestions that somehow you can attribute the accident to the autotrim or the lack of a yoke do not conincide with the transcript of the CVR or the FDR. They are missing the point which is psychological. This ironically is more difficult to understand than simple mechanics and aerodynamics. However, we may postulate what led the PF to take the course of action he did but we will never know for sure. The one thing is if this accident induces AF to improve its training and procedures the loss of life will not have been in vain.

[HazelNuts39]

HN39, how would you justify the vertical acceleration trace in the Perpignan case has never been published ?CONF iture; I don't recall having particularly missed it when reading the report. Anyway, this thread is about AF447.

[HazelNuts39]

The response of the airplane in the first minute is shown in high resolution on pages 41 and 42 of Interim #3. Something odd struck me in the 'simulation' traces on page 42. According to the associated text, "it was agreed that, initially, the simulation would be confined to the longitudinal axis, without introducing turbulence." If the simulation is without turbulence, what caused the variations of alpha prior to 02:10:05, without noticeable variations in pitch, V/S and Az, and why did the elevator move after 02:10:00? Why differs the THS position from that recorded?

[CONF iture]

It is good observation HN39.

Also why simulated elevator position shows signs of oscillation movement already before 02:10:00 ?
Why that time period is not represented in page 41 graph ?
Why only left elev is represented ?
Do left and right elevators really move together or graph on P108 could be better … ?


"Airbus conducted a simulation of the aircraft behaviour based on the theoretical model and on the actions of the PF (sidestick and thrust)."
Why not then represent PF's actions (sidestick and thrust) on those P41-42 graph ?

I don't recall having particularly missed it when reading the report. Anyway, this thread is about AF447.
Correct. But investigation led by the same investigation body and again some data of interest are not published.

In a few words :
Why does the Judge withold data from the proceeding ?

[DozyWannabe]

I think the simulator has a small, random amount of air resistance worked in - even when flying in "clear air", there are still small "bumps", just as in the real world. The (simulated) FBW systems therefore add very small corrections to compensate.

Obviously the simulation is not going to be 100% precise, but as long as the general trend follows suit the exercise is valid. You'll note that the real elevator and G traces diverge from the normal significantly more, and more frequently than those on the simulated traces - this appears to indicate the effect of turbulence.

CONF - the final report on this accident is still some time from completion, give them a chance. Having said that, given the number of ambulance-chasing vultures that would attempt to twist the data to suit their commercial ends, I'm not surprised that the release of data is limited.

[HazelNuts39]

I think the simulator has a small, random amount of air resistance worked inSeems like a very reasonable explanation, but why is it virtually absent until 02:09:50, then growing in amplitude until 02:10:05, when it similar to that encountered by AF447, but without effect in Az and elevator?

[DozyWannabe]

A reaction to autoflight disengagement perhaps? In all honesty I don't know, but the deviations around the normal prior to manual input still seem to be less than those encountered on the actual trace - it's just that because the AoA ended up at such an extreme value, the vertical element of the graph is compressed in it's early stages.

[dnrobson]

This goes to the heart of the question that is continually in my mind. With a descent rate of 10,000 fpm for over 2 minutes why did the crew not realise they were stalled ? Even during the last few seconds of the flight the CVR indicates anguished surprise that the aircraft was going to crash. Was this parameter not prominently displayed on the PFD unwinding rapidly downwards ? If not why not ?

[chrisN]

dnr, FWIW, I have recalled tentative conclusions drawn by others in my post no. 1646. Briefly, some ATPL’s thought PF confused the status with overspeed. There are several bits of evidence to support that, though I think it is only a theory, but seems to me the best one at present.

So once fixated upon that, he kept pulling back, presumably trying to pull out of a dive and failing to understand why it wasn’t working.

[DozyWannabe]

So, I was talking about some research this weekend. I got in touch with an old Aero Engineering pal of mine from Uni and he managed to wangle us some spare sim time at his facility in the wee hours inbetween training sessions. What we had was an A320 sim rather than an A330, which comes with some key differences - the most obvious of which is the lack of Alternate 2, the nearest equivalent being Alternate without speed stability, and a different underlying architecture past a certain point.

Due to time constraints we could only run each experiment once, preceded by some familiarisation time handling the sim manually in Normal Law, albeit at low level, following the FDs around basic turns and level changes.

The first experiment involved setting the conditions to night IMC with CBs in the vicinity, having set the autoflight to take us to 35,000ft and hold us there. We had a friend of his who is a TRE sitting in the LHS to provide guidance and monitor what we were doing. He then failed the ADCs, leading to autopilot disconnect and a drop to Alternate (without speed stability) and we tried to follow through and maintain a 15 degree pitch angle. Things we noted:


I'd suspected it would involve considerable effort to hold the sidestick there for a significant amount of time, but I was genuinely surprised at just how much.
The zoom climb occurred exactly the way we expected
The Alternate Law (no speed stability) on the A320 seems to have a hard trim limit of 3 degrees nose up
It was definitely possible to hold the aircraft in the stall with 3 degrees of nose-up trim and full back stick, but it required effort
The aircraft wanted to nose down and recover itself, and with about 10 degrees of nose-down maintained with the sidestick at the moment we passed about 30,000ft, we managed to effect a recovery with the speed coming back up to a point where we could level out safely at about 20-25,000ft judging by the standby altimeter.


The second experiment was the same as the first, but as my pal had noted, the A320 has a hard limit of 3 degrees NU trim available via autotrim in the secondary Alternate Law. We tried again, this time winding in full nose-up trim manually just prior to the point of stall. This time:


The aircraft seemed more willing to hold pitch with the trim at full-up, but to hold it at 15 degrees still required considerable effort
We had to add a touch of rudder (on the TRE's advice) to control the roll.
Despite full nose-up trim, we elected to start a recovery as we came down through about 35,000ft this time, just to see if it was possible using sidestick only
Following the same 10 degree nose-down sidestick demand as before, the trim rolled forward with the sidestick demand, returning to around neutral within about 5-8 seconds, and we came out of the stall as before.


Based on this, as far as the A320 is concerned at least, recovery is possible using autotrim via sidestick only even when the trim has been manually wound fully nose-up. Given more time we'd have liked to see what happened attempting recovery at lower altitudes, but the general take-away seems to be that with sufficient forward sidestick demand it is possible to recover from stall even with trim forced to where it's not supposed to be.

Of course, these were purely technical experiments. Not only was this a sim session with only pride at risk, but we all knew what was coming and had a pretty good idea of how to get out of it. This does not and cannot compare to a situation where you're trying to get out of it for real, especially with the added handicap of limited manual flying experience.

Whether the A330 behaves differently I don't know, but I've called in my favours for now and am eternally grateful to the people who made it possible. Someone else is going to have to take that on.

[xcitation]

Big thanks DWB, well balanced post.
Was it configured heavy, I recall 447 was a tad over loaded on take off?
Also did you find out what max rate of decent you had when stalled?

[BOAC]

I recall 447 was a tad over loaded on take off? -aha! A new rumour! Goodie.

[john_tullamarine]

Thread #7 starts here (http://www.pprune.org/tech-log/468394-af-447-thread-no-7-a.html#post6793958)