Air Algerie loses contact with its plane leaving Ouagadougou

[Machinbird]

The French readers have had an advantage on the English readers up until the last week.
The English version of the interim report is available here:
http://www.bea.aero/docspa/2014/ec-v...0724.e1.en.pdf

The thing that jumps out at me is that the autopilot modes were changed a number of times after the initial departure from controlled flight. I cannot imagine trying to recover from a departure from controlled flight using any autopilot mode, so it seems that there was an autopilot cripple at the controls.

Seems like the loss in airspeed was due to icing of the EPR sensing system causing loss of thrust which was not handled by sufficiently lowering the nose.

With the data provided, there is a lot to pick apart.

[Machinbird]

IGh
Apologies about the confusion regarding engaging autopilot. Yes, it did not re engage. What I saw was this:
Someone in the cockpit was an autopilot commando with a broken sword. Worse than useless action in a departure from controlled flight.

[vmandr]

page 10 of the BEA report, shows significant difference between flight plan route and ATC clearance (re-route via EPEPO). Seems was not questioned by crew. It also seem wx was better via NY-UM608. RIP.

[Naali]

epr indications

MD80 Epr indications may be very interesting when Pt probe gathers ice at altitude. Normally,it does not need warming,but in certain conditions it may freeze up,even at altitudes. EPR-clocks go pretty wild then... Ref. Roulis at dash245.

[Naali]

Perhaps someone with more knowledge , would like to explain to me what means in interim report that at some time,something reverted to Mach mode for 30 seconds and still wasn,t holding that mode. .74 sounds reasonable at level but fluctuations do make my mind wander...of what happened in there,in the cockpit. .Changes with A/T,would be enough of a reason to get attention. Speed changes from .742 to .740,would cause throttle movements,likely to be observed. MD80,s autothrottle-coupled autopilot, has a bad ? programming,- to react without polarisation or expectations ,so it doesn,t use throttle as human would do.It sure tells the passengers also,that we move these things with power,fluctuating from none to all. I had much work to understand MD80,s Flight guidance systems,and by time learned to override some of them,to revert back to basics..Three miles equal thousand feet as anywhere ,disregarding Vnav ...So we humans might still be able to look farther forward than anything else. No one wants to have an accident to happen,and every pilot wants to fly,any airplane he or she gets trapped on. Sadly,we make mistakes of our own. Human mind is still better to find every Mis,-than any legal control to find guardian or limits to protect us from doing things,even mistakes. Accidents are never,just happening. They have their rooting from many years ,establishing possibilities for expectations to be true. Maths,sadly prove some of the saddest results,without trying to turn the info , upside down for any purposes. And we live in it..

[Centaurus]

A point to make on using simulators for upset training in large aircraft. Currently as far as I know there are no simulators where this can be done

I think you are missing the point. Even if simulators had the required fidelity for the full expected range that are unusual attitudes, the problem would still be the instrument flying ability of the pilot. Reading back over many loss of control accidents after the aircraft had got into an unusual attitude, every one was in IMC or at night with no horizon.

That meant the pilot lacked basic instrument flying skills including recovery from unusual attitudes with reference to his flight instruments. One research paper noted that some pilots failed to use full appropriate control movements to level the wings for example. This is discussed in depth in the Maneuvers Section of the Boeing 737 Flight crew Training Manual. For example one statement in the section says: "With airspeed as low as stick shaker onset, normal roll controls - up to full deflection of ailerons and spoilers - may be used".

In most cases the unusual attitude has been the result of poor instrument flying ability by the pilot. All Level D full flight simulators are capable of conducting full 360 degree rolls on instruments. In other words the instruments correctly indicate attitudes flown. It is true that stick forces may not be correctly represented in some simulators but instrument interpretation never changes. Until pilots are taught how to recover from unusual attitudes in a simulator on instruments then loss of control in IMC will inevitably happen from time to time.

[Naali]

I am quite surprised to look at the graphs. We never tried,(if i know)to duplicate some fighter-like manouvres in our sim. And i don,t know if it even could simulate those. But some figures You show,are very logical for an airplane ,built to fly within quite narrow limits up there. Like all,i guess. -We just never had a chance to see how it behaves if you stall it at altitude. Might be an expensive way to train. What i am targeting,is that training would have been more,of telling also mishandlings,-even by the info and basic vee-wing anomalies at treating it wrong. Basic knowledge is more than Gold when everything goes eerie and very strange things start to happen... - I don,t know if it,s just a feeling but 87,felt to me more like an airplane than else versions. -To pick a thought perhaps far away,You must be comfortable with the airplane you fly. Otherways,it may be a compromise every day -and everyone knows that this marriage won,t last long. Meaning also that if You are beginning to be complacent,indifferent,and too much as "used to,"as an attitude,it might be one of the roots... Telling that,i remind that i,m telling of myself,and of no one else.

[ManaAdaSystem]

I've trained unusual attitude recoveries and high altitude stall, including deep stall, in the sim a number of times.
Yes, the sim may differ from a real world event, but up is still up, and down is still down.

[Lonewolf_50]

Centaurus: I think you've put your finger on the core issue.

My experience is that an instrument scan, if not often used, atrophies. It takes a bit of work to get it going again. The time to do that is not "when you need it the most" but before you need it again.

Final report hopefully will have a bit more granularity in the analysis. (I don't yet understand what is going on with the EPR changes, but I have not had a chance to go through it a second time. Sometimes one sees something else after the initial read).

[despegue]

The only way to improve your instrument scan and stick and rudder is to bloody switch off the Autopilot, Flightdirectors and autothrottle and FLY the aircraft raw data.
Any airline discouraging or Forbidding this should lose its AOC.
Any pilot afraid of this should lose his license and work on the ground.

It is THAT simple.

[Machinbird]

Trying to make sense of all those charts.
Looks like they began to recover coming through 20,000 ft only to enter a secondary stall. The subsequent beginnings of the next recovery were interrupted by impact with the ground, but they had begun to level the wings again. The secondary stall point is probably the hairpin turn on the ground track.
The EPR charts show some significant power loss/reductions leading up to the departure.

[Centaurus]

The only way to improve your instrument scan and stick and rudder is to bloody switch off the Autopilot, Flightdirectors and autothrottle and FLY the aircraft raw data.
Any airline discouraging or Forbidding this should lose its AOC.
Any pilot afraid of this should lose his license and work on the ground.

It is THAT simple

Agree wholeheartedly. Unfortunately, it is almost certain the vast majority of airline operators actively discourage their crews from any sort of manual raw data practice even if conditions are entirely appropriate.

[RetiredF4]

Because they think, that the majority of their crews are not able to practice raw data flying in an safe enough way with passengers on board? And because retraining them to a standard were this practice with passengers on board would be safe enough is too expensive?

[_Phoenix_]

Machinbird,

Looks like they began to recover coming through 20,000 ft only to enter a secondary stall.

We are missing the AOA parameter graph, in order to know if there was a stall, a spin or a spiral dive. It looks they never recover fist time though

Also the interim report is missing the rudder input...

It is interesting to know if they tried to stop the 5 deg/second rotation, before pulling - see pitch graph vs. magnetic heading graph - in 18 sec from 180 to 90

1.11.2.1 Parameter decoding
The data contained in the FDR raw file was decoded using the documentation provided by
the aeroplane operator and manufacturer. The set of recorded parameters could be decoded
and used, except for the following parameters:
 Angle of attack;
 Control column position;
 Control wheel position (LH/RH) (lateral position of the control column);
 Rudder pedal position.
The angle of attack decoding function as specified in the documents provided by the operator
and the manufacturer was not valid. Further work is being carried out with the manufacturer
in order to determine and validate the conversion function.
Control wheel and rudder pedal positions are additional parameters that were not recorded
when the aeroplane was delivered.

[Machinbird]

We are missing the AOA parameter graph, in order to know if there was a stall, a spin or a spiral dive. It looks they never recover fist time though

Yes, it would have been nice to have the AOA.
The secondary stall seems to have happened between 01:46:12 and 01:46:15.
Look at the roll angle leading up to time 01:46:15, look at the g trace which peaks about that time, look at the change in the slope of the pitch angle trace at that time leading to a pitch down, look at the elevator trace which peaks at 01:46:12 just before the secondary stall. Look at the effect on the airspeed which has a peak at that time and then starts back down.

Looks to me that they hurried the pull out from the dive and overdid it.

[Machinbird]

The angle of attack decoding function as specified in the documents provided by the operator
and the manufacturer was not valid. Further work is being carried out with the manufacturer
in order to determine and validate the conversion function.

Had an ominous thought.
Supposing the AOA probe (or some component downstream) was the wrong part number. Then you would very likely have this type of effect on the AOA flight data. Hopefully the maintenance records will give a clue on this.

The ominous part is: what else might be receiving bad AOA data?

[_Phoenix_]

The ominous part is: what else might be receiving bad AOA data?

Good question. How about slats and THS?
All slats have been found at crash site?

[Machinbird]

An earlier discussion re autoslat deployment charcteristics. http://www.pprune.org/tech-log/43624...-question.html
Apparently not available from a clean wing.

[ItsMeFromEarth]

The EPR charts show some significant power loss/reductions leading up to the departure.

Erroneous EPR-Limit triggers Autothrottle-retard of Thrust Levers:

PT2 probe Icing could be a reason

[dc9engineer]

A/P Disconnect

I haven't seen a discrete S/W parameter listed. The A/P is disconnected by S/W Pusher but the pusher is inhibited in slats retract therefore the disconnect at about 1:45min and 31sec was not due to S/W. As for the reason for not extending the slats from a clean config, this was due to concern for a failure causing slat extension at cruise.