PPRuNe Forums - CRJ down in Sweden

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)

SHK has shared some FDR and CVR data in this statement.... SHK webaite: http://www.havkom.se/en/

Quote:

Well from that admittedly limited information it sounds like some dramatic structural failure..

I do not read the comments that way at all. Seems as if an IRU failed and started the nose pitching down and perhaps the zero g caused momentary loss of oil pressure. The real question seems to be what was the crew seeing on their PFDs. There was a comment mid way down the presentation that seems to indicate that the crew tried to re-engage the autopilot!!!??? (If true, then that was a big part of the problem.)
Total duration of the event was about 90 seconds.

Final report is due in December. Lets hope that a lot of the details get filled in before then.

Machinebird

I read both the Swedish and English statements and nowhere did I find any reference to autopilot being re-engaged unless there are some subtleties about Mach trim that is linked only to autopilot re-engagement.

I watched the webcast from the presentation of the report. HAVKOM pointed out that four FDR channels (pitch, roll, ground speed and mag heading) were incompatible with other data after the event started. All these channels are generated by the IRU.
So they made a graph showing both the FDR pitch data and their "best effort estimate" of the actual pitch.
The graph shows that the FDR pitch suddenly indicated up. After 2-3 secs the estimate indicates a pitch down which increases, with -1g negative pitch acceleration.

To me, this would fit a scenario where the PF sees a pitch up on the PFD and reacts by pushing down and trimming nose down (as also recorded on other FDR channels). Overspeeding occurs after approx 15 seconds.
Note that the investigators did not confirm such a scenario, but did not deny it either.

Extracted from the preliminary accident report:
http://www.havkom.se/assets/reports/SRL-2016_01e.pdf

Quote:

After one hour and ten minutes of flight, during the pilots approach
briefing, the pilot in command, who was seated in the left seat,
exclaimed a strong expression. Immediately thereafter the aural
warning for autopilot disconnect (named Cavalry Charge) was
activated. The disconnection was also confirmed by FDR data. The
warning sounded continuously during the following 18 seconds.
After the autopilot disengagement FDR data indicates that both left
and right elevators moved towards aircraft nose down position. Left
and right side angle of attack went towards negative values. The
aircraft entered a descent with vertical acceleration values that reached
a negative load corresponding to -1G.
After a few seconds of negative G-load the aircraft's warning system
was activated by a so-called Triple Chime
10
followed by an aural
warning (synthetic voice) for low engine oil pressure on both engines.
FDR data shows that the trim to the movable horizontal stabilizer was
activated manually and the trim position went from -0.9 degrees nose
up to 1.7 degrees nose down. An aural signal of the stabilizer trim
movement (Stab trim clacker), signifying a manual input longer than
three seconds, was activated in connection with this. Immediately
thereafter a warning for high bank angle was activated.
After 17 seconds from the start of the event, the maximum speed
(VMO), 315 knots was exceeded. The over speed warning was
activated and the vertical acceleration turned to positive values.
Another 16 seconds later, the first officer transmitted a “MAYDAY”
message that was confirmed by the air traffic control. The indicated
airspeed then exceeded 400 knots and the stabilizer trim was
reactivated and dropped to 0.3 degrees nose down. The Pilot in
Command called "Mach trim"
11
after which engine power was
reduced to idle.
During the further event the last valid FDR value shows that the speed
continued to increase up to 508 knots while the vertical acceleration
values were positive, with maximum values of approximately +3G.
FDR data shows that the aircraft's ailerons and spoilerons mainly were
deflected to the left during the event.
Radar data and the accident site position indicate that the track was
changed about 75 degrees to the right during the event.
The crew was active during the entire event. The dialogue between the
pilots consisted mainly of different perceptions regarding turn
directions. They also expressed the need to climb. The aircraft
collided with the ground one minute and twenty seconds after the
initial height loss.

Mr. Snuggles,
The interview in Dubai had this comment in it: (please make allowance for google translate) I do not see any such reference in the preliminary report.

Quote:

Eleonor Norgren: There are no ready answers. But the crew regarade that something was not as it should, and coupled the autopilot. The process must have been under a minute and a half.

Machinebird

It is to me very very clear that google translate was not correctly tuned for this translation.

Your quote:

Quote:

Eleonor Norgren: There are no ready answers. But the crew regarade that something was not as it should, and coupled the autopilot. The process must have been under a minute and a half.

is actually already partially translated by me in an earlier post. The original Swedish writing is as follows:

Quote:

Eleonor Norgren: Det finns inga färdiga svar. Men besättningen regarade på att något inte var som det skulle och kopplade ut autopiloten. Förloppet ska ha varit under en och en halv minut.13:21, 9 March 2016

?Besättningen reagerade och kopplade ur autopiloten? - P4 Norrbotten | Sveriges Radio

My translation would absolutely be

There are no ready answers. The crew reacted on something that was not as it should and disconnected the autopilot. The process (/event/incident) should have taken less than a minute and a half.

Clearly a mistake from automatic translation.

MrSnuggles,
Thank you for the clarification. That makes better sense.

Looks like the guy flying was trying too hard to chase an illogical PFD display.

If the plane has been flying in stable flight with the trim set, why would you be running in a bunch of trim manually once control has been dumped in your lap? Zero g in cruise? That can't be good !

Sit back and size up the situation before you set out on your course of action. At least that is what I was taught.:hmm:

Picked this comment out of the report:

Quote:

FDR data shows that the aircraft's ailerons and spoilerons mainly were
deflected to the left during the event.

Looks like aeroelasticity might have sabotaged their attempts to level the wings once the speed became excessive. Assuming aeroeleastic effects (wing twist) were in play, they might still have been able to level wings by neutralizing aileron and applying gentle rudder input.

Not trying to defend or praise the crew as the report is still quite sparse. However, as pilots it's drummed into us to always trust your instruments and not your senses.

Machinbird, as you say going from perfectly trimmed to an illogical scenario in a few seconds should have alerted them to the fact that maybe the instruments were giving dodgy readouts. Surely a quick glance at the standby instruments would've confirmed they were straight and level to start with?

If this had happened in the day or with visible references outside I suspect the outcome would've been far more positive. All the holes lined up on this occasion...

Champ

Quote:

The process must have been under a minute and a half.

Fairly dramatic in any case - my money is still on some sort of structural failure as the trigger event.

Actually the first deviation from normal on the FRD data released was the pitch showing constant nose up with a rate of about 30 degrees in 4 seconds, until almost 90 degree nose up after 22s.
Second sign, A/P disconnect about one second after pitch moving upwards.
Until AP disconnet all other values vas normal, 1G and no pitch or roll commands or movements.
Does not look like structural failure. There is a clear connection between elevator position, AoA and G all the way down.
And its quite clear that the pitch values(from IRSIRU) is corrupt.

Essential information (CVR & FDR) from the preliminary report for everyone's convenience:

CVR

(Throughout the transcript, different alarms, chimes, and warnings are continuously heard)

23:19:22 (CPT): [expletive]
23:19:29 (FO): [expletive]
23:19:30 (CPT): [expletive]
23:19:33 (FO): Come on
23:19:35 (CPT): Come on, help me, help me, help me
23:19:35 (FO): Turn right
23:19:36 (FO): What
23:19:37 (CPT): Help me, help me
23:19:38 (FO): Yes, I'm trying
23:19:40 (FO): Turn left, turn left
23:19:43 (FO): Turn left
23:19:44 (FO): No
23:19:50 - 23:20:09: FO transmits the standard mayday call and repeats it several times
23:19:59 (CPT): Mach trim
23:20:01 (FO): Trim, trim a lot
23:20:06 (FO): Turn left, turn left
23:20:14 (CPT): We need to climb, we need to climb
23:20:15 (FO): Yeah, we need to climb
23:20:16 (FO): Turn left, turn left
23:20:17 (CPT): No, continue right, continue
23:20:19 (CPT): Continue right
23:20:20 (FO): Ok
23:20:22 (CPT): No, help me, help me please
23:20:23 (FO): I don't know, I don't see anything
23:20:24 (FO): I think you are the right to correct [sic]
23:20:25 (CPT): Ok
23:20:26 (CPT): Ok, ok, ya
23:20:28 (CPT): [expletive]
23:20:31 (FO): [expletive]

FDR

(Apparently the recorded pitch angle is not compatible with the aircraft's motion. Recalculation of pitch angle is shown beneath the FDR)

https://i.imgur.com/ScI7umC.png

Recalculated (more likely) pitch angle information from TAS, altitude, and AoA:

https://i.imgur.com/8WukfMI.png

Quote:

Immediately thereafter the aural
warning for autopilot disconnect (named Cavalry Charge) was
activated. The disconnection was also confirmed by FDR data. The
warning sounded continuously during the following 18 seconds.

When either pilot disconnects the autopilot the calvary charge aural will only alert once for about 2-3 seconds.
When the autopilot disconnects automatically because of a failure it will continue to sound the calvary charge aural until a pilot manually selects the autopilot disconnect switch and that will silence the aural.

It sounds to me like the autopilot disconnected automatically.

Runaway trim or a hyd hardover in pitch leading to UA?

:ooh:

From my reading of the report I'd still think that a major failure was the cause of the accident. Sure, the crew were trying to control the aicraft to the bitter end; but so were the Alaska Airlines guys when their stab broke loose. They had zero chance. With any element of control, I'd expect a couple of partial recoveries, yet they went straight down.

I think that the investigators have some idea of what they are looking for. When the weather improves, I think that they will return to the accident site to confirm what they suspect.

Three bad FDR parameters

Wrist watch: Actually, three parameters were listed as not compatible (on p.23 of the report): färdhastighet (ground speed), tippvinkel (pitch angle), and rollvinkel (roll angle).

if they suspect htp runway then it needs to come out a.s.a.p., but also, what could have caused it.

As we can see on the posted FDR-data, pitch information was leading(but most certain, not true).
Elevator pos, trim pos, AoA and G was not affected until after A/P disconnect.
Does'nt look like any other failure but IRU/IRS.

Possible chain of events

So...to elaborate on AAKEE:s earlier summary:

1) A/P disconnect (due to incorrect pitch indication from IRU?).
2) PF pushes nose down 1s later (due to incorrect pitch indication on PFD?).
3) g loading and AOA follows.
4) PF pushes nose down multiple times in coming seconds.
5) Stab trim increases towards nose down pitch.
6) Alt decreases, speed increases until overspeed.
7) At overspeed warning PF pulls nose up (realizes instrument error?) but does not recover (aeroelasticity?).
8) RIP.

If you look at the blue line of indicatet pitch it seems like when captain is diving with an estimated 30 to 40 degrees down the iru in fact is telling him that this is working because its comming down. Then the estimated pitch is reduced. Then it is going way up again and he is diving again to repeat what he is thinking is the correct recovery. Scary!