Go Back  PPRuNe Forums > Flight Deck Forums > Rumours & News
Reload this Page >

TAM A320 crash at Congonhas, Brazil

Rumours & News Reporting Points that may affect our jobs or lives as professional pilots. Also, items that may be of interest to professional pilots.

TAM A320 crash at Congonhas, Brazil

Old 2nd Oct 2007, 08:31
  #2621 (permalink)  
PBL
 
Join Date: Sep 2000
Location: Bielefeld, Germany
Posts: 955
RWA,

we've been through a lot of this already when discussing the FDR. It appears we have a couple new people, who maybe have not seen the FDR transcript. I will refer to
http://www.rvs.uni-bielefeld.de/publ...s/TAM3054.html
for a link to it, the CVR, and a Why-Because Analysis from the available data.

I'll restrict myself to the technical details.

Originally Posted by electricjetjock
The pilot flying left the thrust lever in the CLB detent NOT the aircraft or designers.
Originally Posted by RWA
That's maybe 80% probable, but not certain. All we know is that the FDR recorded the lever as being in CLB.
A more accurate statement would be that it is 99.9999999% probable, but not certain. I believe that mechanism is a 10^(-9) certification item and I have a query out on that to one of the four JAA certification teams.
Originally Posted by RWA
Try as I might, I've never been able to fully to understand AB's various 'modes'
The scare quotes are a good idea. "Mode" is a technical term that refers to AP and AT operations.

Here may be the point to respond to ChristiaanJ's comment that analogue and digital FBW are functionally the same. Yes, you can theoretically reproduce all digital logic with analogue logic or cables, pulleys and bell cranks, because the "logic" is basically design, which just happens to be more trivial to implement in digital electronics than with pulleys and bell cranks. However, digital FBW and digital autopilots have *modes* and I don't know any planes with analogue FBW that do. (That may not mean much, since I am not an expert on military HW, but that was one of the main design differences when I was working on the SIFT verification 23 years ago.) Modes are a design feature to handle complexity (both combinatorial and cognitive).

Originally Posted by RWA
I believe that another possibility is that the systems would have considered any setting above 'Idle' as being the next detent up (i.e. CLB).
That's too vague for me to be sure what you mean. Here are the relevant FCOM citations
Originally Posted by Power Plant: 1.70.30 p2
If the thrust lever is set between two detents, that FADEC selects the rating limit corresponding to the higher detent.
Just what the FADEC does with this rating limit is another issue. I think the relevant point here is what counts as "Idle" position for the ground spoilers when you are landing, and the answer is
Originally Posted by Flight Controls: 1.27.10 p13
For the ground spoiler logic, idle signifies: Thrust lever position < 4 or < 15 when below 10 ft [radio altitude]
Originally Posted by RWA
The possibilities are ...... or he may have knocked it forward while reaching over it to get hold of and lift the reverser latch on No. 1.
No. This is not a relevant possibility. The FDR shows no significant variation of any sort in TLA #2 from approach to end of recording.

Originally Posted by electricjetjock
He also did not immediately apply full manual braking when autobrake failed.
Originally Posted by RWA
I believe that according to the FDR, the pilot(s)' first step was to recycle the spoiler switch. That's pretty strong evidence that (having got reverse thrust) they thought that they had a spoiler malfunction.
No. First manual braking. Full braking started at 18:48:34 and pedal position was max at 18:48:36 (sampled once a second). Spoilers were unarmed between 18:48:35.5 and 18:48:36 and rearmed between 18:48:47.5 and 18:48:48 (sampled twice a second).

Originally Posted by RWA
my understanding is that applying the brakes at 140 knots would have been next to useless (and could very well have caused more problems than it solved).
We have been over the braking very thoroughly. Manual braking began at 120kts (they lost 10 kts in the first few seconds without) and I calculated an average deceleration of 1.26 m/s/s and at times it was up to 2 m/s/s. Far from being "next to useless", it would have stopped the plane had they had another X meters (estimates vary here), or would have reduced the GS to EMAS-entry speed with a couple hundred more.

PBL
PBL is offline  
Old 2nd Oct 2007, 08:35
  #2622 (permalink)  

Mach 3
 
Join Date: Aug 1998
Location: Stratosphere
Posts: 622
That's the least understandable feature of all. The PF was a senior captain with over 10,000 hours, and a training pilot at that.
Just a guess, but I doubt whether these kind of errors (not closing thrust levers, or, for example, not setting altimeters correctly) correlate well with experience.

In fact, in our particular airline, I am led to believe that a significant proportion of our level busts involve training flights i.e., with at least one highly experienced crew member.

I also suspect that their analysis will not yield much helpful insight.

IMHO, the focus has to be how we improve the collective system propensity to fail safe.
SR71 is offline  
Old 2nd Oct 2007, 08:35
  #2623 (permalink)  
Per Ardua ad Astraeus
 
Join Date: Mar 2000
Location: UK
Posts: 18,582
DW - the Cali accident involved speedbrake and not ground spoilers, and I am not aware of any system change to prevent that.

(Re Post #2541): Assuming they 'went around' from the runway, what would have been their chances of correctly diagnosing the problem? After all, they would need to plan a landing at a diversion with no g/spoilers. That could be a problem.

Would closing down No2 clear the fault?
BOAC is offline  
Old 2nd Oct 2007, 08:39
  #2624 (permalink)  
 
Join Date: Nov 2001
Location: Somewhere in the Tropics UTC+7 to 9
Posts: 450
Forgive me if I have missed it in previous explanations - IF there had been those unlikely multiple failures in the sensors and with No2 T/L actually at IDLE but indicating CLB, what could the crew do to deploy the ground spoilers?
Have they touched down and committed to the reverse or not?
If not? Go around first and then think... *easier said than done!*
If yes? It's a difficult call, whatever answer is given is with the benefit of hindsight. Here we go into the can of worms unfortunately.

If one checks spoilers before reverser, it can buy you time to make the right decision, or delay action that can lead to your death...
A. Land, spoilers nada... decide Go/Stop. Or
B. Land, spoilers nada... decide... hey wait, what's with engine #2? GO!

Or... the craziest suggestion...
C. Land and select reverse... spoilers nada... then what? Flaps to 2 or 3 and speedbrakes... *better than nothing*

*Note: Spoilers None, not Braking None!*

Now is the TL sensor failure a total failure or an "unknown" failure. The safest would be TOGA and see, or max rev and see. but again, it's that select reverser immediately that's causing me the headaches... as it leaves them with little or no choice!

But to prevent all this, what's there in the system that would warn the crew of a TL position sensor failure?

Just out of curiousity, how risky to normal ops would taking out the speedbrake inhibit on config full on the bus? (You CAN use speedbrakes on flap40 on the 737 if you're a cowboy... but it's pretty scary and definitely not recommended unless you need to rely on your "superior skills")...
On this it's only a switch not a cable link... and therefore even if you pull it, the positive thrust would push it back in!

Excuse me if none of this makes sense...

PK-KAR
PK-KAR is offline  
Old 2nd Oct 2007, 08:42
  #2625 (permalink)  

Sun worshipper
 
Join Date: Dec 2001
Location: Paris
Posts: 494
Tyro,
For any warning the priority is a/c flight path - e.g on an EFTO it is not a case of doing absolutely nothing until 400' agl... you raise the gear, trim, engage AP etc. I don't see moving a T/L to Idle as any different in this situation - PF should have done it anyway.
Agreed !
That's covered by the *Normal procedures applied* criterium.

BOAC,
I agree with your last post, though I'm not holding my breath waiting for the "oriental friend". Impossible scenario barring a major re-wiring and re-programming specifically done for this specific series of failure.
As for your question, there is no manual way to extend the ground spoilers in any configuration and no manual way to extend the spoilers in config *Full* and that's valid, whatever extra likely or unlikely failures in the sensors you might think of.
Lemurian is offline  
Old 2nd Oct 2007, 08:43
  #2626 (permalink)  
 
Join Date: Jul 2002
Location: UK
Posts: 3,182
Closing down No.2 would, I believe (will need to check), clear the 'fault' (still not sure about this term, but will go with it for now...). I suspect that given the option to go around in the weather they were in, believing that there was a problem with the braking system they'd likely have diverted to an airport with a longer runway.
DozyWannabe is offline  
Old 2nd Oct 2007, 08:46
  #2627 (permalink)  
PBL
 
Join Date: Sep 2000
Location: Bielefeld, Germany
Posts: 955
Originally Posted by PK-KAR
what's there in the system that would warn the crew of a TL position sensor failure?
The short answer is: quite a lot. It is a highly-redundant system.

PBL
PBL is offline  
Old 2nd Oct 2007, 08:49
  #2628 (permalink)  
 
Join Date: Nov 2001
Location: Somewhere in the Tropics UTC+7 to 9
Posts: 450
But digital FBW, like digital autopilots, has modes and I don't know any planes with analogue FBW that do.
Try the Tu204... If Quadruplex FBW fails, it reverts to a 3 channel analogue FBW.

suspect that given the option to go around in the weather they were in, believing that there was a problem with the braking system they'd likely have diverted to an airport with a longer runway.
And saved the day! *Hindsight self-warning!*

The short answer is: quite a lot. It is a highly-redundant system.
Cheers, Where can I find it in the FCOM? I need to read it to better understand it.

So, if it was a failure, it should have had a warning, would this warning be inhibited on the final stages of the approach? *Just so that it's clear*

PK-KAR
PK-KAR is offline  
Old 2nd Oct 2007, 09:30
  #2629 (permalink)  
PBL
 
Join Date: Sep 2000
Location: Bielefeld, Germany
Posts: 955
Originally Posted by PBL
But digital FBW, like digital autopilots, has modes and I don't know any planes with analogue FBW that do.
Originally Posted by PK-KAR
Try the Tu204... If Quadruplex FBW fails, it reverts to a 3 channel analogue FBW.
I didn't know the Tu204 had an analogue FBW fall-back, thanks. But if it is a fall-back system, I doubt it has modes. Fall-back systems are designed to be simple, not complex.

Concerning reliability of the thrust lever system,
Originally Posted by PBL
The short answer is: quite a lot. It is a highly-redundant system.
Originally Posted by PK-KAR
Where can I find it in the FCOM? I need to read it to better understand it.
The ECAM warnings about power plant items are, of course, in the power plant section, 1.70. Controls and Indicators are 1.70.90; Warnings and Cautions are 1.70.90 p13, p14, 15.
FCOM itself is at http://www.smartcockpit.com/pdf/plan.../systems/0022/

Can you give me an address where I can send your bill?

The details of the design of the thrust lever quadrant and the thrust command system in general are not FCOM items. You might look at the note [email protected]:21.12 on the TL mechanism, and pictures from [email protected]:22.25 for the mechanism.

bsieker wrote something on TL-associated ECAM messages: [email protected]:11.56

Originally Posted by PK-KAR
So, if it was a failure, it should have had a warning, would this warning be inhibited on the final stages of the approach? *Just so that it's clear*
Now it's your turn to research it. Let us know.

PBL
PBL is offline  
Old 2nd Oct 2007, 10:03
  #2630 (permalink)  
 
Join Date: Sep 2007
Location: Portugal
Posts: 39
PK-KAR
Now is the TL sensor failure a total failure or an "unknown" failure. The safest would be TOGA and see, or max rev and see. but again, it's that select reverser immediately that's causing me the headaches... as it leaves them with little or no choice!
The split second decision I know I could make facing a failure, would be to immediately shut down eng#2 (fuel lever off), as it would solve part of the problem and still leave me with good enough hydraulics. And I probbably would do it without PNF confirmation, which although in the ground, would be a bit unorthodox concerning recommended task sharing. The decision I know I would't make is to go-around, as getting airborne with a dubious flt control problem (although not in that chapter, throttles ARE flt controls as well in my picture) would be a bad decision. All this would take as much time as reading the first 6 words of this paragraph, and assuming there was a detected failure condition, wich we don't know if it's the case in this accident.

I think that in this case, we are looking either at a very basic human failure, or a system malfunction, that's still to be unveiled. If there was the human failure, maybe AI can develop a way to mitigate the effect of this kind of events, but that's not very meaningfull to me, as grosse errors will allways occur, unfortunately. If there was a system malfunction, then AI could have a major problem, because it would probbably show a weak point in the phillosophy of the design, and that might be unmaneageble, in such a widespread aircraft generation. One thing is to work on a new design (you know the A380 has it's sidesticks interconnected and moving together?), tottaly different is to work on a flying model.

electricjetjock
Yes we are all hoping an accident like this will not happen again but the main causes at the moment appear to be pilot induced so we should all be saying " I HAVE LEARNT ABOUT FLYING FROM THIS" I do not want to make those mistakes!!!
Absolutely right! But this applies to all HF related accidents. Unfortunately, new catastrophic situations will allways occur, and we can only learn from those that allready happened. That's why, if this is a human error primary cause accident, it is not much different from other accidents, because failing to retard 1 throttle at touchdown is not AI responsability. If there is a system failure, then we will have a serious problem.

DozyWannabe
Oh yes it did. If the two computers linked to the yokes had different amounts of pressure set to disengage the autopilot, the autopilot would read as disengaged on the side with the lower pressure setting, but would read still engaged on the other side. This led to the world's first widebody disaster, Eastern Airlines EAL401 over the Everglades in Florida.
That's one way to see that accident. Another is that there were 5 guys in that cockpit, all looking at tiny inop green light, and nobody was flying the aircraft at a 2000 feet night visual circuit pattern. The cockpit discipline that comes built in on Airbus types, is for me the major AI achievement. The L1011 Everglades accident was primarily a bad CRM accident. Of course there was also the AP disconnection force issue. But I said that the Tristar had few bugs, not any bugs at all

Cheers
3Ten is offline  
Old 2nd Oct 2007, 10:03
  #2631 (permalink)  
PBL
 
Join Date: Sep 2000
Location: Bielefeld, Germany
Posts: 955
Originally Posted by SR71
I doubt whether these kind of errors (not closing thrust levers, or, for example, not setting altimeters correctly) correlate well with experience..........
I also suspect that their analysis will not yield much helpful insight.
I suspect that also.

Some evidence may be gleaned from looking over the 2,500 posts on this thread, many of which are attempts at amateur psychology. If there is insight to be gained, I suspect it would have appeared here by now.

Originally Posted by SR71
IMHO, the focus has to be how we improve the collective system propensity to fail safe.
I agree.

Good starts have been made publically by TAM, in prohibiting landings at CGH in the wet, or without 2 TR OP.
And by the government/airport authority in planning to improve the overrun facilities at the end of 35L.

And, on might think, by Airbus, the JAA, and the FAA, in reminding everyone to reduce thrust levers to idle on each and every landing (if we believe that reminding people of such basic items is efficacious. We have reason to doubt it.)

PBL
PBL is offline  
Old 2nd Oct 2007, 10:23
  #2632 (permalink)  
 
Join Date: May 2000
Location: Glorious West Sussex
Age: 72
Posts: 1,019
Question for 3Ten and Lemurian..
Just imagine that your next A320 landing is with one T/R INOP, on a wet runway, at night. As you select reverse the CRC sounds, master warning lights flash, and ECAM Level 3 warning

ENG ENG 1 THR LEVER ABOVE IDLE appears in red.
I am interested to know what you will do...
a) Stop the aircraft using manual braking, set the parking brake, and say "ECAM actions"
b) Put your hand on No 1 T/L and check it is retarded to Idle or Reverse
c) Something else

Tyro

Last edited by TyroPicard; 2nd Oct 2007 at 10:39.
TyroPicard is offline  
Old 2nd Oct 2007, 10:56
  #2633 (permalink)  
 
Join Date: Sep 2007
Location: Portugal
Posts: 39
Answer b) probbably.

But that's beside the point, because retarding both throttle levers is such a basic principle, that if you'd want to provide warnings to avoid such basic mistakes, you would have to come up with dozens of new warnings.

CRC >> PITCH EXCEEDING 10 DEGREES! (for tail strike avoidance, for instance)

CRC >> VERTICAL SPEED EXCEEDING 720FPM (for hard landing avoidance)

Where does that leave the pilots??

And that would corrupt the ECAM warning phillosophy, ECAM warnings are to be used as described, if one starts to deviate in design, confusion is to arise. There are other type of warnings that are oriented for immediate actions, like the WINDSHEAR warning on the PFD, or the AUTOLAND LIGHT, but these aren't ECAM warnings.
3Ten is offline  
Old 2nd Oct 2007, 11:41
  #2634 (permalink)  
 
Join Date: May 2000
Location: Glorious West Sussex
Age: 72
Posts: 1,019
Thanks 3Ten
There are other ECAM warnings that you react to immediately - Engine Fire below V1 for example. I know that you don't perform ECAM actions immediately, but I think the same principle is at work.
And we already have the Retard warning to remind us of something we do every flight...
Tyro
TyroPicard is offline  
Old 2nd Oct 2007, 11:42
  #2635 (permalink)  
 
Join Date: Nov 2001
Location: Somewhere in the Tropics UTC+7 to 9
Posts: 450
Gdangit PBL!
A Bill? Send it to Kartika Airlines HQ... they owe me some cash I'm sure they can offset your bill. LOL!
I asked coz I'm not 320 rated but love reading their manuals. After previously scouring 1.70, and being battle fatigued in another forum, I just needed to get someone to tell me its somewhere in 1.70 rather than me reading the wrong chapter and wonder where the hours disappeared to!

Will do that reading.

3Ten,
Thanks for your view of the "hypothetical decision" and its assumptions.

Where does that leave the pilots??
On a leash next to the watchdog? HEAVEN FORBID!

PK-KAR
PK-KAR is offline  
Old 2nd Oct 2007, 12:35
  #2636 (permalink)  
 
Join Date: Aug 2007
Location: Brazil
Age: 67
Posts: 131
Fuel Flow behavior

Need help understanding this.

On the FDR we can see that FADEC "sends" more fuel only to engine # 2, even before any TL is moved.
Why is that?

On the 2nd previous landing we don't see that "behavior".
Why is that?

Why there is no N1 recordings (on the FDR)?

I can bet that if we could cross fuel flow info with N1 readings something unusual MAYBE would show...

Rob
Rob21 is offline  
Old 2nd Oct 2007, 13:11
  #2637 (permalink)  
 
Join Date: Jul 2007
Location: Germany
Posts: 533
Rob21,

regarding fuel flow:

Fuel flow is only sampled every 4 seconds, all we can say is that it rises between 18:48:21.5 and 18:48:25.5. For all we know it might have risen only at 18:48:25.

We've been over sampling before, please read a few hundred posts back


Bernd
bsieker is offline  
Old 2nd Oct 2007, 13:28
  #2638 (permalink)  

Sun worshipper
 
Join Date: Dec 2001
Location: Paris
Posts: 494
Tyro,
There are other ECAM warnings that you react to immediately - Engine Fire below V1 for example. I know that you don't perform ECAM actions immediately, but I think the same principle is at work.
And we already have the Retard warning to remind us of something we do every flight...
Interesting discussion.
Back to the Bible : "Any alarm below V1 calls for a rejected takeoff procedure" (specifics as per SOPs ).
One doesn't have to read anything, one has to react first by initiating an RTO.
The actions are done - once again - when one has achieved a stabilised, controlled path.
The very same principle applies to a catIII autoland : *Alarm !-GoAround !, Flaps 3...etc...*, we'll see the specifics later !
The main objection I have with this new warning - believe me, I'm not alone -is that it is against all the failure management that we have been trained for.It is an exception in the rule and as such, leads to confusion of interpretation, especially during a phase where we don't need any.
As a matter of fact, you seem to agree with me when you write :
It will need suitable training to highlight the fact that it exists - IMHO by instructor demonstration as you should not ask a trainee to make such a mistake on landing
(my stress).
As for the *Retard* call, don't you think it has become a "corruption" of the original intent of the designer :
Back to the training manual " Flare shoul be initiated at 30 ft..." If it's done, one would not hear the aural. Of course, some clever souls have found that it is perfectly timed to flare / reduce thrust with this *CUE* in order to achieve a soft landing...Problem is : when I hear it on a 321, I know that this landing is not likely to be a soft one.
Regards

Last edited by Lemurian; 2nd Oct 2007 at 13:32. Reason: spelling
Lemurian is offline  
Old 2nd Oct 2007, 13:38
  #2639 (permalink)  
 
Join Date: Jul 2000
Location: Germany
Posts: 90
Concerning the issue, that the ECAM-Warning 'ENG ENG 1 THR LEVER ABOVE IDLE' would not be feasible, because it can't be dealt with 'procedurally' during the landing roll. I have to disagree.

When the A320, it's ECAM/FWC and the basic ECAM-Handling were designed, a couple of warnings/procedures dind'nt yet exist, either because AI didn't know or admit they were necessary. These warnings/procedures found their way into the books 'the hard way'. Unfortunately some of them didnt't fit into the procedural model of the FCOM-preface. Of course they were introduced anyway.

A textbook example is the 'Loss of Braking' procedure. Consistent of 100% 'Memory Actions' which have to be performed by heart and immediately. Not even an ECAM-Warning to trigger the actions, nevertheless the procedure works quite well in the SIM.

For those not type-rated, he we go:

LOSS OF BRAKING

x If autobrake is selected
__Brake Pedals.........PRESS PF
x If no braking available:
__Reverser...............MAX PF
__Brake Pedals.......RELEASE B
__A/SKID&N/W STRG........OFF B
__Brake Pedals.........PRESS PF
__MAX BRK PRESS......1000PSI B
x If still no braking:
__Parking Brake.....SHORT PF
__AND SUCCESIVE APPLICATION

So I guess another ECAM-Warning (btw: with only one action!) wouldn't be that bad and might have saved the day for the TAM-Crew.

Regards, MAX
Mx Reverse is offline  
Old 2nd Oct 2007, 13:39
  #2640 (permalink)  
 
Join Date: Jul 2007
Location: Germany
Posts: 533
Originally Posted by PK-KAR
But to prevent all this, what's there in the system that would warn the crew of a TL position sensor failure?
there are ECAM warnings, but what is more important, if we are talking about this accident, in case of a thrust lever angle sensor problem, the FADEC commands idle power for the affected engine, if the aircraft is on the ground and not in a take-off run (i. e. thrust before failure was above CLIMB).


Bernd
bsieker is offline  

Thread Tools
Search this Thread

Contact Us - Archive - Advertising - Cookie Policy - Privacy Statement - Terms of Service - Do Not Sell My Personal Information

Copyright 2018 MH Sub I, LLC dba Internet Brands. All rights reserved. Use of this site indicates your consent to the Terms of Use.