Go Back  PPRuNe Forums > Flight Deck Forums > Tech Log
Reload this Page >

AF 447 Thread No. 11

Tech Log The very best in practical technical discussion on the web

AF 447 Thread No. 11

Closed Thread

Old 25th Mar 2013, 22:09
  #1 (permalink)  
Moderator
Thread Starter
 
Join Date: Apr 2001
Location: various places .....
Posts: 6,226
AF 447 Thread No. 11

Thread part -

This thread series started out of an earlier thread which starts here and finishes here.

Another, slower moving, thread on the subject and covering the period from the original thread, above, and the start of thread #4


Total posts to date 17634 .. with in excess of 2.7 million views overall.

Links to the various BEA reports are given below. If I have missed any of the useful papers, please PM me with the URL and I can include it.

(a) BEA site - French, English
- Report link page - French, English

(b) Interim Report (No, 1) Jul 2, 2009 - English

(c) Interim Report No. 2 Dec 17, 2009 - English
- Update Dec 17, 2009 - French, English

(d) Estimating the wreckage location Jun 30, 2010

(e) Wreckage search analysis Jan 20, 2011

(f) Briefing and associated update May 27, 2011
- Briefing - update French
- Briefing - update English
- Briefing - update German
- Briefing - update Portugese

(g) Interim Report No. 3 July 2011 - French, English

(h) Links to final report Jul 5, 2012 and associated documents.

Miscellaneous pertinent links -

(a) Airbus Operations Golden Rules
(b) ALPA FBW Primer
(c) C* and Civil Transports - Cranfield
(d) Longitudinal Flight Control Design - RAeS
(e) Longitudinal Stability: Effect of High Altitude and CG - Boeing
(f) pitot static system performance - USN (Pax River) FTM
(g) The Problem of Automation: Inappropriate Feedback and Interaction, Not Over-Automation. Donald A. Norman UCSD
(h) Upset Recovery - 16MB zip file
(i) Ironies of Automation. Lisanne Bainbridge UCL
(j) Cognitive Capability of Humans. Christopher Wickens Uni Illinois
(k) Trust in Automation: Designing for Appropriate Reliance John D. Lee, Katrina A. See; Human Factors, Vol. 46, 2004
(l) Training for New Technology. John Bent - Cathay Neil Krey's CRM site


Search hint: You can search PPRuNe threads with a filter in Google by using the following search string example -

ths af447 site:http://www.pprune.org/tech-log/

This will search for mentions of THS in the AF447 threads of tech log only.

Just change the THS in the string to whatever you want to look for. This allows one to search for any term or phrase of interest throughout the threads.

Adding the site:URL end part is the magic that restricts Google to only searching in Tech Log.

This filter technique is absolutely wonderful and can be used generically to find things of interest in PPRuNe - appears to work OK in the PPRuNe search function as well.

In respect of Google searching, JenCluse has added some suggestions -

a) indenting the text block with a (one) Tab, *and*

b) emphasizing the fact that it is a search text block with some manner of . . .
<SearchText>, or
"SearchText"

Another search engine approach tailored for AF447 can be found here.
john_tullamarine is offline  
Old 25th Mar 2013, 23:28
  #2 (permalink)  
 
Join Date: Aug 2003
Location: Sale, Australia
Age: 74
Posts: 3,829
Simulators Still Not Accurate Enough On Stalls

As experts struggle to identify why the crew of Air France 447 lost control of their A330 over the South Atlantic Ocean nearly four years ago, the industry is also still struggling to develop the precision data needed to accurately reproduce a stall in a Level D simulator. The lack of accurate stall data limits entry and recovery practice because the computers running the simulators have no idea how the aircraft will actually perform.

Safety experts believe better data is needed to properly prepare pilots for a portion of the aircraft’s performance envelope that was once thought easy to avoid.

At a recent conference held at the Royal Aeronautical Society in London, officials from both Airbus and Boeing joined forces to explain the situation to date as well as where the industry still needs to go. Airbus test pilot Terry Lutz believes the day may be coming when pilots will need to hand over more control to onboard computers when the situation becomes too chaotic. This is reminiscent of the blue “level” button in use aboard the four-place Cirrus SR22 piston single that automatically brings the aircraft back to a wings-level attitude even if the autopilot is turned off.

Boeing engineer Paul Bolds-Moorhead reiterated the monumental task of developing accurate lift and stall data in the high-altitude regime, where the stall and unusual-attitude behavior of transport aircraft is typically never tested.

Simulators Still Not Accurate Enough On Stalls | Aviation International News
Brian Abraham is offline  
Old 26th Mar 2013, 09:19
  #3 (permalink)  
 
Join Date: Mar 2010
Location: South Korea
Age: 57
Posts: 113
Airbus test pilot Terry Lutz believes the day may be coming when pilots will need to hand over more control to onboard computers when the situation becomes too chaotic.
Why not address the root cause? Provide better pilot training and make the pilot-machine interface simplier and less confusing so situations dont become so chaotic.

Edit. Sorry guys, I should delete this post. This has been gone over hundreds of times.

Last edited by Cool Guys; 26th Mar 2013 at 09:58.
Cool Guys is offline  
Old 26th Mar 2013, 09:59
  #4 (permalink)  
 
Join Date: May 2010
Location: PLanet Earth
Posts: 745
Originally Posted by Cool Guys View Post
Why not address the root cause? Provide better pilot training and make the pilot-machine interface simplier and less confusing so situations dont become so chaotic.
Why not do both?
Trying to find a way of having a better grip on aircraft behaviour after things went very wrong still seems a positive thing to me.

Furthermore this move could also help in laying the foundation for developping automatic stall/spin recovery technology similar to what's available in the F-18. Considering today's volume of Flight Control systems in modern airliners this should not be impossible once sufficient aerodynamic data is available.
For Recovery that might be a more realistic approach than training all Airliner crews worldwide in doing 'advanced aerobatics' in IFR conditions...

Curious to see if the manufacturers will go more extreme in test flights in the future?
Maybe even Remote Controlled, similar to the QF-4 Drones?
In a 10 Billion$ Development project these costs should be rather negligeable.
henra is offline  
Old 26th Mar 2013, 22:20
  #5 (permalink)  
 
Join Date: Jul 2002
Location: UK
Posts: 3,181
Originally Posted by PJ2
The other notion that Vaughn pioneered (but which we in this business are familiar with by other names) is the "normalization of deviance". For those new to the notion, one way of expressing the meaning is, the reducing of margins of error in standardized proven systems because the standard can successfully be reduced while maintaining sufficient margins of error. (There are other ways of expressing this of course!).

So rather than nefarious activities behind engineers' backs, most managers could claim to be onside with the safety people but they also knew that they had to be mindful of schedules, budgets, regulatory affairs, government politics and public perceptions. As you would expect these are very bright and aware people but none of that guarantees that phenomenon such as normalizing standards through "reasonable justifications" is the right thing to do. Often it is seen as "amoral", and calculated towards pedantic goals only in hindsight.
Nail on the head.

While us folks on the shop floor like to kvetch about management and accounting, the truth is that there's rarely any generalised malice in their intentions. Whether we're talking about NASA in the case of the Challenger and Columbia disasters, the FAA and McDonnell-Douglas in the case of the DC-10, or De Havilland and the British Government in the case of the Comet, the fact is that most of the time the dangers were not realised because the decisions were mostly in the hands of people who did not fully understand the consequences should something outside their experience go wrong.

As I understand it, in the case of the Columbia breakup, the engineer who first advanced the foam strike theory had to fight tooth-and-nail to get it tested - not just with management, but with other engineers!

Originally Posted by RetiredF4
I wonder myself, in the briefing was nothing new on the planet, mostly stuff what expierienced old school pilots learned about stalls amd falls from the beginning regardless whether it was civil or military.
Well, even line pilots learn about stalls and recovery when they do their initial PPL training. The issue seems to be that this knowledge was not sufficiently revised once they'd earned their seat on an airliner.
DozyWannabe is offline  
Old 27th Mar 2013, 15:33
  #6 (permalink)  
PJ2
 
Join Date: Mar 2003
Location: East of the sun, West of the moon
Age: 70
Posts: 2,345
Dozy;
Re, "Whether we're talking about NASA in the case of the Challenger and Columbia disasters, the FAA and McDonnell-Douglas in the case of the DC-10, . . .", others may accept that the two cases are similar but just for the record, for a host of reasons that have to do with individual, group and corporate behaviours and how knowledge was formed and subsequently reified within the two respective organizations, I would never include the DC10 case in any category that suggests that the two (NASA/Challenger, McD/DC10) are examples of the same phenomenon or are of the same character; in my view they are not. For a good understanding of the difference between the two, The DC-10 Case has to be read alongside
Vaughn's book Vaughn's book
.

I certainly don't want to re-argue anything here and there are elements of the same behaviours in terms of the normalization of deviance and other phenomena discussed in Vaughn but a key difference is, for example, expressed by Peter French: "There can be little doubt that many engineers and managerial personnel at McDonnell Douglas (and Convair) knew, well before the Paris crash, of the potential for a Class IV hazard* due to defective design of the DC-10 cargo door latching system and the floor structure.", *A Class IV hazard is a hazard involving danger to life. (The DC-10 Case; A Study in Applied Ethics, Technology, and Society, ed. John H. Fielder, Douglas Birsch, SUNY, 1992, New York, p.178).

Last edited by PJ2; 27th Mar 2013 at 16:11.
PJ2 is offline  
Old 28th Mar 2013, 02:23
  #7 (permalink)  
 
Join Date: Aug 2005
Location: fl
Posts: 2,561
Remember how simple it was when us old guys learned how to fly? Hand fly the first few thousands of hours, learn how to use the AP, but know how to turn it off if it isn't right. Look at the FD and see if it wants to go the same way you do and if it disagrees just go the way your clearance goes. Very simple and everything is safe. Now, we have children of the magenta line flying.
bubbers44 is offline  
Old 12th Apr 2013, 15:59
  #8 (permalink)  
 
Join Date: Jun 2011
Location: France
Age: 38
Posts: 14
stall recovery in AF447

I just read the leaked (French) experts report on AF447 (I won't put the link here, just google it) and it explicitly mentions that the plane may not have been recoverable at all because of the alternate 2b law, which limits the deflection of the elevator when the stick is full nose down deflection. They showed that even when one of the pilots put the nose down for two seconds (and then neutral) the elevator did not move much.

As a contribution factor of the accident, the report states:

6.4: alt2b flight law, which is hybrid, does not guarantee longitudinal stability for speeds below the stall speed (VS1g)

6.5: The auto trim of the THS is not deactivated below the minimal flight speed (VLS) in alt2b flight law.

Based on these, I think the flight crew was probably in an unrecoverable situation (no matter what they would have done), unless maybe deactivating the ADRs to go into direct mode (as in the recent EVA air incident).

Being a faculty in systems engineering (on a totally unrelated domain), I am shocked by these conclusions. The A330 cannot stall in normal mode, but why do we need a crash to understand that the ALT2B mode is putting pilots into trouble during a stall? They must have done tests during certification. I'm sure they must have stalled the plane in alternate mode to see how it worked...

I know that pilots are to blame etc, but this sheds a different light. Even if the pilots had been professional test pilots, maybe the plane would still have been irrecoverable.
predictorM9 is offline  
Old 12th Apr 2013, 17:26
  #9 (permalink)  
 
Join Date: May 2010
Location: PLanet Earth
Posts: 745
Originally Posted by predictorM9 View Post
I just read the leaked (French) experts report on AF447 (I won't put the link here, just google it) and it explicitly mentions that the plane may not have been recoverable at all because of the alternate 2b law, which limits the deflection of the elevator when the stick is full nose down deflection.
Did you read the previous Threads on AF447?
If so you would have seen that this topic has been discussed ad nauseum.

Conclusion: We will not know for sure if
it would have been recoverable once the AoA was >>30°.
Nor will BEA, even Airbus will probably not know for sure if it was recoverable past that point.
Some argued it was rather likely it would have been technically recoverable (I tend to belong to this group) some concluded it wasn't.

Drawing a conclusion from looking at 2s of SS input is pointless.
It took more than half a minute of predominantly NU input to initiate this condition. It is safe to assume getting the Trim and Elevator down again would have taken at least the same dedication in the opposite direction. This hasn't taken place in this case so it's all guess work.
Airbus might have a clue at least regarding the logic for the deflection of the control surfaces. But even that won't give a defnitive answer re the recoverability once being at the extreme AoAs.

But:
No Airliner is tested nor required to be recoverable in this attitude. Even the trusty old 737 might not be in certain conditions.
This is so far outside any tested and validated Flight Envelope and no one seems willing to sacrifice an airliner plus crew in order to find out during flight test.
Would you volunteer ?

Conclusion: Don't fly at AoA's of 45° !!!
Simples.

Last edited by henra; 12th Apr 2013 at 17:27.
henra is offline  
Old 12th Apr 2013, 17:48
  #10 (permalink)  
 
Join Date: Jul 2009
Location: France - mostly
Age: 78
Posts: 1,689
Originally Posted by PredictorM9
it explicitly mentions that the plane may not have been recoverable at all because of the alternate 2b law
I've read it (the expert group's report) too. Where does it say that?

Last edited by HazelNuts39; 12th Apr 2013 at 17:52.
HazelNuts39 is offline  
Old 12th Apr 2013, 18:29
  #11 (permalink)  
 
Join Date: Jun 2011
Location: France
Age: 38
Posts: 14
Drawing a conclusion from looking at 2s of SS input is pointless.
It took more than half a minute of predominantly NU input to initiate this condition. It is safe to assume getting the Trim and Elevator down again would have taken at least the same dedication in the opposite direction.
What the report says is that the full deflection of the elevator down was not achieved with this input, because of the alt2b law. While the same law gives the full deflection of elevator up when the pilot was pulling the stick.

And 2 seconds is not a lot, but in Quantas 72 that's exactly the time that was needed to flung the passengers up at -0.8g. 2 seconds (that's on the cvr). Granted it is not a stall and all, but the 2 seconds nose down input, at full authority would probably have improved things for them.

No Airliner is tested nor required to be recoverable in this attitude. Even the trusty old 737 might not be in certain conditions.
This is so far outside any tested and validated Flight Envelope and no one seems willing to sacrifice an airliner plus crew in order to find out during flight test.
Would you volunteer ?
I'm not a pilot anyway, but a 707 was crashed in the eighties for some fuel test, maybe they could do a similar test over an ocean with a remotely piloted airliner. After all, this might help calibrating models for simulations, and thus give an insight to all planes. It is not an A330 specific issue.
I hope in 2013 we have the capability of simulating the unsteady flow around wings, even in the transonic regime. What is lacking is probably just some experimental data to validate the computations.

Conclusion: Don't fly at AoA's of 45° !!!
Simples.
Yes but all that is easier when you have an AoA indicator. As you mentioned in your post you are probably safe below 30 degrees, which is not that much.
predictorM9 is offline  
Old 12th Apr 2013, 18:34
  #12 (permalink)  
 
Join Date: Jun 2011
Location: France
Age: 38
Posts: 14
I've read it (the expert group's report) too. Where does it say that?
It says:

"Dans cette phase de vol, la poussée maximale appliquée a un effet à cabrer non négligeable ,' le
THS est en butée à cabrer et s'ajoute donc à l'effet à cabrer; la loi de pilotage Nz (à accélération
constante) limite le débattement des gouvernes.
Toutes ces conditions réunies ne permettent plus la récupération du décrochage."

page 55 of the report. The Nz law is I guess the alt2b. They say that the elevator authority was limited.
predictorM9 is offline  
Old 12th Apr 2013, 20:00
  #13 (permalink)  
 
Join Date: Jun 2009
Location: NNW of Antipodes
Age: 75
Posts: 1,328
They say that the elevator authority was limited.
Obviously they [the group of experts] didn't read, or understand the post by Owain Glyndwr in Thread No.6.
mm43 is offline  
Old 12th Apr 2013, 20:21
  #14 (permalink)  
 
Join Date: Jul 2009
Location: France - mostly
Age: 78
Posts: 1,689
Toutes ces conditions réunies ne permettent plus la récupération du décrochage
That is an opinion, referring to the conditions at 02:13:56 when the airplane was descending through 5000' at 10,000 fpm.
The Nz law is I guess the alt2b. They say that the elevator authority was limited.
That statement needs to be qualified, I think.
HazelNuts39 is offline  
Old 12th Apr 2013, 20:38
  #15 (permalink)  
 
Join Date: May 2010
Location: PLanet Earth
Posts: 745
Originally Posted by predictorM9 View Post
What the report says is that the full deflection of the elevator down was not achieved with this input, because of the alt2b law. While the same law gives the full deflection of elevator up when the pilot was pulling the stick.
It is due to Load Factor law (aka Nz law) which applies in all Flight Laws except Direct Law not only in Alt2b. Alt2b is insofar different as all 'protections' are lost.

That said Full Nose Down command in an Nz law should correspond to a G Load Factor significantly below 1g even if the reference Speed is pretty low. This should result in a Nose Down Elevator response finally. Taking into consideration that average G Load in a developped stall will be close to 1g an Nz law should still allow ND Elevator. As there are G Load fluctuations in the dynamic evolution of the stall they might have been unlucky that during these 2s of ND input the temporary G load was on the low end of the range. Unfortunately I don't have the traces in front of me atm. I don't see why Full sustained ND input should not lead to a significant ND Elevator deflection in an Nz Law !?

And 2 seconds is not a lot, but in Quantas 72 that's exactly the time that was needed to flung the passengers up at -0.8g. 2 seconds (that's on the cvr). Granted it is not a stall and all, but the 2 seconds nose down input, at full authority would probably have improved things for them.
That was at a rather high speed and starting in a mostly neutral THS and Elevator.

I'm not a pilot anyway, but a 707 was crashed in the eighties for some fuel test, maybe they could do a similar test over an ocean with a remotely piloted airliner. After all, this might help calibrating models for simulations, and thus give an insight to all planes. It is not an A330 specific issue.
I guess no one is keen on showing crashing airliners these days. They want to convey the image that these things don't fall out of the sky...

I hope in 2013 we have the capability of simulating the unsteady flow around wings, even in the transonic regime. What is lacking is probably just some experimental data to validate the computations.
I'm not so confident. I would think there is quite some iterations between calculation and actual data required to come to a reliable prediction.

Yes but all that is easier when you have an AoA indicator. As you mentioned in your post you are probably safe below 30 degrees, which is not that much.
No disagreement here.

Last edited by henra; 12th Apr 2013 at 20:40.
henra is offline  
Old 13th Apr 2013, 09:03
  #16 (permalink)  
 
Join Date: Jun 2011
Location: West of Offa's dyke
Age: 82
Posts: 472
I hope in 2013 we have the capability of simulating the unsteady flow around wings, even in the transonic regime. What is lacking is probably just some experimental data to validate the computations
. I'm not so confident. I would think there is quite some iterations between calculation and actual data required to come to a reliable prediction.
I ain't quite that easy! It is (relatively) straightforward to establish the 'static' characteristics (lift, pitch, rolling moment due to sideslip etc.) well beyond the stall using wind tunnel tests - to the best of my knowledge the theoretical methods available still struggle with nonlinear separated flows, but I've been out of the business for a while.

The problem is that to adequately simulate the aircraft's motion one needs to have the 'dynamic' characteristics also (eg rolling moment due to rate of roll etc.) and I certainly don't think CFD will be able to handle that for a while. Some of these are extremely nonlinear in the stall regime and cannot be easily simulated.

For a good description of the problems see AIAA 2005 - 5933.

Obviously there will be research aimed at improving this situation, maybe using dynamic models.

Last edited by Owain Glyndwr; 13th Apr 2013 at 09:07.
Owain Glyndwr is offline  
Old 13th Apr 2013, 23:09
  #17 (permalink)  
 
Join Date: Jun 2011
Location: France
Age: 38
Posts: 14
That said Full Nose Down command in an Nz law should correspond to a G Load Factor significantly below 1g even if the reference Speed is pretty low. This should result in a Nose Down Elevator response finally. Taking into consideration that average G Load in a developped stall will be close to 1g an Nz law should still allow ND Elevator. As there are G Load fluctuations in the dynamic evolution of the stall they might have been unlucky that during these 2s of ND input the temporary G load was on the low end of the range. Unfortunately I don't have the traces in front of me atm. I don't see why Full sustained ND input should not lead to a significant ND Elevator deflection in an Nz Law !?
It should, but from the data on the leaked report, when the copilot puts the stick full nose down for two seconds, the elevator goes from 30 degrees pitch up to 20 degrees pitch up. How fast is the elevator supposed to move? If it is 5 degrees per second then ok they had some authority.

I think this alternate law is a big problem anyway. They should revert to direct law in case of event like this: if you are stall, with the stick at neutral, then the plane will demand 1g and will keep the stall.

On an related note, the report says that
"En cas de perte d'au moins deux ADR, la stabilité basse vitesse est perdue, il est surprenant que le trim automatique ne soit pas désactivé au voisinage de a PROT, alors que c 'est le cas en loi ALT1"

The experts are surprised that the auto trim doesn't stop trimming the plane up once alpha prot is reached, as it is the case in ALT1. I am too.
predictorM9 is offline  
Old 14th Apr 2013, 07:46
  #18 (permalink)  
 
Join Date: Jul 2009
Location: France - mostly
Age: 78
Posts: 1,689
The experts are surprised that the auto trim doesn't stop trimming the plane up once alpha prot is reached, as it is the case in ALT1. I am too.
Alpha prot is a function of Mach. No valid airspeed - no alpha prot. Latched.

Last edited by HazelNuts39; 14th Apr 2013 at 08:47.
HazelNuts39 is offline  
Old 14th Apr 2013, 10:55
  #19 (permalink)  
 
Join Date: May 2010
Location: PLanet Earth
Posts: 745
Originally Posted by predictorM9 View Post
It should, but from the data on the leaked report, when the copilot puts the stick full nose down for two seconds, the elevator goes from 30 degrees pitch up to 20 degrees pitch up.
Hmmm, I'm looking at the traces now but I have some difficulties spotting the point where 2s of Full ND input are applied. Could you direct me to it?
I do see one short spike of half ND input and later one very short spike of 3/4 ND. What is my mistake?

Edit: I think I see which part of the trace you are referring to.
2:12:32 ?
If you look at the other elevator angle changes you will notice that they don't change very swiftly, even in the NU direction.
It took ~5s to get from 15° to 30° NU elevator angle.
Also not every single movement of the SS is followed by a dramatic change in elevator angle (or any at all). So the speed and onset of elevator angle change seems pretty much consistent with the other angle changes. There is quite some dampening visible in the traces between SS input and elevator movements.
I attribute the magnitude of the slow reaciotn to the ND input you mentioned also to this dampening.
The elevator movement slows down when the ND inputs reduce/cease so it is difficult to predict how far and how fast the elevator would have continued to move from this relatively short sustained SS command.

As I said before: You will not come to a generally accepted conclusion as to what would have happened had they continued to command full ND.
I stand by my tendancy to conclude that the elevator would have moved to a clear ND position. As you would expect even for an Nz law. It just doesn't happen in 2s.

Last edited by henra; 14th Apr 2013 at 11:28.
henra is offline  
Old 14th Apr 2013, 13:23
  #20 (permalink)  
 
Join Date: Jul 2009
Location: France - mostly
Age: 78
Posts: 1,689
From BEA's Final Report, para. 2.2.5:
In the specific case of alternate 2B law, some coefficients used in the longitudinal flight control law become speed-independent and are set for the maximum speed for the aeroplane configuration (330 kt in clean configuration). This hardly modifies the behaviour of the aeroplane in comparison to normal law, but can nevertheless induce an unusual response dynamic when the aeroplane has an abnormally low speed for the configuration.
Nevertheless, the pitch attitude reduces from 8.5° NU to 2.5° ND in about 6 seconds. (The Nz law uses a mix of Nz and pitch rate in its feed-back term).

Last edited by HazelNuts39; 14th Apr 2013 at 14:36.
HazelNuts39 is offline  

Thread Tools
Search this Thread

Contact Us Archive Advertising Cookie Policy Privacy Statement Terms of Service