I think Machinbird has probably got it.

Davis talks of superstalls and T-tails (yes, I finally found my copy) but AF447 possibly got into something much more horrible than that - possibly loss of pitot info, possibly degraded FBW as a result and for all we know subsequent engine issues if the AOA got really excessive. As he/she said lets hope the info is still available.

ttcse "... your basic swept wing tends to tip-stall first. "

Say what !!!????
.

alph2z Ttcse's comment is correct. Aft swept wing aircraft begin to stall at the wing tip and the stall progresses forward and inward toward the fuselage. The aerodynamics guys can twist the wing and play games with the leading edges to modify this somewhat but at high enough angle of attack it will happen.
This tip stall causes a nose up pitch moment as the lift contribution from the ends of the swept wing decreases.
I've had a lot of practice over-rotating off cat shots and I can tell you from personal experience that it used to take up to full nose down stabilator to get the nose moving back down where it belonged.:}

Sorry if .... did they find the FDR, CVR etc ...

high altitude stalls
 

I like this quote from the opening sentences of D.P. Davies discussion on "Stalling" (but must admit I'm taking it completely out of context):
In many minds, including mine, stalling is something you get close to during takeoff and and landing, i.e. at relatively low altitude. In thirty-odd pages related to stalling, mr. Davies devotes only a single (not quite accurate) sentence to the effect of altitude:
The point I wish to make is that stall at high altitude involves fundamentally different (transonic) phenomena compared to low altitude. For example, for the A330 in clean configuration, alpha max is of the order of 14 degrees up to Mach 0.275, 8 degrees at Mach 0.6, and 4 degrees at Mmo.

Regards,
HN39

Surely "stalling" is what occurs when you exceed alpha max, regardless of whether or not that's at low altitude (takeoff/landing), or at high altitude.

Even though the fundamental reason for the stall may be different at high/low altitude in terms of the fluid behaviour, exceeding alpha max is a stall?

From this http://www.smartcockpit.com/data/pdf...light_Laws.pdf

follows that when in alternate law, all AOA protection is completely lost with dual ADR fail or ADR disagree, which was the case as we know.

HN39's numbers on AOA limits at altitude are interesting! Add to that lost AOA protection, then given the conditions they would have encountered... there's not much margin left.

Eugene, sure, exceeding alpha max will always end up in a stall. However, as all those of us know who fly, a perfect landing is a stall from an inch altitude, so that's where you want that to happen.

The trouble with the alpha max limit at altitude is that it gets so small, as HN39 has written. 4 degrees is not a lot of room in a turbulent environment.

I recall seeing photos of the SA 295 crash in 1987, where the landing gear tyres were still intact at a depth of 4,900 metres (16,100 ft). They also successfully recovered the CVR (but not FDR) from that crash. So in 2011, it would be somewhat easier to recover parts from a slightly shallower depth of 13,000 ft and compared to 1987's technology.

Yes they did and that's where all the controversy started and still runs to this day.

BEA manager interview
 

interview of Jean-Paul Troadec, BEA manager.05APR2011

ouest-france.fr - Rio-Paris*: un mois pour remonter épave et corps


.../...

Q : Cette zone proche de la dernière position connue n'avait pas été explorée. Pourquoi ?
A:- Entre la dernière position connue et l'impact, il s'est écoulé cinq minutes. Le Rio-Paris aurait pu parcourir une distance très importante. Au départ, nous n'avions pas de raison de penser qu'il était pratiquement à l'aplomb de cette dernière position.

Q : This area near the last known position was not explored. Why ?
A: Between the last known position and the impact, five minutes elapsed. The Rio-Paris flight could have cover a very important distance. At the beginning we had no reason to think that he was almost straight above this last position.

Est-on certain qu'il n'y a pas eu explosion en vol, mais décrochage ?
- L'explosion paraît très peu vraisemblable. Les débris légers retrouvés à la surface indiquaient déjà, par leurs positions symétriques, que l'avion avait touché l'eau en ligne de vol et en étant entier. Il n'était pas préparé à l'amerrissage car les passagers n'étaient pas équipés de gilets de sauvetage. Quant à l'hypothèse d'un décrochage, il est trop tôt pour se prononcer.

Q : Are we sure that there was no in-flight explosion, but stall.
A : An explosion seems very unlikely. Light debris found at the surface already indicated, with their symmetrical positions, that the airplane had touched the water in line of flight and was entire. It was not prepared for a sea landing because passengers were not wearing life jacket. About the stall hypothesis, it is too early to give an opinion.

Remonterez-vous tous les corps des passagers ?
- L'identification des victimes, quand elle est possible, s'impose à nous. Certaines familles le souhaitent, d'autres pas. Mais l'on ne peut pas remonter les corps de façon sélective. Les familles vont être associées à cette ultime phase des recherches. Nous allons les réunir prochainement et leur détailler la façon dont cette mission va être conduite.

Q: Will you retrieve all passengers bodies.
A: Victims identification when possible is essential to us. Some families wish this, others no. But we cannot retrieve bodies in a selective manner. Families will be associated to this last searches phase. We will meet them together soon and detail them the way this mission will be managed.
.../...

The wreckage is at 3900 meters depth. Three boats are considered for this operation:
René Descartes owned by France Télécom Marine, based at Brest, France. Another based in Cyprus and a third owned by Alcatel Lucent. The selected ship will be on the scene in one month.

Here is the main point that is almost ruling out an upset at 02.10 causing the crash where the wreckage is actually lying.

It certainly doesn't take 5 minutes to "deep stall" from cruise level to sea level, when in the meantime, the aircraft is circling at the same spot and will keep this attitude at impact.
Something else happened in that sequence that we still don't know.
CVRs and FDRs are really the key to fully understand it.

I'll bet that they will recover the engines.

I think, this are the windows in seat row 29...31 the first behind the emergencydoor, grity
http://www.bea.aero/fr/enquetes/vol....s/wreckage.jpg

Interesting discussions about deep stalls, but do the revelations of the last 48 hours change anything? ;)

Machinbird,
For the benefit of this pedestrian pilot, what is a "cat shot", and are you talking about a F-102/ F-106 (delta-wing)?

Quote from SaturnV (post #3000):
"...there was no lightning with this particular complex; none detected by satellite, nor by ground stations. Apparently lightning is infrequent in the ITCZ."

Before anyone gets the idea that sub-Saharan Africans (for example) know little about lightning (!), my understanding and experience suggest ITCZ lightning is less likely over the oceans.

I'm interested in the idea that some serious atmospheric changes may have been experienced if they got far enough to emerge from the north face of that Cb. As I've said before, this would have left them very poorly placed if they were already trying to fly thrust/attitude with unreliable airspeed indications.

Chris

PS (Edit):
takata,
Maybe the five minutes involved a tear-drop?

From takata's list, here are the three recovery ships that are being considered. EDT ARES is presently off Alexandria Egypt.

Rene Descartes
RENE DESCARTES - 9248100 - Vessel's Details and Current Position

Île de Bréhat
ILE DE BREHAT - 9247053 - Vessel's Details and Current Position

EDT ARES
EDT ARES - 9130755 - Vessel's Details and Current Position

EDT ARES aircraft recoveries:

EDT and Phoenix Recover Yemenia Flight Debris

Aug 20, 2010

EDT Ares assists with search - Offshore Shipping Online

@takata: Let's look at what we know.

- The last position received was at 02:10:34. That's 20 seconds after the AP switched off and problems obviously had begun.
- At 02:12:51, ADR disagree triggered, with concurrent loss of AOA protection

- Assuming they had taken evasive action shortly after 02:10:34 by initiating a left turn, they would now (~2.5 minutes later, at 02:12:51) be just about overhead where the wreckage was found and would probably be just about coming out of the left hand course reversal

- Enter a deep stall / flat spin at that point, facilitated by lack of AOA protection, pitch and power flying in turbulent conditions in a turn

- Assuming ~350ft/s descent rate in flat spin, impact with water would have happened ~1m40s later (if still at FL350 when entered)

- time predicted for impact: 02:14:31, that's 6 seconds off from when transmissions ceased. Very close.

My understanding is that the system switched to "Alternate law 2" from the begining at 02.10. Then, "AOA protection" was lost from the start of the sequence anyway.

Why would they initiate such an "evasive action" at 02.10 while they had to keep flying "pitch and thrust"?... this would not be very smart without a real need to do so. During all the similar sequences documented so far (more than 36), not a single crew did take such an initiative to make an "evasive action" only due to unreliable airspeed data.

Nothing is telling us for sure that the aircraft impacted at the end of the ACARs transmition (02.15). In fact, without engine power, no more ACARs would be sent anyway.

Only alpha floor is lost in alternate law 2. AOA is still monitored and protection now applies to stall speed rather than AOA. Protection is lost only if VSG1 can no longer be calculated (which I think was not the case, happens when slat/flap information is not available and GWT is no longer known), or when dual ADR fail or disagree is triggered.

In our case, it looks as if at least stall protection was available until ADR disagree triggered.

I agree evasive action in the form of a turnaround is virtually unheard of (in fact it's usually discouraged for reasons we can discuss in another thread). But we don't know what picture was painted on the radar. Nor what went on with the aircraft, maybe a something evolved that prompted them to return to the nearest landing opportunity which would have been behind them (I think?)

At first, I would like to thank you all for this really interesting thread !

After reading your posts I was thinking of an unrecovered spin as there is not much distance between LKP and crash site.

But ZeeDoktor did the maths and found 350ft/s... Not likely to be a spin...
That makes it 21.000ft/min almost a 200kt vertical speed (free-fall?). What about the fact that BEA hypothesis is a close to horizontal water contact.

I have further questions some people following this thread might answer :

What are the consequences of a high speed stall ? On a swept back wing ?

My guess are : First you lose altitude then static pressure rise and compressability effects are weaker.
Assuming this, would the airfoil recover lift by itself after a high speed stall ?

What if only one wing enters a high speed stall ? Self destruction might be more likely than spin at that speed isn't it ?

Well. but the late triggering of "ADR disagree" at 02.12 is obviously what caused ALT2, Autopilot OFF, Autothrust OFF, Rudder limiter control, etc. And this is what happend in the first place at 02.10 because of those Pitots reading errors... There is not so much possibility for the system to switch to ALT2 in any case.

The fact that this maintenance message was transmitted much later in the ACAR sequence doesn't mean that the system was not already in error before its maintenance time stamping or you'll have to find an explanation why ALT2 was already triggered at 02.10.

@takata: Eh bien, you're right the timestamp of the ACARS message is not necessarily related to when the fault occurred. The sequence of events up to the ADR disagree message however would be conducive with a disabled pitot and the corresponding crew action such as manual AP OFF.

@chaz: You don't need that high a speed. The point is I don't know how fast a flat spinning A330 will descend. Any number between 7000 to 20000 ft/min sounds reasonable to me: If you assume FL350 to FL0 in 5 minutes, you just need 7000ft/min ROD. If you take the highest number I've seen (taken from someone elses calculations on the previous page, look for the diagram with velocity vectors), the numbers can still be plausible.

Hopefully we'll know for sure soon. I was just (for my own curiosity) trying to satisfy possible events between LKP and what we now know as the wreck position.

AP OFF was certainly not due to manual action: AP was switched OFF by the system, hence, this relevant ACAR was sent (no ACAR would be sent if this was due to a manual switch). It means that before this point, AP was still ON, and after this point, it was automatically switched OFF because ALT2 was triggered (hence, because ADR disagree due to pitot errors)...

Engine damage
 

Re Chaz: Pulkovo flight 612 stall. From FDR data; stall from 11,677m reached vertical velocity of 94m/s around 5730m, decreasing to 72m/sec just prior to impact. Time of decent was 126sec.

Re the engines. Unless I am very much mistaken, the LPC fan blades are gone, the front sections of the outer nozzle containment shroud ring has gone, and what we see in both pictures are the remnants of the guide vanes which form the rear part of the bypass nozzle. Surely, this level of destruction is suggestive of a high engine shaft rpm at impact.

Compare this with the crush damage sustained on engine 2 for TW 800 and the still intact nature of the LPC fan, where the NTSB concluded that there was minimal amount of low-pressure rotor rotation at impact.

General Electric cf6-80
ImageShack® - Online Photo and Video Hosting

ImageShack® - Online Photo and Video Hosting


CROSS SECTION
ImageShack® - Online Photo and Video Hosting

FLIGHT INTERNATIONAL: DETAILED SECTION
ImageShack® - Online Photo and Video Hosting

REAR VIEW OF FAN SECTION
ImageShack® - Online Photo and Video Hosting


TW800 ENGINE 2 - IMPACT CRUSH DAMAGE
ImageShack® - Online Photo and Video Hosting

@Chris Scot I believe he means the catapult launch from an aircraft carrier.

Chris Scot
For the benefit of this pedestrian pilot, what is a "cat shot", and are you talking about a F-102/ F-106 (delta-wing)?

By takata:

"Well. but the late triggering of "ADR disagree" at 02.12 is obviously what caused ALT2, Autopilot OFF, Autothrust OFF, Rudder limiter control, etc. And this is what happend in the first place at 02.10 because of those Pitots reading errors... There is not so much possibility for the system to switch to ALT2 in any case."

Rudder Limiter. Fast Vertical Speed fall. Did the vertical stab last untill impact?

I don't agree with the post above engine assessment.

Suffice it to say that the high quality photos available to the investigators would be enough for them to make a determination. The photos we are looking at do not have resolution to assess the blade condition

The TWA engines came off during the spin to the ground and thus would be no higher in RPM than a very low windmill (they fall flat).

My interest is the ten or so seconds of flight before the error codes were set and not at what the engines looked like when they hit the water.

I think the only things you can conclude from the photos of the one engine are these:

1. The fan rotor and fan casing are both gone.
2. Most of, if not all of, the fan bypass stator vanes are gone.
3. The titanium fan frame and struts are visable but damaged.
4. The one piece turbine rear frame containing the rear engine mounts (attachment point to the pylon) is gone as well as the rear portion of the LPT casing exposing some turbine rotor blades.

TD

So far, all the rudder damages analyzed were fully consistent with an impact as it is described by the investigators findings. But one is free to believe whatever theory he likes more than any stuctural analysis (instead of checking all the published reports about it).

The structural analysis is telling that the damages were mostly due to vertical compression all over the airframe (and symetrical) with secondary horizontal forces and very few rotational forces. It doesn't mean "Fast vertical speed fall" but that the vertical pressure was higher than horizontal forces.

This is consistent with a stalled level slightly pitched up airframe at impact.

Now, what caused this end stall?
Did this aircraft departed from controled flight at cruise level or did something else happened in between that we are still not aware of?

Very true, my mistake - I mistook the stator vanes for the rotary vanes.

Now, something else than the "deep stall theory" should not be ruled out at this point.

Even if unpowered, such an airframe can glide very far (more than 120 NM from FL350) when it is able to trade off altitude vs distance. It would defintively not fall from the sky like that if it could remain wings level, with all its thrust or if it had enough altitude.

It would take a fair amount of structural damages or an incapacited crew to go down like that (like that?). More likely, before impact, she was left at low altitude without thrust and may have unexpectedly stalled at some point (hence no preparation for ditching).

There was already in the past some problems related with those engines (CF6-80E1 in this case CF6-80E1A3) ice/water ingestion:

@takata - You're clearly right that if the aircraft can maintain wings level at a suitable attitude, it can glide for a long time from FL350.

What if the aircraft entered a storm cell? The huge updraft/downdraft velocities inside the storm system could make the level aircraft lose altitude very rapidly?

I only fly gliders, not large aircraft, but I've flown in pretty rough conditions where we were wings level at normal attitude experiencing 10kt downdrafts (which is a very big downdraft for a glider at little more than 1000ft!). Under those conditions you can lose altitude very rapidly even if the aircraft is in a normal, non-upset condition.

Another speculation
 

Hello everyone – first time here - always wanted to be a pilot, but became an electronic engineer, then a teacher.

This is a fascinating thread, and I've learned a huge amount, especially about stalls in the last few pages. I have a question for the experts:

I recently saw a program on TV that explained the phenomenon of super cooled water vapour which can exist in very pure air, in the absence of particles to condense around. The program narrator shower how ice spontaneously form around an object (he used a real pitot tube) when introduced to supercooled water vapour in the lab

If this did happen, it would be more than a pitot tube or three that iced up, especially the wings and the leading components of the entire craft. (Some manner of) stalling would have occurred, but the subsequent loss of altitude would have cause the ice build-up to melt, and the possibility would exist for the aircraft to become manageable again, would it not?

What are the chances that the plane crashed in a belated but vain attempt by the pilots to pull out of a dive? Is it true that ACARS transmits error conditions, but not the reversal of a error condition back to OK?

Right, but we still don't know yet what were precisely the real weather conditions during this flight. As I said above, something still unknow happened between FL350 and the aircraft final attitude before impact at sea level. All we know for sure is that it took at least 5 minutes (maybe more) to go down and that the weather (ice, rain, storm) was certainly the primary factor above anything else.

All those pitot tubes don't freeze in normal/severe weather conditions, neither a dual engine flameout would be caused, neither such an aircraft would be lost with all hands from its cruise level.

This is why we should learn much more from the recorders (if they are found) and the wreckage analysis (including those engines remants).

Missing fan assembly
 

@Lomapaseo

Sorry, missed your earlier post about non visible blades. Missing fan assembly and casing could also be explained by shearing forces as you suggest, rather than high rotational energy disintegration.

As you say, let's wait for better photos.

takata

I doubt very little controlled gliding was done during the descent. Rather I suspect that immediately following the onset of their upset, the airplane was quick to begin a loss of altitude, with a corresponding increase in vertical speed as forward ground speed declined. The apparent fact that flight attendant stations (seats) were not occupied (on impact) and the lack of any kind of cabin preparedness gives credibility to the quickness of events and the violence of the upset .

http://d.yimg.com/a/p/ap/20110404/ca...xmh7N33euD5Q--
what can be the struktur right forward of the wing?
part of the wing or part of the fuselage? grity

Sure. But the real question is when did this final upset (causing the final crash, possibly stalled) occured?
At 02.15, 02.16, 02.17? Nobody knows actually.
Just before 02.15, at least one engine was obviously still running (certainly both) as ACARS were still transmitted...
Now, at impact time, this aircraft seems to be controlable in the case she was still powered. Was she?
How could we know for sure that she impacted at this exact time if she was not powered anymore?
I'm just pointing that this airframe is able of gliding unpowered but will need some altitude to do so, as well as being able to build up some speed to restart its engines.
When short of both, she certainly will go down.

Come on everyone - we have VERY LITTLE more information now than we had 3000 posts ago. Let's focus on the 'search' and now 'recovery' rather than another 3000 'maybes'. I don't think I could face wading through another 3000 flat spins/stalls/ glides/turnbacks etc etc while trying to find mm's gems.

Personally I suspect the tail section (and FDR) are sadly a long way away from this wreckage - I hope I am wrong.

Maybe I'm just getting confused, which wouldn't be the first time. However, my understanding is that alternate law does not maintain any of the protections in pitch, except maneuver protection. (ref. FCOM 3.04.27 P6). The normal "High angle of attack protection" is replaced by a new protection called "Low speed stability". (ref. ACA Memo P6). An aural "STALL, STALL, Stall" warning sounds at low speeds, but may also sound at high altitude where it warns that the aircraft is approaching the angle of attack for the onset of buffet. (ref. FCOM 3.04.27 P6). BEA report #2 explains that stall warning is triggered when the AoA exceeds a certain threshold that varies with Mach number. Low speed stability is described in FCOM 1.27.30 P2:My interest in collecting these quotes is just to get this complicated subject clear in my own mind. My curiosity concerns in particular the first sentence of the Low Speed Stability description: "At low speed, a nose down demand is introduced in reference to IAS, instead of angle of attack, ...". Does this apply at high altitude? What if the reference IAS is erroneous, i.e. 2 ADRs providing similar low airspeeds? Was the gear possibly down and locked before impact?

This hypothesis would be known quite fast as they have said that it would take only a few days to complete the whole reconnaissance and geo-localization of the debris field.

If the tail separated from the airframe before impact, why was the cabin still pressurized? Do you mean that it happened at low altitude or that it separated without releasing the oxygen masks or without compromising the cabin pressure?
If so, it would not be that far from the main debris field anyway.