Air France Jet Crash: Are Commercial Jets Safe Against Lightning?

[Borek]

Data from both recorders have been successfully read:

http://online.wsj.com/article/SB10001424052748703509104576326792484813816.html

 

[JaredJames]

I think caution is certainly required, otherwise the media may turn it into a fly-by-wire witch hunt. The beginnings of which you can see at the end of that article.

 

[AlephZero]

I think caution is certainly required, otherwise the media may turn it into a fly-by-wire witch hunt. The beginnings of which you can see at the end of that article.

I expect a certain type of aviation website will turn it into a Boeing-vs-Airbus witch hunt, but I hope PF is above such things.

In any case, the days when autopilots were a big box of mechanical gubbins called "George", apparently designed by Heath Robinson and Rube Goldberg, have long gone, so "fly by wire" isn't a very useful description.

 

[JaredJames]

Think I may have missed the point, but fly by wire is the name not the description. Nothing wrong with it I can see.

 

[AlephZero]

OK, maybe I was reading too much into your "fiy by wire witch hunt" based on what I know but you may not.

Both Airbus and Boeing use "fly by wire", but there is a difference in design philosophy. In general terms, Airbus refuses to let the pilot attempt to fly the aircraft outside of its design envelope, period. Boeing normally does the same, but there is a "panic button" that let's the pilots attempt to do anything they like.

Whether the pilots actually know better than the computers how to fly the plane without breaking it is a separate question, and in any case there is no direct mechanical control linkage that completely bypasses the plane's electrics.

If the root cause of the AF crash was indeed failed pitot tubes as has been speculated, then all this is fairly irrelevant, because if you don't have a working airspeed indicator, and you can't make an estimate by looking out of the window, the outcome of any action would be pretty much a matter of luck.

 

[JaredJames]

For the record I know a lot about fly by wire and the boeing / airbus differences.

Irrelevant or not, the media love this sort of thing and won't let little issues such as irrelevance and facts get in the way. This is all my comment was about - again, something you see hinted at in the end of that article.

 

[russ_watters]

So it sounds like the cause was faulty speed indication leading to a stall, then pilot error in recovering: http://content.usatoday.com/communi...rance-flight-experienced-35-minute-freefall/1

 

[mheslep]

Andre, any comments? This sounds like gross pilot error. Idling the engines? What about GPS for approximate air speed?

 

[AlephZero]

Before we start speculating about pilot error, the first question is whether the aircraft type was designed to be recoverable from a full stall even with all systems working properly (which clearly they were not). Recovering from a stick-shaker or audible stall warning is a very different situation from the full monte.

Also we also don't know the exact weather conditions at the time, but there were reports of large weather systems in the area. It is not impossible that the "climb" was caused by flying into the top of a thunderstorm head, not by pilot action.

Passenger planes the size of an A330 are not strong enough to be thrown around the sky like basic trainers or military fast jets, for the basic economic reason that they are never meant to be used that way.

What about GPS for approximate air speed?

It is worked at all, GPS would only give you ground speed. If you were flying through a thunderstorm, the difference between that and airspeed could quite easily be 100 knots.

 

[JaredJames]

Before we start speculating about pilot error

Out of curiosity, do they record pilot input on these aircraft? If so, it would certainly indicate whether the pilot did it or not.

The reports indicate that they know the pilot input the control movements to cause the various stages.

‘The inputs made by the (pilot flying) were mainly nose-up,’ the report notes, indicating a potential pilot error. Generally, pilots would push the nose down to recover from a stall.

http://blog.seattlepi.com/aerospace...oward-pilot-of-doomed-air-france-airbus-a330/

They report the engines were throttled back to idle, that is recorded is it not? So a pilot obviously interacted there to do that much (wrong action in a stall).

EDIT: From the FAA regulations (http://rgl.faa.gov/Regulatory_and_Guidance_Library/rgFAR.nsf/FutureEffective/d5e92f897c8dbc92862576d200740e33!OpenDocument ), it is clear that flight recorders are required to record pilot inputs. So if the pilot did pitch up approaching the stall, it would be known from the data on the recorder.

EDIT 2: You could certainly use the GPS to check speed if desperate. It may not be your airspeed, but if you knew the weather conditions and the ground speed, it would be a darn good indication to help you gauge the danger of stall.

 

[hypatia]

This has been interesting to follow, thanks for keeping it "real information".

 

[russ_watters]

Before we start speculating about pilot error, the first question is whether the aircraft type was designed to be recoverable from a full stall even with all systems working properly (which clearly they were not). Recovering from a stick-shaker or audible stall warning is a very different situation from the full monte.

Granted, but assuming the article is accurate in the statements about the control inputs (as noted above, yes, control inputs are recorded), the actions of the pilot couldn't have been more backwards:

Two minutes after that, the emergency began abruptly when the autopilot shut itself off. "I have the controls," one of the co-pilots said.

Almost immediately, the pilot pulled the jet into a climb, rising from a cruising altitude of 35,000 feet to 38,000 feet.

If pilots lose accurate speed information, they are taught to fly level and maintain the same power setting until they can diagnose the problem.

However, at some point the pilot reduced the jet engine power to idle. Even as the jet's forward speed slowed and it entered a dangerous stall, the pilot continued to try to pull the nose up, the BEA said.

Based on that, my previous post wasn't quite accurate: this is basically saying that the pilot caused the stall. Except for doing it on purpose, the only thing I can think of that would make a pilot make such control inputs is for the instruments to tell him he was going too fast. Still, he was apparently acting against his training.

 

[AlephZero]

The official interim report (English translation):
http://www.bea.aero/fr/enquetes/vol.af.447/point.enquete.af447.27mai2011.en.pdf

Better to try to follow that (not easy!) than use journalists' attempts at explaining it.

According to the BBC, the full report will probably not be poblished till "next year".

 

[AlephZero]

Granted, but assuming the article is accurate in the statements about the control inputs (as noted above, yes, control inputs are recorded), the actions of the pilot couldn't have been more backwards: Based on that, my previous post wasn't quite accurate: this is basically saying that the pilot caused the stall. Except for doing it on purpose, the only thing I can think of that would make a pilot make such control inputs is for the instruments to tell him he was going too fast. Still, he was apparently acting against his training.

They report the engines were throttled back to idle, that is recorded is it not? So a pilot obviously interacted there to do that much (wrong action in a stall).

Selective reporting. At the first stall warnig (which may have been wronigly triggered by the incorrect speed indication) the engines were set to takeoff/go-around thrust, and responded with 100% fan speed. The control inputs were nose down which stopped the unexpected climb.

What happened after the second round of instrument failures is not so clear, but it would appear that by the time the engines were set to idle (at 10,000 ft on the way down) the situation was completely unrecoverable in any case.

 

[Andre]

It occurs to me that the descriptions here, reflecting what is actually known, are too fragmented to judge the cause of the accident. of course, it's tempting to think of the classic killer, freezing/blocking of the pitot tube, causing the speed indication to increase with altitude. One would expect that system redundancy would deal with that nowadays.

It occurs that the pilots could have made a better analysis of the situation. Throttles idle is a pretty gross reaction, which can only be justified in a steep nose down dive - which could have been rejected by the high angle of attack.

I read that the pilots are not presented with a raw angle of attack indication. This is amazing for a fighter pilot, to whom the angle of attack is the most important feedback on the dynamix of the aircraft. Accurate angle of attack information could have prompted the pilots to take the right decisions.

But in the end, I would not be surprised if the root cause would be the failure to avoid the thunderstorm after all.

 

[Andre]

... At the first stall warning (which may have been wrongly triggered by the incorrect speed indication) ...

Hmm the stall warning in the aircraft I flew were all triggered by angle of attack sensors as indicated airspeed is not the only factor determining the stall. G-load is an important second factor. But aircraft are supposed to stall at a certain -constant- angle of attack all the time. So knowing your AoA can be a life saver and hence my surprise that 'the angle of attack is not presented to the pilot'.

 

[AlephZero]

It occurs to me that the descriptions here, reflecting what is actually known, are too fragmented to judge the cause of the accident.

If the flight recorders survived as well as the press reports claimed, there will be much more raw data than was published in that 4-page report.

But as with any scientific investigation, "data" is not "information" till it has been verified. I would expect there that will be a long process with several steps along the way.

First, to get a best estimate of the kinematics of the aircraft through the event - what path it flew, and at what orientation. There are clearly some issues to be resolved there with inconstient and missing data.

Second, to get a best estimate of the control surface positions and engine power output. The report does not mention any particular issues there, so that may be simpler than step 1.

Then comes the hard part, which is figuring out how to make all that data consistent with the laws of physics. Specifically, what wind speeds (horizontal and vertical) was the plane flying in, to make the control inputs produce the path actually flown. That could well mean several months of computer simulations, and possibly some ground testing or test flying to validate the computer models if they are being used outside of their normal range of input conditions.

Only then are you close to being able to guess why the flight crew did whatever they did - and since they are not available for interview, it may never be more than a guess.

On the other hand a simplistic "cause: pilot error" doesn't achieve anything. Even if the cause was pilot error, you want to stop the same type of error happening again.

But in the end, I would not be surprised if the root cause would be the failure to avoid the thunderstorm after all.

From the small amount of data in the report, one of the first unanswered question is: from flying level at a normal cruisng speed and making a gentle turn, how did the aircraft get into a 7,000 ft/min climb for 3000 ft after the first stall warnings? My best guess would be the first par of stall warnings were both false, and the plane had just flown into the top of a thunderstorm with a wind speed of at least 80 to 100 mph upwards.

I read that the pilots are not presented with a raw angle of attack indication. This is amazing for a fighter pilot, to whom the angle of attack is the most important feedback on the dynamix of the aircraft. Accurate angle of attack information could have prompted the pilots to take the right decisions.

The flight crew do have the artificial horizon and the climb and turn rates displayed, which are logically equivalent to the AOA, and for normal passenger flying more useful to meet the basic operational requirement - don't spill the first class passenger's drinks, especially if they are major shareholders in the airline.

 

[Andre]

On the other hand a simplistic "cause: pilot error" doesn't achieve anything. Even if the cause was pilot error, you want to stop the same type of error happening again.

About pilot/human/crew coordination error: If there is no clear sign that there was something wrong with the aircraft, then by elimination, some error was made by somebody. But the most simple one is, getting too close to a thunderstorm.

From the small amount of data in the report, one of the first unanswered question is: from flying level at a normal cruisng speed and making a gentle turn, how did the aircraft get into a 7,000 ft/min climb for 3000 ft after the first stall warnings? My best guess would be the first par of stall warnings were both false, and the plane had just flown into the top of a thunderstorm with a wind speed of at least 80 to 100 mph upwards.

I agree on the updraft, maximum climb rates at that altitude would be more like 500-1000ft/min, I would think. But how about the increase in angle of attack when you hit an updraft like that? With the inertia of the aircraft being slow to react, that would certainly have potential for approaching critical angles of attack. So that actually supports the idea of an unusual strong updraft.

The flight crew do have the artificial horizon and the climb and turn rates displayed, which are logically equivalent to the AOA, and for normal passenger flying more useful to meet the basic operational requirement - don't spill the first class passenger's drinks, especially if they are major shareholders in the airline.

I beg to differ, the angle of attack is the first to react on any pitch input or/and any turbulence/disturbance. You're seconds ahead in the control cycle and much easier to avoid chasing controls, using AOA, rather than airspeed which is much better for major shareholders. http://www.migman.com/sw/F-16_Aggressor/Landing.php

Edit: ignore the landing directions in that last link, that may work in a flight simulator but not in reality.

 

[Greg Bernhardt]

I just found this really good but haunting play by play
http://www.popularmechanics.com/print-this/what-really-happened-aboard-air-france-447-6611877?page=all

 

[turbo]

Scary.

 

[russ_watters]

Chilling.

 

[russ_watters]

So it sounds like the cause was faulty speed indication leading to a stall, then pilot error in recovering: http://content.usatoday.com/communi...rance-flight-experienced-35-minute-freefall/1

So I was wrong: the pilot error was much worse. Greg's article basically says that aside from sporadic and not overly serious airspeed indication faults, there was nothing wrong with the plane: nothing affecting it's flyability. The crash was caused by the one co-pilot applying full back pressure to the stick for nearly the entire event and neither the other copilot nor the captain recognizing it (except for a brief moment during the first stall) until it was too late.

All other issues relate to whether the other pilots - or indeed the airplane itself - could have prevented him from flying an airworthy airplane into the ocean. So sad.

 

[mheslep]

Somehow, automated warning systems are again being dismissed by humans in control. It seems they were oblivious to the dozens of stall warnings meaning they've come to not trust them somehow. That period where the stall warning went away because of attitude, later to return, while the aircraft was always in stall didn't help.

I see similarities to the subway crash in Washington, DC some months ago: automated signal failure, engineer oblivious, one train crashed into another.

 

[Greg Bernhardt]

New article from the Telegraph

With the report into the tragedy of Air France 447 due next month, Airbus’s 'brilliant’ aircraft design may have contributed to one of the world’s worst aviation disasters and the deaths of all 228 onboard.

http://www.telegraph.co.uk/technology/9231855/Air-France-Flight-447-Damn-it-were-going-to-crash.html