News update on Air France crash in the Atlantic

[Zinger]

Can't answer for sure about flight auto-page display. Since the JFK A300 American Airlines loss of vertical tail due to captain overcontrol, Airbus added a rudder control travel limiter with airspeed, and want the pilot to also see rudder control position for that reason (mostly in direct law where protections are removed). It was always there but rudder control criticality to flight safety had been emphasized since.

Hardy, with proper piloting this all becomes secondary. The AF447 was in my opinion a piloting problem from the decision to not change course away from the tropical storm, to the pro-spin controls consistently held, to the water. Simply wrong actions which indicate aircraft system and fight control characteristics, pilot selection, up-bringing, company culture and currency training, all mediocre or worse, and surely negligent.
One has to be a design engineer to understand complex system malfunctions, and an experimental test pilot to master the three main control law with variants so that he can effectively and safely overcome their problems. Next these design engineers will convince the politicians of underworked workforces, that the way to go is to develop 850 passenger remotely controlled transports.

[Zinger]

To add to the Jo'burg incident:
a. Training: Proper airline training and qualifications, should incorporate the following: Even with a type certificate for all variants, there should be some form of first flight instruction or supervision. I flew 1988 Frankfurt to Eilat on the brand new B757-200, the captain previously flew B767-200s. In the jumpseat next to mine was an airline check captain.
b. Pilot technique: During liftoff, the goal is to rotate to a pre-determind attitude at a predetermined rate of pitch to avoid tailstrike while attaining positive climbout. The pilot primary control reference is therefore pitch rate, attain a pitch attitude as a function of time. All other information, especially what control movement, is relevant as background information but secondary for piloting technique. Otherwise 14,500 of runway will be too short. Secondly, you are heavy, new model, high altitude runway. Accelerating too slowly in your opinion, no instructor there- go TO/GA early. After flight debrief the takeoff so the next one is better. Next, with so many CRM tasks during the roll, the PM is engulfed and therefore much less helpful in identifying problems early.

[frumpy]

Frumpy, that link you posted to the voice recorder seems to be down.

Any other way of accessing it?

Seems to be down. I think it's similar to this one:
http://uk.reuters.com/article/2011/07/30/france-crash-crew-idUKLDE76S1IY20110730

[John H Watson]

The A330 flight control system incorporates two indication types for this:
a. A sidestick position indicator of the same seat position, first page. Used mainly for adjustment/. calibration

Correct me if I'm wrong, but isn't this indicator purely an armrest position indicator and has nothing to do with sidestick position indication?

[Hardy Heinlin]

Hardy, with proper piloting this all becomes secondary.

Assuming every pilot would pass the checks that you suggest, could we then assume that everyone of them will be perfectly skilled at all times and physically and mentally fit in every minute? Can you foresee every possibility, everything that could go wrong in the aircraft or in the pilot's mind?

In that sad situation, if I understand the text correctly, the left seat pilot and the Captain on the observer seat didn't know that the other pilot pulled on the stick all the time. That's why I wondered why there's no single indication to allow cross-checking. What if the PF is suddenly incapacitated during a critical flight phase without stall protection? If there was a conventional yoke, the PNF would see the PF's absurde action immediately.


|-|

[Richard McDonald Woods]

Jamie said:

trend information gathered from FDM (Flight Data Monitoring) system: a 'devil' box which records if you did something stupid e.g. taxi too fast, but also gear-down too late, unstabilized approaches, overrotate etc. etc. thousands of parameters are recorded

.

Will PSX have some form of FDM on a per flight basis?

[Zinger]

John,
yes it is. The indication most useful to understand aircraft response, especially during flight contol system failures and control law changes, some without pilot awareness, is control surface position.

Hardy,
regarding anticipation of circumstances: The designer works with FAR/ JAR requirements. An aviation operator in charge of proficiency and resulting safety, has to anticipate all this. I can only repeat that my approach has one advantage, it brought complex opeartions under my responsibility for decades and tens of thousands of flights by hundreds of pilots to zero fatal and maximum mission accomplishment, at zero added cost, actually with cost reduction, a result of effective methodology.
I agree with you about your logic of sidestick postition indication.  For incapacitation, there is a control display system to switch sidestick controller sides.

As captain of AF447, assuming taking a pilot seat impractical, I'd ensure full sidestick controller forward pressure by PF plus forward THS trim, until angle of attack returned to normal for attitude/ altitude leveling. There seems to have been ample time for that.

[Hardy Heinlin]

Jamie said:

trend information gathered from FDM (Flight Data Monitoring) system: a 'devil' box which records if you did something stupid e.g. taxi too fast, but also gear-down too late, unstabilized approaches, overrotate etc. etc. thousands of parameters are recorded

.

Will PSX have some form of FDM on a per flight basis?

The data is available in real-time via network. There is no recorder inside PSX. Recorders can be connected externally.


|-|

[Jeroen Hoppenbrouwers]

I think that most people now agree, especially with the captain's comments seconds before impact, that neither monitoring pilot had any idea at all that the pilot flying had been pulling the nose up intentionally the whole way down. Intentionally as actively -- not necessarily by conscious decision, I fear.

This may be a first. A pilot making a weird input for a moment, that's no problem (and cannot be prevented anyway). But the system simply was not designed with this situation in mind. The stick position averager by itself would not help sufficiently to overcome input to the limit stop. Only if the monitoring pilot would have used the takeover button, he would have noticed that suddenly the plane became totally controllable.

I presume that there will be a new procedure which includes a check list/procedure item to actively announce and use the takeover button by the monitoring pilot, both to check the pilot flying and his/her side stick (which may be stuck).

Which cues me to ask: is there a procedure for a stuck side stick anyway, just as there is a procedure for a stuck yoke (pull hard, if not, pull harder and shear the connection links)?


Jeroen

[Jamie]

Jikes, after reading the CVR transcription, I hereby take back all my previous comments :shock:

[Phil Bunch]

Another update - the authorities/investigators have found that improved pilot training is needed, based on this crash.  

http://www.bloomberg.com/news/2012-07-05/air-france-447-crash-probe-finds-pilots-lacked-proper-training.html

I find it hard to argue with this report's conclusion, but I would add that better pitot tube icing detection and correction would also be helpful, if technically feasible.  Perhaps that's an ongoing, in-progress effort for all aircraft pitot tubes???  The mere fact that they can ice over is presumably plenty enough motivation to make better airspeed instrumentation systems.

========================
Excerpts:

Aviation safety specialists have said the trio were probably bewildered by erratic instrument readings and may have done the opposite of what was needed to keep the jet from crashing. The pilots had only three and a half minutes to avert disaster as the jet fell toward the ocean at a speed of 180 feet (55 meters) a second.

"The startle effect played a major role in the destabilization of the flight path and in the two pilots understanding the situation," today's report found.

The pilots of Air France flight 447 that crashed into the Atlantic Ocean three years ago lacked the right training to respond to a surprise scenario, the French aviation investigator said in its final report of the incident.

The investigation uncovered "profound loss of understanding" in the cockpit in a moment of surprise, when the aircraft went into a stall and lost lift, the report found. The pilots lacked training for stall scenarios, and the authority recommended that flight simulation training be reviewed.

"The dual failure of the expected procedural responses shows the limits of the current safety model," the French BEA authority said in its report. "The crew, whose work was becoming disrupted, likely never realized they were facing a 'simple' loss of all three airspeed sources."

Airbus, which helped fund the search for the recorders, has said that the aircraft was responsive throughout its descent into the Atlantic Ocean. The pilots were required to take over the controls after the auto-pilot disengaged because of faulty speed readings caused by iced-up sensors. That occurrence alone could not explain a crash, Airbus has said.

The BEA is the latest accident investigation agency to call for improved training for pilots in how to recognize a stall. So-called loss-of-control accidents, which include stalls, are the biggest cause of crashes and deaths around the world, according to statistics from Boeing Co. (BA)

"The BEA report describes a crew who acted in line with the information provided by the cockpit instruments and systems, and the aircraft behaviour as it was perceptible in the cockpit," the airline said. "The reading of the various data did not enable them to apply the appropriate action."

The BEA raised several recommendations that indicate shortcomings in how information is displayed to pilots on the A330, in particular the so-called flight director pilots rely on to fly the aircraft. When the autopilot disengaged, the flight director disappeared and other warnings sounded. That led the pilots to make wrong control inputs, although the BEA also found a dozen instances where pilots reacted properly.

Aviation safety specialists have said the trio were probably bewildered by erratic instrument readings and may have done the opposite of what was needed to keep the jet from crashing. The pilots had only three and a half minutes to avert disaster as the jet fell toward the ocean at a speed of 180 feet (55 meters) a second.

"The startle effect played a major role in the destabilization of the flight path and in the two pilots understanding the situation," today's report found.

Pilots are trained to avert stalls, which occur when an aircraft slows enough that its wings lose lift, by dropping the nose to increase speed. Instead, the flight data showed that the pilot at the controls for most of the last minutes consistently angled the jet nose higher.

Airbus, which like Air France is partly owned by the French state, has said the flight-recorder readings support the technical flawlessness of the wide-body aircraft. Air France has called "misleading" the fact that the stall warning alarm went on and off repeatedly as the plane moved in and out of stall, responding pilot directions in the cockpit.

Still, regulators should require specific pilot training on high-speed stalls, which isn't currently mandatory and which neither of the Air France co-pilots had received, the previous report said. That report also called for manufacturers to consider making available a reading of the so-called angle of attack, which defines the angle between air flow and the longitudinal axis of the aircraft.

In the case of the crashed Airbus, the angle, which wasn't visible to the pilots, always remained above 35 degrees during the descent.
========================
EDIT:  a complementary article in the UK's Guardian has a somewhat different perspective.  Here's an excerpt and a  link:

However, the plane was in a stall instead. A basic manoeuvre for stall recovery, which pilots are taught at the outset of their flight training, is to push the yoke forward and apply full throttle to lower the nose of the plane and build up speed. But because the pilot thought the plane was diving, he nosed up.

http://www.guardian.co.uk/world/2012/jul/05/air-france-crash-ruling-pilots

While the pilots could have prevented the crash, were they adequately trained?  The confusing part for me is that one's most basic flight training includes stall recovery and stall recognition.  Yet that's such a different thing when a stall occurs during high-altitude cruise and when one's instrumentation isn't credible...

Is it really all or mostly the pilots' fault?  Will future pilots react quickly and effectively with better training and with this accident now on the record?

I always feel somewhat silly when I comment on such aviation matters since they have such a deep technical component and I do not have detailed technical knowledge or experience.

[Richard McDonald Woods]

I have always been surprised at how so much of an aircraft's safe functioning is predicated on pitot tubes functioning correctly.

[Phil Bunch]

"Pilots a generation ago would have done that and understand what was going on, but (the AF447 pilots) were so conditioned to rely on the automation that they were unable to do this," he said. "This is a problem not just limited to Air France or Airbus, it's a problem we're seeing around the world because pilots are being conditioned to treat automated processed data as truth, and not compare it with the raw information that lies underneath."

http://www.guardian.co.uk/world/2012/jul/05/air-france-crash-ruling-pilots

What information?  No airspeed, no info!?

Would a GPS provide a fairly accurate ground speed estimate?  Would doppler radar then be able to roughly estimate air speed by correcting the ground speed, at least during rainy weather?  But then I think about the "coffin corner" where a very accurate value is required.  

I guess it all comes back to pilot education about high altitude stall recognition and recovery.

[John H Watson]

but I would add that better pitot tube icing detection and correction would also be helpful,

Probe heat is applied continuously at high levels in the air and it was expected to cover all situations, so detection was not deemed necessary. However, heat was simply insufficient on these probe types.

All probes of this type should have been replaced by now (under FAA mandatory airworthiness directives). Thales type replaced by Goodrich or similar.

Rgds
JHW

[Richard McDonald Woods]

So are we any clearer what the pilots should have done in this situation?

[Jeroen Hoppenbrouwers]

*should* have done: just fly the airplane -- it was not at all broken. Go by pitch and power, use cross-cues from all other working instruments and stay stable. Cruise flight is cruise flight.

After half a minute, the pitots would thaw and it would have been passed off as an incident for the techs.

In hindsight...


Jeroen

[Shiv Mathur]

But if you re-read the BEA report:

"The BEA report describes a crew who acted in line with the information provided by the cockpit instruments and systems, and the aircraft behaviour as it was perceptible in the cockpit," the airline said. "The reading of the various data did not enable them to apply the appropriate action."

[Zinger]

I repeat my previous view that the airline industry need to reinvent their aircrew selection, training and qualification criteria. The Airbus cockpit to me appears a trap more than a place of work. No pflaster solution will work.

[Jeroen Hoppenbrouwers]

Final report seems to be out.
http://www.skybrary.aero/index.php/A332,_en-route,_Atlantic_Ocean,_2009_%28LOC_HF_AW%29