On-line Opinion Magazine…OK, it's a blog
Random header image... Refresh for more!

Preliminary Air France Findings

The BBC reports on the first finding released from the recovered data and voice recorders: Air France Rio crash: Jet plunged in minutes

The Air France jet which crashed into the Atlantic en route from Rio in 2009 stalled and fell in three and a half minutes, French investigators report.

The air accident investigations bureau (BEA) found the crew had struggled with contradictory speed readings just before the plane crashed.

The BEA statement did not look at the causes of the crash but one theory is that the jet’s speed probes failed.

Flight AF 447 went down on 1 June 2009 after running into an intense high-altitude thunderstorm, four hours into a flight from Rio de Janeiro in Brazil to Paris.

One of the instruments showed “a sharp fall” in air speed as the plane entered a zone of turbulence, the stall warning sounded and the autopilot and auto-thrust disengaged, the BEA said in its statement.

As the plane slowed, it climbed to 38,000 ft (11,600m).

Based on the information it looks like as they were attempting to avoid the turbulence they were caught in an updraft, as 38,000 feet is an dangerous altitude for a passenger aircraft, and then lost lift and went into a spin.

Their instruments were giving contradictory information, so there was no obvious solution. In clouds, there would be no visual clues to help them figure out what the aircraft was actually doing. The data recorder shows a spiral into the ocean.


1 Badtux { 05.28.11 at 1:10 am }

Based on where the tail fin fell vs. where the rest of the plane was found, it appears quite likely they lost the tail at some point in all of this. Hmm, Airbus jets have no history of doing that, eh?

Personally, I think that a jet that has a history of tail fins snapping off because the pilot did an aggressive rudder movement has a problem, but I’m just a flightless waterfowl so what do I know?

– Badtux the Flightless Penguin

2 Steve Bates { 05.28.11 at 10:41 am }

I have never flown a plane bigger than a small model, but even from that experience I know that if you stall and lose airspeed, the likely result is disaster. With a model, you can see the damned thing. With a commercial aircraft, you need instruments that work reliably even in the worst conditions. Of course once the tail broke off, it was all over.

3 Bryan { 05.28.11 at 12:50 pm }

The data shows that the aircraft went down with a nose up attitude, which tends to indicate that they never figured out that the stall was real and they needed to put the nose down and “floor it”. It took it 3½ minutes to go from 38,000 feet to the surface, which is about 120mph … essentially terminal velocity with the aircraft falling nose up. It dropped like a rock.

There’s now way of knowing when the vertical stabilizer separated, it may have been on entry to the water, or the impact of smacking the water. The aircraft would have “flown” through the water to its resting place.

4 Badtux { 05.28.11 at 2:05 pm }

The vertical stabilizer was located 40 miles from the main debris field. Sorry, but this broke-back jet did *not* fly 40 miles underwater…

My suspicion is that the same updraft and turbulence that forced the jet up to 38,000 feet also tore off the stabilizer. If the jet was going 400mph at the time of the microburst encounter, this would explain why the tail was found 40 miles away from the main body of the jet… it got tore off first. It would also explain some of the confusion that was in the cockpit… an Airbus without its vertical stabilizer is virtually unflyable, the machine would have been responding entirely differently from what the crew and more importantly the fly-by-wire system expected, and if the fly-by-wire system had a case of brain freeze from all this weird data it was getting and started feeding weird things to their displays or blanked their displays , they would have literally been flying blind, since the fly-by-wire system is what flies an Airbus, not the pilot. The only things that you can control on an Airbus that’s not fly by wire is the trim tabs and the engine thrust… and they *did* push the engine thrust all the way.

Note that Airbus has been denying for years that there is a problem with their tail, yet tails keep falling off. Tails don’t fall off of any other passenger airliner. Just sayin’ :twisted:.

– Badtux the Flightless Penguin

5 ellroon { 05.28.11 at 2:07 pm }

Having flown recently, this is so scary and sad. It’s always in the back of one’s mind when travelling that the flight one’s on might be the last one. So three minutes going down seems like it would have been an eternity…

6 Badtux { 05.28.11 at 4:19 pm }

Okay, found an analysis by an actual Airbus driver. He says the instrument readings show the tail was still intact when all hell broke loose, likely it broke off when the pilot kicked the rudders trying to break the spiral because the first thing that happened when the instrument panels blanked out was that the rudder limiters got turned off. And he says that according to what data he’s seeing, the computer shut down *everything* and *all* the alarms were going off — the overspeeding alarms, the stall alarms, *everything*, and all the panels were doing was scrolling error messages faster than the pilots could read them. He says that if altitude (fed by air pressure gauges that apparently kept working) said they were going down and the last speed reedings before things blanked were fine and attitude isn’t available because the attitude vanes froze over just like the speed pitots, it’s not surprising that they tried to keep the nose up and put on speed… unfortunately, at the attitude they’d gotten the nose to, no amount of juice was going to get them out of the stall, but they had no way of knowing they’d gotten the attitude so wrong.

So basically: The electronic instruments virtually all went blank, there were no mechanical steam gauge backups that would function in a thunderstorm, and it was pitch black outside so no horizon. At that point all the pilots could do was guess at the true situation and respond to the guessed situation. Unfortunately they guessed wrong :(.

7 Bryan { 05.28.11 at 10:11 pm }

The other thing that have people wondering is that the data recorder reported that the throttles were at idle the whole descent – it pancaked in a stall with the nose up at 35°. It was probably assisted by a downdraft on the other side of the updraft that took them to 38,000.

The “black boxes” are in the tail section, and the debris field on the bottom is fairly compact. The vertical stabilizer is a composite and floats, as did most of the early debris that was recovered. It took days to get ships to the location, so the debris had time to drift from the point for impact.

They need something simple in the cockpit that will work when the instruments lie, like a pendulum in a jar or a centering bubble indicator like I have on my tripod. There’s a lot to be said for the old ball attitude indicator.

I hated it when we had to go to 40,000 and above, because you have to resort to oxygen and in the event of decompression, you have to use pressure breathing, i.e. the oxygen is entering under pressure and you have to work to exhale. Standard breathing doesn’t working in the near vacuum of those altitudes and you won’t have time to put a mask on before passing out. We didn’t stay up there long, the engines barely function, but it was really tiring.

8 Badtux { 05.29.11 at 2:48 pm }

Actually, the first thing they did was put on 100% throttle to try to shut up the stall warning. But remember, the overspeed warning was going off *too*. So the next thing they did was go to 20% throttle to shut down the *overspeed* warning. When that didn’t work, they went to something like 40% throttle, which would generally result in the aircraft gently descending if it were in a slightly nose-up position, but clearly the crew didn’t know their nose was up in stall position (no instruments, remember?), they had no airspeed indicators working, no attitude indicator working, the fact that the aircraft was descending rapidly while the *overspeed* warning was going off to them meant that the nose was pointing way down and the correct thing to do was therefore to pull up. They had to guess as to whether the overspeed warning or the stall warning was correct. They guessed wrong :(.

Or as the Airbus driver put it: “They were *so* fucked.”

– Badtux the Flightless Penguin

9 Bryan { 05.29.11 at 8:05 pm }

I checked again and see that the article said that throttles were at idle on impact, my bad.

I still don’t understand how they didn’t know they were in a stall – their stomachs would have told them they were free-falling. In a stall you increase power and, if you have altitude, you point the nose down to gain speed. They never went below a 35° up angle. It makes no sense. That’s a primary flight training emergency drill for recovery from a stall.

I wasn’t there, so I’ll have to wait to see what the investigation board says, but it looks like the pitot tube icing and pilot error will probably get equal billing.

10 Badtux { 05.30.11 at 3:08 pm }

They knew they were falling. Their altitude gauge was still working, it apparently is based on radar and air pressure, not on the pitot tubes or attitude vanes. The question is, *why* are you falling? The overspeed *and* the stall warnings are *both* going off, and the consoles are scrolling gibberish error messages past you faster than you can read, and there’s no horizon visible out the windows to tell you what your attitude is, and your altitude says you’re falling. You have two possibilities — either you’re nose-down in a spin overspeeding towards the ground and you need to pull up to get out of the spin, or you’re nose up in a stall and you need to push down to get out of the stall. They chose the “we’re nose-down in a spin overspeeding towards the ground” option because the last thing they saw before the computer went whack was that they were at a too-high altitude with plenty of speed. Then the computer kicked out and both the stall *and* overspeed warnings started howling. They had to choose which alarm was correct. They chose wrong. I suppose that’s “pilot error”, but really, given the utter lack of any usable instruments that would tell them their attitude, it would have been a guess either way.

In short, minus some way of knowing their attitude, they were pretty well f*cked. This is a case where an old-timey gyroscopically-stabilized mechanical attitude steam gauge would have saved their bacon. But they don’t have one in an Airbus cockpit… the closest is a tiny bubble level used to tell whether you’re parked on level ground so that the fuel load balancing can be done right, but it would have been useless in an aircraft being tossed all around in the middle of a thunderstorm.

– Badtux the Flightless Penguin

11 Bryan { 05.30.11 at 4:33 pm }

Badtux, I’m looking at the authorized English version of Official Report [PDF] and there is no mention of any indication of a worry about overspeeding.

You must be reading something else, or someone is making guesses, because there is nothing that would suggest high speed as a factor.

They reduced speed when they started climbing on the updraft, but that was as close as anything I can find for labeling high speed as an issue.

12 Badtux { 05.31.11 at 12:09 am }

I’m not sure where the Airbus pilot in question got the bit about the overspeed sensor *also* going off, he does have access to data not in the public report though. 38,000 feet is a dangerous altitude for an airliner, combined with turbulence and cold temperatures you are quite near “coffin corner”. It is again an issue of what did the pilots think they were facing — did they think they had hit “coffin corner”, stalled, and were now in a nose down attitude gaining speed? That’s what their actions certainly seem to indicate that they believed. Absent a working attitude indicator, with the screens scrolling error messages past them faster than they can read, they were flying blind.

One thing the Airbus pilot mentioned is that when the computer kicks out, *all* feedback on the controls ends. You literally don’t know where the center is, because the feedback is provided via electrically controlled selenoids that provide “resistance” similar to what would be provided by mechanical controls acting hydraulically upon control surfaces in a traditional airliner and once the computer kicks out, all you have is one of those floppy computer joysticks that gives you no real feedback at all. Did the pilots think they had the controls centered in a neutral attitude when they were actually pulling back on the controls? Again, we don’t know, at least not from the published report.

One thing I’ll point out: As with the USAF, the BEA likes to pin things as “pilot error” whenever they can. The summary we were provided is *not* a full transcript of what happened, it is an edited one. Selectively? Again, we don’t know…

But the one thing that struck me from what that Airbus pilot said was just how f*cked those pilots were once the computer started scrolling error messages rather than showing them attitude and speed. They were operating in the realm of blind guess then, Sully himself couldn’t have gotten out of that situation. No feedback on the controls so no idea what position the control surfaces were in. No attitude. No speed. Contradictory error messages scrolling across the screens. The pilots did everything wrong, but they were fucked whatever they did, because if they’d chosen the other extreme (push forward on the controls, push throttles to 100%) they would have ended up arrowing into the ocean too. Without any way of knowing what their angle of attack was, there was no way that jet was going anywhere but where it ended up.