Quote from CONF_iture:
"During high AoA demo where full back stick has to be maintained the FCS is constantly working at maintaining alpha max, therefore the FCS will not use a thrust increase or additional aport of energy to increase the speed but to increase the pitch and the FPA."

Okay, then what increases first - the FPA, or the pitch?
If the pitch, how can that be done without initially exceeding alpha-max?

Hard to tell ... who came first, the hen or the egg ?
The FCS mission is to maintain alpha max whatever the amount of energy brought to the equation. The goal is alpha max and punctual under or overshooting of the alpha max value is part of the process too, the time for the FCS to absorb a variation of energy.

That's only partially true - you're looking at it from the assumption that Alpha Max has already been achieved, in which case the system will maintain it as speed continues to wash off, though eventually if airspeed is not sufficient to fly, ground contact is a certainty (however as the Habsheim and Hudson incidents showed, the aircraft will "fly" right up until impact). This is why, as OG correctly pointed out, Alpha Floor is an integral part of the combined systems protections - and why disabling it effectively transfers the responsibility for maintaining airspeed and thrust solely to the flight crew.

The part you've now skipped over twice is how the EFCS works to *achieve* Alpha Max if it's not there already, and the rate at which it does so seems to be contingent on the mode (pitch normal vs. high AoA protection) and how the latter mode brings the phugoid damping aspect into the equation.

It seems to be generally accepted that the TOGA thrust was applied too late, as was the SS back-stick command. In the case of the latter, HN39 brings up the interesting fact that the back-stick command was not only late, but somewhat hesitant as well.

To me, OG's discovery regarding the phugoid damping system and HN39's careful parsing of the DFDR data regarding application of back-stick both provide good technical explanations of why acquisition of Alpha Max was slower than it might have been with more favourable conditions (as you pointed out, Bechet achieved it much more quickly and HN39 reckons that was because he was still in pitch normal at the time). [EDIT : It also raises the slightly ironic possibility that if AF296 had pulled the SS back to attain Alpha Max earlier in the sequence (before crossing the threshold), the more rapid loss of airspeed would have necessitated thrust application sooner and might have put them in a better position.]

Add to that the low-energy state of the aircraft in general, and I'd say we now have more technical detail, but that ultimately it comes back to the original conclusion - the aircraft was too low, too slow and corrective action was applied too late.

Yo!

Direct hit, Doze.

It all comes back to a poorly executed, poorly planned/practiced "display".

Sheesh.

Even 200 feet and adding power to maintain decent speed should have "satisfied" the crowd, most of whom did not understand what the 'bus was trying to demonstrate.

The guy screwed up. All the flight control laws discussion is "noise".

I would have been fired had I done something like that unless I demonstrated the manuever with HHQ review. The Birds and Angels do hundreds of practices, and all of their maneuvers are reviewd/approved by HHQ. Look at the Angels' loop with the gear down. Then see the inverted passes with a wingie tucked in real nice on the other plane.

The flight control laws on the 'bus seem very realistic for a heavy. The engine performance is about like what I saw back in 1971 when moving to a "fan" ve5rus a pure turbojet motor. Big deal. Spool up time is a bit longer, so you must anticipate, and not wait until you realize that your power setting is too low for what you are trying to do - go around, maintaining proper descent on a precision approach, etc. Sheesh.

Heh - while I must admit that I was going for a collegial tone rather than aggression, your support and sentiment are very welcome all the same.

To be honest, and I've said this before, I've gained so much more understanding of the systems design as it relates to flight mechanics from this thread that the discussion was definitely worth having anyway in my book.

And to play a bit of Devil's Advocate, while "screwed up" is a reasonable - if strongly-worded - conclusion to draw if you're looking purely at the conduct of the flight, if I'd been looking at it at the time I would have been deeply troubled by the apparent lack of oversight from - and shoddy preparation by - the airline. Obviously he was one of their top pilots on the A320 programme, and that implies a certain amount of trust and leeway - but he wasn't the only AF Captain in that position (indeed Capt. Bechet, who headed the investigation, was of equal seniority on the same programme). To me it just beggars belief that management didn't seem to require a peer review of the intended flight plan, because I'd be very surprised if such a review wouldn't have considered the level of risk unacceptable. At the very least I'd expect such a review to have triggered discussions with Airbus as to how it compared to their own practices.

Good practice, no question. But that dreadful B-52 accident demonstrated that even military practice can lapse if the person involved has a track record of making things difficult.

Anyways, nowt more to add - cheers again!

Quote from gums:
The guy screwed up...

As a general observation, I think even the A320's strongest critics in this discussion have never disagreed that no flypast should be conducted with passengers, that 100 ft is too low at any speed, and that this one was ill-conceived, badly planned, and poorly executed - in a number of ways. The Karsenty video proves that beyond reasonable doubt each time we replay it.

...All the flight control laws discussion is "noise".

Given the general consensus that has been evident since the first couple of pages of this thread, some of us have taken the opportiunity to try and improve our understanding of flight at high AoAs in general, and the characteristics of the Airbus alpha-prot mode in particular - including aspects of its certification. Not only was this a unique accident in public transport, but all the gathered evidence provides unique opportunities for study.

Quote from Dozy Wannabe:
"... ultimately it comes back to the original conclusion - the aircraft was too low, too slow and corrective action was applied too late."

Assuming you are referring to the execution in relation to the plan, you are right on your first and last "conclusions", but wrong again on the second. The reason the crew decided to keep the thrust at idle - and then left it there until too late - was that the a/c arrived over the display area too fast - not too slow. :ugh:

Important observation. Agreed.

Of course - perhaps I should have qualified the statement as "*ended up* too slow". I've certainly never argued that the aircraft wasn't fast and high on the approach initially!

Agree with Chris and Doze.

He was too fast, and had to keep power low to finally reach max AoA or whatever some want to call it. Then was slow getting the power back in.

As with others here, I have learned a lot about the "laws" and details about the nuances of the "laws". IMHO, they ain't so straight forward as some might think, huh?

Well, from a technical standpoint they have to be complex by their very nature - they were designed to be as transparent and easily-learned as possible for the end-users (i.e. pilots) though. Of course, there's only so much simplification one can do!

That said, I think it's fair to state that at or below 100ft RA at an unfamiliar airfield is definitely not a suitable time to start probing around the limits of the systems, one's experience with those systems or both - regardless of whether or not you've got pax behind you!

Quote from Dozy Wannabe:
" - perhaps I should have qualified the statement as "*ended up* too slow". I've certainly never argued that the aircraft wasn't fast and high on the approach initially!"

Even that statement would require to be qualified with something like "for the current power setting". During 35 pages of reading and contributing enthusiastically, you have evidently not fully grasped the fundamentals of the basic scenario, nor those of the developing argument.

The game plan was to display the a/c in stabilised flight at an AoA of alpha-max, meaning that - using a fairly high thrust-setting - the airspeed would have remained roughly constant at Valpha-max in level flight and during a spectacular go-around.

The essence of the accusation repeatedly made by CONF_iture is that the a/c failed to achieve alpha-max after the PF called for it. That means that its IAS remained ABOVE Valpha-max until the treetops slowed it down. He argues, understandably, that the slight, residual surplus of speed above Valpha-max could have been traded more immediately for height.

Contributors who are expert in aerodynamic performance have explained that the refusal of the EFCS to respond immediately to the stick-command ** was due to the lack of thrust, and the decaying IAS. The latter may have been exacerbated by a slight loss of headwind component at about t -2: that crucial moment when the N1s were still low, although spooling-up.

Therefore, the unqualified statement "ended up too slow", is simplistic, to say the least. Strictly speaking, it is incorrect and misleading. I'm sure that's not what you intended?


** (that is, increase the AoA as soon as physically possible to the commanded value of alpha-max)

A fine and for me definitive response Chris Scott, thank you.

Because that's part of the alpha max demo and that's where our discussion with OG initially started.

Obviously you still didn't get what alpha max is ...

1. you simply cannot "wash off speed" and "maintain alpha max" at the same time
2. Airspeed is always sufficient to fly at alpha max

As a start, neither Habsheim or Hudson were at alpha max.

Alpha floor is an auto thrust function only and has nothing to do with maintaining speed for a flight crew.

I have not skipped anything, but the BEA obviously has.
The US and Spanish teams have underlined the lack of pitch authority given to the pilots - They made some research and proposed possible explanations.
The BEA did ... nothing : "normal functioning of the aircraft"

That part is a ALL chapter on its own as Asseline pretends TOGA was applied 4 seconds earlier than the BEA has concluded ... But we can keep that chapter for later as we have already enough to discuss on alpha max for now.

Actually it is more specific : the elevator movements show that the FCS had no intention to deliver anything closer than 2.5 deg short of alpha max.
And the BEA has not provided any reason for such restriction.
But I mainly agree with the way you write the conclusion :
"That means that its IAS remained ABOVE Valpha-max until the treetops slowed it down. He argues, understandably, that the slight, residual surplus of speed above Valpha-max could have been traded more immediately for height."

Then our "experts in aerodynamic performance" are in total contradiction with the BEA who conducted the simulated flight at the hands of Bechet.

We need something more substantial than an hypothetical 2 kt wind shear to justify a drastic difference in the FCS response between the simulated flight and the Habsheim one.

Confiture,

In my post #605 I noted the difference between Bechet's simulation and the accident flight. Bechet went to full side stick before entering alpha-prot, and Asseline after, 3 seconds later.

The difference between the two flight control modes is decribed in the Final Report, 1.16.1.2 Lois de pilotage de l'Airbus A 320:

If you see a contradiction with the BEA report please explain where you see it.

EDIT:: Phugoid damping is a feature of the alpha-protection mode, because an airplane is prone to develop phugoid motions when constrained to a commanded angle of attack. Flying an airplane to a commanded pitch attitude does not result in phugoid motion, and therefore pitch-command mode does not require phugoid damping.

HN39,
I do salute your effort to provide what the BEA and Airbus should have dealt with ... but as you quote part of the Report, absolutely nothing in 1.16.1.2 details or even mentions why the FCS response in its capacity to command alpha max should be different whether full back stick is applied before or after Alpha Prot is reached.
The paragraph in question had no intention to deal with such aspect but is purely an attempt to justify why what they call the de rotation law or the ground effect simulation law could not have negatively interfered during the event.

Confiture,

There is no discussion of alpha-protection phugoid damping in the BEA report, it doesn't go into that level of detail. Many more details have been discussed in later reports of BEA and other investigators.

I'm well aware of your views regarding the BEA, but where is the "total contradiction"?

Then why do you quote at length a paragraph that says nothing on the topic if you want to use it to justify your hypothesis ... ?

Before, at, and after thrust levers are advanced to TOGA position, both Bechet and Asseline suffer from the same "lack of thrust and decaying IAS"

• Our "experts in aerodynamic performance" explain that's the reason why the FCS did not comply
• The BEA demonstrates how the FCS fully complies

It is what I call total contradiction ...

The Bilbao and Hudson investigations were of a very different scope though. Bilbao highlighted an unforeseen 'gotcha' (for want of a better phrase) in the EFCS design when subjected to a tail gust - Airbus themselves admitted as much - and therefore a degree of research would have to focus on what precisely caused this unwanted behaviour. The Hudson investigation would have been interested in focusing on every detail of what was an exemplary bit of piloting, and therefore would want to go over the technical aspects with a fine-toothed comb. It's also worth pointing out that the behaviour of the phugoid-damping aspect of HAP mode was known in the industry by that time as a result of Bilbao, and would have been something they'd want to test anyway.

Overall, what distinguishes those incidents from Habsheim is that they occurred on what were routine ops, and in both cases the airmanship from the crews couldn't be seriously faulted.

With Habsheim, what the BEA were faced with was a system-wide failure. Even setting the airmanship question aside, there should have been red flags raised within the airline and by the crew well before the aircraft ever left the ground.

Ultimately, the main aim of the investigation was to ensure that something like that never happened again. I'm happy to concede that an argument could be made that the BEA missed an opportunity to look further into the phugoid-damping aspect in the course of that investigation, however at the same time when looking at the circumstances of the accident holistically, that would have been a small, tangential issue at most when compared to the obvious shortcomings of AF regarding the handling of the operation and the conduct of the flight itself.

I would also be inclined to argue that Capt. Asseline himself bears some responsibility for the way that part was covered - after all it was he who made the decision to disengage from the investigation as soon as the scope moved in a direction he didn't like. I think it's a distinct possibility that if he'd been able to swallow his pride and continue to work with Bechet and the other investigators, then more technical detail may have been uncovered. But by retreating behind his lawyers, openly briefing the press against the investigation with claims of a cover-up and an increasingly bizarre veil of excuses, he left the investigators with no option but to try and work things out on their own.

You're parsing things again. The BEA said that the EFCS behaviour was *normal*, it did not (as far as I know) say anything about complying with the demand.

Unfortunately the report in that sentence uses the term "Alpha Prot" (which as you have stated is a value rather than a mode) to refer to High AoA Protection mode (to give the mode its correct nomenclature in English Airbus documentation), which can cause confusion. If you'll excuse Google Translate again:

If you combine this with what HN39 was saying:

It can therefore be deduced that - based on the assertion that it was the phugoid-damping feature which slowed acquisition of 17.5deg AoA - the difference between pulling the SS back in HAP mode versus pitch normal is that there is no phugoid-damping in pitch normal, thus less potential impediment to the rate the system can acquire 17.5deg AoA.

PS - @Chris Scott - many thanks for the clarification.

Quote from Dozy Wannabe:
"[...] the difference between pulling the SS back in HAP mode versus pitch normal is that there is no phugoid-damping in pitch normal, thus less potential impediment to the rate the system can acquire 17.5deg AoA."

No. The EFCS would never permit the AoA of alpha-prot (in this case 14.5 deg) to be exceeded in Normal Law without introducing High-AoA Protection mode. Therefore, to imply that alpha-max might be achieved in a shorter time from Normal Law than from an AoA in High-AoA Protection mode is a non-sequitur.

Perhaps I should elaborate. While pulling the SS all the way back in Pitch Rate Normal will, as you say, engage HAP mode upon reaching 14.5 degrees, the rate at which pitch-up is commanded will - if I have this right* - be somewhat more rapid than if the aircraft has been put into HAP mode with partial back-stick for some time before full back-stick is commanded. Based on discussions offline, I believe the phugoid-damping aspect of HAP mode relies in part on airspeed deltas recorded at certain time frames prior to that point. Because, as HN39 pointed out:

It is therefore reasonable to infer that the EFCS only uses these deltas (and possibly only stores them) after HAP mode is triggered.

One possibility is because the change in pitch is more rapid from Pitch Rate Normal, there is less time for these deltas to be interpreted as phugoid motion by the EFCS (and because pitch changes due to phugoid motion tend to be more gradual). Certainly in the sim, full back-stick from straight and level seemed to induce a very rapid pitch change.

Another (which could complement the first) is that the decaying airspeed along with the more minor pitch change due to the hesitant back-pressure prior to 3 seconds before impact could also be interpreted as phugoid motion. The damping would then progressively introduce pitch-down in order to counteract this, increasing the time required for the full back-stick command to deliver Alpha Max.

In any case, the damping feature requires a period of time to come into effect, and the faster the aircraft is rotated, the less time that feature has to make any immediate changes to the elevator command.

[* - As stick deflection in Pitch Rate Normal commands rate of pitch change rather than AoA or load factor]

Capn Bloggs in another thread:
Fits well into our thread indeed.

The simulator experiment alone demonstrates what they didn't want to say.

Amen

So what do you think happened, and what is it you think "they" didn't want to say?

A clear restriction in the FCS that prevented the aircraft to get the optimal performance when most needed. If it is part of the normal functioning, then everything should be apparent in the decoding of the algorithms ...

So are you then suggesting that a fix was devised in secret and subsequently applied to the simulator and aircraft Bechet used for his own testing?

If so, given the timeframe, I don't think that's plausible - but as always I can't be certain. Having said that, why would there be a deliberate restriction there in the first place? It all seems a bit too cloak-and-dagger to me.

EDIT : Furthermore, in practical terms it seems unlikely - the software engineering processes made extensive use of regression testing and other techniques that maintained a record of changes made to the implementation over time. If a change was made, then every software engineer working on the project for Airbus at the time would have to have known about it, and with a team of that size it would be a massive gamble that they'd all keep their silence in perpetuity. As with all conspiracy theories, the fact that so many people would have to have known about it and subsequently keep schtum is a major roadblock to plausibility.

Additionally, if such a restriction was found and fixed, surely it would have been better PR for Airbus to tell the world that they had found and fixed the problem, would it not?

EDIT 2 : As for "decoding" the algorithms, the program source code is machine-generated from a graphical layout system - machine-generated code tends to be very difficult to reverse-engineer. You'd have better luck with the schematics, though you'd still need a qualified EE/Systems Engineer to decipher them for you.

From a software engineers' perspective I don't think that's anything out of the ordinary. Modular software design tends to inherently create "dependencies" between modules, so if a change is made in one module, then there will usually be at least some code affected in other modules - sometimes in quite surprising areas, depending on how the system architecture is laid out.

Your point also adds to why I think a surreptitious update scenario is unlikely, because it's not just Airbus's own software guys that would have to keep quiet, you've then got the sim engineers who have to install and test the update, the aircraft techs who had to do the same on the A320 Bechet used, and so on. Any one of them blowing the whistle at a later date would do far more long-term damage to the reputation of the company and the project than simply "'fessing up" to a mistake - as proven by the case of MD's DC-10 and the notorious "Gentlemens' Agreement".

Asseline had to face a restriction in the FCS that Bechet did not.
Now, suggest whatever you like ...

1. I figure Airbus has all the necessary ressources
2. It was not a time to reveal or detail any shortcoming
3. Airbus preferred propagating the myth that the envelope protection had prevented the aircraft from a catastrophic stall

No, something was different, we don't know what. We do know that Bechet's handling of the SS (emphatic and direct) was different to Asseline's (hesitant). I still think that's a better place to look than a nebulous "restriction" or "bug" in the software for which there is no evidence of it's existence or removal before Bechet came to do his tests.

The problem with Asseline is that he's been invested in this idea of a cover-up for so long that neither he nor his supporters are capable of seeing it any other way.

@DozyWannabe & CONF_itureAlgorithms need to be encoded not decoded. They describe the system in the nearest form of natural language (math included:p).Shematics already need to know System theory tools. Pilots need to know algorithms. Asseline seemed to don't know them. Did Chris Scott knew them?

The data show he's not necessarily wrong ... but Dozy has been invested in this idea for so long that no Institution and/or Corporation would ever protect their interest First that he is unable of seeing it any other way.

Not at all, but based on the complete lack of any evidence that there was anything untoward going on, no whistle blown at 25 years and counting, Dozy thinks it's a fairly unlikely possibility.

Also, it's a pretty tenuous claim that pretending there was no issue when there was one would be in the interest of the corporation or the institution involved. Bear in mind that the last company to try that ended up with their project given the nickname "Death Cruiser", and eventually went out of business - their assets becoming part of Boeing.

Obviously they got out from it pretty well as not only they managed to hide the FCS restriction but in the meantime they pushed for their favorite myth :
"the envelope protection has prevented the aircraft from a catastrophic stall"

A little bias seeping through?

Airbus was a little more astute than MD when it came to sweeping issues under the carpet. This was the lesson they learned from the DC-10s.

Nope, just a reasonable point as to why keeping things schtum is more likely to cause further problems down the line.

Sorry, I don't see what that has to do with "sweeping... under the carpet". The DC-10 got a bum rap because of the notoriety of the early accidents, but its later record was actually as good as or better than many other types. It was the damage to their reputation as a result of being caught trying to solve the cargo-door problem while avoiding an AD (and failing to fix it on the THY DC-10 that crashed) that caused the loss of confidence.

As you said - Airbus tended to be savvy and astute PR-wise. They'd have been unwise to go down the same route as MD did in the early days of the DC-10.

@CONF - *or* there was no "restriction" in the FCS, and the aircraft did indeed prevent a stall, because it didn't stall! The stall might still have happened after clearing the trees without active protection...

I'm wondering if you think that somewhere in the code, the A_MAX variable was transposed with A_PROT. Because of the way the systems were developed, this is unlikely because the unit and regression tests (both written separately from, and chronologically before, the code itself) would also have had to carry the same error, otherwise the tests would fail. The presence of regression tests also means that a later change would have shown up in the reports - something that every software engineer on the project would have seen. As OK465 says, the change would also have shown up in the sim logs, so everyone involved in the update would also have had to be sworn to secrecy. To me that's just too many potential loose ends for Airbus to consider taking the risk.

*or* one day you might understand what alpha max is ...

CONF, with all due respect - if you could just ease off on the insults and listen to what I'm saying, it might help us get somewhere.

I know what Alpha Max is, I know it wasn't reached and I suspect that the reason it wasn't has something to do with a combination of phugoid damping and Asseline's tentative stick input compared with Bechet's. I can't be certain, in fact none of us on either side of the argument can - all we can do is try to work out what's likely and what is less likely.

The assertion that the protections may have "prevented a stall" is based on Asseline's control inputs prior to impact, had they been applied to an unprotected aircraft. Without protections, full back-stick would have caused the pitch (and AoA) to increase regardless of whether there was sufficient airspeed. Without an AoA indicator to guide him, it's likely that the instinctive pull would have caused an unprotected aircraft to exceed alpha-stall (to use your nomenclature). Of course, Bechet's inputs during the test were even more immediate and positive, so he'd likely have stalled an unprotected aircraft as well.

I'm still learning about the aero side of things, but what I'm certain of is the methods used in developing the FCS software - and I'm telling you that it is *extremely* unlikely that there would have been a logical error limiting AoA to the A_PROT value, because there were around three layers of test harnessing that would have caught such an error, and each of those layers would have had to separately been developed with the same mistake - hence why I believe it's so unlikely.

EDIT:

Furthermore, whether the aircraft did or did not clear the trees is really just a footnote in terms of the investigation as a whole. Clearing the trees would have avoided the deaths, meaning Asseline would not have been prosecuted for manslaughter, but the existence of the video, whether it showed a crash or a close call, would still likely have ended his career. The investigation as a whole was more concerned with the lack of oversight from the airline and the decisions made during the approach phase than it was the mistakes made during the conduct of the last few seconds of the flight - as such, Airbus would have had no reason to cover up any technical problem. As the MD experience proved, it's far better PR for a manufacturer to put their hands up to a mistake and fix it than to risk being exposed trying to hide it.

EDIT 2: I'm surprised that you haven't picked up on the fact that I've gone from thinking phugoid damping wasn't a factor to being fairly convinced that it was. There's nothing wrong with changing one's mind if new information becomes available!

At a guess, I'd say the sim run that generated that particular data was not an attempt to duplicate the accident sequence, but to determine the limits of system behaviour. In this case they were probably trying to ascertain what difference more prompt and decisive action may have made, if any.

If I've understood previous posts on the subject, the simulator available at that time was still being refined in terms of aerodynamic response when compared to the real aircraft. If this is correct, then sim runs would really only have been of use in terms of defining systems behaviour. The real world run over the Toulouse runway would probably have been more along the lines of an attempt to duplicate the flightpath.

Take it as an insult as you wish, it was merely the observation that contrary to what you're pretending, you still don't get what alpha max is, and a stall could not have happened after clearing the trees as alpha max is not alpha stall.

Don't bother thinking for me, just read what I write, you will spare me theories I did not ask for.

Good for you ... what I was expecting is a BEA + Airbus analysis.
By the way if phugoid damping was involved, they could have detailed why Bechet was miraculously spared from it ... ?

@OK465

Actually, if you have access to a development simulator you wouldn't involve pilots at all; you would simply set up the initial conditions and feed in the actual control movements [sidestick; TLA]. This is what Aerospatiale did to produce the comparison shown on p13 Annex X of the BEA report.

The calculated response in the last couple of seconds (from just after the time where the IAS is marked as 114kt) is rather interesting - [Sidestick held fully back at -16deg, TLA at TOGA position]

A simulator stays a simulator today, since they always have commands which don't exist on the plane and use metalanguage :bored:

EDIT 1 : "real system" replaced by "plane", but it is the case too if you simulate a rocket or another computer,aso.