TAM A320 crash at Congonhas, Brazil
Join Date: Aug 2005
Location: Cloudbase
Posts: 149
Likes: 0
Received 0 Likes
on
0 Posts
ok, I'll leave my quiet corner, even though I'm not a pro, but there are a few points I'd like to make:
#1: looking at how long it took this list to come up with a precise description of what the AB systems do with regards to breaking/spoilers/TR when the TL are in a non-idle position, the pilots flying this airplane had just a few seconds to analyze the situation and come up with the right answer. (that is, assuming that one of the TL was really not only not in the idle detent but so far forward that it prevented breaking action + 22.5°?)
#2: on moving or non-moving thrust levers: on the Boeing, the TLs are a means of bi-directional communication. If the pilot has on hand on the levers, he'll get direct tactile feedback of what the plane's current thrust setting is. That's an additional, very direct channel of communication besides the gauges or displays. AB has eliminated this direct feedback channel. So the pilot has to obtain thrust information through other sensory input. Given the fact that today's cockpits are rather quiet, probably only visual by reading and understanding the data on the displays. In a stressful situation like this one no doubt has been, this will take infinitely longer than registering the tactile feedback from a moving lever.
#3 assumptions of situations: many of the rules by which aircraft systems work do assume certain situations. TLs idle, main landing gear depressed, wheel rotation means we have landed and wheel breaks and spoilers can be activated/deployed. If one or more of the above inputs are not present, the situation becomes ambiguous. How can the system assess the situation when on TL is in full reverse and the other in CLB? How can the system understand what the PF's intentions are with this?
The way I understand the diagram provided further up, the logic is quite simple: all the factors are ANDed together and if one is not "TRUE", the whole situation is not "TRUE"... In this case: one TL out of the area that is considered "TRUE" for landing, you're clearly not landing.
This binary logic might be overly simple for humans who operate, and more importantly err in a continuum of analog values.
So what I'm saying is this:
even though the original error might have been a human factor (and no, I'm not stating this as a fact, just as a hypothesis), the chain of events unfolding seems to be one of miscommunication between the airplane's systems and the pilots operating it.
If it should indeed be concluded that this accident was due to human error, then I do hope that system/human communication and system complexity are listed in the "contributing factors" section.
pj
#1: looking at how long it took this list to come up with a precise description of what the AB systems do with regards to breaking/spoilers/TR when the TL are in a non-idle position, the pilots flying this airplane had just a few seconds to analyze the situation and come up with the right answer. (that is, assuming that one of the TL was really not only not in the idle detent but so far forward that it prevented breaking action + 22.5°?)
#2: on moving or non-moving thrust levers: on the Boeing, the TLs are a means of bi-directional communication. If the pilot has on hand on the levers, he'll get direct tactile feedback of what the plane's current thrust setting is. That's an additional, very direct channel of communication besides the gauges or displays. AB has eliminated this direct feedback channel. So the pilot has to obtain thrust information through other sensory input. Given the fact that today's cockpits are rather quiet, probably only visual by reading and understanding the data on the displays. In a stressful situation like this one no doubt has been, this will take infinitely longer than registering the tactile feedback from a moving lever.
#3 assumptions of situations: many of the rules by which aircraft systems work do assume certain situations. TLs idle, main landing gear depressed, wheel rotation means we have landed and wheel breaks and spoilers can be activated/deployed. If one or more of the above inputs are not present, the situation becomes ambiguous. How can the system assess the situation when on TL is in full reverse and the other in CLB? How can the system understand what the PF's intentions are with this?
The way I understand the diagram provided further up, the logic is quite simple: all the factors are ANDed together and if one is not "TRUE", the whole situation is not "TRUE"... In this case: one TL out of the area that is considered "TRUE" for landing, you're clearly not landing.
This binary logic might be overly simple for humans who operate, and more importantly err in a continuum of analog values.
So what I'm saying is this:
even though the original error might have been a human factor (and no, I'm not stating this as a fact, just as a hypothesis), the chain of events unfolding seems to be one of miscommunication between the airplane's systems and the pilots operating it.
If it should indeed be concluded that this accident was due to human error, then I do hope that system/human communication and system complexity are listed in the "contributing factors" section.
pj
on moving or non-moving thrust levers: on the Boeing, the TLs are a means of bi-directional communication. If the pilot has on hand on the levers, he'll get direct tactile feedback of what the plane's current thrust setting is. That's an additional, very direct channel of communication besides the gauges or displays. AB has eliminated this direct feedback channel. So the pilot has to obtain thrust information through other sensory input
Per Ardua ad Astraeus
Join Date: Mar 2000
Location: UK
Posts: 18,579
Likes: 0
Received 0 Likes
on
0 Posts
Kit d'Rection KG - there are several points here. The accident was caused by something that happened on the runway, not after leaving it. Only the FDR (and CVR) will determine this. Any witness marks will only show the position of various controls at impact and not where they were before touchdown and during the landing roll - agreed? What happened to the controls between the a/c leaving the paved surface and impacting the building is of little relevance. Who knows what inputs were made in those appalling few seconds? Hence the use of the words "relatively little value" and "Only where there is no FDR reading of throttle position on the runway will witness marks etc be relevant." which you no doubt saw - and it obviously is with reference to the throttle lever positions.
The intention was not to address professional accident investigators (like you?) but to try and dissuade the others from pointing out the full left rudder deflection on the remnants of the fin etc, and now where the lever was (possibly?).
I agree that accident investigation is not for 'pilots' per se, but I actually do have experience of such (No FDR). What are your qualifications, and do you have any useful input on the results so far?
Regarding the 'mod' bit, the intention at all times is to try and keep the thread focussed, and threads like these easily run away as I can see from the number of posts here deleted by the seniors.
The intention was not to address professional accident investigators (like you?) but to try and dissuade the others from pointing out the full left rudder deflection on the remnants of the fin etc, and now where the lever was (possibly?).
I agree that accident investigation is not for 'pilots' per se, but I actually do have experience of such (No FDR). What are your qualifications, and do you have any useful input on the results so far?
Regarding the 'mod' bit, the intention at all times is to try and keep the thread focussed, and threads like these easily run away as I can see from the number of posts here deleted by the seniors.
Join Date: Jul 2007
Location: Brazil
Age: 52
Posts: 21
Likes: 0
Received 0 Likes
on
0 Posts
Was this the first leg of the flight deck crew on that particular airplane on that day?
Join Date: May 2007
Location: Both Emispheres
Posts: 226
Likes: 0
Received 0 Likes
on
0 Posts
Is not with the intention of stirring polemics that I reported a media source about "TL position in the wreckage". It is quite possible that it is just an unsubstantiated rumor, and not being an expert myself, I can also concede that whatever "witness mark" of this type is irrelevant for determining causes of the accident.
This said, it is fact is that now, one week after the accident, there are leaks about the investigation and are going in the direction of thrust asymmetry. Another of these rumors are directly sourced from preliminary reading of FDR as the link from a Paulista newspaper:
http://www.estadao.com.br/estadaodeh...imp26535,0.php
A análise preliminar da caixa-preta de dados do vôo 3054 da TAM indica que houve erro dos pilotos no uso dos manetes do Airbus A320.
Now, I'm the first to agree that before the release of the official report, all kind of rumors can be dismissed as media speculations with possibly a vested interest in blaming pilots.
At the same a serious discussion exists for the very purpose of dissecting any kind of objective information in the search for a conclusion that can possibly anticipate, dissent or present a broader and independent opinion from the official report. And I don't think that dismissing "facts" and lines of thinking just because they come from media helps in that.
This said, it is fact is that now, one week after the accident, there are leaks about the investigation and are going in the direction of thrust asymmetry. Another of these rumors are directly sourced from preliminary reading of FDR as the link from a Paulista newspaper:
http://www.estadao.com.br/estadaodeh...imp26535,0.php
A análise preliminar da caixa-preta de dados do vôo 3054 da TAM indica que houve erro dos pilotos no uso dos manetes do Airbus A320.
Now, I'm the first to agree that before the release of the official report, all kind of rumors can be dismissed as media speculations with possibly a vested interest in blaming pilots.
At the same a serious discussion exists for the very purpose of dissecting any kind of objective information in the search for a conclusion that can possibly anticipate, dissent or present a broader and independent opinion from the official report. And I don't think that dismissing "facts" and lines of thinking just because they come from media helps in that.
**********************************************
Quote:
The 'position' of cockpit controls is of relatively little value when there is an FDR trace. Several of us have cautioned about 'running away' with these pieces of information. Those who have seen cockpits after relatively high-speed impacts will know what I mean. Only where there is no FDR reading of throttle position on the runway will witness marks etc be relevant.
***************************************************
Words clearly written by someone who's not an investigator... All evidence is of value, even where there is a full FDR trace, for a number of reasons: correct selection in flight deck, control fails to respond (ergo pilot did not make selection); sensor attached to control or linkage, not surface (ergo surface position not, in fact, recorded), etc.
All witness marks are vital to a trained investigator - correctly interpreted they tell a story which must be heard...
Just as non-pilots are being cautioned to stay away from this discussion, may I suggest that flying an aeroplane doesn't qualify a pilot as an investigator?
(PS, as you're a Mod, I'm surprised at your post...)
The investigators have many things to look at and the appropiateness of each of the gathered facts must also be judged in each individual accident
Join Date: Jul 2005
Location: btw SAMAR and TOSPA
Posts: 566
Likes: 0
Received 0 Likes
on
0 Posts
The aircraft already landed 2 times in Congonhas before the accident, with some other crew. The new crew took over the plane, and flew to Porto Alegre and back.
Join Date: Jul 2007
Location: Brazil
Age: 52
Posts: 21
Likes: 0
Received 0 Likes
on
0 Posts
lomapaseo,
Yes. According to a TAM press release, the TR was disabled four days before the accident.
" So the flight deck crew landed with the reverser inop at Porto Alegre shortly before their return to CGH and did well"
are we sure?
are we sure?
Join Date: Jul 2007
Location: You get one guess!
Posts: 18
Likes: 0
Received 0 Likes
on
0 Posts
START OF FACTUAL STATEMENTS
The investigators will not find much use in looking at the cockpit controls as there is no recognisable cockpit. Edited note: at least not reported as being found and take away for investigation
The aircraft was flown by the crew on the previous sector. The Number 2 thrust reverser had been locked out for a period of time.
If you look at some of the photographs circulating in Brazil, you will find three brake packs ON. You might see one brake pack OFF.
The designated captain has been identified.
THIS IS THE END OF THE FACTUAL STATEMENT.
THIS IS THE START OF THE SPECULATIVE STATEMENT
If you look carefully at the photographs, you might see why the local talk is of post impact damage leading to brake release.
END
Comments:
Please separate out interesting talk of Airbus related technical failures from the facts surrounding this accident. If it is speculation, please say so.
According to an investigator friend in PM conversations relating to this accident, and he is ok to quote this..
DFDR data is not perfect. There is a lot of misinterpretation by "beginners" relating to (1) forgetting to correlate the data and ensure that it was wired up the right way round (2) forgetting at exactly which point in the data frame second the data were recorded and thus ending up comparing not one second intervals across data streams but 1.8 and 0.2 second intervals (3) forgetting that DFDR time needs to be correlated with "real" time as sometimes the clocks run fast or slow in some older recorders. The investigator does not know what type was fitted to this aircraft to know if (3) applies in this case.
Our Man.
The investigators will not find much use in looking at the cockpit controls as there is no recognisable cockpit. Edited note: at least not reported as being found and take away for investigation
The aircraft was flown by the crew on the previous sector. The Number 2 thrust reverser had been locked out for a period of time.
If you look at some of the photographs circulating in Brazil, you will find three brake packs ON. You might see one brake pack OFF.
The designated captain has been identified.
THIS IS THE END OF THE FACTUAL STATEMENT.
THIS IS THE START OF THE SPECULATIVE STATEMENT
If you look carefully at the photographs, you might see why the local talk is of post impact damage leading to brake release.
END
Comments:
Please separate out interesting talk of Airbus related technical failures from the facts surrounding this accident. If it is speculation, please say so.
According to an investigator friend in PM conversations relating to this accident, and he is ok to quote this..
DFDR data is not perfect. There is a lot of misinterpretation by "beginners" relating to (1) forgetting to correlate the data and ensure that it was wired up the right way round (2) forgetting at exactly which point in the data frame second the data were recorded and thus ending up comparing not one second intervals across data streams but 1.8 and 0.2 second intervals (3) forgetting that DFDR time needs to be correlated with "real" time as sometimes the clocks run fast or slow in some older recorders. The investigator does not know what type was fitted to this aircraft to know if (3) applies in this case.
Our Man.
Join Date: Nov 2006
Location: usa
Age: 79
Posts: 24
Likes: 0
Received 0 Likes
on
0 Posts
The statements on the flight to Porto Alegre and back to Congonhas by the same flight crew might carry more weight if we could confirm that the landing in Porto Alegre was indeed a common and uneventful one, and that the PF for the Alegre landing was the same PF for the landing at Congonhas.
Join Date: Jul 2007
Location: You get one guess!
Posts: 18
Likes: 0
Received 0 Likes
on
0 Posts
Reference post 621 about forward thrust.
Please remember that the "effective" runway for rollout has a 1% downslope on it. You can work this out from the diagram on the reverse of the SBSP Aerodrome Chart in the AIP. The resurfacing work probably did not alter the elevation profile of the runway significantly, so the figure should still be valid. Do not take the difference in threshold elevations but read the statement carefully.
I wish to make no factual statement about the difference that this would make, but it is a contribution to your vector analysis.
Please remember that the "effective" runway for rollout has a 1% downslope on it. You can work this out from the diagram on the reverse of the SBSP Aerodrome Chart in the AIP. The resurfacing work probably did not alter the elevation profile of the runway significantly, so the figure should still be valid. Do not take the difference in threshold elevations but read the statement carefully.
I wish to make no factual statement about the difference that this would make, but it is a contribution to your vector analysis.
Join Date: Jul 2005
Location: btw SAMAR and TOSPA
Posts: 566
Likes: 0
Received 0 Likes
on
0 Posts
and that the PF for the Alegre landing was the same PF for the landing at Congonhas.
Just my thoughts
Join Date: Jul 2007
Location: Paris, France
Posts: 14
Likes: 0
Received 0 Likes
on
0 Posts
2) I also suspect that Veja has polled PPRune and elsewhere and made an 'intelligent' guess at this.
If is that so, the brazilian leading news magazine with 1,2 million issues/week, the world’s fourth in circulation, gave credit to PPrune. As a moderator, you know that it should be a reason of proud for the site and it’s members. But the << 'intelligent' guess>> could also came from the magazines’ sources, which looks like that you does not know the serious of its reputation. Anyway, your commentary is not really about aviation, but journalistic investigation.
Last edited by A22; 29th Jul 2007 at 15:19.
Join Date: Nov 2005
Location: UAE
Age: 45
Posts: 113
Likes: 0
Received 0 Likes
on
0 Posts
Guys,
Having experience on both 737 and A320, and trying to "translate" this to the Boeing guys, I can say that what happened on this accident (speculation so far) is something like disengaging the Auto-throttle on final approach and not moving the TL to idle at flare and landing roll out..
Simple and bizarre like that, so that is why I do not believe that this could happen (still think that something else played a significant role on this accident).
Believe me, the Bus is a very easy airplane. We call it a "no sweat" airplane
(when you get to the hotel after several sectors, you are still smelling good!).
Having experience on both 737 and A320, and trying to "translate" this to the Boeing guys, I can say that what happened on this accident (speculation so far) is something like disengaging the Auto-throttle on final approach and not moving the TL to idle at flare and landing roll out..
Simple and bizarre like that, so that is why I do not believe that this could happen (still think that something else played a significant role on this accident). Believe me, the Bus is a very easy airplane. We call it a "no sweat" airplane
(when you get to the hotel after several sectors, you are still smelling good!).
Join Date: Jul 2007
Location: Here
Posts: 52
Likes: 0
Received 0 Likes
on
0 Posts
@ lomapaseo
" So the flight deck crew landed with the reverser inop at Porto Alegre shortly before their return to CGH and did well"
are we sure?
are we sure?
PR-MBK touched down at 4:40pm and remained 36 minutes on the ground.
I believe I read the PF remained the same for flight 3054, but I can't find anymore the source to confirm that, sorry.
Join Date: Feb 2005
Location: flyover country USA
Age: 82
Posts: 4,579
Likes: 0
Received 0 Likes
on
0 Posts
Believe me, the Bus is a very easy airplane. We call it a "no sweat" airplane
Our man;
DFDR data is not perfect. There is a lot of misinterpretation by "beginners" relating to (1) forgetting to correlate the data and ensure that it was wired up the right way round (2) forgetting at exactly which point in the data frame second the data were recorded and thus ending up comparing not one second intervals across data streams but 1.8 and 0.2 second intervals (3) forgetting that DFDR time needs to be correlated with "real" time as sometimes the clocks run fast or slow in some older recorders. The investigator does not know what type was fitted to this aircraft to know if (3) applies in this case.
Fully agree. Experience, careful interpretation and in-depth knowledge of the recording equipment in use, the LFLs programmed (which include parameter timings) and coordination with manufacturers is part of the process of achieving high levels of accuracy, and that must all happen before the interpretive process can even begin. Reading FDR information is not at all a straightforward process.
I take Kit d'Rection KG's point seriously as well, because that is the way accident investigation proceeds - very often, wreckage "arrangement" provides subtle but important clues which can then be coordinated with DFDR and CVR information, (keeping in mind the crucial point made in your post re data analysis). Tell-tale marks on analogue instrumentation can reveal last readings (MK 747 at YHZ, Concorde at Gonesse), scratch/puncture/tear marks which reveal crash-sequence etc.
Using all available data to determine relationships from which an interpretation of the accident's antedents may be derived is important but such changes with each accident. In this case BOAC's point is spot on. Tell-tales on the thrust levers may or may not provide clues as to relative positions but we don't know here what condition the pedestal is in. Thrust lever position would almost certainly change during the accident sequence, with all the consequent scratches and other damage thereby inflicted. Sophisticated microscopic techniques can possibly determine scratch sequence but in this case impact and fire damage may be too great to reliably determine such. There are no cables so cable-stretch points and tell-tale marks on bell-cranks/pulleys are not available. FADEC memory is non-volatile but survival of the micro-processors may be in question, given the fire.
The unfortunate aspect of untrained, inexperienced yet unbridled speculation is that for many others seeking answers, such activity creates many false opportunities for "confirmation bias", impeding true comprehension of what happened. Such speculation here, taken seriously by media personnel only trying to do their job, can badly mislead the public, causing enormous personal pain for families and can make a mockery out of what could be good investigative journalism.
Professional, trained accident investigators and the accident investigation process itself are far less prone to such biases as the process is exhaustive and "open" to all interpretations/theories until the data, in all its forms, provides the most likely cause(s). The human factors of bias in investigations and the cautions taken therein, are well-understood by those who do it for a living.
None of this process described deals with or accomodates the broader organizational aspects which may or may not play a crucial role in creating an environment conducive to an accident. Investigations into corporate culture, training priorities and footprints, corporate predispositions and priorities (or not) to fostering a safety culture are all part of any investigation as are the immediate human factors which may have led a perfectly sentient, highly trained crew to mis-perceive or simply not see circumstances which would permit a crew to interpret an extremely high-risk event as "normal". No crew sets out to end up this way, so the human factors question which must be answered is, "what, in the moment, 'made sense' to the crew which caused them to act in the way they did?" We who sit with both time and huge amounts of data need to re-build the few seconds in which such interpretations were made by professional, highly-experienced airmen.
DFDR data is not perfect. There is a lot of misinterpretation by "beginners" relating to (1) forgetting to correlate the data and ensure that it was wired up the right way round (2) forgetting at exactly which point in the data frame second the data were recorded and thus ending up comparing not one second intervals across data streams but 1.8 and 0.2 second intervals (3) forgetting that DFDR time needs to be correlated with "real" time as sometimes the clocks run fast or slow in some older recorders. The investigator does not know what type was fitted to this aircraft to know if (3) applies in this case.
I take Kit d'Rection KG's point seriously as well, because that is the way accident investigation proceeds - very often, wreckage "arrangement" provides subtle but important clues which can then be coordinated with DFDR and CVR information, (keeping in mind the crucial point made in your post re data analysis). Tell-tale marks on analogue instrumentation can reveal last readings (MK 747 at YHZ, Concorde at Gonesse), scratch/puncture/tear marks which reveal crash-sequence etc.
Using all available data to determine relationships from which an interpretation of the accident's antedents may be derived is important but such changes with each accident. In this case BOAC's point is spot on. Tell-tales on the thrust levers may or may not provide clues as to relative positions but we don't know here what condition the pedestal is in. Thrust lever position would almost certainly change during the accident sequence, with all the consequent scratches and other damage thereby inflicted. Sophisticated microscopic techniques can possibly determine scratch sequence but in this case impact and fire damage may be too great to reliably determine such. There are no cables so cable-stretch points and tell-tale marks on bell-cranks/pulleys are not available. FADEC memory is non-volatile but survival of the micro-processors may be in question, given the fire.
The unfortunate aspect of untrained, inexperienced yet unbridled speculation is that for many others seeking answers, such activity creates many false opportunities for "confirmation bias", impeding true comprehension of what happened. Such speculation here, taken seriously by media personnel only trying to do their job, can badly mislead the public, causing enormous personal pain for families and can make a mockery out of what could be good investigative journalism.
Professional, trained accident investigators and the accident investigation process itself are far less prone to such biases as the process is exhaustive and "open" to all interpretations/theories until the data, in all its forms, provides the most likely cause(s). The human factors of bias in investigations and the cautions taken therein, are well-understood by those who do it for a living.
None of this process described deals with or accomodates the broader organizational aspects which may or may not play a crucial role in creating an environment conducive to an accident. Investigations into corporate culture, training priorities and footprints, corporate predispositions and priorities (or not) to fostering a safety culture are all part of any investigation as are the immediate human factors which may have led a perfectly sentient, highly trained crew to mis-perceive or simply not see circumstances which would permit a crew to interpret an extremely high-risk event as "normal". No crew sets out to end up this way, so the human factors question which must be answered is, "what, in the moment, 'made sense' to the crew which caused them to act in the way they did?" We who sit with both time and huge amounts of data need to re-build the few seconds in which such interpretations were made by professional, highly-experienced airmen.