Air France 447 - On topic only!

[KurtMc]

Peace

[Fear_of_Flying]

Originally Posted by ca787546

Good day everyone,
First of all, I haven't been to this forum in quite some time, but I am wondering if someone could please point me towards a copy of the CVR/FDR transcript, just so I'm up to speed before actually posting something dumb, irrelevant or repeating someone else's points.

Thanks,
Cheers,
DNB

Welcome back. There is no copy of those transcripts available yet, just an update and excerpts based on what they've learned from them. That update is here:

http://www.bea.aero/en/enquetes/flig...may2011.en.php

A report will come out later this summer. As for saying something dumb, irrelevant or repetitive, we wouldn't have it any other way.

Edit: see what I mean, I just repeated KurtMc's post.

[EconomyClass]

I guess there's enough fudge room in the available facts to argue for quite some time about what should or could have been done. I'll only say I hope they don't cop out for financial reasons because a lot of lives that will fly all airlines in the future hinge on getting the right remedies. And if I'm not mistaken, Rio is going to get a massive influx for the Olympic games, perhaps a volume of traffic it has never seen. Nothing like straining a system to discover its weak points. Some people might look at this incident and decide they'd rather watch the games on TV than depend on airline management to do the decent thing.

[Fear_of_Flying]

Originally Posted by KurtMc

How did Captain Kirk overcome his Kobayashi Maru situation...

He reprogrammed the computer? Very clever.

People did the same thing for weeks after Colgan, claiming there was no way the pilots could have just reacted so wrongly to the crisis before them. Yet they did, and the underlying causes were things like scheduling fatigue, inadequate time on type, inadequate training, etc.

While what you say might explain why the AF447 pilots had trouble controlling the aircraft and managing systems, it fails to explain why they did not carry out the memory items or use proper stall recovery procedure, which is what most directly seems to have led to the accident.

Air France is a large carrier with plenty of resources. If there are deficiencies in how they train their pilots, one wonders if these problems would turn out to be industry-wide. If, on the other hand, it should turn out that AF gets high marks for its training programs, where exactly would we go from there? Tweaking the AB automation does not seem like it would have prevented this accident.

[Fear_of_Flying]

Originally Posted by Fear_of_Flying

He reprogrammed the computer? Very clever.

People did the same thing for weeks after Colgan, claiming there was no way the pilots could have just reacted so wrongly to the crisis before them. Yet they did, and the underlying causes were things like scheduling fatigue, inadequate time on type, inadequate training, etc.

While what you say might explain why the AF447 pilots had trouble controlling the aircraft and managing systems, it fails to explain why they did not carry out the memory items or use proper stall recovery procedure, which is what most directly seems to have led to the accident.

Air France is a large carrier with plenty of resources. If there are deficiencies in how they train their pilots, one wonders if these problems would turn out to be industry-wide. If, on the other hand, it should turn out that AF gets high marks for its training programs, where exactly would we go from there? Tweaking the AB automation does not seem like it would have prevented this accident.

Actually, I take that last part back. Making the automation ever more foolproof would prevent similar accidents. For that short list of memory item situations Evan mentioned, let's just program them into the computer, and you push the correct button to initiate the appropriate action should the need arise. That is where AB philosophy is heading, and the human factor will always be beyond our means to safeguard.

[Gabriel]

Originally Posted by guamainiac

Theoddkiwi, are you getting info from the BEA report. I thought that said that as he gained altitude he was applying forward pressure .. that seemed to slow the VSI from 7000 to 700

The sequence seeems to be:
- "I have the plane"
- Right roll.
- Left roll correction combined with Pull up command.
- Stall warning sounding twice in a row.
- "We lost the speeds, alternate law"
- Beyond 10 degrees of pitch.
- + 7000 fpm
- Nose down.
- + 700 fpm.

Originally Posted by BEA

The speed displayed on the left side increased sharply to 215 kt (Mach 0.68 ). The airplane was then at an altitude of about 37,500 ft and the recorded angle of attack was around 4 degrees.

The sharp increase in the indicated airspeed (from less than 60 knots) seems to indicate that the pitot cleared and started to show the real indicated airspeed. No need to mention that 215 kts is SLOW!!!

Anyway, the speed MUST have gone pretty down by then due to the steep climb, no addition of thrust, and high altitude (maybe above what the plane can sustain, what doesn't mean that it can't transiently reach and then descent all under control).

Then:
- AoA increases to 6°.
- Stall warning again.
- TOGA thrust applied.
- PILOT KEEPS PULLING UP!!!!
- Peek at 38000ft, 16° pitch, 10° AoA.
- The plane starts to descend in a fully developed stall.

[Gabriel]

Originally Posted by ultraflight

A question please: would the Captain, not being strapped in, be glued to the ceiling while they were descending at 10.000 fpm?

I don't think so.
The vertical speed is not what glues you against the seat or the ceiling. It's the vertical acceleration. I don't think that they ever went to far from the normal 1G, and it takes a negative G to launch you up. In fact, a negative G will need negative lift, something that will not happen unless the pilots pushed way nose-down, which is exactly what they did not do, rather the opposite.

[Gabriel]

Originally Posted by Darren Howie

Still the 10000000000 dollar questin is why did they not conduct the unreliable speed drill.

I don't know, but the BEA investigated another dozen of UAS events and in none, zero, nada, of them the crew followed the UAS procedures. And that included one that happened AFTER AF went down and a UAS event was being already blamed!!!

Maybe the question is why Airbus pilots (and maybe Boeing pilots too?) are systematically not applying the UAS memory items and procedure in UAS events.

[Gabriel]

Originally Posted by EconomyClass

Are there recorded incidents where airspeed was lost on this route and the plane recovered safely? I've assumed all along that there's a whole history, that this is a departure from normal, but maybe there's something other than the total loss of plane, crew, and passengers that is extraordinary. Also, I assume this storm zone girdles the earth. That means a flight from LA or Tokyo to a southern hemisphere destination faces the same problems.

I don't know in this exact route, but yes, several in the ITCZ.
Within the investigation of this accident, the BEA investigated a dozen of other UAS events that didn't end in an accident. In NONE of them the crerw followed the UAS procedures. In some of them the stall warning activated.

[Gabriel]

Originally Posted by KurtMc

I've been thinking about root causes here, again because I really don't believe three pilots with more than 20,000 hours (9.8 work-years in the front seats) combined flight experience could simultaneously all go 'deer in the headlights' at once, without adjoining factors...

So far, I've boiled it down to the following root causes:

(1. 7.)

I'd agree, and I agree in general with many of your concerns.

Except that, in this particular accident, the pilots seem to have needed no help from the airplane and systems to crash it. It's not that they tried to apply the UAS procedures and failed. They failed from time zero, by pulling up 10 degrees nose up and, on top of that, do it with no addition of thrust.

I've said in a previous post that, while I still don't like some things of the Airbus' approach to automation and systems, with the pilots' actions this accident would have happened in the A330, the B-777, the DC-8 or the C-152.

[EconomyClass]

Here's another way to approach it. Assuming there is some sort of feedback that would make the pilots' actions correct, what would that feedback be? It makes more sense to infer the perceived conditions from their actions than to take an outside view of the real conditions and then puzzle over why they didn't react to those. None of us has any idea what they really saw or felt. So its a leap of faith to assume they saw or felt what we think they should have.

[Gabriel]

Originally Posted by EconomyClass

Here's another way to approach it. Assuming there is some sort of feedback that would make the pilots' actions correct, what would that feedback be?

That's a hard question. Maybe valid, maybe not.

What was the feedback that triggered the Colgan response?

The only thing I can think now that would fit the AF initial actions would be an overspeed (or increasing speed with a trend of getting soon to an overspeed) indication, with the pulling up being an attempt to slow down.

But it looks highly unlikely. The two speed indications recorded in the FDR showed a loss of speed (erroneous), so I don't think that the third one failed in the opposite way. Plus, a recover from an overspeed would involve pulling back on the throttles, something that didn't happen (if it did, I would not understand why the BAE didn't mentioned it).

[Evan]

Originally Posted by KurtMc

Not yet available, BEA thinks it deserves to wait until mid summer to release?

I've been thinking about root causes here, again because I really don't believe three pilots with more than 20,000 hours (9.8 work-years in the front seats) combined flight experience could simultaneously all go 'deer in the headlights' at once, without adjoining factors...

So far, I've boiled it down to the following root causes:

1. Loading a long-haul aircraft to absolute max take off weight (in an effort to save a refueling stop), despite available weather data indicating that the ITCZ was experiencing a massive buildup.

2. Determination to penetrate weather system, rather than divert, likely would not have been made in the presence of better radars/usage and real time satellite data showing updraft/downdraft/water content/temperature gradients of the storm systems the aircraft was entering.

3. Flight systems which at cruise adjust the CG to a rear-ward state, to achieve reductions in fuel expenditures by trimming the entire horizontal tail to compensate. Systems which are then essentially necessary to maintain static stability, unless the pilots are very experienced in flying aft-CG craft in a direct or degraded law FBW configuration which entirely lacks tactical sensation, and has no envelope protections.

4. The linear dropout of flight controls, without efforts by the PRIMS to hand the plane over to the pilots in a more flyable condition. The computers can't had the plane over 'safe and level, at 5 degree pitch, in climb detent, with CG moving forward via trim tanks, and trims returned to zero degree positions'?

5. Instead, the pilots are faced with multiple out of attitude configurations, in turbulence, at coffin corner, and are expected to resort to memory items when the SHTF, and the AP drops out without nary a care about the attitude and trim configs of the plane it's handing over to the pilots? It just gives up?

6. Pitot tubes prone to icing up and causing the AP/AT to drop out in numerous previous crossings of the ITCZ by the same carrier and others using the same type of aircraft? Then deemed to be a low priority for upgrading for several months by the same carrier?

7. A failure to install a BUSS or ISIS system which could provide a glass-cockpit pilot with basic analogue-esque flight parameters, if the AP/AT decide it's time to exit stage left?

How did Captain Kirk overcome his Kobayashi Maru situation...

I get your point (also made by others) that it would seem safer if, when detecting UAS, the autoflight should first revert to a forward CG for natural static stability, cancel load factor demand on the THS trim and apply the memory items. In theory I would agree. But there are things to be considered. This isn't a Boeing. The entire flight control system was not designed around positive static stability, for some very good reasons. It relies on the FCMC to determine an optimum aft CG for fuel burn and range (within the safe envelope). There is no electrical pitch trim control in the cockpit, no pickle switches. Electric pitch trim is entirely controlled, far more precisely, by the FCC's. Manual trim is a purely mechanical link to the hydraulics, and not intended for normal flight when the FCC's are available. The Airbus is designed to be exceedingly stable at 1G flight, correcting for attitude, speed and configuration changes, and exceedingly stall proof to compensate for this lack of positive stability and direct 'feel'. UAS, and the subsequent loss of alpha protections, presents a serious gap in the defenses, but is typically transient, lasting only a few minutes, and was not a real consideration for cruise level flight when the Airbus systems were developed; we are only now beginning to recognize the phenomena. Once airspeed data is restored, the protections remain lost, but stall and speed warnings return and autoflight can be restored, minimizing the risk. Everything comes down to that critical minute or two when airspeed data is unavailable and memory items are crucial.

But I also think memory items alone are not sufficient, especially when systems are available to make these calculations and control inputs. At the very least I think there should be clear prompts on the PFD for pitch and power settings. The BUSS system was designed along these lines, using a green range based on computer calculation of the remaining sensor data. But it proved to be unreliable and even detrimental at high altitudes, and so Airbus has since recommended that it not be used above FL250. For this reason, I think there are trans-systemic problems that we aren't aware of that make it safer to hand over control to the pilots. But I see where system improvements can be made in how that handover is coordinated. On such a sophisticated aircraft, nothing should be left to memory.

The weight issue isn't significant. Yes, they were at MTOW, but only .93t above the fuel order required without an RIF flight plan. At 205t, they were in no danger at FL350. In fact, FL370 was their planned altitude at that point. They were exercising caution.

The pitot changeover issue turned out to be of little significance, since similar UAS events have since been reported with the newer Thales BA probes, and even some with the Goodrich probes. Apparently there is no probe out there that can be 100% reliable in this kind of weather phenomena.

They had ISIS installed of course, but there was this ACARS message:

- .1/FLR/FR0906010211 34220006ISIS 1,,,,,,,ISIS(22FN-10FC) SPEED OR MACH FUNCTION,HARD

This just means that the speed bugs on the ISIS speed scale were flagged. HARD just means it lasted more than 2 seconds.

When both pilots say they have no valid indications, I'm assuming they are referring to airspeed. Apparently all three speed scales were flagged.

[EconomyClass]

Originally Posted by Gabriel

That's a hard question. Maybe valid, maybe not.

What was the feedback that triggered the Colgan response?

The only thing I can think now that would fit the AF initial actions would be an overspeed (or increasing speed with a trend of getting soon to an overspeed) indication, with the pulling up being an attempt to slow down.

But it looks highly unlikely. The two speed indications recorded in the FDR showed a loss of speed (erroneous), so I don't think that the third one failed in the opposite way. Plus, a recover from an overspeed would involve pulling back on the throttles, something that didn't happen (if it did, I would not understand why the BAE didn't mentioned it).

I think we know some things about the Colgan pilots that we don't know about the AF447 pilots. We know that the industry sector they work in had a lot of pressure to complete flights because they didn't get paid for aborted flights. We know that there were experience problems. But I'm thinking this set of pilots had flown under very similar conditions for an unknown number of flights. So all along I've expected that they reacted in some way consistent with their past experience. That's what makes me question the diagnosis that they made the "wrong response" to whatever it was they saw or felt. It is easy to sit on the sidelines and dictate a correct response when you know exactly what the conditions were, but we don't sit in that cabin getting the sense data they were getting.

[Evan]

Originally Posted by EconomyClass

That's what makes me question the diagnosis that they made the "wrong response" to whatever it was they saw or felt. It is easy to sit on the sidelines and dictate a correct response when you know exactly what the conditions were, but we don't sit in that cabin getting the sense data they were getting.

The situation was clearly unreliable airspeed above FL100. The response is clearly 5° pitch and CL thrust. They clearly made the wrong response, because there is only one right response.

[Deadstick]

Originally Posted by Evan

The situation was clearly unreliable airspeed above FL100. The response is clearly 5° pitch and CL thrust. They clearly made the wrong response, because there is only one right response.

Great analysis, coming from a non pilot.

[fmedina]

I was surprised to see that captain returned to cockpit, but was not really mentioned besides the fact that he returned. Once in the cockpit, who calls the shots, the guy in the seat or the ranking officer? Would the jr. co-pilot who seemed to be flying the plane be taking orders from sr. pilot or does it work differently?

Thanks,

[EconomyClass]

Well, I'm not all that confident that the final report will be accurate. And if the bird did something bizarre, it may not even matter. But as I say, when the world starts pouring into Rio, if it can happen, it probably will. Then the whole argument will erupt again. Arguing is a great amusement, but I'm not sure it is ever much of a solution-finder. It'd be a real shame if all those people died for nothing but their decision to step on that plane.

[Evan]

I am wondering why the FMA and the EFCS couldn't be more helpful in this situation. I made some quick sketches to illustrate what I mean, so maybe not exactly like this but something along these ideas.

In the first one the FMA simply informs the pilot of the current power setting and prompts the memory items. The red items would be flashing. If actual power setting is below full CL power, the MOVE THRUST LEVERS message flashes in red.

The second image is after moving the trust levers out of thrust lock and back into CL, or if the actual power was already at full CL. The optimum settings are displayed in amber. When actual pitch and thrust settings match these values, they turn to green.

[Evan]

Taking it a step further, this is a version where the A/P remains active in a 'UAS' mode. Lateral mode maintains basic heading and the EFCS sets the optimum values for pitch and thrust and then holds these until the situation clears. A/P Vertical mode would be like a FPA mode. UNRELIABLE AIRSPEED would flash in red.