PPRuNe Forums

PPRuNe Forums (https://www.pprune.org/)
-   Rumours & News (https://www.pprune.org/rumours-news-13/)
-   -   CRJ down in Sweden (https://www.pprune.org/rumours-news/572882-crj-down-sweden.html)

RAT 5 16th Dec 2016 13:34

Oh Dear! Really? Idle descent engine failure is a gotcha for that, but takeoff???

Herod 16th Dec 2016 13:43

Standard brief in my old company: "Confirmed by two or more parameters"

MurphyWasRight 16th Dec 2016 14:20


Would you immediately react accordingly (switch off AP, pull back on power, pull up nose - or whatever you do) or would you say "Oh balls! The fricken computer's on the fritz again!".
This suggests that one problem with current training is the (understandable) focus on situations that -do- require quick reaction such as failure at V1.

The 'muscle memory/quick action critical' learned though is likely carried over to more benign situations leading to immediate reaction rather than contemplation.

Combined with the high reliability of modern systems that diminish the "fricken computer's on the fritz again" thought since failures are rare and net result is what is being discussed on this thread.

Livesinafield 16th Dec 2016 20:30

The testing is a bit of a joke these days anyway for LPC/OPC everyone knows what is happening and they know what is going to happen, V1 cut etc Single engine ILS RTO, all that rubbish.

Don't think i have ever come out of an LPC/OPC and actually learnt anything...just come out glad to have go through the "script"

Pre dep briefs are a total waste of time aswell "before 80 i will do this after 80 we will stop for this etc after V1 we carry on and this happens that happens...how about what happens if both PDF's switch off as you enter IMC at 500 feet ? But we never will talk about that because it will never happen.... because an engine failure exactly at V1 as it is taught will happen ?? right

khorton 16th Dec 2016 22:17


Standard brief in my old company: "Confirmed by two or more parameters"
That is all well and good if you suspect a failure. But in this case no one suspected that the left attitude display had failed.

The attitude miscompute flags on the PFD on the CRJs are much less prominent than I would prefer. They are small amber boxes around amber text a bit offset from the centre of the PFDs. I'd prefer much larger text, right in the centre of the attitude display.

_Phoenix 17th Dec 2016 01:00

I read the report, the essential sections. Something puzzles me
"In a normal seating position both right and left PFD could be seen from both left and right seat respectively without turning of the head."

"The analysis is therefore based on the assumption that the information displayed on PFD 1 was the same information as recorded by DFDR and that the information displayed on PFD 2 is consistent with the first 23 seconds of the simulation. This part of the simulation has been considered to have a very good conformity with the actual event."

http://www.mediafire.com/convkey/1cb...za382fdvzg.jpg

Out Of Trim 17th Dec 2016 03:58

When the PFD 1 failed due to the data miscompare; why is the Autopilot designed to just auto disconnect straight away? If they programmed it to also pause and announce PFD or PITCH CROSSCHECK REQUIRED - without an auto disconnect then they would probably have been able to handle the failure easily and remained in straight and level flight.

Computers can detect mismatched data and compare all sources of that data extremely quickly, so why not use them to announce what has failed to improve situational awareness.

It seems to be that the immediate auto disconnect is what leads to instinctive pilot inputs, as nobody is now flying the aircraft. When no immediate action would be safer.

That may have saved this crew and indeed the AF.

peekay4 17th Dec 2016 19:58

@_Phoenix

t0 is actually at 23:19:20 UTC (2 secs prior to transcript start)

At t15 the Copilot was probably looking at PFD2 and made the correct call.

But then at t17 the Captain pleaded "Help me, help me", so its possible that after this time the Copilot turned his head towards the the Captain and both were looking at PFD1. At t24 the Copilot may have looked back at PFD2 and uttered "No".

@Out Of Trim

The autopilot cannot determine what's being displayed to the pilots. The AP only knows that there were conflicting servo commands. When the AP disconnected, it left the aircraft in a stable, straight and level attitude.

Intrance 17th Dec 2016 21:14

@Out of Trim:

Although that sounds like a plan initially, since the AP will follow FD1 when the LH pilot is PF, keeping the autopilot on will try to follow the FD1 anyway and still go nose down for example. If it switches to another mode like ALT or PTCH it could lead to more confusion ("why is it doing that"). A disconnect does not necessarily do anything to your flightpath. The plane is trimmed, power is set and very little variables change.

In the mean time, there will be at PIT flashing on the PFD and the EICAS should list EFIS COMP MON caution message. These two combined lead to a QRH checklist, which basically lists "compare EFIS and ISI, switch both PFD to output from the reliable data side (IRU2) and proceed with IRU failure procedure". Please note, this is paraphrase and I'm 700/900 rated, not 200. But I can't imagine there being much difference in the basic procedure.

This is also one of the things that surprised me. They should have had an EFIS COMP MON caution message which would have been a clear indicator to take a moment and cross check regardless of whether the PIT was visible or decluttered on the PFD. Cancel caution light, check EICAS, state the message displayed.

Easy in hindsight of course and we can read all the challenges the crew faced. It just seems like this is one that could have been so easily avoided which makes it extra sad and perplexing that it had this outcome.

SE210 18th Dec 2016 10:03

Air France 447, Colgan 3407, Air Algerie 5017 (Swift Air), West Air 294.

Loss of control accidents happening in darkness - involving fatigue.

From my experience as an instructor and examiner, high altitude loss of control scenarios (particular in darkness) are critical. One thing that can not be simulated in the simulator is fear.

It is essential to avoid the fully loss of controls incidents, since the likelihood of recovery in my opinion is 50/50.

alf5071h 18th Dec 2016 12:53

For those who wish to regress to hindsight, or ask why did't they..., I don't' understand..., etc, see the links to documents from Griffith University:-

http://www.icao.int/Meetings/LOCI/Pr...Strategies.pdf

Surprise - Startle
'... an appraisal that a situation is threatening and is beyond the immediate control of the individual'
'... significant impairment in information processing for up to 30 seconds.
... tasks such as attention, perception, situational awareness, problem solving and decision making can be markedly impacted. Communication is often disorganised and incoherent for some time'.


If this crew were severely startled, as is most likely, then none of what was recorded on the CVR and inferred to be communication may be relevant; similarly any inference of action from the FDR. These might be seen as illogical to us now (hindsight bias), but for some unexplained reason they may made sense to the crew at the time. What were those reasons; training, experience, safety promotions, ...
Also in this event there appears to be more than one opportunity for startle. Unexpected engine oil warning, stab trim and bank angle alerts, over speed warning; perhaps a never ending startle effect or at least multiple distractions as to where to direct attention.

Note the aspects in the presentation about 'Individual differences and defences, and organisational problems and strategies, which relate to the industry's problems noted in earlier posts.

http://www98.griffith.edu.au/dspace/...pdf?sequence=1

'... pilots have been surprised or startled by some event, and have as a result either taken no action, or alternatively have taken the wrong action, which has created an undesired aircraft state,
... have been far from optimal in their handling of unexpected events and rather than utilizing their skills, training and knowledge, have underperformed at exactly the time when these skills were most needed'.


Note the latter point, which relates to regulatory assumptions that the pilot will be able to manage such situations, yet research indicates that inherent limited human performance prevents this.

Training is a central issue, not that trainers are underperforming, but that operators are increasingly constrained by regulatory 'ticks in the box'.
Also relating to this are the delays in instigating change partly due to reactive safety management which requires 'evidence' from accident events. Because there are so few extremely rare events, and that each differ in nature, evidence is lacking. Also, if some events are judged on outcome (what happened,) then false patterns can be found - loss of control, which can result in inappropriate safety activity.
Alternative analysis, to the point of speculation, could identify patters which relate to 'why' the accident occurred, considering the contributions of man-machine-environment, such as surprise, alerting system design, certification process and regulatory assumption.
The accident report makes a good effort to move from 'what' to 'why', but falls down with the recommendations; more calls and communication, perhaps because 'startle' is not an engineering 'fact'.

There is little value in individuals, trainers, or operators beating themselves up in forums like this without safety leadership from the top, the regulators. But these organisations appear to out of step with reality of operations; deciding to use reactive safety and looking for 'evidence' from rare events. Evidence which might be found in everyday operations, the normality of operations, expectancy of success, training requirements, overly standardised procedures, and safety by regulation.
Or are the regulators to surprised to realise?

PJ2 18th Dec 2016 16:22

alf5071h, thanks for your views, and the links on "startle".

andrasz 18th Dec 2016 22:54

Excellent post alf5071h, many thanks!


One contributory cause not emphasised by the report but very significant is the fact that the failure happened during the landing brief, when effectively noone was flying the plane. The attention of both PF & PM were elsewhere, this amplified the startle factor.

TowerDog 19th Dec 2016 00:01

Which is why we state during a brief: "You have the airplane".
Somebody is always supposed to be flying, nothing new, old rules.
We get paid well and have earned respect for being able to handle "startle" factors.
That is our job for :mad: sake.
Still not impressed.

Capn Bloggs 20th Dec 2016 06:34

Killed by Automation...
 
What a tragedy; I dread to think what must have been going through those two pilot's minds for the 90 seconds it took to spear into the ground.

In my mind the cause of this is clear: the inability of the captain to resolve the dud attitude indication because of a lack of basic IF flying skill caused by the automation policies of the regulators and probably the company. If the simulation is correct, he was pushing and trimming nose down with the speed on (and eventually through) the barber-pole, and altitude plunging. He didn't have the brainspace to "Say hang on, there might be something strange going on here, what's the Standby showing?".

Additionally, the FO didn't have the confidence to say "look at my PFD, look at the Standby, your PFD is wrong!"

I was bitterly disappointed with the report; the emphasis on the standard calls is completely misguided. Aeroplanes do not fly by mouth, they fly by hands-on. Adding yet another layer of verbal diarrhoea is not going to fix what is a fundamental problem that is easy to fix. Hardly any mention was made of the failure of the crew to fly the aeroplane, nor of the automation policies of the regulator and the company, nor of specific simulator training/practice (or lack thereof) in basic IF.

"Startle effect" has been mentioned. It is difficult to train for this. However, in my view, the major defence is being able to recognise an abnormal attitude without being overwhelmed. This can only be done, in large part, by being skilled in basic IF, so that all your brain space is not consumed by the shock, with some brain space available to step back and "Say hang on, what is really going on here? How can we be 30 nose-up with the speed increasing and altitude descending like a lead balloon?". Spare brain space will also allow a pilot to receive inputs from the other pilot instead of load-shedding all external information.

I suppose we can only be thankful that that aeroplane wasn't full of people...

Towerdog: :=

ATC Watcher 20th Dec 2016 07:38

alf5071h : excellent info . Thank you .

I wonder how many operators have implemented the upset recovery training package (UPRT) designed a few years ago with blessing of ICAO. There was a Notice of Proposed amendment by EASA to that effect in 2015 , but I do not think it is yet mandatory ( but stand to be corrected ).
An advanced version also contained the word ' COMPARE!" before taking any action on the controls If I remember.

Very appropriate here.

RAT 5 20th Dec 2016 09:43

Which is why we state during a brief: "You have the airplane".
Somebody is always supposed to be flying, nothing new, old rules.

"Startle effect" has been mentioned.


IMHO if the former had been adopted the latter would not have happened. There would not have been any 'startle'. The PF, in tune with his instruments, would also have had a in tune 'seat if the pants' and there would have been a conflict of senses; and hopefully a WTF moment and a pause. This would have been true for both LHS & RHS. Easy to say from here, I know, but I think it might have prevented a rushed response. A good lesson to learn.

Timmy Tomkins 20th Dec 2016 11:49

Contributory causes
 
As ever a lot of useful comment on here. I have only flown one jet type with EFIS and liked it for mass info and ease of use but...it does take away the automatic scan that used to be the norm (and often disadvantage) of the old T panel and I feel that was a factor here.

I have had two ADI failures on conventionally instrumented aircraft and they were non events. It was instantly obvious what had happened without needing to think about it; even though one occurred in the first 500 feet of an IMC departure. The procedure to deal with it was also clearly understood - "ADI failure, you have control"and we carried on easily enough.

That the autopilot disconnected didn't help these guys either as they then used useful brain cells controling the a/c and sadly not in the best way but maybe the EFIS design was part of the reason why. The investigators seem to think so.

Bring back the scan as part of training procedures? Get into the habit of cross checking with the gyro standby during departure and approach for example, so that at least some form of cross checking becomes automatic.

Olympia 463 20th Dec 2016 15:30

I am in awe of the knowledge and interpretation skills here. I think that the key thing to be learned out of all this is to always 'fly the aeroplane', but if you get so little time on the stick these days it may be impossible to keep up your skill. As a glider pilot (don't all laugh) and instructor (now retired) I was lucky, I made 2200 sorties and flew nearly 1000 hrs, every second of it hands on. Solly was a glider pilot too and you can see where that got him. Maybe compulsory gliding needs to added to the training syllabus. For glider pilots every landing is a CFIT.


All times are GMT. The time now is 17:12.


Copyright 2018 MH Sub I, LLC dba Internet Brands. All rights reserved. Use of this site indicates your consent to the Terms of Use.