TAM A320 crash at Congonhas, Brazil

[Lemurian]

Rob21

I don't claim anything here as truth, but yes, I can permit myself to be open to the possibility of mechanical failure. I am also open to the possibility of pilot error. You are the one claiming my theory is one to a million.

I was referring to this :

So, again, IMHO "desacelera" replied by a "não dá" is an indication that something wrong happened with the right TL, not a simple "forgotten TL" thing...

As I said, it could have been about anything.

Finally, you claim I didn't answer any of your tecnical questions. This is not a contradiction? Why would someone like you want to know anything from the last one here who can know anything?

Actually, they were technical points which have never been picked-up by either of you. Not discussed, not answered. You just switched onto another angle.

As for cuecas, mine are quite safe as I have not concentrated on just ONE possibility - i.e a faulty TLA feed-back - in order to exonerate the flight crew.
I hope I have shown that I've kept a much larger picture in mind.

As you are in the mind for reading, please have a look at THIS FAA study on "The interfaces between flight crews and modern flight deck systems"
Maybe then you'd understand some of our concerns, although it is rather long.

Regards
P.s : Sorry, PBL, I stole it from your archives, but you already know that I have been a rather keen reader of your works.

[PBL]

Originally Posted by Rob21

I think you are manifesting your frustrations to the wrong person.

Not at all. Some people are very sceptical that a thrust lever quadrant, a piece of safety-critical equipment, worked as advertised, and then

Originally Posted by Rob21

The guy who started this "FW3" thing was the V.P. of safety for Airbus, in the presence of the press and broadcasted nationwide.

..believe everything they see on television.

PBL

[3Ten]

Tkanks for your reply, DozyWannabe, I'll take it that those here who know the A320 systems have ruled out the theory.

By the way, am not afraid of FBW, just of Airbus FBW, of throttles that don't feedback, and sidesticks not interconnected, that's all. And I know a lot of A320 pilots who like the acft, but agree with me. There are other ways of designing FBW systems, and I guess Airbus already knows that. But that's beside the Congonhas accident, for the moment.

Cheers

[PK-KAR]

If so, is there any way that Airbus' software could deduce which of these options a crew intends to take by analysing the combination of crew control selections in operation (TL positions, autothrust, reverse thrust, autobrakes, manual braking, flap/slats, etc) at this point in the landing? Alternatively, if just after touchdown, the aircraft software detects a combination of control selections which would preclude a safe 'go around', would it be safer, in this case, if the software made the decision for the crew and executed a full stop landing procedure (automatic selection of both engines to Max Reverse, deployment of GS, autobrake operation and activation of any other devices available for slowing the aircraft down)?

The thing is, if one engine is not in idle or reverse, the computer needs to decide whether it is a landing or a go-around. If one wants to do a single engine go-around then the aircraft must enable it to do so. If one (for some absurd reason) determines that after a reverser has been selected, a single (or all) engine go-around is safer, then it must be able to do so. Once on the ground, it is the crew who decides to continue the landing or go-around, or veer off and find a ditch to stop the plane... not the computer. If both reversers have been selected, spoilers would deploy anyways as it is obvious one wants to stop.

I suppose the main problem would be in determining the precise combination of control settings the software should use when deciding to execute a full stop landing procedure. Would the combination of 'Main Gear Compression' AND 'Max Reverse on at least one engine' AND 'Manual brake application by crew' AND 'Aircraft speed above 72kts' be a sufficiently strong indicator of the crew's intention to perform a full stop landing?

1 engine in reverse and 1 engine in forward above idle is no-man's land and neither airbus nor boeing wants to systematically decide / automate a decision on that grey area. I guess the reason is that it is impossible to determine what the crew wants to do in such a situation, and it only takes a car/cow to intrude on a runway as the aircraft is touching down at 500m away from the touchdown to determine in court whether such a system is prudent or not. Hands can slip on a panic, but shouldn't when performing a normal procedure (such as, selecting idle on flaring).

It might help tired and stressed crews when time is of the essence and they are unable to figure out quickly enough why their aircraft is failing to decelerate.

If they're tired enough to not bring both engines to idle, they're probably too tired to find the button in time!


3Ten,

- Has it been established wether the eng 2 throttle was actually (phisically) in the Climb position? By wreckage evidence or any other mean? At the date of the Airbus telex, Agust 2nd, the wording "recorded" is relevant enough for me, it states the recorded data.

I think the physical evidence melted in the fire...
However from the FDR data, the previous landing in CGH by the same crew (presumably performed by the same PF) showed the #2 thrust lever being left in idle and the #1 in reverse. Then, the following landing in Porto Alegre (presumably with the other guy as PF) shows that both thrust levers were recorded in reverse during the landing. In the fateful and final landing into CGH from Porto Alegre, the #2 lever was left in climb.

My vote goes with the #2 physically in the CLB position... sadly...

PK-KAR

[go-si]

All,
I think PK-KAR has hit the nail on the head. That the computers could not predict what was required of them, they could not read into the minds of the pilots.
That said, they should have been making it obvious (loudly and clearly) that there was a throttle mis-match, forcing the pilots into making an affirmative decision.
Si

[PAXboy]

3Ten

- Has it been established wether the eng 2 throttle was actually (phisically) in the Climb position? By wreckage evidence or any other mean?

Non-pilot but I have read every post in the thread.
As far as I can recall in all of this: The throttle quadrant is being inspected and no public (certainly nothing reliable) has been stated about it's physical condition.

In much discussion about the possibility of physical failure of the throttle lever and it's connected push rods and electronic sensors - we have had graphics of the system (both mechanical and electronic) and consensus is that failure of the TL is remote. Possible, of course, but remote. Until the physical evidence is reported, we cannot know.

I shall mention (quietly) once again that if video recording had been in place in commercial air transport a/c, then we would (eventually) be able to see how the TLs were handled. The report (in a couple of years time) will tell us the most likely story and then the lawyers will take ten years to argue it out.

[lomapaseo]

Non-pilot but I have read every post in the thread.
As far as I can recall in all of this: The throttle quadrant is being inspected and no public (certainly nothing reliable) has been stated about it's physical condition

I bet you missed a few posts

Throttle quadrants are not reliable depending on the level of aircraft damage' twisting, etc. Things to check also include the throttle linkage at the engines, internal fuel control cams and of course the easy way is to read out the FADEC chips on the engines, possibly having captured basic readings when a major event happened.

Having heard nothing, I'm betting that confirmation has been already made.

[ChristiaanJ]

Originally Posted by PK-KAR

My vote goes with the #2 physically in the CLB position... sadly...

So does mine.

And I would like to see the discussion to go a bit more to how that could be prevented.
And not by computer software trying to guess what's going on, please.

As an ancient, I would vote for the autothrottle system driving the TLs mechanically, not controlling the system somewhere downstream.
I'd also vote for an OMC override for the spoilers....

Short of that, is there really no way to simplify MELs and MEL procedures and all that?
Like:
"With one TR locked out, you do NOT land under such and such conditions."
Or.... no subtle and confusing instructions...
"On touchdown, retard both TLs to idle, then to reverse."
Period.

With maybe a "note", saying
"If one of the TRs is inoperative for any reason, the slight increase in thrust between idle and reverse thrust will not have a significant impact on the landing distance."

And maybe not even that.
".... retard both TLs to idle, then to reverse."
Period.

[gpvictor]

However from the FDR data, the previous landing in CGH by the same crew (presumably performed by the same PF) showed the #2 thrust lever being left in idle and the #1 in reverse. Then, the following landing in Porto Alegre (presumably with the other guy as PF) shows that both thrust levers were recorded in reverse during the landing. In the fateful and final landing into CGH from Porto Alegre, the #2 lever was left in climb.

The crew of the previous flight (BH-CGH) was not the same. The PF worked during the previous years for Transbrasil and Passaredo, fliying on Boeing 737, and, if he really did a mistake, it is possible that this fact could in some way have contributed to it. But, in any case, it should be stressed that TAM's pilots did follow different procedures for the same kind of situation.

[marciovp]

Translations

And please (you and marciovp) start referring to the FWC upgrades by their proper designations, such as H2F3, and not acronymic kludges misinvented by goodness knows who.

We did not do it PBL. We just listened from the TAM President and the Representative of Airbus. Now if Airbus says something to Brazilians and something else for the rest of the world, this is not our fault...

because of someone in Brazil telling the thread what a newspaper (mis)reported about what the president of an airline said to a group of politicians just *a little silly*.

You don´t seem to understand or worse, don´t want to understand. Not only the President PBL but the representative from Airbus...when they were interrogated by the CPI of the House (yes, a group of politicians but they knew of the software and asked about it... to the proper people). Be fair.

..believe everything they see on television.

Well PBL why do you want to say that this was not so just because it was in TV? Let me repeat...( I am getting tired...) the TV showed these people answering questions, including the representative of AB. I guess only PBL knows the truth... I already said that I shy way from people who are sure of things...owners of the truth... sounds grandiose to me.

That said, they should have been making it obvious (loudly and clearly) that there was a throttle mis-match, forcing the pilots into making an affirmative decision.

This is what I think...

[alf5071h]

RWA thanks for the update (#2363). I agree that the recommendation is logical in the context of the TransAsia accident, and I am not questioning the report. The Retard call is not a ‘warning’ in the normal design sense, it is an advisory call similar to that which could be given by PNF, which can stop at any time (or even be omitted).
Some views consider the call unnecessary. During normal operations a thrust lever asymmetry might be best detected by the failure to achieve reverse, a condition for which the crew are alerted.
Problems arise when one reverse is MELed and when using the old crew procedure, aspects which have been discussed previously.
Thus, (with positive speculation), if Airbus considered alternative means of alerting and discussed these as options with the Taiwanese, they could have been reported as either proposals for action (planned action) or just considerations for other methods for remedial action; I doubt that we can fully understand this or anything else which could have been be lost in translation.

With further speculation, Airbus may have concluded that as the recommendation was only ‘to review’, and that the primary finding in the accident report related to training and crew co-operation, then a training solution (procedural) had greater relevance. Hence the change to the MEL DDG procedure requiring both TLs to be selected to reverse.
The procedural solution might be seen as having advantages of not changing normal operations, and (as I understand the system), provide an additional awareness / error check during MEL ops; i.e. both TL in REV results in 2 annunciations (amber/green). The procedure addressed the problem directly (crew not moving TL) and did not stick a ‘band-aid alert’ over the error.
I am not debating the pros or cons of thinking this way, nor suggesting that a procedure is better than a design change, but these are the type of issues that design teams, safety pilots and management (humans, subject to HF) have to judge at that time. Thus the speculation might aid some contributors to consider how a manufacturer might address the problem at that time and provide a response based on information in the accident report.

[Lemurian]

Christiaanj

"If one of the TRs is inoperative for any reason, the slight increase in thrust between idle and reverse thrust will not have a significant impact on the landing distance."

With a locked reverser, the affected engine won't go beyond idle thrust.

is there really no way to simplify MELs and MEL procedures and all that?
Like:
"With one TR locked out, you do NOT land under such and such conditions."
Or.... no subtle and confusing instructions...
"On touchdown, retard both TLs to idle, then to reverse."

MELs are generally quite straightforward.
On this very subject, some MELs read thus : "Operations on contaminated runways are forbidden"
Operating procedures belong there, but newly published ones, until their inclusion are to be found on the aircraft technical briefing that everycrew consults - and takes on-board - at the flight preparation.
To be more precise, for each item on the MEL, one would find :

• The MEL item number referring to the systems universal ATA chapter ...
• The authorised max period for repair as A,B... A being a no-go
• The number of systems
• The minimum number of systems required, referring to airworthiness.
• An asterix denotes a dispatch requirement (like "Do not dispatch on contaminated runways)
• A maintenance procedure, citing the MM numbered protocol.
• A crew Operational procedure.

That ATA number is then reported on the Tech Briefing as a list of *TECHNICAL TOLERANCES* submitted to the Flight Captain's approval.His judgemnent is final. (in my country).
The same ATA number is also reported in the ATL -*Aircraft Technical logbook*-with the title of the item.

[RWA]

Originally Posted by alf5071h

Thus, (with positive speculation), if Airbus considered alternative means of alerting and discussed these as options with the Taiwanese, they could have been reported as either proposals for action (planned action) or just considerations for other methods for remedial action; I doubt that we can fully understand this or anything else which could have been be lost in translation.

alf, it's well worth downloading and reading the whole Report - it's admirably clear, well set out, and appears to have been well-translated: which reduces the scope for speculation.

The actual Safety Recommendation reads:-

"1. Reviewing the design of stop mode of Retard warning sounds or accommodating other warning methods to ensure that the warning will continue before the thrust levers are pulled back to Idle notch after a touchdown has affirmed.ASC-ASR-06-03 –006"

The operative word in that, to my mind, is 'ensure'. Back in 2004 the Taiwan Board appears to have identified the key problem, spelled out the probability that it might happen again, and recommended action that would have gone a long way towards preventing a recurrence. To quote the body of the Report:-

"In this situation, one of thrust control lever is not in proper position but the aural "RETARD” alert already stopped. The aural “RETARD” alert should continue or there should be other ways to remind the pilots of pulling back thrust control lever to reduce the probability of an accident caused by human error."

If Airbus later found that the new warning that they themselves had proposed couldn't be made to work, or was inadvisable, or whatever, they should have so informed the safety authorities; and gone on trying to devise one. There is no evidence so far that they did that.

Originally Posted by alf5071h

With further speculation, Airbus may have concluded that as the recommendation was only ‘to review’, and that the primary finding in the accident report related to training and crew co-operation, then a training solution (procedural) had greater relevance. Hence the change to the MEL DDG procedure requiring both TLs to be selected to reverse.

As explained above, their duty was to 'ensure,' not just 'review.' And I for one can't see how the 'both levers to full reverse' thing is relevant. If, as pointed out by the Taiwanese and other investigating boards, it was possible for highly-experienced pilots to neglect the need to bring both levers back to 'idle,' what would be the point in telling them to adopt an even more complex (and counter-intuitive) procedure?

Surely they'd have been just as likely to 'neglect' that too? As far as I'm concerned, on the available information, a continued warning would have been a much better solution?

[armchairpilot94116]

I have a feeling that report is going to be in the hands of many a lawyer.

[Lemurian]

alf5071h

With further speculation, Airbus may have concluded that as the recommendation was only ‘to review’, and that the primary finding in the accident report related to training and crew co-operation, then a training solution (procedural) had greater relevance. Hence the change to the MEL DDG procedure requiring both TLs to be selected to reverse.

I am not sure that the procedure change was related to the TransAsia accident.After some research I did on this subject, it seems that it came from a query from some airlines operating types from both manufacturers as a way to have common procedures...It became an Airbus SOP later, probably after some extensive study on risks. However, the original procedure had to do with an *Inop T/R*, not a *De-activated T/R*.

The procedural solution might be seen as having advantages of not changing normal operations, and (as I understand the system), provide an additional awareness / error check during MEL ops; i.e. both TL in REV results in 2 annunciations (amber/green).

I understand your point, and I agree with you, on the proviso that with a locked-out T/R, there won't be an amber *Reverse* annunciation, as it is related to the sleeves being in transit. As the said sleeves are physically prevented from moving, one would have then a blank on the engine ECAM. And your ECAM aspect would then be *GREEN/Blank*.

[PBL]

Marcio (and Rob21, I suppose), there seems to be something about your culture that I do not yet understand. I ask a question. You can't answer it (neither can anybody else) except through what you have seen on television and in the newspapers. So I say: yes, but I want to know about the paperwork, which must exist if this story is to be real.

And apparently this leads Marcio to conclude lots of things about my supposed character, such as

Originally Posted by marciovp

Well PBL why do you want to say that this was not so just because it was in TV? .... I guess only PBL knows the truth... I already said that I shy way from people who are sure of things...owners of the truth... sounds grandiose to me.

Well, we have words for this kind of contribution and I'm sure there are in Portuguese also. But I can't quite find one which fits. Anyway, there are lots of things wrong with it.
I'll point out two.

First, I have only asked a question. I have neither affirmed nor denied what was said on TV. Second, since no one has answered my question, it seems perverse to infer from that that "only PBL knows the truth" - after all, I have just said I don't!

I do remember the appearance of Malinge before the parliamentary committee, indeed I had direct access to people who were there. The consensus was at the time that, when asked, briefly, about the warning message, he did not give a decisive answer, and rather seemed to steer away from the question. And now, a month or so later, here are two technology sceptics repeating some acronym he apparently uttered as though it was Exhibit number 1!

Let me try to smooth the waters with an analogy. Suppose somebody tells me there is a big fire in town which causes a lot of damage. The politicians are there; the press is there. I want to know about it. I go to the firemen's log of emergency calls. There is nothing about it. And there is nothing about it in their incident register. Their incident register in the last few days goes from 2007-1521, say, to 2007-1700. One of the people reading the newspapers tells me the incident is called 4-22. I call up all the firemen I know, and can't find anyone who even knows anyone who was fighting the fire. And I do this for a month with the same result.

What should I conclude? That there is something wrong with my personality? Or should I rather conclude that things don't quite appear to be as they have been reported by some?

PBL

[3Ten]

Thank you all for your answers.

I can accept that a TLA signal transmition failure is highly unlikely, but the pilot's actions of leaving 1 throttle lever behind also aren't all that likely to me. Any pilot here, if never experienced assymetric reverse thrust in real life, for sure has experienced it in the sim. So:

- Landing, #1 eng idle, #1 eng reverse
- #2 eng accelerates to climb thrust, Reverse noise mask noise of #2 accelerating
- This means IMMEDIATE strong assymetric thrust after touchdown.

How would pilot react? Pilot would probbably start by canceling reverse, to try to stay on the rwy, even if the speed was too high. With reverse canceled, he would most likely hear the other eng at high power, and probably go for thath throttle as well, eng noise is a much more effective warning than the "RETARD" call. This is valid from touchdown anytime to the crash, whenever pilot would feel he was loosing directional control. The FDR recording and the general trend seems to indicate that the crew went all the way down to that building without ever doing nothing, and that, to me, is also unlikely.

That could have hapened, I know, but for me is no more likely than the mechanical failure. I travelled in a TAM A320 cockpit a couple of weeks before the accident, a instructon flight with 2 captains, and I observed perfectly good standards. I have to admit that I left that cockpit with a much better impression about aviation in Brasil. Of course, this is worth what it's worth.

Lemurian, in my MEL, the asterisk means that the failure must be plackarded in the cockpit.

Cheers

[Lemurian]

Christiaanj

And I would like to see the discussion to go a bit more to how that could be prevented.
And not by computer software trying to guess what's going on, please.

Our main subject has to do with Throttle handling.
Throttle handling abnormalities are treated at length during sim sessions, at length and with all the configurations one could think of.
However, we have two philosophies about single engine operations :

• one which leaves the *dead* engine throttle alone at idle, and
• one which advocates, after the *Shut-Down* procedure has been completed, to re-gather both throttles and operate them as in normal operations. In this case, the flying pilot has to remember that throttle actions are in fact to be used with an amplitude that is (about) double the normal.

I have to say that I am in favour of the second solution...
BUT
Things start falling down if the auto-throttle is used :

• In the first choice, if it's an Airbus FBW airplane, that's the configuration, easy to use as at A/THR disconnection, both T/Ls would be at idle...procedure is straightforward. If it's another brand of airplanes, the autothrottle motor and the friction will slowly drive that T/L out of idle, one would have a stagger (The *benefit* of moving throttles falls apart, already). So One would need to re-gather them, then disconnect the autothrottle, retard...etc...
• In the second choice, all airplanes would have throttles to-gether, but it's against AI philosophy (the dead T/l should be in idle) and on the other brand, the moving *dead* engine throttle once again indicates a n engine output that is not coherent with it's state.

As one can see, the choices are not that black and white, are they ?
Now the warning.
I am not really keen on the addition of another warning, especially one that tries to save a situation that is already complex in terms of crew failure interpretation. The AI proposal would only appease the polotical branch of the society and the uninformed public.
What I would like to see is

• A continued *RETARD* voice call, which will then become
• *RETARD ENGINE ONE *-or TWO- after the identification of one missing T/L in the *Idle* detent, along with a warning and caution chime.

Have to go,
Cheers all.

[bsieker]

Originally Posted by 3Ten

Thank you all for your answers.

Thank you for your input. There's a perspective that's not been discussed in quite this form.

I can accept that a TLA signal transmition failure is highly unlikely, but the pilot's actions of leaving 1 throttle lever behind also aren't all that likely to me.

Nobody says it very likely in day-to-day operation. And three occurrences in 55+ million landings is a pretty low rate. However, two of those did not result in a catastrophe (in fact, all people aboard survived both previous incidents), and it has been established that this was in fact what the crew did.

- Landing, #1 eng idle, #1 eng reverse
- #2 eng accelerates to climb thrust, Reverse noise mask noise of #2 accelerating
- This means IMMEDIATE strong assymetric thrust after touchdown.

A number of points, if I may:

I understand you haven't been reading all the posts, and probably cannot be expected to, by now, but a good summary is available at PBL's compendium, along with a link to causal graph (called a Why-Because Graph).

- Engine #2 did not accelerate to climb thrust. It accelerated to compensate for the reduced, and then reverse thrust of engine #1 to maintain approach speed, A/THR still being active. EPR for CLIMB thrust estimated from previous take-off graph to be around 1.28, actual thrust reached a peak at 1.26, and decreased again, before freezing at A/THR disconnect at around 1.18.

- There was no strong yawing motion from the asymmetric thrust, look at the plot for magnetic heading and lateral acceleration. So no clue to alert the pilots to it. Noise, as you say, would be masked by the reversed engine #1. They did expect slightly asymmetric thrust, and given the runway condition probably weren't suprised, either, on encountering slight directional instability.

(The runway not being grooved, pooling and varying levels of standing water might be expected. This seems to be confirmed by varying longitudinal acceleration after initiation of manual braking.)

How would pilot react? Pilot would probbably start by canceling reverse, to try to stay on the rwy, even if the speed was too high.

They stayed on the runway until very late, despite asymmetric thrust.

That could have hapened, I know, but for me is no more likely than the mechanical failure.

We know that it happened twice before, unlike this extremely unlikely mode of thrust quadrant failure.

How do you derive relative probabilites of mechanical/electronic failures versus crew error?

Due to its redundancy and development techniques, an undetected(!) electronics failure is extremely remote.

For an undetected(!) mechanical failure, the following would have to happen (please refer to the drawings of the thrust lever system, posted here some time back, to understand what I'm talking about):

- The mechanics (arms, joints, links, pushrod, ...) between the artificial feel unit and the throttle control unit would have to fail,
AND
- At the same time the arm of the TCU would have to get stuck, without moving the slightest bit, at the CL position.

I don't know what friction the TCU arm has, but probably very little, and so it would likely move after the pushrod failure. If moving out-of-range, this would be detected as a failure condition, if moved at all, it would have been recorded on the FDR.

You be the judge of the likelihood of this simultaneous failure of two mechanical parts.

Before someone says it, a failure between thrust levers and AF-unit would be felt immediately by the crew, or one thrust lever might even move by itself, since the detents and main friction component are not in the levers, but in the AF-unit.

(I stand corrected on this one by someone with more hands-on knowledge of the thrust command system ...)

I travelled in a TAM A320 cockpit a couple of weeks before the accident, a instructon flight with 2 captains, and I observed perfectly good standards. I have to admit that I left that cockpit with a much better impression about aviation in Brasil. Of course, this is worth what it's worth.

Good to hear. However, we can deduce some training issues at TAM by observing that the pilots used two different ways of handling the thrust levers at touchdown on the two previous landings of this particular airframe. Once they set both to reverse, the other time only the one with the operative thrust reverser.


Bernd

[DozyWannabe]

Also worth bearing in mind that some of the Brazilian contributions on here have been from the legal community rather than aviation. I do wonder if there's a chance that some of them are hoping for bigger settlements from an Airbus-sized corporation than they would get from a TAM-sized one?

Let's just categorically state that *no-one* is impugning the ability of the dead pilots, to my knowledge. Even if it does turn out conclusively to be pilot error that turned this accident from likely to inevitable, I don't think there's a pilot out there who hasn't made a potentially deadly mistake in the heat of the moment.