CONF...

If the DFDR and CVR are retrieved and readable, when all is said and done your words may seem prophetic.

Hi,

That's a very good interpretation of this picture .. good deduction
As a VFF I fly in many parts of the world ... nice aeras and bad aeras ..
I experience more than enough very bad turbulences .. (mainly at the same aera of the AF447 path and also in Asia and India aeras)
Those very bad turbulences never come suddenly but gradually .. for finish in a apotheosis (rice on the ceiling .. etc .. :) )
So .. what about a foolish speculation .. that in fact ... AF447 was not in very bad turbulences at the time of the event ???

No turbulence
 

jcjeant wrote :

Concur.

Turbulence not necessary to explain LOC in this instance.
CB inadvertent penetration not necessary either.
Less than awake crew not necessary either.
Experience not an issue (test pilot experience maybe).

All of this an opinion, formed after extremely detailed and careful study (by several knowledgeable people). But then, it must be wrong, or else...

On most modern aircraft if we're honest, CONF... Even venerable airframes like the 737 have, in their latest incarnations, developed some previously unknown behaviours (such as the Radalt/Autothrottle problem highlighted by the Turkish Schiphol accident). The manufacturers do tend to know about them, but they rarely if ever publicise them until they're absolutely sure they know what it is.

As such I hope some usable data can be extracted and we can finally get some answers rather than the ring-around-the-roses we've been having on the subject for the last couple of years...

Maybe I'm missing something -- and it was quite a while ago that the ACARS messages were picked through --, but I read this:

as meaning that the message refers to the same event, and the time difference is due to the relevant priorities of messages waiting to be transmitted.

Imho, this is NOT "foolish speculation", it is just plain evidence.
Consider these two points :
- read the LH crew (on the same route a few minutes earlier) statement : standard ITCZ crossing routine ;
- obviously, the "fasten belt" sign was "OFF" : hence the number of bodies recovered (SLF and CC), galley condition etc.. ;

all this suggests that the ride was not too rough.

From what we have here, (a routine ITCZ crossing), there is, IMHO, nothing "bizarre" with the captain having his rest time.
F/O's are fully qualified to deal with this kind of event.
If the captain MUST be on the F/D each time bad weather is forecasted, then why bother with R/P ?

About sleepiness in the cockpit :
Once you have completed your selcal check with the ATC in charge, you have NO "degree of radio trafic" for weather deviations by other A/C.
Requests and clearances go through HF, which you do not monitor, or through CPDLC.

Finally some scepticism of the urban myth that is this thread. Let us hope it will extend into the work of those whose "report" is sacred.

"Two highly qualified pilots, experienced and familiar with the route, flying a first rate ship for a flag carrier...."

My bias has been clear, and I am unrepentant. There was a structural "event", the computer got confused, or the a/c was rendered in some way uncontrollable, and not by the pilot(s).

UAS is patent in this family of a/c. Difficult recoveries are as well. Perpignan, and Asseline's wreck. The a/c is notorious for making a smooth recovery from UAS difficult (EASA AD).

In my firm opinion, almost all of the proffered evidence skirts the issue.

GerardC

If I was the captain, I would probably have had lot of confidence in my buddy in the RH seat as he had more experience on the SA route, 39 rotations against my 16. But I think I would have liked to take a peek at the Wx before taking a nap. I would also have expected him to listen in to other traffic, especially those ahead of us, on 121.5 or whatever.
The third guy, had the least experience of all, and it was suggested that he moved into the LH seat when the captain left. Is that significant?

Weather Radar
 

Just last night got around to watching PBS NOVA "Crash of Flight 447," which aired on 20 Feb. There seems to have been considerable discussion of it, judging by snippets, but there must have been a lot of it at the time, (now page 138 of prior thread) censored by the mods, as there are incomplete references and very few posts about then. That is disturbing.

Nova tried to explain the events leading up to the crash. I'll address only the Wx radar here.

Their expert, from NCAR part of NOAA, said the Wx radar on AF447 had only 50 mile range. Strike one. It's 320 miles, if the storm is dense enough to matter.

Nova showed a small/medium sized cell in the plane's track toward the major line of storm. Nova said the small cell would obscure the radar view of the major storm, causing them to stumble into it.

Poppycock/BS/Howash! The Wx radar on 447 returns were calibrated for rainfall intensity. When the intensity of return is high, an algorithm called "Path Attenuation Compensation" kicks in to assure calibrated display. When the storm is too intense for penetration without reserve to see the full picture of the "storm behind the storm," the Wx radar puts up a yellow band, called PAC Alert, at the outer range ring of the display.

I have great respect for Public Broadcasting in the US, but this program failed miserably. I wonder if major sponsor, billionaire David H. Koch, didn't have too much editorial input?

Salute!

I tend to agree with Bear. The opinions of Svarin and Jct are ones I also agree with.

I only post here as I was flying FBW systems before most, if not all folks here. Try 1979.

We can talk about all the ACARS transmissions and speculate on ocean currents and such, but the cause of the crash is clear to this old pilot - loss of control.

Richard Bach placed it like that - loss of control due to factors beyond human control, or big pieces coming off the plane or control linkages freezing/breaking. etc.

We can sort out the ACARS messages later, once the recorders are recovered, particularly the CVR. I want to know/understand what in the hell did the crew do? What did they see/feel?

The chance that pitot-static air data became unreliable is high. And as SV and JC pointed out, it's possible to have the phenomena occur well before entering turbulence. Hell, the X-31 had a problem in the clear, blue skies over California.

I shall be the last pilot here to blame the crew. If they encountered conflicting indications that things were going south and could not react in time, then the rest is history. Loss of control.

I do not subscribe to the theories that loss of control surfaces or other "parts" were the cause of the crash. Parts may have come off well before impact, but that was just part of the sequence. What in the hell happened at the beginning? Go see the Columbia loss data. Things were happening well before the thing came apart. The crew was unaware, and if they knew then they prolly couldn't do anything except pull up slowly and then bail when at a lower mach/altitude. They didn't know, but last transmission was the beggining of a question by the Commander that was cut off.

Of particular interest to me is the sequence of "reversion" of the flight control system when something bad happens. The Airbus does not have the same sequence of reversion modes that the F-16, B-2, F-22 and Shuttle have. Those planes focus upon crew control for a vastly different mission than commercial airliners, and have few modes that use nav data and approach/takeoff/go-around/cruise to "help".

K.I.S.S. x K.I.C.S.
 

We can separate Researchers, Engineers, Designers, and Technology Professionals in two groups:

One, the people that think a lot before implementing something and always uses an Strategy (from the Greek military commander Stratego that before the battles always had a well made plan). This group normally have SIMPLE SOLUTIONS for complex things.

Kelly Johnson used this approach to design THE ANGEL, capable to fly near the coffin corner, implemented in just 80 days.

He used the Keep It Simple Stupid approach for the U2 R&D and it´s implementation.

The other group uses the Keep it Complex Stupid approach. And at pressure from the bosses deliver anything...

Let´s imagine the scenario of:

1) Crew decided to keep course going through WX.
2) Entered an unexpected WX condition.
3) Had pitot´s blocked.
4) Entered unusual attitude.
5) Stalled and descended at unusual attitude.
6) Eventually leveled (almost) and hit surface

Question:

Is that model too simple?

Finger crossing to locate FDR/CVR and be readable.

Wht´s lacking in this analysis?

I doubt it. The ACARS Message was timestamped 213 as opposed to the initial one which was following the ECAM messages. They were timestamped 210.

Somewhere in the old Thread we learned that once an ECAM message has been posted it will not be sent again. Therefore the original message for disconnection of AP / A/THR was not repeated even if it occured a second time.

This points to a separate action i.e. potential re-engagement and second automatic disconnection of AFS.
This concides timewise with the loss of ACARS signal and the Cabin Vertical Speed advisory, which points to a severe (second) upset following/connected to this potential second disengagement of Auto Flight System(s).

The sequence could indeed match to the scenario mentioned in the AD.

From my point of view the general chain of events as described is quite likely pretty close to what happened in broad terms. Much of the discussions here revolve around the details behind.
E.g. :
- Why did they fly into the weather ?
- Why didn't the manage to keep it flying using Pitch&Power ?
- Why didn't they recover within 35000 ft of altitude ?
- Did automatic flight controls play a role in the upset/Non-recovery ?

And those are for me the really intriguing questions begging for answers by the FDR /CVR.

Pitot Probability
 

Back to the discussion of why the ACARS wasn't consulted first when the plane was overdue:

I remember reading early on that Maintenance in CDG noted the Airspeed fails reported by ACARS, and moved the plane that was AF447 up in priority to get the new pitots upon arrival.

It would be easy enough to automatically alert company Dispatch whenever a Flight Essential or Flight Critical component were flagged over ACARS. Also, maybe the ACARS should be programmed to transmit every minute as long as there is a Critical component flagged as failed?

Regarding Satcom vs. HF data transmission: the link is error-free, when it's working. HF works everywhere beyond VHF range, and is pennies to dollars less expensive than Satcom. HF continues to evolve, and a major data rate improvement was announced recently by Collins.

Ok, let´s look at each point:

1) Crew decided to keep course going through WX.
2) Entered an unexpected WX condition.
3) Had pitot´s blocked.
4) Entered unusual attitude.
5) Stalled and descended at unusual attitude.
6) Eventually leveled (almost) and hit surface

1) Several possible reasons:
a) Radar in auto mode. The best is to adjust manually the antenna elevation to characterize better the pattern ahead.
b) Shadow from first CB area (as PBS suggested)
2) n/c
3) A fact
4) A very probable fact as per ACARS msg´s
5) The same of previous point
6) A fact

Finger crossing to locate FDR/CVR and to be readable.

Thanks for replying, Rgds

That's not a secret that the Airbus is more automated and 'protected' than any others ...

Try a similar Radalt malfunction with the 320, it won't only interfer with the A/THR but also with the Flight Controls !
Very interesting to study.


Clearance - No clearance - we take the deviation we need - The information on the weathear deviations will be broadcasted or/and monitored directly on VHF 123.45 that everybody should syntonize ... unless I'm wrong and the appropriate frequency is different in that area (?)

Graybeard

I wouldn't go so far into wondering.

Programs like this are conceptialized based on written reports, various experts contacted, scripts written against what some experts postulate, then deeper experts are consulted for support or refute. Then talking heads that can explain a theory that the script is written against are asked to explain during filmng.

They were ammendable to making changes in their script if a talking head can't support it.

Just like with PPrune you have to be able to read between the lines to comprehend its veracity

Wx Radar Factor
 

RR_NDB
1 a) The Wx radar on 447 did not have auto tilt. They had the Collins WXR-700, not the newer WXR-2100.

b) As I explained in detail above, it was not possible for the major line to be shadowed by the intervening cell. In this case, PBS "expert" was a ground radar guy. Rockwell Collins, the maker of 447's radar, was not listed in the credits. If RC didn't want to get involved, they surely could have referred Nova to a knowledgable source.

Here are some possibilities:

1. The storm did not look bad enough to deviate around.
2. The line of storms was so long and uniform, it would not seem to matter where they penetrated.
3. The pilots did not have enough training in Wx radar ops.
. a. Nobody was looking at the radar returns.

From the ACARS and some great data via PJ2, and others, the a/p drops involuntarily when Law reverts 'out' of Normal to ALT. The Attitudes from the a/c computer that would cause the a/p to drop on its own when exceeded are here also. I recall that this a/c will not re-enter NORMAL out of ALT1/2 or DIRECT unless and until the ship has landed, and can be indexed (I think). So if from ACARS the a/p dropped by command from (Computer) ADIRU disagree, then I can't imagine why auto would be allowed to re enter the room even if invited.

Likewise, if DIRECT was attained, and RTLU and other minimum protections were operating, I can't easily imagine this accident happening unless something additional occurred (Svarin's nifty catch). The first and most likely candidate would be a/p trying to horn in after teasing the front with reacquired Bars and a less than sixty second a/s agree 2x2. But the pilots would know that NORMAL LAW is u/a, so why try to latch a/p only to see it booted (rejected) by the computer again??

Less likely, but possible, attitude excursions were attained that exceeded the a/p working limits and this caused the drop?

Less likely still, accompanying attitude extremes were disparate airstreams transiting the static and pitot ports which may have caused the TCAS and Windshear bunk that got sensed, as well as (actual) corrupt airdata??

These attitude extremes would be Wx and causative of UAS? Then, as the pilots recovered from UAS, abrupt and mechanically incoherent Pitch commands caused upset??

Too fanciful.... UAS, recovery, then the activity addressed by the (too late AD).

The more I try to understand this a/c's LAW, the more senile I believe I may be.

Graybeard

1. The storm did not look bad enough to deviate around.
2. The line of storms was so long and uniform, it would not seem to matter where they penetrated.
3. The pilots did not have enough training in Wx radar ops.
. a. Nobody was looking at the radar returns.

1. The storms were bad enough for the flights before and after to deviate
2. Then why not return to Brazil
3. I think that 39 SA rotations since being in the A330/A340 division is sufficient to be able to use the Wx effectively
3a. No answer to that possibility.

ALCS65, they went down 400 miles from Natal (which I'd half expect to be VHF ACARS equipped and 160 miles from Fernando de Noronha (which might be ACARS equipped.) They were pretty much out of VHF communications for quite a period of time. Off hand I cannot find a reference to either of them supporting ACARS. I'd estimate they were a good hour out from any meaningful VHF contact, maybe (much) more. I'd expect the cut over to satellite would happen at or some modest time after the switch from VHF flight control communications to HF.

Released documentation reports the last several position reports as well as the standard ACARS all, they say, through Inmarsat's AOW bird.

Electromagnetic messaging from aircraft
 

I get the impression that some commentators here are making the assumption that HF/VHF/UHF radio signals from AF447 would have been equivalent to that of trailing a broadband internet cable behind the aircraft.

Radio amateurs would love to have an antenna 40,000 feet up in the air but realize that in the midst of an electromagnetic storm, signals even 500 metres away might be unintelligible to a receiver due to noise, propagation anomalies and polarization rotations regardless of error correction protocols even if one is dumping the full American 1.5 kW into the ether.

Does anyone know if the QRN at the critical moments of the AF447 flight obliterated RF communications or whether the sequence of the ACARS messages was affected by poor reception?

One might fault PBS for repeating without independently checking, but that was originally a BBC production broadcast in the UK on 2010/05/30. I believe it was re-narrated (different accent of narrator) and possibly re-edited for the PBS release on 2011/02/16.

I believe 1., 2. are more likely than 3. or 3a. The whole WX deviation thing may be a red herring. In the animation that PJ2 posted, we can see the ELY flight flying without any deviations. In fact, it looks like it flew through a cell comparable to the one AF447 is projected to have flown into.

I think AF447 was unlucky to have entered a UAS condition. (quite independent of weather conditions.) AF knew this was always a possibility, no matter how remote. Hence, the pitot replacement program.

Two other possibilities I'd like to add to the above list, if I may:

4. Crew noticed something anomalous with airspeed and were fixated on understanding the situation.

5. Crew noted the weather ahead but deemed it a lower priority issue than something else. (If so, what is the USD 64K question)

Could this be the last actions of a crew desperate to get out of it...

Brazil - VHF ACARS
 

Most of Brazil is covered by SITA/Datacom VHF ACARS, and the eastern corridor is completely covered. This coverage extends to a station located on the offshore island of Fernando de Noronha (SBFN). The ACARS frequency is 131.55MHz, and this coverage was in place at the end of 2008.

AF447 was in range of the SBFN secondary radar until about 250NM or about 10 minutes after passing INTOL. With an omnidirectional 3dBi VHF antenna at SBFN, reliable data comms should have been achievable to at least 180NM with the a/c at FL350.

"With an omnidirectional 3dBi VHF antenna at SBFN, reliable data comms should have been achievable to at least 180NM with the a/c at FL350."


In HF/VHF/UHF in the middle of a Cb?

ARINC assumes a coverage of about 245nm for an aircraft at FL300. Both Natal (NAT) and Fernando de Noronha (FEN) are ACARS RGS's, but I put NAT at 590nm and FEN at about 420nm.

It does look like AF447 would have passed through several VHF ACARS sites prior to being out of range. Still hoping someone with first hand knowledge can comment on whether the VHF to SATCOM ACARS switch happens automatically as mentioned in the manual, or if maybe on a flight on this route it just uses SATCOM the whole time.

RGS List:
RGS list
(Note: Not a complete list.)

Map:
http://datalink.sas.se/ground/images/sita_eumea.pdf

Thanks for getting me to dig into the books again, mm43. Yes, tangential range is 300 miles or the equivalent in nm. Intelligibility would likely suffer some at about 200 miles. They were easily out of communications for some time before things pickled.

kilomikedelta - at 200 miles out they were not in a storm. That came up 200 miles further on. And the map I checked distances on had the LKP mislocated. They had it south rather than north of the equator. Teach me to double check things rather than wing it.

Salute!

Too much about ACARS and WX radar on the jets. And times for ACARS data not correlated with the flight recorders yet.

As Cog pointed out, at least one other plane flew thru the wx.

SOMETHING happened and the result was loss of control. I do not blame the crew.

I have flown thru the ocean CB's and they are nothing compared with a big CB over Kansas, USA. Whatever happened was before penetration of a line of storms. I use the Captain's body recovery to make the point that "things were going well until......"

Before I quit posting here for awhile, I shall bet that some sensor or input to the flight control system went haywire and the crew could not figure it out in time to do something constructive. My God, I just went thru the tech order for the jet and it is layer upon layer of inputs to the control system. Sheesh. Despite the plane's basic aero stability compared to the military jets that have FBW, it looks like even the most basic control laws still use many inputs from sensors. And some of those could be FUBAR.

The CVR will be more revealing than the basic flight data recorder, IMHO.

What did the crew see?

What were the flight conditions? When things went south, what was the environment? Were they pitching up and down and rolling left-right? I don't think so or the Captain would have been back in the cockpit.

SOMETHING happened before encountering the storms. The ACARS transmissions will soon be correlated with the flight recorders. We may be able to paint a picture of the final minutes or seconds. Findings will help to prevent another occurrence.

Although the Columbia breakup did not capture much crew talk, the flight dynamics were faithfully recorded. That tragedy took place very quickly despite indications on the recorder that something was terribly wrong 4 or 5 minutes beforehand. The crew didn't know, but at the last transmission they wanted to ask about something versus tire pressure in one of the wheels that the ground control was interested in.

sorry for the rant, and will stay silent for a bit.

P.S. The IAF guy on board Columbia was a student of mine in the F-16. I have a patch from that IAF bunch, and the IAF only sent one small group to check out in the first operational FBW jet in the world. So I have a special interest in incidents involving FBW planes, having been one of the first to fly the damned things.

Hi, KMD this is CW, :hmm:

With the proper selection of QRG (band) HF would be VERY RELIABLE even with high QRN and the poor current HF antennas.

I NEVEER lost a QSO in several thousands hours of HF operation using 200 W out sometimes lowering to 1.8 MHz (phone) due severe QSB.

In just one case due a VERY STRONG WX System i faced almost complete blackout of powerful MW BCST STN 80nm from my PSN but this is RARE.

I don´t believe, based in ACARS msg´s, they tried to use comm, HF or other.

They should be looking to other priorities.

The info on QRN at the night i think was lost and i think was not a factor.

On good antennas look at the trailing wire (30,000 ft) used by the 707 of USN for VLF to subs.

Unfortunately the current HF antennas are VERY POOR.

ok do I have this right?
 

A working scenario is - the FMC disables the autopilot because of unreliable airspeed measurement by the clogged pitot tubes. Once that is done, it should not be possible to re-engage the autopilot manually. Since that apparently did happen, the implication is that the crew re-engaged the autopilot after some initial upset, and this played right into the trap that was mentioned by the recent AD. The crew think they have recovered, and instead they have guaranteed the worst sort of stall, a flat spin basically.

So all talk about being asleep and weather and all that is just not relevant - because the problem here was systemic, even down to the over-reliance on programming instead of piloting. Ok that was a speculation.

JD-EE;
260NM and 174NM.

RR_NDB: What basis do you have for that statement, Mac?

Appreciate the apology Svarin - accepted.

OK. If I understand correctly, several of you are thinking:
1. Initial UAS leading to upset, leading to loss of AP
2. Manual recovery
3. Re-engage AP - at a lower altitude and several minutes later
4. Catastrophic LOC due to inappropriate pitch up issue covered in recent directive.

Now if you had a significant upset followed by a recovery, would you not expect over a period of some minutes:
1. Some form of radio communication
2. Pax and CC secured

Or do you suppose that the initial upset was sufficiently violent that anyone not secured was incapacitated (and stayed unsecured)? In which case, would the pilots really try and re-engage AP having gained a degree of manual control?

Am I misunderstanding you all here?

'But the pilots would know that NORMAL LAW is u/a, so why try to latch a/p only to see it booted (rejected) by the computer again??'

So if it couldn't/didn't get the a/p back, what was Svarin's ACARS message about ?

That's one possibility according to the AD.


The other is:

Short sequencing of a likely scenario
 

Gentlemen,

here is my own offer of a scenario sequence. All of this to be taken as an opinion. I would not bet my life on it, but it is the best I could come up with, with great outside help :

1 - high altitude ice crystals (outside CBs but inside thin layer of high altitude clouds around them - red returns on infrared satellite pictures).
2 - ice crystals block pitot probe drainholes (over-reading of IAS)
3 - Unreliable airspeeds -> A/P kicks OFF, along with other stuff
4 - a nasty invisible trap is set through FCPC2, no notice to the crew
5 - crew recovers from UAS without any upset
6 - UAS appears over, F/D bars reappear (see AD)
7 - crew reengages A/P, thinking the problem over
8 - trap is sprung, A/P kicks OFF again in violent upset (hard pitch-up)
9 - upset is not recoverable or not recovered (I expect 99.9% of airline pilots to fail that one)
10 - upset ends up as a deep stall or slow spin to the surface

This means the a/c likely did not crash right after end of ACARS sequence. The Cabin Vertical Speed message looks (to my opinion) like a consequence of FMGEC1 fault triggering a CPC reset or change of configuration, not a real descent (other messages exist for relief valves and other conditions).

Comments, questions and counters welcome.

I've been trying to imagine the mechanism by which re-engagement of the autopilot at corrupted airspeed would cause a violent upset. The AP controls altitude, so the first assumption one needs to made is that the altitude at re-engagement differs from the selected altitude. Let's assume it is lower. The AP will then command movement of the elevator control to initiate climb to capture the selected altitude. The AD suggests that the initially commanded movement of the elevator could be too abrupt when the real speed of the airplane is much greater than what the AP 'thinks' it is. The response of the airplane would then be more abrupt than what the AP is trying to achieve, which is, I suppose, a schedule of vertical acceleration and vertical speed. Wouldn't the AP adjust its command before it results in an uncontrollable attitude?

Goodness.

ELY 010 was on a parallel track to AF447. AF 447 was on Airway UN873. ELY was on UN866, over 100 NM to the west.