OT: Air France 447

[svelte]

This whole story has fascinated me since it happened, so I’ve been watching it closely. Now the transcript of the last two minutes of flight has been released.

Chilling really that those in the cockpit argued about the proper course of action right up until the plane hit the water. The pilot in control was apparently still pulling up when he should have been pointing the nose down to build up speed and get out of the stall.

http://www.cnn.com/video/#/video/bestoftv/2011/10/14/am-quest-air-france-447.cnn

Someone definitely needs to look into how the Airbus 330 can be improved in situations like this. I think we as humans have assumed that more automation is the answer to reducing air crashes. What I don’t think we’ve adequately addressed is that automation has also reduced pilot’s skills at reacting to unusual occurrences and emergencies. Not sure what the answer is, but it will be the focus of airplane design and pilot training for years to come.

And then there’s this story:

http://jalopnik.com/5844646/a-passenger-airplane-nearly-flew-upside-down-because-of-a-dumb-pilot

[kev374]

interesting, I just watched the documentary of this on Netflix.

[AK]

Just a matter of time before the very industrious Airbus astroturfing crowd shows up here to attack Boeing’s safety record and “unfair” U.S. subsidies …

But I digress. Yeah the crash of Air France 447 really does cut to the heart of the interaction of man and machine. I don’t work in aerospace (much to the benefit of the flying public) but I run into the same problem with software user interfaces … make them too simple and people lose engagement to the point where data entry errors actually increase.

[CDMA ENG]

[quote=kev374]interesting, I just watched the documentary of this on Netflix.[/quote]

Actually I just watched it as well on Netflix about 2 months ago.

The largest take-away from the show was that the pilots failed to execute proper prodcedure and thus were entirely resonsable. In the documentary, Nova I think, they threw the exact same failures at senior instructors in the cockpit. These instructors were not briefed to what they would be doing and had recovered the aircraft in about 15 seconds. Thus providing evidence that the failure to recover was due to a lack of proper response.

The recovery efforts they invoked were the same Airbus bragged about years ago when the debuted a new series about the aircraft being stall proof.

What the instructors did was to raise the nose, contrary to what most pilots would do in a stall, and to go flaps down with the engines to 85 percent power. This places the aircraft in a very stable configuration where the plane’s wing will generate enough lift to maintain flight in all conditions.

The reason, they explained, that you dont go to full power and point the aircraft down has everything to do with the wing design. The aircraft’s wing, if it is going to fast, start to build a shock wave over the leading edge and interupts the airflow to the extent that the wing will lose lift. The used another term for this but I know the concept is also call compressibility. That is why they pitched the nose/wings up so that wings will create lift while not inducing compressibility.

Air France is trying to quash this information because they know they have a huge lawsuit coming.

This also goes directly to what the captain of the East River landing said about the lack of experienced pilots.

Makes you think…

CE

[AK]

[quote=CDMA ENG]What the instructors did was to raise the nose, contrary to what most pilots would do in a stall, and to go flaps down with the engines to 85 percent power. This places the aircraft in a very stable configuration where the plane’s wing will generate enough lift to maintain flight in all conditions.[/quote]

Interesting … Sounds a lot like what the captain in the Colgan Air crash did, except the copilot simultaneously raised the flaps.

[CDMA ENG]

[quote=AK][quote=CDMA ENG]What the instructors did was to raise the nose, contrary to what most pilots would do in a stall, and to go flaps down with the engines to 85 percent power. This places the aircraft in a very stable configuration where the plane’s wing will generate enough lift to maintain flight in all conditions.[/quote]

Interesting … Sounds a lot like what the captain in the Colgan Air crash did, except the copilot simultaneously raised the flaps.[/quote]

So here is a youtube video of the prodcedure. Difference here is that that the computer is still in control of the throttles becasue it is getting data from the pitot tubes. In 447 they were not…

http://www.youtube.com/watch?v=cggDnHVDoCA&feature=related

I actually found a four part series on youtube. Watch this video at the nine minute mark…

http://www.youtube.com/watch?v=JqTPbPlyX5E&feature=related

The scariest thing about this whole video is that at attitude the wing only has about 20 knot between stall and compressibility. Its a very tight airspeed to have to maintain and now everytime I hear the motors on a aircraft kick back it spooks me!

CE

[OwnerOfCalifornia]

A new article from Popular Mechanics. An interesting (and chilling) read.

[quote]The recovery efforts they invoked were the same Airbus bragged about years ago when the debuted a new series about the aircraft being stall proof.

What the instructors did was to raise the nose, contrary to what most pilots would do in a stall, and to go flaps down with the engines to 85 percent power. This places the aircraft in a very stable configuration where the plane’s wing will generate enough lift to maintain flight in all conditions.[/quote]

A nitpick. This procedure is strictly a reaction to unreliable airspeed indication, not an actual stall recovery technique. By configuring the aircraft as described, it will remain in the flight envelope and not stall. The linked article explains that in ‘normal’ law, the A330 will remain in the flight envelope regardless of the pilot inputs to the controls (autopilot or not). Since airspeed had failed, the aircraft was operating in ‘alternate’ law, and all control inputs were taken literally.

[Coronadoflyer]

For the unreliable airspeed you would set a power setting, like 88% N1 and a nose attitude, 2 – 3 degrees nose up and go from there. This would have saved Air France, but it is easy to be a monday morning quarterback.

[ucodegen]

[quote CDMA ENG]The reason, they explained, that you dont go to full power and point the aircraft down has everything to do with the wing design. The aircraft’s wing, if it is going to fast, start to build a shock wave over the leading edge and interrupts the airflow to the extent that the wing will lose lift.[/quote]Compressibility effect occurs trans-sonic/supersonic not low speed. This is why there were several accidents when trying to break the sound barrier. The shock wave prevented sufficient airflow from getting to the control surfaces. Next Miramar airshow, take a look at the leading edges on F18, F16s. You’ll find they are very narrow/thin. Compare that to a commercial sub-sonic plane, which has a fairly thick leading edge and to private planes which also have a fairly thick leading edge compared to their size and weight. The thicker leading edge gives an advantage due to a larger range of Angle of Attacks(AOA) that the wing will operate over (angle of attack being the angular difference between a line through the width of the wing and the vector representing the direction that the air is impacting the leading edge).

Compressibility is also the reason why supersonic aircraft try to have a smoothly transitioning cross section – ie have a coke bottle shape where the fuselage actually gets a little thinner in cross section when near the wings.

What happens at high AOAs is that as you approach stall, the laminar flow on the back(top) of the wing actually detaches from the wing surface. When that happens, you loose lift from the wing. There are all sorts of tricks done to prevent this. Next time you are on a flight, take the window seat next to the top of the wing. You’ll see small little blades sticking up from the top, near the chord of the wing(highest point, thickest in cross section) going lengthwise along the wing. You’ll notice that they are at a slight angle to the airflow. This is to create a vortex near the top wing surface that ‘glues’ the laminar flow down to the surface of the wing. If on the flight, you are on takeoff or landing and there is a moderate amount of moisture in the air – you can actually see the vortexes. Another thing to note, is that when flaps on commercial aircraft are deployed near their fullest, the leading edge of the wing actually pulls away from the wing and drops down. This is to bleed high pressure air from under the wing and ‘jet’ it close to the surface to ensure laminar flow at high Angle of Attacks.

There is another piece of the wing that you may see move. It will be near the fuselage of the wing is a large panel that tilts up from the top of the wing. Most of the panels will be in close to the fuselage. This is the ‘spoiler’ and it does what its name says. It ‘spoils’ the airflow, killing lift and increasing drag (also acting as a brake) when it comes up. It is useful at low speeds because the ailerons at the tips don’t have as much a bite on the airflow at low speeds. It has more of a ‘brake’ effect at higher speeds. They are mostly on the part of the wing nearest to the fuselage because they have a very large effect on the aircraft. If they were further out to the tip, they could easily flip or roll the aircraft.

[ucodegen]

[quote flinger]By configuring the aircraft as described, it will remain in the flight envelope and not stall. The linked article explains that in ‘normal’ law, the A330 will remain in the flight envelope regardless of the pilot inputs to the controls (autopilot or not). Since airspeed had failed, the aircraft was operating in ‘alternate’ law, and all control inputs were taken literally.[/quote]This ‘normal law’/’alternate law’ behavior is something I find kind of freaky. There should be some sort of ‘audible’ or ‘visual cue’ that it is operating in ‘alternate law’, not just a warning for something else. It is like driving a car that behaves one way under ‘computer assistance’ and another without – (Stabilitrak works this way – and is kind of freaky in adverse conditions.).

What gets me is the following:

[OwnerOfCalifornia]

[quote=ucodegen]This ‘normal law’/’alternate law’ behavior is something I find kind of freaky. There should be some sort of ‘audible’ or ‘visual cue’ that it is operating in ‘alternate law’, not just a warning for something else. It is like driving a car that behaves one way under ‘computer assistance’ and another without – (Stabilitrak works this way – and is kind of freaky in adverse conditions.).[/quote]

There is supposedly an audible enunciation for the switch to alternate law. Furthermore, the aircraft was necessarily in alternate law given the unreliable airspeed indication–and the pilots presumably understood this.

[quote=ucodegen]What gets me is the following:

It looks like no-one looked at the artificial horizon, so they didn’t seem to notice the extremely high AOA. This device functions separately from the pitot tube.[/quote]

The artificial horizon displays the aircraft pitch and roll attitude, not the AoA. This instrument wasn’t necessarily displaying anything catastrophic, even after the plane was hopelessly stalled with a 40° AoA.

[quote=ucodegen]

It also doesn’t seem that they paid attention to the dive-climb and altitude indicators. Combined with high AOA and the altitude spinning downward with a high decent rate – should have clued everybody as to what was going on. Commercial pilots are supposed to be IFR rated. The pitot tube is used for altitude, but only the side vent, not impact pressure. They did get the pitot tube working part way through (giving them good airspeed indication and altitude indication).. and started flying normally.. then with all instruments working, they crashed the plane with poor flight decisions.[/quote]

The static port is located on the fuselage somewhere (certainly on older general aviation aircraft), nowhere near the pitot tube. While it could have iced over, nothing indicates that it did. The pilots were seeing descents as high as -10,000 feet/min. They were certainly aware of what was happening to the aircraft. But…

[quote=ucodegen]

The youngest pilot, the one in control, seems to have his driven more by panic than conscious rational decisions. When told by the superior in the cockpit to release the stick from full-back position, that this was part of the problem, this individual grabs the stick and slams it into full-back position again after hearing that they are nearing the ground. If the plane kept horizontal and made contact with the water, it would have been more of a pancake landing – and more people may have survived.[/quote]

Yes, it appears the young FO reacted tragically to the information he did still have. Aeronautical decision making (ADM) has become a critical new addition to pilot training, even starting at the basic level for private pilot students.

Also, I believe the plane did contact the water in an almost level attitude–but descending at -10,000 feet/min. That is not survivable, and the plane disintegrated on impact.

[quote=ucodegen]

I am amazed that the A330 flight control system doesn’t consider having wildly differing inputs on the flight control as an unusual condition and proceeds to ‘average’ the controls instead. There should be some kind of feedback to the other pilots stick as well as the ability to disable one of the ‘seats’ (imagine a control that decides to fail – would want to disable it). The old systems had the yoke move together.. so you could tell if the other pilot was giving irrational input.[/quote]

I believe Boeing aircraft still have coupled controls. The uncoupled Airbus controls are certainly a point of controversy regarding the AF 447 accident.

[ucodegen]

[quote flinger]There is supposedly an audible enunciation for the switch to alternate law. Furthermore, the aircraft was necessarily in alternate law given the unreliable airspeed indication–and the pilots presumably understood this.[/quote]I didn’t see any indication of the ‘alternate law’ warning going off on anything that was written. Considering that other warnings were going off in the cockpit, it might have been viewed as a distraction. Just the same, I do find that a control system that has an ‘alternate’ behavior to be a bit freaky. The transition between ‘normal’ and ‘alternate’ is not necessarily predictable nor controllable.

The problem I had with Stabilitrak is that it will grab the outside front brake if the tail starts moving out(sliding).. Under some conditions, this is actually the wrong thing to do. I had a problem when driving in Alaska.. where Stabilitrak cut in at the wrong time, making a slight low speed tail out condition worse.. then it decided that it didn’t know what it was doing.. releases the brake and I start to bring the vehicle straight.. then it decides.. “oh yes, I know what to do now”.. and grabs the other front brake.. causing rapid yaw in the opposite direction.. back and forth. I ended up locking up the brakes to stop it. – and yes, I do know how to correct a tail-out slide(doh.. turn into it and don’t touch the brakes)

[quote flinger]The artificial horizon displays the aircraft pitch and roll attitude, not the AoA. This instrument wasn’t necessarily displaying anything catastrophic, even after the plane was hopelessly stalled with a 40° AoA. [/quote]
It would when combined with the altitude indication and ascent/descent indicator. +40° and dropping in altitude makes a pretty convincing case for a stall and high AoA. This is why I got the impression that nobody was really looking at the instruments. The high pitch indicated by the artificial horizon would have also clued the senior pilot into the fact that the junior may have had the control all the way back.

[quote flinger]The static port is located on the fuselage somewhere (certainly on older general aviation aircraft), nowhere near the pitot tube.[/quote] Some have it combined.. port on the front is impact, port on the side is static (pitot-static tube).,see also http://www.grc.nasa.gov/WWW/k-12/airplane/pitot.html https://secure.steenaero.com/Store/site/product.cfm?id=B6853D82-102E-BB20-1868CF89D388401B http://store.diydrones.com/product_p/ac-0001-04.htm . In the case of the A330, Static ports are in a different area from the pitot tube and probably weren’t subject to the icing.

[quote flinger]Also, I believe the plane did contact the water in an almost level attitude–but descending at -10,000 feet/min. That is not survivable, and the plane disintegrated on impact.[/quote]I would have to disagree on this. The last thing the junior pilot did @ 2000 feet was to pull all the way back on the stick… again. At 10000f/m, 2000feet equates to more then 10 seconds. Enough to change orientation. Tail contacting hard, whipping the front down and a huge bending moment on the fuselage would cause tearing on the bottom of the fuselage with disintegration aided by forward motion relative to water hitting the tears in the fuselage. Realize that the landing on the Hudson (flight 1549), the plane was doing close to 120mph forwards.. or 10,560ft/min. The forward motion is likely to twist the wings and try to pull the nose down as the engines first make contact with the water (which is more like a solid at 120mph).

[KSMountain]

[quote=ucodegen][quote CDMA ENG]The reason, they explained, that you dont go to full power and point the aircraft down has everything to do with the wing design. The aircraft’s wing, if it is going to fast, start to build a shock wave over the leading edge and interrupts the airflow to the extent that the wing will lose lift.[/quote]Compressibility effect occurs trans-sonic/supersonic not low speed. This is why there were several accidents when trying to break the sound barrier. The shock wave prevented sufficient airflow from getting to the control surfaces. Next Miramar airshow, take a look at the leading edges on F18, F16s. You’ll find they are very narrow/thin. Compare that to a commercial sub-sonic plane, which has a fairly thick leading edge and to private planes which also have a fairly thick leading edge compared to their size and weight. The thicker leading edge gives an advantage due to a larger range of Angle of Attacks(AOA) that the wing will operate over (angle of attack being the angular difference between a line through the width of the wing and the vector representing the direction that the air is impacting the leading edge).

Compressibility is also the reason why supersonic aircraft try to have a smoothly transitioning cross section – ie have a coke bottle shape where the fuselage actually gets a little thinner in cross section when near the wings.

What happens at high AOAs is that as you approach stall, the laminar flow on the back(top) of the wing actually detaches from the wing surface. When that happens, you loose lift from the wing. There are all sorts of tricks done to prevent this. Next time you are on a flight, take the window seat next to the top of the wing. You’ll see small little blades sticking up from the top, near the chord of the wing(highest point, thickest in cross section) going lengthwise along the wing. You’ll notice that they are at a slight angle to the airflow. This is to create a vortex near the top wing surface that ‘glues’ the laminar flow down to the surface of the wing. If on the flight, you are on takeoff or landing and there is a moderate amount of moisture in the air – you can actually see the vortexes. Another thing to note, is that when flaps on commercial aircraft are deployed near their fullest, the leading edge of the wing actually pulls away from the wing and drops down. This is to bleed high pressure air from under the wing and ‘jet’ it close to the surface to ensure laminar flow at high Angle of Attacks.

There is another piece of the wing that you may see move. It will be near the fuselage of the wing is a large panel that tilts up from the top of the wing. Most of the panels will be in close to the fuselage. This is the ‘spoiler’ and it does what its name says. It ‘spoils’ the airflow, killing lift and increasing drag (also acting as a brake) when it comes up. It is useful at low speeds because the ailerons at the tips don’t have as much a bite on the airflow at low speeds. It has more of a ‘brake’ effect at higher speeds. They are mostly on the part of the wing nearest to the fuselage because they have a very large effect on the aircraft. If they were further out to the tip, they could easily flip or roll the aircraft.[/quote]
What a great post. Thanks ucodegen.

[KSMountain]

1) The avionics warned “stall stall stall” something like 75 times. Yet the word never came up in the cockpit conversation. Language issue? Sensory overload? Was it an issue that they didn’t “believe” the warning? If you truly internalized “stall” the recovery actions would have been more obvious.

2) Perhaps the root issue really was that the other pilots didn’t know what was going on with the right hand sidestick. Make that more obvious? You could say “detect conflicts between left and right” but that won’t be enough. In a disciplined cockpit only one pilot will “have the controls” at a time, which was generally the case here. They handed off verbally. The problem was when the right hand guy was in control the inputs were improper. How to detect that – even *more* software and alarms? Difficult situation. Yes there was a time when left and right pilots were both controlling but that was only a small portion of the accident, near the end, as I understand it.

3) I’m not sure there’s an angle of attack indicator in that aircraft. Others on this board may know. But even if the cockpit said “angle of attack, angle of attack, angle of attack”, they were ignoring the other warning that said “Stall”.

4) Even though it was mentioned above that “they were descending at 10000ft/min, they knew what was going on”, the popular mechanics article said they “incredibly” were debating whether they were descending.

5) When envisioning this accident, keep in mind this was at night over the middle of the ocean, in a storm. Probably no visual references whatsover. No horizon.

6) We can blame it on the “stupid guy in the right hand seat”, but I guess you could have a guy with a brain fart in the left hand seat. What to do? Averaging the inputs won’t be enough. Difficult problem.

[ucodegen]

[quote KSMountain]1) The avionics warned “stall stall stall” something like 75 times. Yet the word never came up in the cockpit conversation. Language issue? Sensory overload?[/quote]Could also be because they discounted the stall warning because the pitot tube initially froze over (which could cause the warning to occur erroneously), and then did not keep an eye on the airspeed indicator (which uses the pitot tube) to know if it ‘thawed’. NOTE: Added this after adding my ref to the A330 flight manual. It turns out the A330 also has a GPS based speed indicator (pg230).
[quote KSMountain]3) I’m not sure there’s an angle of attack indicator in that aircraft. Others on this board may know. [/quote]There generally is not. It takes a special pitot tube to get that information directly. It can be derived using GPS track info and orientation though. The military training pods are able to get that info either directly or using the GPS data and orientation (ACMI pod = the missile like thing they put on F14’s in Top Gun). The A330 has the AoA data (I suspect derived info) but does not present the AoA data to pilot.
[quote KSMountain]5) When envisioning this accident, keep in mind this was at night over the middle of the ocean, in a storm. Probably no visual references whatsover. No horizon.[/quote]That is why I mentioned IFR. The IFR tests involve flying, landing and takeoff with a hood over you so that you can’t see out the window, only using the instruments. Commercial jetliner pilots are supposed to have completed it. (IFR = Instrument Flight Rules).

Here is another link on the incident:
http://blog.seattlepi.com/aerospace/2011/05/27/report-may-point-blame-toward-pilot-of-doomed-air-france-airbus-a330/

One of the physical warnings of a stall is that the ailerons loose effectiveness, leading towards a tendency of the plane to want to roll. Note the end of the 4th bullet up from the bottom: “The airplane was rolling side to side at up to 40 degrees.“. Last bullet indicates that they crashed with a ground speed of 107knots. That is very near the lower end of the A330’s flying envelope. If the junior pilot had not pulled back on the stick when they noticed they were 2000 feet from the ground, they might have flown out of it. The engines were already completely spun up (the TOGA setting is literally ‘floored’). Minimum takeoff distance for an A330 is 2590meters/7500feet, but that is with an initial speed of 0. They already had 107knots. Takeoff/landing speeds are normally around 145/160knots. It would have been close and colorful though.. From what I remember, min takeoff distance also includes a safety factor for a mid takeoff engine failure, so it includes distance from ‘decision point'(145knts takeoff) to stopping the Aircraft on the runway. Minimum stopping distance is about 1080m/3240feet from touchdown to stop.

Some more info:
http://www.smartcockpit.com/data/pdfs/plane/airbus/A330/misc/A330-A340_Flight_Crew_Training_Manual.pdf
and even more info..
http://www.smartcockpit.com/data/pdfs/flightops/aerodynamics/Getting_To_Grips_With_Weight_and_Balance.pdf

[Author]

Posts