Air France 447 - On topic only!

[Evan]

Originally Posted by Gabriel

"We" only had three total pressure source in disagreement. "Airbus" made a disagreement between the ADRs from that. As far as we know, there was nor reason to think that static pressure, outside air temperature, and AoA were in disagreement too.

There might be a reason. Until we see the FDR plots, we can only assume. But if the culprit was ice accretion due to an emergent warm air mass containing supercooled water, then the TAT sensors might have also been affected. Could that not result in incorrect altitude readings? I am very anxious to see the temperature data, for one thing.

I think both Airbus and the certifying authorities were very wary of making the leap to FBW back in the 1980's, and stipulated that when any redundancy fails in the ADIRU systems, the FGMC can no longer be trusted to fly the plane and the pilots must take over.

But train them, for chrissake. They have not yet designed the plane a pilot cannot kill.

You might have a good point. It's debatable. I am thinking of something along those lines, a way for the computers to make these calculations and assure that these things get done correctly. My idea is more about guidance than direct control. I sent you some of that in a PM earlier this week. But enough for now...

The PF was the most experienced on type. He had only Airbus type ratings, with the majority (4,479 hours) of his experience on the A330/340. He should have had a good instinct for handflying the A330. I'm really looking forward to learning how such a pilot could not be prepared for something like this.

[Evan]

Originally Posted by Harding

What is this "memory items"? At first i thought it was a emergency checklist of some kind - but i changed my mind after reading the report. Seems like they actually didnt know they where in an emergency for quite a while?

'Memory Items' are the emergency procedures a pilot must have already committed to memory, and be able to perform without any prompts from checklists. They serve to stabilize the aircraft so that the checklist items can then be performed. Like learning a musical instrument, these things require repeated training and practice.

In this case, the memory items were 'apply 5° pitch' and 'Set CLIMB thrust'. This was not done.

[Gabriel]

Originally Posted by Evan

As, suspected, they were slowing for turbulence penetration just prior to the A/THR disconnect, so thrust lock would have frozen a power setting below the one (CL) indicated by the thrust levers.

It seems that the speed reduction started some two minutes before the UAS event. I guess that the AT can deal with a change in 0.02 of Mach in less than that, so the thrust should have been back to a "sustained flight" level by the UAS.

The PNF had no trouble in identifying the UAS condition, so we can put that one to rest.

I agree.

Is this the automatic pitch trim in action? A smoking gun?

Yes, but not only auto trim. We don't have the elevator position plots yet, but let me bet that they will show an increasing level of ANU deflection too.

If the pilot was holdong the stick even a bit back from neutral, then he was commanding more than a load factor of more than 1. The FCCs will then command elevator nose-up to get the commanded load factor and ANU trim to cancel the need of elevator deflection. With the slow speed and stall protections removed, the FCC will have no problem applying nose-up trim into a stall, and once the trim is fully up, it will keep trying to produce the commanded load factor with elevator. Of course that a load factor of more than 1 is not achievable when flying at less than the stall speed, no matter how much nose-up command you apply with elevator and trim.

Thus the stall warnings cease. Is this a smoking gun? Did Airbus engineers not foresee a situation where UAS might be the cause of such low airspeed, where AoA data is still valid and very critical in maintaining stall warnings?

My thought too. Why not just make AoA invalid in GROUND mode?

[/quote]The FDR has them at 40° AoA and -10,000 ft/min V/S, but only 15° pitch with 100% N1 thrust. How is that possible?[/quote]
Why not? There's a lot, a BIG LOT of drag at 40º AoA.

Or is it possible that the pitch indications were erroneous, both on the PFD and on the FDR?

I don't think so. That would be a fully independent failure (a set of new independent failures indeed). As close to impossible as it gets.

And why did they subsequently bring thrust back to IDLE in a developed stall, when airspeed was obviously needed? Did the PF do this or the Capt?

Of course I don't know, but I have a couple of possible scenarios:
- Lack of better ideas: Nothing is working so I'll just try different settings.
- Upset recovery training. In under-wing-engines planes, thtust has a nose-up pitching moment. In upset recovery for thsese planes it's taught that, in extreme cases where the elevator and trim are not affective to reduce the AoA, reducing power can help. While I wouldn't expect someone to reduce thrust trying full nose-down command first, there are a lot of things that I would not have expected but still happened.

I am not surprised to hear that the pilots did not do the memory items correctly. That seems to be what killed them. Obviously, there was an industrywide training deficiency on this issue

Agreed.

But I can't accept that such an experienced crew would not be well versed on general stall avoidance and recovery procedure either. They must have been deprived of vital positional awareness beyond airspeed data to get into the mess described in the report.

I don't think so. It's happened before. More than once. Please add this accident to my list in the stall-rant thread.

Evan, a final word to you:
You were mainly right. While I still think that some things in the Airbus systems are potentially dangerous (like the load factor demand with no protections and the thrust-lock feature), this event was mainly very poorly managed.

- UAS was recognized, but the memory items not done (as in a dozen of other more lucky cases).
- The thrust levers were not moved for a while. Here the thrust-lock feature seems to have been a real issue.
- If they had been distracted while the plane was happily pulling up into a stall (to keep 1G in a diminidhin speed situation), then I'd balme the load facotr demand with no protections issue. But not. This crew pulled up, climbed (and I'm assuming they had good pitch and altitude information), stalled, and kept pullng up. Stick shaker shaking means stall. +15 deg pitch and -10,000 ft/min means stall. And they kept pulling up. The really amazing thing is that at one point, when they were at some 40 deg of AoA, they pushed down a bit and the plane, even at that amazingly AoA well beyond any imaginable envelope, responded by diminidhing the AoA. But then they pulled up again!!!

As I've said, I don't ike some things in the Airbus systems, but they don't seem to have played a critical role in this accident. By the way this crew responded to the event, it looks like they would have stalled and crashed an A330, a B-777, a DC-8 or a Piper Tomahawk the same (but not an Ercoupe)

I'm angry.

[Gabriel]

Originally Posted by EconomyClass

Other question I can't find the answer to is: Once the plane starts to fall, all efforts now become useless?

No.

The controls available for the pilots have no effect on what the plane does?

Incorrect. THey do have effect.

It said over 3 minutes falling. Was that three minutes of using the controls that no longer were working?

No. It was three minutes of using the controls in a wrong way.

They only had to f*=/%king push down!!!!
And then set the pitch and thrust for unreliable airspeed (if they had done this first, they would have never started to fall in the first place, so ther would have been no need to push down to recover from a stall that would and should have never happened)

[Gabriel]

Originally Posted by Harding

What is this "memory items"? At first i thought it was a emergency checklist of some kind - but i changed my mind after reading the report. Seems like they actually didnt know they where in an emergency for quite a while?

No. They correctly noticed the unreliable speed at once.

There is a two-fold procedure for an unreliable speed event:

1- Fly the plane at a given initial pitch and thrust that ensures that the plane remains safe. This are the memory items because you have to know them by hart. There's no time to go grab the book and look up for the correct procedure.
2- Now, with the plane stabilized as said above, one pilot keep flying the plane according to these two memoty item values, and the other calmly grabs the book, looks up for unreliable speed, and with a few parameters (aircraft weight, altitude and temp) look up in a table for fine tunned values of pitch and thrust that not only will ensure safety but also will keep the altitude.

[guamainiac]

Ummm, Gabriel, they had a VSI of 7000 fpm at altitude which sounds much higher than could be anticipated by pitch which was partly nose down and may .. note word may ... have partly arrested the ascent to 38,000.

What is say perhaps, the best rate of climb for that ship?

[Myndee]

Can someome answer my question? Do I have it right that to recover from a stall is to trade altitude for airspeed (nose down - not nose up?) Why were these pilots making nose up inputs?

I feel terrible for the pax on 447. Just as we thought, there was no merciful loss of consciousness. These poor souls suffered terribly for the last three minutes of their lives.

My question to all of you is: Was this avoidable? Was this aircraft 100% recoverable?

Someone needs to make me feel better about this. I am flying for the first time on a 319 next weekend. I'm really nervous now.

[Gabriel]

Originally Posted by guamainiac

Ummm, Gabriel, they had a VSI of 7000 fpm at altitude which sounds much higher than could be anticipated by pitch which was partly nose down and may .. note word may ... have partly arrested the ascent to 38,000.

The pilot pulled up first. The pitch increased from 2.5 to beyond 10 degrees. That's a 7.5º change in pitch which, initially (until the plane starts loosing speed) translate into a 7.5º change in trajectory. If you are flying horizontaly at say 450 knots TAS and bend that thrajectory 7.5º up you get a vertical speed of some 58 knots, which is some 5,900 ft/min. We are in the ballpark of the 7,000 mentioned.

And again, the pitch DID increase to beyond 10 degrees, and a raising mass of air doesn't do that by itself (rather the opposite).

What is say perhaps, the best rate of climb for that ship?

At that altitude near the service ceiling? Not much. But that would be a sustained climb.

When you trade speed for altitude things are different. Since you mentioned a gider, what is the climb rate of a glider in the absence of raising air? There is no rate of climb, rather a rate of descent. That doesn't mean that the pilot can't pull up and trade knots for feet.

[Myndee]

Also, could someone kindly post a summary of what happened in laymens terms for those of us who do not 100% understand all of the lingo. Thanks.

[AJ]

Originally Posted by Darren Howie

Hi AJ
The aircraft was a 2006 build and the BUSS hit the line in late 05.

The aircraft was delivered to Air France in April 2005 so, as Evan points out since, it wouldn't have had the BUSS by your implementation date of late 2005. That shouldn't mean a loss of control, but it certainly would have helped if the BUSS was installed.
I recently got to fly a volcanic ash encounter in our A380 sim, with the ash blocking the probes and flaming out all four engines. It was still an aeroplane.....lower the nose, glide, exit the ash and relight. The BUSS was, as you say, instinctive and allowed one crewmember to attempt relights without concern the other guy would pitch up and stall!
We survived....even with an un-Airbus trained pilot! (BTW, it didn't stop me mocking everything relentlessly, it's a Ford vs Holden, Melbourne vs Sydney type thing)

[Gabriel]

Originally Posted by Myndee

Can someome answer my question? Do I have it right that to recover from a stall is to trade altitude for airspeed (nose down - not nose up?)

"Trade altitude for speed" is not universal.
The correct answer is "reduce the angle of attack".
You are right that that typically involves a nose down command (or at least to releave some of the back pressure).

Why were these pilots making nose up inputs?

For the same reasons that other pilots did the same thing dozens of times. Because they were wrong. Think Colgan.

But as always (or nearly) is the case, incorrect stall recovery is the last link in the chain. They should have never put the plane in the boundary of a stall to begin with.

The correctly identified the unreliable speed and had two memory items to apply: Pitch 5º and climb thrust. They did none.

I feel terrible for the pax on 447. Just as we thought, there was no merciful loss of consciousness. These poor souls suffered terribly for the last three minutes of their lives.

Maybe it was not that bad. I mean they knew that there was something odd, strange sounds and movements on the plane. But I doubt the plane ever went far away from 1G. I bet the people got fraightened but never really "knew" that they were going to die beyond hope, and next thing they were all died on impact. Except maybe the pilots, who probably knew at some point that they were not going to make it.

My question to all of you is: Was this avoidable? Was this aircraft 100% recoverable?

Yes and yes.

Someone needs to make me feel better about this. I am flying for the first time on a 319 next weekend. I'm really nervous now.

If this serves... I fell ungry. The pilots performed very badly. I also don't like some ways Airbus approaches its automations and systems.

Yet, I'd fly in an Airbus anyday. Remember these were the first lives lost in an A330 in service, after probably million of flights already. It took a critical and very unlikely combination of system failures (due to the unreliable speed event) and pilots' failures to bring it down.

[Gabriel]

Originally Posted by Myndee

Also, could someone kindly post a summary of what happened in laymens terms for those of us who do not 100% understand all of the lingo. Thanks.

- The pitot tubes were blocked or otherwise rendered ineffective.
- That triggered an unreliable speed event.
- The crew correctly identified the unreliable speed event and the fact that the autopilot and autothrottle had disconnected.
- A pilot took manual control of the plane.
- He should have applied the memory items for unreliable speed: Point the nose 5º above the horizon and set the throttles to climb thrust.
- Instead, he didn't set climb thrust and pointed the nose more than 10º above the horizon.
- The plane climbed, lost speed, and stalled.
- The pilot never recovered from the stall because he didn't apply the correct control inputs, as you know by now, so the plane kept falling stalled for 38,000ft and three minutes.

I think that that very much summarizes it.

[Theoddkiwi]

I think many of you are looking too far into what happened, perhaps looking for something that is not there.

I see nothing in the report apart from the Unreliable airspeed as a failure of the aircraft (the Engines were fine, the Flight Controls responded properly, All the data on their PFDs eg Altitude was correct apart from speed) and or any other influencing factors such as up drafts etc.

I am with Gabriel. I am angry. This was a completely botched recovery of something that should have been reasonably easy to deal with.

This aircraft crashed simply because the at the least the two Pilots flying had no idea what to do and when the Captain arrives not much seemed to have changed either. That is damning on them and VERY damning of Air France and its crew training.

Of the aircraft systems that probably didn't help the Autotrim probably didn't help, but it just did what it was designed to do. The Air New Zealand A320 crash in France revealed a similar outcome. The Stab drove to full nose up. But in this case i am sure had they pushed the little spring loaded joystick forward the Autotrim would have driven the Stab back to a more neutral position.

What is crazy is that a part of me is wondering if this crew had not touched the controls at all, that this crash might never have happened. I mean the aircraft would have been trimmed pretty close to perfect and the engine thrust would have been about right too. Its not like when AP and AT disengage, the aircraft will just spin wildly out of control.

Just regarding the AOA sensors at low airspeed, they need to have sufficient airflow over them to be considered accurate enough, this would be the same for all aircraft that use the wind vane type.

[guamainiac]

Too far into it?

The PF during the inflight briefing tells the Captain that because the temperature at altitude is falling more slowly than forecast they won't be able to fly there (higher).

Moments later they are ascending to 38,000 where things start to come apart with stall warnings.

Performance of the wing and flight in general has two salient atmospheric/geo factors. Temperature and altitude ... climb performance suffers, safe ceiling suffers.

Look at the radar returns from the weather they were flying not over .. but in.

[Fear_of_Flying]

Originally Posted by Theoddkiwi

I think many of you are looking too far into what happened, perhaps looking for something that is not there.

I see nothing in the report apart from the Unreliable airspeed as a failure of the aircraft (the Engines were fine, the Flight Controls responded properly, All the data on their PFDs eg Altitude was correct apart from speed) and or any other influencing factors such as up drafts etc.

I am with Gabriel. I am angry. This was a completely botched recovery of something that should have been reasonably easy to deal with.

This aircraft crashed simply because the at the least the two Pilots flying had no idea what to do and when the Captain arrives not much seemed to have changed either. That is damning on them and VERY damning of Air France and its crew training.

Of the aircraft systems that probably didn't help the Autotrim probably didn't help, but it just did what it was designed to do. The Air New Zealand A320 crash in France revealed a similar outcome. The Stab drove to full nose up. But in this case i am sure had they pushed the little spring loaded joystick forward the Autotrim would have driven the Stab back to a more neutral position.

What is crazy is that a part of me is wondering if this crew had not touched the controls at all, that this crash might never have happened. I mean the aircraft would have been trimmed pretty close to perfect and the engine thrust would have been about right too. Its not like when AP and AT disengage, the aircraft will just spin wildly out of control.

Just regarding the AOA sensors at low airspeed, they need to have sufficient airflow over them to be considered accurate enough, this would be the same for all aircraft that use the wind vane type.

I am inclined to agree with you. This is like Colgan, only worse. With Colgan, it was one terrible panic reaction, all over in a few seconds. Here, they had two things the Colgan pilots never had: time and altitude. Yet they apparently botched it all the way down. They were not, to the best of my knowledge, fatigued from an inhumane schedule, nor were they straight out of a get-your-certification-in-6-weeks flight school. You quickly run out of excuses for these guys.

We have people here who will immediately start looking for phantom explanations (i.e. with Colgan, maybe they were reacting to tail stall) but the glaring facts, as we know them up to now, seem to paint a pretty clear picture.

[MCM]

You quickly run out of excuses for these guys.

That is why it is vital to look deeper than just the pilots.

This was an experienced crew, who would have gone through training and checking many many times. The usual pilot training excuses, such as new on type, inexperienced in the aircraft involved, just don't apply.

So, WHY do two experienced aviators make the mistake?

You need to look at everything - from their training program right through to the aircraft design and data presentation.

These were not rogue pilots paying no regard to their aircraft, and so we need to know why they were not prepared for this scenario.

I think the AF training department is about to get a few BEA analysts paying them a very long visit.

[guamainiac]

You can't compare the localized strata of ice of Colgan and the boiling tropical systems that rocket up to the upper flight levels. In size, in force and complexity. Weather phenomena whose genesis is the warm oceans that are sending tropical convective activity so high. Never knowing if you are in what developmental phase and what you will encounter by way of it building maturity or some of the down drafts as it decays.

Ice that will disable not only the pitots but the engines.

[Theoddkiwi]

Originally Posted by guamainiac

Too far into it?

The PF during the inflight briefing tells the Captain that because the temperature at altitude is falling more slowly than forecast they won't be able to fly there (higher).

Moments later they are ascending to 38,000 where things start to come apart with stall warnings.

Performance of the wing and flight in general has two salient atmospheric/geo factors. Temperature and altitude ... climb performance suffers, safe ceiling suffers.

Look at the radar returns from the weather they were flying not over .. but in.

The AT and AP disconnected when they were at 35,000 because of the UAS. not because the plane launched into the air because of a massive up draft.

The only reason they reached 38,000ft is because the Pilot Flying pulled back on the joystick and commanded a 10 degree pitch up.
The report to me clearly shows the pilot initiated the climb. There is no indication that they said hey we have suddenly climbed to 38'000 ft all by itself, in fact there seems to be no indication that they were concerned about altitude during this maneuver.

To me the reason the stalled is because they didn't increase engine thrust during a climbing maneuver and slowed to what seems to be less than 60kts.
Basic airmanship, to climb your supposed to use PAT (Power - Attitude - Trim)

[guamainiac]

MCM, these storms in this area are common but how did that storm fare in contrast to the norm.

It seems like the Captain elected to take his rest because this area and the weather, well it seems to be SOP, and this is how a crew gets the experience and confidence .. when the old man takes a nap.

Will a look at the magnitude of the convection in terms of strength (flow and lapse rate) factor?

Note that the normal escape seemed to be over the tops but as they said to the Captain ... the weather at altitude was not cold enough to fly there. Yet it seemed like they were forced there?

Did convention fail them? Ernie Gann ... Fate is the Hunter and the hunting is done where the numbers converge.

[Theoddkiwi]

Originally Posted by guamainiac

Ice that will disable not only the pitots but the engines.

There is no indication that the engines suffered any performance deficit. They clearly state the engines responded normally to the the input commands at all times.

The engines had nothing to do with this accident except that the pilot didn't try to use them until it was too late, and then bizarrely set them to idle.