Robots Dive Deep To Solve Airliner Crash Mystery

coondoggie writes "A small squadron of undersea robots has begun to conduct a 4-month, 3,900 square mile search of Atlantic Ocean bottom looking for the deep-sea wreck site of and black boxes from Air France Flight 447 which crashed off the coast of Brazil nearly two years ago. The Air France plane was flying from Rio de Janeiro to Paris, when for exact reasons that remain a mystery, it crashed into the Atlantic Ocean on June 1, 2009, taking with it 228 souls."

Comments

Reasons unknown??

[Ark42]

Isn't this the flight that flew right into a huge huge storm that was obscured on their radar by a smaller storm which was safe enough to fly through. As soon as the larger storm was in view, it was too late to change course and fly around it. I heard the most likely case is extreme icing of the sensors that monitor airflow, causing autopilot to disengage as the plane no longer knew its own speed. Without any way to know the current speed, the plane lost altitude and crashed, due to a small window of safe speeds that don't result in altitude loss.

Re:Reasons unknown??

[bradgoodman]

I heard of this same sort of thing happening once to a plane. What happened was that the plane was just painted. During the painted process, they put masking tape over the Pitots (holes/ports used to measure air pressure). They forgot to take the tape off, and when they were in flight, the airspeed, altitude, and stall warnings all went crazy from the erronious pressure readings on the clogged/covered pitot tubes. Result was bizarre instrumentation - overspeed and stall warnings at the same time, etc. They wound up crashing from confusion. Perhaps icing in the pitot tubes were causing a similar thing here.

Re:Reasons unknown??

[martyb]

Not necessarily. Even without accurate airspeed readings, the pilots should have still been able to maintain safe airspeed by setting the engines to a specific power output and trimming to a specific angle of attack. Probably pilot error (i.e. being distracted with alarms and not remembering to adjust throttle and angle....) but without that box it's hard to really know.

Honest Question: Why in this day and age do we still have to chase down a black box? More and more airliners now provide in-flight internet connections. Couldn't they just transmit it as well as record it to the black box? TFA says this search is costing them $12.5 million. That would pay for a lot of upgrades and support for this.

Continuous Transmission? Send all of the recorded data to both the black box and some remote data center, too. If this is too much to transmit continuously, then maybe a subset of the data? I know planes are becoming increasingly complex and automated, so there's probably loads more data that *could* be considered for transmission. Still, something is better than nothing (what we have now.) Pick some subset of the available data and send it periodically.

Burst Transmission? Instead of a continuous stream of data, when the pilot (or plane) detects a "dangerous condition", it starts sending a high-speed burst of accumulated data, and continuously until things look "normal" again. Say the plane takes a sudden 200-foot drop in altitude. Or banks unusually sharply. Or... whatever. Just ignore the values that appear 99.9% of the time, and only trigger outside that normal range. (numbers pulled out of thin air; pick whatever works best.)

At this point, there's nothing much to go on. Imagine if we had the last few minutes' airspeed, altitude, as well as settings for the flaps, rudder, and engine would be an enormous improvement over what we've got now. I suspect the pilots' unions might raise a concern about monitoring and potential for it to be help against them, but I could also imagine some kind of escrow mechanism where the data is sent and stored, but only to be accessed upon certain, predefined circumstances.

Admittedly, this is quite rough. I'd like to think that there is at least some part of this which could be implemented in parallel to the provision of internet access on planes. I'd appreciate it if anyone who knows more about these things could comment on the viability of this and/or the technical limitations/challenges which I'm missing here.

Re:Reasons unknown??

[jamesrt]

I heard of this same sort of thing happening once to a plane. What happened was that the plane was just painted.

The plane crash being referred to is this one: http://en.wikipedia.org/wiki/XL_Airways_Germany_Flight_888T

Re:Reasons unknown??

[Slutticus]

Actually, the plane transmitted quite a lot of information to AirBus HQ as it was going down, but that system (ACARS) is quite outdated i think. The last three minutes of the transmissions gave a wealth of data related to alarms and faults that were occurring (i.e. inconsistent airspeed readings, excessive vertical speed, autopilot information, etc...). It would be interesting to see how much voice data could be reliably transmitted in a situation like that.

Re:Souls?

[jamesh]

I'm not religious but I think the concept of the soul at the very basic level is valid. It's the program running in your head that is you. That's about where it ends though - I don't believe in any way that the program keeps running once the hardware fails, outside of the bits of your program that have rubbed off on the other people that you interacted with along the way.

pitot probe failure most likely cause.

[bongey]

I write flight diagnostic software , special software the tries to determine the root cause after number of BIT/and OR ACARS messages. I was especially interested in this flight. Thank god the air bus aircraft sent the ACARS messages otherwise we would have no idea what happened to it. Nova video is pretty convincing. Especially when in the flight simulator , and they cause a simulated air speed failure. The exact same ACARS messages are produced by the simulator that were produced by flight 447. http://www.pbs.org/wgbh/nova/space/crash-flight-447.html . NOVA concluded bad weather caused the failure of the air speed sensors (pito tubes). Air speed sensor failure cause the auto pilot to fail, which turned the cockpit into a christmas tree of error and warning lights. Finally, pilot error in which they didn't react quickly enough .The pilots had to react quickly enough and apply just the right amount thrust and pitch to avoid a dramatic stall. The plane final injuries were consistent with dramatic stall, literally falling strait down out of the sky.

Re:pitot probe failure most likely cause.

[by (1706743)]

First thing I did when I opened this thread was Ctrl-F for "nova". I know nothing about the aeronautics, but I too found this to be a very convincing explanation.

What really struck me as odd was that (as I recall from the Nova video) planes are out of communication from land when in the middle of the ocean. With humanity's level of satellite technology (not to mention radio-wave-bouncing-off-of-atmosphere-skillz), this just seems weird.

Re:pitot probe failure most likely cause.

[Anonymous Coward]

I managed to get one of our more experienced pilots to follow the rabbit down the wrong hole in a similar way in the sim yesterday. It was easy, too. I gave the local pitot and static sources a bad pressure, and then faulted the B bus to kill off the good sensor. The pilot assumed that he was getting good data from the remaining sensors and failed to notice bank angle creeping as the AP pitched down to maintain speed in climb mode. What's 3 degrees of pitch when you've got cascading faults? Every few seconds the computer would spit out another fault; as something exceeded a time limit for being out tolerance.

The plus side is that mine has dual-string flight systems, while the bus has triple-string systems, so the two failures I input won't cause the same scenario. However, if they did encounter severe icing and iced up the pitot tubes, the aircraft could have departed controlled flight before any severe faults showed. My pull it out of my ass guess, though, is that they had an electrical fault before the pitot tubes iced, and didn't have a fair chance.

Nova Documentary on Flight 447

[catchblue22]

This is an excellent Nova documentary on the disappearance of Flight 447. It is interesting how investigators were able to give a reasonable hypothesis as to what happened, even without the black boxes. The long and the short of it is that they think super-cooled liquid water from a serious thunderstorm overcame the pitot anti-icing heating systems, freezing over all of the pitots and thus depriving the computer of airspeed data. The computer probably panicked, suddenly switching off the autopilot (they did get data from the computer, as its satellite uplink gave some telemetry). Pilots are capable of flying without airspeed readings, but only if they react quickly. They think that prior to flying into a severe thunderstorm, the computer automatically reduced thrust, in order to slow down in anticipation of turbulence. The problem is that the only pilot feedback that the thrust was reduced would have been a tiny circle on a computer monitor...there is no physical feedback in the throttle levers in Airbus planes. The computer then probably switched off the autopilot, overwhelming the pilots with a sequence of warnings. The thrust likely remained at 70% and the pilots probably didn't realize it. After a minute so the airplane may have lost so much airspeed from the low thrust that it became unflyable, in effect causing the crash.

Give this Nova episode a try...it is very detailed, going into many technical aspects of airplane design.