Some additional news on the Air France Atlantic crash

[Phil Bunch]

http://www.bbc.co.uk/news/world-europe-13426928

I found this to be an odd news release, essentially saying nothing.  Perhaps I simply don't understand how to interpret a "no new recommendations" statement...

In any event I suspect won't know too much until something closer to a full investigation report is available.

[edit:  new link to a better-written summary added:]

http://www.businessweek.com/news/2011-05-17/air-france-black-boxes-show-no-initial-fault-with-airbus-jet.html

[torrence]

Hi Phil,

Thanks for the link.  Yeah - lots of strong emotions and legal complexities to this tragedy.  My personal guess is that extreme turbulence produced some sort of inappropriate response from the autopilot-human pilots combination - not 'the plane broke' nor pilot error in the sense of 'they pulled the wrong switch'.  It's possible that this may end up looking like that case a few years back in NY where the vertical stab came off because many pilots didn't know/weren't trained to realized that aggressive (but not forbidden) rudder input could break the thing (as well as I can remember).  As you say I think we will hear little until the final report.  Hopefully with the voice and flt recorders both they should be able to sort out what happened and make things (somewhat) safer in the future.

Cheers,
Torrence

[Phil Bunch]

I thought the airspeed indicator/pitot tube system(s) were a prime suspect?...

My guess is now that if they had found airspeed indicator failure evidence they would have reported it.  By saying almost nothing, I am guessing that the airspeed indicator systems were not clearly at fault, at least as of the time being.  My uninformed reasoning is that if they had found evidence of faulty airspeed indicators, they would have to report such an impression promptly even if it was just a probable finding.  This is just a guess, though - I don't know much about what is required at this very early stage of investigation, so soon after uploading the black box data.

It's great that they were able to upload the black box data - that should help the investigation immensely.

[torrence]

The discussion of background in your link suggests that the pitot tubes could have caused problems but should not have led to the loss of the ship in normal circumstances, unless either the autopilot had other problems or the crew mishandled the response to inaccurate airspeed.  Hence the sensitivity about 'pilot error' I think.  

I can't read much into not changing their previous recommendations.  The pitot tubes probably should have been changed out in any case and I suspect the statement just means there is nothing else suspicious enough to warrant any other action at this time.  I'll not familiar with all their rules either but I think if they find something that could conceivably have led to problems and some airworthiness action is indicated, they release it even if the findings are still preliminary.  

Cheers,
Torrence

[Zinger]

I don't know more, but have a few thoughts about information released so far:
a. Why did the flight not divert around the thunderstorm?
b. It has been reported that the autoflight system shut down, so I am assuming that the aircraft was being flown by the crew manually. Next, the airplane hit the ocean waters. The combination of the last two factors leads me to the probability of the crew loss of control.
c. Difficult to trust information so heavily impacted by ulterior motives.

[Phil Bunch]

An update, with news story seeming to emphasize air speed sensor issues again (excerpt):

"Air France 447 May Have Stalled on Sensors"

"Air France Flight 447's flight recordings show the aircraft slowed to a stall after its airspeed sensors failed while the two co-pilots were at the controls, two people with knowledge of the investigation said."

http://www.bloomberg.com/news/2011-05-23/air-france-flight-447-said-to-have-stalled-after-airspeed-sensor-failure.html

If so, I'm surprised this wasn't mentioned as a part of the previous news release.  One's suspicions about the credibility of mass media news stories are stimulated...we'll need to wait for the at least an interim report, as usual.

[Zinger]

Regardsless of the specific causes, and not having piloted an FBW Airbus, I have concerns regarding some incorporated design features. Specifically-
-the flight control system features which modify system characteristics (control laws) when problems arise.
-a self centering side controller which moves about one millimeter and thereby gives the pilot no idea of control surface position.
- flight envelope automated protections which limit pilot ability to control the aircraft in certain situations.

I may be uninformed, unaware of frequent Boeing aircraft Pitot tube icing during cruise.

The thing I like least of accident follow-up is the sniping as WSJ mentioned of parties involved, and the politics which override any common sense in the mass media investigation reports.
At the end of the day there appears to be enough information to determine the root cause of the crash.

[OKD]

This airbus technologies have always been in the back of my mind.

I just wonder its advance stuff, are they just too smart for human to handle / understand, where seems to me that the "inflight computers" have minds of their own...

For example, those who can recalled the 2 airbus incidents at Nagoya and Taipei in the mid 80s, where they both took place during the approach phase. If I am not mistaken (correct me if I am wrong), that the pilots failed to understand the technologies back then, and subsequently the computers "took over"the operations of the airplane while the pilots were "fighting" the opposite command of the the computers.

Zinger's first point of concers re: control laws speaks volume of the advancement of the technologies, which I also share with him 100%.

[Jeroen Hoppenbrouwers]

I think that in general, automation improves efficiency to such an extent that organisations are prepared to put up with the negative aspects. Look all around: automation does cause calamities to run out of hand much quicker. But as long as the figure below the line is higher with automation, the world automates.

Jeroen

[Zinger]

I am a strong proponent of automation, when it is done correctly. Example of good and poor in my kitchen. As a young married couple we chose a French SAUTER gas range, with a timer controlling one flame, integrated into that flame control. Excellent when doing unattended cooking. Now I have a newly designed brand name oil-less electric fryer, alas its timer doesn't shut off the flame, just rings a bell. What if the neighbor upstairs suddenly needed help to overcome a serious leak? Easy to guess, the French fries could become pretty dark... maybe also the kitchen wall  
Let's enter a transport aircraft cockpit, look at the landing gear extend/ retract control, apart from the very latest, such as B777 and modified B744, why are there 3 handle positions where the gear has only two desired positions? How many decades were required to change this? why? Should a pilot care what is happening to hydraulic fluid or limit stops, or should he mostly supervise aircraft operation while free to maintain situation awareness and make appropriate decisions?

I used simple but useful cockpit automation from the outset of my flying to a limited extent at a time it was just evolving, and later designed cockpit automation which is still flying.  When first entering a modern glass cockpit it looks so clear and stunningly uncrowded with those missing rows of steam gages.  But look at the amount of ACARS messages AF447 transmitted out by itself, and the amount of warnings the pilots got, and needed to manage. Wrong design which can kill. Good design inhibits irrelevant warnings between V1 and liftoff. This feature should also be applied while in stall. It should prevail in redesigning cockpit automation. The concept of genious engineers who build such complex systems in the name of efficiency, leads to the aircraft being in control while the pilot should be, from entering the cockpit until leaving it. He can't be both a skilled captain AND aircraft system's doctor, cpu's should take care of the latter. No process should inhibit main pilot decisions and control inputs. If that control with its range and authority was put there in the first place, it was for a reason. The reason is well defined in FAR and in the technical specifications governing handling qualities as written by the aviation authority.  Today's pilot needs to have substantial reduction in task workload, better more simplified display, so as to retain constant situation awareness. A display system which cannot alert quite a few pilots to below-glidepath approach is not doing its job, and worse.

[SwissCharles]

..... No process should inhibit main pilot decisions and control inputs. If that control with its range and authority was put there in the first place, it was for a reason. ...

Couldn't agree more! Mother nature can always throw 3+ more cards into a game that even the most clever engineers hadn't foreseen and thus were not 'covered' by their design.

I certainly don't want to launch into a Boeing vs Airbus argument but I doubt that AB would be able or willing to change their design philosophy to a point where the above quoted 'hierarchy' would be re-established. And: How far on this route did Boeing go with their 787's? I dunno..

So they will continue to 'patch' for each and every shortcomings which - and thats the cost which would also have to be taken into account when balancing automation savings against operating costs - mostly come after accidents or major mishaps only. But thats how this world operates...

Charles

[Zinger]

I think it is all about understanding the pilot needs, abilities and limitations to perform. You then design a system able to provide the information easily and clearly. If you display to a busy pilot 26 consequent warnings in mid air, he will not likely understand, prioritize and correctrly act in a short time while also flying the aircraft and managing the crew and cabin. The PM in such situations should automatically be delegated higher authority to start and stop things, excluding engines and gear, otherwise he would be waiting for approval from someone with task overload. Many abnormal procedure tasks can be semi-automated,  such that a checklist auto-runs but waits for a manual input or action (e.g. fuel valve shutoff) before continuing.

As another example, why can't aircraft systems which incorporate more than 2 independent airspeed indicators, use a simple routine which compares their readings and indicates a deviation of one from the average value? It could be even done with 2 independent airspeed indicators, assuming a good ground speed value had been calculated for a while. Just this would have helped save at least one B757 I remember.

[Jeroen Hoppenbrouwers]

Mu guess: airplane customers (buyers) tend to not have a pilot's license. They have an MBA.

[Jeroen D]

Some more news:

http://www.cnn.com/2011/WORLD/americas/05/27/air.france.447.crash/index.html?hpt=T1

Doesn't paint a pretty story.

[Phil Bunch]

Some more news:

http://www.cnn.com/2011/WORLD/americas/05/27/air.france.447.crash/index.html?hpt=T1

Doesn't paint a pretty story.

This news report seems to be more informative, although one doesn't know how trustworthy or complete.  It sounds like they didn't trust the standby instrumentation.  At 38,000 feet and cruising speed, there isn't much room for fooling around with airspeed, as best I can recall the "coffin corner" situation that is present under these conditions.

Is a 10,000 fpm stall-generated airliner descent "reasonably recoverable"?  Do the Airbus systems tend to fight recovery from a high-altitude, airspeed-related stall?

It was a little strange that CNN didn't know that "pilot PF" just means "pilot flying".  At least that's what I assume that "PF" meant in this situation...

[Zinger]

I don't klnow about the intervention of flight control system protections affecting pilot stall recovery attempts. But I believe that when you set max continuous thrust and level attitude, an aircraft would stall only when the wing surface accumulated so much snow that lift was seriously degraded, In such situation I'd descend a few thousand feet so the cruise airspeed regime is wider and allows for larger errors resulting from both airspeed indications and AOA, as well as provide more thrust available.

I studied and trained in departure from controlled flight and sustained out-of-control flight in a subsonic jet (North American T-2C Buckeye), including steady continuous deep stalls where angle of attack is approaching 90 degrees and the aircraft is falling literally vertically at the mentioned rate with little forward speed. This condition is recoverable in such aircraft using standard recovery technique. The elevator maintains sufficient control effectiveness to sustain the condition with near full aft control, and reacts to forward control motion as expected.

[Jeroen D]

I believe it does mean pilot flying and notice they also used PNF. I don't think CNN has a clue. They probably thought it was the pilot's initials.

Would stall recovery at these altitudes and cruising speeds be part of normal pilot training practice. I know stall can occur in any flight regime, but I tend to associate it more with airplanes at lower altitudes going through manoevres.

Jeroen

[Zinger]

Having read today's BEA accident findings, to me it looks like the pilots flew the aircraft into a stall, recovered then renetered a deep stall. Due to wrong, pro-stall control inputs they rapidly descended in a deep stall, For 3:30 minutes, from FL380 to hit the water. No further stall recovery control input was recorded. Further I believe pilot actions to have been only minimally affected during the stall by airspeed and AOA errors.
My summary:
Main cause: Poor pilot flying technique- inability to recover a stall with ample time, altitude and largely- functional aircraft.
Secondary, contributing factors:  Short term erroneous airspeed measurement discrepancy (disappeard after a minute), saturation of the crew by a multitude of warnings, sounds and messages, complex control laws,  the captain not being at the controls.

[Michel Vandaele]

In todays modern cockpit management,  PF = Pilot Flying and PNF = Pilot Non flying,  now also called PM = Pilot Monitoring.

As long the aircraft is at the gate / non powered,  you have the Capt / F/O mode. Once the aircraft is under its own power, then PF/PM comes in.  Of course the captain is still generally in charge.

You can't expect from CNN - press, they know these techinal terms  

B. RGds
Michel

[Hardy Heinlin]

Pete Frankenheimer and Paul Nordberg Fripp got a new friend.