CRJ-200 down

RobertL18C wrote:

Would the PFD have not shown an ATTITUDE fail display, and the CAS shown an IRU FAIL or MISCOMPARE signal?

If only one IRU failed, one of the PFDs would still read correctly and the failed PFD could be switched to the good IRU?

Yes, and I guess one of things the SHK is trying to figure out is why this didn’t happen.

The final report was published yesterday. Not much new technical information but it did answer one of @RobertL18C’s questions.

RobertL18C wrote:

Would the PFD have not shown an ATTITUDE fail display, and the CAS shown an IRU FAIL or MISCOMPARE signal?

It did, but with the rapid increase in indicated pitch, after a few seconds the captain’s PFD switched to “declutter” mode where much of the information was removed. This included the miscompare flag! However the simulator for the aircraft model did not remove the miscompare flag in the same situation.

As for the second question:

If only one IRU failed, one of the PFDs would still read correctly and the failed PFD could be switched to the good IRU?

The operator didn’t have any procedures for callouts in connection with abnormal attitudes so there was most likely a split between the two pilots’ understanding of the situation combined with cognitive overload. There was no effective communication between the pilots until 13 seconds after the start of the upset. At that point, the likely presentation on the two PFD’s was this:

The cause of the accident, according to the investigation, was that “operational conditions were not sufficient to catch a fault in a redundant system.”
Contributory causes were given as

• There was no effective system for managing and communicating warnings or emergencies.
• The instrument system gave insufficient guidance on the resulting malfunctions.
• The initial manoeuvre that led to negative load is likely to have affected the pilots’ ability to handle the issue rationally.

The investigation also had some criticism of the SAR operation.

This is the panel of the airplane:

There are three artficial horizons. One of them broke. Autopilot went off due to mixed incoming signals. Captain thought the plane was going up looking the wrong data displayed in front of him. He corrected the attitude according to wrong data and the plane dived. Looking at all of these or crosschecking would have shown one artificial horizons displaying differently. If the captain had noticed this he would have probably got level again trusting the two out of three. He probably never checked them. Co-pilot said “Come up” seeing the correct attitude in front of him but the captain did not recover probably because he was looking the big artificial horizon in front of him showing wrong attitude. There was very little time to recover and realize what is going on after the plane was put to dive and trimmed according to the wrong data.

AirV wrote:

Looking at all of these or crosschecking would have shown one artificial horizons displaying differently.

Yes. This is not the first time that a fully functional aeroplane was flown into the ground because the one faulty instrument of three was trusted instead of the other two. Without thinking much, I remember Birgenair flight 301 (https://en.wikipedia.org/wiki/Birgenair_Flight_301) and Alitalia flight 404 (https://en.wikipedia.org/wiki/Alitalia_Flight_404).

Now it would be easy to say “this can’t happen to me”, but in the middle of the night doing the third of fourth sector of a long day? Half asleep, occuied with something different (over that part of the world, 500NM direct-to-legs with a frequency change every quarter hour are normal). Suddenly the autopilot disconnects, you look startled at the AI in front of you which shows all-sky and you instinctively do what you always trained… After a couple of seconds both primary instruments are off their limits in opposite directions. A nightmare! Who thinks about the standby instrument in this situation, an indicator you will (or will not) have used on a few occasions during your simulator trainings. I am pretty sure that quite a few flight crews would have crashed in that situation as well.

This would seem to be a lesson to have two AIs both in the primary field of view, so a malfunction of one is fairly obvious.

I got a milder version of this coming back from Colmar. The KI256 AI would itself up into a ~10deg UP indication, which pitched the autopilot down (didn’t disconnect it) but I got the ALTITUDE warning when a few seconds later the plane had departed from the preset altitude by 200ft. In IMC there would have been some potential for the wrong action to be taken. The condition lasted about 10-20 mins.

what_next wrote:

Yes. This is not the first time that a fully functional aeroplane was flown into the ground because the one faulty instrument of three was trusted instead of the other two. Without thinking much, I remember Birgenair flight 301 (https://en.wikipedia.org/wiki/Birgenair_Flight_301) and Alitalia flight 404 (https://en.wikipedia.org/wiki/Alitalia_Flight_404).

Interesting observation but I think those cases were harder in that there were only two instruments (ASIs in Birgenair’s case, ILS receivers in Alitalia). While in either case, crosschecking with other instruments to find the culprit was theoretically possible, it was hard, particularly in the Birgenair accident as the A/P was reacting to the faulty instrument’s input. I’m also reminded of the Bandeirante accident out of Leeds.

In this latest case with the CRJ200, it seems that two of three instruments were functioning normally. Thus I think the SHK’s recommendation is spot on:

Ensure that a general system of initial standard calls for the handling of abnormal and emergency procedures and also for unusual and unexpected situations is implemented throughout the commercial air transport industry.

bookworm wrote:

Interesting observation but I think those cases were harder in that there were only two instruments (ASIs in Birgenair’s case, ILS receivers in Alitalia).

Birgenair had three ASIs (one standby). The crew did their usual callouts during the take-off run and noticed already then that they were not in agreement. Yet the captain decreed that his one was the good one… The Alitalia flight had only two ILS receivers and again, the captain decided that his one must be the good one. Cross checking with DME readings, sort of a third instrument (which I am sure was part of their SOPs just as everybody else’s), did not even cross his mind.

bookworm wrote:

Ensure that a general system of initial standard calls for the handling of abnormal and emergency procedures and also for unusual and unexpected situations is implemented throughout the commercial air transport industry.

I guess we will see something like this coming soon. But not easy to implement in an upset situation where every second counts. In the end, it will be the captain’s decision which indication to follow. Just like before…

LeSving wrote:

The pilot’s job is to fly the plane, not to be a observer,…

In a modern highly automated airliner the pilot’s job really is that of an observer, like it or not. SOPs of a typical airline call for “positive rate – gear up” and “autopilot – engage” before passing 1000ft AGL. Upon landing, the last action before touchdown is “autopilot – disengage” at minimum or even below, if not autoland. Even in our business operation, on every single flight I have to call out “NON STANDARD: manual flight to FL 100” (above that we are not supposed to fly manually at all) if I want to do any flying myself. And this will be getting even worse in the future. Maybe rightly, because a computer would have compared those three AIs and reacted to the indication of the two correct ones, not the single at fault.

How come the pilots did not spot the increasing airspeed, decreasing altitude, etc?

Maybe glass cockpits make that less obvious, assuming the main PFD attitude has failed.

Perhaps the airframe broke up due to aerodynamic forces, which would happen within maybe 10-20 seconds after pitching down at 60+ degrees.