Good for you. Since the topic we are discussing is AF447 accident and not some informative video, did you read BEA's third interim report? The one with DFDR and CVR readouts?
You are claiming that AF447 entered the storm cloud, which is at odds with Nz trace. Would you mind explaining the process that enabled you to heartily disagree with official interim report?

The ice crystals that are main suspect in the AF447 pitot blockage are found near active cells but not within them. Avoiding the radar returns won't keep you clear of the naughty ice.

Thermostat said:
Indeed we should. So, will you share with us the name of this program? :)
It sounds like the NOVA special (PBS) that came out before Interim Report #3. (Or was it before #2?) Though highly speculative, I thought it was really well done, considering they didn't know what the hell they were talking about! :)
It didn't posit a horizontal splashdown, even. :rolleyes:

Here is a quote from a declassified U.S. Army Missile Command, Redstone Arsenal, AL, document by Dorathy Stewart dated 27 June 1990 and entitled -

"Occurrence and Detection of Supercooled Water in the Atmosphere"

- with reference to - "Supercooled Water; Aircraft Icing; Clouds; Climatology".
and another;The last sentence of each quote is important, as we are dealing with events that had their beginnings in relatively clean air above an expansive maritime environment.

So it would appear that the military had at their disposal some definitive data on this form of icing over 20 years ago.

As BEA puts it on page 30 of Interim Report No.1:
IMHO the heating of the pitots is designed and has been demonstrated to function in any quantity of supercooled liquid water likely to be encountered. The icing detectors would have detected the presence of significant quantities of liquid water.

Clandestino,

For distribution of liquid water and ice particles in convective clouds, read IR#3 page 50.

Recovery trajectories
 

Well, I thought so too, but changed my mind after looking at Owain Glyndwr's data. With Owain's help I did a similar analysis and got similar results shown in the graphs below. The second graph shows how AoA slowly reduces until the airplane is unstalled.
http://i.imgur.com/1cftr.gif?1
http://i.imgur.com/Z2Saq.gif?1

The BEA's comments are accepted as such, but at the time of making them they hadn't completely ruled out that possibility. I have pointed out that the known science doesn't rule it out either.

In respect of Owain Glyndwr's stall recovery scenario, your follow-up is appreciated. It follows what I had in mind when creating graphics showing the relationship of elevator and THS angles to the relative airflow - some time back. As the pitch is reduced, the applied thrust leads to increasing IAS which rapidly reduces the the AoA and the wing attains the lift to allow the recovery. At the time I had envisaged no more than 10° ND would be required to get the wing flying again, and that appears to have been borne out.

assumed trajectory for recovery
 

Would somebody explain, on what facts the following happens in the first 12 seconds of the above graph:

Pitch is reduced from +15° to -10° = 25° change
FPA increases from -22 to -34 = 12° change
AOA decreases from +35 to +22 = 13° change

Speed in this timeframe is not specified, but is esentially important for generating lift. Drag in this AOA regime has an influence as well, where can it be found in the graph?

On what ground would the AOA change (13°) be grater than the FPA change (12°) by changing pitch (25°)?
Wasn´t the initial AOA according BEA higher than 40°?


Out of my own expierience with such a high initial AOA and such low speed the pitchdown initially increases the FPA by the same amount, the pitch is decreased,, giving only a minimal change in AOA and an increase in FPA and descent rate.

The expierience in the A330 sim from PJ2 and another Airbus pilot make me believe, that this magic graphs miss some vital points. Also all accepted procedures for combat jets as well as the new updated stall recovery procedures speak of the need to place the stick forward until the stall warning stops, which would not be the case until 45 seconds into the recovery like shown in the graphs.

RetiredF4,

To begin with your last point: If the nose-down pitch rate is maintained until the stall warning stops, that would occur at t=21 seconds and 27 degrees ND pitch. Level flight would be achieved at t=42.5 seconds, FL262, 324 kCAS. I suspect that Owain Glyndwr did not show that scenario in his graph because he found it difficult to imagine that an ATP would deliberately do that in an airliner.

Owain explained the 'Performance Model' in his post #87:
The 'Performance Model' consists of lift, drag and thrust data as functions of relevant parameters, and the trajectory then follows from Flight Mechanics, i.e. Newton's laws. It is driven by the specified variation of pitch attitude versus time. As opinions differ regarding the the achievable rate of pitch change, Owain's graph shows trajectories for pitch rates of 1, 2 and 3 degrees/second. These rates bracket the values observed by PJ2 in the simulator, and the responses of the airplane to elevator movement observed in the DFDR traces of AF447.

"Wasn´t the initial AOA according BEA higher than 40°?" The recorded AoA at 02:11:45 was just about 40°. In my graph I chose to stay with the recorded pitch attitude and calculated flight path angle, which result in AoA=36° for symmetrical flight in still air. The difference between that value and BEA's is attributable to the bank angle of 30° existing at that moment.

and there are drag-data available for those flown AOA´s, despite the fact, that this area was neither tested nor evaluated?

RetiredF4;

The pre- and post-stall aero data have been obtained from analysis of the DFDR data. Within the normal flight envelope these match authentic Airbus data. Outside that envelope the verification consisted of running the model through the recorded time histories of pitch and N1, and comparing the resulting 'model' trajectory (the violet line in Owain's graph) to the actual DFDR trajectory (the yellow/red line).

I'm glad we are bringing up this subject again. Pointing the nose below the horizon adds effective thrust along the aircraft axis in the amount of sine(negative pitch angle) x weight of the aircraft. In effect you are powering out of the stall with sufficient power to let the aircraft accelerate despite the high induced drag.

This probably isn't the most efficient way to get an airliner out of a stall, but for one that does not have an AOA indicator, it has a lot to recommend it in that the likelihood of secondary stall is drastically reduced. You just have to get the nose low enough initially so that you can actually accelerate. Surprisingly, angles of as little as 10-15 degrees nose low seem to be sufficient for the A330 to accelerate, despite having a very high initial AOA.

In the military aircraft I've flown, we were always concerned with departures from controlled flight (spins), thus the importance of unloading the wing promptly before you began to generate unintended rates of yaw/roll.

With an aircraft that is (relatively) stable in the stall, you just have to avoid doing something stupid with the controls.

I hope someone (like NASA) does further study on this method of stall recovery.

OK465,
Sorry, I forgot to label the secondary axis. Yes, it is Nzw, the acceleration normal to the flight path in 'g'.

Thanks. (deleted the post with the question, thinking some more about it.)

Why doesn't Nzw go to 1 at zero FPA?

Due to not enough test flights, knowledge of A330-200 limits and rate limitations of flight envelope is not sufficient.

Loss of speeds gets a critical situation to the aircraft's system design, reconing and protections. So the system reversed to Alternate 2B Law.

At this time we find a total lack for man-machine interface in terms of effective aircraft observability.

In that moment the restitution of human's classic hand and high altitude skills is not possible due to bad selected, bad teached, undertrained concierge-pilots. Don't forget their questions have been throwes off since 1/4 century by airlines, national and former european civil aviation autority.

That situation means that UAS is first an unsolved SYSTEM problem (despite inertial system exists on the aircraft which may not be clogged by ice ?) which will get a human UAS unsolvable problem,

Recording more and more parameters in the black boxes would be useless at preventing next SUCH accident.

Engineers and pilots have first to coordinate on the ground to elaborate transfer from process to SOPs.

We do not know the total conversation (due to BEA decision) and why1, why2, why3, why4, why5, and how the crew pulled to stall.

But, as they are not teached to, AF pilots could not invent "unloading the wing" technic, nor imagine computed but not displayed AoA, to replace the airframe shape in the airflow to recover.

1. None of the three simulations reported in the thread, and very well compilated by TurbineD, can get accreditation from test pilots, as simulators have not enough information. So no available SOPs can be written on these simulator experiences basis.

2. Owain G. takes only the half of the problem as stall is not a speed problem but an AoA problem :i.e. a free faller diving at 300 km/h falls without lift, only drag, but may deplace himself forwards with felt , strong lift, 40° pitch ND, falling 200 km/h changing only a little his arms positions and Cg position.

The deepstall is another case were you have speed and not enough lift.

3. On the other side, the "unloading the wing" technic, that Machinbird pulled like a magician from his large fight piloting science, is used in 0g certified Airbus flights to train astronauts. An that technic combined with official performance calculations reported by Hazelnuts39 show that THAT TECHNIC , and only that technic, could have been used with AoA and training, to overcome the stall with relieble new SOP.

In conclusion, developing these two points, further AF447 may be avoided. ;)

Franzl,

Good questions that deserve a reasoned reply. Sorry to have got out of sequence, but domestic duties got in the way.

I assume we are not doubting that stuffing the nose down will eventually give increased EAS and reduced AoA, the debate being the mechanism by which one achieves that end state.

I rationalised it this way:

Suppose we start with pitch +15, Gamma -25, AoA +40 and instantaneously pitch down through 25 deg to -10. (this is a thought experiment so we can do that)

AoA remains at +40 for the moment as you suggest, but that means gamma goes to -50. EAS hasn't changed, neither have CL and Cd at constant AoA so drag and lift are unchanged. However we now have a large additional gravity component so the aircraft will accelerate in absolute terms, i.e.TAS will increase, again as suggested in a later post. But in addition the aircraft is now descending more rapidly into denser air so there is a double effect on EAS, and the thrust is also increasing due to altitude effects which also helps a bit although this is not anything like as important as the gravity effect since the maximum thrust one might get at say FL350 is only about 7%W whereas a 25 deg FPA shift is worth 42%W.

If EAS increases at constant AoA the aircraft will develop an upwards acceleration normal to the flight path and thence an upward velocity component which, with the increased airspeed will result in a reduction in FPA. You can see this in HN39's traces. [It will also of course experience an increase in drag which partially offsets the benefits, but I think this only partially offsets them]

A reduction in FPA, WITH PITCH HELD CONSTANT, will give a reduction in AoA.

Cd is very sensitive to AoA above the stall, so the drag reduces more than the effect of increased EAS and the whole response is "speed unstable" as it were, with an increasing acceleration and reduction in AoA as time goes on until the AoA is reduced to below S/W limits.

That is, I think, the basic recovery mechanism, but in the early seconds the exact response depends on a changing balance between the various effects, and these in turn depend on the exact time history of applied pitch, so it is really quite difficult to answer your questions about the first twelve seconds.

Drag (and lift) in this case were calculated by analysis of the traces extracted from the BEA cartoons, the only assumption needed being how thrust varies with altitude and RPM and there are well established scaling rules for that. The justification for staying with these values is that the actual trajectory can be calculated by inputting just pitch and power, the output being speed and altitude. Hence the purple line on my original graph.


I don't think there is any conflict with your point about holding forward stick until the S/W stops, the only qualification is that I assumed the pilot flew constant pitch.


I don't think PJ2 will be upset if I say that in my conversations with him he confirmed that on at least one occasion he executed a recovery from 30 deg AoA using 10 deg ND pitch - it took about 15000 ft.

@Owain Glyndwr
"That is, I think, the basic recovery mechanism, but in the early seconds the exact response depends on a changing balance between the various effects, and these in turn depend on the exact time history of applied pitch, so it is really quite difficult to answer your questions about the first twelve seconds."

In the early seconds of each partial recovery, wasn't the crew presented with a "re-invigorated" STALLWARN?

Having sussed all the napkin calcs presented here, would the PF go immediately to ND? Or would he secure TOGA thrust, and "maintain altitude" with aft stick?

The DFDR tells us little of theoretical benefit to our crew from possibilities that were never entertained at any level of practicality.

Call me a wise ass, but I venture to say the crew was so absorbed in trying to "PREVENT" a STALL, that they spent no time trying to recover from one. Again, the focus is shifted from the problems that caused the crash to hypotheticals re: after the fact suppositions? IMHO.

If you have read my previous posts you will be aware that I avoid any comment on what pilots might or might not have done - I leave that to more qualified observers.

The DFDR is not intended for the use you imply, but you cannot say that 10 deg ND is a theoretical possibility outside the levl of practicality since they actually went there just after 02:12:45

With great respect I am saying that 10 degrees NOSEDOWN was not done for reasons discussed at length, specifically, a proposed fear of Overspeed, and the fact they didn't suss STALL.

I think you misunderstand my DFDR remarks.

At 2:12:45, at 10 degrees NOSE DOWN, were they there due command, or airframe trying on its own to recover from STALL?

IIRC, as looking at the traces will tell you, at 02:12:45 they were at 10 deg nose down with full up elevator and flight idle power. The aircraft, as I have said before, has no natural nosedown pitch at the stall. It responds to applied elevator and thrust. In this case it seems that reducing thrust to idle or thereabouts produced enough pitch to get to -10 deg.

No natural tendency to lower its nose in Stall? Are you sure? I thought that was one of the cues for a Stall suss. Is there something I am missing re: CG?

Thanks OG
 

Thank you for your kind and detailed answer. I have nothing to add to the main part of your explanation, it is along my thinking.

Imho it is even more difficult, when we consider that the aircraft was not that stable at all. Well, it did not depart into some kind of spin, but looking at the traces we had rolling, banking and pitching movements of the aircraft, and those might have been present due to pilot action or despite pilot reaction on aircraft behaviour. Those aircraft movements and associated flight control deflections cause aditional drag and prolong any kind of recovery.

The assumption of instantaneous pitch down by 25° assumes, that enough authority is available even with thrust at TOGA, or when thrust would be in idle we would have to consider the later application of thrust, its effect on pitching moment and the spool up time, which again would influence the acceleration. We assume, a pilot in this situation can decrease the pitch in a most expedite way by 25° and maintain exactly -10° pitch over a time period of more then 30 seconds, while still stalled despite the fact, that his aircraft is in a degraded mode of operation (not to talk about the stress of the crew itself). Those 30 seconds with an unwinding altimeter would be like ages.

I buy that value, it is a fifty percent increase against the posted values in the graph. If we further consider, that the average descent rate at that time was around 15.000´/min (at 02:11:45 at 35372 feet, 22 seconds later at 02:12:07 at 29736 feet, and that those values where present without stuffing the nose down (which would increase initally the descentrate), we can assume an aditional altitude loss of at least 5.000 feet for time delays due to this unplanned stall situation any crew would be confronted with even after this discussion here.

In a practical sense we would end up somewhere between 20.000 and 10.000 feet when starting the recovery close to FL 350 and we might crash with the same procedure when stalling at FL 250.

My harsh assesment (no offence intended) would be, it might look good on a piece of paper in a warm office, but i doubt the practical use of it.

Reduce the AOA until stall warning stops, and the sooner it stops, the better it will be.

I´m not saying that those calculations are wrong in any way, but i dont think they will lead to a practical recovery technique.

Stall recovery and AoA one mo' time
 

Glad to see we're back to recovery possibilities and maybe even probabilities.

Liked 'bird's and Retired's responses, a lot.

Unlike that little jet I flew long ago, the Airbus does not appear to have that "bucket" in the pitch moment characteristics. In other words, no "deep stall" and some semblance of pitch control by the humans besides those inputs otto is cranking in. Otherwise, why would the pilot even need to hold back stick? And it also appears that a few nose down( or even a few less nose up) inputs resulted in actual pitch changes, no?

The flaky AoA indications available and the system's disregard of AoA for a short time didn't help, but seems the speed indications came back fairly soon.

So I throw my hat in with 'bird and Retired and others postulating that a moderate nose down command and increase in thrust made recovery possible, and not requiring 20,000 friggin' feet. In other words, it was possible to fly out of the stall.

But ya gotta realize you are stalled! And seems some warnings were being blared on and off for the whole time. Apparently, the clue light never came on.

With the crappy AoA indications, all ya got is altitude and IAS/EAS/CAS ( pick one) to show you are making progress. No big deal, as once nose down 5 or even 10 degrees, the speed indications now available would show a slow but steady increase. And as O.G. points out, drag decreases immensely when lowering AoA from extreme values. I would imagine seeing a few knots increase every few seconds.

If I don't see some drastic changes in training and warnings to the crews about stalls and stall recoveries, I shall be greatly disappointed.

(earlier reply withdrawn).

That was caused by an error in the mathematical model. Thanks for pointing out the error. The erroneous graphs have been replaced by corrected versions.

Graphs
 

Would the performance graphs have looked different if they had been made following the suggestions in AB's" Safety First" -Page 10 issue Jan 2011 ?

These include :

"Autopilot and autothrottle...Off
Nose down pitch control...Apply until out of stall...
Nose down pitch trim...As needed ...( ...this will also address autopilot induced full back trim.)
Thrust.... As needed.....for engines installed below the wing applying maximum thrust can create a strong nose up pitching moment if the speed is low......
Speed Brake...Retract
Bank... Wings level Apply gentle action for recovery to avoid secondary stalls, then return to desired flight path."

The main differences appear to be Thrust and some contributors seem to think that 10 -15 degrees ND might be more appropriate or enough.

Lyman

Yes, you are missing something. I gave a description of stall on modern airliners is Post #294 of Thread 5.
Franzl

Yeah I agree the lateral behaviour complicates matters. A lot of that motion might be the natural response of the airframe when taken to those levels of AoA, not helped by the fact that the wing mounted roll controls would have been virtually ineffective and that the fin/rudder effectiveness was also compromised by their high effective sweep at high AoA. Not sure about drag, although there was a lot of scatter in the results which could have been disguised roll/sideslip effects, but also there might have been induced AoA effects.

Hey, I wasn’t assuming the aircraft could be pitched down instantaneously! That was a gross simplification to get the explanation across more easily. Neither would I assume that a pilot would in practice be able to maintain an exact nose down pitch. The sums, such as they were, were intended to give some idea of what MIGHT have been possible if mental blocks hadn’t got in the way not as actual recoveries.

The starting conditions included the rate of descent, so that bit shouldn’t apply. My typical calculation for recovery from FL350 took about 12000 ft, and I was fairly happy to say that looked reasonable against 15000 ft especially when one considers that the simulator modelling was probably even less representative than mine.
Let me put it clearly – my sums are intended to illustrate the sort of recovery that MIGHT have been possible. I am only too aware of their shortcomings. But even so I think they are a good indication that, as Gums says, the aircraft could have been flown out of it – if only!


Well I leave that to others more skilled than I in practical piloting.

Reduce the AOA until stall warning stops, and the sooner it stops, the better it will be.:ok:

If the A330-200 is balanced similarly to the A320, the nose will pitch down upon stall as long as the CoG is not set too far aft (it was well within the limit as I recall). In the A320 sim I had to work quite hard on the sidestick to keep the nose at 15 degrees up both on approach to and during the stall.

Lyman, if they were trying to avoid stalling, they were going about it in a very unusual manner - in fact they were doing precisely the opposite of what they should have been doing. Coupled with the "crazy speed" remarks, I think the PF, at least, was fighting a non-existent overspeed.

"Well I leave that to others more skilled than I in practical piloting.

Reduce the AOA until stall warning stops, and the sooner it stops, the better it will be."

They decided to hold the side stick full back for some reason and continued to do so with impact with the Atlantic Ocean. Several minutes was not enough time for them to figure it out. Too bad the passengers didn't understand who would be commanding their aircraft that day. How can you justify that lack of skill?

Dozy sayeth:
Well, whyinheck would they be similar? Two entirely different airframes, aren't they?

Designed to respond in as similar a manner as possible though. Remember type conversion between the two is supposed to be largely a matter of learning new numbers (and the odd difference in systems behaviour).

I'd have tested in an A330 sim, but there wasn't one available.

Also worth bearing in mind that according to the DFDR traces, the nose did come down below the horizon several times during the stall, even with the THS at 13 degrees nose up. None of the nose-down inputs were enough or held long enough to have commanded the pitch-down achieved, so I think it's reasonable to assume that the A330 is fairly benign in terms of stall characteristics.

Dozy. Yes, she wanted to fly, and without lift and tail to keep her flying, she will fall, nose first, of course. Agreed. So it becomes ever more important to suss the "place" of the crew. HF.

(Bold is mine)

That is certainly an intriguing statement. :)

Well, the sim I was in certainly got the gist of it, even if the numbers might not have been that exact. Specifically the loss of effective aileron control once established in the stall, necessitating use of rudder (albeit *very* gingerly) to level the wings. We know that the crew didn't try using rudder until very late in the sequence, and that some very aggressive aileron roll inputs were made, which might have mad a bad situation worse.

We all know that when that far outside of the flight envelope, the data is based on extrapolation from the numbers gleaned in flight testing rather than real numbers - but that doesn't mean the sim cannot be a useful learning tool.

I don't doubt that the PF and the rest of the crew were doing everything they could think of to rescue the situation - the question is why did their training not encompass the *correct* things to do?

Dozy.
Their training no doubt did include recovery from STALL, if not the 330, other airworthy examples that would translate to 330. The training that lacked was high altitude manual handling, and a complete understanding of the UAS. The rough handling can be laid at the PF, but the lack of a proper syllabus to include a fully understood UAS was not given, if even available. I think with gentle handling the rest may not have happened. There is a very suspicious lack of training re: STALL recognition, but as said, it shouldn't have STALLED.

No?

Very good, and I am not saying it ironically, I really think it's good we are starting to discuss reports and not media hype.

So let's take this line of discussion a bit forward; page 50 of interim 3 deals with general distribution of water and ice particles in convective cloud, as introduction to principles of operation of Wx radar.

Interestingly, BEA findings on the conditions that were most likely encountered by AF447 crew is also on page 50 but of interim 2.

The crux of the matter is not "doing" but rather "recognizing" the correct thing to do. I have not an ounce of doubt that the crew believed they were doing their best to save themselves up to the very end. That their perception of the situation had 180° phase shift relative to reality is tragic and it must not be simply written off as "pilot error". To paraphrase sir Winston Churchill: "The cost of solving the AF447 mystery must be reckoned in neither money nor manpower."

I don't think it is being (and I don't think you do either). The BEA's output thus far seems to be much more nuanced than that. Combine that with the sheer amount of time spent on this (by which I mean if it was considered a simple case of pilot error, the final report would have followed hot on the heels of interim #3), along with the collaboration between Airbus and Boeing regarding UAS and stall procedures - and I think we're looking at a landmark report here, once it finally arrives.

Dozy

There is a big difference between an uncontrollable nose down pitch when the aircraft stalls in the classic manner and the need to hold back stick in ever increasing amounts as AoA is increased.

The former is a sort of instability the latter a natural characteristic of a stable airplane. The A330 and A320 are alike in this respect and if you go back to that pseudo-pitching moment curve I posted many moons ago (and which mm43 referred to more recently) you will find a continuous relationship between elevator (sidestick) position and AoA.

bubbers44

I'm not trying to - as I said I leave that to others who know more about it. Doesn't stop me having an opinion though!

OK465

Not meant to imply anything more than that the simulator aerodynamic model would have been validated up to the stall but not beyond and that (for us) the simulator modelling beyond that is unknown so we cannot assess how realistic the response might be.
In that respect, and in that respect only, my simplified model might be equally representative to the simulator in that there was at least an attempt to use realistic post stall aerodynamics.

Dozy,

If he was fighting an overspeed, why would he select TOGA?

How speed ?
Aircraft operating in an environment has three dimensions
Vertical speed ?
Horizontal speed ?
Lateral speed ?
The combination of the three ?
I've said in other posts
The extracts of the BEA CVR show a dialogue crazy .. or as coming from people who are not in a same place
"You are going up
I'm going up?
I then descends"
And this continues throughout the recording
So when we study the CVR .. it does not really help understand what has happened ... instead it gets complicated