Airbus pic

haud

, a

and,

le that

ason

anyway.

sulted in

nd see if

s after takeoff,

chine

but there

xt -

Well, the very first Airbus fly-by-wires did have some issues that could=20 lead to them ignoring pilot input in favor of doing what the computer=20 thought prudent.=20

In , T wrote in part:

I do see that even now very occaisionally a polt in a fly-by-wire aircraft that has such a computer would disagree with the computer.

I have seen cockpits a few times, and I largely notice lack of a "horn button" on the object that is analogous to and largely being a "steering wheel". I think that is a good place to put a "pilot override" button. That button allows the pilot to do what the computer wants to disallow - with logging to recordable media in the nose, tail, and midsection of the aircraft. Pilots would get trained in use of the "pilot override button" - as in use it only when expectations are high that doing so is necessary. Unnecessary or unsuccessful use of this button leads to disciplinary action. If unnecessary use of this button causes no harm, disciplinary action is milder. If use of this button leads to *bad things*, then the pilot gets demoted in a humiliating public display, even if posthumously.

However, I have the impression that aircraft have a high-enough rate of being made well-enough to give little opportunity for improvement by suggestions by outsiders.

- Don Klipstein ( snipped-for-privacy@misty.com)

On a sunny day (Mon, 28 Jan 2008 18:03:17 -0800 (PST)) it happened Le Chaud Lapin wrote in :

IIRC that was a *landing*, that pilot always claimed the software did not give him power when he asked for it. Although Airbus claimed there was no soft error, the soft was changed after this. It is from that case right? It sucked in some tree tops :-) Here is the *real* video, from different viewpoints:

aud

nd,

e that

son

nyway.

ulted in

d see if

after takeoff,

hine

ut there

t -

The plane's terrain database obviously did not account for the height of the trees.

I like the idea of computer control with pilot override, and the location you suggest is not a bad place to put it.

There is another use for computers in general aviation that many pilots seem to be refusing to consider:

Computers are good for more than just auto-pilot. There are a huge number of control functions that might or might not involve flying the aircraft that could be done by computers, electronics, and electro- mechanical actuators.

Mechanical engineers are accustomed to achieving closed- and open- loop control functions using mechanical parts and mechanical techniques. They spend countless hours tweaking and re-tweaking trying to find just the right mechanical part to add to other mechanical parts to get the right control they want. They worry about weight, temperature, deformation, dirt, lubrication, balance, proper synthesis, all kinds of things. It should be intuitively apparent to them that, by opting to go with mechanics instead of electronics, they severely limit their reachability in the system space. If one replaces many of those functions with actuators and computers, not only would the system be less prone to wear, but very arbitrary points in the system space could be reached.

They also claim that electroncis are expensive compared to mechanical parts. This is simply not true, IMO. What they mean is that, if you take a $275,000 airplane and add a $15,000 computer to it because your friend implored you to modernized your control functions with a computer, the cost of the aircraft jumps to $290,000. Duh!

That's not what I'm suggesting. Incrementally "modernizing" an aircraft (or an automobile) like this is silly, IMO. It's like paying $2500 for a CD player in a Lexus. Why? No one at Lexus will ever convince me that there is something so special about their CD player that makes it worth $2500.

I'm suggesting an entire system rethink, so that when you are done, the cost of the entire airplane, including the computer (which should cost < $1000), is less than $100,000, because many of the problems that were present before become irrelevant in the new model. This, of course, implies that the new set of problems, mostly electronics related, are less burdensome than the old mechanical problems, which I believe to be the case.

-Le Chaud Lapin-

aud

riffi=3D

nd,

e that

son

nyway.

ulted in

d see if

after takeoff,

hine

ut there

The displays are obvciously presented their images by a graphics computer of some sort. There is also likely a separate smaller processor tied in with that to handle associated tasks like backlight brightness control and reading what appears to be a touchpanel.

Yes.

Most certainly did.

...

speed is low,

ting increase in power

I noticed that. That means the pilot was landing at far less than full power. The flap configuration also appears to be wrong for landing, and it even looks like the flap setting that he did have was done far too late.

In any case, whoever is saying that there is mystery about what happened here is lying, IMO.

When a pilot is about to land an aircraft, during the day, especially one of that size, they must be uncommonly focused. If he is about to crash, he *KNOWS* it. You can see DEATH coming at you at 200km/hour. You don't just sit there thinking, "Oh...dee dee dum...ho hum..we're all about to die..." you start:

1. ...uttering expletives (IIRC, research has shown that "sh*t" is the most popular expletive just before death).
2. ...being highly repentent to God, even if you're an atheist.
3. ...dislodging from your intestines in a most explosive manner whatever you had for breakfast.

And if you have been trained correctly as captain to maintain your composure during a crash...

1. ...going hrough the crash landing sequence with crew and passengers: "BRACE! BRACE! BRACE!"

The cockpit voice recorder will have recorded all of this. From the video, it appears that the pilot would have known at least 3km or more from the crash site that he was far too low. That's plenty of time to utter the words "SH*TTTTT" (or "MERRRRRRRDE!!")

Did they even say what he said just before crashing?

-Le Chaud Lapin-

Or perhaps the data is from 1960, and the trees grew at little.

Or maybe there were no trees there at all, and the tree-huggers planted them to make a point:

Break our reliance on oil!

-Le Chaud Lapin-

On a sunny day (Tue, 29 Jan 2008 07:59:12 -0800 (PST)) it happened Le Chaud Lapin wrote in :

I dunno, I did see a direct interview with that pilot many years ago here on TV. He was no longer flying, at least not for Airbus I think. He maintained it was a software problem.

I am not that negative about pilots, sure 's*it' may well come out of my mouth too, faced with a Darwin Award like situation you cannot avoid, but in my experience (personal) I go into some sort of super concentration, I had this with cars.... electricity too. You cannot 'play' cool dude, in my view, you gotta automatically give all you have. Some panic, but it is skill only that saves.

In a modern cockpit there can be soo much info coming at you, all those TV screens, that I really think a computer can take better decisions in many cases (if programmed right, and that is the issue). Bugs in software (or basic shortcomings or wrong program flows, can happen, but if they put some effort in it should be possible to fly by voice control :-)

Computer: 'Engine 1 on fire, extinguishing it, relax and enjoy the trip'. ... time passes Computer: 'Sorry could not put out fire in number 1, do you have a parachute?'

Just fixed some nasty bug in my PIC programmer. in the code sometimes some instruction was not executed... did a hex dump finally, and one location was always partly zero..., so whatever was there, was skipped.... Took me a day, especially to figure it was not an PIC asm coding error, and to really go after it. Need to persist. They should finally make good software. But you cannot deny, and look the other way, and claim it is OK, if Airbus did that, then they damaged themselves and the pilot.

Yes, I agree. I only mention that people say "sh*t" to say that one can use that to determine at which point during final approach the pilot knew there was an issue. I had that experience on my Honda VFR-800

).

the key issue, IMHO...if it's programmed right, whole new worlds of opportunity will be opened.

That's way I code also. For all the testing procedures, etc...I think only people of certain mindsets should be allowed to write code for critical components. There should be a psychological exam given to each prospective engineer to make sure s/he understand the seriousness of what s/he is doing.

I also have a strong suspicion that there is a correlation between people's engineering thought process and toilet hygiene.

I wouldn't put it past them. That's what happened with Space Shuttle. Twice.

-Le Chaud Lapin-

especially the seat covers, they could have found something less terrible guess the seats were certified 50 years ago and they never thought to find something else

-Lasse

that

anyway.

resulted in

see if

after takeoff,

machine

there

I believe the airbus that crashed in 2001 lost its tail because the pilot made a manoeuvre that pushed the tail past its design limits. In that case it would have been smart if the computer had refused to do so.

On the other hand if trying to avoid a crash or collision you'd probably want the computer to what you want taking your chances hoping the plane holds together

-Lasse

It would seem to me that this would be yet another tree to hang the crew for "pilot error". "He touched the horn => the plane crashed".

Dunno, ever see a DC-10? There were a *lot* of outsiders telling them it was crap, from day one.

I don't think a 380 uses coins for its tolls. Will it lift that much? ;-)

You guessed wrong.

Software control assumes that the programmer in advance could foresee every possible event and situation (reality is a bitch :). And that the user supplied the correct parameters. And that instruments work correctly (and is not plugged by maintaince personell..). If anything of this goes wrong output will be bad (s*it in -> s*it out system).

Also one have to consider failure modes. Anything like tin whiskers, sun flares, made in china(tm), thunder, transients etc..

So a pilot override where a minimal amount of redudant control systems exists between the pilot and the airplane sounds like a good idea.

The Aeroflot Flight 593 the Airbus plane did something the pilots were not aware of. And if they had a drop-computer-I-run-the-show-button to establish a basic fallback mode where they know exactly what the plane was upto. They might have prevented the fatal crash.

Some aircraft investigations made it obvious that: * Seeing if the enginees is on fire might be useful (out view normally) (normal instruments didn't provide useful data) * Seeing passenger cabin, especially if there's smoke etc.. * Abnormal heat detectors in electrical spaces.

I guess this aircraft doesn't come cheap. So I wonder if they have spent some effort to get these features into it?

Yes, and an enormous number of other things could be possible if a computer were the centerpiece of control, things that, if done now using conventional means, not only would the proposition seem weird to the designer, but the cost would gro exponetially.

Take a flasher, for example.

The mechanically-oriented hardware designer thinks..."Ok, I need to read about appropriate timing circuits, those that meet FAA regulations, what effect the environment will have on the delay element, how much current should be supplied to achieve proper intensity, etc., failure rates, special circuits to tell me if it failed, etc."

The software-oriented designer thinks..."Whatever. as long as you give me minimum circuitry, not only can I hit the FAA specs, but I can do in software for *ZERO* incremental material cost pretty much anything you want, like flashing at whatever rate you specify, many in fact, say, 10,000 different rates, even though that's silly, variable intensity, that depends on environment, whose sensors are, of course, not hard-wired, but again fall under software control, and all the data you need about failure rates. I can do this not only for the flashers, but almost any device on the aircraft."

The different in cost of these two modes of thought is simply incomparable.

ome

They probably did. But, IMO, the entire aircraft industry (and automobile industry), is plagued with an infestation of self- preservation.

Why do for $10,000 what is currently done for $100,000 when the $10,000 method will obviate your job?

And then there is simply the matter of ignorance. I know for a fact that a large computer maker has routinely preyed on the ignorance of customers sold $4,000 computers to relatively aflluent individuals who were simply looking for an $800 machine for basic Internet surfing. It was policy - "If an ignorant customer calls and does not understand computers, make sure you burn up their wallet, irrespective of what they think they want."

If a designer has never experienced the power of software control of commoditized CPU's and sensors versus the tedium of one-shot customized designs, how can they know?

-Le Chaud Lapin-

The reason for suggesting these things is that a simple camera can show in no uncertain terms that, well the reason for the weird readings and motor not wanting to start is that it's *on fire*. And getting filled with vulcanic ash. The pilots were not aware of these issues until told upon landing..

Another incident is smoke filling the passenger cabin while pilots were unaware of the issue ie fire. A camera could show smoke, and t3rror1sts.

It's not about price, but that pilots ought to have these things available.