Investigators Suspect Computers Doomed Air France Jet 403
Posted
by
timothy
from the scary-thought dept.
from the scary-thought dept.
DesScorp writes "Investigators working with the wreckage of Air France flight 447 believe the aircraft suffered cascading system failures with the on-board computers, eliminating the automation the aircraft needed to stay aloft. 'Relying on backup instruments, the Air France pilots apparently struggled to restart flight-management computers even as their plane may have begun breaking up from excessive speed,' reports the Wall Street Journal. Computer malfunctions may not be an isolated incident on the Airbus A330, as the NTSB is now investigating two other flights 'in which airspeed and altitude indications in the cockpits of Airbus A330 aircraft may have malfunctioned.'"
Automation (Score:5, Insightful)
"The fancier they make the plumbing, the easier it is to stop up the drain." -Scotty
Re:Suspect?.... (Score:4, Insightful)
I dunno, the NTSB usually drags their feet before stating anything. They usually don't make statements about suspicion of what may have happened without specific evidence. This seems like an unusual announcement from them, not their usual style. I wonder if they are compelled to state a truth that they fell won't be properly addressed otherwise. After all, Airbus is built in Europe not the US.
Personally I wonder if they were compelled to state a suspicion that might otherwise not benefit business interests in the US. After all, Boeing is built in the US not Europe.
See how these stupid slurs work in both directions?
Re:GPS-based air speed (Score:5, Insightful)
Because GPS knows nothing about *airspeed*.
A GPS recorded speed of 100mph, into a 50mph headwind = 150 mph airspeed.
Re:Two things (Score:5, Insightful)
Flights are getting more and more automated. It used to be up to the pilot to take off and land, and the autopilot would fly the bit in the middle in good conditions. Now the autopilot takes off and lands too. The pilot is there in case of emergencies. But I would still wager that a computer would statistically be better than a human overall, otherwise the airlines wouldn't deploy this.
This case is of a plane travelling at such high speed and altitude that it only has a tiny window of opportunity between breaking up, stalling, or falling into the tempest below. If the computer systems keeping it in that window fail, then the pilot has little chance of actually fixing things. The alternative is to fly a lot more conservatively, with bigger margins of error. That would mean flying slower, and at lower altitude. Which means longer flights, that burn more fuel, hence cost more.
Re:Short version: (Score:5, Insightful)
Yes, because what we really need is pilots who can program in assembly while rewiring the control panel with their toes. Blindfolded. At mach 15.
You've watched one too many holywood flicks. If your computers crap out while airborne, you don't have time to troubleshoot and diagnose. You just follow the preset procedures, and hope that one of them works before you hit the ground.
Re:Suspect?.... (Score:4, Insightful)
The NTSB is the national transportation safety board. The criticism isn't that they shouldn't share their conclusions, it's that they may be politically/economically motivated to "share" mere suspicions which are detrimental to a foreign aircraft manufacturer.
Re:GPS-based air speed (Score:3, Insightful)
To go for the car analogy:
Imagine a (large) conveyor 100 miles long, stable enough for you to drive on in your car. Now drive from it's start to it's end in one hour. The distance you traveled is 100 miles, right?
Now imagine that conveyor moving in the opposite direction (i.e. towards you) at 50 mph. To still get from your starting point to your destination in an hour, you're doing 150 mph road speed. The GPS will still report 100 mph, but your car's tachymetre will report 150 mph, the wheels will revolve as is necessary to go 150 mph and, if you add 50 mph of headwind, even the air resistance will be equal to doing 150 mph without wind.
In an environment where the you need to stay in a 10 mph zone in order to avoid either stalling, rapid descent, crash, death if going too slow or plane breaking apart in mid-air, rapid descent, crash, death; it's quite helpful to know an accurate measurement. It's like Speed, except the bomb will blow up when your axle speed drops below 145 and the bus will spontaneously disintegrate at 155. Also, there's varying levels of wind. Also, you're driving on slicks. Through some kind of rally track half of which is concrete, the other half sand/dirt and the other half is jell-o.
Speculation (Score:5, Insightful)
Until they know what the actual cause is they should avoid speculation because it does absolutely nothing other then fill media headlines with non-sense.
Re:Suspect?.... (Score:3, Insightful)
Good observation.
The NTSB made an unexpected announcement on the B777 crash in LHR due probably to ice slurrie
in the fuel with uncalled for blame shifting just before the primary investigators in the UK
did their public announcement.
The NTSB going for partisan announcements is a very bad sign directly connected to
Boeing being in dire straits these days. So any published findings of the NTSB
may be completely worthless.
G!
MACC
Re:Suspect?.... (Score:5, Insightful)
The articles are pure FUD, and the summary is worse. The A330 doesn't need computers to "stay aloft" any more than your PSU needs an OS to power your motherboard. The rest of the functionality is pure gravy.
All these hysterical articles about computer failures always forget that the computers are a BONUS, and it is quite frankly becoming less and less insane to start believing in anti-Europe propaganda. It may indeed be true that pilots are becoming too accustomed to their presence, but in the meantime their high uptime has saved more lives from pilot error than the resulting complacency will ever cost.
This case is especially ridiculous, because modern computer controlled aircraft will actually handle sensor failures BETTER than ones without them. Had the Air France plane had the most up to date equipment, the computers would have used other sensors to estimate a safe range of approximate speeds and provided the pilots with a fast/slow indicator. It's even possible that this is exactly what happened, but something else went wrong.
Even if the computers just shut themselves down, it was still the sensor info that was invalid, so how would a plane without computers have fared any better?
It's also complete bullshit to say that the pilots can't override the computers. In normal flight, the computers *aid* the pilots. For example, avoiding a collision is easier in an Airbus, because pilots can just pull the stick back hard and the computers will automatically give the best possible climb performance, closer to the stall speed than a Boeing pilot would ever dare to go. Meanwhile, the pilot can look out the window instead of at an instrument panel!
This is what happened with the "infamous" crash into the forest. The pilot was too low and slow, and when he did pull up, the aircraft didn't "let him" because even maximum performance wasn't enough and the plane would have dropped like a stone had it been a Boeing. The computers probably saved everyone who did walk away from that crash!
IF the computers actually malfunction, they will turn themselves off. If they don't, the pilots can turn them off manually.
Re:Suspect?.... (Score:2, Insightful)
Re:This is why airbii make pilots nervous. (Score:4, Insightful)
> Also, i want a solid mechanical link between the controls I'm pushing on and the control
> surfaces on the wings...
You aren't strong enough to control an A330 with your muscles.
R2 SHUT DOWN ALL THE TRASH COMPACTORS ON THE ... (Score:5, Insightful)
Did the flight computer failure mode fail to (dis)engage? I've heard about the manual control levels that an Airbus flight system degrades through. It looks like the computer wouldn't get out of the way soon enough, so the flight crew kicked it in the head.
They received the airplane in a un-recoverable, un-flyable, disintegrating condition from mach turbulence destroying lift and ultimately the aircraft. (coffin corner [wikipedia.org])
Cascading failures generally occur from a synergy of multiple causes. In this case:
- A narrow flight envelope due to altitude and varying wind-speed in the storm. (had they climbed, trying to avoid the storm?)
- Pilot over-reliance on automated flight assist in marginal conditions.
- Failure of physical airspeed instruments due to severe icing from a massive updraft.
- Increased thrust from engines ingesting water contained in the 100mph updraft. (coffin corner!)
- Altitude increase from 100 mph updraft. (coffin corner!)
- Inappropriate computer control responses, destabilizing flight dynamics, leading to overspeed and unrecoverable loss of lift (mach stall).
- Turbulence and chaos of a severe storm masking the initial flight computer deviations.
Still human error. (Score:3, Insightful)
Like any other part of the plane, the computer is just another instrument designed and manufactured by people. Blame the programmer, the tester, the lack of analysis. The cause of this accident has nothing to do with computers. They just do what we tell them to. Leave them alone.
Re:No manual control? (Score:4, Insightful)
I did RTFA, and from what i understand of it it was impossible to get a reliable reading from the instruments in the cockpit, because the computers were failing and the airspeed-detector was unreliable (what seemed to be the primary cause of the failing of the computers). Manual control is fine, IF you know your altitude, airspeed etc. Try driving a car with blinded windows and a defective speedometer and an unreliable rev-meter.
I am not a pilot, but even I can understand that for manual control one has to have reliable data on what the plane is doing, which is exactly what was missing in this case (if the theory we are talking about is right).
Re:This is why airbii make pilots nervous. (Score:0, Insightful)
Re:Suspect?.... (Score:4, Insightful)
They just investigate and report which is what you want in an investigative body.
What the NTSB doesn't normally do is report unsubstantiated rumor to newspapers about investigations they have no direct jurisdiction over. While their job is certainly to get to the truth of why a plane crashed, in the absence of good evidence they can spin their version whichever way they choose. Unsurprisingly they have chosen to tell the story in a way that is detrimental to the design philosophy of the A330, just as European investigators would tend to blame Boeing if a 767 crashed and no reliable evidence was available as to why it crashed. Being dedicated to the pursuit of truth and being political are not at all mutually exclusive you know.
Re:Automation (Score:4, Insightful)
Sounds nice, but statistically the truth is exactly the opposite here. Over the years, planes have become increasingly safe and reliable with more technology (complexity), accident rates have steadily declined. And even today, the highest-tech aircraft are the safest ones - the big new ones flown by major airlines. Colgan Air 3407 [npr.org] wouldn't have crashed if the pilot hadn't been allowed to nose-up in response to a stall - a patently stupid thing to, which the A330 prevents [slashdot.org] according to another post in this thread.
Meanwhile, on the other side, we have the argument that this Air France A330 crash was due to a software failure that forced the crew to fly without the autopilot. This is theory is highly speculative, yet even if true, all it means is the autopilot is not directly to blame because it wasn't operable during the crash, i.e. humans were in control. So I don't understand the anti-automation spin on this story at all.
Re:A330 -- No Margin for Error (Score:2, Insightful)
There are a couple of aspects about the A330 problems that amaze me:
1. How can an airplane be allowed to carry passengers when the margin to airframe disintegration is so narrow? I can understand falling out of the sky if it stalls, but to be able to tear the airplane apart in level flight? What happened to margin of safety in airframe construction -- or is that whole concept now obsolete?
2. If the airplane can send fault messages home, why don't blackbox data streams get sent as well? At least that way there would be some situation info available as opposed to none.
3. In some ways reliance on flight computers is like reliance on spreadsheets or calculators -- if you do not understand what is going on and are not capable of doing it yourself then you cannot tell if the software is correct. Essentially, if the computer says it is so then it is, and you either survive or not.
1. Don't underestimate the power of wind shear. This plane may have been flying straight and level from the grounds point of view (we don't know that), but it was flying in the middle of a storm according to news accounts, likely experiencing some extreme wind forces.
2. The amount of telemetry and logging data generated by any aerospace system (air or space) is humongous, and even with an aircraft (as opposed to low data rate spacecraft), to large to transmit in real-time. In this case, the system did automatically transmit a wide range of critical telemetry packets which the original designers deigned the most important to transmit in emergency situations. The news articles are vague, but they do mention that those failure messages received were among a much larger set of automated data received.
3. In principal, I completely agree with that. In practice, that's rarely possible. A spreadsheet application can process a file containing 10,000 entries and calculate complex formulas on each one in seconds. Sure the user knows what these formulas do and could do it by hand, but it's not feasible for them to do so in a timeframe that would be useful before the data is outdated. In the manned space program (even in the Apollo days), everything was automated. The "manual" landing sequence was in fact linked to a computer that calculated the correct thrusters to fire based on the pilots desired course, there was no direct control, and no way for a human to calculate in real time exactly which thrusters to fire each second if there was. The pilots of modern airliners must be highly experienced on the principles of flight, but unless they designed the aircraft (and even then), there are often to many variables and control surfaces to monitor to do so without at least some computer assistance. Manual overrides are useful and should be there for redundant single-system failures, but most modern systems are far too complex for a human to be in full control of if all automation fails.
As others have said, computer failure is still only a theory until the black box is recovered.
(Disclaimer: I'm not a pilot or know much about airlines, but I do develop spacecraft flight software)
Re:Automation (Score:5, Insightful)
For years, aircraft manufacturers have had a philosophical debate over who should be in ultimate control of the aircraft. Boeing says that the pilot should be in direct control of the aircraft, and the computer should assist the pilot.
Oft repeated nonsense. The ultimate control of an Airbus, during fault conditions, is Direct Law, where the pilot control inputs are transmitted unmodified to the control surfaces, providing a direct relationship between sidestick and control surface.
http://www.airbusdriver.net/airbus_fltlaws.htm
Re:Suspect?.... (Score:3, Insightful)
Except there's no good evidence here to show that the NTSB is in any way being political; the statement isn't political in and of itself, and there's no evidence that there was any political pressure anywhere being applied.
Here's the facts: other organizations investigating the Air France crash have pointed to possible airspeed malfunctions as a contributing cause. Meanwhile, the NTSB has looked into similar matters and has announced it's looking into two completely separate cases in which it appears that the same kind of aircraft may have had airspeed indicator malfunctions. It has nothing directly to do with the Air France case.
And re: MACC's observation below, the NTSB reported that due to a flaw in the Boeing 777's engines there was an urgent need for a component redesign. I don't see how that's shifting blame away from Boeing at all. (And the British AAIB announced that the incident was probably caused by an accumulation of ice in the fuel system and also caused for a system redesign; that's not wildly different from the NTSB's statement.)
Re:Except... (Score:5, Insightful)
No, you're completely wrong. First of all, hydraulics move the control surfaces on all large aircraft. Nothing else has enough power while being light enough.
Old aircraft controlled the hydraulics with mechanical cables, newer ones with electrical cables (Boeing too). The computers in question are not needed for electrical signaling to the hydraulics systems.
The damage required to make the aircraft completely unflyable would be so severe it would affect any aircraft, and it has nothing to do with how well the computers are working.
When the computers went bye-bye, the pilots had complete control of the aircraft, as designed. Furthermore, the computers didn't malfunction - they turned themselves off because they couldn't trust the damaged sensors, but *neither could the pilots*. To characterize this as a computer problem just because they shut down is stupid and dishonest.
Re:This is why airbii make pilots nervous. (Score:1, Insightful)
As a passenger, I don't want you to have any of those things. If every computer on the plane dies, you won't be controlling anything. You'll be one of the dead computers. The issue that I have is that you're not the best computer for the job to begin with, any more. Even if you didn't spend half your time in flight school drinking and pretending to be Tom Cruise.*
*hint to potential future pilots: If you're not a naval aviator, Top Gun isn't about you. You look like someone repeating a joke that everyone else gets but you.
Re:Except... (Score:3, Insightful)
And nobody is arguing that the fly by wire system is what failed here.
Re:well that's terrifying (Score:3, Insightful)
I assume these kinds of modern planes can't even fly without a computer anymore.
You're wrong. They can.
Re:A330 -- No Margin for Error (Score:1, Insightful)
There are a couple of aspects about the A330 problems that amaze me:
1. How can an airplane be allowed to carry passengers when the margin to airframe disintegration is so narrow? I can understand falling out of the sky if it stalls, but to be able to tear the airplane apart in level flight? What happened to margin of safety in airframe construction -- or is that whole concept now obsolete?
2. If the airplane can send fault messages home, why don't blackbox data streams get sent as well? At least that way there would be some situation info available as opposed to none.
3. In some ways reliance on flight computers is like reliance on spreadsheets or calculators -- if you do not understand what is going on and are not capable of doing it yourself then you cannot tell if the software is correct. Essentially, if the computer says it is so then it is, and you either survive or not.
Disclaimer: I am a pilot, but not an airline pilot. I know enough about airliners to be a hazard, so to speak, but my take on these 3 questions:
1. Because more robust designs weigh more and cost more to operate. Therefore, the most important computer in this equation is the one that figures out how likely something like this is to happen and what it's going to cost vs. doing it this way and accepting the risk. In other words, and without knowledge of what actually happened here, many industrial accidents of all sorts begin on finance people's spreadsheets.
2. Ultimately because aviation authorities don't require it. However, I can also see pilots not liking it very much for good reason. Humans aren't perfect, and our lack of perfection is usually made up for in other ways. I've made mistakes and corrected them while flying. If every one was recorded in great detail, I can imagine what employers and insurance companies would do with that were I a professional pilot. What we'd need to do is get it codified into law that this data could only be used in accident investigations. Given the world's record on data privacy generally, I don't see this being easy to do.
3. Having flown light planes with glass cockpits, I can just say that you have no idea... They do change your way of thinking and reacting. The things I fly at least I can turn the computers off and the plane will still perform normally. Not the case in many things that fly these days.
Re:Suspect?.... (Score:1, Insightful)
Who the hell said a person could do that? Please explain to me how the computer is supposed to look out the window to do the same.
The computers can't continue to operate with faulty sensor readings, so complete control was given to the pilots. What do you expect should happen?
The most modern equipment won't even shut down, but rather extrapolate a safe speed estimate from other sensor input, and present it to the pilots. They could never do so on their own. Computer assistance is safer, period.
The Wall Street Journal story is misleading, IMO. (Score:5, Insightful)
Air France didn't begin replacing the malfunctioning pitot tubes in the Airbus until April 2009, and the tubes were not replaced yet in the crashed aircraft. The computers were not at fault apparently; there is no reason to suspect a computer malfunction.
Notice that the Wall Street Journal article, Computer Failures Are Probed in Jet Crash [wsj.com], says exactly that: "... seemingly beginning with malfunctioning airspeed sensors..." The "airspeed sensors" are the pitot tubes, which in the Airbus have been known for many years to collect ice in unusual conditions, and to stop giving reliable data.
The computers did what they were programmed to do, apparently. They stopped operating when they calculated that the data was bad. At that point the pilots needed to fly the plane themselves. However, the aircraft was operating in what is known in the aircraft industry as the coffin corner" [wikipedia.org]. There was apparently no way a human could fly the aircraft safely at the speeds necessary to get the craft to France in time, since in a severe thunderstorm the airspeed could not be known accurately enough to prevent overstressing the aircraft.
The Wall Street Journal apparently has NO new information. Here is a quote from the article: "The Air France crash could become the first since the 1980s in which U.S. and European investigators try to piece together a probable cause in a high-profile crash without the help of information from at least one of the plane's black boxes -- the digital recorders containing detailed flight data and cockpit conversations from the flight." There is apparently NO honest reason for the Wall Street Journal to publish an article now, claiming "Computer Failures".
Quote from a June 25, 2009 Aviation Week article, EASA: No Action Soon On A330 Pitot Tubes [aviationweek.com] published three days ago: "The pitot tubes have come under fire in the wake of the crash of AF447 because the accident aircraft, an A330-200, broadcast maintenance messages just before all contact was lost, indicating inconsistent speed information and potential problems with the pitot tube."
Should the Wall Street Journal be trusted for financial information? Apparently the publication did NOTHING to stop the present corruption in the financial departments of the U.S. government. Warren Buffett very publically called derivatives "financial weapons of mass destruction" [bbc.co.uk] beginning in 2002. The corruption was caused by the removal of laws designed to prevent fraud, at the beginning of George W. Bush's first term.
Apparently the Wall Street Journal always serves the profit of its advertisers and others in the U.S. financial industry. If publishing the article at this time and in the way it did indicates anything other than ignorance, it could be theorized that someone connected with the publication has investments in Air France or Airbus Industries.
Other similar incidents concerning the Airbus 330 are being investigated, according to a June 25, 2009 Associated Press news release, US panel probes 2 incidents involving Airbus A330s [nola.com]. The Wall Street Journal has access to the Associated Press, obviously. Why did it publish its misleading article two days later, which appears to blame the "computers"? The REAL story is apparently that apparently such incidents with the Airbus are common.
Here
Re:This is why airbii make pilots nervous. (Score:1, Insightful)
And your post is why pilots make passengers nervous.
I want the computer to do it. No offense, but you're only human.
Re:Still human error. (Score:3, Insightful)
If you arbitrarily redefine terms, anything can become anything else...
You're really stretching it to the breaking point, however, as any act of god can be written off as humans not making everything so unbelievably robust as to withstand all possible events.
Re:Suspect?.... (Score:3, Insightful)
Now find a safe runway, cover all your windows with heavy black plastic, disconnect your speedometer and try and accurately maintain 60km/h +- 3km/h over the length of the runway. Ten bucks says you park it in a fence a third of the way down.
That's the situation the pilots were in. No points of reference at all.
Re:Aerospace systems are made by humans, but... (Score:4, Insightful)
What a hideous and offensive generalisation: "everything is programmed using procedural code, and nothing works right anymore." That may be how *some* programmers work, but I give a sh*t, and I write concurrent (and more generally concurrency-safe) code all the time. And I can do that procedurally or by graph reduction or however you like.
As to: "Electrical engineers are trained in how to design things that really work"; do you have any snooty views about all EE grads being better people than all CS grads for example? My first interest was electronics but I don't see a halo.
Any other bigotry about "natural rhythm" or "education shrivelling the uterus".
I must be new here: I expect better reasoned objectivity from someone apparently able to type with reasonable spelling and grammar.
Rgds
Damon
Re:The Wall Street Journal story is misleading, IM (Score:5, Insightful)
tl;dr version:
On two occasions I have been asked, 'Pray, Mr. Babbage, if you put into the machine wrong figures, will the right answers come out?' I am not able rightly to apprehend the kind of confusion of ideas that could provoke such a question.
--Charles Babbage
Re:well that's terrifying (Score:1, Insightful)
This video shows an Airbus pilot switching off the flight computers then barrel rolling an A320:
No it DOES NOT: it shows the pilot rolling a flight simulator. He even says it would not work
in a real A320.
Re:A330 -- No Margin for Error (Score:1, Insightful)
"How can an airplane be allowed to carry passengers when the margin to airframe disintegration is so narrow?"
Because it won't fly otherwise. The design margins are like 110%, otherwise the plane is too heavy to take off. And at such high speeds, control surfaces that could survive full-deflection without destruction would be either too heavy or too small to be useful in other flight conditions like takeoff and landing.
"If the airplane can send fault messages back home, why don't blackbox data streams get sent as well?"
That would be a lot more data, too much bandwidth considering the number of planes in the sky at any given moment. And in 99.9999% of the time, the data wouldn't be useful--- and it would be crowding out the other .00001%. Now, you could argue that a system that senses a severe set of failures could step up its data broadcasts... if the hardware will facilitate that. I'm guessing that they already transmit as much as they can.
"In some ways reliance on flight computers is like reliance on spreadsheets or calculators"
Yes, and no. The pilots in the cockpit understand the basic physics behind flight pretty well. :) In a cascade of failures, however, they might get overwhelmed with conflicting information and be able to sort out how to respond to the problem--- if a successful response is even available.
Remember RFK's crash in 1999? Had he merely turned ON and USED his autopilot, all aboard would have survived the flight. Automation is a double-edged sword. Not to be dismissed lightly, but not to be used without justification, either. For a variety of reasons, today's planes need it and are getting a lot of benefit from it.
Re:Straw man troll (Score:3, Insightful)
a) A barrel roll can be a 1g maneuver, as this video shows: http://www.youtube.com/watch?v=Xp2Uc9XvmjY [youtube.com]
Any aircraft can do it if the pilot is good enough.
b) The article is the one pointing the finger at the machines, nowhere do I see it pointing at the piece of meat who decided to fly through the middle of an enormous thundercloud.
>I'm not sure what you were trying to prove. This video doesn't prove anything.
It clearly shows how many computers are on board, it clearly states that any one of them can fly the entire aircraft, it clearly says that the aircraft is designed so that all five of them cannot fail at once (ie. the aircraft would be in little pieces before that happened). It shows that you can switch computers off and still fly.
All of these Pesky Facts disagree with the article's description of the pilots struggling to reboot the computers on the way down.
Hey, but don't let that stop you frothing at the mouth and completely missing the point, because it's only a simulator (well duh!)
Re:Suspect?.... (Score:3, Insightful)
The thing is, these airplanes don't have a "computer override". There's no function that cuts in, takes away control from the pilots, and decides on its own to do what it wants. There are things called "limiters", which prevent the aircraft from exceeding certain well-defined parameters, but those are pretty rigidly defined mathematically within the control laws of the system, and not some "fuzzy" limit determined at the whim of a computer.
In my experience working on fly-by-wire systems, and from my personal perspective as an engineer and a pilot, a system like this should be designed to revert to "direct mode", where control surface deflection is directly proportional to stick throw (acting essentially like a traditional non-computerized aircraft) in the event of air data loss or if any doubt exists as to the quality of that data.
Re:Suspect?.... (Score:2, Insightful)