TAM A320 crash at Congonhas, Brazil

[SIDSTAR]

Bernd,

As I am not either an expert in accident investigation or aircraft design I bow to your expertise and that of many others on this thread in such matters. However, following this latest accident, I would expect the investigators to recommend Airbus to objectively examine the benefits of the system to see if it can be improved without transferring the problem elsewhere or creating a totally new problem, which I acknowledge is always a possibility. An example of such recommendations would be the UK AAIB's recommendation regarding the ECAM procedure after the BA 320 lost its displays at night.

I was just making the point that I've seen pilots become totally unaware of their thrust setting at almost all phases of flight and this is certainly not helped by the non-moving levers. I agree, better training is part of the answer but the type rating courses that I've seen do not really allow for much of this basic training in the simulator due to the huge number of system failures that must be at least touched on during 7 or 8 sim sessions. While this could be demonstrated in the simulator, the real place where it needs to be emphasised over and over again is in the line training phase. However, even with the point ingrained in a pilot's psyche during line training, operating the system day in day out without any problems can instill a sense of complacency. As professional pilots we all have a duty to try to be aware of such dangers.

I just feel so sorry for these two pilots who clearly were thrown out of their normal "comfort zone" on this landing. The confusion in their minds must have been appalling and it appears that hey did react within a reasonable timeframe when they realised that they weren't decelerating. Unfortunately it was too late on this occasion.

There but for the grace of God go you or I!

[bsieker]

marciovp,

As PBL has already pointed out, making observations about event timing at a precision higher than the sampling rate is impossible, maybe we should hold a short course "Reading Time-Discrete Data Plots". Drawing a linear interpolation between the points implies a continuity in the data that's not there.

I'm also not sure what point George Rochas is trying to make.

There are some points to consider about autothrust operation:

- Thrust in engine #1 was reduced
- Slightly before the dramatic increase around 18:48:26, there is a notable drop in longitudinal acceleration at 18:48:24.2, corresponding to a recorded drop in speed around the same time
- Autothrust is especially responsive to speed variations during approach, trying to compensate quickly even for small deviations, since approaches are usually flown rather close to minimum speed. This submode of operation is known as "approach autothrust", in this mode, precise speed-keeping has priority over comfort considerations, safely to keep the speed above stall speed.

Taking this into account, the reaction of the engines is perfectly normal.

Even with two engines operating, sudden thrust increases shortly before touchdown are not unusual, especially in gusty conditions. Again I recommend watching some landing videos, youtube has a wide selection.

My impression is that George Rocha, this talented Jet pilot feels that the software did not work very well because even eith the LEFT motor reversed and the pilot aplying brakes, it still behaved as if the plane wanted to go around

Yes, this has been widely discussed here, almost from the beginning. The problem still being that the aircraft cannot read the pilot's mind so has to follow control inputs, because there is no obvious Right Thing to do.

And the aircraft was not behaving as if it was tryng to go around; going around requires a whole different region of thrust, closer to sustained EPR 1.4, and not a peak at 1.26, and then settling at 1.18.


Bernd

[peterperfect]

Does anyone know if the runway re-grooving at Congonhas has been completed ? I last saw a projected date of 8th Sep. Any updates ? Thanks.
pp

[Dream Land]

Is it even barely possibly that the pilot at CGH didn’t know this, or didn’t recall the point under the pressures of landing on a short wet runway with only one operating reverser? And therefore, in his own mind, used the buttons to order ‘autothrust disconnect’ and then interpreted that as ‘finished with engines,’ and concentrated only on making sure that he got hold of the correct lever for the operating reverser and moved it into ‘max. reverse’ as soon as possible? by RWA

My guess is no, not a dumb question, you seem to be very educated on the system, I doubt if a person can make it to the left seat without being more familiar with the A/TH system.

[FlightDetent]

Originally Posted by RWA

So, as I see it, even if the wheels are down, the ONLY way to reduce thrust on a given engine to ‘idle,’ and keep it there, is to retard the lever all the way to the ‘idle’ stop?

Exactly. In fact, by reatarding TL to "idle", you will initially reduce the "dead-band" between TL position and actual thrust setting, then limit the autothrust commanded N1 setting to TL position and when at idle the autothrust will disconnect. Unless you do so, A/THR will attempt to maintain speed indefinitely. KILL-THE-THRUST is normal operation at every single landing. There is no way any A320 pilot could attempt and succeed landing other way around.

Originally Posted by RWA

Is it even barely possibly that the pilot at CGH didn’t know this, or didn’t recall the point under the pressures of landing on a short wet runway with only one operating reverser? And therefore, in his own mind, used the buttons to order ‘autothrust disconnect’ and then interpreted that as ‘finished with engines,’ and concentrated only on making sure that he got hold of the correct lever for the operating reverser and moved it into ‘max. reverse’ as soon as possible?

The buttons are not used and not needed in normal operations - this landing was a normal operation, albeit demanding and with little room for error. You do not need buttons to disconnect A/THR for landing, you do not need them to disconnect A/THR to salvage a too high flare, you do not need them when little extra thrust is required. You do not need them for go around thrust. The system is a complicated piece of engineering because it is designed to be operationally VERY simple.

The way you finish with the engines, is you put them in idle. The grip witnessed on the video makes a lot more sense on different autothrust designs. n.b: technically speaking. I am nowhere near being authorized for Madeira landings - definitely no secondguessing here.

If metal gets bent and people hurt, there are lesson to be learned. Personally I feel lost in here, there's nothing to grasp, I cannot come up with anyting to change that would really had made a difference.

[bsieker]

Originally Posted by RWA

http://www.youtube.com/watch?v=3OtcaFkpUpk

It shows that the pilot’s grip on the handles looks slightly odd – only the forefinger is actually hooked over the levers, the thumb and middle finger are gripping the sides of the levers.

The grip in this video looks like the pilot keeps ready for go-around, by keeping the thumb and middle finder behind the levers to be able to push them quickly and decisively all the way forward. Note that contrary to some general SOPs for landing, he does not select reverse at touchdown, but waits for the spoilers call instead.

Some pilots here have remarked that this is also their preferred way on short runways, because, although this carries a small LDR penalty, it keeps open the GA option in case something goes wrong.

I researched this a bit and found that there are in fact red buttons on the outside of each lever, which trigger ‘autothrust disconnect.’ That is presumably the reason for the odd grip, so that the pilot could disconnect the autothrust before flaring – even though he would have been planning to pull the levers back to ‘idle’ in any case.

You only need one button, most comfortably I guess pressed with the thumb. It is a bad idea to disconnect autothrust explicitly before pulling the levers to idle on landing, because thrust would then rise to meet the lever position. After pulling the levers to idle autothrust disconnects, but can be re-activated, but in either case, thrust would be limitied to the lever position, i. e. idle.

From information earlier in the thread, we all know that, with the A320 setup, autothrust disconnect does not automatically reduce engine thrust to ‘idle.’ If the throttles are above ‘idle’ the thrust merely ‘defaults’ to the last recorded thrust setting (which in this case was presumably the power required to maintain landing speed on final approach).

Careful here. There are four distinct ways autothrust can disconnect:

1/ by pulling the levers (both) to idle. This is the normal way during landing. Thrust is reduced to and stays at idle.
2/ by pressing the so-called "instinctive disconnect" pushbutton at the thrust lever side. Thrust rises to meet the lever position. Standard procedure when using this method is to first move the levers to match the actual thrust setting, but it is not technically necessasry.
3/ by pressing the A/THR pushbutton on the glareshield. Thrust is locked until the levers are moved, after which thrust meets lever position.
4/ by an internal failure condition. Thrust is locked until the levers are moved, after which thrust meets lever position.

(4/) was the most likely way in this accident.

So, as I see it, even if the wheels are down, the ONLY way to reduce thrust on a given engine to ‘idle,’ and keep it there, is to retard the lever all the way to the ‘idle’ stop?

Yes, the above notwithstanding.

Is it even barely possibly that the pilot at CGH didn’t know this, or didn’t recall the point under the pressures of landing on a short wet runway with only one operating reverser? And therefore, in his own mind, used the buttons to order ‘autothrust disconnect’

No. In that case, thrust would have risen to the lever position (maximum climb power, around EPR 1.28). And A/THR disconnection would have been recorded in the FDR graphs. Neither is there.


Bernd

[PAXboy]

SIDSTAR

I would expect the investigators to recommend Airbus to objectively examine the benefits of the system to see if it can be improved without transferring the problem elsewhere or creating a totally new problem, which I acknowledge is always a possibility.

Unfortunately, even rigorous testing may not test the combination of events that causes an incident. So, you might change something in the equipment configuration and test it for six months and declare it safe. Perhaps five years later, you will discover that a new set of circumstances has occurred and they met the new weakness in the system that you made.

However, you could change nothing and the exact same fault might be repeated in five years. That is the nature of the business of aviation.

[alemaobaiano]

Does anyone know if the runway re-grooving at Congonhas has been completed ? I last saw a projected date of 8th Sep. Any updates ? Thanks.

The work has been completed but as of Friday the authorisation to reopen the runway had not been received. This is probably related to the ongoing legal battle over documents submitted to the courts by a director of ANAC during the "closed when it rains" situation some months ago. Everybody wants this to be legal and above board this time around with no disputes over who said what.

ab

[marciovp]

Grooving

Does anyone know if the runway re-grooving at Congonhas has been completed ? I last saw a projected date of 8th Sep. Any updates ? Thanks.
pp

Yes, from what I read, the job has been done. They are now thinking about the soft cement on 150 meters at the end of the runway (I guess on the sides?...). And some are talking about the steel horizontal net...

Guarulhos main runway is being repaved... it will take a while so the pilots are using the secondary runway (shorter).

It seems, so far, that the Air Force will remain with Air Traffic Controll (people and equipments). The federal government has released moneys
for the Armed Forces (including the Air Force).

Slowly things seem to be improving, thanks God.

Yes, this has been widely discussed here, almost from the beginning. The problem still being that the aircraft cannot read the pilot's mind so has to follow control inputs, because there is no obvious Right Thing to do.

But with the pilot doing a reverse and trying to brake the plane could there be a doubt on what he wanted to do? Couldn´t the software be programed for this situation?...

[bsieker]

Originally Posted by marciovp

But with the pilot doing a reverse and trying to brake the plane could there be a doubt on what he wanted to do? Couldn´t the software be programed for this situation?...

I believe it could, to a point. Firm braking being the key factor here. Some data mining needs to be done to establish what would unambiguously constitute braking instead of "parasitic" brake pedal deflections while making rudder inputs.

I will shortly publish an extension to the ground spoiler-extension logic, taking braking into account, while allowing for one above-idle thrust lever reading.

Great care must be taken, because ground spoiler deployment in the wrong moment can be just as disastrous as lack of ground spoilers when they are needed.

I expect it to be controversial, and hope for a constructive and fruitful discussion.

Watch PBL's web site for news.


Bernd

[Dani]

I guess it could, but it contradicts completly the Airbus design philosophy. In spite of being highly electronically based, Airbus aircraft never do something themselves! You might be surprised.

Example: If you have an engine failure, the software could easily do everything for your, Rudder deflected, MCT set, Master off, ...

Boeing in 777 for example does automatic rudder compensation in case of OEI (if I'm right). Saab does automatic APU start. MD-11 is extensively automated when it comes to switchings in the hydraulic and pneumatics (to replace the FE from DC-10).

Airbus' intention is, if you take the pilot out of the loop, he doesn't know what's happening anymore. That's why you have to brake yourself. And you can revers one engine and go on TOGA on the other. The system is completly pilot-oriented.

Dani

[EMIT]

Taking over

A couple of pages ago there was some question/discussion about the apparent take-over of control during the landing roll.

If both pilots had given simulteneous inputs on the sidesticks, there would have been a "DUAL INPUT" callout.

If one had pressed the take-over pushbutton on a sidestick, there would have been a "PRIORITY LEFT" callout (or RIGHT, for that matter).

If CM2 did give control inputs while CM1 was busy with braking and cycling speedbrake lever, I think this would be a fairly natural tendency with death looming about 15 seconds away. Forgetting an SOP call ditto.

[bsieker]

EMIT,

good point about the "PRIORITY RIGHT" automatic callout. I had forgotten about that.

It does not appear in the WARNINGS AND CAUTIONS section, so do you have any idea in which flight phases (if any) it might be inhibited? Or which other aural signals might have priority?

(We found out analysing another accident, that the "DUAL INPUT" warning, even though FCOM says it is not inhibited in any flight phase, has a lower priority to the EGPWS "pull up" call-out, and was thus inhibited.)


Bernd

[SIDSTAR]

Dani,

Just to clarify the AI position: Airbus changed the method of dealing with a locked out reverser last year but gave no reason for the change. Perhaps it was linked to the previous incidents? The revision to the MEL clearly now shows that on landing with a locked out reverser, you pull reverse on both engines even though this will result in an increase (small they say) in forward thrust on the engine with the locked out reverser.

[EMIT]

Dual Input

Bernd,

Off the top of my head, those calls are never inhibited (if you mean inhibition based on flight phase), but indeed they may be overruled by calls of even higher priority like (GPWS) PULL UP.

EMIT.

[gpvictor]

Some lay questions

I'm not a pilot (my domain is psychology), but I like aviation. I care also for my safety, so I stay informed: in fact, I were in Brasil when the crash occurred, and it was quite impressive.
I wonder why Airbus did maintain a somewhat complicated landing procedure with one locked reverse (first both to idle, then only one to reverse) that the two TAM's pilots followed, probably in a wrong way. Sidstar says that they changed it on the MEL, but, after two crashes (Philippines and Taiwan) with similar dynamics (and probably some other occurrences where the pilots discovered quickly the error), a change in the software i.o. to unify all landing procedures, with or w/out reverse, should be implemented. When you are relying more on perceptions than on logic (and it occurs), the image of two asimmetric levers can be misleading.
And I would ask to the pilots flying on A320 if a warning when the engines of the plane are pushing one in the opposite direction of the other could be useful

[EMIT]

Of Books and Procedures

SIDSTAR,

MEL changes are never accompanied by "reasons why". Even stronger, since MEL is a book, or a file on our LPC laptop, that we only have available on board (not as "at home documentation" like FCOM and QRH), we as pilots see the MEL only when we have a dispatch deficiency on the aircraft.

Whether or not the procedure for de-activated reverser was changed, was probably not known by the pilots. The procedure as it was at their time, of course must have been viewed by them.

Strange thing to realize is that, in case of a reverser failure happening during the flight, e.g. in case of hydraulic system failure on G and/or Y system, ECAM and QRH will only state REV 1(/2) INOP. You will receive no guidance on what to do with the thrust levers! (I mean whether to pull into reverse range or to leave in idle). Remember that in such a case, MEL is not applicable, so you are not required to consult it.

Couple this with perhaps very good FCOM knowledge (about the increase in fwd thrust when T/L is moved into reverse range) or experience from SIM sessions where this phenomenon may have made the difference between stopping on, or just beyond the end of, the (simulator) runway, and there may be the ground for wanting to leave the "offending" thrust lever in idle, rather than taking it along into reverse range as the latest AI MEL instructions tell you to do.

EMIT.

[RWA]

Originally Posted by Dream Land

I doubt if a person can make it to the left seat without being more familiar with the A/TH system.

Originally Posted by FlightDetent

KILL-THE-THRUST is normal operation at every single landing. There is no way any A320 pilot could attempt and succeed landing other way around.

Quite agree, Dream Land, Detent. But I'm trying to 'think outside the box'; it's let's say 80% certain, on the FDR evidence, that a highly-experienced pilot made a mistake. The only reasons I can think of for that are 'overload' (too much else to think about) or a less than full understanding of the interacting systems.

Originally Posted by bsieker

Careful here. There are four distinct ways autothrust can disconnect:

That's what's 'bearing in on me' - four different methods, but only ONE (TLs to 'idle') that actually does what you want. Does anyone know why the systems got that complicated - in particular, why the red 'instinctive disconnect' buttons are provided at all, when in fact (as I understand it) they don't kill the power but merely 'freeze' the thrust where the autothrust had left it?

My 'starting point' (from the rare occasions when I was let loose on anything expensive enough to have an autothrottle) is that in those days you disconnected the A/T quite early in the approach. Thanks to moving throttle levers you knew exactly what the remaining power setting was.

I've learned, from chatting to pilots, that the technique nowadays is to leave the A/T on much longer, almost into the flare; but the whole idea of that 'bothers' me somehow. Since flaring inevitably involves a slight but progressive reduction in airspeed, aren't you just about asking for the engines to give you a 'kick in the pants' at the very moment when you least need it, unless you get the timing of the 'retard' exactly right?

Originally Posted by bieseker

The grip in this video looks like the pilot keeps ready for go-around, by keeping the thumb and middle finder behind the levers to be able to push them quickly and decisively all the way forward.

Have cordially to disagree on that point, bieseker. As far as I know, all he'd need if he wanted to apply full power would be the heel of his palm behind the levers. Unless all these 'systems' are even weirder than I've gathered to date.

[peterperfect]

alemaobaiano,
Thanks for the update on the runway re-grooving situation. pp

[bsieker]

Originally Posted by RWA

That's what's 'bearing in on me' - four different methods, but only ONE (TLs to 'idle') that actually does what you want.

This is rather looking at it backwards. At the flare your aim is not to disconnect autothrust, but to cut engine power. So the most natural thing is to move the levers to idle. That autothrust also disconnects is "merely a courtesy detail."

Does anyone know why the systems got that complicated - in particular, why the red 'instinctive disconnect' buttons are provided at all, when in fact (as I understand it) they don't kill the power but merely 'freeze' the thrust where the autothrust had left it?

One does not turn off autothrust to kill power, but to have manual thrust control.

The red buttons don't freeze thrust. They adjust thrust to match the lever position and are used if you want manual thrust control in the IDLE-to-CLIMB range. Adjust the levers to meet actual thrust, push the button, you have it. (or just push the button and be prepared for a sudden change in thrust. As the FCOM puts in in other places, this is probably "discouraged if passenger comfort is a priority.)

(N. B. Disconnecting A/THR with the illuminated pushbutton on the FCU, which does freeze thrust, is non-standard.)

Have cordially to disagree on that point, bieseker. As far as I know, all he'd need if he wanted to apply full power would be the heel of his palm behind the levers. Unless all these 'systems' are even weirder than I've gathered to date.

Well, that was just a blind guess anyway. It looks like a position I would find most comfortable and safe when anticipating to push as quickly as possible. I use similar grips on the gearshift-lever of my car sometimes, which others also find strange-looking, but I happen to feel comfortable with.

The fingers reach a bit farther than the heel of the hand, so you'd have to move your arm/body a bit less.

You'll notice that he uses the same grip to get from reverse back to forward idle (going a bit past and pushing firmly back to idle stop, as other pilots said is normal.)


Bernd