TakataI agree, it is not easy to explain, thus my reservations about that scenario. But lets think a bit about the characteristics of a pitot tube that has iced up and closed the drain hole.
With no airflow through the tube (drain hole frozen), and with a blocked inlet, the heaters must be able to add to the total energy in the tube and thus begin to melt ice. At some point, the elbow area of the tube should melt and allow the tube to drain. With appropriate coincidences you might get a temporary return of actual pitot indications (plural) long enough to set the RTL unit to the correct value and still get into trouble by virtue of the nose attitude. But it is a long shot. Too much has to happen synchronously.
The best RTL fit is with something like a dynamic loss of control. There the aircraft pitches up into a high speed stall, freezing the RTL unit at the pre-existing correct value, since the high AOA triggers ADR disagree results similar to the Perpignan A-320. It appears that the aircraft will not over G in this situation based on HN39's analysis. Only question is how you get enough loft to account for the known time of flight. Of course, no one has published A330 flight characteristics for AOA in the 50-60 degree range. The wind tunnel tests may never have been run.

RTLU
 

Now don't let the facts get in the way of a good story.:{
BEA Interim Report No.2
We know from the AOC positions that the aircraft was proceeding at M0.82 and at what seemed a constant GS of 467KTS. However, when changing to Alt Law the last valid airspeed data caused the RTLU to limit rudder angle to +/-7.9°. A higher level, e.g. FL362 would also provide the same CAS, and I am sure that HN39 will be able to come up with a suitable graph for higher levels.

In all probability the a/c was still at FL350 and had slowed slightly in "light chop". We don't really know, and the RTLU may be telling us something different, so is this another "red herring" or something that should be considered.

If you really want to consider this case, then, you'll need TWO pitot tubes iced up the same way and exactly at the same time for your system not rejecting the wrong data (but it would reject the third pitot, displaying the good airspeed, as wrong) -ie. something easy by pressing a button on the sim panel, but not so replayable in real life (possible though as Airbus worried).
I'm really unable to follow you here: any "high speed stall" won't freeze the RTLU, only wrong air data will.
Also, if the aircraft would wrongly pitch up, the end risk would be a low speed stall if both wrong airspeed retained were in reality lower than the system would think... not a high speed one.
Next, if your pitots are really clogged for inducing the system in error, you'll need to clean them again before they would record some wrong values due to any "stalled/spin" attitude (and now trigger unreliable airspeed, freezing RTLU at Mach .80, as mm43 quoted above).
Moreover, Perpignan is a really bad case of reference as this whole sequence was way too short as to even close a correlation window and to valid or reject any airspeed discrepancies [meaning, you can have many intermittent UAS alerts popping up in the cockpit without having a real/solid UAS event being triggered at the maintenance computer level]

This whole explanation is way out of reality. This is mostly the only thing acertained during this flight: they had suffered some REAL airspeed issues and the aircraft systems seems to have responded correctly by switching to Alternate Law 2.

The question remaining is not how the system performed but how the crew reacted to it. By studying similar unreliable airspeed cases, you'll see that the system is working as expected (as it is predictable) but you can't say the same about the crew reaction.

S~
Olivier

@ Takata, mm43, PJ2, Machinbird, Gums et al...

Can I be allowed to say this?:

(I am SLF and a very frustrated pilot. I should have done my licence years and years ago so that I could work amongst the likes of you people here.)

I would like to say that I am so incredibly amazed at the dedication you have all shown to finding the cause of this incident. The knowledge being shown by all contributors to this thread is simply astounding. From pilots, to engineers, to radio experts, to airframe experts and weather experts etc... - the collective brains thinking about this in the public domain is truly remarkable.

I am convinced that your collective thoughts are not too far from the truth of what happened. Once all the data is - hopefully - retrieved from the FDR and CVR etc... I am sure many of you will be proven mostly right in your assumptions of what happened to AF447

So - from one small guy who reads this thread daily - I would like to thank all of you on behalf of us.. the public... for your true professionalism and dedication to unlocking the events surrounding the sad demise of AF447.

I sincerely hope that the many authorities (who surely read these forums) take on board what has been learned here, and still to be learned from this incident, and apply many improvements to make flying safer for us all.

My heartfelt thanks on behalf of the PUBLIC...!

May I propose some answers to some of the questions posed here?

"Why did they 'elect' to fly through a CB?" :

1) I cannot see where we have established the exact track of 447 to show that they did?

2) When faced with the ITCZ or any other significant weather feature, sometimes there is no 'easy' way through, and a crew will 'elect' to follow what appears to be the best route (which can prove to be the wrong one).

3) I believe that from the recovered surface wreckage the indications are that there was no 'general' state of cabin preparedness for significant turbulence? This I find surprising. If I were to cross an active ITCZ I would have the cabin secured. An anomaly.

Regarding flight path from LKP to supposed impact point:

I believe we are surmising around 4 minutes between LKP and impact? We have NO idea of where the aircraft went in those '4 minutes'. There is enough time for 4 completely controlled 180 degree turns there. Why are we 'assuming' the aircraft just fell downwards with little forward speed? Why do we not consider the crew may have turned away from whatever 'happened' and may even have been returning to S America?

We seem to be building all sorts of dramatic manoeuvres, and then lines of 'theory' which rapidly gain a life of their own as semi-fact based on very little evidence. Ah! I forgot where I am for a moment..........

Zoom climbs, high-level combat manoeuvres - all very interesting - and I have been there, but why are we getting involved in them until we get the FDR/CVR readings? If they are unreadable, then game on, and this thread will become HUGE!

Takata,

you made the point that abscence of ACARS messages regarding the engines could indicate both engines died off simultanously. After that, ACARS would be inoperable, so the message reporting the engine failure could never be sent.

How simultanously do the engines have to fail, for no ACARS to be sent ? I would think, if they do not fail within 3 s at least the attempt for reporting the failure of the first engine should show up in the logs of the satellite operator ? Do you know of any precedent for the A330 in which both engines died off within such a short timespan ?

We do not yet know actual height, air temperature, windspeed or Mach No. Ground speed is not that precise as the exact time of transmission seemed to vary, apparently slipping by 15 seconds occasionally.

Flight plan was for FL370 M0.82, but would have needed approval to shift to FL370 from FL350. Recommended Mach number for Turbulence Penetration lower (M0.8 or M0.78?).

The BEA statement on RTLU very careful, and correct according to the figure posted by mm43

http://www.pprune.org/tech-log/44963...ml#post6411873

My use of online calculator generates 272KCAS from

M0.80 @ 35,000ft ISA +15 deg C
M0.82 @ 36,200ft ISA +15 deg C
M0.84 @ 37,400ft ISA +15 deg C
M0.86 @ 38,600ft ISA +15 deg C

Since people seem to be posting their full hypotheses, I thought I would have an attempt, as an alternative to grity's fascinating "zoom" idea. Apologies for the length. I also anticipate I am very likely to make mistakes about the aviating side, so stand to be justifiably flamed & hope to be corrected where I deal with those issues.

I am still wondering whether the CB is the initial problem and the pitots were not the precipitating cause, but a very early consequence (and ultimately fatal complicating factor). I've tried to pull together some of the excellent things posted since the beginning of the first thread. Here goes:

Step 1. Aircraft travelling at about M0.82, at FL350 on its planned track.

Step 2. They don't see storm on radar (for whatever reason) or for some reason try to penetrate it (eg including penetrating the smaller front system and not seeing the big one behind it). Crew is otherwise alert - as evidenced by eg the failed SELCAL calls just before.

Step 3. Tim Vasquez's graph shows that 2:10 corresponds with a position in the middle of the larger CB assuming more or less continuation of direction of flight, which is why I choose it as the initiating factor.

Step 4. The first encounter is a severe up or downdraft - aircraft rises or drops very suddenly and very significantly due to air movement or change in air characteristics or both. Given the overall high loss of altitude with short net horizontal distance, I assume a strong downward motion.

Step 5. We get the host of messages, AP off, ATHR off, windshear detection off, alternate law and TCAS fault. Also the flag on cpt pfd spd limit and on f/o pfd spd limit, with the LKP (higher priority message) interposed between them. This is where I am likely to be totally wrong, and stand to be corrected, but could this all be a consequence of a sudden massive altitude change due to the storm?

Note that BEA said of the TCAS fault:
Step 6. Assume that still at around 2:10, we then get pitot icing. (eg encounter large amounts of supercooled water/ice etc). This in turn causes the associated MAINTENANCE STATUS EFCS 2, MAINTENANCE STATUS EFCS 1, Pitot and "ADIRU1 BUS ADR1-2 TO FCPC2" fault (that svarin noted some time back ) messages. We also get a rudder travel limit fault at this time.

BEA said of the ADIRU message (report 2, p37): The pitot message appears to indicate a drop in calculated speed. See report 1:
I seem to recall a suggestion that the rudder limiter message is also a result of unreliable airspeed (ie the sudden drop), but cannot find justification now. As MM43 just noted, BEA wrote that the examination of the actual assembly from the recovered VS showed it was at "7.9° +/- 0.1°. As an example, at FL350, this travel is obtained for Mach 0.8 +/- 0.004, corresponding to a CAS of 272 +/- 2 kt." At this point the aircraft, at least, thinks it is around kcas 272 [strikethrough]M0.80[/strikethrough].

[So at this point -- the a/c has dropped suddenly, and IAS has dropped suddenly, and AP/ATHR are off, plus we are in alt law. Now the pitot failures have started to affect everything else. Pilots also confronted by a bunch of unusual messages all at once?]

Step 7. Moving into 2:11, we then get then get a whole host of failures consequent on the icing - Flag on CPT PFD FPV and F/O PFD FPV. Then speed or mach function and and ADIRU2 fault.

BEA states that the speed or mach function fault could have meant CAS was above or below the relevant limits -- but which of these is the case is not captured on this aircraft (report 1, p53). Given we had a drop in airspeed before, hypothesise that it dropped below the lowest limit to trigger the message.

BEA says of the ADIRU2 fault (report 2, p 38):

Step 8. At this point, indicated airspeed is way off -- probably much too low. It is dark, middle of thunderstorm, turbulent, possibly raining heavily. Aircraft is in alt law, no AP and the computers are failing. The crew has no AoA indication in the cockpit. What would you then do? Nose down either unintentionally due to disorientation, or unintentionally as a result of turbulence or intentionally from a desire to increase speed if they believe the speed has dropped? Or nose up, again either mistakenly, from turbulence or on purpose? Either leading to fast steep dive or nose up deep stall?

Step 9. At 2:12 we have Nav ADR disagree, which is the only message time stamped 2:12. According to BEA (Report 1, p 50) "This message indicates that the EFCSs have rejected an ADR, and then identified an inconsistency between the two remaining ADRs on one of the monitored parameters." There are no other messages time stamped 2:12. Pilots are attempting recovery? Attempting to hold prior speed and AoA?

Step 10. We then enter the period 2:13:14 to 2:13:45 where no ACARS are received. It seems that aircraft orientation prevents ACARS messages being transmitted. (Control has now been fully lost? Possibly severe roll, or spiral? Or a hard dive, or very high AoA stall where the body of the aircraft itself blocks the ACARs transmissions from reaching the satellite?) In any of these cases, losing altitude very fast, horizontal speed decayed (if nose up) or a minimal component of overall velocity vector (if nose down). PRIM1 and PRIM2 faults, and FMGEC faults are generated but are saved up until ACARS can resume. (Aircraft so far out of normal envelope that the computers finally give up the ghost?)

Step 11. When ACARS transmissions resume, the PRIM1 and PRIM2 faults are transmitted. Then the interposed ADR2 maintenance status message (time stamped 214) then the FMGEC message (time stamped 2:13), then finally the cabin altitude advisory (time stamped 2:14). Aircraft is relatively low now. High vertical speed. Little horizontal speed.

Step 12. ACARS messages stop. Note that one reason BEA gave is "loss of electrical power supply: this would imply the simultaneous loss of the two main sources of electrical power generation." Finally, the flame out? Might this suggest high AoA deep stall (though why would they not have flamed out earlier if this was the case) or consequence of steep dive or during a partial recovery from steep dive? In any event, it seems no relight succeeds.

Step 13.Within a minute or so (perhaps longer), the aircraft has hit the water. At this point, aircraft is at level orientation (wings level, nose slightly up, the famous "en ligne de vol") - possibly partial recovery but with not enough altitude left?

it is part of the smal risidual-risk in any security concept with multible equivalent sensors..... IF they are disturbed in an unexpected way, they can be disturbed all together in the same moment (if a tsunami kills the emergency generator, he can kill also the second-third-and so forth....)

Hi,

BEA communication 10 May 2011
10 May 2011 briefing

RTLU consideration
 

Machinbird,

Tail down as in "nose slightly up" as noted in respect to the BEA report of the impact.

Others have addressed the RTLU issue recently and I need to reflect on it further.

Thanks.

P.S. I'm a retired Engineer & Program Manager in the field of cockpit instruments and engines and in those roles have worked at a company interfacing with both Boeing and EADS on a number of commercial aircraft development programs (not the A330).

auv-ee, wouldn't the thermocline bounce the surface noise back up leaving the far sub-surface quiet?

{o.o}

Machinbird, getting very deep into "noise" involves long time averages and usually something akin to a phase locked loop. With very slow Morse code your ear doesn't do as well as circuitry because your ear does not average over one second periods (or longer) very well. The Moon bounce people have some pretty clever software to play with.

Heck, even GPS is designed to work with jamming signals that are up to 100,000 times larger than the desired (P-channel) GPS signal. That's digging the signal out of noise. Again, it uses known signal properties and long term averaging to make it work.

The submarine had no chance of reception unless it's crush depth is WAY below that of anything I could find on the Internet a couple years ago. If the pinger's exact properties were known the long term average MIGHT have worked. But no two pingers will be close enough in operating parameters to characterize them to that degree of accuracy. The real weak signal people use very stable frequency references and time references to make it work.

(That's how you make something about the size of a martini olive talk to a satellite - very slowly.)

{^_^}

Presumptions
 

Quote from takata:
"I'm feeling that we are now back to the early discussions we had two years back, right after the crash."

I have some sympathy with his point, but what do we know now that we didn't on 3/7/2009, after publication of BEA interim Report No 1? Not a lot, really, although much of the evidence therein has now been supported by images from the debris field. People speculate on why the crew allowed the aeroplane to fly into a Cb, but we have as yet no proof that it did, let alone that they did. We don't even have proof that weather was contributory; beyond the probability of pitot icing, which many previous flights had coped with. The only big news is that the sea-level impact took place within 5nm of the LKP.

Also find myself in some agreement with BOAC. Perhaps we are all anxious to record our ideas while we have the chance. But there does seem – in the anticipation of an imminent, recorder-led denouement to this mystery – to be some risky polarisation of theories. Here are a couple more.

Quote from tubby linton:
"Imagine the folowing scenario- pitots ice up generating an overspeed. The aircraft pitches up and the athr retards the power to idle..."

Cannot speak for Tubby, but most of you seem still to be convinced that pitot icing always leads to the related ASI over-reading. This presumption is unfounded. Further homework is called for.

Quote from Machinbird:
"We do have actual wreckage this time and it is located very close to LKP pretty much down track, and indicating a descent to the water that almost guarantees it was in a deep stall."

Another equally dodgy assumption, apparently encouraged by the idea of over-reading ASIs. If I happen to fly my airplane over the final-approach fix at 10,000ft, and subsequently land on the same sea-level runway, does that mean that I had to lose the height in 4 or 5 miles?

Chris

RR NDB

Regarding the long wave stuff - that's REALLY simple minded stuff. All you are doing is reducing the bandwidth on an amplitude detector. A factor of 100 reduction in bandwidth (say 30 Hz from 3kHz) gives you that 20 dB you talk about.

Unfortunately that does not apply here. The pinger is like an old time radar pulse, short and sweet. It has no spectrum spreading that can aid discrimination with a low power pulse spread over time. It has no serious possibility of long term averaging. So that equipment you write about is just about useless.

And that is probably what Thales tried to do with the submarine records. There's no magic they could pull out of recordings. The pinger pulse is too short.

{^_^}

The problem with the short ping time is you cannot get the bandwidth down very far or you attenuate signal energy along with noise energy. The signal energy is coherent and the noise energy is not. So increased averaging times decreases your ability to detect the signal once you hit the threshold. Filter ringing, as it is called, screws you if you try really narrow bandwidths.

Um, and the oscillator chirp and drift mean you cannot use coherency tricks.

{^_^}

Don't you find the evidence posted by BEA compelling? It is pretty clear that AF459 could see a similar picture to that portrayed and took evasive action.

Flight Paths of Flight AF 447 and of the flights that crossed the zone around the same time
http://www.bea.aero/fr/enquetes/vol.....03h45.p14.avi

ELY010 also appears to have crossed "red" but maybe the cloud picture had moved a bit by then, but is it possible AF447 could see a gap? There is the 3NM offset to suggest a small deviation, maybe to find shortest crossing?

sensor_validation,

I think you've answered your own question. Sometimes the only apparent alleyways can be frighteningly narrow, and only a mile or two off track (if you're lucky). They can also turn out to be culs-de-sac. (Think the French don't call them that, preferring "Impasses".)

Regards,
Chris

The BEA web site now shows pictures of one of the engines retrieved from site and they have also retrieved the avionics rack as well so they now also have the quick access recorder and more data to hopefully download.