Please continue the discussion here.

Transasia ATR-72 Crashes In Taipei - Part 2 (by bnatraveler Feb 4 2015 in Civil Aviation)

May all the victims rest in peace.

Ben Soriano

http://www.youtube.com/watch?v=RUQWJai59lo

Video from one of the buildings.

[Edited 2015-02-06 21:36:38]

They were dropping fast when they went past that building.

Can you tell me more about this - do you get a physical text based message on your ECAM easily decipherable from other maybe routine messages? It would be hard to misinterpret that, no?

The message is ENG 1 FLAME OUT or ENG 2 FLAME OUT. difficult to misinterpret but we see what we think sometimes.

Seems Migair54 flies ATR's so maybe he can help - what would happen if you bring the thrust lever back to zero then push back to TOGA after a few seconds? Would the engine complete the first command and completely shut right down & autofeather or is it like a gas pedal in a car and respond to the current position if it had not come to a complete stop?

If the case if with the Power levers like any other turbine it will respond slowly but it will do, But the problem is if they retard the Condition lever and cut off the fuel, that take times, and at low speed and slow you're almost gone...

One of the issues I see here is that the -600 is relatively new machine, so every single incident and accident has to be examined even more carefully than normal because it can be a defect of the software, hardware or procedures, same happened with A380, B787 or A350, even more new technologies.

If this was the case then it reminds us of the British Midland Flight 92 disaster where the pilot shut down the wrong engine and crashed on approach short of the runway.

And the CASA in Málaga also.

Here's an accurate representation of the last half mile of flight in GE format.

Red line is the flight path, the junction of the red and yellow lines is located at the point the plane was seen passing the building in the video from post #1 above. The Transmission line tower is also indicated but you may have to blow the picture up in order to see it.

It appears that the PF was intentionally flying down the river valley, avoiding high ground on the left and right.
That's if there was a PF - there were three Captains on the FD, all three may have been managing the restart of #1.

Has the FDR data for the flight parameters (not the engine parameters) been put out yet?

Many airlines have changed their engine out/engine fire procedures so that nothing really happens below 2500' but climb. This strategy works quite well for jet operators. In the fairly distant past, these engine failure scenarios were memory items with you shutting fuel and bleed valves and shedding generators, all from memory often below 1,000'. Partially because the necessity was gone, partially because what little you gained from performing all of these steps down low was largely offset by doing something really bad like shutting off the fuel to the wrong engine. Now days, you don't throw any switches until you are safely away from the ground. In a jet, there are not really many conceivable scenarios that make you want to do these things early. If you have a simple engine failure, you got all kinds of time to get to a safe altitude before running a checklist. If you have a simple engine fire, while you want to shut the engine down promptly, the thrust of the engine that is on fire, makes you get up to a safe altitude fairly quickly. The only scenario that you MIGHT need to be throwing switches down low is a catastrophic engine failure with an engine that is vibrating enough to be threatening to tear the airplane apart. That is a pretty rare chain of events for that to happen. It would not be normal to be shutting-down an engine below 2,500 feet.

For the turbo-prop operators however, the single engine performance depends on the propeller being feathered. The most difficult training scenario in a 121 training program in my opinion is the engine failure with a negative auto-feather. In the last turbo-prop I flew, the DHC-8, In the even of an engine failure the first thing the crew did was check to verify that the prop did in fact feather. If the auto-feather did not work, then you had to manually shut the engine down and then command the props to feather with the condition levers...apparently there are scenarios where this might not work either, so your last chance was to manually drive the props to feather using the feather pumps (I can't for the life of me remember what the motive force for these pumps were, electric?) All of this is done at tree-top level if you are heavy enough so you are trying to be methodical and careful, but at the same time you know that if you don't get the prop feathered, the airplane will not climb.

I guess my point is that while a jet pilot is trained to climb and then start throwing switches, some failure scenarios in turbo-props lead to some negative training as far as flipping switches down low.

Quoting PassedV1 (Reply 5): some failure scenarios in turbo-props lead to some negative training as far as flipping switches down low.

I tend to agree. Tprops training mentality hasn't changed much in recent decades, even with the advent of reliable auto-feather systems.

This might be the trigger point for regulators and manufacturers to start re-evaluating the statistical risk of these engine out procedures.

As a side note, I would bet anyone a beer that this airplane would have flown with a failed engine AND its prop windmilling/unfeathered..

Quoting LTC8K6 (Reply 2): They were dropping fast when they went past that building.

Quoting PassedV1 (Reply 5): I guess my point is that while a jet pilot is trained to climb and then start throwing switches, some failure scenarios in turbo-props lead to some negative training as far as flipping switches down low.

I hope these issues are raised during the retraining going on with all 71 Transasia ATR pilots as we speak.
http://www.dailymail.co.uk/wires/afp...pilot-retraining-deadly-crash.html

Best,

OMP777X

Quoting migair54 (Reply 3): The message is ENG 1 FLAME OUT or ENG 2 FLAME OUT. difficult to misinterpret but we see what we think sometimes

Unless the story of the whistleblower is correct and, let's assume, engine 1 was was acting up on the previous flight. The pilot might have assumed that same engine is now broken and without checking anything shuts it down.

GE235 flight received take off clearance from SongShan tower at 1051:13, after took off, at 1052:33.8 ATC controller requested the crew to contact Taipei approach. After 5 seconds, (1052:38.3), master warning sounded in the cockpit associated with right engine(2) flame out procedure message on display unit. The crew called it out. At 1053:04.0 the recorded parameters indicated that the left engine power lever(1) was progressively retarded to flight idle. At 1053:24, the left engine condition lever(1) was set to fuel shutoff position resulting in left engine(1) shutdown. Between 1053:12.6 to 1053:18.8, several stall warnings sounded. Flight crew declared an emergency at 1053:34.9 and reported an engine flame out. At 1054:09.2, flight crew called several time for engine restart. At 1054:20 the recorded parameters indicates a restart of the left engine(1). At 1054:34.4, master warning sounded, 0.4 seconds later, CVR recorded unidentified sound. Both recorders stopped recording at 1054:36.

The only thing I do not understand is: why retard ANYTHING.

Back to ME school 101.... Eng OUT.... 1) MAX PWR 2) Identify 3) VERIFY 4) Feather

The ATR should have a T handle for the backup feather system ( if autofeather does not wrk ).
Assuming one eng failed, flamed out or whatever... Full POWER allows you to sort things out while getting the max
Power out of the working engine. If you retard the wrong one, well.

I thought ME training drilled this into the minds of pilots.

Quoting zeke (Reply 9): GE235 flight received take off clearance from SongShan tower at 1051:13, after took off, at 1052:33.8 ATC controller requested the crew to contact Taipei approach. After 5 seconds, (1052:38.3), master warning sounded in the cockpit associated with right engine(2) flame out procedure message on display unit. The crew called it out. At 1053:04.0 the recorded parameters indicated that the left engine power lever(1) was progressively retarded to flight idle. At 1053:24, the left engine condition lever(1) was set to fuel shutoff position resulting in left engine(1) shutdown. Between 1053:12.6 to 1053:18.8, several stall warnings sounded. Flight crew declared an emergency at 1053:34.9 and reported an engine flame out. At 1054:09.2, flight crew called several time for engine restart. At 1054:20 the recorded parameters indicates a restart of the left engine(1). At 1054:34.4, master warning sounded, 0.4 seconds later, CVR recorded unidentified sound. Both recorders stopped recording at 1054:36.

Am I interpreting this correctly? They have a problem on #2 and retard power on #1 and then shut it down?

Quoting flylku (Reply 11): Am I interpreting this correctly? They have a problem on #2 and retard power on #1 and then shut it down?

Yep, another Kegworth style event.

Quoting flylku (Reply 11):

3 Capt in the cockpit....... ! WTF

Quoting xdlx (Reply 10): Back to ME school 101.... Eng OUT.... 1) MAX PWR 2) Identify 3) VERIFY 4) Feather

The ATR72-600 is equipped with ATPCS (automatic takeoff power configuration system) which is armed when power management is in TO mode. ATPCS provides automatic feathering of a dead engine, torque uptrim of the good engine to 100%, and auto rudder trim to counter asymmetric power. My understanding is that power management would normally be switched to CLB at around 400-500 feet, at which point ATPCS protections are automatically disarmed.

According to Zeke's timeline, power management would likely still have been in TO mode at the time of number 2 flameout, with a few seconds remaining before CLB selection. Does anyone know how a switch to CLB while ATPCS reconfiguration was taking place (there's a lag of 2.15 seconds for feathering) would affect the ATPCS reconfiguration. There was a right turn, which would be consistent with number 2 not being (fully) feathered - but the right turn is also consistent with the SID?

Quoting tailskid (Reply 4): It appears that the PF was intentionally flying down the river valley, avoiding high ground on the left and right. That's if there was a PF - there were three Captains on the FD, all three may have been managing the restart of #1.

Quoting xdlx (Reply 13): 3 Capt in the cockpit....... ! WTF

When AF 447 happened I stated that it might have gone down, because there were too many cooks on the kitchen, when in emergency situation sometimes having more personnel on the cockpit is down right dangerous, and we have seen it so many times because the CRM goes out the window and most of the time the extra pilots give more input/distract/act and make things confusing. Amsterdam, AF447, Aeroflot, etc etc etc...

I bet the CVR will show that there was a lot of confusion/supositions and actions taken without following procedures simply due to the fact of too many hands on the deck...

TRB

Quoting TheRedBaron (Reply 15): in emergency situation sometimes having more personnel on the cockpit is down right dangerous

I'm gonna be rebutting this with UA232.

It all comes down to proper CRM and utilizing the resources at hand effectively.

Interesting report in the Taipei Times citing the ASC, which suggests some systems issues: http://www.taipeitimes.com/News/front/archives/2015/02/07/2003610992

According to the report, there was a flameout warning for no. 2, but the engine had not actually flamed out. The engine autofeathered nevertheless.

Quoting Kaiarahi (Reply 17): According to the report, there was a flameout warning for no. 2, but the engine had not actually flamed out. The engine autofeathered nevertheless.

Lets nip this in the bud, go back to the source

http://www.asc.gov.tw/upload/cont_at...e8-7a33-4d04-a3fc-cd43f5451b77.pdf

It is not correct.

Quoting TheRedBaron (Reply 15): because the CRM goes out the window and most of the time the extra pilots give more input/distract/act and make things confusing. Amsterdam, AF447, Aeroflot, etc etc etc...

AF447: when they lost control, there were only two pilots on the flightdeck.

THY 737 at AMS: I don't think the observer pilots made any actions or distracting comments during the accident sequence.

Aeroflot: Which one? Of the relatively recent, well known Aeroflot accidents (A310, 737 Aeroflot nord) I don't know that either had anything to do with too many pilots. Indeed on the A310 it was too few, as the guy handling the aeroplane was not a pilot.

Quoting Kaiarahi (Reply 17): Interesting report in the Taipei Times citing the ASC, which suggests some systems issues: http://www.taipeitimes.com/News/front/archives/2015/02/07/2003610992 According to the report, there was a flameout warning for no. 2, but the engine had not actually flamed out. The engine autofeathered nevertheless.

Seems like it was an in commanded auto feather situation. On the Dash 8 we had a drill for that and it's was treated the same as an engine failure. We also did failure recognition drills during sim so we knew what to look for during a failure. Every failure would show different indications i.e. Torque, ITT, rpm.....

This seems to be a failure to recognize the failure....

Quoting Kaiarahi (Reply 17): but the engine had not actually flamed out.

According to the diagram, the ITT (Inter Turbine Temperature) has a drop first. This happens at the same time the master warning ENG 2 flameout is coming on. To me it looks like a ENG 2 failure or flameout.

Later they shut down ENG 1.

Maybe translation errors at work?

Quoting whiteguy (Reply 20): Seems like it was an in commanded auto feather situation.

#2 flame out with, with low ITT, N, and no FF on #2 would not indicate a false positive in my book.

Quoting zeke (Reply 18): Lets nip this in the bud, go back to the source

My bad. Thanks.

Quoting zeke (Reply 22): Quoting whiteguy (Reply 20): Seems like it was an in commanded auto feather situation. #2 flame out with, with low ITT, N, and no FF on #2 would not indicate a false positive in my book. We are addicted to our thoughts. We cannot change anything if we cannot change our thinking – Santosh Kalwar

Yeah I agree, after looking at the FDR chart again.....

According to the approximate "unofficial" event chart posted by rcair (thread 2 post 171); it seem like engine #1 was cut off about 32 seconds after engine #2 failed. To me, it seems like there was no exceptional stress in shutting down the #1 engine. Even the mayday call came after engine #1 was cut off, which looks to me like the situation was under control after engine #2 had failed.

Is there clear evidence that engine #1 was fully operational. Even though it doesn't make sense, I would have more understanding for an engine #1 shutdown in case of exceeding engine parameters in oil, engine temp or vibration etc.

Aircraft accidents are usually a bit more complex than what seems to be the case here. Any facts pointing in the direction of an actual dual engine failure or any other combination of failures?

Quoting cuban8 (Reply 25): According to the approximate "unofficial" event chart posted by rcair (thread 2 post 171);

Instead of twitter, you can look at the investigators link I posted in reply 18.

Quoting cuban8 (Reply 25): Any facts pointing in the direction of an actual dual engine failure or any other combination of failures?

No, with the attempts to restart engine #1 tends to indicate they realized they shut down the wrong engine, and just started it again before impact.

Quoting whiteguy (Reply 20):

I am sure everyone can remember in basic IFR training VORs have no idea which way you're headed. Once a pilot can remember that, then things become easier, and will not get trap into a term know as reverse sensing scenario.
Could it be the FP or the crew in general of this accident although by all accounts navigation was not an issue in this case, could have been trap in a similar situation and shut down the good engine instead of the problematic engine? Would it be appropriate to classify this event as Misinterpretation of Engine Electronic Mechanical Data (MEEMD?)

[Edited 2015-02-07 12:24:50]

I had some instances on the ground, where something was wrong with the plane, but the pilots started switching around, until I told them to stop and let me do some analysing of the problem. E.g. once we came into SHJ very early in the morning from KDH and left our 747-400 freighter with a local engineer before going to the hotel. When we came back in the evening, there was a christmas tree of fault messages on EICAS, the loadmaster complained that she could not open the cargo doors and the pilots started random switching in the cockpit. I told them to stop and went through the fault messages. I soon noticed an AIR / GRD system fault, so I went downstairs to check the nose landing gear. I found that the strut was almost completely extended and the two AIR / GRD porox. switches on the scissors had engaged into air mode.
As a result the avionic cooling overboard valve had closed (the plane thinking that it was in the air) and the avionics cooling system went to recirculation mode. Not a good idea at 50 centigrades ambient temperature (which probably also caused the massive NLG strut extension). The result was that the avionics boxes started cooking (they were too hot to touch) and the computers resigned, producing all the fault messages. Also the air signal caused power to be removed from the cargo door mechanism. Similarly the APU start was inhibited (B-747 APU can be started on ground only).

So I disconnected the scissors and tied them up with some rope to get the prox switches back into ground mode and then I ordered two ground aircon units. One I connected to the aircraft aircon distribution system to cool down the interior and the hose of the other one I ran through the E/E bay door into the E/E compartment to cool down the boxes (actually the captain was very helpful, he crawled in there and aimed the hose at the avionics racks). In meantime the loadmaster had opened the cargo door and they had loaded a few containers into the forward lower deck, so that the NLG strut got compressed and I could re-engage the scissors.
After about two hours of cooling the fault messages began to disappear one after the other, so that eventually we could fly home. Fortunately no boxes got fried.
The local guy in meantime stood around without having a clue.

But the problem was that the guys first started with their routine flight preparation and then, when it didn't work, just put on switches to see what was happening instead of checking the fault messages. Ok, we were on the ground, so it was not dangerous.

Jan

[Edited 2015-02-07 13:12:21]

Quoting xdlx (Reply 13): 3 Capt in the cockpit....... ! WTF

2 Captains together is bad enough.

Quoting zeke (Reply 22): #2 flame out with, with low ITT, N, and no FF on #2 would not indicate a false positive in my book.

What is the threshold for the flameout warning?

Does the flameout warning literally mean the flame is out?

Or do certain conditions trigger it, even though the flame may not literally be out?

I assume the latter?

My 2 cents might not be worth a plugged nickle, and I think this was basically stated in an earlier thread but a valuable piece of advice I was given in flight school was;

"The first thing to do in the event of an emergency is.....nothing...because doing the wrong thing can be much worse than doing nothing"

That doesn't mean one should just do nothing, but it can give people in command enough time to figure out what the right thing actually is.

Unfortunately, even though they may have flown thousands of hours perfectly, pilots often don't get a chance to make a second mistake.

Unlike in most jobs, there are some things pilots must do perfectly every single time, no mistakes ever, or people die.

Quoting JoeCanuck (Reply 31): That doesn't mean one should just do nothing, but it can give people in command enough time to figure out what the right thing actually is.

Looks like they actually took about 46 seconds to decide to shut off #1?

22 seconds after the warning, they decide to reduce power to #1
24 seconds later, they decide to shut off #1

http://en.wikipedia.org/wiki/TransAs...ht_235#6_February_press_conference

Quoting TheRedBaron (Reply 15): when in emergency situation sometimes having more personnel on the cockpit is down right dangerous, and we have seen it so many times because the CRM goes out the window and most of the time the extra pilots give more input/distract/act and make things confusing. Amsterdam, AF447, Aeroflot, etc etc etc...

Quoting namezero111111 (Reply 16): I'm gonna be rebutting this with UA232.

And with QF 32.

Quoting Kaiarahi (Reply 17): According to the report, there was a flameout warning for no. 2, but the engine had not actually flamed out. The engine autofeathered nevertheless.

Quoting zeke (Reply 18): It is not correct.

Quoting zeke (Reply 22): #2 flame out with, with low ITT, N, and no FF on #2 would not indicate a false positive in my book.

Based on the prelim FDR data graph released, I find it hard to understand how there was a flame out on engine #2, for several reasons:
Looking at #2 engine parameters:
* ITT starts to drop 1.5 - 3 seconds AFTER the Master Warning: ENG 2 Flame-Out;
* ITT drops and stabilizes at around 400C. This is inconsistent with a flame-out condition (should have been lower, and continuously dropping until impact due cooling effect from windmilling and unburned fuel vapourizing- fuel flow did not stop);
* NH, NL and WF only start dropping 1.5 - 3 seconds AFTER the Master Warning;
* NH was around 95% (the scale is not accurate enough to reed within 1%) which seems to correspond with normal take-off power.
NH drops to around 80%, and stabilizes there. To me, it seems very extremely unlikely that NH could stay anywhere near 80% in an engine flame out event. 80% NH would be around Flight Idle (FI) power . . .
* Engine oil pressure was normal throughout the event until impact (oil pumps are driven by NH spool).

Looking at the #2 engine parameters, all parameters suggest a rollback of power to FI at the start of the event, consistent with an MFCU (Fuel Control) commanded power reduction. There is nothing in these parameters that suggest any internal engine problem; no hot section problem (like turbine blade failure), no compressor problem (like birdstrike), no stall, no icing, nothing. The important engine parameters behave as in a normal power reduction sequence.

There is nothing in the published engine parameters that suggest an engine hardware problem.

So it seems like a normal power reduction, with the IMPORTANT exception:
* it was not commanded by the Power Lever (PLA);
* it was not commanded by the Condition Lever (CLA);
* it seems that it also was not commanded by the auto throttle, since Commanded Torque (Objective Tq, as displayed in the FDR chart) was normal throughout the event until impact.

So what was it that commanded the power rollback on #2?
Something apparently triggered the ENG 2 Flame-out warning logic. To me, that something was not the actual engine performance (based on the released FDR data graph). I suspect that the answer should be found in the ENG Flame-out logic.

I have no expertise here, as it is not an engine system (my field of expertise is limited to the engine side). Flame-out warning logic is not designed by P&WC, and it is not part of the (many) parameters and functionalities tested in the engine Test Cell. I therefore assume that it must be an ATR algorithm, buried somewhere in the airframe systems software architecture. While I have deep knowledge of the engine internal hard ware, reasonable level on its control systems and architecture, I have only limited knowledge with the airframe side.

What I did find to be remarkable, is both engine #1 and #2 showed an initial torque increase at event initialization. Shortly after this, engine #2 Tq goes down all the way. This could be an important clue, that I haven’t figured out yet. Here is me thinking out loudly . . .

In this respect, it may be important to understand that the PW127M engine has two fully independent Tq measurement channels: One channel reports to the engine (Electronic Engine Control - EEC), the other reports to the airframe/cockpit. One of these channels (can’t remember which one) communicates with the opposite engine EEC. The auto uptrim (increase of power in case of opposite engine failure) is controlled by the EEC. I also believe that this logic controls the autofeather functionality and communicates with the propeller controller (not an engine control, this is again airframe side)

It may be possible that an error in this channel could activate automatic uptrim and associated autofeather on the opposite engine. This would be engine commanded, not airframe commanded. Perhaps this is somewhere embedded in the ENG Flame-out logic. I can also see that EEC-commanded power would not be equal to the Commanded Torque in the FDR graph, as the latter may be airframe commanded power, communicated to the EEC. If the EEC decides for whatever reason not to accept that command, and overrule, that may not show up on the Commanded Torque in the FDR.
But then again, I’m not familiar with the FDR channel for Commanded Torque, so I do not understand if this is parameter is as commanded by airframe, or by engine EEC.

Quite puzzling at this point why #2 engine rolled back.
Even more puzzling why the crew pulled #1 engine lever.

Rgds,
PW100

Quoting Aesma (Reply 183): **Previous thread** Unless the story of the whistleblower is correct and, let's assume, engine 1 was was acting up on the previous flight. The pilot might have assumed that same engine is now broken and without checking anything shuts it down.

.
.
.

Quoting ComeAndGo (Reply 8): Unless the story of the whistleblower is correct and, let's assume, engine 1 was was acting up on the previous flight. The pilot might have assumed that same engine is now broken and without checking anything shuts it down.

Interesting thoughts.

In addition to this, would the auto-everything in case of engine failure (autofeather, auto-rudder trim etc) remove important cognitive clues as to which engine failed? Not being a pilot, but I understand in the “old days” the drill for twin engine turbo prop was, kick rudder to counteract asymmetric power, increase power on the good engine, isolate failed engine and feather prop. Just the yawing and required hard stepping on rudder provided for very strong cognitive clues and would remove any potential doubt as to what engine had failed.

Rgds,
PW100

Quoting namezero111111 (Reply 16): I'm gonna be rebutting this with UA232.

Some emergencies have plenty of time and some others you don't have it, in fact those with time they have been bale tu run through checklists etc, so although I see your point the situation is different,because of the time, pause of flight and altitude (hence time)

Quoting bueb0g (Reply 19): AF447: when they lost control, there were only two pilots on the flightdeck.

Correct but they got into a stall situation, and the return and extra personnel on the cockpit in my opine contributed to the loss of CRM and in 3 minutes they were in deep doodoo, so same logic applies here a lot of altitude and not much time.

Quoting bueb0g (Reply 19): THY 737 at AMS: I don't think the observer pilots made any actions or distracting comments during the accident sequence.

Also Correct, but the extra personnel in the cockpit "probably" changed the outcome, since they all did not notice the throttles go to idle with disastrous results.

Even British Midland did not have the altitude nor the time to correct the mistake (in this case there were 2 on the cockpit) but they did not followed procedure, and my point is that sometimes having extra guys in the cockpit , changes the way things are done/procedures. due to extra inputs or suggestions in the cockpit.

In the end as most accidents, they are a long chain of mistakes that together form a tragedy.

TRB

Quoting LTC8K6 (Reply 30): What is the threshold for the flameout warning?

I do not know on the ATR.

Quoting LTC8K6 (Reply 30): Does the flameout warning literally mean the flame is out?

Yes, a flameout is a condition where the combustion process within the burner has stopped.

Quoting PW100 (Reply 34): Based on the prelim FDR data graph released, I find it hard to understand how there was a flame out on engine #2, for several reasons: Looking at #2 engine parameters: * ITT starts to drop 1.5 - 3 seconds AFTER the Master Warning: ENG 2 Flame-Out; * ITT drops and stabilizes at around 400C. This is inconsistent with a flame-out condition (should have been lower, and continuously dropping until impact due cooling effect from windmilling and unburned fuel vapourizing- fuel flow did not stop); * NH, NL and WF only start dropping 1.5 - 3 seconds AFTER the Master Warning; * NH was around 95% (the scale is not accurate enough to reed within 1%) which seems to correspond with normal take-off power. NH drops to around 80%, and stabilizes there. To me, it seems very extremely unlikely that NH could stay anywhere near 80% in an engine flame out event. 80% NH would be around Flight Idle (FI) power . . . * Engine oil pressure was normal throughout the event until impact (oil pumps are driven by NH spool). Looking at the #2 engine parameters, all parameters suggest a rollback of power to FI at the start of the event, consistent with an MFCU (Fuel Control) commanded power reduction. There is nothing in these parameters that suggest any internal engine problem; no hot section problem (like turbine blade failure), no compressor problem (like birdstrike), no stall, no icing, nothing. The important engine parameters behave as in a normal power reduction sequence.

Modern engine electronic control units will try to autorelight the engine, it would not surprise me it will try and do a autorelight at flight idle. The conditions for the master warning may not be listed in that FDR readout, the engine has a lot more sensors than what we see in the cockpit.

The -600 has a glass cockpit for engine displays, they would have had TQ, NP, and ITT gauges, all three of these would have shown lower parameters than engine 1 power level was reduced to IDLE. You can see they advanced the #2 power lever before shutting the fuel off to #1, however there was no associated engine acceleration or power to the prop.

I do not agree that this is a normal power reduction, the torque and beta values show that the power was not going to the prop. The PW127M is a 3 shaft design, it could also mean a failure of the third shaft which couples the power turbine to the reduction gearbox for the prop.

Quoting LTC8K6 (Reply 32):

It looks like they had enough time to figure out a proper course of action, and if they had just done nothing except continue to climb out, then we probably wouldn't be talking about an accident now.

I'm more then willing for a final report to come out before pointing fingers but it's difficult to imagine a scenario where killing the one working engine, isn't a terrible lapse of judgement.

It is very important to know 'why' they did what they did, not just that they, likely, made tragic mistakes.

Quoting PW100 (Reply 34): So what was it that commanded the power rollback on #2? Something apparently triggered the ENG 2 Flame-out warning logic. To me, that something was not the actual engine performance (based on the released FDR data graph). I suspect that the answer should be found in the ENG Flame-out logic.

Hmm. Interesting.

Here's a thought. One engine had been changed recently. What if wires in some of the automated features somehow got switched the wrong way in that change? E.g., it really was engine 1 that failed, and the pilots did shut down the correct engine. But the automatic systems auto feathered the wrong engine, the other one.

Not sure if the engine FDR supports this though. IIRC engine 1 was in a correct operating model according to the parameters, before it was shut down by the crew.

Quoting Kaiarahi (Reply 17): According to the report, there was a flameout warning for no. 2, but the engine had not actually flamed out. The engine autofeathered nevertheless.

It could Be, I have seen that before, but I don't remember exactly what maintenance told me about that, I think it was a problem with some sensor in the engine, Nh or TQ.

Quoting xdlx (Reply 13): 3 Capt in the cockpit....... ! WTF

Not usual but one captain under training on the left, instructor on the right and examiner on the jump seat, to check the intructor or both.

On this kind of flights the first step is setting up the roles and making it clear to everybody in cockpit and in the back (Cabin crew), to avoid problems, I'm really looking forward to see how did they manag the situation, procedure and CRM.

Quoting Kaiarahi (Reply 14): The ATR72-600 is equipped with ATPCS (automatic takeoff power configuration system) which is armed when power management is in TO mode. ATPCS provides automatic feathering of a dead engine, torque uptrim of the good engine to 100%, and auto rudder trim to counter asymmetric power. My understanding is that power management would normally be switched to CLB at around 400-500 feet, at which point ATPCS protections are automatically disarmed.

Normally the "Climb Procedure" is done at 400 feet AGL, but for difficult airport can be higher to avoid terrain (Taipei could be one of them),

Quoting LTC8K6 (Reply 32): Looks like they actually took about 46 seconds to decide to shut off #1? 22 seconds after the warning, they decide to reduce power to #1 24 seconds later, they decide to shut off #1

The 22 seconds after the warning are not uncommon, but the 24 seconds after are too much, it seems for me that they're discussing.
On the -600 the visual presentation of the messages of flame out, uptrim and Autofeather are very good and easy to recognize, clear messages with colors, its difficult to made a mistake.

It's a tragedy, but I hope we can learn and correct.

Quoting TheRedBaron (Reply 36): Also Correct, but the extra personnel in the cockpit "probably" changed the outcome,

I think that's a difficult to measure, because the opposite metric - how many accidents did not happen / ended uneventfully due to a third crew member - is missing.
It'd be like trying to measure, for example, how many engine failures a new maintenance program prevented.

Quoting TheRedBaron (Reply 36): In the end as most accidents, they are a long chain of mistakes that together form a tragedy.

  

Quoting AirlineCritic (Reply 39):

*pure speculation*
The most likely candidate would be probably the EEC "crossfeed".
After PW100's post I was also thinking something along those lines, as that would apparently not show up on the FDR either.

Surely they have listened to the entire CVR by now, which should give them some clues as to whether a scenario like that could have unfolded. If they were reasonably sure that something like that would be a possibility, we should have seen an emergency AD of some sort by now though.

[Edited 2015-02-08 00:39:28]

Quoting zeke (Reply 37): The PW127M is a 3 shaft design, it could also mean a failure of the third shaft which couples the power turbine to the reduction gearbox for the prop

Problem with that is, that any type of failure in the third shaft (Power Turbine shaft NP**) will show significant effects on NL spool speed and ITT due to internal aerodynamic unbalances (disturbed component efficiencies), which changes turbine and compressor mapping. And this effect is not there. I can dream the interactive relations between NH, NL, ITT, WF and NP on these machines. Spool speed matching is what I do for a living . . . 

NP speed can be Propeller Speed, or PT Rotor Speed (power Turbine). They are mechanically connected through the RGB at a fixed ration of 16.67. Under normal operation, the PT speed is fixed at 20000 rpm (=1200 rpm prop speed),.

Note that there is no mechanical linkage between Power Turbine Rotor (NP), Low Pressure Rotor (NL) and High Pressure Rotor (NH). The linkage is purely aerodynamic, and is determined by compressor and turbine efficiencies. At engine build, the performance engineer can play around with effective vane area's to control the relative speeds of NH and NL in relation to power output. Once those vane areas are determined, and the engine is assembled, NH and NL spool speed reactions are pretty predictable. And to my eyes, the FDR graph is pretty clear; no signs of internal failure.

Rgds,
PW100

Quoting PW100 (Reply 42): And to my eyes, the FDR graph is pretty clear; no signs of internal failure.

Good to know, sounds like a sound explanation.

I still cannot figure out how the shut the wrong engine down, surely the beta would have given enough yaw to tell them which side is not generating power.

Quoting zeke (Reply 43): I still cannot figure out how the shut the wrong engine down, surely the beta would have given enough yaw to tell them which side is not generating power.

I agree. Perhaps those answers could be found in the Engine Flame-out logic.
Further, I was contemplating to what extent the auto rudder trim would possibly remove some of the natural cognitive clues of engine failure yawing? But I can't offer any opinions here as I have never flown any twin. My experience is limited to 0 and 1 engines . . .  

Quoting PW100 (Reply 44): I was contemplating to what extent the auto rudder trim would possibly remove some of the natural cognitive clues of engine failure yawing?

On some aircraft even if the flight control systems can fully compensate for yaw in an engine out situation they only partially do, specifically for this reason. In other words, there is an introduced side-slip towards the live engine for identification purposes. Not sure however if the ATR would fully compensate for the asymmetrical yaw and remove this feedback.

TransAsia Cuts 122 Flights as Taiwan Orders ATR Pilot Tests

http://www.bloomberg.com/news/articl...s-as-taiwan-orders-atr-pilot-tests

All 71 of TransAsia’s ATR pilots must undergo training and oral and written tests on emergency procedures within the next four days, CAA Director-General Tim Lin said Friday at a news conference. Pilots, who will still be able to fly even if they haven’t yet taken the tests, will later have to practice in simulators, Lin said.

I'd like to know exactly what is being emphasized in their re-training....

Quoting zeke (Reply 43): surely the beta would have given enough yaw to tell them which side is not generating power.

If you look at the flight trajectory after liftoff from the runway, I don't think they were adequately correcting for the asymmetry. They just kept turning towards the failed engine until they shut the other one and started veering to the other side...

I think they weren't 'on the ball'. In more ways than one.

Lay man's question after Kegworth and this one: Isn't it possible to have some sort of warning mechanism if a pilot tries to turn off engine X in a situation where there is an alert for engine Y? So that the pilot is forced to double check and/or has to override a security mechanism?

Quoting vfw614 (Reply 48):

The security mechanism is there. It's called "the other pilot". Whenever you have an engin issue, whether it's an engine failure, fire, torque runaway, etc. the crew should confirm with each other. Example:

Captain: "We have left engine fire, confirm?"
F/O: "Confirm"
Captain: "Left engine shut down drill"
F/O"Left condition lever, confirm? (points to it)"
Captain "Confirm"
F/O "Cut off..." (pulls the actual lever)

etc etc etc.


It's easy to be an armchair pilot, but I've done these drills many many times in the sim. (B1900, B200, LR35). Things get stressful quickly in a cockpit, but that's why we do these rigorous training sessions.

Rule number one in aviation: Aviate, Navigate, Communicate. Under normal circumstances, even an engine failure above V1 should be manageable. These guys were what, 1200 feet AGL when this happened, with the gear and flaps up? You have plenty of time to do things properly. The only thing I would consider "DIRE" and needs attention ASAP is if you have a V1 cut and no auto-feather. Then you need to get on that ASAP, but still confirm you're actually feathering the correct propeller.
Just my two cents...



[Edited 2015-02-08 05:47:21]