Air France Flight 447

[PRB]

In 2009, an Air France Flight 447, an A-330, vanished over the mid Atlantic, flying from Rio De Janeiro to Paris. They still don’t really know what happened, since they never found the data recorders. But the folks from NOVA did their own investigation and arrived at some interesting conclusions (SWAGs...?).
<?xml:namespace prefix = o ns = "urn:schemas-microsoft-comfficeffice" /><o></o>
ATC stopped receiving pilot reported position reports about the time the plane disappeared from radar, but that didn’t worry anybody at the time, since it’s routine to loose contact with trans-ocean flights in this way. As it happens, at this point the plane was fine, as it continued to send automated data link position reports for some time after radar and comms were lost. In fact it was the ending of the automatic position reports that lead searchers to the small amount of wreckage they eventually found.
<o></o>
But what happened? It seems they went down around the time they flew through a line of thunderstorms. And even though they don’t have the data recorders, the A-330 sends flight control data automatically via satellite to Air France HQ for maintenance purposes. This data shows a series of flight control failures just before the data stopped.
<o></o>
NOVA’s theory is that they encountered “super cooled liquid water” which, when it hit the plane, would have overwhelmed the pitot heat system, resulting is loss of airspeed data, and the subsequent FCS failures. This seems thin. Is there any documented case of this happening before? While there are procedures for flying safely with no airspeed indication, NOVA speculated that if it happened at the same moment the crew slowed down to penetrate the thunderstorm, the highly automated FCS of the A-330 might have caused the crew to be unable to keep up with the changing situation in the dark, leading to a stall. There is, after all, only 20 knots, at 35,000 feet between overspeed and stall. Then the speculation continued that since the A-330 is so automated, that the pilots are “not used to” flying manually, and might have been unable to recover from the stall. I find this difficult to believe. Bottom line is nobody really knows what happened. I sure hope they do some day find the “black boxes”. We can’t have modern air liners just up and vanishing over the empty ocean!
<o></o>
** OT **
20 knots separate overspeed from stall in an A-330 at 35,000 feet? Something doesn’t add up here. The NOVA folks showed the pilot, in a sim, dialing back the selected speed from Mach 0.79 to 0.75 while explaining that this is done when entering a thunderstorm. This after they explained to us that there is only 20 knots that separate stall from overspeed. I’m pretty sure Mach 0.75 to 0.79 is more than 20 knots…
** OT **

[HvyEng]

As a former professional aviator with a BS in Aviation safety and currently finishing a BS in Mechanical Engineering, I can say with a qualified background that although I do not agree with their conclusions the math is in the right ballpark. Remember, the speed of sound is based on temperature, not altitude. Thus at 35,000 feet on the ole’ SMOE table gives us a temperature of -54 degrees and a Speed of sound of 576 knots. Therefore hitherto and such, .79X576=455, and .75X576=432, 455-432 (in most textbooks) is 23. Soooooo, you're right, it is more than 20 and they were off by three knots, which for NOVA is a lot. Optimum flight levels are based on climb rate, as are the various ceilings. There is a phenomenon that occurs when your climb rate asymptotically approaches 0fpm as a constant attitude is maintained, in which your cruise, climb, and stall speeds are only separated by a knot or two. We used to call this the coffin corner of the spec-range charts. We generally avoided that area for obvious reasons, but it was usually only 4000 feet or so above our computed "Best range" altitude.<?xml:namespace prefix = o ns = "urn:schemas-microsoft-comfficeffice" /><o></o>
<o></o>
I have seen icing so severe that even with all equipment on high/emergency the windscreen glazed over completely and the fans were shedding so much ice every single acoustic panel was destroyed, but I have never seen the pitot system not keep up. (Not to mention the GPS system does provide ground speed information) There is a reason why pitot tubes will actually cause covers to catch on fire if they are left on (hence the massive press about the faulty elements). I also don't agree with the pilots inability to handle manual operation. Pitch and power are basic pilot skills, and Air France has some of the best pilots the business; which their safety record stands testament too. My personal opinion; A high altitude lightning strike rendered either the FBW (fly-by-wire) direct law system nonfunctional, or the FADEC's reverted into the (I think) third or fourth tier failure mode of "Idle", preventing correct stall recovery of the aircraft. I do not believe there is any sort of "fatal flaw" or conspiracy with the Airbus systems. They build great aircraft and I would fly them in a heartbeat, but I can't imagine any other reason to hold a stall for that long, and from FL350 to sea level they were fighting it for quite a while. God rest their souls, it horrifies me to think of the battle that ensued between man and machine that fateful night.<o></o>
<o></o>
--Dan<o></o>

[PRB]

Interesting, Dan. The first thing I thought of was either lightning taking out critical avionics and flight controls, or turbulance causing fatal damage. Analysis of the pieces that were recovered suggest the plane was intact when it hit the water, so it apparently didn't break up in flight. Also, among the FCS failures reported by the maintenance reporting system was a "exceedance of rate of change of pressure" which suggests vertical speed. I do hope they get to the bottom of it at some point.

[Railrunner130]

While I have no comment (or clue) on the cause, I do find it interesting that the pitot static system was immediately pointed to as a "known" potential cause and (I think) an emergency Airworthiness Directive was issed to replace certain models of the pitot static system. I wonder if it was a kneejerk reaction or if there is more to the story.

[deathfromafar]

I know a few guys who fly Buses & their general consensus is that the Pitot/Sensor system could easily be overwhelmed by ice in the MC that flight was in. The ITCZ in that region was exploding with super cells that day & night. I remember seeing the daytime visible water vapor satellite photo which Meteorologists said was exceptional that day. There was a nearly solid line of convection and development stretching from Africa to across Venezuela and AF447 flew right into the heart of it. The ACARS transmitted 24 messages within a 4 minute period and 14 in the final minute. 3 of the speed data inputs showed a discrepancy of 30 knots in less than one second. That alone makes it clear that the sensors were not functioning normally and likely due to element icing/ blockage. There initially was talk about a lightning/multiple lightning strikes possibly being the cause of the system failures but if I understood what Airbus said, the data ruled that out. What is known is that the FBW FCS went into Direct Control Law and the Autopilot went offline at the exact same moment. Under the Meteorological Conditions and at night, it is quite conceivable that with the conflicting airspeed & altitude data and severe turbulence upset that the crew could have mishandled the plane and caused it to enter a deep stall. The investigators said they have evidence the plane hit the water intact flat and upright.

Regarding accidents due to sensor failures, I recall the Birgenair 757 that crashed in the Caribbean in which one of the Pitot tubes was blocked which caused conflicting speed readings on the ASI's. The sharp and rapid changes in airspeed readings which confused the pilots causing them to make the wrong inputs & thrust settings which led to both a engine stall/asymmetric thrust and stall/spin to inverted position prior to impact.

Bottom line, if you're in IMC/Convection & at night, the instruments go Tango Uniform, all the pilot training and experience in the world is unlikely to save you.

Speaking of the ITCZ, a buddy of mine(who is a member here) and myself have flown through the ITCZ on a few days and nights where things got very rough. On one occasion, we were on a 757-200 and flew through the absolute worst weather either of us has ever seen from the inside of a plane. After we landed, we saw the Captain and his face was pretty pale & diaphoretic. I wasn't in the best shape after that myself!

[Bjoern]

Then the speculation continued that since the A-330 is so automated, that the pilots are “not used to” flying manually, and might have been unable to recover from the stall.

That's the most stupid attempt at finding a contributing factor that I've ever heard.

Regardless of equipment, pilots are still pilots. Judging their skill from the equipment they fly is like degrading them to button pushing monkeys (No offense, apes. You did great during Mercury-Redstone 2 and Mercury-Atlas 5!) who can't do squat without the autopilot.

Yes, pilot error can be a decisive contributing factor, but not because they're ill prepared to fly their plane but rather because they get into situations not or badly covered by training and make the wrong decision due to a lack of experience.

[Bone]

I would like to throw in my practical test in regards to the MACH/KIAS spread. The conditions were:

Altitude = 37,000ft
SAT (Saturated Air Temp, basicly OAT) = -53
TAT (Total Air Temp, which is SAT corrected for A/C skin friction) = -20


Watch as I scroll the speed bug down, but pay attention to it as I scroll it upwards again. Compare it to the KIAS ladder, and you'll see that MACH .75 is equivilent to 241 KIAS, and MACH .79 is equivilent to 258 KIAS...a spread of 17 knots. When you look at the speed bug, on the right hand side is a cut out wedge in the center, and the point of the wedge faces left. That's the "on speed" pointer.


Click on the pic to see the video

[txnetcop]

I believe it came down to this: They flew through what the radar showed as a minor storm, but behind it hidden from radar was this massive monster storm. All three pitot tubes were filled with supercooled ice that even the pitot heater could not handle hence no way of gauging speed for the autopilot so the plane starting shutting down system connected with it. A variance of of only twenty knots plus or minus could have sent the aircraft into an unrecoverable wing stall because the air around the aircraft was extremely unstable. The pilots it seems were overwhelmed by the flood of system shutdown information. Of course there is even more to it than that, but I do believe that more extreme weather training would be warranted. One fix that could have been used was to trim 5 degrees nose up and take power to 85% power or mach .75.
Ted

[NWarty]

Speaking of the ITCZ, a buddy of mine(who is a member here) and myself have flown through the ITCZ on a few days and nights where things got very rough. On one occasion, we were on a 757-200 and flew through the absolute worst weather either of us has ever seen from the inside of a plane. After we landed, we saw the Captain and his face was pretty pale & diaphoretic. I wasn't in the best shape after that myself!

My sister was a FA on a DAL 757 several years ago that went through some really bad turbulence. Passengers on both sides held her down in the aisle as the plane rolled violently. She mentioned the Captain and FO were both white as ghosts.

You guys???

[deathfromafar]

That's the most stupid attempt at finding a contributing factor that I've ever heard.

Regardless of equipment, pilots are still pilots. Judging their skill from the equipment they fly is like degrading them to button pushing monkeys (No offense, apes. You did great during Mercury-Redstone 2 and Mercury-Atlas 5!) who can't do squat without the autopilot.

Yes, pilot error can be a decisive contributing factor, but not because they're ill prepared to fly their plane but rather because they get into situations not or badly covered by training and make the wrong decision due to a lack of experience.

Bjoern, I'd like to make a point on this statement. Yes, Pilots are Pilots but the trending towards more and more automation has clearly created more dependency upon it and there has been much discussion throughout the industry and the regulatory arms on practical changes/updates in training to focus on failures & issues with automated systems. Between the Boeing & Bus Pilots I know, all of them practice manual flying during recurrent sim training but rarely do in air except takeoffs, approach/landings. Still, most of them have a lot of hand flying experience and are able to settle into the control/handling feel of the types they now fly. The Bus however is a different animal in terms of control feel(or lack thereof). It is very sterile compared to conventional control systems. The architecture of the Bus's FBW control system is well thought out and offers a lot of safety. The Pitch, Bank, & Speed protection is excellent. Manual flying of the Bus is absolutely easy once a Pilot is trained/checked out on it. But, the one area of the FBW FCS on the Bus that all the Bus Pilots I know admit they have little experience/training on is Direct Control Law(except takeoffs and landings where DCL is normal). Different airlines who operate the Bus have slightly different procedures and training regimens based on Airbus recommendations but it is fact that little time is spent on sim training in DCL with all the normal FBW safety systems bypassed. None of the guys I know who fly A330/A320/319's have ever flown the planes in DCL in flight(other than takeoff & landing when the FBW system switches automatically between DCL and Protected Mode). So in general, it is absolutely correct that most crews do not have a lot of experience flying the Bus types in DCL. There are plenty of scenarios where that can bite a crew pretty hard. The last recurrent classroom and sim runs my Bus flying friends have gone through have placed a lot of emphasis on this and many other aspects not covered before in their training.

[Bone]

Over reliance on the automation during line flying has put quite a few guys in a pickle come check ride time. It doesn't matter what airline it is, some guys turn on the automation just after take off and then off right before landing. Lazyness, I see it all the time. My former wife has been an A-320 FO for ten years (Northwest...merged Delta), and they've had a huge problem with over reliance on the automation for a very long time.

I hand fly to a minimum of 10,000 ft after take off (sometimes even up to the mid 20's), and click it off on the descent at about 8000 ft...unless it's a low weather approach or otherwise high workload environement.

[PRB]

Just to clear up the 20 knots between mach overspeed and stall, this means that from whatever the optimum speed is, they couldn't slow down or speed up more than ten knots. This doesn't appear to leave much room for slowing down prior to entering a thunderstorm. I've read that the U-2 had a 3 knot window between overspeed and stall at its operating altitude, which is amazing. I was surprised to learn that air liners have such a small "window" of this type at high altitude, although it makes perfect sense when you look at the numbers involved. I'll have to check the size of this window for FS air liners... A ton of very interesting comments in this thread.

[Bone]

Just to clear up the 20 knots between mach overspeed and stall, this means that from whatever the optimum speed is, they couldn't slow down or speed up more than ten knots. This doesn't appear to leave much room for slowing down prior to entering a thunderstorm. I've read that the U-2 had a 3 knot window between overspeed and stall at its operating altitude, which is amazing. I was surprised to learn that air liners have such a small "window" of this type at high altitude, although it makes perfect sense when you look at the numbers involved. I'll have to check the size of this window for FS air liners... A ton of very interesting comments in this thread.

Actually, airliners don't have a small margin....it's pretty darn huge, unless you load up with ice you can't shed, and then that raises your stall speed. Seemingly, it raised Air Frances stall speed to a point where they might have experienced coffin corner.

You have to be higher than the service ceiling to shrink the margins down to coffin corner (or loaded up with ice). In my video above, we were at 37,000ft, and we had a very wide margin between stall and overspeed. The max operating altitude for my jet is 41,000. Below is a shot of the PFD showing 41,000ft. If you look at it you'll see a vertical red bar that starts at 252 KIAS and goes up. We call that the "snake" and it shows you what speed an overspeed starts at. We also have a snake for stall speed, and it looks just like the overspeed snake. If we were to go up to 45,000 feet and higher, you would see the overspeed snake move down the airspeed ladder...250, 240,200, 180 ect...closer and closer to the stall snake. I've taken one of my company's simulators up high enough to where we were wedged between the two snakes (coffin corner). As you can see in this photo, we weren't anywhere near coffin corner. If we were, you would see the stall snake.

If the A-330 was in a coffin corner at 35,000, then he had to be loaded up with ice, and that raised his stall speed artificially. I've loaded up with ice before and the stall snake moved up the KIAS ladder...it's kind of a wicked thing to watch.

[txnetcop]

I believe he was ice heavy as well as the pitot tubes had been clogged quickly with supercooled super-condensed ice I am sure the wing boots cold not keep up with wing supercooled ice either. I guess it is an extremely rare condition...I'm glad I never encountered any. I did have my Cessna 140 fuselage and wings battered and holed by hail once...the pucker factor was pretty high that day. I can imagine how the flood of failures in succession like they experienced would be pretty unnerving.
Ted

[Z-claudius24]

Hi,

AF447 found

http://www.bbc.co.uk/news/world-europe-12953432

[wiltzei]

A recent official update, describing what happened.

http://www.bea.aero/en/enquetes/flig...ght.af.447.php Click on desired language -> link on the following site.

[AndyG43]

A recent official update, describing what happened.

http://www.bea.aero/en/enquetes/flig...ght.af.447.php Click on desired language -> link on the following site.

Mm, very interesting!

[Lateral-G]

BEA initial findings...........

(this is not the full report, but a factual account of the flight with data gleaned from the recovered flight recorders).

http://www.bea.aero/fr/enquetes/vol....mai2011.en.pdf

-G-

[PRB]

Damn... What the heck happened to that plane? Thanks for the update.

[Lateral-G]

Damn... What the heck happened to that plane? Thanks for the update.

Basically is was stalled with little fwd airspeed and fell at over 10K ft/min. It pretty much hit flat like they first postulated.

[Allen]

I haven't followed this much but how the hell dose a airliner get in a flat stall? Its not a P-39.

[stansdds]

After being lost in the Atlantic Ocean for nearly two years, the now recovered flight data recorder has been found to be functioning and is providing a wealth of information about the final minutes of the doomed flight.

Icing in the pitot tubes is likely the start of the disaster, but the pilot at the controls was the second in command. He put the aircraft into a climb, even when the stall warnings were sounding. The result was a stall and a descent rate of 10,000 feet per minute.

http://www.reuters.com/article/2011/...74Q28420110527

[flaminghotsauce]

Simply amazing. If what is being told here is actually what happened, the pilots did the unthinkable and stalled the airplane for 38,000 feet into the ocean without every commanding the nose down, or increasing thrust?!?

This is strangely difficult to swallow. Pilots who are regularly drilled on emergency procedures not recognizing a stall, with the stall warning sounding more than once?

[fliger747]

Back in the "olten days" we used to memorize attitude and thrust settings for level and other flight modes. We don't do this any more, but after staring at the instruments of the whale for some 15 years I have a pretty good idea what they should say even without airspeed or alttiude info. Quite a number of years ago a South American airline planted an otherwise good 757 in the Atlantic from taped over pito-static ports (the plane had been washed).

I don't know enough about the airbus in question to know how much the throttles and flight controls can be made to override the computer Fly-by-wire system.

As Paul Harvey would say.... there must be "The Rest of the Story"..... Both incidents were at night over the ocean.

t

[wiltzei]

Back in the "olten days" we used to memorize attitude and thrust settings for level and other flight modes. We don't do this any more, but after staring at the instruments of the whale for some 15 years I have a pretty good idea what they should say even without airspeed or alttiude info. Quite a number of years ago a South American airline planted an otherwise good 757 in the Atlantic from taped over pito-static ports (the plane had been washed).

t

Do you still check N1's an/or other values at final appoach for example? Regarding the 757, didn't Boeing add unreliable airpeed warning to EICAS after that crash?

- - - - -

By the way mods, there are many threads about same topic, time for a merge?