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

AF 447 Thread No. 7

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

AF 447 Thread No. 7

Old 27th Nov 2011, 17:36
  #541 (permalink)  
 
Join Date: Jul 2002
Location: UK
Posts: 3,093
Likes: 0
Received 0 Likes on 0 Posts
The other explanation is adrenalin. The "startle response" is quickly gaining currency as the biggest potential killer in line flying. With Colgan 3407 you have two tired pilots, one of whom is shocked into pulling up into a stall when the stall horn sounds. West Caribbean 708 had a Captain who was convinced he had a double flameout and concentrated on that, pulling up all the way into the ground. Birgenair 701 had a captain who focused on the initial overspeed warning to the exclusion of all else and pulled the aircraft into a stall.

Then you have success stories - US Airways 1548, BA038, Aloha 243 and United 232 as a few examples where in all cases the crew kept calm, overrode their internal panic switches and found a way to resolve things.

Whether you're flying, driving, doing your job or whatever - after doing it for some time things become routine - if something happens to break that routine then you have to keep your head about you and logically focus your way out of an event, but you've got millions of years of evolution behind the human brain which triggers a "fight or flight" response. Sometimes it works to the good and people have found almost superhuman resolve in themselves to get the job done, but sometimes it does not - people freeze or try random combinations of things they half-remembered in training or heard on the line with tragic results. It's all part of being human and training must address that.

Remember that the this was the PF's first day back on the job - he had just ended his holiday in Brazil and his wife was down the back - if there's a situation more conducive to be startled and try absolutely *anything* to keep that aircraft at the assigned cruise level, I can't think of it.

@Lyman - a phugoid cycle is something that happens over longer periods of time than the turbulence encounters the aircraft was experiencing. Remember JAL123 and UA232 - with dead flight surfaces the cycles happened over the course of about a minute and a half, so it was about 45 seconds of slow climbing followed by about 45 seconds of slow descent. The turbulence encounters and autoflight corrections we're talking about here were on the order of a few seconds at most. The pitch attitude was increasing at A/P disconnect as a result of one of those corrections, and stopped short of 4 degrees nose up because of the A/P disconnect. The elevators don't remain in the position they were in when A/P disconnects, they are set to match the last requested flight path angle, which was about 2-3 degrees nose up.

Last edited by DozyWannabe; 27th Nov 2011 at 18:04.
DozyWannabe is offline  
Old 27th Nov 2011, 17:48
  #542 (permalink)  
 
Join Date: Jun 2009
Location: NNW of Antipodes
Age: 81
Posts: 1,330
Received 0 Likes on 0 Posts
@Dozey;

You are right regarding the A/P disconnect. The RTLU locked because the ADRs disagreed, and I have made an edit to post #526.

I am not going to get into the argument regarding the cause of the climb, other than to reiterate that the elevator and THS responded to the PF's SS stirring and the aircraft pitched into the climb almost immediately at 02:10:09.
mm43 is offline  
Old 27th Nov 2011, 18:15
  #543 (permalink)  
 
Join Date: Aug 2011
Location: Grassy Valley
Posts: 2,074
Likes: 0
Received 0 Likes on 0 Posts
So, a robust updraft. Very warm, sufficient to corrupt Mach value? Unless the increase in VS is due baro, then it is airmass. The point is, the a/c was out of phase, and the pilot was given an a/c that was climbing with the NOSE showing low. His Pull was additive to an input made by AutoPilot("STALLSTALL"), and the confusion began. Yes, he should have been "ready", and managed a better get than he did.

The problems encountered were not complicated, but the timing was.
Lyman is offline  
Old 27th Nov 2011, 18:27
  #544 (permalink)  
 
Join Date: Jul 2009
Location: DFW
Age: 61
Posts: 221
Likes: 0
Received 0 Likes on 0 Posts
Originally Posted by Lyman
How many times had PF been handed an a/c out of autopilot? As many as he had landings logged, most likely. His stick work is a copy of the two vids grity has posted, perfectly acceptable, when approaching to land.
Those two videos show but one pilots control technique. I bet you a cup of coffee that they could have both kept their hand off of the SS and the end result would have been the same. In other words, the Bus doesn't require that kind of SS input in/on final. The pilot was working too much. Those little corrections he made were not likely necessary because the Bus will stay where you left it, and/or return to where you left it without any input. Once you learn too let it correct itself in roll on approach, you find yourself making very few roll inputs.
TTex600 is offline  
Old 27th Nov 2011, 18:34
  #545 (permalink)  
 
Join Date: Jul 2002
Location: UK
Posts: 3,093
Likes: 0
Received 0 Likes on 0 Posts
Originally Posted by Lyman
...and the pilot was given an a/c that was climbing with the NOSE showing low.
It depends on your definition of "climb" - it looks like the tail-end of a small pitch correction due to turbulence that would have ended maybe a second later than it did due to A/P disconnect - the aircraft's pitch attitude was increasing, but all the evidence suggests it was not about to climb significantly, it was simply trying to regain the assigned flight level and pitch attitude.

The pitch attitude was lower than it should have been at A/P disconnect, but it was a matter of 2 degrees - it briefly settled at 2 degrees nose-up after A/P disconnect. That kind of difference is barely noticeable on a modern PFD, let alone steam gauges - it was two points in the blue. It certainly does not require half back stick to correct, which is what the PF did.

He may have mentally been correcting for the bump they felt at disconnect, even though the autopilot was ahead of him - seeing the ADI pointing into the brown for a second might have been a shock. This is where Human Factors comes in. For those who aren't pilots but can drive, would you say you take more care when driving with your loved ones in the car than you do on your own? I know I do. See someone driving erratically or dangerously when you're on your own and it's tempting to just go past him. Do it with family in the car and I'm inclined to leave him to his own devices and back off because it's more than just me at stake. Professional airline pilots live with the knowledge that there are hundreds of people behind them who are depending on their skill - that's part of the job and why I respect them so much. If you're flying a jet that nevertheless has family on board? All of a sudden instincts start to cloud rationality and there's an emotional pull there too (which is why the Aeroflot crash with the children in the cockpit is to my mind the saddest aviation story I've ever heard).

His Pull was additive to an input made by AutoPilot("STALLSTALL")
No - the autopilot disconnected at 02:10:04.6 as mm43 has it. PF inputs begin at approximately 02:10:07 (by your own statement) and the first STALLSTALL does not sound until 02:10:10. There is no "additive input" to the autopilot from the sidesticks.

The problems encountered were not complicated, but the timing was.
It happened in a matter of seconds, no doubt - but the control inputs the PF made were completely disproportionate to what was required from the outset.

Last edited by DozyWannabe; 27th Nov 2011 at 18:58.
DozyWannabe is offline  
Old 27th Nov 2011, 19:06
  #546 (permalink)  
 
Join Date: Aug 2011
Location: Grassy Valley
Posts: 2,074
Likes: 0
Received 0 Likes on 0 Posts
At long last I believe we are closing in on what may be a lay solution. As such, it is of little value, save for some sort of truce. It also may be as close to a resolution as anyone gets, including BEA.

2:10:04.7. A/P quits. Here, the Nose is 4.5 degrees lower than what it should be. With MASTER CAUTION and CAVALRY CHARGE, the PF is brought up short, from whatever his status was, re: SA. A quick glance at PFD, if not already looking, shows NOSE LOW, and VS in descent, perhaps 500fpm. 2:10:07, and he inputs back stick with a ROLL LEFT.

The a/c has started to climb at a/p quit, and there is not a reason to doubt his input causes sufficient NOSE UP to trigger STALLSTALL. His next inputs are NOSE DOWN, so he 'gets it'. The rest depends on what he sees on his screen, feels, (The g is <1 until after his input), and decides. (Hears?).

If there is mitigating data, sufficient to establish his inputs were based on data not seen here, I propose there is a 50/50 we will never see it. That is the way of the world; life is harsh, and seldom "fair".

At long last the discussion is where it should have been from the outset.

With NOSE DOWN, I would submit that it takes a healthy Updraft to cause such an ascent as we see just prior to a/p loss. Was it sufficient to cause the accident? Was there ICE? or only variable TEMPS and a bumpy airmass?

It's been a real slice.
Lyman is offline  
Old 27th Nov 2011, 19:15
  #547 (permalink)  
 
Join Date: Jul 2002
Location: UK
Posts: 3,093
Likes: 0
Received 0 Likes on 0 Posts
No Lyman, the nose was on the way *back up* when the A/P disconnects. The PF may have *seen* the brief turbulence induced nose-low attitude and might have been trying to correct for it but there was no major updraft. The aircraft climbed because the PF's response was incorrect - in a perfect world he would have waited a few seconds to get a feel for what the aircraft was doing before trying to join in.

Hypothetically speaking - put yourself in the PF's position. You have not been trained in manual handling at altitude, most of the 2,000-odd hours on your flight log have been completely uneventful (and not to mention automated), you've been on holiday with your wife and you're now returning home with her in the passenger cabin. The Captain has nominated you as the relief pilot despite being the junior flight crew member on board - you're on top of the world. Suddenly you get an autopilot disconnect alarm followed by an alarming bump in which the ADI showed the nose dipping below the horizon for a second or two. What do you do?

We'd all like to think that we'd be calm and rational and behave as if this was any other flight - but all of a sudden one of your best days has turned into your worst nightmare.

Last edited by DozyWannabe; 27th Nov 2011 at 19:28.
DozyWannabe is offline  
Old 27th Nov 2011, 19:25
  #548 (permalink)  
 
Join Date: Jul 2009
Location: Not far from a big Lake
Age: 82
Posts: 1,454
Likes: 0
Received 0 Likes on 0 Posts
Originally Posted by Gums
my understanding from looking at the roll laws is the roll command works the same in "alternate" as in "normal". The only difference I see is the max roll angle allowed in "normal", not the command, but actual body-referenced roll angle. So what I see is what I was used to - let go of the stick and the roll command is zero deg/sec. Am I missing something?
Gums, What you wrote is correct for Alt1 Law but not for Alt2 Law. Alt2 Law is roll direct, i.e. aileron deflection proportional to stick deflection.

Organfreak, There is a reason each transducer on the stick has its own control mechanism, it is redundancy. That redundancy undoubtedly carries through in the software design of the "Bus" so that it can recognize a bad transducer, but don't ask me details because I don't know. At this point, all I know is that it would be extremely out of character not to cover that potential problem.

[quote Dozy]but the control inputs the PF made were completely disproportionate to what was required from the outset. [/quote]
Dozy, I agree with this summation

Originally Posted by TTex
or the PF made completely inappropriate control inputs, almost impossible control inputs. ...........OR, the PF was just hanging on
TTex, he was essentially reversing his roll input every two seconds. Maximum control input pretty well corresponds to maximum wing down. Just hanging on is probably as good a description as any. Over time the intensity of his inputs begins to decrease as he began to get the "touch".

PF was definitely driving a lateral oscillation by means of his excessive control inputs. This period where he is struggling with roll is when the aircraft became committed to a nose up attitude.

So why didn't PNF at least coach him back on altitude? All I can conclude is that both of their scans were in the toilet. "GO Down" , YGTBSM.

Originally Posted by rudderrat
In ALT Law, it has to be flown like any conventional aircraft in roll. The control input has to be held against any roll tendency until the load is trimmed out.
Thanks Rudderrat. I just wanted to be sure. No lateral pulses then in Alt2.

Last edited by Machinbird; 27th Nov 2011 at 22:54. Reason: remove redundant phrase
Machinbird is offline  
Old 27th Nov 2011, 21:03
  #549 (permalink)  
 
Join Date: Oct 2011
Location: Lower Skunk Cabbageland, WA
Age: 74
Posts: 354
Likes: 0
Received 0 Likes on 0 Posts
Machinbird wrote:
Organfreak, There is a reason each transducer on the stick has its own control mechanism, it is redundancy. That redundancy undoubtedly carries through in the software design of the "Bus" so that it can recognize a bad transducer, but don't ask me details because I don't know. At this point, all I know is that it would be extremely out of character not to cover that potential problem.
Thanks for addressing my question/speculation. I had assumed, apparently wrongly, that there was one TD for each direction of stick travel. Your contention that they are redundant rings true. Damn, and here I thought I'd come up with The Answer.
Organfreak is offline  
Old 27th Nov 2011, 23:56
  #550 (permalink)  
 
Join Date: Jun 2009
Location: somewhere
Posts: 451
Likes: 0
Received 0 Likes on 0 Posts
Artificial feel spring rod as in A330 SS

2 identical rods are installed in each SS.



Last edited by A33Zab; 28th Nov 2011 at 00:11.
A33Zab is offline  
Old 28th Nov 2011, 03:33
  #551 (permalink)  
 
Join Date: Jul 2009
Location: Not far from a big Lake
Age: 82
Posts: 1,454
Likes: 0
Received 0 Likes on 0 Posts
A33Zab, Thank you for posting the drawing of the spring rod (or cartridge).
It appears to be an elegant design in that the same springs work in compression during both extension and compression of the cartridge. Additionally, it appears that one of the springs is of a non-linear design and does not follow Hooke's Law.
Machinbird is offline  
Old 28th Nov 2011, 04:16
  #552 (permalink)  
 
Join Date: Jul 2009
Location: Not far from a big Lake
Age: 82
Posts: 1,454
Likes: 0
Received 0 Likes on 0 Posts
Originally Posted by Lyman
The PF inherited an airframe in phugoid oscillation.
Lyman, not really possible in an Airbus in Normal or either of the A330 Alternate Laws. The flight control system, by managing g, sets a stable flight path on the average.

A phugoid oscillation is a relatively slow oscillation at ~constant AOA, where potential and kinetic energy trade back and forth, but that isn't the way the 'Bus flight control system works.
Machinbird is offline  
Old 28th Nov 2011, 09:01
  #553 (permalink)  
 
Join Date: Jul 2009
Location: France - mostly
Age: 84
Posts: 1,682
Likes: 0
Received 0 Likes on 0 Posts
A330 phugoid

Back in june i posted the A330 phugoid that results when the system in normal law enters high AoA protection and maintains AoA equal to alpha-prot (as in the TC-JDN airprox incident).
HazelNuts39 is offline  
Old 28th Nov 2011, 10:29
  #554 (permalink)  
 
Join Date: Oct 2009
Location: berlin
Posts: 152
Likes: 0
Received 0 Likes on 0 Posts
Machinbird A33Zab, Thank you for posting the drawing of the spring rod (or cartridge).
It appears to be an elegant design in that the same springs work in compression during both extension and compression of the cartridge. Additionally, it appears that one of the springs is of a non-linear design and does not follow Hooke's Law.
I like this forum very much for such a design-picture, thank you A33Zab

but machinbird I am shure, there is a smal mistake in the drawing both springs will work in hook´s law, but the bigger(purple) spring IMO is a few mm shorter than shown, there is more space on the yellow side..... so first you press the weaker green spring and later you have to ad the hooks force of green and purple

so we have a redundant system with two catridges, well gums it can be a herring but if one of the green spring was broken then I am shure this is not easy to detect on the taxiway, the SS feels full ok for left/right/up+down and only the force for 1/3 pull is half of normal........ remember we search for the smalest possible defekt who can explains the action of the PF shown on the FDR!

If Grity's chart of control motions is accurate (and not overly interpolated/smoothed) then PF used an inappropriate control strategy.
the basis of my chart was the datas shown in interimreport3 s.29 (F/O pitch and roll command position) I used 2 valus/sec put both together for 2D and smoothed a splinline over it.... so it is little interpolated but not much,

it looks not as he has made pulses ore something like this in this time.....

much more interesting for the control strategy is if you overlay the sidstickway with the left/right asceleration, which was also left-right-left-right-left in this 10 sec with values +/- 0.1 g
grity is offline  
Old 28th Nov 2011, 14:58
  #555 (permalink)  
 
Join Date: Jul 2009
Location: Not far from a big Lake
Age: 82
Posts: 1,454
Likes: 0
Received 0 Likes on 0 Posts
Non-Linear springs

but machinbird I am shure, there is a smal mistake in the drawing both springs will work in hook´s law, but the bigger(purple) spring IMO is a few mm shorter than shown, there is more space on the yellow side..... so first you press the weaker green spring and later you have to ad the hooks force of green and purple
Grity, it is quite possible that you are correct and there is actually some spring freeplay that is not shown in the drawing. It is also possible to make a non-linear compression spring by wrapping the spring in such a way that the adjacent loops make contact in only a portion of the spring and thus effectively shorten the active portion of the spring. The guys who make springs can do it very consistently, even with 50 year old equipment.
Machinbird is offline  
Old 28th Nov 2011, 15:14
  #556 (permalink)  
 
Join Date: Jul 2009
Location: Not far from a big Lake
Age: 82
Posts: 1,454
Likes: 0
Received 0 Likes on 0 Posts
HN39,
Thank you for the reminder on how you cannot forget completely about the phugoid on an Airbus in Normal Law.

In the AF447 case, the phugoid did not apply since as soon as the AP dropped, it was not in Normal Law. And before the AP dropped it was not in high AOA protection.
Machinbird is offline  
Old 28th Nov 2011, 16:03
  #557 (permalink)  
 
Join Date: Jul 2009
Location: France - mostly
Age: 84
Posts: 1,682
Likes: 0
Received 0 Likes on 0 Posts
Originally Posted by Machinbird
In the AF447 case, the phugoid did not apply since as soon as the AP dropped, it was not in Normal Law. And before the AP dropped it was not in high AOA protection.
Agreed entirely. My intent was no more than to provide an illustration to your post #553.
HazelNuts39 is offline  
Old 30th Nov 2011, 19:59
  #558 (permalink)  
 
Join Date: Jul 2009
Location: France - mostly
Age: 84
Posts: 1,682
Likes: 0
Received 0 Likes on 0 Posts
Another mystery?

Does anyone know what the parameter 'TLU Position (DA)' shown on page 112 of IR no.3 (English) stands for? Since with 1300 recorded parameters it is inevitable that the report must necessarily involve a selection of the relevant parameters, showing this parameter means that somebody thought this one might be relevant.

On page 89 we read at 02:10:18: "The ‘TLU 1 availability’ and ‘TLU 2 availability’ parameters become NOT AVAILABLE" and the traces of these parameters are shown on page 107. Since this time roughly corresponds to the end of the initial 10 second speed monitoring process, I thought that TLU could stand for RTLU (Rudder Travel Limiter Unit). But what then is TLU position?
HazelNuts39 is offline  
Old 30th Nov 2011, 21:17
  #559 (permalink)  
 
Join Date: Jun 2009
Location: NNW of Antipodes
Age: 81
Posts: 1,330
Received 0 Likes on 0 Posts
From page 40 -
* the loss of the normal law between 2:05:10 and 2:10:06. The position of the rudder limiter, set by the rudder travel limiter unit (RTLU) is locked at 2:10:04.5 and this function is recorded between 2:10:17.5 and 2:10:18.5.
NOTE: The report uses "stuck" instead of "locked".

From page 89 -
* 2 h 10 min 18 - The ‘TLU 1 availability’ and ‘TLU 2 availability’ parameters become NOT AVAILABLE.

From page 112 -
* TLU 1 and TLU 2 parameters repeated, and at 2:10:08 the Side Slip from IRS becomes ACTIVE.

I suspect this is related to the Yaw Damper.

I've previously commented that the Rudder Pedal demands (up to RTLU limit) had the Yaw Damper output algebraically added.

Last edited by mm43; 30th Nov 2011 at 21:27.
mm43 is offline  
Old 30th Nov 2011, 22:08
  #560 (permalink)  
 
Join Date: Jul 2009
Location: France - mostly
Age: 84
Posts: 1,682
Likes: 0
Received 0 Likes on 0 Posts
The french text is quite clear:
La position du limiteur de débattement de la gouverne de direction (RTLU) se fige à 2 h 10 min 04,5 et l’indisponibilité de la fonction de limitation de débattement de la gouverne de direction est enregistrée entre 2 h 10 min 17,5 et 2 h 10 min 18,5.
What about the changes of 'TLU position' before 02:10:04.5?

Last edited by HazelNuts39; 30th Nov 2011 at 22:18.
HazelNuts39 is offline  

Thread Tools
Search this Thread

Contact Us - Archive - Advertising - Cookie Policy - Privacy Statement - Terms of Service

Copyright © 2024 MH Sub I, LLC dba Internet Brands. All rights reserved. Use of this site indicates your consent to the Terms of Use.