TAM A320 crash at Congonhas, Brazil

[bsieker]

Originally Posted by Lemurian

PBL is referring to the deceleration achieved by this aircraft taken from the FDR print-out. A very *poor* braking indeed.

Poor, but still more than nothing.

Originally Posted by flyingnewbie10

That's what I thought.
However if (if) he was based on the FDR the information might not be accurate as we still don't know how slippery the runway was.

The figure PBL quoted was observed deceleration, i. e. the deceleration that was actually achieved in this landing (derived from the ground speed plot, and supported by the longitudinal acceleration pliot).

This is independent of how it was achieved, and under what conditions.

What is, is.

What we know is that the crew thought that the runway was slippery. And this could explain their hesitation to apply manual braking before spoiler extension if after all such hesitation showed up to be expected from a crew in the same situation.

If you expect poor braking action, you must brake as soon as possible to get the most out of the available distance. Not delay it.

Originally Posted by marciovp

Also if there was some plans for the A320 to land at Congonhas in a raining slippery night efforts should be made to make the plane as light as possible, no?

No. Taking all parameters into account one arrives at a maximum landing weight for a given configuration/airport/runway condition/... Either from tables or (I guess) the LPC ("Less Paper in the Cockpit") computers.

One can go up to that weight. One does not have to take into account every possible adverse situation, like keeping the thrust on and lack of ground spoilers during roll-out.


Bernd

[slip and turn]

Not directly associated with landing weights but then again heavier might help up to a point when slippery if what follows is true ... I was somewhat informally taught by ex military that in potential hydroplaning conditions a "heavy even if a bit uncomfortable" arrival was SOP because it was doubly useful to cause the tyres to break through the surface water & initiate spin-up & traction, and to get positive WoW at earliest opportunity.

Certainly from basic mechanics I would assume that if there is any missing friction force on touchdown, then there is also a missing WoW vector component. I couldn't put a magnitude on it.

I imagine some of this is formalised in type training?

[HotDog]

Definitely correct. Our company training emphasised landing in conditions of possible aquaplaning. A firm touchdown is necessary to bang through the water, get the weight onto the wheels by pushing fwd on the stick , select full spoilers and max reverse immediately followed by braking action when the aircraft is firmly on three points.

[TyroPicard]

Some thoughts following this from Rob21
So maybe I should refrase my question on when the pilot was expected to apply manual brakes. Autobrake was selected to med, as recommended for slippery runways. Autobraking starts with ground spoiler deployment.
Why not start manual braking also with spoiler deployment?

Three braking strategies for this approach ...
1. Autobrake OFF, apply manual braking as required.
2. Autobrake MED, no manual braking - monitor Autobrake effectiveness and apply manual braking if necessary.
3. Autobrake MED, but apply manual braking as required - if for some reason you forget to brake the a/c does it for you.
Without access to the approach briefing, we cannot tell which of 2 or 3 was planned. Even with access we may still be in the dark.
Had it been my landing, I would have used strategy 2. Had I judged that full manual braking was required from mainwheel touchdown I would have diverted - risk assessment. Therefore I think it likely that this crew thought this landing was normal - but a shade demanding, because of the evidence of a "duck-under" and selection of G/S warning OFF.

Regarding waiting for the Ground Spoilers call... nowhere in my Airbus FCOM/FCTM is this mentioned as a procedure or recommendation, and I cannot recall any training (in two airlines) where this was mentioned. I assume the reason is that in the absence of ECAM warning or other known problem, they will always work when conditions are met. If you know all/some are INOP then you choose a runway of appropriate length.
One of the potential problems is.. if NOT armed the Ground Spoilers extend on Reverse selection as a back-up.. so if you wait for the call before selecting Reverse you might wait a long time...

Thanks to Soga for re-posting the Brazilian release..
The crew approach briefing included a reminder that only the left engine thrust reverser was available.

What we do NOT know is what T/L handling MEL procedure they briefed, if any. I think it important to brief what will happen, what indications you expect. It appears from the FDR that crews used different procedures at different airports.....
Cheers, TP

[Rob21]

PBL,
You forgot to "quote" my whole phrase. What you said, and I did not agree, was that engine #2 had "significant forward thrust" during flare, therefore the situation was not "normal". My full statement was that during the flare, engine #2 had the "expected" forward thrust, and EPR "split" occurred after reverse on engine #1 was selected. And only then problems began, when ATHR was disengaged.

TyroPicard,
Thanks for clarifying my thoughts on this braking issue.
I believe that when the pilot went to the ground spoilers arm switch it was a desperate attempt to have G/S, and he knew he would not be able to stop. If he had had a manual G/S "deployment device", I'm sure he would have used it, instead of clycling the G/S arm switch.
I know, on A320 the "manual G/S deployment device" is TLs at idle. But at that point, TL was or "forgotten" or inop.
Too bad...

[bsieker]

Rob21, please read the thread.

Best to search for the things about which you wish to make claims, before posting unverified stuff about them.

Originally Posted by Rob21

[...] EPR "split" occurred after reverse on engine #1 was selected. And only then problems began, when ATHR was disengaged.

- When exactly problems began is hard to say, some might say all was well until they hit the building, but possibly at diverging thrust, particularly at the sharp increase in forward thrust on engine #2, around 18:48:24, coinciding with selection of reverse thrust on engine #1.

- Autothrust remained active until 18:48:29, around 5 seconds later, and until after EPR of engine #2 had decreased again.

- Autothrust wasn't disengaged, but automatically disconnected due to an internal failure condition. Please review previous posts by Lemurian and myself, the Taipei-Sungshan-Overrun report, p148, and my Why-Because-Graph on the matter. Be sure to read the introductory words by PBL on the analysis.


Bernd

[Rob21]

bsieker,
Please, also read the thread!!
My reply was to PBL, in an answer to his statment that things were NOT normal during flare. I disagree, things were not normal for us, but for the pilots nothing "abnormal" was happening. Only after touchdown and reverse on engine #1 was selected "things" went bad (for the pilots).
For some of us, maybe "things" started to go bad even before takeoff in Porto Alegre.
But thanks for your effort on trying to educate me.

[flyingnewbie10]

The figure PBL quoted was observed deceleration, i. e. the deceleration that was actually achieved in this landing (derived from the ground speed plot, and supported by the longitudinal acceleration pliot).

This is independent of how it was achieved, and under what conditions.

What is, is.

What was: The plane had a given deceleration
What we don't know whether it was or not as yet: The runway was (very) slippery

PBL stated that one can get a given deceleration in a slippery runway. I supposed he was basing his statement on the TAM's PR-MBK FDR.

Then I finished by stating that his statement might not be accurate not because the plane didn't decelerate at that given rate (What was) but because we don't know in what extent that runway was slippery (what we don't know whether it was or not).

If you expect poor braking action, you must brake as soon as possible to get the most out of the available distance. Not delay it.

No controversy about that when you don't have a (very) slippery runway (or don't think you have).

However, I found this article

http://www.jet-jobs.com/articles/hydropln.html

which I am not certain whether it is accurate or not. It deals with a "Minimum Hydroplanning Speed" i.e., a threshold above which you should not apply braking when facing a slippery runway.

Again, I don't know whether the article is accurate or not.

[Lemurian]

Of brakes and slippery

Flyingnewbie

What was: The plane had a given deceleration
What we don't know whether it was or not as yet: The runway was (very) slippery

I don't know how you read the FDR graphs but to me, if an aircraft only achieves a deceleration of 1.2m/s/s with full brake pedal pressure, that braking indeed qualifies as very *poor* and the runway as *slippery*.

It deals with a "Minimum Hydroplanning Speed" i.e., a threshold above which you should not apply braking when facing a slippery runway.

No, it does not say that at all.
Min hydroplaning speed is the speed above which your tyres might face dynamic hydroplaning, whether they are turning or not.
Now the technique, and its explanation :
..."The anti-skid system will not operate until the main wheels of the aircraft spin... don't lock your brakes before touchdown"...
For a better illustration of the phenomenon and recommended piloting techniques , see the links I provided in some of my latest posts : #2165 "BAC wet runway guide", which deal with physics, really, and #2177 "Landing on slippery runways".
Worth a read, really !
Lemurian

[flyingnewbie10]

I don't know how you read the FDR graphs but to me, if an aircraft only achieves a deceleration of 1.2m/s/s with full brake pedal pressure, that braking indeed qualifies as very *poor* and the runway as *slippery*.

Do not forget the forward push from the right engine...

[RWA]

Originally Posted by Lemurian

Yes, I have, both on the sim and in real line operations. There is no problem at all, both on reverse selection and on directional control, even on a wet runway.(

Thanks, Lemurian - pretty conclusive, coming from a serving airline pilot. Except that all the signs are that a vastly experienced pilot (and a training pilot at that) DID make that mistake. What's your personal take on how that highly-unlikely eventuality could possibly have occurred?

Originally Posted by PBL

But it is not as if we are entirely without our own. People who think that the only cause (read "causal factor") was the pilot failing to reduce thrust to idle might do well to read the first few inches of column space of David's article, in which he discusses the runway situation at Congonhas.

Have to disagree to an extent, PBL. This appears to be one accident where the 'primary cause' is pretty clear; the No. 2 TL being left in the 'CLB' detent (or, to be strictly accurate, being recorded by the FDR as being left in the 'CLB' detent, and duly interpreted as such by the systems) resulted in about 55% N1 forward thrust being delivered by the starboard engine throughout the landing run. In addition, and arguably more important, the incorrect TL position prevented ground spoilers and autobraking from being deployed.

That would have made some sort of incident or accident virtually certain.

The pilots applying manual braking a few seconds earlier might have turned a catastrophic overrun into a slightly less serious one; but commonsense says that a serious accident could not have been avoided, whatever the pilots had done, once the original mistake or misreading (both TLs not being retarded to 'idle,' or alternatively the systems not apprehending that they were at 'idle') had been made.

[bsieker]

Originally Posted by Rob21

bsieker,
Please, also read the thread!!

Believe me, I did.

My remark was mostly to your assertion that "autothrust was disengaged". It wasn't. It disconnected automatically. And much later than the thrust increase on engine #2.

The two events are not causally connected, except in the sense that the fact that A/THR disconnected late was a causal factor in A/THR still trying to maintain approach speed, and thus increased thrust on engine #2. The WB-Graph shows causal relations quite clearly. I invite you to take a look at it, if you haven't already. It is not final. I'd appreciate your comments on it.


Bernd

[Dani]

Lemurian, we don't know how slippery the runway was, but we know for sure that engine 2 was still at forward trust. Even with full brakes you will not manage to get any usefull deceleration.

[Lemurian]

Flyingnewbie

o not forget the forward push from the right engine...

and Dani,

we don't know how slippery the runway was, but we know for sure that engine 2 was still at forward trust. Even with full brakes you will not manage to get any usefull deceleration.

Please refer to the Transasia FDR graphs and the discussion on pp 30 to 36 of that report.They achieved better deceleration, obviously.
What's more interesting, though, is that (refer fig 1.11-2, page 34 )

• originally, without any braking to speak of -time tags 30 to 39 -, they had a deceleration of 1m/s/s, therefore due to aerodynamic drag and 1 T/R only.
• Then between Tt 40 and 48, the effects of a skid with a working antiskid system (brake pressure is way too low for the pedal position), but they still achieved 1.3 to 1.6 m/s/s.
• On reaching a less slippery surface,- Tt 49 to 15 -, a deceleration of 2m/s/s was reached, still with the antiskid modulation but now, max brake pressure is achieved.

Unfortunately, the size of the CGH graphs don't allow such analysis.
Therefore, I stand to my arguments : That runway was very slippery.
The Transasia report is found
Here.
Lemurian

[bsieker]

Originally Posted by PBL

People who think that the only cause (read "causal factor") was the pilot failing to reduce thrust to idle might do well to read the first few inches of column space of David's article, in which he discusses the runway situation at Congonhas.

Originally Posted by RWA

Have to disagree to an extent, PBL. This appears to be one accident where the 'primary cause' is pretty clear; the No. 2 TL being left in the 'CLB' detent (or, to be strictly accurate, being recorded by the FDR as being left in the 'CLB' detent, and duly interpreted as such by the systems) resulted in about 55% N1 forward thrust being delivered by the starboard engine throughout the landing run. In addition, and arguably more important, the incorrect TL position prevented ground spoilers and autobraking from being deployed.

I don't think PBL is denying that this is a causal factor in this accident. He is simply contesting the notion that it is the only cause, or even the Primary Cause.

The notion of "Primary Cause" as such is problematic, because it is impossible to create an objective prioritisation of causal factors (see last paragraph).

To make a causal explanation of an event, one needs to ensure that the causal factors that lead to it, are (a) individually necessary, and (b) jointly sufficient.

This means that, leaving thrust lever #2 at CL detent during a landing is most likely a causal factor in this incident. (Using the counterfactual test to determine causality: Had the thrust lever #2 not been left in the CL detent, the accident would not have occured.)

But this is not sufficient. Not all landings where this happens, end in a catastrophe. In fact there is at least one which only resulted in a no-injuries-incident. There may be others, where, except for a longer roll-out distance, nothing remarkable happened. On a long, dry runway, manual braking may well be sufficient.

So there must be other causal factors. The length of the runway comes to mind, as well as its state (presumed "slippery", although this term seems to lack a precise definition), and also the delay in the application of wheel brakes.

PBL has addressed the problem of priorisation of causal factors in his talk at the Ninth Bieleschweig Workshop: http://www.rvs.uni-bielefeld.de/ -> Bieleschweig Workshops -> Ninth Workshop -> Negotiating Accidents. (Direct Link).

Cf. pp 36, 54


Bernd

[PBL]

Originally Posted by Dani

we know for sure that engine 2 was still at forward trust. Even with full brakes you will not manage to get any usefull deceleration.

Here we go again.

For the information of others, the most optimistic estimate of the stopping distance at the recorded deceleration is that of bsieker on 10.08@10:21, who notes an acceleration of 1.5 m/s/s after brake application and calculates a stopping distance of 1728m.

The most pessimistic estimate is that of TripleBravo on 17.08@21:47, who estimates -1.08 m/s/s acceleration over the entire episode for a stopping distance of 2.4 km.

There are lots of airports with runways with LDAs of well over 2.4 km. The airplane would have stopped on the runway on all of those.

Some people such as bsieker and TripleBravo have done some considerable work calculating such figures and redacting them for the rest of us. It is only fair to them to remember what they have done and use their work. And it is sensible to do so, otherwise we do not progress with understanding this accident and its consequences.

PBL

[PBL]

Originally Posted by RWA

This appears to be one accident where the 'primary cause' is pretty clear

RWA,

There is a fairly extensive discussion about selection of some causes over others between myself and TopBunk and Dani on 19.08. Would it be too much trouble to ask you to look over what was said there? I'm happy to discuss causes and selection of "primary" causes when we are working on some level of common understanding.

Note, for example, that had the runway been, say, 3km long, there would have been no overrun. That means that the relatively short length of the runway was a causal factor in the accident. You need therefore to tell me why and how you prioritise not reducing thrust (as a factor) over the length of the runway (as a factor).

PBL

[flyingnewbie10]

originally, without any braking to speak of -time tags 30 to 39 -, they had a deceleration of 1m/s/s, therefore due to aerodynamic drag and 1 T/R only.
Then between Tt 40 and 48, the effects of a skid with a working antiskid system (brake pressure is way too low for the pedal position), but they still achieved 1.3 to 1.6 m/s/s.
On reaching a less slippery surface,- Tt 49 to 15 -, a deceleration of 2m/s/s was reached, still with the antiskid modulation but now, max brake pressure is achieved.

Lemurian,

Do you have an average EPR for the right engine for those time frames ?

P.S.: Should we discard a below average braking system effectiveness ? One of the last maintenance reports for the PR-MBK, if I am not wrong, is related to problems in the braking system.

Editing: Right Engine EPR in the TransAsia accident: 1.08
Right Engine EPR in the PR-MBK accident: 1.18

[flyingnewbie10]

That means that the relatively short length of the runway was a causal factor in the accident. You need therefore to tell me why and how you prioritise not reducing thrust (as a factor) over the length of the runway (as a factor).

The relatively short length (it is really short and has no escape area) is longer than the one for which the A/C is certified to Land.

That being naturally with a normally reduced thrust (not even reversed) after touchdown.

So I would prioritise not reducing thrust (as a factor) and that would be the main (or the real) cause IMHO.

[RWA]

I did follow that earlier discussion, PBL.

Originally Posted by PBL

Note, for example, that had the runway been, say, 3km long, there would have been no overrun. That means that the relatively short length of the runway was a causal factor in the accident. You need therefore to tell me why and how you prioritise not reducing thrust (as a factor) over the length of the runway (as a factor).

With respect, PBL, I did tell you how I prioritise not reducing thrust, by referring to it as the 'primary cause.' By definition, there can only be ONE primary cause.

Sure, the shortness of the runway, the lack of a runoff area, the surface, the weather, possible delay in applying manual braking etc. all contributed to the severity of this accident. They may even have been responsible for making it a catastrophic 'accident' instead of a mere 'incident.' But they are not the features of it which deserve the most urgent attention, from the point of view of avoiding any recurrences.

One of the problems with all crash investigations is that the resources allocated to them, and the notice taken of any findings, tend to be proportionate to the number of people killed. The fact that pilots were capable of making this mistake, and appropriate action which could have been taken to avoid it happening in the future, was very clearly spelled out by the Taiwanese Aviation Safety Authority in their very competent report on the 'carbon copy' 2004 Transasia A320 'no injuries incident' at Taipei. To quote their report:-

"According to the documents provided by the aircraft manufacturer, the reason of “RETARD” stop was caused that the thrust control lever no.1 was at reverse position. When the FWC detected an internal signal of TLA inhibition which includes either thrust control lever at REV position and then FWC stopped the aural “RETARD” alert.

"During landing, the FWC has delivered four aural “RETARD” alerts and then after touchdown two seconds it came to stop. But at this moment the thrust control lever no.2 was not pulled back to idle or reverser position yet. 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."

They included a recommendation on this aspect in their Safety Recommendations and the Report records Airbus' response:-

"According to the Airbus Company

"1. In response to: 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.

"Airbus Company Response:

"Airbus has developed a specific warning when one throttle is set to reverse while the other is above idle. This warning generates an ECAM warning "ENG x THR LEVER ABV IDLE", a continuous repetitive chime (CRC), and lights the red master warning light. This new warning is implemented in the FWC standard "H2F3".

"A Service bulletin will be issued very soon on this subject."

http://www.asc.gov.tw/acd_files/189-c1contupload.pdf

I gather that the modification has been offered to airlines as an option but has not been taken up much so far (if at all). It bothers me, too, that as far as I know Airbus are not even incorporating it in new construction.

With hindsight, if Airbus had 'pushed' the available mod. more - or, ideally, if the relevant authorities had made it the subject of an Airworthiness Directive - the Congonhas crash could have been avoided; or at least had much less severe consequences.

I hope that makes my tendency to concentrate on the 'primary cause' a little more understandable? As things stand, the same sort of accident could happen again tomorrow.